Deletion of astrocytic BMAL1 results in metabolic imbalance and shorter lifespan in mice

2.2. Locomotor activity

Two to three or 15 months after TM treatment, male Bmal1cKO and control mice were single‐housed in cages equipped with running wheels (ENV‐044; Med Associates, Inc). Mice were adapted to the wheel for 3 days in standard light–dark cycles (12:12 hr, lights on at 8 a.m., ZT0). Running wheel activity monitoring started under these conditions during the next 3–5 days. Running wheel activity was recorded in 5 min (min) bins by Wheel Manager software (SOF‐860; Med Associates, Inc) as previously reported (Barca‐Mayo et al., 2017).

2.3. Operant conditioning

Two months after TM treatment, male control and Bmal1cKO mice were subjected to a restricted feeding schedule where the access to standard lab chow (Special Diet Services, UK) was restricted from 16:00 to 18:00 (i.e., ZT8 and ZT10, respectively). Behavioral experiments started 3 days after the restricted feeding schedule and were conducted between 14:00 and 16:00. The behavioral training took place in operant chambers (17.8 cm × 15.2 cm × 18.4 cm) in which two holes were placed on either side of a food magazine (Med Associates, St. Albans, VT). Sucrose pellets (SP, TestDiet, Indianapolis, IN) were delivered into the food magazine when the mouse nose poked into the “active” hole (ANP), whereas a poke into the “inactive” hole had no consequence. Initial nose poke training consisted of three daily sessions of fixed ratio (FR) schedule (FR‐1), in which every active nose poke was rewarded. After the FR1 sessions, mice were trained on a FR‐5 and FR‐25 schedules, in which every fifth or 25th active nose poke led to SP delivery, respectively. After three daily FR‐5 sessions and three daily FR‐25 schedules, mice were trained under a progressive ratio (PR) schedule for 3 days. In the PR‐schedule, the number of ANP required to obtain SP is increased with each completed trial (ANP  =  5 × e^0.2SP) so each successive SP required more ANP and the amount of ANP reflected the effort that was invested in the task.

2.4. Glucose tolerance test

Control and Bmal1cKO mice (at the age and time point specified) were fasted for 16 hr before the glucose tolerance test. Blood was obtained from a tail cut and was assessed for fasting glucose levels using an OneTouch Ultra 2 (LifeScan, Johnson & Johnson) glucometer. Mice then received a glucose solution (2 g/kg body weight) delivered by intraperitoneal injection. At 15, 30, 60, 90, and 120 min after the administration, dried blood was quickly removed from the tail wound and fresh blood was collected again to measure the glucose concentration.

2.5. Determination of serum leptin, insulin, corticosterone, and glucagon

Leptin, insulin, corticosterone, and glucagon serum levels were determined by ELISA using reagents kits and methods provided by Merck‐Millipore (insulin, EZRMI‐13K; leptin, EZML‐82K), Enzo Life Sciences (corticosterone, ADI‐900‐097) and Sigma (Glucagon, RAB0202).

2.6. Determination of GABA/glutamate levels in cerebrospinal fluid

GABA was quantified in mouse cerebrospinal fluid (CSF) samples, collected from the cisterna magna, by UPLC‐MS/MS (Ultra Performance Liquid Chromatography–Tandem mass spectrometry) as we previously described (Barca‐Mayo et al., 2017). Glutamate was extracted from CSF by precipitation with acetonitrile spiked with deuterated Glutamate (D5) as an internal standard (Sigma Aldrich). The analytes were then separated by HILIC chromatography (Hydrophilic Interaction Liquid Chromatography) using a BEH HILIC 2.1X100 mm. column and a short gradient of water in acetonitrile (5–40% in 2 min), with the eluent added with formic acid to a final 0.1% v/v, flow rate was kept at 0.45 ml/min. Glutamate was quantified on a Xevo TQ‐MS instrument operating in electrospray, positive ion mode and following the MRM (Multiple Reaction Monitoring) transitions. Both the column and the UPLC‐MS/MS systems were purchased from Waters Inc. (Milford). Glutamate quantification was performed using a standard calibration curve prepared by serial dilution in artificial CSF and extracted along with the samples. The investigators were blinded to group allocation during experiments.

