Hepatocyte Circadian Clocks Control Cholesterol Metabolism and Protect From Metabolic Dysfunction–Associated Steatohepatitis

Abstract

Background & Aims

The circadian clock synchronizes physiological processes with the 24-hour light–dark cycle. Clock disruption contributes to metabolic disorders, including metabolic dysfunction–associated steatohepatitis.

Methods

We investigated the role of the hepatocyte clock in metabolic dysfunction–associated steatohepatitis using hepatocyte-specific Bmal1 deletion (Hep-Bmal1KO) mice.

Results

Hep-Bmal1KO mice showed faster metabolic dysfunction–associated steatohepatitis progression with increased hepatic cholesterol, inflammation, and fibrosis. Transcriptomic and lipidomic analyses revealed dysregulated cholesterol metabolism in Hep-Bmal1KO mice, marked by reduced expression and disrupted rhythmicity of key cholesterol-related genes. Bioinformatic analyses identified Chrebp as a potential coregulator of these transcriptional changes. In an in vitro model with palmitate exposure and gene silencing, we found that Bmal1, but not Chrebp, regulated cholesterol accumulation, indicating Bmal1’s specific role in hepatic cholesterol metabolism. Translating our findings to a human patient cohort revealed a significantly shifted circadian phase, despite no marked effect on hepatic cholesterol levels in the livers of patients with more advanced liver disease (ie, metabolic dysfunction–associated steatohepatitis) compared with simple steatosis.

Conclusions

Taken altogether, our findings offer a roadmap to understand the hepatocyte clock’s role in metabolic dysfunction–associated steatohepatitis and its potential as a therapeutic target.

Graphical abstract

Figure generated with Biorender. Full licensing can be found here: https://app.biorender.com/citation/69f5b8d8dac4c93c998bb339.

What You Need to Know.

Background

Liver metabolism is regulated by the circadian clock system, and evidence suggests that this system is disrupted during metabolic dysfunction–associated steatotic liver disease/metabolic dysfunction–associated steatohepatitis progression. The role of the hepatocyte clock in mediating lipid metabolism and its contribution to metabolic dysfunction–associated steatotic liver disease/metabolic dysfunction–associated steatohepatitis progression remains poorly understood.

Impact

Using a dietary model of metabolic dysfunction–associated steatohepatitis, we show that hepatocyte-specific Bmal1 deletion selectively disrupts hepatic, but not systemic, cholesterol metabolism, leading to a worse disease progression. In humans, impaired clock gene expression was associated with metabolic dysfunction–associated steatotic liver disease/metabolic dysfunction–associated steatohepatitis progression.

Future Directions

These findings highlight the hepatic circadian clock as a potential therapeutic target and suggest that modulating clock function may enable liver-specific interventions for metabolic dysfunction–associated steatotic liver disease/metabolic dysfunction–associated steatohepatitis.

The circadian clock is a highly conserved temporal system that synchronizes biological processes with the Earth’s 24-hour rotation, enabling organisms to anticipate environmental changes and coordinate physiology according to the cyclic demands of their surroundings.¹ At the molecular level, the circadian system is based on interlocked transcriptional–translational feedback loops of core clock genes that oscillate throughout the day, regulating the expression of numerous gene programs.² A central clock in the suprachiasmatic nucleus coordinates peripheral clocks through multiple pathways, including behavioral (eg, food intake and activity/rest rhythm), humoral (eg, cortisol and melatonin), and neuronal signals (eg, autonomic activity) to align with the external day-night cycle.^3,4

The liver functions as a central metabolic hub and is the main site of carbohydrate, lipid, amino acid, bile acid, and xenobiotic metabolism, which are all under the control of the circadian clock.5, 6, 7 Shift work and other factors, such as daily stress and high-calorie diets, are known to disrupt the circadian clock network, and chronodisruption is a risk factor for the development of metabolic dysfunction–associated steatotic liver disease (MASLD).^8,9 MASLD has become the most common chronic liver disease globally, affecting about 30% of adults in Western countries.¹⁰ MASLD is caused by the liver’s inability to metabolize carbohydrates and fatty acids effectively. Additional imbalances, such as in de novo lipogenesis, beta-oxidation, triglyceride secretion via very low-density lipoproteins, endoplasmic reticulum stress, mitochondrial dysfunction, inflammation, insulin resistance, microbiome, and increased fibrogenesis, promote the transition of MASLD to more severe conditions, such as metabolic dysfunction–associated steatohepatitis (MASH), liver fibrosis, or hepatocellular carcinoma (HCC).^11,12

