Melatonin Does Not Affect the Stress-Induced Phase Shifts of Peripheral Clocks in Male Mice

Abstract

Repeated or chronic stress can change the phase of peripheral circadian rhythms. Melatonin (Mel) is thought to be a circadian clock-controlled signal that might play a role in synchronizing peripheral rhythms, in addition to its direct suppressing effects on the stress axis. In this study we test whether Mel can reduce the social-defeat stress–induced phase shifts in peripheral rhythms, either by modulating circadian phase or by modulating the stress axis. Two experiments were performed with male Mel-deficient C57BL/6J mice carrying the circadian reporter gene construct (PER2::LUC). In the first experiment, mice received night-restricted (ZT11-21) Mel in their drinking water, resulting in physiological levels of plasma Mel peaking in the early dark phase. This treatment facilitated re-entrainment of the activity rhythm to a shifted light-dark cycle, but did not prevent the stress-induced (ZT21-22) reduction of activity during stress days. Also, this treatment did not attenuate the phase-delaying effects of stress in peripheral clocks in the pituitary, lung, and kidney. In a second experiment, pituitary, lung, and kidney collected from naive mice (ZT22-23), were treated with Mel, dexamethasone (Dex), or a combination of the two. Dex application affected PER2 rhythms in the pituitary, kidney, and lung by changing period, phase, or both. Administering Mel did not influence PER2 rhythms nor did it alleviate Dex-induced delays in PER2 rhythms in those tissues. We conclude that exogenous Mel is insufficient to affect peripheral PER2 rhythms and reduce stress effects on locomotor activity and phase changes in peripheral tissues.

The rotation of the earth around its axis imparts daily light and dark cycles that led to the evolution of circadian clocks (∼24 hours, circa-about or around and dies-day in Latin) to form an internal representation of this light-dark (LD) cycle in most organisms (1-4). In mammals, circadian clocks are present in almost every tissue or organ. The suprachiasmatic nucleus of the hypothalamus (SCN) acts as the central clock, while oscillators in extra-SCN brain tissues and peripheral tissues are referred to as peripheral clocks. Ambient light resets and synchronizes the SCN to the LD cycle in the environment. The SCN in turn synchronizes peripheral clocks via a variety of neuronal, hormonal, and physiological signals (5, 6). Consequently, the whole endogenous clock system is kept in pace with the external time (5, 7, 8).

Alterations of the phase angle of entrainment between the SCN and the external LD cycles, between the SCN and peripheral clocks, or among peripheral clocks, may have negative health consequences. This view is supported by, for example, a higher incidence of metabolic disorders and cancer seen in frequent cross-continent travelers and night-shift workers (9-16). Over the last years, we and others have shown that stress, especially repeated or chronic stress, can cause phase dissociation between central and peripheral clocks by phase shifting peripheral clocks without disturbing the phase angle of entrainment of the SCN to the LD cycle (17-21).

Melatonin (Mel) is thought to be one of the output signals of the SCN to synchronize peripheral clocks (5, 22). Its release is controlled by the SCN through a sympathetic neuronal connection to the pineal gland (22, 23). Mel receptors (MT1 and MT2) are widely distributed in many tissues, such as the pituitary, liver, kidney, and lung (24-26). Importantly, some studies have found effect of exogenous Mel on circadian clocks in peripheral tissues such as the retina in mice (27), the pars tuberalis in sheep (28-30), and heart tissue in rats (31). Mel is also suggested to be an antistress hormone that antagonizes the immunosuppressive effects of acute anxiety stress in mice (32, 33), attenuates the adrenocorticotropin secretory response to stress (34-37), and counteracts the effects of exogenous glucocorticoids in rats (38). At the molecular level, Mel affects glucocorticoid receptor (GR) expression and its translocation into nucleus (39, 40), and reduces its transcriptional activity (41).

Here we address whether Mel can ameliorate the social-defeat-stress–induced phase shifts of peripheral clocks, either by its synchronizing effect or by modulating the stress effects. We used transgenic PER2::LUC knockin mice, which produce a PER2::LUC fusion protein that allows for prolonged and continuous tracking of PER2 expression using measurement of luciferase-driven bioluminescence (42, 43). In a first experiment we generated rhythmic melatonin in the blood circulation in Mel-deficient C57BL/6J mice by nighttime–restricted drinking of Mel-enriched water, to see if this reduced the stress effects on locomotor activity and PER2 rhythms in isolated tissues (pituitary, lung, kidney). In a second experiment we added Mel with and without the synthetic glucocorticoid dexamethasone (Dex) to cultured tissues to see if there is a direct effect of Mel on peripheral circadian rhythms.