2.7. Immunofluorescence

Mice were administered ketamine/xylazine (150 mg/kg, 10 mg/kg, respectively) and transcardially perfused with ice‐cold PBS followed by ice‐cold 4% PFA in PBS. Brains were post‐fixed overnight in 4% PFA in PBS and 30 μm slices were prepared Cryostat (Leica). Slices were permeabilized with 0.3% Triton X‐100 in PBS, blocked with 10% goat serum in PBS, and incubated at 4°C overnight with the primary antibody mouse anti‐Glial Fibrillary Acidic Protein (GFAP) 1:1,000 dilution (Sigma, G3893); rabbit anti‐BMAL1, dilution 1:200 (Abcam, ab93806); mouse anti‐S100 calcium‐binding protein β (S100β), 1:1,000 dilution (Sigma, AMAB91038); rabbit anti‐active CASPASE 3, 1:1,000 dilution (Cell Signaling Technology, 9579); mouse anti‐RNA Binding Fox‐1 Homolog 2 (FOX2) 1:500 dilution (Abcam, ab57154); rabbit anti‐KI67 1:200 dilution (Thermo Scientific MA5‐14520). The following day, sections were extensively washed, and incubated for 2 hr with goat anti‐rabbit or anti‐mouse Alexa‐488 or Alexa‐546 secondary antibodies (1:1,000 dilution). Slices were then washed, mounted with Prolong Gold and imaged with an inverted laser scanning confocal microscope (TCS SP5 microscope using a 20× or 40× objective, Leica Microsystems). Quantification and analysis were performed with ImageJ software (Wayne Rasband, NIH), by outlining the hypothalamus from the 4′,6‐diamidino‐2‐phenylindole (DAPI)‐stained image and using this template to measure the relative intensity of the immunostaining. When more than one section was analyzed from each animal, the mean of the measures from consecutive sections was used for that individual.

2.8. RNA isolation and quantitative real‐time RT‐PCR

Total RNA was extracted from hypothalamus using TRIzol reagent following the manufacturer's instructions. RNA was further cleaned using an RNeasy Mini Kit. cDNA was obtained by reverse transcription of 0.5 μg of total mRNA using the ImProm‐II™ Reverse Transcription System following the manufacturer's instructions. Real‐time RT‐PCR was done using the ABI PRISM.7900 (Applied Biosystems). For a 15 μl reaction, 9 ng of cDNA template was mixed with the primers to a final concentration of 200 nM and mixed with 7.5 μl of 2× QuantiFast SYBR Green PCR Master Mix. The reactions were done in duplicates using the following conditions: 5 min at 95°C followed by 40 cycles of 10 s at 95°C, 30 s at 60°C, and 1 min at 70°C. Glyceraldehyde 3‐phosphate dehydrogenase (GAPDH) or beta‐actin transcripts were used as reference controls.

3.2. Bmal1cKO mice showed increased food intake with no alterations in the brain reward systems

Bodyweight results from the homeostatic regulation to balance energy intake and energy expenditure. We monitored, at different times after TM treatment, the daily food intake of control and mutant animals that were kept in normal light–dark cycles. We observed that 1 month after TM treatment, Bmal1cKO animals had comparable food intake than control mice (Figure ​(Figure2a).2a). However, from 2 to 6 months after TM treatment, Bmal1cKOs had significantly increased food intake than controls (Figure ​(Figure2a),2a), therefore preceding their increase in the body weight (observed 4 months after TM treatment, Figure ​Figure1b)1b) and their insulin resistance (detected 6 months after TM treatment, Figure ​Figure1f).1f). Twelve months after TM treatment, when the bodyweight of mutant mice was indistinguishable from control animals (Figure ​(Figure1b),1b), the mutants showed no differences in the food intake compared to controls (Figure ​(Figure2a).2a). On the other hand, we found no differences in the daily activity between control and Bmal1cKO mice 2 months after TM treatment (Figure ​(Figure2b),2b), consistent with our previous observations (Barca‐Mayo et al., 2017). However, at this pre‐obese stage, Bmal1cKO animals had abnormal feeding behavior as shown by their significantly increased food intake at both day and night times (Figure ​(Figure2c).2c). It was shown that the metabolic dysfunctions and obesity of mouse models with genetic disruptions in core clock genes such as Clock, are directly linked to disturbed feeding rhythms and/or excessive daytime feeding (Turek et al., 2005). Therefore, we postulate that the altered feeding pattern of Bmal1cKO mice at the pre‐obese stage results in obesity and insulin resistance at later stages.

