Report

Synchronization of Cellular Clocks in the Suprachiasmatic Nucleus

See allHide authors and affiliations

Science  21 Nov 2003:
Vol. 302, Issue 5649, pp. 1408-1412
DOI: 10.1126/science.1089287

Abstract

Individual cellular clocks in the suprachiasmatic nucleus (SCN), the circadian center, are integrated into a stable and robust pacemaker with a period length of about 24 hours. We used real-time analysis of gene expression to show synchronized rhythms of clock gene transcription across hundreds of neurons within the mammalian SCN in organotypic slice culture. Differentially phased neuronal clocks are topographically arranged across the SCN. A protein synthesis inhibitor set all cell clocks to the same initial phase and, after withdrawal, intrinsic interactions among cell clocks reestablished the stable program of gene expression across the assemblage. Na+-dependent action potentials contributed to establishing cellular synchrony and maintaining spontaneous oscillation across the SCN.

In mammals, most physiological and behavioral events are subjected to well-controlled daily oscillations. These rhythms are generated by an internal self-sustained oscillator located in the hypothalamic suprachiasmatic nucleus (SCN) (1). Electrophysiological studies have demonstrated that the circadian oscillation in the SCN is generated at the level of individual neurons (2). The current model of this oscillator is based on autoregulatory transcription and translation feedback loops of clock genes in which Period (Per) genes occupy a central position (3). The SCN is a heterogenous cellular assemblage of autonomous cellular oscillators that act in concert (4); their measurable outputs, principally neuropeptide secretion (5) and electrical firing rhythms (6), are synchronized. A major unresolved question, therefore, is how the autonomous cellular clocks, and the self-sustained molecular loops upon which they are based, interact across the heterogenous neuronal network of the SCN.

To characterize the nature of synchronization of the molecular clockwork of individual neurons, we monitored continuously, in real time, the dynamics of the core circadian loop in hundreds of individual neurons within an intact SCN neuronal assemblage (organotypic SCN slices held in long-term culture) with a bioluminescence method and a highly sensitive cryogenic cooled charge-coupled device (CCD) camera. CCD imaging of SCN slice cultures from neonatal transgenic mice carrying the mPer1-promoter–driven luciferase reporter gene (mPer1-luc) (7) revealed hundreds of light spots throughout the SCN, indicative of local cellular mPer1 transcription (Fig. 1A). This luminescence occurred in neurons but not in glial cells (fig. S1A). The functional half-life of the reporter luciferase in individual cells was ∼1.2 hours (fig. S1B), confirming adequate temporal resolution to monitor phenomena with 24-hour cycles. Individual cells showed robust oscillations, with variable peak and trough phases (leading to large phase differences, even among nearby cells) that were repeated in every cycle monitored (Fig. 1A). Nearly all luminescent cells (99.2%, or 1168 out of 1177), regardless of their intensity, showed a pronounced circadian rhythm. This proportion was higher than that determined with mPer1-GFP (green fluorescent protein) transgenic mice (8), although phase differences among the SCN neurons were detected by both techniques. This difference may arise from the greater sensitivity of the luciferase technique, the difference in the size of the mPer1 promoter used (7.2 kb versus 3 kb), and/or the use of long-term organotypic cultures rather than acute slices. The overall rhythm of randomly chosen cells from a single SCN slice culture was very stable, with a periodicity of ∼24 hours (24.51 ± 0.03, mean ± SEM, n = 100) (fig. S2). Referencing the daily profiles of optical fiber recording from the SCN of free-moving mPer1-luc animals (9), the peak and trough times in vitro corresponded to subjective day and night times, respectively. In this context, mPer1 transcription in each SCN cell peaks at subjective day with a wide variety of phases and reaches a trough at late subjective night to dawn.

Fig. 1.

