Responses of brain and behavior to changing day-length in the diurnal grass rat (Arvicanthis niloticus)

Abstract

Seasonal affective disorder (SAD) is a major depressive disorder that reoccurs in the fall and winter when day-lengths get short. It is well accepted that day-length is encoded by the principal circadian clock located in the suprachiasmatic nucleus (SCN), but very little is known about day-length encoding in diurnal mammals. The present study utilized the grass rat, Arvicanthis niloticus, to investigate how the circadian system responds to photoperiodic changes in a diurnal mammal that shows day-length dependent mood changes. The animals were initially housed in equatorial day-length (12hr, EP) followed by either long (16hr, LP) or short (8hr, SP) photoperiods. The LP animals showed an expansion of the peak phase of the PER1 and PER2 rhythm in the SCN as well as an extended behavioral active phase. In contrast, the SP animals did not show any compression of their active phase nor a change in the peak duration of PER1 or PER2 expression, compared to those in EP. The results suggest that the circadian system in the diurnal grass rats is less responsive when day-lengths get short compared to when they get longer. The depression-like behaviors were assessed using sweet solution preference (SSP) and forced swimming test (FST). Animals in the SP group showed decreased SSP and increased immobility time in FST as compared to the EP group, suggesting a depressive phenotype. The present study serves as the first step toward exploring the role the circadian system plays in SAD using a diurnal rodent model.

INTRODUCTION

Seasonal affective disorder (SAD) is one of the most striking seasonal changes observed in humans (Rosenthal et al., 1984). SAD is a major depressive disorder that involves recurring depressive episodes in the fall and winter when day-length gets progressively shorter (Howland, 2009). A link between SAD and circadian dysfunction has been suggested by the facts that light is the most salient synchronizer for circadian rhythms and that light therapy is often effective in treating SAD (Rosenthal et al., 1984, Pittendrigh, 1993). Several hypotheses, such as the photoperiod hypothesis, melatonin hypothesis and phase-shifting hypothesis, have been put forward that focus on the putative circadian underpinnings of this disorder (Levitan, 2007). However, how the circadian system in humans responds to photoperiodic changes is not well understood due to ethical and technical difficulties.

In mammals, the principal circadian clock is located in the suprachiasmatic nucleus (SCN) of the anterior hypothalamus (Klein et al., 1991). The SCN is also a fundamental element of a seasonal timer regulating annual rhythms in behavior and physiology (Goldman, 2001). Lesions of the SCN abolish both the daily rhythms in behavior and physiology, and the responses in reproductive function to changes in day-length (Rusak and Morin, 1976). Based on studies conducted mainly on nocturnal species, it is well accepted that in mammals, day-length information is encoded in the SCN (Hofman, 2004, Sumova et al., 2004, Johnston, 2005, Meijer et al., 2007, Meijer et al., 2010). However, changes in day-length have a very different impact on behavioral rhythms between nocturnal and diurnal species (Refinetti, 2004). Therefore, a diurnal model should provide further insight into the photoperiodic responses of the circadian system in humans. The only work addressing the response of the SCN to photoperiodic changes was from Soay sheep, a seasonal breeding species living at temperate latitude (Lincoln et al., 2002). Almost nothing is known about how the circadian system responds to photoperiodic changes in non-seasonally breeding diurnal species.

To fill this gap, we utilized a diurnal equatorial rodent species, the unstriped Nile grass rat, Arvicanthis niloticus. The grass rats show diurnal patterns of general activity and do not show seasonal changes in their reproductive functions (McElhinny et al., 1997, Blanchong et al., 1999). More interestingly, the grass rats show depression-like behaviors under winter-like lighting conditions, when the day-length is shortened to 5hrs or the daytime light intensity is reduced to 50 lux (Ashkenazy-Frolinger et al., 2010, Workman and Nelson, 2011, Yan et al., 2011). The light-dependent depressive responses suggest that the grass rat is a potential animal model of SAD (Workman and Nelson, 2011). In the present study, the daily profile of the clock proteins PER1 and PER2 in the SCN, activity rhythms and depression-like behaviors were examined in grass rats housed in different photoperiods. The results reveal unique patterns of response of the circadian system to changes in photoperiod that are associated with the emergence of depressive behaviors in this diurnal species.

