Neuropeptide Signaling Differentially Affects Phase Maintenance and Rhythm Generation in SCN and Extra-SCN Circadian Oscillators

All experiments were performed in accordance with the UK Animals (Scientific Procedures) Act of 1986 using procedures approved by The University of Manchester Review Ethics Panel.

For this study, mPer2^luc mice [46] were cross-bred with Vipr2^−/− mice [13] to generate a PER2::LUC reporter strain that lacked expression of functional VPAC₂ receptors (mPer2^luc×Vipr2^−/−; herein referred to as Vipr2^−/−). Standard mPer2^luc mice (expressing functional VPAC₂ receptors) from the University of Manchester breeding colony were used as controls (WT). All mice used in this study were adult males on a C57BL/6 background, housed at 20–22°C and humidity ∼40%, with ad libitum access to food and water.

Animals were initially bred and maintained group housed under a 12 h light∶12 h complete darkness cycle (LD; Zeitgeber time [ZT] 0 was defined as the time of lights on). Animals contributing to the LD part of the study were taken directly from these conditions and culled during the mid-late day phase (mean cull phase ZT6.7±1.0 h). Behaviorally screened mice were singly housed in running wheel-equipped cages (wheel diameter 16 cm) under LD for at least 7 days then transferred to constant darkness (DD) for at least 14 days before cull. Analyses of period and rhythm strength (percentage of variance accounted for by the rhythm; %V) of wheel-running activity for mice in DD were made using actograms and Chi² periodograms created with the Analyze9 software package (Stanford Software Systems, Santa Cruz, CA) on the final 14 days before cull. Using the onset of wheel-running activity as a phase marker (circadian time [CT] 12 was defined as the onset of the daily activity bout), DD mice were culled at times spanning the circadian cycle to allow assessment of the effect of culture preparation time on the phase of peak PER2::LUC expression. Vipr2^−/− mice that did not express a significant circadian rhythm were culled at arbitrary timepoints. Mice were classified as either rhythmic or expressing multiple low power rhythmic components (arrhythmic) according to previously defined criteria [56].

Mice were culled by cervical dislocation following isoflurane anaesthesia (Baxter Healthcare Ltd, Norfolk, UK). SCN, Arc/ME/PT complex (combined) and DMH were micro-dissected and cultured as 300 µm thick coronal slices (cut from a consistent rostro-caudal level for each tissue, based on neuroanatomical landmarks and the Paxinos and Franklin mouse brain atlas [57] as previously described [47], [58]. Pituitaries were removed by hand and cultured whole, under identical conditions. Night-vision goggles were used during DD culls to maintain darkness until mice had been euthanased and enucleated.

Brain and pituitary cultures were maintained at 37°C in light-tight incubators (Galaxy R+, RS Biotech, Irvine, Scotland) and total PER2::LUC bioluminescence emission recorded for at least 7 days using photomultiplier tube (PMT) assemblies (H8259/R7518P; Hamamatsu, Welwyn Garden City, UK). Emitted photon counts were integrated for 59 s every 1 min and raw bioluminescence data were processed by subtracting a 24 h running mean to remove long term trends (baseline-subtracted) then smoothed with a 3 h running average. The longitudinal study design employed here allows sensitive identification of low amplitude rhythms in individual animals, such as those of Vipr2^−/− mice. Discontinuous sampling methods, which assess population level trends across a number of individuals, can fail to detect significant variation when individuals are not synchronized to one another or peak-trough amplitude is low [13], [17], [59].

To investigate the effects of gastrin releasing peptide (GRP) and adenylate cyclase signaling on circadian expression of PER2::LUC in WT and Vipr2^−/− tissues, cultures were treated with either 100 nM GRP (Tocris Bioscience, Bristol, UK) or 10 µM forskolin (an adenylate cyclase activator; Sigma, Poole, UK) 3–8 days following culture preparation. Drugs were administered as complete medium changes to fresh culture medium containing the drug but otherwise identical to control medium.

