Exposure to dim light at night alters daily rhythms of glucose and lipid metabolism in rats

The first experiment was in agreement with Dutch laws and the Council Directive 2010/63EU of the European Parliament and the Council of 22 September 2010 on the protection of animals used for scientific purposes. All procedures were also approved by the Animal Ethics Committee of the Royal Dutch Academy of Arts and Sciences (KNAW, Amsterdam, Netherlands) and in accordance with the guidelines on animal experimentation of the Netherlands Institute for Neuroscience. The second experiment was approved by the Ethical Committee for the Care and Use of Laboratory Animals at the Comenius University in Bratislava, Slovak Republic and the State Veterinary Authority of the Slovak Republic (Ro-1648/19-221/3).

The amplitude of the locomotor activity rhythm in the CTRL group was dampened by ALAN exposure (p < 0.01; Figure 1A; Supplementary Table S2). The characteristic bimodal pattern of locomotor activity with rises at the beginning and end of the active period was changed by ALAN exposure. The two-way ANOVA and subsequent post hoc test revealed that the peak at the beginning of the night (ZT13—ZT17) found in controls, was suppressed after ALAN exposure (ZT13: p < 0.05, ZT15: p < 0.001, ZT16: p < 0.001, ZT17: p < 0.01; Supplementary Table S2). On the other hand, a smaller extra peak of locomotor activity was observed during the light period in the ALAN group (ZT1: p < 0.001, ZT4: p < 0.05, ZT5: p < 0.05; Supplementary Table S2). Actograms are shown in Supplementary Figures S3, S4. This change in the daily pattern (F_(1,14) = 30.85, p < 0.001) was also reflected in increased daytime (p < 0.001) and decreased night-time activity levels (p < 0.001) in ALAN animals, without changes in the average 24-h locomotor activity (t₍₁₄₎ = 1.43, p = 0.18; Figures 1B,C).

The RER exhibited daily rhythms (p < 0.001; Supplementary Table S2) with distinct higher values during dark/dim periods in comparison to the daytime in both the CTRL and ALAN groups (F_(1,14) = 369.20, p < 0.001; Figures 1D–F; Supplementary Table S2). In the ALAN exposed animals, RER was elevated from ZT5 to ZT7 compared to controls (ZT5: p < 0.05, ZT6: p < 0.001, ZT7: p < 0.001; Figure 1D; Supplementary Table S2), corresponding with increased locomotor activity. Average 24-h RER was not affected by 2 weeks of ALAN exposure (t₍₁₄₎ = 0.56, p = 0.59; Figure 1F). The RER is determined by the balance between carbohydrate and lipid oxidation throughout the day (Price and Mager, 2020); therefore reflecting substrate oxidation. Overall, ALAN altered the daily rhythms of locomotor activity and RER with an additional peak in locomotor activity in the light phase, as well as a less deep through of the RER.

Body weight (experiment 1: F_(2,28) = 130.10, experiment 2: F_(2,140) = 442.3, p < 0.001) and fat mass (F_(2,28) = 38.61, p < 0.001) increased and lean mass decreased (F_(2,28) = 11.21, p < 0.001; Figures 2A–D) throughout the 2-week experimental period but did not differ between the CTRL and ALAN group.

To further investigate how ALAN modifies substrate oxidation, we analyzed changes in glucose metabolism. Insulin is a crucial hormone regulating glucose metabolism and lipogenesis in the postprandial state. Plasma insulin levels displayed no daily oscillations (CTRL: p = 0.95; ALAN: p = 0.31; Figure 3A, Supplementary Table S3) and were not significantly affected by ALAN (p = 0.80; Supplementary Table S4). In the CTRL regime, plasma glucose levels tended to be rhythmic (p = 0.08) and reached their maximum during the first half of the active phase (Figure 3B, Supplementary Table S3). ALAN exposure abolished this rhythmic pattern (p = 0.55). Glucose is stored in the liver and muscles in the form of glycogen. Hepatic glycogen levels showed daily oscillations in both groups (p < 0.001), but the amplitude of the rhythm was dampened in the ALAN compared to CTRL group (p < 0.01; Figure 3C, Supplementary Table S3).

