Report

Characterizing a Mammalian Circannual Pacemaker

See allHide authors and affiliations

Science  22 Dec 2006:
Vol. 314, Issue 5807, pp. 1941-1944
DOI: 10.1126/science.1132009

Abstract

Many species express endogenous cycles in physiology and behavior that allow anticipation of the seasons. The anatomical and cellular bases of these circannual rhythms have not been defined. Here, we provide strong evidence using an in vivo Soay sheep model that the circannual regulation of prolactin secretion, and its associated biology, derive from a pituitary-based timing mechanism. Circannual rhythm generation is seen as the product of the interaction between melatonin-regulated timer cells and adjacent prolactin-secreting cells, which together function as an intrapituitary “pacemaker-slave” timer system. These new insights open the way for a molecular analysis of long-term timing mechanisms.

Endogenous circannual rhythms drive many long-term cycles in physiology and behavior in long-lived vertebrates (1, 2) including reproduction (3), hibernation (4, 5), migration (6), and pelage growth (7), but the anatomical and cellular bases of such rhythm generation remain a mystery. We investigated whether a circannual rhythm may be generated through a pituitary mechanism, itself dependent on the circadian system. We focused on the anterior pituitary control of prolactin secretion, but similar cell-cell interactions in the brain may govern circannual rhythms for other physiological processes.

Circannual rhythms are self-sustaining under constant conditions of day length, temperature, and food supply, with a free-running period typically less than 1 year. Geophysical cues, including the annual cycle of day length (photoperiod), act to entrain the circannual rhythm to the precise 365-day periodicity of the Earth's year (1, 2). Thus, many long-lived organisms use both an endogenous calendar and a day length–measuring mechanism to adjust physiological state precisely to the seasons.

In sheep, photoperiod-dependent changes in the duration of nocturnal melatonin secretion by the pineal gland synchronize circannual rhythms to time of year (8). Pinealectomy blocks photoperiodic responsiveness and leads to expression of variable and asynchronous long-term rhythms, whereas appropriate replacement with exogenous melatonin mimics the synchronizing effect of photoperiod. Melatonin replacement to simulate only a part of the overall annual cycle is sufficient to induce synchronous circannual rhythms (9). Short-duration daily melatonin signals (8 hours/day) given for 3 months once a year are notably more effective than long-duration melatonin signals (16 hours/day). This indicates that the summer photoperiod is the important zeitgeber for the circannual reproductive rhythm.

The neuroanatomical basis of melatonin-mediated photoperiodic control is well defined (10). In sheep and hamsters, melatonin acts within the hypothalamus to mediate control of seasonal changes in gonadotrophin secretion and gonadal activity and acts within the pars tuberalis (PT) of the pituitary gland to control prolactin secretion and its dependent biology. This differential control is strongly supported by studies in hypothalamopituitary-disconnected (HPD) Soay sheep, where ablation of the neural input to the median eminence and arcuate nucleus blocks the photoperiodic control of the reproductive and metabolic axes but spares the control of prolactin secretion, because of a direct action of melatonin within the pituitary gland (11, 12). The HPD sheep thus provides a unique in vivo model for the study of pituitary gland function in the absence of complex neural inputs from the brain (13).

We investigated whether HPD sheep exhibit a circannual rhythm of prolactin secretion under a constant photoperiod, with the aim of localizing a circannual timer mechanism. HPD Soay rams were preconditioned to short photoperiod (SP) (8 hours light/day) for 16 weeks to entrain the seasonal physiology to a winter state with low blood concentrations of prolactin. The animals were then released into constant long summer photoperiod (LP) (16 hours light/day) for 144 weeks (Fig. 1). This period was selected to cover, potentially, three repeated circannual cycles.

Fig. 1.

Free-running circannual prolactin rhythms under constant LP. (A) Long-term rhythms in blood prolactin concentrations in individual HPD Soay rams exposed to a change from SP (8 hours light/day) to LP (16 hours light/day) at week 0 and maintained on constant LP for 144 weeks. The curves were fitted by a nonlinear sine wave regression procedure. The horizontal bars indicate the timing of the spring pelage molt, known to be prolactin dependent (white bar, full molt; line, partial molt). (B) Prolactin curves for all animals on a common scale illustrating the development of asynchrony.