Materials and Methods

Animals and Housing

Adult 2- to 5-month-old male PER2::LUC knockin mice with a C57BL/6J background from our breeding colony were used. The animals were individually housed in cages with a running wheel. Adult 4- to 8-month-old male CD-1 mice were used as aggressors for the social defeat stress (Charles River). These CD-1 mice were individually housed in a different room where the social defeats took place. All mice were housed under a 12-hour:12-hour LD cycle with lights on at 13:00 (regular light ∼30 lux) and lights off at 01:00 (dim red light <9 lux). Ambient temperature was kept at 21 ± 1 °C and relative humidity at 50 ± 2%. Food and drinking water were provided ad libitum. The experiments were conducted under Dutch rules and regulations and approved by the Central Authority for Scientific Procedures on Animals (CCD, license No. AVD1050020198665).

Methodological Validation Experiment: Inducing a Functional Melatonin Rhythm

To test if nighttime-restricted drinking of Mel-containing water (drinking protocols in experiment 1) can induce a daily Mel in the blood circulation, tail blood (∼50 μL) was collected by cutting the lateral tail vein at 4 different time points of day (early light ZT3, late light ZT9, early dark ZT15, late dark ZT21, where ZT is in hours after lights on; n = 4 for each time point, only one blood sample was taken from the same animal) from C57BL/6J male mice. Blood was collected (by pipette tips) in EDTA-coated tubes (Sarstedt), placed on ice, and centrifuged at 4 °C (revolutions per minute = 14000, 10 minutes) to obtain plasma that was then stored at −20 °C. Plasma Mel was determined with enzyme-linked immunosorbent assay (ELISA, catalog No. G280, RRID:AB_2939041) according to the manufacturer manual (NovoLytiX GmbH).

In addition, to test if nighttime-restricted Mel drinking has true physiological effects in our normally Mel-deficient C57BL/6J mice, we measured the effects on locomotor activity entrainment after a 6-hour advance of the LD cycle. Twenty animals were used for this east-bound-jet-lag protocol, and running wheel activity in the mice was recorded for 5 days before the LD shift and 8 days after the shift (original LD: lights on at 19:00, lights off at 7:00; after the shift: lights on at 13:00, lights off at 1:00). All the mice received regular tap water before the LD shift. Immediately after the LD shift, half the mice had access to night-restricted Mel water, whereas the other half of the mice received tap water (0.15% ethanol, see experiment 1 for details).

The daily activity onset was determined as described by Meerlo et al (1997) (44). Specifically, activity data were smoothened by a 2-hour running average and the activity onset was defined as the time the 2-hour smoothened data exceeded a 24-hour running average. Re-entrainment to the 6-hour advanced LD cycle was defined as the amount of time it took each animal to align its activity onset with the new LD cycle. The day of re-entrainment was the first day of each individual to have the activity onset earlier than ZT6 for at least 3 days (45).

Experiment 1: Melatonin and In Vivo Stress

Experiment 1 was designed to evaluate whether daily exogenous Mel intake can reduce the effect of social defeat stress on locomotor activity and PER2 rhythms. PER2::LUC male mice were randomly assigned to 1 of the following 4 groups: group 1, Mel ± Stress: mice received Mel in their drinking water at night and were subjected to 5 daily repeated social defeat stress; group 2, Mel ± Control: mice received Mel in their drinking water at night and were subjected only to brief daily handling when the animals of group 1 were exposed to stress; group 3, Tap ± Stress: mice received tap water without Mel and were subjected to 5 daily repeated social defeat stress; group 4, Tap ± Control: mice received tap water without Mel and only brief daily handling. The Mel-treated groups received drinking water containing Mel (10 mg/L) and 0.15% ethanol (necessary for dissolving the Mel), while the Tap groups received tap water with 0.15% ethanol only. Both the Mel and Tap groups received Mel or vehicle (0.15% ethanol) water between ZT11 and ZT21 (nighttime-restricted). They were provided with regular tap water during the remainder of the day.

In all groups, running wheel rotations were recorded before and during the stress phase of the experiment to confirm stable daily rhythms and to assess the effects of repeated social defeat. One hour after the last defeat, mice were humanely killed and samples of pituitary, lung, and kidney were collected for culturing and measurement of PER2::LUC rhythms using a LumiCycle as described next.

Social defeat stress

The social defeat stress protocol was the same as used in our previous studies (19, 20, 46). Briefly, in the late dark phase (ZT21-22), experimental mice (C57BL/6J PER2::LUC) were transported to the cages of aggressors (CD-1 mice) in another room. Each social defeat session consisted of 3 phases with a total duration of 20 minutes: initiation phase (5 minutes), physical interaction phase (10 minutes), and ending phase (5 minutes). During the initiation phase and ending phase, the intruder and the resident were divided by a perforated acrylic wall, allowing only olfactory and visual contact. During the physical interaction phase, the attacks were allowed once within 1 minute, otherwise the animals were temporally separated to enable evenly distributed attacks in 10 minutes and avoid uncontrollable biting wounds. Every day when the mice in the social defeat groups were exposed to the stress protocol, the associated control mice were gently picked up and handled for body weight measurement, and their bedding was disturbed.

