Over 90% of Siberian hamsters (Phodopus sungorus) fail to reentrain to a 5-h phase delay of a 16:8-h photocycle. Because constant darkness (DD) restores rhythms disrupted by constant light, we tested whether DD could also restore entrainment. DD began 0, 5, or 14 days after a 5-h phase delay, and the light-dark cycle was reinstated 14 days later. All hamsters exposed to DD on day 0 reentrained, whereas 42% reentrained irrespective of whether DD began 5 or 14 days later. For these latter two groups, tau (τ) and alpha (α) in DD predicted reentrainment; animals that reentrained had a mean τ and α of 24.1 and 8.9 h, respectively, whereas those that failed to reentrain maintained a mean τ and α of 25.0 and of 7.1 h, respectively. Restoration of entrainment by DD is somewhat paradoxical because it suggests that reentrainment to the photocycle was prevented by continued exposure to that same photocycle. The dichotomy of circadian responses to DD suggests “entrainment” phenotypes that are similar to those of photoperiodic responders and nonresponders.
- phase shift
- light-dark cycle
- activity rhythm
circadian rhythms are synchronized to the day-night cycle by the process of entrainment in which photic stimuli adjust the phase of the endogenous circadian pacemaker until it adapts a stable phase relationship to the photocycle. Resetting of the pacemaker proceeds reliably even if the light-dark (LD) cycle is artificially phase shifted by as much as 12 h or if animals are phase shifted repeatedly during reentrainment (3, 6, 9). The behavior of circadian rhythms during reentrainment has been a useful tool to elucidate some general features of pacemaker function such as the temporal relationship between the pacemaker and overt rhythms, the presence of transient cycles during reentrainment, and the asymmetry of reentrainment rates via phase advances or delays (6, 19). These reentrainment phenomena have been explained at a formal level of analysis by a nonparametric model of entrainment in which light produces abrupt daily phase shifts of the pacemaker (19). The response of the pacemaker to light pulses changes across the circadian cycle, but the basic circadian pattern of responses [i.e., the phase response curve (PRC)] bears strong similarities among a diverse number of species. In fact, the shape of the PRC among diurnal rodents is similar to that of nocturnal rodents (7), even in those that entrain primarily via modulation of pacemaker velocity (18).
Phase shifts of the photocycle are invariably followed by phase changes in the circadian pacemaker. We know of no reports in which organisms failed to reentrain to phase shifts of typical LD cycles (e.g., 8–16 h of light/24 h). Siberian hamsters have, however, proven to be exceptional in this regard. These animals are maintained in 16 h light/day and readily reentrained when the LD cycle was phase advanced or delayed by 1 or 3 h, but they failed to reentrain to phase advances or delays of 5 h (23). Hamsters that failed to reentrain free ran for several months or became arrhythmic despite the continued presence of the LD cycle (25). The most consistent feature of free-running activity rhythms in these “nonentrained” Siberian hamsters was that the period [i.e., tau (τ)] of their rhythms increased after the phase shift as it does under constant light (LL). We calculated mean τs from 24.8 to 25.3 h in separate experiments (25) with animals that free ran after the photocycle was phase delayed by 5 h. These values are similar to those reported for Siberian hamsters monitored in LL under similar light intensities [25.3 h (13), 24.7 h (14), and 24.8 h (21)]. We also observed that, after a few months of free running in the photocycle, activity rhythms in some hamsters eventually degrade and become arrhythmic like those of other nocturnal species maintained in LL for long durations (8, 10, 28). By contrast to LL, long-term exposure to constant darkness (DD) does not degrade circadian organization in Siberian hamsters or other nocturnal rodents. Activity rhythms of nocturnal rodents are generally more robust and free run with shorter τs in DD than in LL (4, 5,19). These observations indicate that the effects of a 5-h phase shift on τ and activity rhythm coherence in hamsters are similar to those induced by exposure to LL.