Rhythmic bioluminescence traces were assessed by two experienced, independent researchers, blinded to conditions, to extract the following parameters: period, amplitude and phase of peak bioluminescence expression. Period was assessed using peak-peak and trough-trough durations averaged over as many cycles as possible for each individual tissue explant, discounting the first 24 h of data. At least one full peak-peak or trough-trough cycle was assessed of each explant, though in the majority of cases two or more full cycles were used. Amplitude was measured as the peak-trough difference 24–48 h after culture preparation from baseline-subtracted traces and phase was assessed as the time of the first peak to occur between 24–48 h after culture preparation. Further, the rate of damping of PER2::LUC bioluminescence rhythms was assessed as the number of cycles before smoothed bioluminescence variation reached the amplitude of dark noise (previously determined for each PMT module). The projected rate of damping was calculated for cultures that showed obvious damping of their bioluminescence rhythm but had not fully damped by the end of data acquisition. Statistically significant differences between genotypes were determined using unpaired t-tests performed in Microsoft Excel (p<0.05 required for significance). Rayleigh vector plots (custom software designed in house by Dr T. Brown and El Temps, Dr. A. Díez-Noguera, Barcelona, Spain) were used to assess the significance of phase clustering of peak phases in relation to circadian time (CT) or to time from culture preparation. Responses to forskolin and GRP treatments were assessed by comparison of amplitude 24 h before and 24 h after treatments, not including treatment artifacts. Statistical significance was determined by comparison of amplitude changes between control and drug treatments (2-way analysis of variance with a priori pairwise comparisons).

mPER2^luc (WT) mice behaved in a manner consistent with previous reports, both for this strain [46] and non-mPER2^luc WT mice (e.g. [20], [60], [61]). All WT mice (n = 9) entrained to the LD cycle, confining activity to the dark phase, and all free ran with a strong circadian rhythm in DD (mean period 23.82±0.05 h and rhythm strength (%V) of 58.51±5.34%), starting from the time of activity onset under LD (Fig. 1A–1B).

Circadian locomotor activity of Vipr2^−/− mice was not altered by the PER2::LUC reporter transgene; mice expressed locomotor activity rhythms consistent with previous descriptions for non-mPER2^luc Vipr2^−/− mice [20], [62]. Briefly, Vipr2^−/− mice (n = 13) limited their activity to the dark period of the LD cycle and on release into DD began activity almost immediately, defining a large phase advance from the timing of LD activity (Fig. 1C, 1E). In DD, Vipr2^−/− mice expressed the continuum of behavioral phenotypes common for this genotype, from robustly rhythmic with a single dominant component of locomotor behavior (Fig. 1C–1D), to apparent arrhythmicity, often with multiple, low power periodic components (Fig. 1E–1F). Approximately 50% (6/13) of Vipr2^−/− individuals were classified as rhythmic in DD expressing a mean period of 22.47±0.63 h and rhythm strength of 24.14±3.70%. Both period and rhythm strength of Vipr2^−/− mice were significantly different to those of WTs (p<0.05 and p<0.0001, respectively).

All SCN cultures recorded during this study expressed strong circadian rhythms of PER2::LUC bioluminescence that remained robustly rhythmic for the duration of recording (at least 7 days in vitro; Fig. 2A). The period of rhythms expressed by Vipr2^−/− SCN cultures did not differ from WT SCN rhythms, in cultures prepared from mice under either LD or DD conditions (both p>0.05; Table 1). The amplitude of Vipr2^−/− SCN rhythms was, however, significantly lower than that of WT SCN in cultures from mice housed in both lighting conditions (p<0.01 and p<0.05 for LD and DD respectively; Table 1; Fig. 2A). No significant difference was found in either period or amplitude between behaviorally screened rhythmic and arrhythmic Vipr2^−/− mice in DD (p>0.05; Fig. 2A).

Consistent with earlier work [44], [46], [47], Arc/ME/PT, DMH, and pituitary cultures from WT mice showed significant circadian oscillations in PER2::LUC bioluminescence (Fig. 2B–2D). Examination of period, amplitude and rate of damping revealed no significant differences between the rhythms of WT and Vipr2^−/− mice expressed in the Arc/ME/PT, DMH or pituitary gland, in tissue collected from mice housed in either LD or DD (Table 1; all comparisons p>0.05; Fig. 2B–2D). Further, no significant differences were found in any of these tissues between the rhythms expressed by rhythmic and arrhythmic Vipr2^−/− mice (all p>0.05; Fig. 2B–2D).