Running wheel activity

Running wheel rotations were recorded and stored in 2-minute bins by an automated computer system (Circadian Activity Monitor System [CAMS], designed by Cooper, INSERM U486) (46). Data were extracted with Actoview (version 4.0) and further exported to Excel for calculation of daily and hourly activity counts (46, 47). Double-plot actograms were generated in Actoview and R studios (version 2022.07.1 + 554). Wheel rotations were analyzed for 2 time blocks consisting of 5 baseline days and 5 stress days, respectively. Twenty minutes of running wheel rotations of the social defeat hour were deleted from the nondefeated control mice since running wheel revolutions were not recorded during the stress treatment in the social defeated mice.

Experiment 2: Melatonin and In Vitro Dexamethasone

Experiment 2 was designed to assess a direct effect of Mel on peripheral clocks and its interaction with glucocorticoid signaling. Male C57BL/6J PER2::LUC mice were provided regular tap water and singly housed with running wheels, to consolidate their daily rhythms (48). The animals were undisturbed until the moment of euthanasia and tissue collection at ZT 22-23. This timing was consistent with experiment 1, and it was also the time when mice had circadian trough levels of corticosterone (20). Pituitary, lung, and kidney cortex were collected for culturing in regular luciferase recording medium for 3 days (42, 43). Then the culture dishes containing tissues were gently removed from the LumiCycle and drug solutions (10 μL/1 mL) (Mel, Dex, Mel + Dex) or vehicle (dimethyl sulfoxide [DMSO]) were added to the culture dishes (see “Drug application”). They were then resealed and returned to the LumiCycle in their original slots. The culture dishes were kept in the LumiCycle for another 3 days without a drug washout.

Drug Application

Melatonin (Sigma-Aldrich Corp) was first dissolved in DMSO and diluted to 10 μM in phosphate-buffered saline. A total of 7 to 12 μL of this Mel solution was added to the culture medium (10 μL/1 mL) to reach a final concentration of 100 nM of Mel (27). Dex was first dissolved in DMSO and subsequently diluted to 100 μM in phosphate-buffered saline. A total of 7 to 12 μL of this Dex solution was added to the culture medium (10 μL/1 mL) to reach a final concentration 1 μM of Dex (49). A total of 0.01% DMSO was prepared with the same procedure and used as vehicle. Drugs and DMSO were purchased from Sigma-Aldrich Corp.

Tissue Culturing and Measurement of PER2 Rhythms

In experiment 1, the procedures for tissue preparation and in vitro measurement of PER2 expression were similar to a previously described procedure, with minor adaptations (19, 42). In brief, after decapitation and blood collection, the head and body of the mouse were placed on ice for transportation to the culture room, where dissection and collection of peripheral tissues were performed in the light. Samples were taken from pituitary, kidney, and lung. Sella turcica containing pituitary of the sphenoid bone was excised from middle cranial fossa and immediately placed in a dish (60 × 15 mm, Greiner bio-one) with chilled cutting medium. Then the whole pituitary was separated under a long working distance microscope and transferred on membrane insert (MilleCell, PICMORG50, Millipore) in a culture dishes (35 × 10 mm, Greiner bio-one) with 37 °C prewarmed recording medium (1.2 mL). For lung and kidney samples, the left lung and kidney were collected in the chilled cutting medium and further cut into smaller pieces (∼8 mm³) by disposable scalpels. These pieces placed in culture dishes with 37 °C prewarmed recording medium (700 μL) without membrane inserts. The cutting and recording medium used in the present study was the same as published standards (19, 42). Finally, the dishes were sealed with grease (Molykote 111 Compound) and cover glasses (40 mm in diameter, thickness No.1, VWR), and placed in a LumiCycle (Actimetrics Inc) photon top counter for 5 days for in vitro culturing (∼37 °C) and PER2::LUC bioluminescence recording (43).

Phase and period of PER2::LUC bioluminescence rhythms of each tissue culture were processed and analyzed with LumiCycle software (version 3.002; Actimetrics Inc) as described in our previous study (20). For calculation of the bioluminescence rhythm amplitude, we slightly extended the data range beyond the 24- to 96-hour in the recording time frame to ensure that it included 3 complete circadian bioluminescence cycles. The amplitude of cycle 1 is calculated as the bioluminescence counts of (first peak—first trough)/2 of the baseline-subtracted data, and so forth for the amplitudes of cycle 2 and 3.

In experiment 2, the naive animals were humanely killed at ZT 22-23 and pituitary, lung, and kidney were collected following the same protocol as in experiment 1. As described in the experimental design section, the tissues were first cultured in regular recording medium as in experiment 1 for 3 days. Then drug solutions were added to the culture dishes, which were then restored in their original slots for the recording of another 3 days.