Open in a separate window

Figure 2

Bmal1cKO mice showed increased food intake with no alterations in the brain reward systems. (a) Daily food intake was determined in control and Bmal1cKO mice after 1, 2, 3, 6, and 12 months of TM treatment. Animals were maintained in 12 hr:12 hr light–dark cycles and fed ad libitum with a standard mouse chow. Data are represented as mean ± SEM (n = 8, paired t‐test, **p < .01 vs. control animals). (b) Activity waveforms for control (n = 8) and Bmal1cKO (n = 7) mice, 2 months after TM treatment, in 12 hr: 12 hr light–dark cycles. Activity counts are expressed as the average amount of activity in 5 min bins. Data plotted is given in ZT, such that ZT0 = lights on. The value expresses the means + SEM. (c) Food intake was determined in control and Bmal1cKO mice after 2 months of TM treatment. Animals were maintained in 12 hr:12 hr light–dark cycles and fed ad libitum with a standard mouse chow. Data are represented as mean ± SEM (n = 8, paired t‐test, *p < .05 vs. control animals). (d) Number of nose spokes performed by control and Bmal1cKO mice, 2 months after TM treatment, during fixed ratio (FR), and progressive ratio (PR) sessions in the operant conditioning test. Data are represented as mean ± SEM (n = 4 for controls and n = 11 for Bmal1cKOs). (e) Mean ± SEM breakpoints in control (n = 4) and Bmal1cKO (n = 11) mice, 2 months after TM treatment, in the operant conditioning test [Color figure can be viewed at wileyonlinelibrary.com]

The main brain region regulating homeostatic food intake is the arcuate nucleus (ARC) of the hypothalamus through its connections with other hypothalamic nuclei and extra‐hypothalamic brain areas (reviewed in Coll, Farooqi, & O'Rahilly, 2007 and in Blouet & Schwartz, 2010; Dietrich & Horvath, 2009; van Vliet‐Ostaptchouk, Hofker, van der Schouw, Wijmenga, & Onl‐Moret, 2009). However, other brain circuits involved in the rewarding effects of food such as several limbic (nucleus accumbens, amygdala and hippocampus) and cortical brain regions (orbitofrontal cortex, cingulate gyrus, and insula) are also implicated in hedonic‐driven food consumption and obesity (reviewed in Coll et al., 2007 and in Belgardt, Okamura, & Brüning, 2009; Goldstone, 2006; Rolls, 2008). To address whether the increased food intake of the mutant was due to an alteration in hedonic system, we subjected to control and Bmal1cKO mice (after 2 months of TM treatment) to a progressive‐ratio schedule of food‐pellet reinforcement, a commonly used measure of reward strength (Hodos, 1961, 1963). Animals were trained to nose‐poke in an active hole to get a food reward under FR1, FR5, and FR25 schedules. No significant differences were detected between Bmal1cKO and control mice in the acquisition of the task (Figure ​(Figure2d).2d). Moreover, in the PR session, in which mice had to nose‐poke an increasing number of times to get the same reward (breakpoint), Bmal1cKO mice poked as control animals and earned the same rewards (Figure ​(Figure2e).2e). This result suggests that increased food intake in Bmal1cKO mice is due to an alteration in the hypothalamic homeostatic system and less likely to hedonic search for food.