Circadian disruption impacts the liver and contributes to MASLD development. For example, chronic jetlag exposure of wild-type (WT) mice leads to steatosis in all animals and HCC in up to 9% of them. HCC rates are even higher in circadian clock mutant mice because of elevated androstane receptor (Car, also known as Nr1l3) signaling.¹³ In line with this, mice kept for two weeks on a choline-deficient high-fat diet (cdHFD) develop MASH and display a notable 4-hour phase advance in core clock genes in the liver compared with chow-fed mice, although clock gene amplitudes remain largely unchanged. At the same time, many clock target gene rhythms exhibit substantial changes in phase and amplitude.¹⁴ Along with liver disease progression—from MASLD to fibrosis—circadian rhythms deteriorate further.⁶

In this study, we examined the role of the hepatocyte clock in MASH. Targeted Bmal1 deletion in hepatocytes resulted in significant liver transcriptome rewiring under control conditions, which was potentiated under MASH conditions. The absence of a hepatocyte clock worsened MASH progression, leading to higher levels of hepatic cholesterol, inflammation, and fibrosis. Predictive analyses identified carbohydrate-responsive element-binding protein (CHREBP) as a potential co-regulator of Bmal1-driven effects. Notably, genes affected by Bmal1 knockdown, but not Chrebp, were associated with cholesterol but not triglyceride metabolism. Similarly, Bmal1-silenced hepatocytes treated with palmitate, an in vitro model of MASLD, showed increased cholesterol levels compared with control cells. In a patient cohort, patients with MASH presented with a shifted circadian phase in comparison to patients with simple steatosis.

Taken together, we show a role of the hepatocyte circadian clock function in MASH progression, emphasizing its protective effects as a potential therapeutic target.

Results

Hepatocyte-Specific Bmal1 Deletion Reduces Amplitude, Advances the Phase of Circadian Gene Expression, and Disrupts Core Clock Gene Rhythms in Mouse Liver

The Hep-Bmal1KO mouse model, created more than 10 years ago,¹⁵ allows to specifically investigate the influence of hepatocyte clocks on circadian physiology. We collected liver samples from young adult Hep-Bmal1KO mice and age-matched controls (Hep-Bmal1WT) of both sexes every 4 hours over a 24-hour cycle (Figure 1A). To identify the presence of rhythmicity in messenger RNA (mRNA) profiles, we employed a suite of analytical tools, including Jonckheere–Terpstra–Kendall (JTK)_cycle, Metacycle, CircaN, and DryR, together with a stringent statistical threshold (P < .01). There were 2756 and 1327 genes that were exclusively rhythmic in livers of Hep-Bmal1WT or Hep-Bmal1KO, respectively, whereas about 50% of the rhythmic transcriptome (3224 genes) were rhythmic in both groups (robustly rhythmic genes) (Figure 1B and C). Rhythmic genes, identified in at least 1 group, were then integrated into CircaCompare to identify rhythm parameter changes between genotypes. Robustly rhythmic genes showed a general reduction in amplitude and phase advances in Hep-Bmal1KO livers (Figure 1D). Similarly, a decrease in amplitude and a tendency for a phase advance (P = .14) were identified for the core clock genes (Figure 1D). Reduced amplitudes were identified in genes from the positive (Bmal1(Arntl), Clock, and Npas2), negative (Crys and Pers, except for Per1), and auxiliary loops (Nr1d1/2 and Dbp) of the clock transcriptional–translational feedback loop (Figure 1E).

Figure 1.