Raw bioluminescence data were processed and analyzed with LumiCycle software (version 3.002; Actimetrics Inc). The first 12 hours of data in culture were excluded because the tissue bioluminescence during this period may exhibit fluctuations due to dissection and culture medium exposure (19, 20, 50, 51). Bioluminescence data were detrended by subtracting a centered 24-hour running mean from each data point. The final range of detrended data selected for analysis was 12 to 132 hours after the start of bioluminescence recording. The time of drug administration was 71 hours (±30 minutes) after the start of bioluminescence recording. Period and phase were separately calculated before and after the drug treatment. The detrended data from 12 to 67 hours and from 72 to 127 hours were separately processed by fitting a dampened LM sin fit curve in the LumiCycle software, which would return the period of that part of the recording. Only the samples with a “goodness of fit” value greater than 70% were selected for further analysis. Peak 3 of each fitting curve (12-67 hours and 72-127 hours), which was closest to the drug treatment time, was selected as the phase marker and expressed in hours after the last lights on perceived by the mouse.

The period and peak 3 before and after drug treatment was then used to compare the period change and phase shifts. Period difference was calculated as period (after treatment – before treatment) and phase shift was calculated as peak 3 (after treatment – before treatment) (52).

Statistics

In the methodological validation experiment, plasma Mel at different times of day were compared using one-way analysis of variance (ANOVA) followed by the Fisher least significant difference test. In the east-bound jet lag protocol, the activity onset of ZT hours every day between Mel and tap drinking groups were compared using 2-way repeated-measures ANOVA followed by the Sidak test. The total number of days necessary to achieve re-entrainment of the jet lag LD between Mel and tap drinking group was assessed by unpaired an t test.

In experiment 1, total daily running wheel counts were subjected to 3-way repeated-measures ANOVA with between-subjects factors stress (social defeat vs control) and Mel (Mel vs Tap), and within-subjects factor Time (baseline days 1-5 and stress days 1-5). F tests were performed using the Geisser-Greenhouse correction when necessary, resulting in noninteger values for the degrees of freedom. To test for differences in daily profiles of activity, hourly running wheel counts were subjected to 3-way repeated-measures ANOVA with stress (social defeat vs control) and Mel (Mel vs Tap) as between-subjects factors, and time (hours of the day) as a within-subjects factor. The amplitude of PER2::LUC rhythms was analyzed using 3-way repeated-measures ANOVA with between-subjects factors stress (social defeat vs control) and Mel (Mel vs Tap), and within-subjects factor cycles. Period and phase of PER2::LUC rhythms were analyzed using 2-way ANOVA with stress factor (social defeat vs control) and Mel (Mel vs Tap) followed by Tukey test.

In experiment 2, period changes and phase shifts were analyzed using a 2-way repeated-measures ANOVA comparing before and after treatment as the “within subject factor” and Dex and Mel treatments as between subject factors. Dex and Mel treatments were indicated by 2 dummy variables (Dex and Mel) that were both 0 for the DMSO control treatment and both 1 for the Dex + Mel treatment. All statistical significance thresholds were set at α = .05.

Results

Methodological Validation Experiment

Effects of nighttime-restricted melatonin drinking on the plasma melatonin levels

The plasma Mel levels in the Mel-treated groups showed a daily rhythm with a peak in the early dark phase (Fig. 1, F_3,12 = 4.31; P < .05), which was significantly different from the early light (P < .05), late light (P < .05) and late dark phase (P < .01). No difference of Mel was found among the other 3 groups.

Figure 1.

Plasma melatonin levels in mice with night restricted melatonin drinking. Mice received melatonin (10 mg/L) in their drinking water in the night phase (ZT11-ZT21) for 3 days, then blood was collected and plasma melatonin was measured by enzyme-linked immunosorbent assay at different time points (mean ± SEM, n = 4). Plasma melatonin levels in the early dark phases were significantly different from the levels at all other time points, *P less than .05, #P less than .01.

Effects of Nighttime-restricted Melatonin Drinking on Re-entrainment Rate

To evaluate the effects of nighttime Mel drinking on locomotor rhythms, we applied the “east-bound jet lag” paradigm, in which the LD cycles were 6 hours phase-advanced, to Mel and tap drinking animals. Mice with Mel drinking subjected to an abrupt 6-hour advance of dark onset re-entrained faster than mice with tap drinking as assessed by daily activity onset phase (Fig. 2A and 2B, F_1,17 = 7.28; P < .05), and by the total number of days necessary to achieve re-entrainment (Fig. 2C; P < .01). Sidak's test indicated the difference between 2 groups on day 2 (P < .01), day 3 (P < .001), and day 4 (P < .05) during the 6-hour phase-advanced LD cycles.

Figure 2.

