Targeting of the circadian clock via CK1δ/ε to improve glucose homeostasis in obesity

Experimental procedures were licensed under the 1986 Home Office Animal Procedures Act (UK), and approved by The University of Manchester animal welfare committee guidelines. All experimental procedures were carried out in accordance with the above licensing and guidelines. Male C57BL/6J andmice were purchased from Charles River (UK) and Harlan (UK) respectively. mPER2::luc transgenic micewere kindly provided by Joe Takahashi (University of Texas Southwestern) and subsequently bred locally. All animals were maintained in 12 h:12 h light:dark (LD) with an ambient temperature of 20–22 °C, with food and water supplied. ob/ob ad libitum ³⁴

For DIO studies, 8–10 week old mice were providedaccess to high fat diet (HFD; 60% energy from fat; DIO Rodent Purified Diet, IPS Ltd) or normal chow (NC) for 2, 8, or 16 weeks. Prior to tissue collection, mice were placed in constant darkness (DD) and tissues collected every 4 h (n = 5–10/time-point/diet condition). In selected experiments, body temperature (T) and locomotor activity were recorded using radiotelemetry (Data Sciences International), and metabolic gas exchange measured by indirect calorimetry using the CLAMS system (Columbus Instruments). Food intake was monitored using the Labmaster Metabolism Research Platform (TSE systems), with meal size and feeding events (see) recorded over 6 days. Forstudies, mice were maintained on NC throughout, and locomotor activity was recorded in home cages using infrared beam-break sensors. ad libitum ob/ob b ³⁵ Supplementary Fig. S1

DIO mice (16 weeks of HFD) and matched NC-fed controls were injected daily with vehicle (20% 2-hydroxypropyl β-cyclodextrin; Sigma) or the selective inhibitor of CK1δ/ε, PF-5006739 (10 mg/kg/day, s.c.) at ZT10. This dose of PF-5006739 and timing of administration was based on previous work, and selected to achieve a CNS target occupancy above 50% for CK1δ/ε(). After 3 weeks of drug treatment, mice were fasted for 6 h before administration of glucose (1 g/kg i.p., n = 7–10/group). Blood glucose was monitored with an Accu-chek Aviva glucose meter (Roche), and serum samples collected at 0 and 30 min post-glucose administration. Mice were then maintained on PF-5006739/vehicle dosing for a further 14d prior to tissue collection. Forstudies, mutant and control mice (8 weeks of age) were dosed PF-5006739 as described above for 14d, following which a GTT was performed (20 h fast followed by 1 g/kg glucose i.p). Serum insulin and adiponectin levels were determined by ELISA (Millipore and R&D Systems, respectively). ³⁶ Supplementary Table S1 ob/ob

hybridisation was performed as previously described. Briefly, brains were rapidly dissected and frozen. Coronal sections (12 μm) were collected using a cryostat freezing microtome, and stored at −80 °C until processing. Theprobe was kindly provided by Prof. Urs Albrecht (University of Fribourg).andprobes (primers listed in) were cloned into pGEM-T easy. All probes were synthesized in the presence ofP-uridine triphosphate (MP Biomedicals). Hybridization was visualized by film autoradiography (Kodak BioMax MR film), and OD determined using 3–4 sections per mouse, and 5 mice per time-point/group. In situ Per1 Bmal1 Rev-erbα ³⁷ Supplementary Table S2 33

Peripheral tissues were rapidly dissected, snap frozen, and stored at −80 °C until use. Hypothalamic blocks were micro-dissected from 300 μm coronal brain slices. RNA extraction and qPCR was performed as previously described(qPCR primers listed in). For protein extraction, frozen tissues were homogenised in Tissue-Protein Extraction Reagent (Peirce) containing Protease Inhibitor Cocktail (Roche) and 1 mM PMSF. Protein levels were determined by SDS-PAGE and Western blot analysis using anti-PPARγ (C26H12, Cell Signalling, 1:500) and anti-GAPDH (FL-335, Santa Cruz Biotechnology, 1:1000) antibodies. ³⁸ Supplementary Table S2

Foranalyses of circadian rhythms, 8 week old male mPER2:luc mice were fed HFD or NC for 16 weeks. mPER2-dependent bioluminescence was recorded from gWAT samples maintained at 37 °C using a Lumicycle (Actimetrics) as described previously. Amplitude and phase of the second peak post-culture was determined for each animal (n = 4 mice/condition; 16 samples/mouse). To determine the effect of PF-5006739 on mPER2::luc expression, cultured gWAT were treated with vehicle (water) or PF-5006739 (0.08–50 μM), 5d post-culture. Amplitude was measured at the first peak post-treatment using Lumicycle recording, or bioluminescence pre- and 12 hr post-treatment using a GloMax-Multi Luminometer (Promega). ex vivo ³³

Data are presented as mean ± standard error (SEM). Students’ t-test was used when two groups were tested, one-way ANOVA with Dunnett’s multiple comparisons or two-way ANOVA followed by Sidak post-hoc analyses were used when more than two groups and/or factors were analysed. Statistical determination of acrophase was performed by linear harmonic regression using CircWave v1.4 software. ³⁹