Deletion of Bmal1 in hepatocytes dampens and shifts the phase of the rhythmic liver transcriptome. (A) The design of the in vivo experiment is depicted. (B) A Venn diagram shows the number of rhythmic genes in Hep-Bmal1WT and Hep-Bmal1KO mice. (C) Heatmap represents the exclusive rhythmic genes of Hep-Bmal1WT (left) and Hep-Bmal1KO mice (right). (D) Boxplot shows the difference in amplitude and phase between genotypes for the robustly rhythmic and core clock genes. (E) The diurnal profile of core clock genes is depicted. N = 4 per ZT, both male and female mice. ‘R’ symbol indicates rhythmicity, whereas ‘NR’ signifies arrhythmicity detected by CircaCompare.

Loss of Bmal1 in Hepatocytes Dampens Messenger RNA Rhythms and Lowers Overall Expression Levels of Metabolic and Transcription Factors

Detailed rhythm parameter analysis using CircaCompare identified several genes with altered mesor, amplitude, and/or phase (differentially rhythmic genes [DRGs]). A total of 2168, 629, and 450 genes showed changes in mesor, amplitude, or phase, respectively (Figure 2A; Supplementary Table 1). In Hep-Bmal1KO livers, genes with increased mesor showed enrichment for biological processes associated with biosynthetic activities, including amino acids, isoprenoids, sterols, and organophosphates. Additionally, enrichment was seen for processes of structural assembly (eg, ribosome and subunit biogenesis) and DNA repair (eg, double-strand break repair). DRGs exhibiting reduced mesor were enriched for lipid biosynthesis as well as glycerolipid and carbohydrate metabolism (eg, monosaccharide metabolic processes). Additional processes associated with energy-associated pathways (eg, generation of precursor metabolites), membrane transport (eg, monocarboxylic acid transport), and response to cholesterol were also identified. Genes with a reduction in amplitude were enriched for processes related to circadian rhythms and various metabolism-related processes (eg, alpha-amino acid, polysaccharide, carboxylic acid, and vitamin B6 metabolism). Conversely, genes that gained amplitude were enriched for immune functions (eg, including myeloid cell differentiation and the regulation of the adaptive immune response) and some metabolic processes, such as fatty acid and organic acid biosynthesis, in addition to steroid and alcohol metabolism. Among those genes with a phase advance, biological processes related to energy metabolism (eg, carbohydrate and purine metabolism, and adenosine triphosphate biosynthetic processes) and lipid storage were overrepresented. Conversely, genes with a phase delay were enriched for metabolic processes related to ketones, lipids, fatty acids, and cholesterol metabolism (Figure 2B; Supplementary Table 1).

Figure 2.

Differential rhythm analysis of the diurnal liver transcriptome in Hep-Bmal1WT and Hep-Bmal1KO mice on a NC diet reveals disrupted circadian rhythms and decreased expression of key transcription factors. (A) UpSet plot depicts the number of DRGs. (B) Enrichment analyses of the identified DRGs are depicted. (C) Diurnal expression profile of selected DRGs is depicted. (D and E) Predictive transcriptional factor analyses using Ch3A3 for DRGs with reduced amplitude or mesor. (F) Diurnal expression profile of TFs identified using Ch3A3 algorithm is depicted. Changes in additional rhythmic parameters are described in bold letters. N = 4 per ZT, both male and female mice. ‘R’ symbol indicates rhythmicity, whereas ‘NR’ signifies arrhythmicity detected by CircaCompare.

Because lipid metabolism was strongly ranked in the enrichment analyses, this prompted us to further investigate this pathway. Most of the effects identified in lipid metabolism were associated with reduced mesor, especially in lipid biosynthesis (Dgat2, Lpin1, Lpin2, and Scd1) and oxidation (Cpt1a, Acadl, Acox1). Similarly, cholesterol metabolism-associated DRGs (Ldlr, Lrp6, Ces1d, Ces1e) showed reduced mesor (Figure 2C). The rate-limiting enzyme HMGCR, encoded by Hmgcr, showed increased mesor and a phase delay in Hep-Bmal1KO (Supplementary Table 1). Considering the role of BMAL1 as a transcription factor (TF), we hypothesized that its loss would affect the rhythmic expression of secondary TFs and impair the expression of lipid- and cholesterol-associated genes. We tested this hypothesis by predicting putative TFs associated with amplitude down DRGs using the ChEA3 algorithm.¹⁶ By applying stringent criteria (score <400 and amplitude down), we identified 14 candidate TFs. As expected, Bmal1 (also known as Arntl) was one of these transcription factors exhibiting the greatest reduction in amplitude. We also identified several other clock-associated TFs, including Clock, Bhlhe40, Nr1d1, Nfil3, and Rorc, which is in line with the pivotal role of BMAL1 in sustaining circadian clock function (Figure 2D). Subsequently, we focused our analysis on DRGs with reduced mesor that have a direct role in energy metabolism by using the Kyoto Encyclopedia of Genes and Genomes database (KEGG). ChEA3 prediction in this subset of genes revealed 25 TFs, among which 12 were identified (eg, Chrebp (Mlxipl), Nr1i2, Nr1h4, Nr1i3, Hnf4a, and Creb3l3) as key lipid and cholesterol metabolism regulators, all demonstrating reduced mesor (Figure 2E and F).