Robust and synchronous oscillation of mPer1-luc bioluminescence in the SCN slice culture. (A) Circadian fluctuation of luminescence from individual SCN neurons. The luminescence signals from four neighboring neurons are indicated by arrowheads in the upper panels. The signal intensity of bioluminescence from each cell, integrated for 1 h, was quantified and plotted in the graph below the panels. The colors of the lines correspond to those of the cells indicated in the upper panels. The trough time of the total SCN bioluminescence was set to time = 0. The bioluminescent image of the whole slice culture containing a paired SCN is shown at the lower left. ov, original third ventricle; ooc, original optic chiasma. Scale bar, 10 μm in the upper panels and 100 μm in the left panel. (B and C) Temporal changes in bioluminescence signals from 100 SCN cells randomly chosen in wild-type (B) and mPer1-luc–bearing mCry1/mCry2 double-knockout mice (C). Time 0 was set at the trough time of the total SCN bioluminescence in the wild-type mouse and 11 hours after the start of monitoring in the mCry double-knockout mouse.

Mice show complete behavioral arrhythmicity when the cryptochrome genes mCry1 and mCry2, indispensable components of core clock loop, are impaired (10, 11). SCN slices from mPer1-luc transgenic mice lacking CRY proteins (mPer1-luc-mCry1–/– mCry2–/–) showed constant cellular luminescence rather than the overt synchronous rhythm detected in wild-type animals (Fig. 1C and fig. S3), demonstrating that the arrythmicity observed in the animal and across the SCN network after genetic disruption of core clock loop was not due to desynchrony across an otherwise rhythmic population, but rather arose from cessation of the individual cellular oscillators.

In organotypic culture, the SCN slice retained its nuclear morphology as in vivo (fig. S4A) (12, 13). Luminescence signals first appeared in the dorsomedial periventricular part of the SCN near the third ventricle, then spread out to the central part of the SCN after 4 to 8 hours, and eventually spread to the ventral portion after 12 to 15 hours (Fig. 2A; movie S1). Hourly profiles of 1228 cells revealed that 49.3% exhibited peak luminescence between 11 and 13 circadian hours after the trough time of the overall SCN bioluminescence (1 circadian hour = 1/24th of the individual circadian period) (Fig. 2B). When phase differences across the three SCN zones (dorsal, middle, and ventral) were examined, most dorsal SCN cells showed earlier peak times (61.1% of cells had peaked by the eighth circadian hour after the trough time of total SCN bioluminescence), but hardly any of the middle and ventral cells showed earlier peak times (13.7% and 7.2%, respectively). The topography-dependent sequential expression was clearer when we plotted the sequential peak phases every 2 hours in a slice (fig. S4B). Because a comparable dorsomedial-to-ventral spread of mPer1 mRNA signals was also observed in adult mouse SCN in vivo (fig. S5), it is likely that the topographically specific and orchestrated multiphase oscillations of mPer1 transcription rhythms are inherent properties of the SCN.

Fig. 2.

Spatially organized synchronicity of cell rhythm in the SCN and the effect of surgical disconnection. (A) Hourly images of dorsomedial-to-ventral spread of bioluminescence within the SCN slice. The trough time of the total SCN bioluminescence was set to time = 0. Scale bar, 100 μm. (B) The temporal changes in total SCN bioluminescence [the peak value was adjusted to 100 and expressed as means ± SEM (n = 4)] and the range of peak times of individual cells in the whole SCN (n = 1228), and the subdivided dorsal (n = 246), middle (n = 654) and ventral (n = 328) SCN portions. The x axis represents circadian hours (1 hour = peak interval/24 hours) from the trough time of the total SCN bioluminescence. (C) An SCN slice cut horizontally at the upper one-third. Bioluminescence of each subdivision was recorded in a different chamber. The cells in the dorsal subdivision (n = 25) lost their coordinated rhythm, although those in the ventral subdivision (n = 89) retained it. Scale bars, 100 μm.

To address whether the dorsomedial cells can drive oscillation within the remaining ventrolateral cells, we performed a surgical knife-cut across the SCN slice in the horizontal plane at the upper one-third of the SCN (Fig. 2C). Each subdivided slice was cultured in different incubation chambers, and luminescence was measured. Although each cell showed a clear rhythm, cell synchrony was completely distorted in the dorsal slice subdivision, which included the original phaseleading cells. The cells in the ventral slice cut piece, however, continued to cycle with clear synchronicity, similar to that observed before the cut. This demonstrates that the dorsomedial population of “leading” cells cannot synchronize its cellular rhythms when disconnected from the rest of the SCN and is not a necessary driver to the ventrolateral cells. Despite the completely disrupted synchronicity, the amplitude of rhythmic luminescence in each cell in the upper slice subdivision did not dampen even after several cycles. Desynchronization does not, therefore, lead inevitably to reduced amplitude within individual cellular oscillators.

