Hierarchical organization of the circadian timing system Steensel, M.J. van

Citation

Steensel, M. J. van. (2006, June 21). Hierarchical organization of the circadian timing system.

Retrieved from https://hdl.handle.net/1887/4418

Version:

Corrected Publisher’s Version

License:

Licence agreement concerning inclusion of doctoral thesis in the

_{Institutional Repository of the University of Leiden}

Downloaded from:

https://hdl.handle.net/1887/4418

Chapter 1

Organization of cell and tissue pacemakers

in the circadian timing system;

A comparison between species

Department of Molecular Cell Biology, Laboratory for Neurophysiology, Leiden University Medical Center, P.O. Box 9600, 2300 RC Leiden, The Netherlands

SUMMARY

INTRODUCTION

As a consequence of the rotation of the earth, light and darkness alternate in a 24-hour fashion. Most organisms throughout the plant and animal kingdom possess a circadian timing system that allows them to cope with and anticipate to daily changes in their environment, such as nutrient availability or the presence of predators. By integrating environmental information, the pacemaker signal reflects a reliable representation of the day-night cycle, which is communicated to the body. For many species, a main circadian pacemaker has been identified in discrete regions in or close to the brain. These regions include the suprachiasmatic nuclei of mammals (Ralph et al., 1990), the basal retinal neurons of the marine mollusk Bulla gouldiana (Block and Wallace, 1982), the pacemaker neurons in the eye of the mollusk Aplysia californica (Jacklet, 1969a; Jacklet and Geronimo, 1971; Jacklet et al., 1996), the lateral neurons of

Drosophila (Ewer et al., 1992; Frisch et al., 1994; Helfrich-Förster, 1998) and

the optic lobes of the cockroach and the cricket (Page, 1982; Tomioka and Chiba, 1984). For the zebrafish, the presence or location of a central circadian pacemaker has not been revealed. Possibly, many cells and tissues in the zebrafish contain circadian clocks (Cahill, 2002).

Rhythm generation occurs within individual cells of the pacemaker and is based on positive and negative molecular feedback loops consisting of several clock genes and their protein products. For mammals, the positive feedback loop comprises the proteins of the clock genes Clock and Bmal1 (see for review, Reppert and Weaver, 2002; Lowrey and Takahashi, 2004). These proteins heterodimerize into a protein complex, which translocates into the nucleus and activates the transcription of the clock genes Period (Per) and

Cryptochrome (Cry). PER and CRY proteins subsequently form a complex that

blocks CLOCK/BMAL1-induced transcription by interaction with CLOCK and/or BMAL1, thereby forming the negative feedback loop. Posttranslational mechanisms, such as protein phosphorylation, play an additional role in rhythm generation and period determination. The full length protein product of timeless (tim) may be part of the negative feedback loop, but its precise function remains to be determined. The positive and negative feedback loops are connected to each other via REV-ERBα, which inhibits Bmal1 transcription. Transcription of

Rev-erbα is activated by the CLOCK/BMAL1 complex and inhibited by PER/CRY through its interaction with CLOCK/BMAL1.

review, see Panda et al., 2002; Stanewsky, 2002). In the zebrafish, the proteins of two Bmal genes can bind with CLOCK. Moreover, three per and six cry genes have been identified. The protein products of Cry1a, Cry1b, Cry2a and

Cry2b can block the activation of transcription by mammalian CLOCK/BMAL1

(for review, see Cahill, 2002). Together, the data indicate that for mammals,

Drosophila and zebrafish, clock genes are well conserved between the species.

For both Drosophila and mammals, there are indications that electrical activity of pacemaker neurons is essential for maintaining the cell-autonomous circadian oscillation. Electrical silencing of Drosophila pacemaker neurons by targeted K+ _{channel expression stops normal freerunning TIM and PER}

oscillations in the pacemaker neurons (Nitabach et al., 2002, 2005). Moreover, application of tetrodotoxin (TTX) to an SCN slice suppresses the amplitude of single cell Per1–bioluminescence rhythms (Yamaguchi et al., 2003). While these observations indicate that coupling between molecular and membrane processes may be critically involved in rhythm generation, intracellular coupling mechanisms fall outside the scope of the present review.

The molecular feedback loop in individual cells appears to provide an adequate explanation of the oscillatory properties of single pacemaker cells. Dissociated mammalian SCN cells display a rhythm in the neuronal firing rate (Welsh et al., 1995) and individual basal retinal neurons of Bulla show circadian changes in membrane conductance (Michel et al., 1993). While generation of circadian rhythmicity is accomplished at the single cell level, it has become apparent that other attributes of the circadian system arise at the tissue level, as a consequence of interactions within and between neuronal networks. Coupling occurs at many levels of organization within the organism, for example between single cell oscillators, between groups of synchronized single cell oscillators, and between the clock and its target areas. Although the molecular basis of circadian rhythmicity has been well reviewed, the neurophysiological and system-level properties have received less attention. Consequently, the present review focuses on the progress made in understanding circadian rhythms at the tissue level, including responsiveness to environmental stimuli, peripheral signals from the body, and signals from other parts within the clock. Taking advantage of a comparative analysis, we have included a number of animal species, namely fruit flies, crickets, cockroaches, snails, zebrafish and mammals. By comparing the pacemaker structure, coupling pathways and interaction with peripheral signals we will attempt to identify the general principles of organization involved in circadian timing. Specifically, we will evaluate to what degree network properties contribute to the characteristics of the circadian timing system.

DROSOPHILA

Drosophila pseudoobscura and Drosophila melanogaster are two of the most

the anatomical location of the fly pacemaker were aided by the discovery of three different mutants of Drosophila melanogaster causing arrhythmicity or a long or short period in eclosion and locomotor activity. These phenotypes all resulted from mutations of a single gene, called period (per) (Konopka and Benzer, 1971). Using mosaic analysis, i.e. a technique where a mutation is induced in a portion of cells, Konopka and colleagues were able to demonstrate that the brain is the locus of the action of per in controlling the circadian rhythm in locomotor activity (Konopka et al., 1983). Subsequently, expression of per in a group of neurons called the lateral neurons (LNs) was found to be sufficient for robust circadian locomotor rhythmicity (Ewer et al., 1992; Frisch et al., 1994). Moreover, most flies of the “disconnected” (disco) mutant strain of

Drosophila, which lack functional LNs, are arrhythmic in their locomotor

behavior under constant darkness. In the occasionally observed rhythmic disco fly, one or more LNs with their projections are retained (Helfrich-Förster, 1998). These data indicate that the LNs are important for the expression of circadian rhythms in Drosophila behavior.

Figure 1. Location of the different subsets of clock cells in the Drosophila brain

s-LNv = small ventral lateral neurons, l-LNv = large ventral lateral neurons, LNd = dorsal lateral

neurons. DN1, DN2, DN3 are three subsets of dorsal neurons. (Modified from Helfrich-Förster, 2003).

Two major subtypes of LNs are present in the Drosophila brain: the ventral lateral neurons (LNvs) and the dorsal lateral neurons (LNds) (Figure 1). The

LNvs project to the accessory medulla, an area located at the base of the

peptide pigment dispersing hormone (PDH) (Helfrich-Förster, 1995). PDHs are a family of octadecapeptides, which were first isolated from crustacean species. Their homologue in insects is referred to as pigment dispersing factor (PDF) (see e.g. Helfrich-Förster and Homberg, 1993). Most flies carrying a null-mutation of the pdf gene or PDF cell ablations are arrhythmic in constant darkness, confirming the importance of these neurons for locomotor rhythmicity and suggesting a role for PDF in their output (Renn et al., 1999; see also Helfrich-Förster et al., 2000). Interestingly, individual LNvs in pdf-mutant flies are

rhythmic, but desynchronize after several days in constant darkness. In addition, the phase and amplitude of the rhythms in the LNds are altered in

these mutants. These data indicate that PDF-induced synchronization between individual LNvs and PDF-mediated signaling from the LNvs to the LNds are

required for the expression of rhythmic behavioral activity (Lin et al., 2004; see also Peng et al., 2003).