Our findings demonstrate the multifaceted consequences of disrupting the hepatocyte clock in mice fed with normal chow (NC) ad libitum. Notably, the significant reduction in mesor observed in clock target genes associated with lipid and cholesterol metabolism was associated with expression changes in metabolism regulator TFs.

Loss of Bmal1 in Hepatocytes Worsens Metabolic Dysfunction–Associated Steatohepatitis

In light of the significant reduction in the expression of several metabolically relevant TFs, we postulated that Hep-Bmal1KO mice would exhibit heightened susceptibility to MASH development. To test this, we subjected Hep-Bmal1KO and -WT mice to a cdHFD, which was established to induce MASH.¹⁴ We conducted untargeted lipidomics at the peak time (Zeitgeber time [ZT] 6, light phase) of endogenous triglyceride and cholesterol levels.¹⁷ A pronounced diet effect on lipid levels was observed irrespective of genotype. For instance, an increase in cholesterol esters, diglycerides, triglycerides, and phosphatidylglycerols, as well as a reduction in acyl-carnitine, ceramide, lysophosphatidylcholine, phosphatidylcholine, lysophosphatidylethanolamine, and phosphatidylethanolamine species was identified (Figure 3A; Supplementary Table 2). Lipids influenced by both diet and genotype included cholesterol esters, plasmalogen phosphatidylcholine, plasmalogen phosphatidylethanolamine, and sphingomyelin species (Figure 3B; Supplementary Table 2). Increased cholesterol levels in Hep-Bmal1KO livers were confirmed by an independent method (Figure 3C). In stark contrast to hepatic lipidomics, serum lipidomics analysis (at ZT10) indicated a minor role of the hepatic clock in the regulation of serum cholesterol levels, affecting only a few species. These included 1 triglyceride and phosphatidylinositol, which were upregulated in Hep-Bmal1KO fed with cdHFD (Figure 3D). Pathologic MASH scoring revealed no marked differences in steatosis and ballooning but higher inflammation scores in Hep-Bmal1KO compared with Hep-Bmal1WT mice (Figure 3E and F). Moreover, Hep-Bmal1KO mice fed with cdHFD exhibited a nearly 10-fold increase in peri-sinusoidal fibrosis compared with Hep-Bmal1WT controls (Figure 3G).

Figure 3.

Deletion of Bmal1 in hepatocytes elevates hepatic cholesterol levels, exacerbating inflammation and fibrosis. (A) Heatmap depicts all lipids identified with a diet effect. (B) Lipids with a diet and genotype effect are shown by a heatmap. Liver samples (male and female) were obtained from ZT 6. Normalized lipid species levels and their identity are shown. (C) Evaluation of total hepatic cholesterol across genotypes and diets. (D) Serum lipidomics was performed on samples from ZT10 and is shown as a volcano plot (P < .05). (E) MASLD activity score of Hep-Bmal1WT and Hep-Bmal1KO fed with cdHFD is shown. (F) Representative images of ballooning (upper left), inflammation (lower left), fibrosis (upper right), and fibrosis quantification (lower right) are depicted. (G) Fibrosis score quantification. For histologic analysis, samples from the light (ZT 2–10) and dark phases (ZT 14–22) were used. Scale bar, 100 μm. For (D–F), samples were taken from Hep-Bmal1WT (N = 14) and Hep-Bmal1KO (N = 15) mice, both male and female, at ZT 6, 10, 18, and 22.