Astrocytes have been recently postulated as key contributors to energy balance regulation and obesity due to the discovery of hypothalamic inflammation and gliosis, particularly in the ARC nucleus, in obese rodents and humans (reviewed in McNay, Briançon, Kokoeva, Maratos‐Flier, & Flier, 2012; Thaler et al., 2012; Valdearcos, Xu, & Koliwad, 2015). Remarkably, it was reported that BMAL1 is a potent regulator of astrocyte activation or gliosis (Lananna et al., 2018; Musiek et al., 2013). Therefore, we hypothesize that BMAL1 may connect dysregulation of circadian function to hypothalamic astrogliosis leading to the increased body weight and metabolic alteration of Bmal1cKO mice.

3.3. BMAL1 deletion in astrocytes globally impairs the hypothalamic molecular clock

To test our hypothesis, we crossed Bmal1cKO mice with a Cre‐inducible Red fluorescent reporter mouse line (Td‐Tomato), as we previously reported (Barca‐Mayo et al., 2017). We first ascertained the specificity and efficiency of GLAST‐Cre‐mediated BMAL1 deletion in hypothalamic astrocytes, by quantifying the co‐immunolocalization of TOMATO with the astrocyte markers Glial fibrillary acidic protein (GFAP) or S100 calcium‐binding protein β (S100β) and BMAL1 in Glast‐Cre‐Td‐Tomato (control) or Bmal1cKO‐Td‐Tomato animals. As expected, 2 months after TM, virtually all of the Cre‐recombined cells in the hypothalamus, as revealed by TOMATO, exhibited the stellated morphology characteristic of astrocytes, as well as immunoreactivity for GFAP and S100β (Table S2), consistently with previous observations (García‐Cáceres et al., 2016). BMAL1 was expressed in 48.5 ± 4.78% of TOMATO positive astrocytes in the ARC nucleus of control animals. In contrast, this proportion was reduced by approximately 60% in Bmal1cKO‐Td‐Tomato mice (paired t test, p = 8.39 × 10⁻⁶; Figure ​Figure3a,b).3a,b). Similarly, the percentage of GFAP or S100β positive astrocytes expressing BMAL1 was significantly reduced by 60.1 and 62.3%, respectively, in the ARC of Bmal1cKO‐Td‐Tomato mice (paired t test, p = 6.9 × 10⁻⁸ for GFAP and p = 8.88 × 10⁻⁵ for S100β), compared to controls (Figure ​(Figure33a).

Open in a separate window

Figure 3

BMAL1 deletion in astrocytes globally impairs the hypothalamic molecular clock. (a) Upper left panel, a reduction of BMAL1 positive cells was observed in the ARC nucleus of Bmal1cKO mice compared with control animals 2 months after TM treatment (Y‐axis represents the percentage of total BMAL1‐positive cells in the ARC nucleus). Data are represented as mean ± SEM (n = 4, paired t‐test, ***p < .001 vs. control animals). Upper right panel, a 60% reduction of BMAL1‐positive cells was observed in the population of Td‐TOMATO‐positive cells of Bmal1cKO compared with control animals (red, paired t‐test, ***p < .001 vs. control animals). A 36% reduction of BMAL1‐positive cells in the population of Td‐TOMATO‐negative cells was found in Bmal1cKO compared with control animals (green, paired t‐test ***p < .001 vs. control animals). Data are represented as mean ± SEM (n = 4). Percent of BMAL1‐positive cells was significantly reduced in GFAP (lower left panel) or S100β (lower right panel) positive astrocytes in the ARC nucleus of Bmal1cKO‐Td‐Tomato mice compared with control animals. Data are represented as mean ± SEM (n = 4, paired t‐test, ***p < .001 vs. control animals). (b) Representative micrographs of BMAL1 immunostaining in the ARC nucleus of control or Bmal1cKO‐Td‐Tomato animals. Scale bar, 25 μm. (c) Analysis of clock transcripts (Bmal1, Cry1, Per2 and BMAL1 target, Dbp) in the hypothalamus of control and Bmal1cKO mice after 2 months of TM treatment. Experimental data were cosine fitted. The ZT24 time point is the ZT0 time point, shown again. Data are represented as mean ± SEM (n = 5–6, paired t‐test, *p < .05 and **p < .01) [Color figure can be viewed at wileyonlinelibrary.com]