Regulation of locomotor rhythmicity by two groups of neurons

Many strains of Drosophila express bimodal patterns of locomotor activity in light-dark conditions, with a first component around lights-on (morning component) and a second around lights-off (evening component) (Figure 2). Generally, strong anticipatory behavior is present before the light-to-dark and dark-to-light transitions. Under constant dark conditions, some flies display unimodal activity rhythms, whereas the behavior of others remains bimodal (Helfrich-Förster, 2000). In constant light, wildtype flies become arrhythmic (Konopka et al., 1989). In contrast, cryb mutant flies, which lack cryptochrome

as one of the light input factors, remain rhythmic in constant light (Emery et al., 2000a; Helfrich-Förster et al., 2001) and, at higher light intensities, express two activity components that freerun with different periods (Yoshii et al., 2004). Using immonohistochemistry, it was demonstrated that PER expression in the LNvs of cryb mutants in constant light correlates with the short period

component in behavior (Yoshii et al., 2004). Moreover, PER expression was measured in a dorsally located group of neurons, which included the LNds and

another subset thought to be involved in circadian rhythmicity: the dorsal neurons (DNs) (Figure 1). PER expression in this dorsally located group shows bimodal peaks, correlating with the long and short period components in behavior. The data suggest that Drosophila locomotor rhythms are regulated by two circadian oscillators that respond differently to light. Note that the PER expression rhythms remained bilaterally synchronous in these flies, suggesting relatively strong bilateral coupling. This can also be inferred from the presence of direct neuronal projections between the accessory medullae (Helfrich-Förster, 2004).

In addition to the study by Yoshii and colleagues (2004), other studies provide evidence for discrete functions of different groups of pacemaker cells in the control of Drosophila locomotor rhythmicity. Flies in which the PDF+ LNvs are

anticipation, whereas lights-off anticipation is retained (Stoleru et al., 2004; see also Renn et al., 1999; Blanchardon et al., 2001). Moreover, flies without CRY+PDF- cells, which mainly comprise the LNds, do not show lights-off

anticipation (Stoleru et al., 2004). Correspondingly, per0 _{mutant flies in which}

PER is selectively expressed in LNvs show clear morning, but not evening,

anticipatory behavior (Grima et al., 2004). In animals with PER expression in LNvs and LNds, both morning and evening activity components are present.

Together, these data suggest that the LNvs generate the morning activity

component of Drosophila, whereas the evening activity component may be mediated by the LNds.

Interestingly, PER expression in LNvs, but not LNds, results in circadian

rhythmicity in constant darkness (DD) (Grima et al., 2004). Moreover, ablation of PDF-expressing LNvs abolishes rhythmicity, whereas flies with ablated

CRY+PDF- neurons are rhythmic in DD (Stoleru et al., 2004). These results indicate that the morning oscillator in the LNvs is sufficient to generate circadian

rhythmicity in constant darkness, whereas the evening oscillator by itself is not able to induce sustained rhythmicity in locomotor activity under this condition.

Figure 2. Average distribution of the locomotor activity of Drosophila (n = 55) during seven days

under a light-dark cycle

White and black bars indicate the activity during the light and dark phase, respectively. The locomotor activity pattern is bimodal, with anticipatory activity present before the dark-to-light and the light-to-dark transitions. ZT = Zeitgeber Time. (Modified from Taghert et al., 2001).

Two subsets of LNvs exist: the small and the large ventral lateral neurons

(s-LNvs and l-LNvs) (Figure 1). Selective PER expression in only the l-LNvs does

to the s-LNvs, robust rhythmicity is not always observed in the l-LNvs (Veleri and

Wülbeck, 2004). Whereas the available data suggest that the s-LNvs (together

with the LNds) contain the principal circadian pacemaker, the l-LNvs may play a

role in coupling mechanisms (see also Veleri and Wülbeck, 2004; Helfrich-Förster, 2005).

The role of the DNs is still relatively unknown. They may contribute to the generation of the evening activity component, but they may also play a role in synchronization to the external light-dark cycle. For example, disco flies, which lack functional LNs, are able to entrain to light-dark cycles (Helfrich-Förster, 1998). Moreover, animals that express PER only in the DNs are able to synchronize their behavior to light-dark cycles, but do not show circadian rhythmicity under constant darkness (Veleri et al., 2003). Future experiments have to clarify the role of the DNs in freerunning and entrained locomotor rhythmicity in Drosophila.

Eclosion and the A-B-oscillator model

One of the first reports suggesting the presence of multiple circadian oscillators came in the late 1950’s from Pittendrigh and coworkers. They employed

Drosophila pseudoobscura as a model system to study the rhythm in eclosion in

populations of flies. In a 12:12 light-dark cycle, eclosion of Drosophila occurs approximately three hours after lights-on (Pittendrigh and Bruce, 1959; Pittendrigh, 1965; Saunders, 1976). After initiation of rhythmicity, eclosion is rhythmic under constant darkness and shows a freerunning period close to 24 hours (Skopik and Pittendrigh, 1967; Saunders, 1976). Single light pulses affect the phase of the eclosion rhythm and induce phase delays when presented during the early night and phase advances when given during the late night. These light-induced phase shifts do not reach their full magnitude immediately. Instead, the magnitude of the phase shift increases in the course of several cycles. During the transient cycles, no further light stimulation is required, indicating that the magnitude of the steady state phase shift is already set at the moment of the light pulse (Pittendrigh et al., 1958). Note that it is especially the phase advances that do not reach their full magnitude immediately. There is an asymmetry in the response of eclosion to phase advances and delays.

either phase advancing or delaying light and 2) entrainment of the B-oscillator by the A-oscillator (gradually in the case of phase advances), by temperature or by both (see also Pittendrigh and Bruce, 1959).

The question arises whether the oscillators of the LNvs, LNds and DNs, which

are now thought to play specific roles in the control of locomotor rhythmicity, could also be responsible for the observed effects in the timing of eclosion. Certain genetic mutations affect eclosion and locomotor rhythmicity in a similar manner, supporting the concept of a common pacemaker (Konopka and Benzer, 1971; Dushay et al., 1989; Dushay et al., 1990; Konopka et al., 1991), but others do not (Newby and Jackson, 1991; Newby and Jackson, 1993; Wülbeck et al., 2005). Phase shifting induced by light pulses occurs almost instantaneously in locomotor activity rhythms (Dushay et al., 1990), which is in contrast to the transients in eclosion, but it must be noted that these data were derived from different Drosophila strains. Support for the involvement of different oscillator systems comes from studies that show differences in freerunning periods of locomotor behavior and eclosion within Drosophila

pseudoobscura and Drosophila melanogaster (Engelmann and Mack, 1978;

Sheeba et al., 2001). Moreover, the Drosophila circadian system can be synchronized by light administered during early developmental stages, when LNds and most DNs are not yet developed (Brett, 1955; Sehgal et al., 1992;

Kaneko et al., 1997).

Whereas the LNs are required for the timing of eclosion (Blanchardon et al., 2001), the presence of an intact molecular clock in these neurons is not sufficient (Myers et al., 2003). Cells of the prothoracic gland, a peripheral tissue involved in development, show a rhythm in per-driven bioluminescence, which is directly entrainable by light (Emery et al., 1997). When the molecular clock in these cells is disrupted, eclosion rhythmicity is absent, whereas locomotor rhythmicity in adult flies remains. A model was proposed, in which a circadian clock in the LNvs secretes PDF and signals indirectly to a second circadian

clock in the prothoracic gland. The prothoracic gland would subsequently control the timing of eclosion (Myers, 2003; Myers et al., 2003). Note that this model shows conceptual similarities with the A- and B-oscillator model. In conclusion, while circadian control of eclosion is specifically dependent on the prothoracic gland, locomotor activity is not. However, both eclosion and locomotor rhythmicity rely on the functional integrity of the LNs.