The disruptive effects of long-term bright light exposure on activity rhythms are not permanent and can be reversed by reinstating the LD cycle or transferring animals to DD (8). Even rats that are arrhythmic under LL can regain circadian organization within 1 wk of transfer to DD (10). Body temperature and sleep-wake rhythms that are eliminated by long-term exposure to bright light are restored within a few days of reinstating the LD cycle and then free run when subsequently held in dim light (12). The tendency of DD to ameliorate the disruptive effects of LL on rhythm organization led us to hypothesize that exposure to DD could restore normal entrainment in our nonentrained hamsters. The present study tested whether reentrainment could be restored to Siberian hamsters that failed to reentrain to a phase-delayed photocycle by housing them in DD for a brief period of time before returning them to the LD cycle.
Siberian hamsters (Phodopus sungorus sungorus) were bred in our colony from original stock provided by Dr. Irving Zucker of the University of California, Berkeley, CA. Animals were maintained three to four per cage from birth in a 16:8-h LD cycle (lights on at 0200 PST) and ambient temperature of 22°C. Housing conditions and illumination specifications in the hamster colony and experiment rooms were as previously described (25). Light fixtures contained two cool white (4,100°K) fluorescent tubes (Philips 40 W); light intensity was 350 lx in the center of the room and 10–45 lx on the cage floors depending on cage location in the room. Light measurements were made with cage lids and water bottles in place and food hoppers filled; the light meter lens (ILM 350, Iso-Tech) pointed upwards from the cage bottom for these measurements. Animals were provided cotton batting for nesting material; food (Purina chow #5015) and tap water were available ad libitum. All procedures were carried out in accordance with the guidelines of the American Physiological Society for animal care and use in research (1).
Activity was measured by passive infrared motion detectors as described previously (25). Each detector is mounted directly above the tip of the water bottle sipper tube, and the coverage pattern is configured so that activity levels primarily reflect drinking behavior but also include locomotor activity that occurs directly under the sipper tube. The temporal resolution for detecting successive bouts of activity is 1–2 s between bouts. Activity bouts were summed in 10-min intervals and stored on computer.
Activity rhythm periods were determined by a standard deviation-based periodogram analysis (11) of 10-day intervals during entrainment and 5-day intervals during free runs. The shorter intervals during free runs were necessary because τ changed noticeably over a 5- to 7-day period at several phases of these experiments. Peaks in the periodogram were deemed statistically significant if they exceeded the 99.9% confidence interval limit. Q values (i.e., power) associated with τ were used to compare rhythm coherence as described previously (25). Activity rhythms were considered entrained or free running based on periodogram analysis. Because a 10-min sampling interval was used for data acquisition, periodogram analysis occasionally estimated rhythm period to be 23.83 or 24.17 h (i.e., 24.00 ± 1 h sampling interval) in entrained animals. Rhythms were considered entrained if the period estimate was 23.83, 24.00, or 24.17 h and if daily rhythm onsets maintained a stable phase relationship to the LD cycle. The phase relationship to the LD cycle was considered stable when daily activity onsets occurred within a range of 60 min over 10 successive days. Although day-to-day variability in timing of activity onsets is typically <30 min, this definition allowed for occasional outliers in daily activity onsets. Rhythms were considered free running in the LD cycle if periodogram analysis estimated τ to be either <23.83 or >24.17 h.
Mean waveforms of activity data were constructed from the same 5- or 10-day data series used for periodogram analyses as described (25). Each waveform represents the mean activity in 10-min intervals over 24 h if the rhythm was entrained or over a period equal to τ if the rhythm was free running. The waveforms of individual animals were used to calculate α and the phase difference during entrainment. Activity onset was defined as the first 10-min interval in a circadian cycle when the number of activity bouts increased above the daily mean activity level and was sustained at or above that level for ≥30 min. Activity offset is the time when the number of activity bouts remains below the daily mean for ≥30 min. α is the time interval between activity onset and offset.
ANOVA or t-tests were used to determine differences among groups and changes in dependent variables over time (repeated measures). Dunnett's or Tukey's correction, where appropriate, was applied for all independent post hoc pairwise comparisons. Differences in the proportion of animals that reentrained under various lighting regimens were evaluated with the Chi-square distribution for goodness of fit. All group values are expressed as means ± SE. Differences were considered significant if P < 0.05.