Loss of Hepatocyte Bmal1 Alters TF Gene Network Rhythms in Metabolic Dysfunction–Associated Steatohepatitis

To understand how the loss of Bmal1 in hepatocytes alters cholesterol metabolism and MASH development, we compared daily transcriptional signatures between livers of Hep-Bmal1WT and Hep-Bmal1KO mice under MASH conditions. Rhythm parameter analysis identified several hundred DRGs that exhibited changes in mesor (1631), amplitude (441), or phase (376) (Figure 4A). Enrichment analysis of DRGs with decreased mesor revealed pathways related to lipid and fatty acid metabolism, small molecule catabolism, xenobiotic metabolism, and the urea cycle. Conversely, enrichment of DRGs with increased mesor yielded processes such as wound healing, repair, and chondrocyte proliferation. DRGs with reduced amplitude were enriched for cell–cycle regulation, DNA replication, cytokinesis regulation, DNA damage response, and circadian rhythms. Processes related to the immune system, including virus response, were overrepresented in DRGs with increased amplitude. DRGs associated with a phase delay were most enriched for circadian processes, whereas those with phase advancement were enriched for cellular damage response and the Wnt pathway (Figure 4B; Supplementary Table 3).

Figure 4.

Differential rhythm analysis of the diurnal liver transcriptome in Hep-Bmal1WT and Hep-Bmal1KO mice on a cdHFD shows that Bmal1 deletion disrupts the expression of cholesterol-related genes. (A) UpSet plot depicts the number of DRGs. (B) Enrichment analyses of the identified DRGs are depicted. (C) Diurnal expression profile of selected DRGs associated with cholesterol metabolism. Changes in additional rhythmic parameters are described in bold letters. (D) Predictive transcriptional factor analyses using Ch3A3 for DRGs with reduced mesor. (E) Diurnal expression profile of TFs identified using the Ch3A3 algorithm is depicted. ‘R’ symbol indicates rhythmicity, whereas ‘NR’ signifies arrhythmicity detected by CircaCompare. ‘R∗’ symbol shows a tendency of rhythmicity (P < .1). N = 4 per ZT, both male and female mice.

Based on our lipidomics findings, we performed a targeted analysis of cholesterol metabolism-associated DRGs. Several genes involved in cholesterol uptake (Ldlr), intracellular cholesterol transport (Npc1), bile acid secretion (Cyp7a1), and cholesterol metabolism (Ces1b, Ces1d, Cyp1a2, Dhcr24, Mttp) showed significant reductions in mesor. The Hmgcr profile showed no alteration in any rhythmic parameter (Supplementary Table 2). Additionally, reduced amplitudes of mRNA profiles for Npc1 and Cyp7a1 were observed with profiles classified as being non-rhythmic in Hep-Bmal1KO livers (Figure 4C). We employed the ChEA3 algorithm to identify putative TFs associated with DRGs exhibiting decreased mesor. This analysis was also performed on a subset of energy metabolism-associated DRGs with decreased mesor. We consolidated both lists to focus on shared TFs implicated in metabolism (Figure 4D). A total of 12 TFs were identified (Figure 4E), among which Chrebp (Mlxipl), Creb3l3, Cebpa, Hnf4a, and Nr1h3 were top-ranked.