Autonomous oscillators in the periphery

Indications for the autonomy of the peripheral oscillators came from a study showing that transplanted Malphighian tubules maintain their original phase, despite the fact that the host flies were entrained to a reversed light-dark cycle before transplantation (Giebultowicz et al., 2000). In the normal situation, Malphigian tubules may be directly entrained by light, as peripheral insect tissues can be immediately phase shifted by a shifted light-dark schedule, also in absence of the brain (Giebultowicz et al., 1988; Giebultowicz et al., 1989; Giebultowicz and Hege, 1997; Hege et al., 1997; Plautz et al., 1997).

Figure 3. Expression of PER-reporter in Drosophila Malphigian tubules under constant

darkness

In both intact and headless flies, PER expression shows a circadian rhythm. On the bottom of the figure, the light-dark regime is indicated, with shaded bars representing the subjective day. (From Hege et al., 1997, Used by permission of Sage Publications, Inc.).

Circadian rhythmicity in certain behavioral and physiological processes may be mediated by peripheral oscillators, rather than by the central pacemaker. For example, flies expressing per only in the LNs are rhythmic in locomotor behavior, but not in olfactory responses (Krishnan et al., 1999). Mutant flies in which the LNs or PDF is absent are arrhythmic in their locomotor behavior, whereas olfactory behavioral rhythms persist (Zhou et al., 2005). These data indicate that olfactory rhythms are not mediated by the central pacemaker in the LNs. Indeed, it has been observed that the presence of an intact genetic clock in only the antennal neurons is sufficient to drive circadian rhythmicity in olfactory responses (Tanoue et al., 2004).

Conclusion

Taken together, the data indicate that circadian rhythmicity in different behavioral and physiological parameters of Drosophila may be mediated by (partially) different sets of circadian oscillators. The two components of rhythmic locomotor behavior are mediated by the LNvs on one hand and the LNds and

tissues of Drosophila contain circadian oscillators, which can be light-responsive and which can drive rhythmicity in physiological parameters such as olfactory responses. Coupling or synchronization occurs between individual pacemaker cells and between pacemaker centers and often involves PDF.

COCKROACHES AND CRICKETS

Most crickets and cockroaches show nocturnal patterns of behavioral activity that persist under constant darkness (Leucophaea maderae and Periplaneta

americana: Roberts, 1960; Roberts, 1962; Gryllus bimaculatus: Tomioka and

Chiba, 1982; Gryllodes sigillatus: Abe et al., 1997; but see Sokolove, 1975a for

Teleogryllus commodus). Several studies point to the optic lobes in the brain as

the site of the main circadian pacemaker for behavioral activity rhythms in the cockroach (Nishiitsutsuji-Uwo and Pittendrigh, 1968a; Page, 1982) and the cricket (Sokolove and Loher, 1975; Tomioka and Chiba, 1984; Abe et al., 1997). Moreover, the optic lobe pacemakers seem to be the crucial element for the expression of rhythmicity in a range of other functions, such as cockroach ERG amplitude and olfactory responses as well as cricket stridulation and spermatophore formation (Loher, 1974; Sokolove and Loher, 1975; Wills et al., 1985; Page and Koelling, 2003; but see Chang and Lee, 2001).

Within the cockroach optic lobe, a small area, called the accessory medulla (Figure 4), is innervated by a set of nearby located PDH immunoreactive neurons (Homberg et al., 1991). Transplantation of the accessory medulla with associated PDH immunoreactive neurons restores circadian rhythmicity in optic lobe-ablated cockroaches (Reischig and Stengl, 2003), suggesting that the accessory medulla is the specific location of the cockroach pacemaker. Moreover, the data indicate a role for PDH immunoreactive neurons as pacemaker cells.

No general agreement exists on the specific location of the cricket circadian pacemaker. Interesting similarities exist, however, between insect circadian systems (see for review, Helfrich-Förster et al., 1998; Helfrich-Förster, 2004). Similar to the cockroach and Drosophila, a group of PDH- or PDF-immunoreactive neurons is present in the cricket optic lobe, with projections into the accessory medulla. Based on the importance of PDF positive neurons for behavioral rhythmicity in Drosophila and the cockroach, and on information obtained in several lesion studies, the accessory medulla with associated PDF neurons is suggested as the location of the cricket circadian pacemaker within the optic lobe (Figure 4) (Tomioka and Chiba, 1984; Homberg et al., 1991; Abe et al., 1997; Helfrich-Förster et al., 1998; Lupien et al., 2003; Helfrich-Förster, 2004; but see Okamoto et al., 2001; Tomioka and Abdelsalam, 2004).

al., 2003; Reischig et al., 2004). Moreover, a recent study proposed a role for PDF in synchronization of accessory medulla neurons in the cockroach (Schneider and Stengl, 2005). The multiple functions of PDF or PDF positive neurons correspond with observations in Drosophila, suggesting that PDF plays a central role in insect circadian timing systems.

Figure 4. Location of the cockroach and cricket circadian pacemaker

Schematic drawings of the brain of the cockroach (A) and the cricket (B), showing the lamina (La), medulla (Me) and lobula (Lo) of the optic lobes. Black dots indicate the optic lobe circadian pacemakers, which are located in the accessory medulla. (Modified from Helfrich-Förster et al., 1998).

Bilaterally paired optic lobe oscillators

cockroach circadian pacemaker system contains two bilaterally located, mutually coupled oscillators.

Several PDH immunoreactive neurons directly connect both accessory medullae, providing anatomical evidence for the presence of a neuronal coupling pathway in the cockroach (Reischig and Stengl, 2002). These data indicate an important role for PDH in coupling of bilateral pacemakers in this species, which can also be inferred from the phase delays it induces upon injection near the accessory medulla (Petri and Stengl, 1997). Computer simulations suggest that a phase advancing substance is involved in the coupling pathway as well (Petri and Stengl, 2001), but this substance remains to be identified. Interestingly, there are indications that transport of circadian information between the bilateral optic lobes is mediated by a different pathway than contralateral photic entrainment. The transport of circadian information is thought to be regulated by the PDH-positive cells that directly connect both accessory medullae. Contralateral light-entrainment could be controlled by another, PDH-negative, group of neurons (Reischig et al., 2004).

Interestingly, two components can be observed in the locomotor behavior of a small percentage of cockroaches under certain circumstances, e.g. in constant light or following a low temperature pulse. It will be important to identify the anatomical loci for these components (Roberts, 1960; Wiedenmann, 1977a; Wiedenmann, 1980).

Similar to the cockroach, each of the cricket optic lobes is able to drive rhythmicity in overt behavior (Loher, 1972) and the freerunning period of locomotor behavior is changed by removal of one of the optic lobe pacemakers (Okada et al., 1991). Unilateral blinding and subsequent exposure to constant light or light-dark cycles with a long, e.g. 26-hour, period gives rise to the expression of two components in the behavioral rhythms, each of which originates from one of the optic lobe pacemakers (Figure 5) (Wiedenmann, 1983; Tomioka et al., 1991; Tomioka, 1993; Ushirogawa et al., 1997). The data indicate that the two bilateral oscillators in the optic lobes of the cricket are only weakly coupled. Two components of cricket rhythmicity can also occur spontaneously, especially under constant light conditions, and as a result of low temperature pulses or phase shifts of the light-dark cycle (Wiedenmann, 1983; Wiedenmann and Loher, 1984; Tomioka and Chiba, 1987; Tomioka et al., 1991).

received by the dorsocaudal area of the compound eye (Tomioka and Yukizane, 1997). A subset of medulla bilateral neurons (MB-4) has its receptive field in this area of the compound eye, suggesting that this subset is involved in the transmission of photic information (Yukizane et al., 2002). Future studies may reveal which subset of neurons is involved in the transport of circadian information between the bilateral optic lobe pacemakers.

