The bear circadian clock doesn’t ‘sleep’ during winter dormancy

Using a variety of experimental approaches in captive bears and observations of naturally denning bears, we now provide the most compelling evidence to support the operation of a circadian clock during winter dormancy in brown and black bears. This was confirmed first by demonstrating that activity and Tb rhythms free-ran in constant conditions, a feature that also reveals their endogenous nature. Second, the biological clock was reset or shifted when exposed to single pulses of light applied at discrete times of the circadian cycle. Third, daily symmetrical light exposure caused activity bouts to coalesce and assume a 24 h periodicity. Fourth, fibroblasts obtained from dormant bears expressed robust clock-gene rhythms. Moreover, the period of this molecular rhythm closely matched that of the bear’s own activity and Tb rhythms but only under conditions where bear serum was used. Importantly, our findings of activity rhythms in captive bears were corroborated in wild bears. The present results are therefore entirely consistent with our earlier report of circadian rhythmicity in captive dormant bears [32] and together reveal that rhythmicity is a normal facet of torpid bear physiology.

Compared to the daily rhythms expressed in captive bears during the active (feeding) season, the activity rhythms of torpid bears were much weaker. They were however virtually identical in strength to their denning counterparts in the wild, indicating that the general conditions we used in our isolation experiments were comparable to those experienced by wild bears. Yet, based on the consistency of our findings it is surprising that others have not observed similar rhythms in hibernating bears, especially since those studies used bears exposed to environmental light and temperature fluctuations [20, 21] as some of our captive (ambient) and wild bears were. Aside from a species difference (black vs. grizzly), which seems unlikely given that we saw rhythms in a hibernating wild black bear, at least three possible explanations for this difference can be envisioned, although they are not necessarily mutually exclusive. The first relates to the low amplitude/strength of the activity rhythms in our captive bears (5–10 %), which if present in earlier studies may have been associated with correspondingly low temperature rhythms and thereby escaped detection due to methodological differences. Arguing against this is the fact that our temperature rhythms were actually more robust than the activity rhythms and were easily detected even when we used identical analytical methods as the earlier studies (e.g., Lomb-Scargle periodogram; data not shown) [21, 32]. Second, subcutaneous Tb rhythms could reflect a fundamentally different output of the circadian clock than core Tb measured using intraperitoneal implants. We also view this as unlikely since activity, a faithful reflection of master clock integrity, remained rhythmic and was phase-locked to temperature, even under different photic conditions. The third possibility relates to differences in den temperatures between studies. In the case of the Alaska and Wyoming black bears, outside temperatures and corresponding den temperatures frequently dropped below 0 °C for extended periods [20, 21] causing bears to shiver [21]. Thus, shivering may have masked an underlying circadian Tb rhythm as a result of the bear’s need to thermoregulate. Since our bears were held at a constant 7 °C and had relatively large body masses (125 kg) when entering winter dormancy, this presumably placed them above a Tb that eliminated the need for shivering [21], allowing rhythms to be unmasked. This hypothesis however, remains to be tested. A final possibility, related to the third, is that poor body condition upon entering the den could have masked or suppressed Tb rhythms in an effort to conserve energy. This is based on observations from a single juvenile wild female brown bear (213052006) who had only an estimated 8.4 % body fat in August at the time of capture. This bear entered the den in November (adiposity unknown) and promptly ceased to exhibit any clear rhythmicity (Additional file 2: Figure S4). Then, in March, rhythmicity reappeared coincident with dramatic re-entrainment to the daily light: dark cycle visible as daily delays in activity onset until activity bouts were fully synchronized to dawn and dusk around the end of April. Although not conclusive, the data from this single case are indicative of a somewhat earlier than normal den exit perhaps due to the depletion of fat reserves and the need to obtain food.