We also studied the effects of ALAN on rhythms in glucose metabolism on the molecular level (Figure 3D). The expression of hepatic glucose transporter 2 (Glut2) was rhythmic in both groups (p < 0.001) and ALAN phase-advanced this rhythm by approximately 2 h compared to controls (p < 0.01; Figure 3E, Supplementary Table S3). After entering the hepatocyte, glucose is phosphorylated by glucokinase (GCK). The expression of Gck mRNA was rhythmic with the peak in the middle of the active phase in both the CTRL and ALAN groups (p < 0.001; Figure 3F; Supplementary Table S3) and tended to be higher after the ALAN exposure (p = 0.05; Supplementary Table S4). Glucose release from glycogen is regulated by glycogen phosphorylase (PYGL). Hepatic expression of Pygl mRNA tended to be rhythmic with the peak in the middle of the light phase in the CTRL group (p = 0.09; Supplementary Table S3). ALAN rats showed the daily rhythm of Pygl mRNA with tendency to higher amplitude compared to controls (p = 0.09; Figure 3G, Supplementary Table S3). Transcription factor, forkhead box protein O1 (FOXO1), which activates expression of genes from the gluconeogenic pathway in a time-of-day dependent manner (Santoleri and Titchenell, 2019), was rhythmic at the mRNA levels with the maximum in the middle of the active phase in the control animals (p < 0.05; Supplementary Table S3). This rhythm was suppressed by the ALAN exposure (p = 0.54; Figure 3H; Supplementary Table S3). Moreover, Foxo1 mRNA was up-regulated in the ALAN group as confirmed by two-way ANOVA (p < 0.05; Supplementary Table S4). Altogether, these results indicate disturbed glucose uptake and processing in the liver. In the adipose tissue, glucose uptake is mainly facilitated by GLUT4, but its gene expression was arrhythmic in both groups (CTRL: p = 0.45; ALAN: p = 0.57; Figure 3I; Supplementary Tables S3, S4).

Lipid metabolism is under strong circadian control; therefore, we determined daily rhythms of triacyglycerols and cholesterol, adipokines and expression rhythms of genes involved in lipid metabolism. In CTRL animals, plasma triacylglycerols and cholesterol levels exhibited daily rhythms with their peaks at the end of the dark period and the beginning of the light period, respectively (p < 0.05; Figures 4A,B; Supplementary Table S5). This rhythmic pattern was lost after 2 weeks of ALAN exposure (triacylglycerols: p = 0.59; cholesterol: p = 0.24; Supplementary Table S5). In the liver, triacylglycerol concentrations were arrhythmic in both CTRL (p = 0.74) and ALAN groups (p = 0.10; Figure 4C; Supplementary Table S5). In contrast, hepatic cholesterol content exhibited a daily rhythm with no significant differences between CTRL (p < 0.01) and ALAN animals (p < 0.05; Figure 4D; Supplementary Table S5), but two-way ANOVA revealed higher plasma cholesterol levels and a tendency towards higher hepatic cholesterol levels in the ALAN than the CTRL group (p = 0.06; Supplementary Table S6). Plasma LDL-cholesterol showed a daily rhythmicity in both groups (CTRL: p < 0.05; ALAN: p < 0.05), and this rhythm was phase-advanced in the ALAN-exposed animals (p < 0.05; Figure 4E; Supplementary Table S5).

Adipokines, such as adiponectin and leptin, regulate food intake and metabolism and their rhythmicity was modified by ALAN. Plasma leptin concentration showed significant daily rhythm in the control animals (p < 0.01) which was dampened in the ALAN group (p = 0.07; Figure 4F; Supplementary Table S5). Interestingly, arrhythmic adiponectin levels in the control animals (p = 0.46) tended to gain rhythmicity after ALAN exposure with the peak at beginning of the light phase (p = 0.08; Figure 4G; Supplementary Table S5).