Equal numbers of male and female hamsters were used in all three experiments. Data were combined by gender because no significant sex differences were found in any dependent measure. Animals were separated and housed singly in the same photoperiod as in the animal colony (LD 16:8, lights on at 0200 PST). Fourteen days later, the LD cycle was phase delayed via an extension of the light phase by 5 h (lights on at 0700) and animals were divided into three groups (n = 12 each). The first group, designatedDD0, was exposed to DD beginning at the end of the 5-h extension of the light phase and held there for 12 days. The second and third groups, designated DD5 and DD14, were moved to DD at 5 or 14 days, respectively, after the phase shift and kept there for 14 days.
We chose intervals of 5 and 14 days because hamsters free ran with τs of ∼25 h in our prior studies (23, 25). τ of 25 h produces a delay in activity onset of 1 h/day, which allowed us to predict that activity onsets would eventually coincide with the new phases of dark and light onsets at 5 and 14 days, respectively, after the phase shift. The duration of DD exposure was determined from pilot data in which normal patterns of activity were either restored within 5–7 days of DD exposure or remained disrupted despite several weeks of DD. At the end of DD, the LD cycle (16:8 h) was reinstated with lights on at 0200 for DD0 and DD14 animals and at 0700 for DD5 hamsters. There was no difference on several quantitative measures among DD5 and DD14hamsters during DD. We, therefore, reinstated the LD cycle at a different time for DD5 hamsters so that distribution of phase relationships to the LD cycle would be great enough when these two groups were combined to test whether the phase relationship between activity onset and the LD cycle influenced the probability of reentrainment.
Circadian organization under LD and DD conditions.
The mean times of activity onsets were calculated for free-running animals on the day that each group went into DD. The mean times of activity onsets among DD0, DD5, andDD14 animals were 0.3 ± 0.2 and 0.3 ± 0.2 h after dark onset and 0.3 ± 0.4 h after light onset, respectively, and did not differ significantly (ANOVA,P > 0.05). The number of days that elapsed between the phase shift and exposure to DD had a significant impact on the number of animals that reentrained (χ2 = 12.7,P < 0.01). All DD0 hamsters reentrained, but only 42% of DD5 and 42% of DD14 hamsters reentrained to the photocycle (n = 5 of 12 fromDD5 and DD14 groups).
All DD0 animals had robust activity rhythms and free ran in DD (Fig. 1). τ was >24.0 h for the first 5–7 days of DD and then shortened to close to 24.0 h for the rest of the time in DD (Fig. 1). α did not change significantly during the experiment in these animals (1-way ANOVA,P > 0.05; Fig. 2). AllDD0 animals reentrained within a few days after the photocycle was reinstated. Entrained activity onsets were, however, delayed by more than 2 h compared with activity onsets before the phase shift (paired t-test, P < 0.001; Fig.3). This delay was not accompanied by a decrease in α (paired t-test, P > 0.05; Fig. 3), rather, DD0 hamsters extended their active phases into the first 2 h of the light phase (Fig. 3).
All 12 DD5 and 11 of 12 DD14 animals had detectable free-running rhythms after the phase shifts that were >24 h (Figs. 4 and5); one DD14 animal became arrhythmic and remained so for the remainder of the study (not illustrated). By contrast, only seven DD5 and eightDD14 animals continued to free run throughout DD exposure. τs of these hamsters were significantly shorter during the first 5 and last 5 days in DD compared with their baseline values [1-way repeated-measures ANOVA, DD5:F(2,6) = 6.10, P < 0.05;DD14: F(2,7) = 3.91,P < 0.05; Fig. 2]. Circadian rhythms were not detectable by periodogram analysis during the first 5 days of DD in the remaining five DD5 and three DD14 animals. Rhythmicity spontaneously returned, however, to three of theseDD5 hamsters during the last 5 days of DD (e.g., Fig. 4,left); these animals had a mean τ of 24.2 ± 0.1 h. α was significantly greater during DD exposure compared with postphase shift values in free-running DD5 andDD14 animals [1-way repeated-measures ANOVA for each group;DD5: F(2,6) = 5.67,P < 0.05; DD14:F(2,7) = 8.54, P < 0.01; Fig. 2].
Predictors of reentrainment.