Figure 5. Schematic representation of double plotted actograms of cricket behavioral activity

rhythms

Black bars indicate the presence of behavioral activity. When crickets are unilaterally blinded on the day of transfer from light-dark (LD) to constant light (LL), the behavioral activity pattern expresses two components that freerun with different periods. Removal of the intact optic lobe (indicated by the asterisk in A) results in disappearance of the behavioral activity component with the long period, whereas removal of the blinded optic lobe (indicated by the asterisk in B) abolishes the short period component. This indicates that the two behavioral activity components are mediated by the bilaterally located optic lobe pacemakers. (Based on Figure 4 of Tomioka et al., 1991).

Other oscillators or pacemakers in cockroaches and crickets

Pittendrigh, 1968a; Sokolove and Loher, 1975; Helfrich-Förster, 2004). More experiments are required to determine the role and importance of the pars intercerebralis relative to that of the optic lobe pacemakers in the control of overt rhythmicity.

Data on oscillators outside the brain in these species are sparse. A few studies have reported that the rhythm in endocuticle growth continues in the cockroach Blaberus craniifer (Blaberus fuscus) after removal of the optic lobes and also in isolated culture of the cockroach leg (Lukat, 1978; Weber, 1985; Weber, 1995). The rhythm is not affected by exposure to light pulses or shifts of the light-dark schedule (Wiedenmann et al., 1986). These data indicate that the circadian rhythm in cuticle formation is not mediated by the optic lobe pacemaker or by other secondary brain pacemakers and suggest that independent peripheral circadian oscillators may be present in the cockroach.

Conclusion

Cockroaches and crickets share great similarities in their circadian systems. Bilaterally located, mutually coupled, optic lobe pacemakers regulate rhythmicity in most behavioral and physiological processes. PDF neurons and PDF play a central role in the cockroach and cricket circadian systems as pacemaker neurons, but also in output, coupling and synchronization mechanisms. Whereas several studies point to the existence of a secondary brain oscillator in the cricket and the cockroach, its importance in the control of rhythmicity in these species remains to be determined.

MOLLUSKS

The snails Bulla gouldiana and Aplysia californica have been used extensively in the study of circadian rhythmicity. Aplysia californica is diurnal, whereas Bulla

gouldiana is active during the night (Kupfermann, 1968; Block and Davenport,

are located close to the optic nerve and are able to produce action potentials (Jacklet et al., 1982; Herman and Strumwasser, 1984; Jacklet and Barnes, 1993; Jacklet et al., 1996). Although it remains to be determined whether individual pacemaker neurons of Aplysia are able to generate circadian rhythms, the available data indicate that they are the location of the Aplysia circadian pacemaker.

Action potentials generated by Bulla and Aplysia pacemaker cells are synchronized with CAPs in a one to one fashion, suggesting that pacemaker cells fire in synchrony with one another. These findings indicate that coupling mechanisms exist between individual Bulla or Aplysia pacemaker neurons (Jacklet et al., 1982; Jacklet and Colquhoun, 1983; Block et al., 1984). Coupling is strong, as hyperpolarization of one basal retinal neuron can result in the inhibition of the entire population (Block et al., 1984). Application of a high-magnesium / low-calcium solution, which blocks chemical synapses, does not block CAPs (Jacklet, 1973; Jacklet et al., 1982; Block et al., 1984), indicating that coupling mechanisms between basal retinal neurons of Bulla as well as between Aplysia secondary or pacemaker neurons are electrotonic. This can also be inferred from the presence of gap junctions between neurons in the

Aplysia and Bulla eye (Luborsky-Moore and Jacklet, 1977; Jacklet and

Colquhoun, 1983) and of dye coupling between Bulla pacemaker neurons (Jacklet, 1988).

Bilateral ocular pacemakers in Aplysia and Bulla

It has been established that the bilateral ocular pacemakers of Bulla are mutually coupled (Roberts and Block, 1983; Roberts and Block, 1985; Block et al., 1986; Page and Nalovic, 1992). Coupling of the Bulla pacemakers is strong in that the eye rhythms remain in phase for long durations of constant darkness. In addition, when a phase shift is induced in one of the eyes, interaction with the other eye results in a decrease of the initial phase difference between the eyes in the course of several days. This is accomplished by phase shifts of both eyes in opposite directions (Roberts and Block, 1983; Roberts and Block, 1985). Severance of the cerebral commissure prevents resynchronization of the eyes, indicating that the cerebral commissure and the optic nerves are part of the coupling pathway (Roberts and Block, 1985; see also Lacroix et al., 1991). The compound action potentials may constitute the coupling signal, as resynchronization between the eyes appears to be mediated by optic nerve activity. Moreover, CAP generation by monocular illumination induces phase shifts in the opposite eye (Block et al., 1986). The BRN neurotransmitter used in this coupling pathway is likely to be glutamate (Michel et al., 2000).

coupling between the Bulla ocular pacemakers is not strong enough to maintain synchrony in all preparations (Page and Nalovic, 1992).

In contrast to Bulla, the electrical impulse activity from the bilateral optic nerves of Aplysia is not well synchronized (Roberts et al., 1987). Housing of these animals in constant darkness eventually results in out-of-phase rhythms in the bilateral eyes of a single animal (Hudson and Lickey, 1980; Lickey et al., 1983). This may explain the weak behavioral rhythms of Aplysia under constant conditions (Kupfermann, 1968; Lickey et al., 1983) and indicates that the bilateral circadian pacemakers in Aplysia are only weakly coupled. Indeed, exposure of one Aplysia eye to a light-dark cycle results in entrainment of only the exposed eye (Lickey et al., 1976) and after a chemically-induced phase separation the two eyes do not resynchronize even after several cycles of interaction (Roberts et al., 1987). Roberts and Block (1983) have suggested that the difference in coupling strength between Bulla and Aplysia is related to these animals being nocturnal and diurnal, respectively. Diurnal animals are exposed to the synchronizing effects of environmental light, and therefore strong endogenous pacemaker coupling would be less important.

Modulation of the ocular pacemaker by extra-ocular processes

As indicated above, the ocular pacemaker is affected by the contralateral ocular pacemaker, especially in Bulla. Besides this influence, the Bulla pacemakers are also affected by other tissues or processes, such as the brain. For example, the period of the impulse activity rhythm of the Bulla eye, measured when the eye is isolated from the brain, is shorter than when the eye is attached (Page and Nalovic, 1992). Moreover, after effects (i.e. longterm changes in the period) are more prominent in isolated eyes than in eyes that are measured in connection to the brain (Page et al., 1997). Modulation of the pacemaker by the brain may involve the action of FMRF-amide, as axons in the optic nerve are immunoreactive for FMRF-amide and terminate near the basal retinal neurons (Jacklet et al., 1987; Roberts and Moore, 1987). Application of FMRF-amide to the eye results in a decrease of CAP activity, induces phase shifts in the CAP rhythm and modulates light-induced phase shifts (Jacklet et al., 1987; Colwell et al., 1992). FMRF-amide affects the ocular circadian system by modulation of a delayed rectifier potassium current and a second conductance, thereby regulating the excitability of the basal retinal neurons (Michel et al., 2002). These data indicate that FMRF-amide is a neurotransmitter of the projection from the brain to the ocular pacemaker in Bulla.

different reentrainment rate has been obtained for eyes that are kept in vitro and

in vivo (Eskin, 1971), as well as a difference between attached and isolated

eyes in the size of phase shifts induced by a transition from LL to DD (Prichard and Lickey, 1981). A serotonergic pathway projects to the eye in Aplysia (Olson and Jacklet, 1985; Takahashi et al., 1989) and studies demonstrating effects of serotonin on e.g. the potassium current of pacemaker neurons, on CAP activity, and on the phase of the CAP rhythm, provide indications for a role of serotonin in mediating the effects of the brain on the pacemaker neurons (Corrent et al., 1978; Colwell, 1990; Barnes and Jacklet, 1997).