Next, we measured daily profiles of genes involved in fatty acid transport, synthesis, β-oxidation and ketogenesis (Figure 4H). One of the first steps of lipogenesis is catalyzed by acetyl-CoA carboxylase α (ACACA). The rhythmic expression of Acaca (p < 0.001) tended to be phase-advanced in the liver of ALAN compared to the CTRL rats (p = 0.09; Figure 4I; Supplementary Table S5). Moreover, two-way ANOVA revealed higher Acaca expression in ALAN than CTRL rats (p < 0.05; Supplementary Table S6). Daily gene expression of fatty acid synthase (Fasn) was rhythmic in both groups (CTRL: p < 0.05; ALAN: p < 0.001), whereas, another lipogenic enzyme, stearoyl-CoA desaturase 1 (Scd1), did not show a rhythmic pattern (CTRL: p = 0.94; ALAN: p = 0.78; Figures 4J,K, Supplementary Table S5). Transcription of these lipogenic genes is controlled by the transcription factor sterol regulatory element-binding protein 1c (SREBP1c) whose mRNA was arrhythmic in the CTRL group (p = 0.43) but tended to gain rhythmicity under the ALAN conditions (p = 0.09; Figure 4L; Supplementary Table S5). Similarly, ALAN-exposed rats showed a rhythmic pattern in the hepatic expression of lipoprotein lipase (Lpl, p < 0.001) and 3-hydroxy-3-methylglutaryl-CoA synthase 2 (Hmgcs2, p < 0.01; Figures 4M–N; Supplementary Table S5) with a peak at the beginning of the resting period, whereas no significant rhythms were detected in the CTRL group (Lpl: p = 0.22; Hmgcs2: p = 0.54). Daily gene expression of fatty acid transporter (Cd36) and acyl-CoA dehydrogenase medium chain (Acadm) was arrhythmic in both the CTRL (Cd36: p = 0.35, Acadm: p = 0.96) and ALAN groups (Cd36: p = 0.33; Acadm: p = 0.87; Figures 4O–P; Supplementary Table S5). Nocturnin (Noct) is a circadian-regulated protein, which has a role in lipid metabolism, its expression lost the distinct rhythmic pattern in the adipose tissue of the ALAN animals (CTRL: p < 0.01; ALAN: p = 0.17; Figure 4Q; Supplementary Table S5). However, ALAN did not affect the expression of perilipin1 (Plin1) (CTRL: p = 0.92; ALAN: p = 0.44; Figure 4R; Supplementary Table S5), the enzyme that envelops lipid droplets and prevents lipolysis in adipocytes. Taken together, ALAN profoundly affected control of lipid metabolism. Dim light during the whole night resulted in the loss of rhythmicity in plasma triacyglycerols and cholesterol, as well as Noct expression in the adipose tissue, and phase-advanced plasma rhythms of LDL-cholesterol and hepatic Acaca. Surprisingly, some parameters (adiponectin, Lpl, Srebf1c, Hmgcs2) even gained rhythmicity.

Metabolic sensors play an important role in the interplay between metabolism and the circadian clock. Therefore, we explored how ALAN impacts their daily pattern. Peroxisome proliferator-activated receptors (PPARs) are metabolic transcription factors activated by fatty acids and transcriptionally regulated by the clock (Oishi et al., 2005; Grimaldi et al., 2010). Under the CTRL regime, Pparα was rhythmically expressed with the peak during the first half of the active phase in both the liver (p < 0.001) and adipose tissue (p < 0.05; Figures 5A,H; Supplementary Table S7). Importantly, ALAN advanced the Pparα rhythm in the liver (p < 0.01) and abolished its rhythmicity in the adipose tissue (p = 0.62). On the other hand, ALAN did not affect rhythm of Pparγ expression in the liver (CTRL: p < 0.01; ALAN: p = 0.06) or adipose tissue (p < 0.05; Figures 5B,I; Supplementary Table S7). The peroxisome proliferator-activated receptor gamma coactivator 1α (PGC1α), which interacts with PPARs and other transcription factors, participates in the regulation of gluconeogenesis and other metabolic pathways (Puigserver et al., 2003; Wang et al., 2020). In the liver, distinct daily rhythms of Pgc1a were found in both CTRL and ALAN groups (p < 0.01), but ALAN phase-advanced the acrophase from the middle of the active to the end of the resting phase (p < 0.01; Figure 5C; Supplementary Table S7). On the other hand, Pgc1α expression in the adipose tissue was arrhythmic in both the CTRL (p = 0.89) and ALAN regime (p = 0.38; Figure 5J; Supplementary Table S7). Two-way ANOVA revealed a stimulatory effect of ALAN on Pgc1α expression in the adipose tissue (p < 0.05; Supplementary Table S8). Another important transcription factor and cholesterol sensor, liver X receptor α (LXRα) (Rui, 2014), lost its rhythmic expression in the liver after 2 weeks of ALAN (CTRL: p < 0.01; ALAN: p = 0.10; Figure 5G; Supplementary Table S7).