Re-entrainment of diurnal rhythm after 6-hour phase advance of the ligh-dark (LD) cycle. A, Representative double-plot actograms showing re-entrainment of wheel-running activity after a 6-hour advance of the LD cycle in mice drinking melatonin containing water (left) or tap water (right). The LD cycles before and after the shift are indicated by black and white bars above the panels (white = light phase, black = dark phase; upper LD bar = before the shift, lower LD bar is after the shift). Light and dark are also indicated inside the graph panels with yellow and gray-shaded areas, respectively. The numbers above the LD bars indicate local times. The solid red lines indicate the time of providing mice with melatonin-containing water or tap water (0.15% ethanol). Dashed red lines indicate the time of replacing the melatonin or ethanol containing tap water with regular tap water. B, Running wheel activity onsets on baseline days −4 to 0 (lights on 19:00, lights off 7:00) and post-LD advance day 1 to 8 (lights on 13:00, lights off 1:00). C, Number of days to full re-entrainment after the 6-h LD advance. mel = melatonin, *P less than .05, #P less than .01, $P less than .001.

Experiment 1

Effects of melatonin and social defeat stress on running wheel activity

A clear suppression of running wheel activity can be seen during the dark phase without affecting the daily activity onset following social defeat in both the mice drinking Mel or tap water (Fig. 3A). There was no difference in daily activity between groups during baseline (Fig. 3B). For the 5-day social defeat phase of the experiment (see Fig. 3B), an effect of day (F_2.34, _74.79 = 20.45; P < .001), stress (F_1,32 = 38.55, P < .001), and a day × stress interaction (F_4,128 = 18.43; P < .001) was found, but no effect of melatonin (F_1,32 = .02; P = .89).

Figure 3.

Effects of repeated social defeat stress and oral melatonin on running wheel activity. A, Representative double-plot actograms of mice from the 4 experimental groups: tap control (upper left), tap stress (lower left), mel control (upper right), and mel stress (lower right). Red lines indicate when brief daily handling or social defeat stress occurred. Activity suppression can be seen after the daily defeat in the stressed mice but not after brief handling in the control mice regardless of melatonin or tap drinking. B, Total running wheel activity per day during 5 baseline days (−4 to 0) and 5 social defeat days (1 to 5). Social defeat suppressed activity in both melatonin and tap drinking mice. Three-way repeated-measures analysis of variance indicated a general stress effect in the social defeat days. Panel Cand D show wheel-running activity per hour during baseline and stress days, respectively. Social defeat suppressed the hourly activity in both melatonin and tap drinking mice in the dark phase. Data represented as mean daily running wheel rotations. Symbols represent mean ± SEM. The red arrow in panel D indicates when the daily stress or handling occurred. mel = melatonin. Significant effects of stress: $P less than .001.

Running wheel activity showed a time-of-day variation, but the profiles were not different among the 4 groups during baseline days (Fig. 3C). During the stress days (Fig. 3D), in addition to the time-of- day variation (F_2.28, _72.95 = 157.9; P < .001), an effect of stress (F_1,32 = 37.18; P < .001) and an hour × stress interaction (F_23,736 = 13.42; P < .001) were found, but not effect of Mel (F_1,32 = .01; P = .91).

Effects of melatonin and social defeat stress on PER2::LUC expression in peripheral tissues

Representative bioluminescence traces of pituitary, lung, and kidney tissues showed clear and robust PER2 rhythms during the recording period, although amplitude differences can be observed (Fig. 4A-4C). For each of the tissues, the stress groups showed delayed PER2 phases as compared with their controls, irrespective of whether they were drinking Mel or tap water.

Figure 4.

Effects of repeated social defeat stress and oral melatonin on PER2 expression in peripheral tissues. Representative in vitro PER2::LUC bioluminescence traces from A, pituitary; B, lung; and C, kidney samples taken from control mice and social defeat stress mice with either melatonin containing drinking water or tap water (0.15% ethanol). Values are plotted as 24-hour running mean baseline subtracted photon counts per second. Recordings are from 12-108 hours after the start of the culture. Time is indicated in hours since the last lights on perceived by the animal. Average amplitude of PER2::LUC bioluminescence rhythms in D, pituitary; E, lung; and F, kidney samples from control mice and social defeat mice with either melatonin or tap water (0.15% ethanol). Amplitude values are plotted as mean photon counts per second of (peak—trough)/2 of each cycle. Two-way repeated-measures analysis of variance indicated a general suppressive effect of stress on the amplitude of PER2 expression in the pituitary. G, Period and H, phase values of PER2::LUC rhythms from pituitary, lung, and kidney from control mice and social defeat mice with either melatonin or tap drinking (0.15% ethanol). The phase was calculated as the time of the first onset of the trace, defined as its first incremental baseline crossing of the sampling time (24-96 hours of recording). C, control; mel: melatonin; S, social defeat stress. Yellow dots represent individual mice and black dots represent group means ± SEM. Red lines indicate the comparison pairs. #P less than .05, $P less than .001.

Besides the clear dampening of the PER2::LUC rhythm amplitude in all tissues, we found an additional suppression of amplitude in the pituitary of social defeated mice (Fig. 4D-4F, F_{1, 36} = 11.60; P < .01).