Extra-ocular pacemakers in Aplysia and Bulla

For both Aplysia and Bulla, it has been suggested that a circadian oscillator outside the eyes may mediate locomotor behavior. Entrainment of Bulla to different photoperiods reveals that the phase relationship between the behavioral rhythm and the rhythm in ocular action potentials changes with changing daylength (Roberts and Xie, 1996). Similarly, the amplitude of freerunning Aplysia behavioral rhythms and the phase difference between the ocular rhythms does not correlate significantly in all cases (Lickey et al., 1983; Jordan et al., 1985). Moreover, freerunning locomotor rhythms of Bulla and

Aplysia can persist (for several days) under constant darkness after removal of

the eyes or section of the optic nerves (Block and Lickey, 1973; Lickey et al., 1976; Lickey et al., 1977; Block and Davenport, 1982; Lickey et al., 1983). Whereas indications exist for a role of the cerebral ganglia as part of the Aplysia circadian system (Roberts and Block, 1982), the exact location of an extra-ocular oscillator or pacemaker, also in Bulla, its relative importance in the control of overt rhythmicity and the mechanisms by which it affects behavior or physiology remain to be determined.

Conclusion

Circadian pacemakers of both Aplysia and Bulla are located in the eyes close to the optic nerve. Individual basal lateral neurons or pacemaker neurons within one eye are electrically coupled to one another. Bilateral coupling mechanisms are present between the two retinal clocks, but in Aplysia these are weak. In addition, the brain may affect the ocular pacemakers. There are indications for the existence of extra-ocular oscillators or pacemakers, but more research is required to establish their location and their role in the mediation of overt rhythmicity of these two snails.

ZEBRAFISH

Only recently have chronobiologists focused attention on the zebrafish (Danio

rerio) as a model system to study circadian rhythmicity in vertebrates. Most

day (Hurd et al., 1998). These rhythms are regulated by a temperature compensated circadian clock (Hurd et al., 1998), but no central, master pacemaker has been identified in the zebrafish (e.g. Cahill, 2002; Tamai et al., 2003). The pineal and the retina release melatonin in culture, the pineal producing robust melatonin rhythms for at least five cycles in constant darkness (Cahill, 1996). In both the eye and the pineal, the clock genes zfClock, zfBmal1 and zfBmal2 are expressed in a rhythmic manner. Interestingly, many other tissues in the brain and in the periphery show robust gene expression rhythms as well (Whitmore et al., 1998; Cermakian et al., 2000). In culture conditions, peripheral tissues (heart and kidney) remain rhythmic and are directly entrainable by light-dark conditions (Whitmore et al., 2000). Rhythmicity has also been observed in zebrafish cell lines under light-dark conditions. In constant darkness, however, rhythmicity dampens over the course of several cycles (Whitmore et al., 2000; Pando et al., 2001; Carr and Whitmore, 2005). This was shown to be a result of desynchronization between rhythmic single cells, which express a wide range of periods. The cells are light-sensitive and light resets them to a common phase and reduces the range of periods that is expressed. It was suggested that the peripheral zebrafish clocks do not need to be highly precise, as they are strongly affected by daily light (Carr and Whitmore, 2005). Taken together, the available data indicate that many zebrafish tissues contain independent, autonomous and light-sensitive circadian clocks.

MAMMALS

Multiple, coupled, single cell oscillators in the SCN

Figure 6. Period length

Distribution of the period lengths of the rhythms in A) wheel running activity of individual mice, B) Per1-luc bioluminescence measured from cultured SCN explants and C) spontaneous firing rate of single SCN neurons. The average period (± standard deviation) is indicated in the individual panels and is similar for mice, explants and neurons. The period distribution, however, is broader for single SCN neurons than for explants and mice. (From Herzog et al., 2004, Used by permission of Sage Publications, Inc.).

whether they are coupled to one another, as coupling does not necessarily result in identical phases. While this remains to be determined, it is clear that individual cells or small subpopulations do not reflect the circadian discharge or gene expression pattern of the SCN as a whole. Instead, it appears that SCN electrical activity or gene expression rhythms are a composite tissue property.

Multiple mechanisms have been proposed for coupling between SCN neurons (for review, see Michel and Colwell, 2001), one of which is GABA (Liu and Reppert, 2000; Shirakawa et al., 2000; Albus et al., 2005). Application of the GABAA antagonist bicuculline blocks transmission of information between the

dorsal and the ventral SCN. Pulsed application of bicuculline revealed that endogenous GABA inhibits neurons in the ventral SCN, but excites a substantial number of dorsal SCN neurons, suggesting that coupling between the dorsal and the ventral SCN is asymmetrical (Albus et al., 2005). The rhythm in GABAergic activity is dependent on VIP and both VIP and the VPAC2 receptor are required for synchronization of a subset of SCN neurons (Itri et al., 2004; Aton et al., 2005).

Synchronization between SCN neurons also involves Na+_{-dependent action}

potentials, as it is strongly affected by application of tetrodotoxin (Honma et al., 2000; Yamaguchi et al., 2003). In addition, coupling within the SCN may occur via gap junctions. This is suggested, among others, by the observation that dye coupling in the SCN can be blocked by the gap-junction blocker halothane (Colwell, 2000) and by the absence of electrical coupling in the SCN of animals that lack the gap junction protein connexin 36 (Long et al., 2005). Gap junctions may be important for short-distance coupling and less so for coupling between regions (Colwell, 2000; de la Iglesia et al., 2004a; Albus et al., 2005; Long et al., 2005).

Behavioral arrhythmicity induced by desynchronized single cells

Longterm exposure to constant light may lead to behavioral arrhythmicity (Pittendrigh and Daan, 1976; Daan and Pittendrigh, 1976a). In the SCN of LL-arrhythmic animals, rhythms in Per1-promotor driven GFP fluorescence and PER2 protein expression are severely disrupted (Figure 7) (Beaulé et al., 2003; Amir et al., 2004; Ohta et al., 2005). Individual SCN neurons, on the other hand, are robustly rhythmic in Per1:GFP, but the rhythms show large differences in phase (Ohta et al., 2005). These data suggest that the molecular clock in individual cells functions normally in these animals and that the LL-induced behavioral arrhythmicity is caused by desynchronization between rhythmic pacemaker cells.

Coding for daylength: single cell or population response?

Figure 7. Individual SCN cells remain rhythmic after longterm exposure to constant light

A) Double plotted actogram of wheel running activity of a mouse under constant light. Black marks indicate the presence of wheel running activity. The animal gradually loses rhythmicity after longterm exposure to constant light. B) The Per1:GFP fluorescence signal from the SCN of an LL-arrhythmic mouse shows disrupted rhythmicity with a very low amplitude. C) The

Per1:GFP fluorescence signals of individual SCN cells from the SCN of Figure 7B show strong

rhythms with high amplitudes, but a wide phase dispersal. (From Ohta et al., 2005, Used by permission of Macmillan Publishers Ltd).

Per2 mRNA and PER1 protein is longer under long than under short daylengths

(Messager et al., 1999; Steinlechner et al., 2002; Sumová et al., 2002; Carr et al., 2003; Johnston et al., 2003; Sumová et al., 2003; Tournier et al., 2003; de la Iglesia et al., 2004b). The expression patterns of the clock genes Per3, Cry1,

Cry2, Clock and Bmal1 show changes in phase, duration or amplitude (Carr et

al., 2003; Sumová et al., 2003; Tournier et al., 2003).