Next, we focused on the rhythmic expression of a number of additional genes encoding enzymes responsible for energy metabolism. Hepatic expression of Nampt remained rhythmic after ALAN exposure (p < 0.001), but the rhythm tended to be phase-advanced in comparison to the CTRL group (p = 0.05; Figure 5D; Supplementary Table S7). In the adipose tissue Nampt mRNA rhythm was not affected by ALAN (CTRL: p < 0.001; ALAN: p < 0.01; Figure 5K; Supplementary Table S7). NAMPT is an enzyme involved in NAD⁺ synthesis that regulates SIRT1 activity (Levine et al., 2021); therefore, we also analyzed daily oscillations of Sirt1. We found the peak in the first half of the active phase in the liver (p < 0.05) and at the light to dark transition in the adipose tissue of CTRL animals, respectively (p = 0.09; Figures 5E,L; Supplementary Table S7). Interestingly, Sirt1 rhythms were abolished in the liver (p = 0.93) and adipose tissues (p = 0.46) by ALAN exposure. However, protein expression of SIRT1 did not change throughout the day (F_(3,40) = 1.27, p = 0.30) and did not differ between the CTRL and ALAN groups in the liver (F_(1,40) = 0.59, p = 0.45; Supplementary Figure S5E).

An important enzyme controlling energy-producing processes during fasting is AMP-activated proteinkinase (AMPK). Two-way ANOVA revealed that the protein expression of the phosphorylated form of AMPK (pAMPK) was elevated in both the light and dim light period in comparison to the CTRL group (F_(1,40) = 9.37, p < 0.01; Supplementary Figure S5C). However, no significant differences between the ALAN and CTRL groups were found for the unphosphorylated AMPK (F_(1,40) = 2.27, p = 0.14; Supplementary Figure S5B) or the ratio of pAMPK to AMPK (F_(1,40) = 0.52, p = 0.47; Supplementary Figure S5D). The daily rhythm of hepatic Ampka2 mRNA levels was not affected by ALAN (p < 0.001; Figure 5F; Supplementary Table S7). Our results indicate that dim ALAN may disrupt the daily rhythms of some metabolic sensors which could potentially disturb connection between the clockwork and metabolism.

Finally, we analyzed the effects of ALAN on clock gene rhythms in the liver and adipose tissue. In both tissues, all analyzed clock genes displayed daily rhythmicity in the CTRL and ALAN groups (Figures 6A–J; Supplementary Table S9). The daily rhythm of Bmal1 expression peaked at the beginning of the resting period in CTRL rats, but its phase was advanced after exposure to ALAN in the liver (p < 0.01) and adipose tissue (p < 0.05; Figures 6B,G; Supplementary Table S9). Per1 peaked through the first half of the active period under the CTRL regime in both tissues (p < 0.001; Figures 6C,H; Supplementary Table S9). In the adipose tissue, ALAN reduced the amplitude (p < 0.05) and phase-advanced the Per1 rhythm (p < 0.01), whereas no changes were observed in the liver (p < 0.001; Supplementary Table S9), indicating tissue-specific response to ALAN. Moreover, two-way ANOVA revealed a tendency to lower Per1 expression in the adipose tissue of ALAN compared to the CTRL animals (p = 0.05; Supplementary Table S10). Daily rhythmicity of Nr1d1 tended to be phase-advanced in the liver after ALAN exposure (p = 0.06; Figure 6E; Supplementary Table S9), but was not changed in the adipose tissue (p < 0.001; Figure 6J). At the protein level, two-way ANOVA revealed daily rhythms in the hepatic REV-ERBα expression (F_(3,40) = 3.19, p < 0.05), but no effect of ALAN was detected (F_(1,40) = 0.02, p = 0.90; Supplementary Figure S5F). Together, the data show that all investigated clock genes remained rhythmic after ALAN exposure, and only the rhythms of Bmal1, Per1 and Nr1d1 showed small phase-advances in a tissue-specific manner.