To determine whether the synchronization mechanism across the SCN is intrinsic to the circuitry of the assemblage or is a stochastic and unpredictable emergent property of the circuit, cell clocks were stopped and then restarted. Treatment of SCN slices with cycloheximide (CHX), a protein synthesis inhibitor, starting 3 hours after peak luminescence, did not move the subsequent phase of the luminescence until it had been applied for 12 hours. However, once applied for more than 18 hours, it delayed the luminescence time-dependently (Fig. 3, A and B). Therefore, only long-term CHX treatment stopped the molecular clocks, and its elimination allowed the clocks to restart. After 36 hours of CHX treatment, we could not detect PER1 and PER2 immunoreactivities in the SCN, but CLOCK and BMAL1 staining were high in the nuclei of SCN cells (fig. S6A). This result suggests that the half-life of the positive factor proteins, CLOCK and BMAL1, are long, whereas those of the PER1 and PER2 proteins are short and, thus, are absent 36 hours after the introduction of CHX. It may be that when the CHX is eliminated, the remaining CLOCK and BMAL1 proteins start the molecular loop by initiating de novo transcription of mPer1 and mPer2 to initiate the negative feedback phase of the cycle.

Fig. 3.

Resetting and reorganizing the synchrony of cell rhythm with cycloheximide (CHX). (A) The total SCN bioluminescence measured with the photon-counting camera after the addition of 10 μg/ml CHX to the culture medium for 6 hours and 36 hours, starting 3 hours after the peak time. (B) Analysis of peak phases after inhibitor treatments of various durations, which was reproduced in two subsequent assays. The peak time at each cycle is shown as a dot. (C and D) Recording of bioluminescence of representative individual cells (C) (see inset photo for localization; scale bar, 100 μm) and 40 SCN cells (D) before and after the application of CHX (36 hours)., Nearly all cells reached a peak 3 hours after the drug had been washed out. (E) Correlation of the peak times of individual cells (n = 177) before CHX application (38 to 56 hours) with those after CHX washout (182 to 194 hours).

Individual cells exhibited a variety of peak-trough rhythms with large phase differences before application of CHX (Fig. 3C). After 9 hours of the application, the luminescence levels were sharply reduced to the previous trough value, and after 24 hours the signal was undetectable against background. After the drug had been washed out, the first peak appeared after 3 hours, simultaneously in all previously asynchronous cells. Restored cellular rhythms evolved after the washing, and the phase difference among the individual cells gradually recovered. The analysis of rhythms of randomly chosen cells from the whole slice culture under these conditions showed that all cells were reset to the luminescence zero-point when CHX was applied for 36 hours, and the phase differences among the cells became evident within 4 to 5 days of their restart (Fig. 3D; fig. S6B). Consequently, there was a highly significant correlation between the respective peak times before the CHX treatment and the peak times after the washout of the drug (r = 0.732, P < 0.001, n = 177) (Fig. 3E). Thus, even though CHX sets all cells to the same initial phase, interactions between the cells reestablish a stable program of gene expression across the assemblage that is determined by intrinsic properties of the network rather than by stochastic events.

To elucidate the role of electrical activity through voltage-dependent Na+ channels in synchronizing oscillatory cells, we used tetrodotoxin (TTX), which selectively and reversibly blocks the channel and inhibits the generation of action potentials without affecting membrane potential, K+ currents, Na+ pump mechanism, or local depolarization of postsynaptic membranes. The in vitro recording obtained from hypothalamic slices demonstrates that SCN action potentials are abolished when 0.5 μM of TTX is added to the bath used in the slice culture (14). The luminescent rhythm persisted in the continuous application of TTX (0.5 μM) (Fig. 4A), consistent with previous electrophysiological data in vivo (15), in fresh slice (16), or in dispersed cell culture (2). However, the amplitude of the rhythm was diminished cycle by cycle. Because the peak-trough amplitude recovered immediately to the pretreatment level when TTX was eliminated at the seventh day, it is unlikely that the long-term application of TTX causes permanent harm to SCN cellular function.