It should be noted that all experiments performed so far describe the response of a population of SCN cells and that at present, it is still unclear whether the observed photoperiodic changes in the SCN are the consequence of changes in

Figure 8. Coding for daylength

Altering phase differences between individual neurons can explain photoperiod-related changes in the multiunit activity pattern. A-C) Simulation of the multiunit activity pattern under short (L:D = 8:16 [A]), normal (L:D = 12:12 [B]) and long (L:D = 16:8 [C]) daylength. The bars above the graphs indicate the light-dark schedules, with white representing daytime and black representing nighttime. In each graph, the distribution of 9 individual firing patterns is given, as well as the resultant multiunit activity pattern. A) When the individual firing patterns are close together, which simulates a short daylength, the resultant multiunit activity peak is narrow and has a high amplitude. B-C) When the distribution of the individual firing patterns is broader, the multiunit activity peak becomes wider and the amplitude decreases. D-F) Measured multiunit activity patterns of rats exposed to short (D), normal (E) and long (F) daylength. The actograms of the individual animals are given above each graph. In correspondence with the simulations, the multiunit activity peak in the SCN is narrow after exposure to a short daylength and the amplitude is high. Under a long daylength, the width of the multiunit activity peak is increased and it has a lower amplitude. (From Schaap et al., 2003, Used by permission of National Academy of Sciences, USA).

the phase relationship between multiple single cell oscillators, or whether these effects occur within single cells. One study that addressed the question how single cell SCN oscillators contribute to the changes in population waveform revealed that changes of the multiunit activity pattern can be explained by a changing phase relationship between single unit activity peaks (Schaap et al., 2003). A close phase relationship between oscillatory neurons corresponds with a short daylength waveform, whereas a broader phase distribution corresponds with a long daylength shape of the multiunit activity pattern (Figure 8). Whether or not single unit activity patterns change under different photoperiods remains to be determined by actual single unit activity measurements. A recent study demonstrated that clock- and clock-controlled gene expression patterns show large phase changes in the caudal SCN, whereas the rostral SCN shows relatively small responses (Hazlerigg et al., 2005). Together, the data support the idea that photoperiodic encoding is a consequence of the structural heterogeneity within the SCN (Schaap et al., 2003; Hazlerigg et al., 2005).

Regional differences within the SCN

The SCN is not a homogeneous tissue and regional differences have been observed within the SCN of many mammalian species. An important distinction has been made between the dorsomedial (shell) and the ventrolateral (core) parts of the SCN (van den Pol, 1980). In general, the core contains cells that are immunoreactive for vasoactive intestinal peptide (VIP) or gastrin releasing peptide (GRP). Cells in the shell typically contain vasopressin (VP) or somatostatin (Figure 9) (Card and Moore, 1984; Aïoun et al., 1998; Abrahamson and Moore, 2001; Moore et al., 2002). A specialization of the core and the shell exists also for the organization of afferent and efferent projections. The hypothalamus and limbic forebrain mainly project to the shell, whereas projections from the intergeniculate leaflet, raphe nuclei and the retina terminate predominantly in the core (and central parts) of the SCN (Moore and Lenn, 1972; Card and Moore, 1984; Moga and Moore, 1997; Leak et al., 1999; Abrahamson and Moore, 2001). A special afferent pathway, the retinohypothalamic tract, will be discussed separately (see Entrainment by

light). Efferent projections to the lateral sub-paraventricular zone (sPVZ)

originate from the SCN core, whereas the SCN shell projects to the medial sPVZ and the DMH (Leak et al., 1999). The direct projections from the SCN to target brain areas are believed to be first steps of more complex pathways that eventually regulate rhythmicity in the brain and the periphery (Aston-Jones et al., 2001; Bartness et al., 2001; Kalsbeek and Buijs, 2002; Buijs et al., 2003; Deurveilher and Semba, 2005).

inconsistencies exist in the available data, it is generally believed that the shell shows robust endogenous expression rhythms, whereas rhythms in the core are damped or absent (Takeuchi et al., 1992; Cagampang et al., 1994; Nishiwaki et al., 1995; Guido et al., 1999; Ibata et al., 1999; Schwartz et al., 2000; Yan and Okamura, 2002). Several single unit electrical activity recordings have confirmed this distinction (Shibata et al., 1984a; Derambure and Boulant, 1994; Jiao et al., 1999; Nakamura et al., 2001; Saeb-Parsy and Dyball, 2003), but multiunit electrical activity in both parts of the rat-SCN slice, as well as VIP

and vasopressin release, show robust rhythmicity (Shinohara et al., 1994, 1995;

Honma et al., 1998b; Schaap et al., 2003; Albus et al., 2005). The cause of these differences remains to be determined.

Figure 9. Regional differences within the SCN

Photographs showing the distribution of VIPergic (left) and VPergic (right) neurons in the rat-SCN. VIP positive neurons are mainly located in the ventrolateral SCN (vlSCN), whereas the dorsomedial SCN (dmSCN) contains many VP positive neurons. OC = optic chiasm, III = third ventricle, bar = 50 μm. (From van Esseveldt et al., 2000, Used by permission of Elsevier).

In the mouse, SCN tissue organization is somewhat different than in the rat. Several studies show spontaneously rhythmic Per1 bioluminescence and PER1 and PER2 protein expression rhythms in both SCN regions (Yamaguchi et al., 2003; Yan and Silver, 2004), but others show that a region in the central mouse-SCN has distinct gene expression patterns (King et al., 2003; LeSauter et al., 2003). In the hamster, a subgroup of tightly packed Calbindin-D28K (CalB)

circadian rhythms in the hamster (LeSauter and Silver, 1999; Hamada et al., 2001; Jobst and Allen, 2002; Kriegsfeld et al., 2004).

Entrainment by light

In mammals, entraining light information reaches the SCN directly via the retinohypothalamic tract (RHT), which contains several neurotransmitters, the most important ones being glutamate and PACAP (Hannibal, 2002; Brzezinski et al., 2005). The photopigment melanopsin plays a role in photic signaling to the SCN and is expressed in a subset of intrinsically light-responsive retinal ganglion cells that innervate the SCN (Gooley et al., 2001; Berson et al., 2002; Hannibal et al., 2002; Hattar et al., 2002). Rods, cones and cryptochromes contribute to photic signaling towards the SCN (Mrosovsky, 2003; Partch and Sancar, 2005) and their relative contribution to photic entrainment is currently under investigation.

The RHT terminates predominantly in the core of the rat-, hamster- and mouse-SCN and more sparsely in the shell (Moore and Lenn, 1972; Card and Moore, 1984; Abrahamson and Moore, 2001). The retinorecipient area shows light-induced changes in the expression of several clock genes and immediate early genes, as well as electrophysiological responses to light or optic nerve stimulation (Groos and Mason, 1980; Shibata et al., 1984b, 1984c; Meijer et al., 1986; Meijer et al., 1992; Guido et al., 1999; Yan et al., 1999; Schwartz et al., 2000; Dardente et al., 2002; Kuhlman et al., 2003; Karatsoreos et al., 2004). In the shell, light-induced changes in gene expression are generally absent, less pronounced or delayed. In the hamster-SCN, light responses of clock gene expression and Fos immunoreactivity are particularly clear in the CalB area (Silver et al., 1996a; Hamada et al., 2001).

Entrainment to the environmental light-dark cycle relies on the phase shifting effects of light. Light-induced phase shifts of e.g. behavior are not complete within one cycle but develop over several transient cycles. To test whether the transients reflect the phase shifting kinetics of the pacemaker itself, double light pulses have been applied. The resultant phase shift of two light pulses can be used to calculate the magnitude of the phase shift, induced by the first light pulse, at the time the second was administered. These studies revealed that the phase response curve shifts immediately in response to a light pulse, indicating instantaneous phase resetting of the underlying oscillator (Pittendrigh, 1981; Elliott and Pittendrigh, 1996; Sharma and Chandrashekaran, 1997; Best et al., 1999; Sharma et al., 2000; Watanabe et al., 2001; see also Takamure et al., 1991).