In our rat experiment, body and fat mass were not affected by 2-week ALAN exposure, which contrasts with the higher fat mass and body weight in mice reported previously (Fonken et al., 2013; Cissé et al., 2017). The altered body mass in mice was predominantly explained by the increased daytime food intake (Fonken et al., 2010). The daily pattern of food intake was altered also in rats, but without the increased body weight and fat mass (Stenvers et al., 2016; Okuliarova et al., 2020; Okuliarova et al., 2022). These latter results from rats are in line with our current study, and we suggest that this species differences can be caused either by differences in light sensitivity or the study designs, for instance the different photoperiods used (14:10 LD cycle in mice and 12:12 LD cycle in rats).

We found that dim ALAN disrupted the bimodal daily pattern of locomotor activity. As expected, in the control animals, locomotor activity was higher during the dark phase with increased activity at the beginning and the end of the dark phase. Mean 24-h activity was similar in both groups, but in ALAN rats, the expected rise of activity at the onset of the dark phase was dampened. Moreover, dim ALAN increased activity in the light phase, especially at ZT1 and ZT4-5. Altered daily rhythms of activity after dim ALAN were also reported in previous mouse (Fonken et al., 2014) and rat (Stenvers et al., 2016) experiments. In rats, these changes could be partially explained by the appearance of a second rhythm with a period of ∼25 h in addition to the standard 24-h rhythm under dim ALAN conditions (Stenvers et al., 2016). Our findings are different from constant light experiments with higher light intensities when animals became arrhythmic (Polidarová et al., 2011; Stenvers et al., 2016; Báez-Ruiz et al., 2017) or constant dim light conditions in which locomotor activity was either free-running or arrhythmic, depending on the previous exposure to the LD or constant light, respectively (Stenvers et al., 2016). This supports our conclusion that dim light at night provides a condition that is principally different from constant light or constant dim light when animals become free running. With dim light at night, a synchronizing stimulus is still present, albeit its properties are weaker than properties of the standard light:dark cycle.

The observed changes in locomotor activity paralleled the changes in the daily rhythms of food and water intake reported previously (Okuliarova et al., 2022). The elevated daytime food intake was reflected in the RER, which was increased around ZT5-7. The increased RER during the light phase after ALAN exposure reflects a decreased lipid oxidation and increased oxidation of carbohydrates. The reduced utilization of lipids to cover energy needs during the sleep period may lead to an increased accumulation of lipids in the liver (Rumanova et al., 2019; Okuliarova et al., 2020). The modified daily patterns of activity and feeding behavior after dim ALAN exposure can be the consequence of the altered rhythmicity of the central oscillator, as we previously reported (Stenvers et al., 2016; Okuliarova et al., 2022).

To further understand changes in substrate oxidation after ALAN exposure, we looked at the effects of ALAN on the molecular mechanisms of glucose and lipid metabolism. Plasma glucose lost its rhythmicity and although hepatic glycogen maintained its oscillatory pattern, its amplitude was dampened after ALAN. At the molecular level, we observed a phase shift in the Glut2 rhythm in the liver of ALAN animals. In line with previous studies (Udoh et al., 2018), control rats showed the peak of Glut2 rhythm at the beginning of the active period when it is important for the efficient uptake and processing of glucose. It seems that Glut2 expression is controlled by the molecular clock as daily oscillations of Glut2 were suppressed in mice with hepatocyte-specific Bmal1 deletion (Lamia et al., 2008; Ando et al., 2016; Udoh et al., 2018). The phase advance of Glut2 expression can relate to the shift in Bmal1 expression in the liver. We assume that under ALAN conditions the uptake of glucose during the daytime is preferentially mediated by hepatic insulin-independent GLUT2 instead of insulin-dependent GLUT4 in the adipose tissue, because its expression was unaffected by ALAN exposure.