The transfer to LP caused an initial synchronous increase in prolactin concentrations. This was followed by a robust cyclical decline and reactivation in prolactin release that persisted throughout the 144-week study. The oscillations dampened in amplitude and became asynchronous (Fig. 1B). The period of this free-running prolactin rhythm, determined by a nonlinear sine wave regression procedure (13), ranged from 37.6 to 46. 9 weeks (mean of 40.9, SD of 2.8 weeks, n =10). The mean period was significantly (one-sample t test; P < 0.0001) different from the period of the sidereal year (13). There was a periodic “spring” molt of the pelage (recorded for the scrotal hair) following each cycle of increasing blood prolactin concentrations (Fig. 1A), consistent with the known role of prolactin in the dermal papilla (10). The molt cycle became progressively less clearly defined in parallel with the dampening of the prolactin rhythm. The testes of the HPD rams remained permanently regressed, and there was no overt cyclicity in other seasonal characteristics, as previously observed after pituitary disconnection (11, 13)

The daily blood melatonin rhythm was measured at regular intervals, and locomotor activity patterns were recorded continuously throughout the 144 weeks under constant LP (13). There was no significant (P > 0.05, analysis of variance) change either in the duration of the nocturnal melatonin peak or in the amplitude of the maximum 24-hour melatonin concentration recorded at eight different sampling occasions under LP. Locomotor activity was consistently diurnal, and there was no evidence of a change in the behavioral patterns with time under LP, or related to the progression of the circannual prolactin cycles. Thus, the circannual rhythm in prolactin secretion in HPD animals is not accompanied by any detectable changes in circadian rhythmicity.

In a second experiment, we tested whether an abrupt change in photoperiod would reset the circannual rhythm. The same HPD animals from experiment 1 were switched to SP (8 hours light/day) for 8 weeks and then reexposed to continual LP (16 hours light/day). The resulting prolactin patterns are shown in Fig. 2A. The exposure to SP inducedarapiddeclineinprolactin concentrations within a week in all animals, and values remained low during the 8-week treatment. The release back into LP reactivated a robust prolactin response and a high-amplitude prolactin rhythm with a corresponding pelage molt in most animals, very similar to that seen at the start of experiment 1 (Fig. 2A). Analysis of the phase of the new circannual rhythm after 8 weeks of SP, assessed by the sine wave regression procedure, revealed that full reentrainment had occurred. The new phase was not dependent on the phase immediately before SP exposure (Fig. 2B). This “type 0” resetting characteristic (14) indicates that the circannual oscillator expressed in the sheep is highly labile to altered photoperiodic input.

Fig. 2.

Resetting the circannual prolactin rhythm. (A) Long-term blood prolactin rhythms in individual HPD Soay rams under constant LP (starting at week 0), following exposure to SP for 8 weeks after prolonged LP for 144 weeks (a resetting protocol). The timing of the pelage molt is indicated (horizontal bar as in Fig. 1). (B) Phase-transition plot showing that the magnitude of the phase-shift in the circannual rhythm induced by SP was unrelated to the previous circannual phase, consistent with a type 0 resetting response.

The pronounced effects of photoperiod on prolactin secretion in the HPD sheep are mediated by altered melatonin secretion (11). We therefore investigated whether the pineal melatonin signal is necessary for the expression of a circannual prolactin rhythm under constant photoperiod. In a new group of HPD Soay rams, the rhythmic secretion of melatonin was permanently blocked by superior cervical ganglionectomy, a procedure that denervates the pineal gland (15). The absence of a nocturnal increase in melatonin concentrations in these “HPDX” animals was confirmed by radioimmunoassay. The groups (HPD and HPDX) were exposed to a change from LP (16 hours light/day) to SP (8 hours light/day), and after 16 weeks returned to LP for 96 weeks. The standard HPD sheep showed the expected increase in blood prolactin concentrations under LP and a decrease in response to SP, with rapid transitions. The switch to constant LP produced a robust circannual prolactin oscillation, with two full cycles completed within 88 weeks (Fig. 3A). The removal of melatonin in the HPDX sheep disrupted both the initial photoperiodic response and the long-term expression of circannual prolactin rhythms (Fig. 3A). Alignment of the prolactin profiles to the time of the HPDX operation revealed that the loss of the LP melatonin signal because of the surgery caused an immediate and progressive decline in prolactin concentrations followed by partial recovery. This is akin to that seen in HPD animals transferred from LP to SP or given a constant-release melatonin implant (16) (Fig. 3B). Thereafter, prolactin concentrations remained in the intermediate range, with a variable long-term pattern in the individual animals. There was no consistent residual prolactin rhythm with a period close to 40 weeks and no regular pelage molt in the HPDX animals. Hence, an invariant LP melatonin signal is required for the expression of a robust, circannual prolactin rhythm in HPD sheep. This inference contrasts with the observations of Thrun and colleagues (17), who suggested that variability in melatonin secretion during continuous exposure to SP might lead to expression of circannual variation in reproductive status.

Fig. 3.

Effect of removing the melatonin signal on circannual rhythmicity. (A) Long-term blood prolactin rhythms in HPD and HPDX Soay rams exposed to changes between SP and LP and then maintained under prolonged LP for 96 weeks. The time of the surgical operations is indicated (ops, arrow). Removal of the melatonin signal blocked both photoperiod responsiveness and circannual rhythm generation. (B) Blood prolactin concentrations in HPD Soay sheep aligned to the time of treatment comparing the effects of removal of melatonin (HPDX), transfer from LP to SP (HPD LP-SP), and treatment with a subcutaneous melatonin implant (HPD Mel) (16). All three treatments effectively blocked the LP melatonin signal and produced a similar winter-like default response.