EXPERIMENTAL PROCEDURES

Animals

The grass rats (Arvicanthis niloticus) were from a laboratory colony established with animals imported from East Africa. These animals are equatorial; 12hr light is the typical photoperiod they experience in their natural habitat. Adult male grass rats (n=82) were kept in a 12:12h light/dark (LD) cycle (equatorial photoperiods, EP), with lights on at 6:00 and lights off at 18:00. Food (Prolab 2000 #5P06, PMI Nutrition LLC, MO) and water were provided ad libitum. All experimental procedures were approved by the Michigan State University Animal Use and Care Committee.

Immunocytochemistry (ICC) of PER1 and PER2 in the SCN

Animals (n=48) were randomly assigned into two groups, with one group moved to long photoperiods (LP, 16:8hr LD) and the other to short photoperiods (SP, 8:16hr LD), by either delaying or advancing the dark onset. Following 4 weeks of housing in their respective photoperiods, animals were anesthetized (pentobarbital, 200 mg/kg, ip) and perfused intracardially with saline followed by fixative (4% paraformaldehyde in 0.1M phosphate buffer) every 4 hr (n=4/time point) throughout the day starting at 2 hr after the dark onset time, which was defined as Zeitgeber time 12 (ZT12). Following the perfusion, brains were removed and post-fixed overnight in the same fixative, then cryoprotected in 20% sucrose for 2 days. ICC was carried out as described previously (Yan and Silver, 2004; Ramanathan et al., 2006). Alternate sets of coronal sections (30μm) were cut through the entire SCN using a cryostat. Sections were incubated with either PER1 (1:5000, gift of Dr. D. R. Weaver, University of Massachusetts, MA. now available as Millipore antibody AB2201) or PER2 (1:5000, Millipore antibody AB2202) for 48 hr at 4°C. Sections incubated with PER1 antibody were processed with avidin-biotin-immunoperoxidase technique using DAB as the chromogen. Sections incubated in PER2 antibody were processed with avidin-biotin-immunoperoxidase technique using DAB enhanced with 4% Nickel Sulfate as the chromogen. Brains from each time point were marked with unique cut on the cortex, and sections at 6 different time points under same photoperiods were processed together as one time series. After the ICC reaction, sections were mounted on slides, dehydrated with alcohol rinses, cleared with xylene, air dried, and coverslipped with Permount (Fisher Scientific, NJ, USA).

Quantitative analysis of ICC results

For quantification, images of serial sections through the SCN were captured using a CCD video camera (CX9000, MBF bioscience, Williston, Vermont, USA) attached to a light microscope (Zeiss, Gottingen, Germany). Numbers of PER1- or PER2-immunoreactive (ir) nuclei were counted bilaterally in 3 mid-SCN sections using NIH Image J program. The counting regions for the SCN were delineated as in previous studies (Ramanathan et al., 2006). The counts from each side of the SCN was expressed as the percentage of the maximum count from the same time series, then the average counts from 3 bilateral regions was calculated to represent the value for each animal. A two-way ANOVA (time of the day × photoperiods) was first performed to assess the effect of photoperiods on the expression for each protein. If a significant interaction was detected, a one-way ANOVA followed by Tukey post-hoc comparison was used to assess the effect of time of the day on the daily expression of PER1 or PER2. In all cases, differences were considered significant when p < 0.05.

Daily rhythms of locomotor activity

Male adult grass rats (4-6 month old, n=18) were singly housed in plexiglass cages (34 × 28 × 17 cm) with motion sensors mounted on top of the cage. General locomotor activities were recorded in 5-minute bins using VitalView (Minimitter, Inc.). After one week of baseline recording and habituating to the apparatus, animals were released from EP into either LP or SP for four weeks. The first two groups (n=6/group) were moved to either LP or SP using the same protocol as that used in the ICC portion of the study, by delaying or advancing the dark onset. The third group (n=6) was also moved to SP but through delaying the light onset. The duration of the active phase, stability of entrainment and phase angle for activity onset and offset were examined in each condition. No significant difference was observed between the two SP groups and the results were pooled together. Stability of entrainment was calculated as the standard deviation of the daily activity onset or offset time. Phase angles were calculated as the activity onset or offset time relative to the light onset or offset time, respectively. The quantification was based on data obtained during the last week of the recording and was analyzed using one-way ANOVA followed by Tukey post-hoc comparisons.