In hepatocytes, glucose is phosphorylated by GCK and can enter one of the pathways: glycolysis, pentose pathway or glycogenesis. During the postprandial state, insulin promotes glycogen synthesis in response to the higher glucose levels. Even though we found no changes in plasma insulin levels in ALAN rats, we did find a lower amplitude of the hepatic glycogen rhythm, which could be explained by the increased food intake during the light period and consequently suppressed need for glucose to be released from glycogen stores during the resting phase. Moreover, the change could be promoted by the altered amplitude of Pygl and overall higher and arrhythmic Foxo1 expression in the liver. Glycogen phosphorylase is the rate-limiting enzyme in glycogen catabolism (Agius, 2015) and FOXO1 promotes transcription of gluconeogenic genes (Santoleri and Titchenell, 2019). Foxo1 is up-regulated in diabetic individuals contributing to hyperglycemia and its inhibition or hepatocyte-specific deletion improves glucose levels (Altomonte et al., 2003; Titchenell et al., 2016). Overall, dim ALAN caused misalignment between rhythms of glucose uptake and processing in the hepatocytes, which in the long-term can negatively affect glucose metabolism.

Lipid metabolism is under strong circadian control. Plasma triacylglycerols and cholesterol lost their rhythmic patterns, and LDL-cholesterol was phase-advanced under the ALAN regime. Moreover, blood and hepatic cholesterol levels were higher in ALAN compared to control rats. In our previous study, we found an increased hepatic lipid content in rats exposed to ALAN (Okuliarova et al., 2020) and this lipid accumulation was even more pronounced when spontaneously hypertensive rats, which are genetically insulin resistant, were exposed to dim ALAN (Rumanova et al., 2019). In the current study, Srebf1c gained rhythmicity with the peak around ZT10, a few hours after the increased daytime food intake. SREBP1c is a transcription factor promoting de novo lipogenesis and such a gain of rhythmicity in Srebf1c and Lpl was also observed when rats were fed six meals a day (De Goede et al., 2018). It seems that in conditions with an attenuated central oscillator functioning the timing of food intake can contribute to the gain of rhythmicity in some genes, because this effect was also found in animals on a high-fat diet (Dyar et al., 2018) or with a disrupted SCN (Kolbe et al., 2016).

The transcription factors SREBP1c and LXRα stimulate syntheses of fatty acids and regulate cholesterol transport (Rui, 2014). Thus, changed rhythmicity of these factors can contribute to the increased plasma and hepatic cholesterol concentrations, which could be further promoted by the gain of rhythmicity in plasma adiponectin.

Metabolism and molecular clocks are mutually inter-connected through metabolic sensors that are regulated by both and ensure fine-tuning of the circadian clock to the metabolic status of an organism. In addition to clock and clock-controlled genes, metabolic sensors can be considered as the accessory elements of the molecular clock. In the liver, ALAN exposure resulted in the arrhythmicity of Sirt1 and Lxrα, while Nampt, Pparα and Pgc1α were phase-advanced. In the adipose tissue, Sirt1 and Pparα became arrhythmic. The loss of rhythmicity and altered daily pattern of these sensors could disrupt the circadian control of metabolism. Sirtuin one adjusts the molecular clock by deacetylation of PER2 and BMAL1 (Nakahata et al., 2008; Levine et al., 2020). In both the liver and adipose tissue, the rhythmic profile of Sirt1 was lost after ALAN, indicating an important pathway via which light pollution can interfere with metabolic control. The oxidized form of NAD acts as the cofactor for SIRT1, whereas NADH inhibits SIRT1 to help energy conservation during the early hours of the resting phase (Levine et al., 2021). Indeed, we found low Sirt1 values at the beginning of the resting phase in the control rats. The transcriptional control of Sirt1 is poorly understood. One of the assumed mechanisms is the regulation via PPARγ:SIRT1 complex (Han et al., 2010); therefore, the misbalanced metabolic state could be one of the factors contributing to the loss of Sirt1 rhythmicity. NAMPT is an important enzyme in the NAD⁺ salvage pathway (Reinke and Asher, 2019; Van Der Spek et al., 2021). ALAN rats showed a tendency towards a phase-advanced Nampt rhythm, which could be explained by the shift of Bmal1 expression since the heterodimer CLOCK:BMAL1 promotes its transcription (Lee and Kim, 2013). Another indirect regulatory mechanism of Nampt is via AMPK through CRY degradation, affecting the heterodimer CLOCK:BMAL1 (Lee and Kim, 2013).