Remarkably, in the mutants, the percentage of BMAL1 positive cells was also reduced by 36% in TOMATO negative cells (paired t test, p = 9.27 × 10⁻⁵; Figure ​Figure3a,b),3a,b), suggesting that deletion of BMAL1 in GLAST positive astrocytes might also impact the clock in other cell populations of the hypothalamus, as we previously reported in cortex and hippocampus (Barca‐Mayo et al., 2017). To evaluate the impact of BMAL11 depletion on global oscillations in the hypothalamus, we quantified rhythmic expression of clock genes in controls and Bmal1cKOs, 2 months after TM treatment, at different ZTs. In control mice, we found rhythmic expression of Bmal1, Cry1, Per2, and BMAL1‐target Dbp, as previously reported (Barca‐Mayo et al., 2017) (Figure ​(Figure3c).3c). However, these oscillations were attenuated in Bmal1cKOs (Figure ​(Figure3c),3c), indicating that the deletion of BMAL1 in astrocytes globally impairs the molecular clock and, therefore, the circadian function of the hypothalamus.

3.4. Bmal1cKO mice showed age‐dependent astrogliosis and apoptosis of hypothalamic astrocytes

Next, we investigated whether the deregulation of the molecular clock in the brains of our mutants lead to reactive gliosis, as previously reported in cortex and hippocampus of constitutive Bmal1−/−, Nestin‐Cre‐Bmal1 or Aldh1l1‐CreERT2‐Bmal1 mice (Lananna et al., 2018; Musiek et al., 2013; Nakazato et al., 2017). Increased GFAP levels are generally regarded as hallmark of reactive glia, therefore we quantified Gfap transcripts or protein by qPCR or immunostaining, respectively, in different brain areas (cortical, hippocampal, and hypothalamic regions) of control and Bmal1cKO mice at 2 and 4 months after TM treatment. In agreement with previous reports (Lananna et al., 2018; Musiek et al., 2013; Nakazato et al., 2017), we did not detect the increased expression of Gfap in cortex or hippocampus 2 months after TM treatment (Figure ​(Figure4a,4a, Table S1). However, 4 months after TM treatment, we found increased GFAP immunoreactivity in different cortical regions and hippocampus of the mutants (Figure ​(Figure4b4b and S1), confirming that Bmal1cKOs developed age‐dependent gliosis. Surprisingly, already at 2 months after TM treatment, we found elevated expression of Gfap in the hypothalamus of Bmal1cKO compared to control mice (Figure ​(Figure4a).4a). This result suggests that BMAL1 deletion in astrocytes leads to a temporal activation of this glial cell type which differs among different brain areas and, importantly that the hypothalamus is, among the different brain regions analyzed, the first in showing reactive gliosis.

Open in a separate window

Figure 4

Bmal1cKO mice showed age‐dependent astrogliosis and apoptosis of hypothalamic astrocytes. (a) Gfap expression in cortex (Ctx), hippocampus (Hc) and hypothalamus (Hpt) of control and Bmal1cKO mice, 2 months after TM treatment. Data are represented as mean ± SEM (n = 5, paired t‐test, **p < .01). (b) Quantification of GFAP fluorescence intensity in cingulate (Cing), visual (Vis), piriform (Pir) cortex, Hpt and Hc of Bmal1cKOs, and controls, 4 months after TM treatment. Data are represented as mean ± SEM (n = 5, paired t‐test, **p < .01 and ****p < .0001 vs. controls). (c) Left panel, percentage of TOMATO‐positive cells in the ARC nucleus of control (Glast‐Cre‐Td‐Tomato) or Bmal1cKO‐Td‐Tomato animals, 2 months after TM treatment. Middle and right panel, percentage of active‐CASPASE 3 cells that co‐localized with TOMATO or FOX2 positive cells, respectively. Data are represented as mean ± SEM (n = 5, paired t‐test, *p < .05 and ***p < .001 vs. controls). (d) Representative micrographs of TOMATO and active CASPASE‐3 immunostaining in the ARC nucleus of control and Bmal1cKO‐Td‐Tomato mice 2 months after TM treatment. Scale bars, 50 and 25 μm in the higher magnification images. Cortex of Dgrc8 (DiGeorge Syndrome Critical Region Gene) knockout mice at embryonic day 13.5 was used as a positive control (right upper panel). (e) Percentage of GFAP (upper panel) or S100β (lower panel) positive cells that are TOMATO‐negative in the ARC nucleus of control and Bmal1cKO‐Td‐Tomato animals. Data are represented as mean ± SEM (n = 5, paired t‐test, *p < .05 vs. controls) [Color figure can be viewed at wileyonlinelibrary.com]