To assess the impact of diet-induced obesity (DIO) on behavioural and physiological rhythms, adult male C57BL/6J mice were maintained on NC or HFD (60% energy from fat) for 16wk, leading to pronounced weight gain and WAT accumulation (). Diurnal rhythms in locomotor activity and body temperature (T) were maintained in HFD-fed and control mice throughout the 16wk study (). However, a reduction in the amplitude of activity and Tbetween light-dark periods was observed in the obese mice, due principally to a reduction in night-time activity and increased day-time T. Assessment of metabolic gas exchange at the end of the study period (16wk HFD) showed a significant reduction in overall rates of oxygen consumption (VO, L/kg/h) and increased rates of energy consumption (kcal/h) in DIO mice (). In line with activity and Tmeasures, diurnal rhythmicity in metabolic gas exchange was maintained, but with a significant reduction in diurnal amplitude. Increased daytime feeding is commonly reported in HFD-fed mice, and we therefore assessed diurnal feeding structure in mice fed NC or HFD. The proportion of calories consumed during the light phase of the cycle was increased upon HFD feeding (). However, this reflected a decrease in food intake during the dark phase, in terms of overall consumption and meal frequency, rather than increased feeding in the day (). Notably, altered nocturnal feeding was observed within days of the mice being placed on HFD (), likely reflecting a homeostatic response to the increased calorific and fat content of the diet, rather than an obesity-dependent modulation of circadian feeding rhythms. Supplementary Fig. S1 Fig. 1a–c Fig. 1d Supplementary Fig. S1 Fig. 1e,f Supplementary Fig. S1 b b b 2 b

Although these studies demonstrate that diet-induced obesity did not overtly disrupt diurnal rhythmicity in the mice or their ability to entrain to the LD cycle, the relative amplitudes of physiological measures were reduced in HFD-fed mice when compared to NC-fed controls.

We next assessed the impact of obesity on clock gene rhythms in a panel of central and peripheral tissues collected from mice fed HFD or NC for 16 weeks. Gene expression analyses of micro-dissected hypothalamic blocks did not reveal any significant alteration in expression of the core clock genesand(). As these blocks contain a number of oscillatory sites aside from the SCN (e.g. arcuate and dorsomedial nuclei), we employedhybridisation to assess clock gene rhythms in the SCN.hybridization analyses revealed a significant damping ofexpression in the SCN of DIO mice when compared to matched NC-fed controls (). SCN expression ofremained rhythmic in the DIO mice, yet exhibited a reduced level of expression relative to NC-fed mice at all time-points and reduced amplitude overall. Despite the damping ofexpression,andexpression within the SCN was similar between DIO and control mice, although both genes exhibited a small advance in acrophase within the DIO mice (,). Bmal1, Clock, Per1, Per2, Cry1, Rev-erbα Rev-erbβ in situ In situ Bmal1 Bmal1 Bmal1 Per1 Rev-erbα Fig. 2a Fig. 2b Fig. 2a Supplementary Fig. S2

In peripheral tissues, analysis of circadian clock gene expression in HFD- and NC-fed mice revealed that robust rhythms were maintained in most tissues examined (,). However, a pronounced alteration of clock gene expression was observed in perigonadal WAT (gWAT). Interestingly, clock gene expression was relatively unaffected in scWAT of obese mice, suggesting that reduced amplitude in gWAT was not due simply to increased expansion and triglyceride storage within the adipose tissues. The differential impact of obesity on clock gene expression in the gonadal and subcutaneous adipose depots was most pronounced inandgene expression profiles. The damping of gWAT rhythms in response to chronic HFD-feeding was maintained, in tissue explants derived from 16wk HFD-fed mPER2::luciferase reporter mice, when compared to tissue derived from NC-fed matched controls (). In addition to damping of gWAT clock gene expression, evidence of altered phase alignment between tissue clocks was also observed in HFD-fed mice. Specifically, a phase advance of transcriptional rhythms was observed in the liver and adrenal gland of HFD-fed mice (~2.0 h phase advance in HFD relative to NC-fed, based on acrophase of the 8 clock genes profiled;), but not other tissues such as skeletal muscle and BAT, indicating reduced synchrony among tissue oscillators. Throughout the peripheral tissue panel, the most profoundly and consistently affected transcript was, which showed no significant time of day difference in expression across the circadian cycle in gWAT of HFD-fed mice (). Interestingly, we did not observe a similar effect in the related gene,, which was largely unaffected by HFD feeding across the eight peripheral tissues examined in this study (). This highlights the differential regulation of the two related nuclear hormone receptors. Fig. 3 Supplementary Fig. S2 Supplementary Fig. S2 Supplementary Table S3 Fig. 3 Supplementary Fig. S2 Rev-erbα, Per1 Per2 ex vivo Rev-erbα Rev-erbβ

Overall, these studies demonstrate a pronounced effect of chronic HFD feeding on expression and synchrony of the molecular clockwork, and highlight the gene- and tissue-specificity of these effects.