Behavioral Analysis of depression-like responses

Animals were singly housed in plexiglass cages (47 × 25 × 20 cm) and were assigned into two groups (n=8/group). The first group remained in EP and the other was put into SP by advancing the dark onset (the same procedure as that used for our analysis of PER1). After 4 weeks, the two tests described below were performed to assess depression-like behaviors. There was a 3-day interval between the two tests.

Saccharin solution preference (SSP) test

Sweet-solution preferences are commonly used to detect anhedonia in animal models of depression (Willner et al., 1987, Papp et al., 1991). Grass rats were supplied with a bottle of 1.0% saccharin (Sigma, MI) dissolved in tap water over 4 days, in addition to the regular supply of water and food. The saccharin concentration was selected based on previous experiments with diurnal sand rats and diurnal grass rats (Ashkenazy et al., 2009, Ashkenazy-Frolinger et al., 2010, Fonken et al., 2012). The amount of saccharin solution and water intake was measured every 24 h. Sweet-solution preference (SSP) was calculated daily as a ratio of saccharin solution relative to total liquid (i.e. saccharin solution + water) consumed. The first 24hr of the exposure was used as the training period for the animals to associate the location of the bottle and the taste, and the SSP over this period was defined as the baseline. There was no difference in the baseline preference between the two groups (EP: 55±8%; SP: 54±6%, t-test: t₁₄=0.16, p=0.88). If the sweet taste is a hedonic cue, an increase in the SSP in the following days is expected. Therefore, changes in the SSP for each of the subsequent testing days over the baseline were calculated for each animal, and were analyzed by two-way ANOVA followed by post-hoc Tukey’s test.

Forced swim test (FST)

This test was performed as described in previous studies using mice (Porsolt et al., 1978) or grass rats (Ashkenazy-Frolinger et al., 2010, Fonken et al., 2012). The test involves two exposures, a pre-test and a test, to a cylindrical pool (35.5 cm tall × 30.5 cm diameter) filled with 25 cm of water maintained at 28-30 °C. Water in the tank was changed after each animal. For the pre-test training section, each grass rat was placed in the water tank for 10 min. Twenty-four hours later the animal was placed in the water again and videotaped for a 5 min test session. The time that animals spent immobile was scored from the videotapes by individuals blind to the experimental conditions of each animal. All the tests were performed between ZT6 and ZT10. The immobility was scored when the grass rats were passively floating without movements other than those necessary for keeping the head above water. Climbing was scored when the animals showed upward movements with their forepaws touching the wall of the water tank. Swimming was defined when the animals showed horizontal movements other than passive floating (Porsolt et al., 1978). The results were analyzed with unpaired t-test.

RESULTS

Photoperiodic effects on the expression of PER1 in the SCN

Daily rhythms of PER1 expression were observed in the SCN of animals housed in both LP and EP, with the highest staining seen around the time of dark onset (Fig. 1A). The number of PER1-ir nuclei was compared between the LP and EP groups (Fig. 1B). A two-way ANOVA revealed a significant effect of time and the interaction between time and photoperiod (photoperiod factor, F_1,46=0.03, p=0.85; time factor, F_5,46=14.2, p<0.01; interaction, F_5,46=4.43, p<0.01). The daily expression profile of PER1 was further analyzed in the LP and SP groups. In LP (Fig. 1B, left panel), one-way ANOVA revealed a significant effect of time (F_5,23=16.76, p<0.01), with a higher level of PER1 expression observed from ZT2 to 14 compared to the other time points (post-hoc Tukey test, p<0.05). In SP (Fig. 1B, right panel), the effect of time was also significant (one-way ANOVA, F_5,23=16.76, p<0.01), with the peak expression of PER1 found at ZT10 and 14 (post-hoc Tukey test, p<0.05).

Figure 1.

PER1 expression in the SCN of grass rats in long-(LP, 16:8hr LD) or short-photoperiods (SP, 8:16hr LD). (A), representative photomicrographs depicting PER1 expression in the SCN of grass rats housed in LP or SP. (B), quantitative analysis showing the number of PER1-ir nuclei in the SCN of animals in LP (left panel) or SP (right panel). The data are presented as mean ± SEM, n=4. The gray lines superimposed to the histograms show the daily profile of PER1-ir in the SCN of animals in EP. The data are re-plotted with permission using data previously published in Ramanathan et al (2006). * indicates p < 0.05. Scale bar, 100μm. oc, optic chiasm.