To gain further insights in this observation, we investigated whether reactivity was restricted to the subpopulation of GLAST positive astrocytes, in two different areas of the hypothalamus: The ARC nucleus, involved in the control of food intake and where inflammation and gliosis was linked to obesity in rodents and humans (reviewed in Valdearcos et al., 2015; McNay et al., 2012; Thaler et al., 2012) and the SCN, the main regulator of circadian locomotor activity in mammals (Ralph, Foster, Davis, & Menaker, 1990; Stephan & Zucker, 1972). Surprisingly, we found a 17% reduction of GLAST positive astrocytes (as revealed by TOMATO) in the ARC of Bmal1cKO‐Td‐Tomato mice compared to control animals, 2 months after TM treatment (Figure ​(Figure4c).4c). Moreover, we detected a significant decrease in the total number of cells per area, as determined by DAPI staining, in the ARC of the mutants 2 months after TM treatment (469.44 ± 39.99 for Bmal1cKOs vs. 585.14 ± 38.05 for controls, paired t test, p = .046). This decrease in the total cell numbers in the ARC of Bmal1cKO mice was not due to a reduction in the number of neurons, as shown by quantification of the neuronal marker RNA Binding Fox‐1 Homolog 2 (FOX2), a homolog of NeuN (Kim, Adelstein, & Kawamoto, 2009; Underwood, Boutz, Dougherty, Stoilov, & Black, 2005) (50.11% + 1.76 for controls and 53.49% + 2.64 for Bmal1cKOs, paired t‐test, p = .29). Indeed, in the ARC nucleus of Bmal1cKO‐Td‐Tomato mice we found a significant increase in the proportion of TOMATO‐positive astrocytes, but not in FOX2 positive neurons, that were also positive for the apoptosis marker active‐CASPASE‐3 (Figure ​(Figure4c,d).4c,d). Additionally, we observed a significant increase in the proportion of TOMATO negative cells that were positive for GFAP, while no differences were found in the expression of S100β among TOMATO positive or negative cells (Figure ​(Figure4e4e and S2). Altogether, these results indicate a selective loss of GLAST positive astrocytes upon deletion of BMAL1 (i.e., TOMATO positive cells), and reactivity in the astrocytes that retained BMAL1 expression (i.e., TOMATO‐negative cells).