To further assess the impact of HFD-feeding on clock gene expression, mice were maintained on NC or HFD for 2, 8, or 16 wk (,). Importantly, damping ofandexpression in gWAT was not observed following 2wk of HFD-feeding, and was most pronounced in mice that had been maintained on HFD for 16wk (). Therefore, damping of clock gene rhythms in gWAT of obese mice was not due to an acute effect of the diet. In animals and humans, obesity-related insulin resistance has been linked to increased immune cell infiltration and pro-inflammatory cytokine production within hypertrophic WAT. We therefore profiled pro-inflammatory markers in the tissues of NC and HFD-fed mice. A profound elevation of gWAT inflammation was observed in 16wk HFD-fed mice compared with matched mice fed NC, and those fed HFD for 2wk or 8wk (). At 16wk of HFD-feeding, expressions of the pro-inflammatory cytokinesand, as well as the macrophage markerwere most profoundly elevated in gWAT (), suggesting that clock gene damping within this tissue may be linked to the development of obesity-related inflammation. Fig. 4a–c Supplementary Fig. S3 Fig. 4a ⁴⁰ Fig. 4b Supplementary Fig. S3 Bmal1 Rev-erbα Ccl2 TNFα F4/80

Components of the circadian clock are also closely and reciprocally linked with the metabolic regulators, PPARα and PPARγ. We therefore examined tissue-specific and temporal dynamics of PPARα/γ expression in mice maintained on HFD (). Similar to the attenuation of clock gene rhythms, a significant reduction inexpressions were observed in gWAT of mice maintained on HFD for 16wk (but not 2wk or 8wk;). Obesity-related changes inandexpression were highly tissue specific, with the expressions of both receptors significantly increased in the liver, yet profoundly attenuated in gWAT of obese mice when compared with matched NC-fed controls (). In contrast to gWAT,andexpressions were not reduced in scWAT (). Tissue-specific changes inmRNA expression in liver and gWAT were mirrored in PPARγ protein expression (), as well as in PPARtarget genes includingand(,), and genes involved in lipid metabolism known to be under circadian control (). ^{5 19 41} Fig. 4c–f Fig. 4c Fig. 4d Fig. 4d Fig. 4e,f Fig. 4c Supplementary Fig. S3 Supplementary Fig. S3 PPARα/γ PPARα PPARγ PPARα PPARγ PPARγ γ Adipoq Fabp4

These studies reveal a pronounced tissue-specific nature of obesity-related clock dysfunction, with gWAT being particularly affected. However, our findings also highlight the fact that damping of the clock in gWAT is accompanied (in terms of duration of HFD-feeding and tissue-specificity) by elevated tissue inflammation and altered/expression. PPARα γ

We have previously shown that daily administration of CK1δ/ε inhibitors can be effective at increasing the amplitude of disrupted and weak oscillators. We therefore assessed whether this approach would be effective in DIO mice using a novel CK1δ/ε inhibitor (PF-5006739), which exhibits improved target selectivity over previous compounds. The impact of CK1δ/ε inhibition on damped rhythms was first examined using gWAT tissue cultures derived from mPER2:Luc mice which had been maintained on NC or HFD for 16wk (). WAT tissue explants were maintained in culture for 5 days, and then treated with vehicle or PF-5006739 (0.4–50 uM). In gWAT tissue derived from either NC or HFD-fed mice, PF-5006739 administration caused a rapid and dose-dependent induction of mPER2::luc bioluminescence (). ^{33 36} Fig. 5a,b Fig. 5a,b

We next tested whether pharmacological targeting of CK1δ/ε could provide a metabolic benefitin models of obesity. DIO mice (16wk HFD-feeding) were treated once daily at ZT10 with PF-5006739 (10 mg/kg/day). Dose and timing of administration were based on prior characterisation of the compound. Following 3 weeks of treatment, no significant impact of PF-5006739 was observed in NC or HFD-fed mice (relative to vehicle-treated mice for the same diet group) with respect to body weight, food intake, day/night feeding behaviour, or diurnal activity profile (,). However, daily administration of the CK1δ/ε inhibitor did lead to a significant improvement in glucose tolerance in response to bolus glucose challenge (glucose tolerance test; GTT) (n = 7/group;). Damping of circadian gene rhythms in peripheral tissues has been reported in leptin-deficientmice, a genetic model of obesity. We therefore tested the ability of PF-5006739 administration to improve glucose tolerance in this model. Similar to studies in DIO mice, PF-5006739 treatment (10 mg/kg/day, ZT10) did not affect body weight, feeding behaviour, or activity profiles inmice, relative to vehicle-treatedmice (,). Yet once again, daily administration of PF-5006739 significantly improved blood glucose profiles during GTT (). No difference in glucose clearance was observed in lean (NC-fed) C57Bl/6J mice treated with PF-5006739 relative to vehicle-treated mice (). in vivo ob/ob ob/ob ob/ob ³⁶ Fig. 5c Supplementary Fig. S4 Fig. 5c ³⁰ Fig. 5d Supplementary Fig. S4 Fig. 5d Supplementary Fig. S4

Taken together these studies highlight the potential benefits of targeting CK1δ/ε to improve glucose homeostasis in obesity.