We were also able to confirm in bears another feature of the circadian clock found in many other species [42–44], namely, a longer circadian period in older animals compared to younger ones. Previous studies had shown that dramatic changes in circadian period occurred around the time of puberty whereupon the clock “slows” leading to an adult circadian period that is generally longer than that of juveniles [44], consistent with our results. Because our younger bears had just entered their fourth year and therefore had not reached full sexual maturity [45] our comparisons are between different aged bears; thus, we cannot ascribe the change specifically to sexual maturation or to a particular hormone. However, future studies could address this by measuring reproductive hormones in blood in combination with estimating circadian period under constant conditions in a captive setting. The differences in rhythm period among life stages of bears could have interesting consequences during winter dormancy. For example, female grizzly bears den with cubs from birth and for several winters afterwards [45–48]. It is conceivable therefore that a female and cubs in a family group denning together in constant conditions would drift out of phase with one another by about 10–15 min per day (based on our current circadian estimates) and in opposite directions. Over time, the net effect of this drift would be overall activity in the den appearing virtually continuous and possibly arrhythmic, eventually returning to a rhythmic condition, and so on. Whether the cubs then influence the mother’s activity rhythm or the cubs are masking the mother’s rhythm is unclear; however, examination of the pre- and post-parturition activity patterns of the two pregnant wild bears in our study (Additional file 4: Figure S2) clearly reveals a reduction in amplitude and virtual arrhythmicity of the mother bear (Fig. 3b,c). This apparent loss of rhythmicity due to reductions in activity corroborates previous findings made in pregnant bears [49]. Additional analysis of hibernating family groups consisting of older cubs would be required to determine whether this phenomenon is limited to cubs-of-the year or is a general feature of denning female bears with offspring.

Responsiveness to an environmental synchronizing cue, such as light, is a hallmark of circadian systems. This was confirmed during winter dormancy in the current study by robust light-induced phase shifts in captive bears as well as light-associated activity changes in denning wild bears. Together, these observations expand our understanding of the behavioral ecology of bears to now include a dynamic responsiveness to a relevant environmental cue, even while dormant. The general pattern of responses to light, i.e., activity onsets delayed when exposed during the inactive period and advanced in the active period, generally mimic those seen in other mammals [50]. However, the marked effect of light to reset the clock to early morning irrespective of when it was applied during the inactive period was unexpected [50] for two reasons. First, the light intensity was low (~200lux). Second, the light duration that caused the largest phase shifts was often the shortest (1 h vs. 4 h). Collectively, these results suggest that bears are exquisitely sensitive to light during hibernation, itself rather surprising since during the active season bears use food to more effectively organize their behavior than light [32]. This apparent shift however could explain their temporal flexibility [51]. Indeed, only short daily (1 h) pairs of light pulses were sufficient to accomplish entrainment, which is consistent with extensive work done in other species [52]. Our findings in wild bears also appear to support these conclusions. However, a remaining question pertinent to denning wild bears is whether they shift their body position towards/away from a light source as a behavioral adaptation or if light exposure occurred incidentally. Given that our wild bears had light sensors affixed to collars on the neck, it is possible that the eyes were facing away from the light source when exposure occurred.