The periods of the PER2 rhythm in pituitary and kidney were not affected by social defeat stress or Mel (Fig. 4G). In the lung samples of the tap-drinking mice, the period of the bioluminescence rhythm was not affected by stress (see Fig. 4G). However, social defeat stress induced a longer oscillating period of PER2 in Mel-drinking mice as compared to the nonstressed control mice (see Fig. 4G; Mel + Control = 23.99 ± .27 hours, and Mel + Stress = 25.86 ± .27 hours; P < .01).

In the pituitary, stress delayed PER2 phase both in tap- and Mel-drinking mice (Fig. 4H; F_1,36= 78.74; P < .001), while Mel by itself did not (F_{1, 36}= 1.63; P = .21). Post hoc testing showed that social defeat stress significantly phase-delayed the PER2 rhythms by approximately 3 hours both in the tap- (Tap + Control =36.02 ±0.27 hours, Tap + Stress =38.83 ± 0.49 hours; P < .001) and Mel-drinking groups (Mel + Control = 36.12 ± 0.28 hours, Mel + Stress = 39.64 ± 0.35 hours; P < .001).

In the lung, stress delayed PER2 phase both in tap- and Mel-drinking mice (see Fig. 4H; F_1,39 = 231.4; P < .001), while Mel by itself did not affect phase (F_{1, 39} = 1.41; P = .24). Post hoc testing showed that social defeat stress significantly phase-delayed the PER2 rhythms by approximately 8 hours both in tap- (Tap + Control = 38.77 ± 0.34 hours, Tap + Stress = 46.29 ± 0.84 hours; P < .001) and Mel-drinking groups (Mel + Control = 39.10 ± 0.47 hours, Mel + Stress = 47.18 ± 0.47 hours; P < .001).

In the kidney, stress significantly delayed PER2 phase (see Fig. 4H; F_1,29= 68.58; P < .001), while Mel by itself did not affect phase (F_{1, 29}= 0.04; P = .84). Post hoc testing showed that social defeat stress significantly phase-delayed the PER2 rhythms by approximately 4 hours both in tap- (Tap + Control = 41.39 ± 0.29 hours, Tap + Stress = 45.24 ± 0.73 hours; P < .001) and Mel-drinking groups (Mel + Control = 41.09 ± 0.53 hours, Mel + Stress = 45.75 ± 0.60 hours; P < .001).

Experiment 2

Effects of melatonin and dexamethasone on PER2 expression in peripheral tissues

To examine the possible direct effects of glucocorticoids and Mel on peripheral clocks, we added different compounds (DMSO, Mel, Dex, Mel + Dex) to the cultured tissues collected from naive, 12:12 LD entrained mice.

Representative bioluminescence traces of pituitary, lung, and kidney before and after different drug treatments show a clear expression response and phase reset after Dex treatment in the lung and kidney (Fig. 5B and 5C) but less so in the pituitary (Fig. 5A).

Figure 5.

Effects of melatonin and dexamethasone on PER2 expression in peripheral tissues. Representative in vitro PER2::LUC bioluminescence traces from A, pituitary; B, lung; and C, kidney samples taken from naive mice and treated with vehicle (DMSO), melatonin (Mel), dexamethasone (Dex) and Mel + Dex. Values are plotted as 24-hour running mean baseline subtracted photon counts per second. Recording traces are plotted from 12-132 hours after the start of the culture. Time is indicated in hours since the last lights on perceived by the animal. Gray arrows indicate when drugs were added to the culture medium. D, Period and E, phase values of PER2::LUC rhythms from pituitary, lung, and kidney from naive mice with different drug treatments. Blue and red dots represent D, period, or E, peak 3 of same individual samples before and after the drug treatment. The mean value of different groups before drug treatment is connected by dashed black lines, and after treatment by solid dark lines. Significances in the 2-way repeated-measures analysis of variance of the before-after interactions with Dex, Mel, and Dex + Mel are indicated with symbols in each panel (*P < .05; $P < .001; NS P > .10).

Average effects of Dex and Mel treatment on period and phase changes within each tissue are quantified (see Fig. 5, Table 1). Two-way repeated-measures ANOVA were used to evaluate the period and phase changes within each tissue over all treatments (Fig. 5D and 5E). As PER2 expression exhibited varying oscillating periods in different tissues when cultured in vitro, the pituitary, kidney, and lung exhibited different phase angles or circadian phases of PER2 when drugs were administered (see Fig. 5A-5D). The timing of the treatment was denoted as the circadian hours since last peak (HSLP) before the treatment. In the pituitary samples, Dex treatment between 3 to 7 HSLP led to a significant decrease in period (F_1,36 = 51.33; P < .0001) and a significant phase delay of PER2 expressions (F_1,36 = 5.10; P < .04). In the lung samples, Dex treatment between 1 and 5 HSLP led to a significant increase in period (F_1,36 = 50.37; P < .0001) and a significant phase delay of PER2 expressions (F_1,36 = 117.17; P < .0001). In the kidney samples, Dex treatment between 16 and 20 HSLP did not induce period changes but led to a significant phase delay of PER2 expressions (F_1,36 = 63.86; P < .0001). No significant effects were found of Mel treatment alone on period (kidney P > .36, lung P > .76, pituitary P > .47), nor on phase (kidney P > .12, lung P > .18, pituitary P > .94). No significant effects were found of Mel when added to the Dex treatment on period (kidney P > .85, lung P > .59, pituitary P > .18), nor on phase (kidney P > .39, lung P > .50, pituitary P > .17).