The phase of peak expression of PER2::LUC was only assessed for cultures collected from mice housed in DD as these individuals were culled at a wide range of times throughout the circadian cycle. This allowed assessment of whether the peak phase of different tissues from WT and Vipr2^−/− mice was associated with a particular CT phase or was reset by cull/culture procedure and expression peaked the same number of hours after cull regardless of the CT phase of cull.

Overall, WT SCN peak phase was significantly associated with CT (p<0.00001; Fig. 3A); rhythms peaked at a mean phase of CT12.9±0.5, with a slightly earlier phase (∼CT11–12) observed in cultures from mice culled during the middle of the circadian day, and a slightly later phase (∼CT14) observed in cultures prepared at other times. Vipr2^−/− SCN, however, were reset by cull/culture procedure and PER2::LUC rhythms consistently peaked at ∼31 h after cull (30.7±0.3 h; SCN peak phases from behaviorally rhythmic (30.8±0.5 h) and arrhythmic (30.6±0.4 h) Vipr2^−/− mice combined), regardless of the circadian phase of cull and culture preparation. Indeed, Rayleigh analysis revealed Vipr2^−/− SCN peak phase from rhythmic individuals to correlate significantly with time of cull (p<0.0001) and not CT (Fig. 3A). Behaviorally arrhythmic Vipr2^−/− mice were not included in these Rayleigh analyses as CT phase could not be calculated for these individuals, however, a separate analysis of the peak phase for arrhythmic Vipr2^−/− SCN cultures also revealed a significant association with time from cull (Fig. 3A; p<0.0001).

Similarly to SCN cultures, WT Arc/ME/PT expressed peak levels of PER2::LUC at a consistent circadian phase (CT19.9±0.7; Rayleigh correlation with CT: p<0.0001) and were not reset by cull/culture procedure (Fig. 3B). Vipr2^−/− Arc/ME/PT cultures, as for Vipr2^−/− SCN, were reset by cull/culture and peaked 31.8±0.7 h following cull (behaviorally rhythmic (32.5±0.9 h) and arrhythmic (31.1 h±1.4 h) individuals combined). Vipr2^−/− Arc/ME/PT peak phase for both rhythmic (p<0.005) and arrhythmic (p<0.05) mice correlated significantly with time from cull (Fig. 3B). While pituitary cultures from WT mice consistently expressed peak levels of PER2::LUC expression at CT17.7±1.4 h (Rayleigh correlation with CT: p<0.01), the phase of Vipr2^−/− pituitaries differed widely between individuals and was not consistently associated with either a particular circadian phase or duration of time following cull (Rayleigh associations with CT (for behaviorally rhythmic individuals only) and cull time (rhythmic and arrhythmic individuals separately), all p>0.05). DMH tissue from both WT and Vipr2^−/− mice reset to ∼30 h after cull (29.7±1.2 h and 30.1±0.5 h, respectively), regardless of the CT cull phase or behavioral rhythmicity of the animal (Rayleigh correlation with time from cull p<0.05, p<0.005 and p<0.005, respectively for WT, behaviorally rhythmic and behaviorally arrhythmic Vipr2^−/− mice).

To investigate the health of tissues after rhythms had damped, cultures were treated with the adenylate cyclase activator, forskolin (10 µM), previously shown to boost rhythms in cultured circadian oscillator tissues [44]. Forskolin treatment induced robust increases in rhythm amplitude in tissues that previously showed either a circadian deficit due to altered VIP-VPAC₂ signaling (Vipr2^−/− SCN; p<0.005; Table 2), or demonstrated rapid damping of oscillations (WT and Vipr2^−/− Arc/ME/PT, p<0.005 and p<0.0005, respectively; WT and Vipr2^−/− DMH, p<0.01 and p<0.005, respectively; Table 2). More robustly rhythmic tissues (WT SCN, WT and Vipr2^−/− pituitary gland) did not show significant increases in rhythm amplitude following forskolin treatment (Table 2), presumably as these tissues had not yet significantly damped at the time of treatment and/or did not suffer any inherent rhythm abnormalities associated with a lack of VPAC₂ receptors.