Photoperiodic effects on the expression of PER2 in the SCN

Daily rhythms of PER2 expression were also observed in the SCN of animals housed in LP and EP (Fig. 2A). The number of PER2-ir nuclei was compared between the LP and EP groups (Fig. 2B). A two-way ANOVA revealed a significant effect of time and the interaction between time and photoperiod (photoperiod factor, F_1,46=0.44, p=0.51; time factor, F_5,46=13.38, p<0.01; interaction, F_5,46=19.55.43, p<0.01). The daily expression profile of PER2 was further analyzed in the LP and SP groups. In LP (Fig. 2B, left panel), one-way ANOVA revealed a significant effect of time (F_5,23=10.9, p<0.01), with a higher level of PER2 expression observed from ZT6 to 14 compared to the other time points (post-hoc Tukey test, p<0.05). In SP (Fig. 1B, right panel), the effect of time was also significant (one-way ANOVA, F_5,23=26.05, p<0.01), with the peak expression of PER2 found at ZT14 and 18 (post-hoc Tukey test, p<0.05).

Figure 2.

PER2 expression in the SCN of grass rats in long-(LP, 16:8hr LD) or short-photoperiods (SP, 8:16hr LD). (A), representative photomicrographs depicting PER2 expression in the SCN of grass rats housed in LP or SP. (B), quantitative analysis showing the number of PER2-ir nuclei in the SCN of animals in LP (left panel) or SP (right panel). The data are presented as mean ± SEM, n=4. The gray lines superimposed to the histograms show the daily profile of PER2-ir in the SCN of animals in EP. The data are re-plotted with permission using data previously published in Ramanathan et al (2006). * indicates p < 0.05. Scale bar, 100μm. oc, optic chiasm.

Photoperiodic effects on daily rhythms in locomotor activity

Data on activity patterns of animals initially housed in EP and then exposed to either LP or SP are presented in Fig. 3. The animals showed clear diurnal patterns in their locomotor activity in all housing conditions. Following the transition from EP to LP (Fig. 3A), the animals showed slightly delayed activity onsets and an extended active phase. SP was produced using two different protocols: advancing the dark onset (SP/AD, Fig. 3B) or delaying the light onset (SP/DL, Fig. 3C). Following the advancement of dark onset, the animals gradually adjusted to the new LD cycle by advancing their behavioral rhythms until stable entrainment occurred approximately 2 weeks later. When day-length was shortened via a delay of the light onset (SP/DL, Fig. 3C), no apparent changes were observed in the daily activity pattern. Following the re-entrainment of daily rhythms to the new LD cycle, no significant difference was observed in the parameters examined between the SP/AD and SP/DL groups. Therefore, the results from the two groups were pooled together for further analysis. Quantitative analysis revealed a significant effect of photoperiod on the duration of active phase (Fig. 3D, one-way ANOVA, F_2,26=9.78, p<0.01). The active duration was higher in LP compared to that in EP and SP (p<0.05, post-hoc Tukey test). However, there was no significant difference in the active duration between the EP and the SP groups (Fig. 3D, p>0.05, post-hoc Tukey test). The stability of entrainment was analyzed for animals in each group. A significant effect of photoperiod was observed for both the onset time (Fig. 3E, one-way ANOVA, F_2,26=5.76, p<0.01) and offset time (Fig. 3F, one-way ANOVA, F_2,26=6.17, p<0.01). Post-hoc comparisons revealed that the onset time was less stable for the SP group compared to EP and LP groups (Fig. 3E, p<0.05); while the offset time was more stable for the LP group (Fig. 3F, p<0.05) without significant difference between the EP and SP groups (Fig. 3F, p>0.05). A significant effect of photoperiod was also observed for the entrainment phase angles (Fig. 3G, one-way ANOVA, onset: F_2,26=48.38, p<0.01; offset: F_2,26=18.32, p<0.01). The SP group showed the largest phase angles while the LP group had the smallest phase angles of entrainment (p<0.01, post-hoc Tukey test).