Table 1.

Overview of average period and phase and standard errors, before and after treatment with control, dexamethasone, and melatonin, for pituitary, lung, and kidney

Tissues	Drugs	Period, h	Phase, h	Time
	DMSO	23.52 ± 0.12	88.96 ± 0.28	Before
		23.24 ± 0.29	91.90 ± 0.33	After
	Dex	23.62 ± 0.08	89.44 ± 0.32	Before
Pituitary		21.55 ± 0.42	90.40 ± 0.91	After
	Mel	23.98 ± 0.18	90.86 ± 0.67	Before
		24.37 ± 0.36	92.99 ± 0.6	After
	Mel + Dex	23.51 ± 0.05	89.03 ± 0.26	Before
		21.24 ± 0.29	90.71 ± 0.71	After
	DMSO	24.16 ± 0.17	93.07 ± 0.53	Before
		24.36 ± 0.14	95.95 ± 0.26	After
	Dex	24.06 ± 0.25	92.27 ± 0.68	Before
Lung		26.06 ± 0.17	104.13 ± 0.36	After
	Mel	24.46 ± 0.23	93.57 ± 0.62	Before
		24.59 ± 0.22	95.89 ± 0.16	After
	Mel + Dex	24.55 ± 0.32	93.28 ± 0.69	Before
		26.78 ± 0.22	103.50 ± 0.32	After
	DMSO	26.93 ± 0.26	102.85 ± 0.61	Before
		25.94 ± 0.53	100.16 ± 0.34	After
	Dex	26.97 ± 0.23	103.15 ± 0.49	Before
Kidney		26.89 ± 0.13	105.84 ± 0.09	After
	Mel	27.42 ± 0.47	104.17 ± 0.77	Before
		25.82 ± 0.39	99.59 ± 0.40	After
	Mel + Dex	27.15 ± 0.23	103.82 ± 0.78	Before
		26.67 ± 0.2	105.95 ± 0.09	After

Abbreviations: Dex, dexamethasone; DMSO, dimethyl sulfoxide control; Mel, and melatonin.

Discussion

In the present study, we confirmed our previous findings that repeated social defeat stress in the late dark phase delayed the peripheral clocks in a tissue-dependent manner. Night-restricted Mel drinking led to rhythmic Mel in the circulation in the Mel-deficient C57BL/6J mice, which functionally accelerated re-entrainment to a jet-lag paradigm. However, rhythmic Mel in the circulation did not influence peripheral clocks in the pituitary, lung, and kidney, nor did it attenuate the phase-delaying effects on peripheral clocks or the activity suppression caused by social defeat stress. In vitro, the administration of Dex significantly affected PER2 rhythms in the pituitary, kidney, and lung, either in period, phase, or both. Also here, the administration of Mel did not influence the PER2 rhythms nor did it influence the shifted PER2 rhythms induced by Dex. Together these findings suggest that Mel does not have strong entraining effects on circadian PER2 rhythms in peripheral tissues, nor does it attenuate the effect of stress and stress hormones on peripheral clocks.

Several studies have applied free Mel drinking in mice and alleged this could mimic the endogenous Mel rhythm with a nighttime Mel peak since mice display most of their drinking behavior at night (53, 54). However, in an earlier pilot study in mice with free Mel drinking as compared with tap water drinking, we did not see a clear nighttime peak in plasma Mel, nor did we see a faster re-entrainment to a 6-hour advance of the LD cycle (data not shown). Therefore, in the present study, we restricted the Mel drinking to the dark phase, which successfully induced a plasma rhythm in Mel with a peak in the early dark phase (see Fig. 1). The re-entrainment to the 6-hour phase advance of the LD was also significantly faster in mice with Mel drinking than mice with tap drinking. This is consistent with the finding that Mel-proficient C3H/HeN mice re-entrain faster to an east-bound jet lag paradigm than Mel-deficient C57BL/6J mice (55). Together these results suggest the Mel drinking protocol we used in this study mimicked an endogenous Mel rhythm and had positive effects on the activity re-entrainment.