GRP treatment to Vipr2^−/− SCN tissue in vitro has been shown to boost and resynchronize rhythms [27], [29] and GRP receptors are expressed in the pituitary gland and arcuate nucleus [63], [64]. We, therefore, investigated whether GRP treatments would induce similar responses from oscillators in the MBH and pituitary. Surprisingly, neither WT nor Vipr2^−/− SCN rhythms were markedly improved by treatment with 100 nM GRP (both p>0.05; Table 2, Fig. 4A). The amplitude of Arc/ME/PT rhythms, however, was increased by treatment with GRP (p<0.05 for both WT and Vipr2^−/− tissue; Table 2, Fig. 4B). Interestingly, though GRP receptor expression has not been reported in the DMH, WT DMH cultures showed an increase in rhythm amplitude in response to GRP treatment (p<0.05; Table 2, Fig. 4C) while Vipr2^−/− DMH cultures failed to do so. The amplitude change of rhythms expressed by WT pituitaries in response to GRP was not significantly different to control treatments (Table 2, Fig. 4D), however, this lack of significance resulted from control treatments inducing an increase in amplitude in 3 of 5 cultures. Indeed, GRP treatments did induce a significant increase in amplitude in WT pituitary cultures (post-treatment amplitude vs. pre-treatment amplitude p<0.05; Table 2, Fig. 4D) and the mean increase in amplitude in response to GRP treatment was 665% of the mean control treatment-induced response. Conversely, neither GRP nor control treatment increased the amplitude of PER2::LUC expression in Vipr2^−/− pituitaries (p>0.05; Table 2, Fig. 4D).

While the SCN innervates the MBH region [53], endogenous rhythms in clock gene expression are present in vitro in the MBH and pituitary gland [44], [46], [47], [65], tissues critical for the regulation of metabolism [49]. Signaling via the VPAC₂ receptor is essential for the generation of high amplitude synchronized SCN clock gene rhythms [13], [26], [27] and for appropriate temporal control of behavior and metabolism [18], [20]. Here, we demonstrate that, despite known alterations of circadian function in Vipr2^−/− mice, the lack of VPAC₂ receptors does not alter period, amplitude or rate of damping of rhythms in PER2 expression in the Vipr2^−/− MBH and pituitary. These rhythms were observed in slice cultures from mice housed in LD and persisted in tissues collected from mice housed in DD, despite the behavioral arrhythmicity of ∼50% of Vipr2^−/− mice in constant conditions. This study provides the first description of extra-SCN neural oscillators in mice lacking either VIP or VPAC₂, while our demonstration of rhythm maintenance in the Vipr2^−/− pituitary gland concurs with a previous report of peripheral oscillations in the liver and heart of these mice [66].

Consistent with previous investigations, both in our laboratory and elsewhere, Vipr2^−/− SCN were found to express significantly lower amplitude clock gene oscillations than WT SCN [25]–[27]. In contrast, the MBH and pituitary do not show this decrease in amplitude. Rather more surprisingly, we found an alteration in the ability of Vipr2^−/− SCN tissue to maintain a consistent circadian phase when cultured in vitro. WT SCN cultures expressed peak levels of PER2::LUC expression at a predictable projected CT phase, consistent with previous in vivo and in vitro investigations of per2/PER2 expression [47], [67]–[71]. Further, the earlier peak phase we observed for WT cultures prepared during the circadian day is consistent with a small phase advance of rhythms in cultures prepared at this time [58], [72].

Conversely, the rhythms of SCN cultures from Vipr2^−/− mice were reset by cull/culture procedure and consistently peaked at the same time after culture preparation, regardless of the circadian phase at which this procedure was performed, illustrating the critical role of VIP-VPAC₂ signaling in the maintenance of robust SCN function. These data suggest that while the strong rhythm generating properties of the WT SCN, at the unicellular and network levels, are sufficient to maintain in vivo phase, the diminished circadian capabilities of the relatively weak and disorganized Vipr2^−/− SCN are unable to resist the phase-shifting influences associated with culture preparation. This presents an interesting parallel with a recent description of in vitro PER2::LUC phase in embryonic mouse liver [73]. Here, the authors find the phase of embryonic liver cultures to be determined by cull/culture time, while the phase of PER2::LUC rhythms in maternal liver is determined by external Zeitgeber time.