As we found reduced astrocyte numbers and increased the reactivity of astrocytes in the ARC after 2 months of TM treatment, we evaluated whether Bmal1 deletion was maintained at later time, specifically at 15 months post tamoxifen treatment. We found that the percent of GFAP positive cells was increased in the ARC nucleus of both control and Bmal1cKO mice with age (Figure S3b) as previously reported (reviewed in Palmer & Ousman, 2018). However, at this stage, Bmal1cKOs showed a significant increase in the percent of GFAP positive cells compared to control animals (Figure S3a,b). On the other hand, the percentage of BMAL1 positive cells in the ARC of control animals was reduced at 15 months after TM treatment compared to 2 months posttreatment (Figure S3c). This observation is consistent with previous reports showing a decline in the clock gene expression with age (reviewed in Hood & Amir, 2017). Despite, at this stage, we observed a reduction in the percentage of total BMAL1 positive cells in the ARC of the mutants compared to control animals, it did not reach significance (two‐way ANOVA, p = .19; Figure S3a,c). However, co‐immunostaining of GFAP and BMAL1 indicated a significant reduction among the BMAL1 positive that were GFAP positive or negative in the mutants (Figure S3d). Therefore, these results suggest that BMAL1 deletion is maintained at later time points after TM treatment. Indeed, despite that we detected a significant increase in the percentage of GFAP‐positive cells that were mitotically active (by coimmunostaining with the proliferative marker, KI67) after 15 months of TM treatment in the ARC nucleus of the mutants as compared to control animals (0.5% vs. 0.17%, respectively), this percentages were very small (Figure S4) and likely biologically irrelevant. Consistent with a previous report in Bmal1nestin−/− mice, these data suggest that the deletion of Bmal1 does not increase the number of astrocytes but rather increases the activation of pre‐existing ones (Nakazato et al., 2017).

Remarkably, in the SCN, no differences between control and Bmal1cKO animals was observed in the percentage of TOMATO (17.84% ± 1.51 vs. 17.23% ± 1.12, respectively, paired t test, p = .74), GFAP (16.47% ± 0.92 vs. 19.15% ± 0.82, respectively, paired t test, p = .58) or FOX2 positive cells (8.38% ± 0.46 vs. 10.5% ± 2, respectively, paired t test, p = .63), 2 months after TM treatment. Consistently, no differences were found in the proportion of TOMATO negative cells that were positive for GFAP between control and Bmal1cKO animals (7.11% ± 0.94 vs. 7.10% ± 0.72, respectively, paired t‐test, p = .99) indicating that BMAL1 deletion in SCN astrocytes do not lead to reactive gliosis 2 months after TM. As wheel‐running activity is widely used as an index of SCN circadian function (Ralph et al., 1990; Stephan & Zucker, 1972), this observation is consistent with the comparable pattern of daily activity in control and mutant animals in 12 hr–12 hr light–dark cycles (Figure ​(Figure2b)2b) indicating that the SCN circadian function was not altered in Bmal1cKO mice 2 months after TM treatment. It was reported that the circadian output measured at the level of circadian locomotor behavior is dampened with age (Nakamura et al., 2011). Consistent with this report, we found that the daily activity of control animals, at 15 months after TM treatment, were reduced as compared to younger mice (2 months after TM treatment; Figures ​Figures2b2b and S5). Remarkably, the circadian locomotor activity of the mutants, at this stage, was reduced as compared to controls (Figure S5).

Our results indicate that BMAL1 deletion in astrocytes lead to a temporal‐ and regional‐specific gliosis within the hypothalamus and, importantly that ARC nucleus is, among the different brain regions analyzed, the first showing astrocyte reactivity and apoptosis. This cascade of events likely accounts for the altered hypothalamic function of Bmal1cKO mice, which might lead to the increased body weight and metabolic alterations of the mutants. Together, these results (i.e., shorter lifespan, glucose imbalance, altered growth curves, age‐dependent gliosis and decrease of age‐related circadian locomotor activity) support our hypothesis that BMAL1 deletion in astrocytes may lead to premature aging.

We thank Dr M. Götz (Physiological Genomics, Biomedical Center, Ludwig‐Maximilians‐University Munich, Germany) for kindly providing the Glast‐CreERT2 mouse line. We thank R. Pelizzoli and IIT‐NBT technical staff for their excellent support. We also thank M. Morini and staff of IIT Animal Facility Central Research labs for assistance in animal experiments. This work was supported by the European Research Executive Agency (REA) through the FP7‐PEOPLE‐2014‐IEF “ASTROCLOCK” (629867) and Fondazione CARIPLO research grant (2015‐0590). O.B.M. has a “Ramon y Cajal” contract (RYC2018‐026293‐I) from the Ministerio de Ciencia, Innovación y Universidades of Spain.