Activity rhythms were compared among animals that did or did not reentrain to determine which parameters of circadian organization best predicted reentrainment. Data from DD5 and DD14groups were combined for this analysis because these two groups did not differ in the proportion of animals that reentrained, or in their mean τs, αs, or times of activity onsets during the last 5 days of DD. Animals that were arrhythmic before or during DD were excluded from these analyses. We predicted that reentrainment would be more likely to occur if the LD cycle was reinstated so that the dark phase coincided with an animal's active phase because light exposure during subjective night can prevent entrainment in these animals (25).
Mean times of activity onsets on the day that the LD cycle was reinstated did not differ among animals that reentrained and those that failed to reentrain (t-test, P = 0.81; Fig. 6). Reentrainment was just as likely to occur whether activity onset coincided with the dark or light phase of the photocycle. There were, however, striking differences in τ and α between these two groups. Hamsters that reentrained to the photocycle had significantly shorter τs during the first 5 (t-test, P < 0.001) and last 5 (t-test, P < 0.05) days in DD compared with animals that did not reentrain (Fig. 7). α was ∼2 h greater in animals that reentrained during the first 5 (t-test, P < 0.001) and last 5 (t-test, P < 0.001) days of DD exposure compared with hamsters that did not reentrain (Fig. 7). The presence of a long α and short τ in DD did not, however, guarantee reentrainment to the LD cycle. Despite having intact free-running activity rhythms in DD, one DD5 hamster and oneDD14 hamster failed to reentrain to the photocycle (e.g., Fig. 4, right).
Most Siberian hamsters fail to reentrain to a 5-h phase delay of a 16:8-h photoperiod and continue to free run for several months despite the continued presence of the LD cycle (23-25). Because only ∼9% of hamsters spontaneously reentrain to a 5-h phase shift, reentrainment rates of 100% for DD0 and 42% forDD5 and DD14 animals indicate a marked effect of DD to promote reentrainment. Because the majority of hamsters free run with stable rhythms after a 5-h phase shift, the failure to reentrain must involve changes either in input pathways to the central pacemaker or within the pacemaker itself. We consider it unlikely that input pathways to the pacemaker were compromised after the phase shift because of the changes in τ and α that occurred when the LD cycle was replaced with DD, which indicates that the pacemaker could discriminate LD from DD. Therefore, the inability to spontaneously reentrain to a 5-h phase shift of the photocycle probably reflects changes within the pacemaker itself rather than to its afferent pathways. We cannot rule out, however, the possibility that photoreceptors may also influence reentrainment. Photoreceptors are able to integrate light information over relatively short periods of time (2, 27), thus it remains possible that LD cycle history may also be encoded in photoreceptors.
The efficacy of DD to restore entrainment was greatly influenced by the interval between the phase shift and start of DD. All animals in theDD0 group reentrained and did not experience changes in α while in DD. This suggests that the extra 5 h of light experienced on the day of the phase shift were insufficient to prevent reentrainment. For the DD0 group, DD may have prevented, rather than reversed, changes in entrainment mechanisms as was the case for DD5 and DD14 animals that reentrained. The critical events that ultimately prevent reentrainment appear to occur within 5 days of the end of the light exposure on the day of the phase shift. In a prior study (22), we found that α was maximally compressed 2 days after a 5-h phase delay and, in the present study, we found that maintenance of a compressed α is associated with a failure to reentrain. It seems likely, therefore, that changes in circadian organization that prevented reentrainment to the photocycle in our previous studies were completed within 48 h of the phase shift.
One unexpected finding from the DD0 group suggests a novel flexibility in how τ and the PRC interact to produce entrainment. All of the animals in this group reentrained with substantial delays in their times of activity onsets. Such delays are typically indicative of increases in τ (19), but τs of these hamsters resumed their normal DD values well before the end of DD exposure and one can assume that τ was not affected by reinstating the LD cycle. Thus, the delay in activity onsets indicates that the 5-h phase delay altered either the shape of the photic PRC or the phase relationship between the PRC and the animal's active phases. Either of these possibilities is incongruous with documented relationships among τ, PRC, and the phase relationship between activity and the photocycle for other species (19). Relationships among these variables may be particularly labile in Siberian hamsters. Recently, it has been shown that two light pulses, delivered on sequential nights, made Siberian hamsters arrhythmic even though the animals were housed in a LD cycle; subsequent access to running wheels restored circadian organization to activity (26). Because this species is unique in many of its responses to typical lighting regimens, it provides a novel system for future studies on the relationships among τ, PRC, and entrainment.