The tendency for stronger oscillators to maintain phase and weaker oscillators to reset to cull time is also seen across the Arc/ME/PT, DMH and pituitary oscillators. In WT tissue, the stronger of the MBH areas investigated here, the Arc/ME/PT, maintains an in vitro phase similar to a prior report of the expression of per2 in these areas [74], whereas the weaker oscillator contained in the DMH is reset by cull/culture. Similarly, WT pituitary, which contains a strong oscillator of comparable strength and robustness to the WT SCN (based on rhythm amplitude and a lack of damping), maintains a consistent in vitro CT phase of peak PER2::LUC expression. Vipr2^−/− pituitary, however, appears to neither maintain a steady phase predicted by locomotor rhythms, nor fully reset according to cull/culture time. This lack of resetting of the Vipr2^−/− pituitary differs from the properties of the Vipr2^−/− SCN and indicates a reduced importance of VPAC₂ signaling for the maintenance of pituitary rhythms – supported by the absence of any significant alteration to period, amplitude or rate of damping in Vipr2^−/− pituitary.

While the phase of Vipr2^−/− DMH responds to culturing in a similar fashion to WT DMH tissue, it is intriguing that WT and Vipr2^−/− Arc/ME/PT and pituitary are differentially sensitive to the phase-resetting stimuli associated to culture preparation. This inability of Vipr2^−/− Arc/ME/PT and pituitary to maintain an in vitro phase that correlates with the phase of behavioral rhythms is surprising given that these tissues show no alteration in period, amplitude or rate of damping of PER2::LUC expression. This raises the possibility that VPAC₂ signaling within these oscillators impacts on other aspects of rhythm strength/robustness that we have not detected here. Further, bioluminescence imaging studies and microdissections of the individual rhythmic components of the Arc/ME/PT complex (Arc, ME, PT and the ependymal cell layer of the 3^rd ventricle; [47]) will be necessary to elucidate the relative contribution of these oscillators to both rhythmicity of the Arc/ME/PT complex in Vipr2^−/− mice and its increased sensitivity to phase-resetting compared to WT tissue. The increased sensitivity of Vipr2^−/− Arc/ME/PT and pituitary oscillators to stimuli associated with culture preparation presents an intriguing parallel with the increased sensitivity of mice lacking VIP-VPAC₂ signaling to photic stimuli [17], [75].

Responses to GRP treatment were variable, with no consistent genotype differences across the tissues examined, or tissue differences between genotypes. Surprisingly, neither WT nor Vipr2^−/− SCN cultures responded to GRP treatment with an increase in the amplitude of PER2 expression, as has been shown previously [27], [29]. These studies, however, differ in crucial technical details to the current investigation, measuring electrical activity with chronic GRP infusion and measuring per1 expression in neonatal tissue, respectively. Indeed, the failure of GRP to boost rhythm amplitude of clock gene expression in vitro has previously been reported in cultured extra-SCN brain tissue [76]. Our observation of increased WT DMH rhythm amplitude following GRP treatment is of note given the lack of previous reports of GRP binding sites/GRP receptor expression in the DMH. This response is presumably mediated by previously undetected bombesin-like peptide receptors [77] in this area.

The data presented here demonstrate that SCN and extra-SCN circadian oscillations are maintained in the absence of VIP-VPAC₂ signaling. However, the Vipr2^−/− tissues assessed here are differentially capable, compared to WT tissues, of maintaining a consistent circadian phase when cultured in vitro. Most surprisingly, the phase of PER2 oscillations in Vipr2^−/− SCN is reset by culture preparation. Despite the maintenance of period, amplitude and rate of damping in the Arc/ME/PT, DMH and pituitary, the differential resetting effects of culture preparation on WT and Vipr2^−/− phase in these areas presumably reflects circadian deficiencies not assessed here which may contribute towards the aberrant behavior and metabolism associated with Vipr2^−/− animals.