Figure 3.

Behavioral responses of grass rats to photoperiodic changes. (A, B, C), representative double-plotted actograms of three grass rats that were housed initially in equatorial photoperiods (EP,12:12hr light/dark) and then to long photoperiods (LP 16:8hr light/dark, A), to short photoperiods by advancing dark onset (SP/AD, 8:16hr light/dark, B) or to short photoperiods by delaying light onset (SP/DL, 8:16hr light/dark, C). The bar graphs show the duration of active phase (D), the stability of entrainment for activity onset (E) and offset (F), the entrainment phase angle for activity onset and offset (G) of animals under each photoperiod. Gray shadow indicates the dark phase. The data are presented as mean ± SEM, n=18 for EP, n=6 for LP and n=12 for SP, respectively. * or # indicates p<0.05.

Depression-like behaviors induced by short photoperiods

To determine if the SP condition in the present study altered the affective state, the animals housed in SP and EP were first examined using the sweet solution preference (SSP) test (Fig. 4A). The baseline SSP in the first 24hr of the exposure was not different between the EP and SP groups (EP: 55±8%; SP: 54±6%, t-test: t₁₄=0.16, p=0.88). Over the following 3 days of exposure, there was a marginally significant effect of photoperiod on the SSP, with diminished levels in the SP compared to the EP groups (Fig. 4A, F_1,42=3.82, p=0.057); this variable was not affected by day of testing (F_2,42=0.36, p=0.70) or by an interaction (F_2,42=0.06; p=0.94). The results suggest that the animals in the SP group did not develop a “liking” to the sweet taste as much as those in the EP group, indicating signs of anhedonia in the SP animals. The animals were also tested in the FST (Fig. 4B). There was a significant difference in the time of immobility and climbing between the EP and SP group (t-test, p<0.05). No difference was observed in the time spend swimming (t-test, p>0.05).

Fig. 4.

Short day-length induces depression-like behavioral responses in grass rats. (A), changes in the sweet solution preference (SSP) in animals housed in SP and EP. (B), the duration for immobility, climbing and swimming behaviors during FST. The data are presented as mean ± SEM (n=8). * indicates p < 0.05.

DISCUSSION

The present study assessed the photoperiodic responses of the circadian system and the depression-like behaviors in a diurnal non-seasonally breeding species, the unstriped Nile grass rat, Arvicanthis niloticus. The results revealed that in long days, the diurnal grass rats show well-entrained behavioral rhythms and broadened peak duration of the PER1 and PER2 rhythm within the SCN, as expected from previous findings in nocturnal rodents and diurnal sheep (Nuesslein-Hildesheim et al., 2000, Lincoln et al., 2002, Sumova et al., 2004, VanderLeest et al., 2007). However, the grass rats did not respond effectively to a change from an equatorial photoperiod to short-days with respect to either the duration of their active phase in behavior or the elevated phase of PER1 or PER2 expression within the SCN. Furthermore, the grass rats housed in SP showed increased depressive responses compared to those in EP.