It should be noted that an immediate phase shift of the light-pulse phase response curve may not represent the behavior of the SCN as a whole. Several studies have now directly addressed the question whether or not transient phase resetting may be the result of slowly responding SCN components. Reddy and coworkers reported immediate resetting of the rhythms in Per1 and

Per2 expression in the SCN, but slow phase shifting of the Cry1 rhythm, in

studies observed differences between the ventral and dorsal SCN regions. The rhythm in Per1 expression was shown to phase advance and delay immediately in the ventral SCN, but slowly in the dorsal SCN, resulting in a temporal desynchronization between the SCN regions (Nagano et al., 2003). A more detailed spatial analysis within the SCN revealed transient desynchrony also within ventral and dorsal SCN regions during phase resetting (Nakamura et al., 2005). Albus and coworkers (2005) observed bimodal electrical activity patterns in the ventral and dorsal SCN in a slice preparation after a 6-h phase delay of the light-dark cycle (Figure 10). One component showed a large and immediate phase delay, whereas the other showed almost no phase shift. In the course of several days, the two components merged, resulting in a unimodal, 6-h phase delayed electrical activity peak. Separation of the dorsal and ventral parts by a knife cut resulted in unimodal peaks in both regions. In the ventral SCN, the

Figure 10. Transmission of phase information between the ventral and the dorsal SCN

peak was completely delayed, whereas in the dorsal SCN, the peak was not shifted (Figure 10). The molecular and electrophysiological studies demonstrate that the ventral SCN phase shifts quickly and the dorsal SCN phase shifts slowly (Nagano et al., 2003; Albus et al., 2005; Nakamura et al., 2005). The presence of the bimodal peaks in electrical activity in intact slices indicates that phase shift information is transmitted from the ventral to the dorsal SCN and vice versa. In contrast, molecular expression seems to remain region specific.

Onset and offset of mammalian behavioral activity show different phase resetting characteristics in response to photic and non-photic stimuli and in response to a shift of the light-dark cycle (Sisk and Stephan, 1981; Honma et al., 1985; Elliott and Tamarkin, 1994; Meijer and De Vries, 1995; Vansteensel et al., 2003a). Also the evening rise and morning decline of the rhythms in hamster melatonin and rat pineal N-acetyltransferase (NAT) activity respond with different rates to phase shifting light pulses (Elliott and Tamarkin, 1994; Illnerová and Sumová, 1997). The differences in phase resetting of the onset and offset of behavioral activity and melatonin regulation may relate to differentially responding subregions within the SCN. In horizontal slice preparations of the hamster-SCN (but not the rat- or mouse-SCN, Burgoon et al., 2004), two neuronal activity components occur around projected dawn and dusk (Jagota et al., 2000; Burgoon et al., 2004) and shift differentially in response to glutamate (Jagota et al., 2000). It is possible that these two components observed in the SCN neuronal activity of the hamster correspond to the ventral and dorsal oscillators observed in the rat-SCN (Albus et al., 2005) and that, in the hamster, they desynchronize in a specific (horizontal) preparation of the tissue.

Interaction between the left and right SCN

Splitting

Splitting may occur in animals that are kept in constant light for prolonged periods of time. During splitting, the behavioral activity splits into two components that freerun temporally with different periods until they reach a stable phase relationship. Pittendrigh and Daan (1976) proposed that the two behavioral components originate from two functionally different oscillators, one controlling the behavioral activity onset (E) and the other controlling the offset (M). The phase relationship between the two oscillators was supposed to determine the length of the activity time (α) and resting time (ρ). According to the model, the oscillators have distinct freerunning periods that are differentially responsive to light. Normal rhythmicity and splitting were suggested to reflect two metastable coupling modes between the underlying oscillators.

Figure 11. Splitting and out-of-phase SCN nuclei

A) Actogram of the wheel running activity of a hamster under constant light conditions. Days are plotted underneath each other and black marks indicate the presence of wheel running activity. After a number of days in constant light, the behavioral activity splits into two components, which subsequently freerun ~12 hours out-of-phase (arrow). OVX = ovariectomy, E2 = implantation of estradiol benzoate capsule, asterisk = sacrifice. B) Coronal brain section of a behaviorally split hamster, processed for c-Fos immunoreactivity and showing asymmetry between the left and right SCN nuclei. Only one SCN nucleus is intensely stained for c-Fos. 3V = third ventricle, OC = optic chiasm, scale bar = 500 μm. (From de la Iglesia et al., 2003, Used by permission of Society for Neuroscience).

11) (de la Iglesia et al., 2000; de la Iglesia et al., 2003; Mendoza et al., 2004; Ohta et al., 2005). Electrical activity recordings in the SCN of split hamsters revealed bimodal patterns in the left and right nuclei (Zlomanczuk et al., 1991), indicating that the antiphase molecular expression rhythms of the left and right SCN may be integrated at the level of the electrical activity. The findings do not agree with the original proposition that the two behavioral activity components arise from functionally different E- and M-oscillators. The E- and M-oscillator model may remain useful, however, to guide experiments on photoperiodicity.

Effects of extra-SCN tissues on the SCN pacemaker

SCN neuronal activity measured in vitro shows smooth rhythms in electrical impulse frequency. In contrast, in vivo recordings in freely moving animals show

Figure 12. SCN neuronal activity, slow wave activity and vigilance state

a large variability in electrical activity that is superimposed on the circadian waveform (Meijer et al., 1997). The difference indicates that the SCN electrical activity is affected by the presence or absence of connectivity with the central nervous system. Indeed, it has been found that behavioral activity induces a decrease in the multiunit activity in the SCN of the hamster and the rat (Meijer et al., 1997; Yamazaki et al., 1998; Schaap and Meijer, 2001). Also vigilance state affects SCN neuronal activity: During REM sleep, SCN electrical activity is increased, whereas it is decreased during non-REM sleep (Figure 12) (Deboer et al., 2003).

Behavioral stimuli, such as novel wheel-induced running, and sleep deprivation can induce phase shifts in behavioral activity rhythms (Mrosovsky et al., 1989; Antle and Mistlberger, 2000), which possibly involves identified projections from several brain areas to the SCN (Pickard, 1982; Moga and Moore, 1997; Abrahamson and Moore, 2001). The neurotransmitters of these projections induce phase advances when applied during the subjective day, for example neuropeptide Y (NPY) and serotonin (5-HT) agonists (e.g. Huhman and Albers, 1994; Horikawa and Shibata, 2004), and result in changes in the expression levels of Per1 in the SCN (Horikawa et al., 2000; Yokota et al., 2000; Fukuhara et al., 2001; Maywood et al., 2002). While a causal relationship between Per1 suppression and non-photic phase shifting has been indicated (Hamada et al., 2004), some stimuli induce phase shifts without concurrent changes in Per1 expression levels (Poirel et al., 2003; Vansteensel et al., 2005), suggesting that they affect the clock via intracellular pathways that do not involve Per1.

A role for extra-SCN tissues or processes in reentrainment

unknown. However, reentrainment to a shifted light-dark schedule is accelerated if animals and also humans exercise at the new phase (Mrosovsky and Salmon, 1987; Barger et al., 2004). Moreover, in pinealectomized animals (Quay, 1970; Finkelstein et al., 1978), in animals that have low levels of melatonin (Kopp et al., 1998), and in animals that are deficient in neuronal PAS domain protein 2 (NPAS2) (Dudley et al., 2003), resetting occurs more rapidly than in control animals.

Figure 13. Dissociation between Per1 bioluminescence and SCN neuronal activity in vitro and

in vivo after a phase advance of the light-dark cycle

ZT 6 of the unshifted or shifted light-dark cycles is indicated by vertical lines. The light-dark cycles to which the animals were exposed before recording are indicated above the graphs, with white indicating lights-on and black indicating lights-off. A) Per1 bioluminescence (counts / s, indicated every minute) rhythms on the day before a 6-hour phase advance of the light-dark cycle (cont.) and on days 1, 3 and 6 in constant darkness after the shift. The Per1 bioluminescence rhythm is immediately and completely phase shited on the first day after the advance and remains shifted on days 3 and 6 after the shift. B) SCN neuronal activity (in Hz, plotted every 10 sec), recorded from a brain slice in vitro, phase shifts for several hours on the first and third day after the shift of the light-dark cycle, but returns to the unshifted phase on day 6 after the shift. C) In vivo SCN neuronal activity (in Hz, plotted every 10 sec) does not phase shift at all on days 1, 3 and 6 after the shift of the light-dark cycle. Episodes of multiunit activity containing movement artefacts were not plotted, resulting in the gaps in the data. (From Vansteensel et al., 2003b, Used by permission of Elsevier).