In the present study, we observed a significant suppression of locomotor activity in defeated mice (see Fig. 3B and 3D), consistent with our previous studies (19, 44, 56). An earlier study by Kopp et al (57) reported that injection of Mel (1 mg/kg/day, intraperitoneal) counteracted the reduction of spontaneous locomotor activity after chronic mild stress exposure in mice (57). However, such buffering effects of Mel on stress-induced suppression of activity was not found in this study (see Fig. 3B and 3D). One possible cause is the different routes of administration of the exogenous Mel and the different dosing of the Mel: intraperitoneal 1 mg/kg/day in Kopp et al (57), vs orally 0.06 to 0.10 mg per day in this study. Another explanation could be that the social defeat stress used in our study was much stronger than the chronic mild stress exposure applied by Kopp at al (57), so that Mel was unable to counteract the stress effect in our study. In addition, exogenous Mel actually induced a sedative or hypnotic effect in mice, rats, and hamsters, which resulted in slight reduction in locomotor activity (58-60). We also observed a slight reduction of running wheel activity after Mel drinking, but only after stress (see Fig. 3B and 3C), although this difference was not statistically significant (see Fig. 3B). Together, the effect of Mel on locomotor activity appears to be minor, and might even further reduce locomotor activity in stressed mice.

Confirming previous work, we found that 5-day social defeat stress in the late dark phase significantly delayed PER2 rhythms in the pituitary, lung, and kidney (see Fig. 4) (17, 20). We hypothesized that rhythmic Mel at physiological levels might prevent or reduce the stress effects on peripheral clocks, either by directly entraining the peripheral clocks or by acting as an antistress hormone interfering with GR translocation to the nucleus (39, 40). However, we did not see a significant effect of Mel on the PER2 rhythms in any of the tissues examined, neither in the stressed mice nor in the control mice. In line with the present study, Owino et al (2016) (61) suggested that the removal of Mel receptors produced only subtle effects on the rhythmic expression patterns of clock genes within skeletal muscle, liver, and adipose tissue in mice. We have not found other studies on how Mel manipulations in mice interact with stress on the circadian clocks in the periphery.

In this study, night-restricted Mel-drinking resulted in a peak of Mel in the circulation in the early dark phase (see Fig. 1). It is possible that PER2 rhythms are less sensitive to external stimuli during this period, since stress in the late dark phase shifted PER2 rhythms but did not in the early dark phase (20). Therefore, we performed an in vitro study (experiment 2) to examine if there is a direct effect of Mel on the circadian clocks of tissues collected in the late dark phase, with a dose of 100 nM that was proven to induce phase shifts in cornea (27, 62). In our in vitro experiment, we found a clear effect of Dex on the PER2 rhythms in the kidney, lung, and pituitary, either on the period or phase, or both (see Fig. 5). The extent to which phase or period are affected by Dex in these different tissues may depend on the capacity of the tissues to sustain the signaling pathway of the GRs involved. However, Mel had no effects on the PER2 rhythms in any of these tissues nor did it interfere with the effects of Dex on the PER2 rhythms. It is possible that the Mel causes phase shifts of PER2 in the cornea (27), and that the absence of Mel effects in the tissues of this study are due to tissue-specific differences in Mel sensitivity. Indeed, Mel receptor varies in density among different brain areas and different tissues (24, 63). Some articles found effects of Mel on clock gene expression in various tissues, including PER2 in the SCN of mice (64), PER2 in the cardiovascular system in rats (65), and PER2 or Cry1 in the pars tuberalis of rats and sheep (29, 31) (30) but no such findings have been reported for the tissues that were the topic of the present study.

It should be noted that a phase-response curve has been established for the effects of Dex and Mel on PER2 expression (27, 66-68). One concern is that the lack of synchronization of peripheral clocks with Mel in this study was due to drug timing being at the “dead zone” of the phase-response curve. In this study, Mel was administered at 1 to 5 HSLP (lung), 3 to 7 HSLP (pituitary), and 16 to 20 HSLP (kidney) of PER2 expressions (see Fig. 5A-5C). It is not excluded that all these time points are still located in the “dead zone” of the phase-response curve. However, the observations that PER2 were not responding to Mel whether during its peak (lung), offset (pituitary), or trough (kidney) suggest the effect of Mel on Per2 is minor. Nevertheless, future experiments examining the effects of Mel administered at different circadian phases of PER2 expression in various tissues are needed to gain complete insight on the synchronization capacity of Mel on various peripheral clocks under normal and chronic stress conditions.

Conclusion

Our experiments show that repeated social stress and glucocorticoid stress hormones can shift PER2 rhythms in peripheral tissues. In addition, exogenous Mel did not affect the PER2 rhythms in peripheral tissues and was not able to prevent these stress- and glucocorticoid-induced phase shifts of peripheral clocks.

Abbreviations

ANOVA

analysis of variance

Dex

dexamethasone

DMSO

dimethyl sulfoxide

GR

glucocorticoid receptor

HSLP

circadian hours since last peak

LD

light-dark

Mel

melatonin

SCN

suprachiasmatic nucleus of the hypothalamus

Funding

This work was supported by a scholarship from the China Scholarship Council and support from Hunan Normal University, China. The equipment used was funded by Aard- en Levenswetenschappen, Nederlandse Organisatie voor Wetenschappelijk Onderzoek grant 834.11.005 “Tracing Time” to Roelof Hut.

Disclosures

The authors have nothing to disclose.

Data Availability

Data from this study are available at the FigShare repository (69).