Light or photoperiod information is conveyed to the SCN through direct and indirect retinal projections (Moore, 1973). In habitats where the timing for lights on and off changes daily over the seasons, the circadian clock of organisms has to integrate photoperiodic information in order to maintain entrainment to the LD cycle (Pittendrigh and Takamura, 1989). Accumulating evidence on metabolic rate, electrical activity and clock gene expressions indicates that the day-length information is encoded by the SCN (Schwartz et al., 2001, Hofman, 2004, Sumova et al., 2004, Johnston, 2005, Meijer et al., 2007, Meijer et al., 2010). To explore the responses of the SCN to photoperiodic changes in grass rats, the expression of PER1 and PER2, the protein product of two core clock genes (Per1, Per2) were analyzed (Fig. 1, 2). The results revealed time-of-day dependent patterns of PER1 and PER2 in both long- and short-photoperiods. The daily rhythm of PER1 and PER2 in the SCN of grass rats under 12:12hr LD cycle (EP) has been characterized in a previous study (Ramanathan et al., 2006) and the data are re-plotted as line graphs in Fig. 1 and 2 to ease visual comparison. For PER1, an expansion in the peak duration was observed in LP (Fig. 1B, left panel), while the peak expression in SP was observed at ZT10 and 14, the same as that seen in EP, without showing apparent compression in the peak duration (Fig. 1B, right panel). The expression of PER2 under EP was high at ZT10 to 18, with a 40% reduction at ZT18 compared to ZT14 (Ramanathan et al., 2006). In LP, the number of PER2-ir nuclei reached the peak at ZT6 and remained the highest level until ZT14 before declining at ZT18, indicating an expansion in the elevated phase (Fig. 2B, left panel). In SP, the expression of PER2 started to increase at ZT10, was highest at ZT14 and 18 and declined at ZT22 (Fig. 2B, right panel). There was a delay in the peak level expression of PER2 in SP compared to that in EP, but the duration of the peak phase seemed to be similar to that in EP. Although the expression of PER1 and PER2 are generally synchronized under EP (Field et al., 2000, Ramanathan et al., 2006), slightly different responses of these proteins to photoperiodic changes have been observed previously (Nuesslein-Hildesheim et al., 2000). In Siberian hamsters housed under LP or SP as in the present study, the peak duration in LP was longer for PER1 compared to PER2 (15 vs 9hr), and there was a delay in the raising phase of PER2 compared to PER1 in SP (Nuesslein-Hildesheim et al., 2000). Consistently, different responses in the peak duration and phase between PER1 and PER2 were also observed in diurnal grass rats (present study). In addition to Per1 and Per2, numerous other core clock genes i.e. Cry1, Cry2, Clock, Bmal1, CK1ε, Rev-erbα have been identified (Reppert and Weaver, 2001, Ko and Takahashi, 2006). These genes and their protein products constitute the molecular clock through an interlocked transcriptional/translational feedback loop (Hastings et al., 2003, Okamura, 2004, Ko and Takahashi, 2006). While the expression of a single component in this inter-locked feedback loop could serve as an indicator for the overall circadian oscillation of the molecular clock (Yamazaki et al., 2000, Yamaguchi et al., 2001, Yoo et al., 2004), analyzing other clock genes and their protein products in addition to PER1 and PER2 could reveal a more completed picture of the photoperiodic responses of the circadian oscillations within the SCN. It should also be noted that the temporal resolution of sampling (4hr) of the present study also put some constraints for detecting small differences between patterns of PER expression in the SP and EP groups. Even with these caveats, the results described here clearly show that PER proteins, especially PER1, a key element of the molecular oscillator within the SCN of grass rats is considerably less responsive when day-lengths get short than when they get longer.

The asymmetrical responses of the circadian system to LP and SP were also reflected on the rhythms of locomotor activity (Fig. 3). In contrast to the animals in LP that showed an extended daily active phase, those in SP did not show any compression in their active phase compared to animals in EP (Fig. 3A-D). The expansion and compression of the active phase in short and long photoperiods, respectively, has been well documented in nocturnal species (Nuesslein-Hildesheim et al., 2000, Lincoln et al., 2002, Refinetti, 2004, Sumova et al., 2004, VanderLeest et al., 2007). In general, diurnal species show opposite responses and expand their active phase in long days and compress it in short days (Sulzman et al., 1982, Challet et al., 2002, Lincoln et al., 2002, Lahmam et al., 2008). In a previous study in which grass rats were given access to running-wheels, the duration of the active phase also showed an incomplete compression (Refinetti, 2004). The grass rats are equatorial and their circadian system has been shaped by the 12:12 LD cycle in their natural environment. It is likely that there is a “floor effect” in their photoperiodic responses of daily rhythms, such that the duration of general activity in the grass rats cannot be further compressed when day-length changed from 12 hr to 8hr. In contrast, when day-length changed from 12 to 16 hr, the active duration expanded to the entire light phase within a few days. The results seem to suggest that the circadian system in grass rats is less adaptive to SP compared to LP. It is noteworthy that when the animals were transferred from EP to SP, different protocols produced different behavioral responses (Fig. 3B, C). When the change occurred via an advance in dark onset there was a gradual shift in the times of both activity onset and offset, whereas when the light onset was delayed the daily behavioral rhythms were not affected. The results are consistent with the findings of Dr. Refinett’s study (Refinetti, 2004), and suggest that non-parametric entrainment by phase delaying light exposure around light offset is critical for this species. The parameters related to the entrainment of daily rhythms were also analyzed (Fig. 3E-G). The results show that in laboratory conditions, the daily rhythm of this equatorial species is best entrained in LP as indicated by the stability (precise onset and offset time) and the phase angle of entrainment. On the other hand, the entrainment in SP is less stable especially with the activity onset time, which is more variable or less precise compared to that in EP and LP conditions (Fig. 3E). The SP group also showed larger phase angles compared to the other groups, indicating the longer duration of activities into the dark phase (Fig. 3G).