PPARs represent an important connection between the circadian clock and metabolism and are regulated by the clock (Oishi et al., 2005; Grimaldi et al., 2010), the availability of fatty acids (Botta et al., 2018) and SIRT1 (Kosgei et al., 2020). While PPARα controls energy producing processes, PPARγ stimulates lipogenesis, especially in the adipose tissue. Rhythms of Pparγ were not affected by ALAN exposure in neither the liver nor adipose tissue. On the other hand, the Pparα rhythm was phase-advanced in the liver and completely lost in the adipose tissue. The daily rhythm of the transcriptional coactivator Pgc1α, which interacts with both PPARs, was also phase-advanced in the liver and unchanged in the adipose tissue after ALAN. PPARs, together with PGC1α, promote Bmal1 transcription via PPAR-response element of the promoter (Oishi et al., 2005; Kriebs et al., 2017) and transcription of Pparα and Pgc1α is regulated by CLOCK:BMAL1 complex. PPARα stimulates energy production via the promotion of the transcription of genes encoding enzymes involved in the β-oxidation of fatty acids and ketogenesis in the liver. In the adipose tissue, PPARα controls fatty acids oxidation and improves insulin sensitivity (Goto et al., 2011; Takahashi et al., 2017). Therefore, its disrupted rhythmicity can interfere with insulin signaling. Since PPARα and SIRT1 are known for the stimulation of energy producing processes, the suppression of their rhythmicity or a phase advance could disturb necessary links between the liver and adipose tissue, compromising the ability of adipose tissue to store excess of lipids.

Pparγ expression was unaffected on the mRNA level, but its protein nuclear translocation is facilitated by nocturnin in the adipose tissue (Kawai et al., 2010), and Noct expression became arrhythmic. It is likely that the misalignment between gene expression rhythms of metabolic sensors in the different tissues will have a negative effect on metabolic homeostasis and the balance between the molecular clock and metabolism.

Despite the attenuated rhythmicity of the central oscillator and its hormonal outputs after exposure to dim ALAN, the daily oscillations of clock genes in the liver and adipose tissue were largely preserved with only small changes in their acrophases or amplitudes (Bmal1, Per1, Nr1d1). Our data are in line with other models with a disturbed central oscillator, also showing maintained daily clock gene rhythms in the liver and adipose tissue, which were often phase-advanced (Husse et al., 2014; Kolbe et al., 2016; Kolbe et al., 2019; Van Der Spek et al., 2021). This phenomenon is observed only under the LD regime and not in constant darkness, suggesting an important role for behavioral and endocrine rhythms in the synchronization of peripheral clocks. Our previous study using the same lighting conditions (Okuliarova et al., 2022) demonstrated that the daily rhythm of food intake was preserved, but attenuated and the corticosterone rhythm was phase-advanced after ALAN exposure. Both rhythms are well-known to contribute to the rhythmicity of peripheral clocks. Indeed, corticosterone is known to adjust the phase of daily rhythms in peripheral tissues (Balsalobre et al., 2000; Pezük et al., 2012) and also food intake represents a strong Zeitgeber for peripheral clocks, especially the liver clock, independent of the LD cycle (Damiola et al., 2000; Oosterman et al., 2020). Thus, the attenuated circadian signaling of the central clock due to ALAN is sufficient for peripheral clocks to keep ticking, but they are not able anymore to synchronize all rhythmic metabolic processes in the different organs due to their different sensitivities to the conflicting information from the environmental light/dark and behavioral sleep/wake rhythms.