The likelihood of reentrainment in DD5 and DD14animals could be predicted from changes in τ and α during DD exposure but not from the time of activity onset when the photocycle was reinstated. Because our previous studies showed that 5 h of light exposure at night had such a marked effect on rhythm organization (25), we were concerned that reentrainment would be less likely if the LD cycle was reinstated in such a way that the animals' active phases were coincident with the light phase. This hypothesis was not supported. Hamsters that were active during the light phase of the photocycle were equally likely to reentrain as those that were active during the dark. By contrast, changes in τ and α that occurred in DD were good predictors of whether animals would reentrain. Although τ increased and α decreased after the phase shift in allDD5 and DD14 hamsters, these effects were reversed in DD only in those animals that subsequently reentrained to the photocycle. This expansion of α was complete and stabilized before the end of DD, thus, it is unlikely that longer DD exposure would have increased the number of animals that reentrained.
We examined the interaction between the LD cycle and activity rhythms after the phase shift to determine whether the pattern of light exposure was contributing to the maintenance of the free runs. Siberian hamsters that failed to reentrain generally free ran with τs that ranged from 25.0 to 25.5 h and αs close to 7.5 h (23,25). This combination of τ and α, in the presence of an 8-h dark phase, means that these animals were exposed to light during part of their subjective night on nearly every day of their free run. This pattern of light exposure is important because, in Siberian hamsters, light exposure that is limited to subjective night increases τ to values that are similar to those observed in our previous studies of unentrained animals (14) and may have additional effects on entrainment. Therefore, it is possible that disruptions in entrainment were initiated by the phase shift and sustained by the animal's interaction with the photocycle. Hamsters in theDD0 group, for example, were not exposed to light after the phase shift and readily reentrained.
The individual differences that could account for variability in response to DD among the DD5 and DD14 groups remain unclear to us but may be related to the general phenomenon of “photoresponsiveness” of photoperiodic animals. Within populations of photoperiodic species, there are often subsets of animals that fail to respond to the decreases in daylength that signal approach of winter (i.e., nonresponders). Siberian hamster nonresponders generally have longer τs, shorter αs, and lower PRC amplitudes in short daylengths than do responders (20). The animals in this study that failed to reentrain had τs and αs in DD that were similar to those of nonresponders, whereas animals that reentrained were more similar to responders. Our colony of hamsters was derived from breeding stock in which approximately one-half of the animals were photoperiodic nonresponders (16, 17). This is intriguing because one-half of the DD5 and DD14 animals maintained short αs and long τs in DD as one would expect for photoperiodic nonresponders and suggests that our entrainment phenomena are related to photoperiodic responsiveness. Although it is possible that these photoperiodic phenotypes may underlie circadian changes observed in DD, it does not explain why 90% of the entire colony failed to reentrain to a 5-h phase shift of the LD cycle (25). In that situation, photoperiodic responsiveness does not appear to differentiate “reentrainers” from “nonreentrainers.” There is an additional important difference between the animals that failed to reentrain in this study and photoperiodic nonresponders. α of Siberian hamsters that are nonresponders in short daylengths increases consistently by several hours when hamsters are housed in DD (15); these animals subsequently undergo gonadal regression (15). If our nonentrainers were also nonresponders, α would have increased in DD, but it did not. Thus, the available evidence does not provide a decisive link between photoperiodic and entrainment phenotypes. Nevertheless, we are currently breeding populations of photoperiodic responders and nonresponders to directly investigate the relationships among these phenotypes.
The authors thank P. Franken and J. Larkin for comments on an earlier version of this manuscript and J. Panta for animal care.
This work was supported by National Institutes of Health Grant MH-60385.
Address for reprint requests and other correspondence: N. F. Ruby, Dept. of Biological Sciences, 371 Serra Mall, Stanford Univ., Stanford, CA 94305-5020 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published August 22, 2002;10.1152/ajpregu.00362.2002
- Copyright © 2002 the American Physiological Society