Fig. 4.

Tetrodotoxin disrupts the synchrony and suppresses the oscillatory activity of cell clocks. (A) The suppression of SCN bioluminescence during T TX treatment (0.5 μM for 7 days), and recovery of luminescence after T TX washout. (B) Quantitative PCR with TaqMan probes of mPer1, mPer2, and GAPDH genes (left), and immunocytochemistry of mPER1 and mPER2 proteins in slices (right). For the PCR, slices were taken at the peak luminescent times at the third day after the application of the drug. The mean value before the application was defined as 100 in each probe and expressed as means ± SEM (n = 6). For immunocytochemistry of mPER1 and mPER2 proteins, we examined slices at 0, 4, 12, and 16 hours after the peak bioluminescence, with or without T TX treatments. Scale bar, 10 μm. (C) Single-cell bioluminescence (upper, 52 cells), and the number of cells peaking at every hour (lower). (D) Single-cell bioluminescence profiles of representative six cells before, during, and after the T TX treatment. Each colored line represents the luminescent signal from each cell. Time scales are common in all figures. (E) Correlation between the peak times before T TX application (40 to 52 hours) and after T TX washout (272 to 284 hours).

To determine whether the reduced circadian amplitude was accompanied by the suppression of mPer1 and mPer2 genes, we measured mPer1 and mPer2 mRNA and protein in the slices (Fig. 4B). When mRNAs were measured at their expected peak time (the peak time of luminescence), mPer1 and mPer2 levels had been reduced after TTX treatment (P < 0.01, n = 6), in contrast with the minute change in glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Protein expression was also suppressed. After TTX treatment, mPER1 and mPER2 immunoreactivities were abolished at their expected peak time (4 hours after the peak of luminescence), whereas both were expressed in control slices (Fig. 4B, right panels). These findings suggest that the Na+-dependent action potential is an essential element of the free-running molecular clock of pacemaker neurons.

By analyzing luciferase luminescence at the single-cell level, we found that TTX broke the harmonized expression of the luciferase rhythm. This desynchrony became evident day by day (Fig. 4, C and D), accompanying the suppression of the amplitude of luminescence. At the sixth day of TTX treatment, synchrony of the cell rhythm across the slice was severely disrupted (Fig. 4C), reflected by the increased variation in the peak time of cellular luminescence. This result suggests that the Na+-dependent action potentials, which influence both spike production and voltage-dependent intracellular signal transductions, are needed for establishing the synchronized program of clock gene expression between cells. After washing out TTX, the amplitude of each cell rhythm and the phase order among SCN cells were restored. At the third cycle of TTX withdrawal, the phase order was generally recovered as before (Fig. 4D), and there was a strong correlation between the peak times before and after (r = 0.704, P < 0.001, n = 48) (Fig. 4E). Thus, the intrinsic property of the assemblage was not influenced by the preceding loss of synchrony by TTX.

We have demonstrated that a stable ensemble SCN rhythm is orchestrated within an assembly of cellular clocks that are differentially phased and that sit in a distinct topographic order in the SCN. Recently, several reports suggested the importance of intercommunication among pacemaker neurons by means of neuropeptides for the rhythm generations in Drosophila (17) and mice (18). This view is also consistent with the report that electrical silencing of pacemaker neurons in flies eliminates free-running molecular rhythms as well as behavioral rhythms (19). Because treatment with TTX distorted cell synchrony and suppressed clock gene expression, Na+-dependent action potentials appear to be crucial both for intercellular synchronization and for maintaining the cell-autonomous circadian oscillation in the SCN. Elucidation of the mechanisms of synaptic communication and other dependent signaling events between and within SCN neurons (20, 21) will provide the necessary definition of how the core molecular loop and the electrical signals at the cell surface interact to sustain temporal and spatial order across the SCN network.

Supporting Online Material

www.sciencemag.org/cgi/content/full/302/5649/1408/DC1

Materials and Methods

Figs. S1 to S6

Movie S1

References

References and Notes

View Abstract

Navigate This Article