Oscillators in the periphery

when cultured in isolation of the SCN (Tosini and Menaker, 1996; for review, see Tosini and Fukuhara, 2002). Moreover, the expression of Per1, Per2 and

Bmal1 is rhythmic in peripheral tissues, such as in the liver and the heart, and in

several brain regions (Balsalobre et al., 1998; Yamazaki et al., 2000; Akhtar et al., 2002; Abe et al., 2002; Izumo et al., 2003; Mühlbauer et al., 2004; Fukuhara et al., 2005; Prolo et al., 2005). Peripheral rhythms were often shown to damp (i.e. express a decline in rhythm amplitude) within several cycles in absence of the SCN. Measurement of bioluminescence from individual rat-1 fibroblasts transfected with mBmal1::luc, however, or from individual primary fibroblasts from mPER2::LUC-SV40 mice, reveals robust circadian rhythms. Similarly, robust rhythms in Rev-erbα-VNP fluorescence have been obtained in individual NIH3T3 fibroblasts (Nagoshi et al., 2004; Welsh et al., 2004). These data suggest that damping of circadian oscillations at the population level may result from desynchronization between rhythmic cells.

Figure 14. Persistent rhythmicity in isolated peripheral tissues of the mouse

MPER2::LUC bioluminescence recordings from SCN, liver and lung tissues show sustained circadian rhythmicity for more than 20 days in culture. (From Yoo et al., 2004, Used by permission of National Academy of Sciences, USA).

can be observed in the Per1 gene expression of the olfactory bulb of animals turned arrhythmic by constant light or by an SCN lesion (Granados-Fuentes et al., 2004a) and in the electrical activity of individual cultured olfactory bulb cells (Granados-Fuentes et al., 2004b). Notably, in LL-arrhythmic and SCN lesioned animals, the phases of the rhythms of the peripheral tissues are desynchronized to a large extent, both within and between animals (Yoo et al., 2004; Granados-Fuentes et al., 2004a). It is concluded that the SCN is required to synchronize self-sustained oscillators in the periphery.

Food-entrainable oscillator

Restriction of food availability to a relatively short interval during every cycle induces food-anticipatory behavior, i.e. strong increments of behavioral activity shortly before the food becomes available (Mistlberger, 1994). When animals are completely food deprived after a period of restricted feeding, food-anticipatory behavior persists for at least several cycles at the predicted phase (Mistlberger, 1994). This phenomenon also occurs in SCN-lesioned animals (Stephan et al., 1979; Boulos et al., 1980; Mistlberger, 1994; Marchant and Mistlberger, 1997), which indicates that it is mediated by an oscillator independent of the SCN. At present, the location of the food-entrainable oscillator is still unknown (for review, see Mistlberger, 1994; Stephan, 2001; Davidson et al., 2005) and it is not impossible that food anticipation results from the interplay of multiple structures or tissues that are distributed over the body. In response to a phase shift of the light-dark cycle, food anticipatory behavior shifts to the new phase (Rosenwasser et al., 1984; Stephan, 1986a; Ottenweller et al., 1990), which indicates that the food-entrainable oscillator is affected by or coupled to the light-entrainable oscillator of the SCN.

LL-arrhythmic animals (Lamont et al., 2005). Differences between the two groups of studies may arise from a difference in the number of cycles that animals are kept in the restricted feeding schedule (Boulos et al., 1980; Castillo et al., 2004), species differences (Mistlberger, 1994), differences in the endogenous period τ (Stephan, 1986a, 1986b; Abe and Rusak, 1992; Mistlberger, 1993; but see Castillo et al., 2004) and the presence or absence of caloric restriction (Challet et al., 2003; Mendoza et al., 2005). Identification of the location or pathway employed by the food-entrainable oscillator will be helpful to determine the reciprocal coupling mechanisms between the light- and food-entrainable oscillators.

Conclusion

The mammalian pacemaker of the SCN contains multiple, single cell oscillators that are synchronized. Each SCN nucleus contains two anatomically and functionally different regions, the ventrolateral SCN or core and the dorsomedial SCN or shell. These regions are coupled and show different properties during reentrainment to a shifted environmental light-dark cycle. A number of attributes observed in mammalian behavior and physiology have now been explained at the tissue level of the SCN. Arrhythmicity in behavior, induced by constant light, is a result of desynchronized rhythmic single cell SCN oscillators (Ohta et al., 2005) and coupling between these single cells is thought to render increased precision of rhythmicity (Liu et al., 1997; Herzog et al., 1998; Honma et al., 1998a; Herzog et al., 2004; Honma et al., 2004). Antiphase oscillations of the left and right SCN nuclei explain splitting of behavioral rhythms (de la Iglesia et al., 2000; de la Iglesia et al., 2003; Mendoza et al., 2004; Ohta et al., 2005) and also daylength may be encoded by changing phase relationships between individual SCN cells or SCN regions (Schaap et al., 2003; Hazlerigg et al., 2005). These findings emphasize that, for proper functioning of the circadian timing system, synchronization and coupling mechanisms within the SCN are indispensable. Interactions also exist between the SCN and the periphery. Tissues and processes outside the SCN have direct effects on the SCN neuronal activity and phase shift the pacemaker. Independent oscillators are present in the periphery of the SCN that may have some degree of autonomy, but which seem to require signals from the SCN to remain synchronized.

COMPARISON OF SPECIES

Synchronization between pacemaker cells

1982; mammals: Michel and Colwell, 2001). Within the mammalian SCN, small subpopulations (< 10 neurons) appear to be coupled and oscillate in phase. These subpopulations can show large phase differences with adjacent populations (Schaap et al., 2003). Also recording of clock gene expression rhythms revealed the presence of groups of SCN neurons that are in phase, but show large phase differences with other populations (Quintero et al., 2003; Yamaguchi et al., 2003; Nakamura et al., 2005). Similarly, neurons of the cockroach accessory medulla are organized in assemblies that are synchronized and in phase. GABAergic synaptic interactions are involved in synchronization within the assemblies, whereas synchronization between different assemblies is thought to be mediated by PDF (Schneider and Stengl, 2005). PDF also plays a synchronizing role in the Drosophila circadian system (Peng et al., 2003; Lin et al., 2004). Synchronization in the mammalian SCN is mediated by, among others, VIP and GAP junctions (Long et al., 2005; Aton et al., 2005), whereas coupling in the snail eye occurs through electrotonic mechanisms (Jacklet, 1973; Jacklet et al., 1982; Block et al., 1984).

The importance of coupling for overt rhythmicity is indicated by the arrhythmic behavior of Drosophila carrying a pdf-mutation and mammals that are exposed to constant light for extended periods of time (Peng et al., 2003; Lin et al., 2004; Ohta et al., 2005). In addition, loss of VIP or its VPAC2 receptor, or loss of the gap junction protein connexin 36 leads to disrupted behavioral activity rhythms of mammals (Long et al., 2005; Aton et al., 2005). These data underscore that, besides rhythm generation, neuronal synchronization is required to produce coherent rhythmic output signals.

Regional oscillators

The two main subregions of the SCN, the core and shell, are dissimilar in their neuropeptide contents and in the organization of efferent and afferent projections. The photic input projects mainly to the core SCN, which may relate to the finding that light-induced phase resetting occurs faster in the ventral than in the dorsal SCN (Nagano et al., 2003; Albus et al., 2005). GABA transmits phase information between the ventral and dorsal SCN and exerts a different effect in the two regions (Albus et al., 2005). At present, it remains unclear whether the existence of distinct subregions within the pacemaker serves a certain function and why these regions have different phase resetting kinetics. It is possible that such an organization is more resistant to short term environmental stimuli, but is optimized to follow slowly developing changes, which may be the ones that are biologically important.