The results of PER1 expression in the SCN and daily locomotor activities collectively suggest that the circadian system in the grass rats is less effective in making adaptive changes to short days compared to long days. We then hypothesized that the inadequate response of the circadian system to short days could be one of the mechanisms underlying depressive behavior. It has been previously reported that, in grass rats, a 5:19hr LD cycle induced increased anhedonic responses and increased immobility time in forced swimming test compared to those housed in EP (Ashkenazy-Frolinger et al., 2010). Nonetheless, whether an 8:16hr LD could produce similar effects was unknown, and therefore, was tested in the next experiment. Due to the high variability between individual animals, the effect of day-length on sweet preference was only marginally significant (Fig. 4A, p=0.057). The FST revealed significant differences in immobility and climbing times between the EP and SP groups (Fig. 4B). The results collectively confirmed the depression-like behavioral responses for animals housed under 8:16hr LD condition (Fig. 4). These typical depression-like behaviors associated with short days observed in grass rats mirror the depression symptoms seen in human SAD patients (Rosenthal et al., 1984), suggesting a great potential for the grass rats as an animal model for SAD (Workman and Nelson, 2011). Although many hypotheses focusing on the chronobiological mechanisms have been proposed to explain the etiology of SAD, how seasonal photoperiodic changes affect the brain clock in humans and more specifically in SAD patients remains a mystery. The present study examined the responses of the circadian system and depression-like behaviors following the same photoperiodic manipulation. Our results show that following 4 weeks in SP, the grass rats developed depression-like behaviors compared to those in EP. A question that needs to be addressed is why the short-day animals show increased depression-like behaviors compared to the controls in EP. In other words, what are the roles played by the circadian system or the SCN in SAD? One hypothesis is that the depressive behaviors are caused by mechanisms independent from the SCN or circadian system (LeGates et al., 2012). Alternatively, although not mutually exclusive, the depression could be derived from a mismatch between the internal timing generated by the SCN and the environment. One of the advantages of having an internal clock is that the organisms can predict and prepare for the changes in the environment (Pittendrigh, 1960). In short days when the day-length information is not effectively integrated into the daily time-keeping function of the SCN, the animals would anticipate dawn and dusk at a time prior to the actual dawn and later than the actual dusk. Although the underlying neural mechanisms are yet to be discovered, it has been shown that when the internal clock is running at a time in conflict with the environment, it can lead to many health consequences including mood changes (Abbott, 2003, Hastings et al., 2003, Haus and Smolensky, 2006).

In summary, using the equatorial diurnal grass rats Arvicanthis niloticus, the present study showed that the circadian system of the grass rats is less responsive to short photoperiods compared to long photoperiods, and the animals in short days develop depression-like behaviors. The neuropathology of SAD is not well understood, but likely involves many factors beyond the circadian system (Levitan, 2007). Developing an adequate animal model will help to provide insights into the neurological mechanisms underlying this condition (Flaisher-Grinberg et al., 2011, Workman and Nelson, 2011, Kronfeld-Schor and Einat, 2012).

Highlights.

Diurnal grass rats develop depression-like behaviors in short days.

Grass rats show longer duration of elevated PER1 and PER2 in the SCN and expansion in their active phase in long days.

Grass rats are less responsive to short days in the daily rhythm of PER1 and of behavior.

Abbreviations

EP

equatorial photoperiod

FST

forced swimming test

LD

light/dark

LP

long photoperiod

SAD

seasonal affective disorder

SCN

Suprachiasmatic nucleus

SP

short photoperiods

SP/AD

short photoperiods by advancing dark onset

SP/DL

short photoperiods by delaying light onset

SSP

sweet solution preference

ZT

zeitgeber time