The maintenance of circadian rhythmicity and light entrainment by bears during winter dormancy may be important to optimize metabolic function. For example, metabolic rate could be directly proportional to circadian strength and could explain why rhythms disappeared (or were masked) in bears from earlier studies, i.e., their metabolic threshold was reached. In fact, the reduction in rhythm strength we observed in torpid bears(~50 %) is similar, but not as low as the metabolic suppression (60–75 %) previously reported for bears [5, 53, 54] suggesting that the putative threshold may not have been reached in our bears. Alternatively, even a weak (low amplitude) circadian rhythm may be necessary to maintain lower metabolic rate until challenged with extreme thermal or other demands. This would not be entirely surprising given the mounting evidence indicating that loss of rhythmicity and desyncrhony are associated with adverse metabolic outcomes [34, 37, 55]. Indeed, recent evidence suggests that rhythmicity in biological gene expression networks of eukaryotes as diverse as yeast, fruit flies, and mice serve to optimize metabolic costs, in part by modulating rhythm amplitude [38]. Thus, it is likely that a threshold of circadian amplitude exists beyond which the presumed cost/benefit relationship is lost. For example, it’s possible that a torpid, anorectic, denning female bear loses rhythmicity when faced with the additional metabolic demands of lactation. Similarly, a bear entering the den in poor body condition (low adiposity and unable to produce cubs if female [56]) might dispense with circadian rhythmicity altogether to maximize survival. Irrespective of these possibilities, the reduction in rhythm amplitude we observed, rather than a complete loss of rhythmicity could be viewed as evidence to support a metabolic optimization strategy. Nevertheless, precisely how rhythmicity, Tb and metabolic costs in shallow heterotherms are related remains to be determined.

Using data obtained from a small number of wild denning bears fitted with light sensors we were able to determine that those bear dens received periodic light exposure despite significant snow cover (based on environmental monitoring station, camera data at den sites and site visits to the U.S. black bear den). We now can confirm that wild hibernating bears are exposed to the same stimulus – light (albeit of different intensities), that was used in our captive studies to shape activity and Tb patterns. Thus, winter dormancy in bears provides a valid, physiologically relevant, condition in which to further explore the influence of light on the circadian clock independently of the confounding influence of food entrainment [32] and over many months. This ability to explore clock function without the contaminating influence of food anticipatory activity [57] holds great promise for our basic understanding of how central and peripheral clocks are organized.

Overall, our behavioral and physiologic results provide strong support for the existence of a functional circadian clock in torpid bears during winter dormancy. Although the Tb rhythms differ significantly from daily torpor bouts seen in some birds [58], they do suggest a lower Tb set point in torpid bears and one that is defended. Indeed, earlier findings in black bears by Tøien [5] were interpreted to indicate that Tb cycles were the result of a “regulated” process, distinct from a passive hypothermic response [59]. The similar maximum daily Tb of our bears housed in constant temperature versus the bear exposed to natural temperature changes would support this conclusion. Thus, these earlier findings combined with our current ones would suggest that bears are capable of maintaining a tightly regulated torpor around a lower Tb set point and over multiple timescales. Together, these features appear to place winter dormant bears somewhere between “true” hibernators and shallow heterotherms [59].

A recognized caveat to the interpretation of activity and Tb rhythms is that these measures represent ‘outputs’ of a central (brain) clock and are therefore subject to misinterpretation as a result of masking or other factors. Thus, it is necessary to confirm clock operation in other ways, such as by examining the molecular clock directly. Circadian clocks are distributed in tissues throughout the body [60] and these provide a readily available means to assess clock operation [61]. We therefore collected skin fibroblasts from winter dormant bears and infected these cells with a lentiviral clock gene (Bmal1) construct linked to luciferase enabling a real-time luminescent readout in vitro [61]. Our findings of robust rhythms in fibroblasts obtained from the same animals whose activity and Tb rhythms were confirmed during winter dormancy now provides direct proof that a functioning biological clock is an inherent feature of dormant grizzly bear physiology. Our findings stand in stark contrast to those in European hamsters (Cricetus cricetus) whose clock “stops” during hibernation [19] and to arctic reindeer (Rangifer tarandus), who don’t hibernate, but lose their behavioral and molecular rhythms in winter when measured using similar techniques to ours [62]. Somewhat to our surprise, the period of bear fibroblast rhythms required the bear’s own serum to match that observed in vivo. These results suggest that humoral factors may play additional roles in maintaining circadian integrity of the entire metabolic engine of these animals. Because we performed our culture experiments at 37 °C and since lower culture temperatures also appear to influence fibroblast circadian period in a homeotherm [63] it remains to be determined what influence different culture temperatures and serum combinations have on these bear rhythms.