It has been suggested that circannual rhythms emerge as a subharmonic of circadian rhythms through “frequency demultiplication” (1). This idea has not gained favor, however, as the period of circannual rhythms appears to be independent of the period of the daily light-dark cycle on which animals are held (1, 18, 19). Given these findings, we prefer a model in which the dependence of the circannual prolactin rhythm on the circadian melatonin signal reflects a permissive requirement for circadian input. Thus, the circadian melatonin signal is required for circannual behavior to emerge, but does not determine the period of that behavior.

Two types of mechanism can be invoked to account for the generation of a circannual rhythm in the HPD sheep. The first is one in which the pituitary cells are driven by a timer in the brain or elsewhere that is responsive to pineal melatonin and that produces a rhythmical signal to control the pituitary gland. The pituitary is thus downstream from the putative circannual pacemaker. The second mechanism is where the timer exists within the pituitary gland and melatonin acts directly in the pituitary tissues. The latter model is consistent with the data from the HPD sheep where the surgery isolates the pituitarygland andblocks all known hypothalamopituitary-dependent changes in seasonal physiology (13). Melatonin receptors are highly expressed in the PTof the pituitary gland in HPD sheep, as in the normal animal (20), and can thus act as the local target for melatonin. The PT produces a prolactin-releasing factor modulated by melatonin that regulates the synthesis and release of prolactin by the lactotrophs that are located in the adjacent pars distalis (PD) of the pituitary gland (21). The local administration of melatonin close to the PT in sheep, using microimplants or programmed infusions, markedly affects prolactin secretion, with rapid suppression and slow recovery after 8 to 12 weeks, paralleling thechange seen in response to SP (10). These treatments produce no detectable increase in the concentration of melatonin in the peripheral circulation. The result provides the strongest support for a localized timer mechanism within the pituitary. Other studies demonstrate that prolonged exposure to a fixed photoperiod produces changes in PT function [glycoprotein hormone, subunit α, and thyroid-stimulating hormone, subunit β, gene expression (22, 23)] and in prolactin-releasing factor production (21) that preempt the endogenous cycle in prolactin secretion. This indicates that it is PT cells that drive the cycle in the lactotrophs.

We propose that the intrapituitary communication between PT photoperiod-relay cells and PD lactotroph cells is central to the propagation of a circannual prolactin rhythm. This is illustrated schematically in our working model for circannual rhythm generation within the pituitary gland (Fig. 4). We know that the nocturnal melatonin signal regulates circadian rhythms in clock gene expression in the mammalian PT (24, 25), and that clock gene rhythmicity continues to reflect the ambient light-dark cycle, even during prolonged photoperiod treatments when photorefractoriness develops (22, 26). This indicates that control depends on a stable melatonin signal and that the melatonin rhythm may be permissive because of its gating effect on processes in the PT that depend on clock gene expression. By contrast, endogenous changes in the PT appear to drive the circannual cycle in prolactin secretion. Moreover, we have shown that temporarily blocking prolactin secretion with bromocriptine fails to perturb the phase of this endogenous cycle (27). This favors our view that the PT cell may be the circannual pacemaker for this system (Fig. 4).

Fig. 4.

Working hypothesis for a tissue autonomous circannual timer. The model proposes a pacemaker-slave mechanism operating within the pituitary gland. The pacemaker is the PT cell that receives circadian gating through the nocturnal melatonin signal, by means of the melatonin MT1 receptor, and has interval timer properties. The slave is the PD lactotroph cell that has stochastic and heterogeneous properties and must be synchronized as a population of cells by a positive stimulus from the PT. The propagation of circannual oscillation is presumed to depend on feedback signals with long time-delays.

Lactotrophic cells are known to exhibit stochastic variability in prolactin gene expression and secretion over circadian and noncircadian time scales (28, 29), and the heterogeneous population of lactotrophs is known to be activated by LP by means of a PT prolactin-releasing factor (30). We therefore propose that the function of the PTcells is to provide a coordinating signal for the lactotroph population that generates a slave response (Fig. 4). According to this hypothesis, ablation of the melatonin signal results in desynchronized activity within the lactotroph population, accounting for the winter-like default pattern in prolactin secretion, as well as loss of coordinated circannual rhythmicity, as observed in the HPDX animals (Fig. 3). Further studies will be required to determine whether long-term rhythmicity is an intrinsic property of the melatonin-regulated PT cells or whether it emerges through feedback interactions within the pituitary, or possibly involves peripheral prolactin target tissues.

Supporting Online Material

www.sciencemag.org/cgi/content/full/314/5807/1941/DC1

Materials and Methods

Figs. S1 and S2

Tables S1 and S2

References

References and Notes

View Abstract

Navigate This Article