Hepatocyte Clock-Driven Transcriptional Programs Regulate Cholesterol Metabolism

Considering that Chrebp/Mlxipl was identified as the top TF-associated with the DRGs exhibiting reduced mesor in our MASH model and its established role in regulating carbohydrate and lipid metabolism,¹⁸ we hypothesized a cooperative interaction between Bmal1 and Chrebp in regulating MASLD/MASH progression. To directly test this interaction, we used an in vitro setup with murine hepatocytes treated with palmitate to induce steatosis.^19,20 Following 48-hour palmitate treatment, a transcriptome analysis was conducted to assess the impact of palmitate-induced steatosis on gene expression. A total of 853 genes were upregulated and 975 downregulated by palmitate (bovine serum albumin [BSA] vs Palmitate; Padj < .1) (Figure 5A and B; Supplementary Table 4). Enrichment analysis of the upregulated genes revealed involvement in ribonucleoprotein complex biogenesis, ribosomal RNA processing, protein folding, and post-transcriptional regulation. In contrast, downregulated genes were enriched in pathways related to inflammatory response, cytokine biosynthesis, and lipid, carbohydrate, and steroid metabolism. Notably, palmitate treatment recapitulated several processes observed in Hep-Bmal1WT mice fed a cdHFD, with both conditions exhibiting upregulation of pathways related to cytoskeleton organization, protein transport and folding, and cellular responses to stress. In contrast, processes enriched among downregulated genes in both datasets included pathways involved in lipid metabolism, small molecule biosynthetic processes, and, expectedly, circadian rhythms (Figure 5C; Supplementary Tables 3 and 4).

Figure 5.

An in vitro model of MASLD demonstrates that the Bmal1-regulated transcriptional program specifically controls cholesterol metabolism in immortalized hepatocytes, without impacting lipid metabolism. (A) Schematic of the experimental design. (B and C). Volcano plots displaying genes with increased (red) and decreased (blue) expression, accompanied by enrichment analysis (N = 4 per group). (D) qPCR validation of knockdown efficiency for Bmal1, Chrebp, and combined Bmal1-Chrebp (BOTH) (N = 4 per group). (E) Dual staining analysis of triglycerides and cholesterol via flow cytometry (n = 7–8 per group). Letters denote statistical significance determined by 1-way ANOVA followed by Tukey post-test. (F) Heatmap illustrating gene expression clustering (N = 4 per group), with distinct clusters highlighted in response to gene knockdown (KD). (G) Representative plots showcasing unique Bmal1- and Chrebp-regulated targets. (H) Enrichment analysis of the clusters identified in panel (F).

To explore a putative interaction between Bmal1 and Chrebp, we conducted combinatory silencing experiments using endoribonuclease-prepared small interfering RNA (esiRNA). The aim of this experiment was to examine the transcriptional effects of silencing Bmal1 or Chrebp, as well as to assess the impact of dual silencing. By studying transcriptome changes in Bmal1- and Chrebp-silenced cells, we aimed at inferring a synergistic or antagonistic relationship between these genes. Twenty-four hours after esiRNA transfection (48 hours postpalmitate treatment), transcriptome analyses were performed. Transcriptome data confirmed the successful silencing of Bmal1 and Chrebp, which was further validated by quantitative polymerase chain reaction (qPCR) (Figure 5D; Supplementary Table 4). DESeq2 analyses identified 1377 differentially expressed genes (DEGs) in response to Bmal1, Chrebp, or dual silencing (likelihood ratio test; Padj < .1). Dual-labeling triglycerides and cholesterol staining revealed that silencing Chrebp alone, or in combination with Bmal1, significantly reduced triglyceride levels, indicating a Chrebp-driven effect. However, only Bmal1 silencing resulted in elevated cholesterol levels in murine hepatocytes (Figure 5E), mirroring the in vivo phenotype observed in cdHFD-fed Hep-Bmal1KO mice (Figure 3B and C).

To identify genes exclusively regulated by Bmal1 or Chrebp, we filtered the previously identified DEGs using a threshold of a 25%-fold change (log2 of 0.32), ensuring they were unique to either the Bmal1 or Chrebp dataset (Figure 5F). This analysis yielded 640 DEGs, which were categorized into 6 distinct classes: shared with similar regulation (63 DEGs), shared with opposite regulation (173 DEGs), genes exclusively upregulated upon Bmal1 silencing (34 DEGs) (Figure 5G), genes exclusively downregulated upon Bmal1 silencing (33 DEGs) (Figure 5G), genes exclusively upregulated upon Chrebp silencing (230 DEGs) (Figure 5G), and genes exclusively downregulated upon Chrebp silencing (107 DEGs) (Figure 5G).

Comparison of the genes coregulated by Chrebp and Bmal1 in the same direction (shared cluster, Figure 5F) revealed enrichment of protein acetylation and positive regulation of immune effector process (Figure 5H; Supplementary Table 4). In contrast, a distinct set of genes showed opposite regulation when either Chrebp or Bmal1 was silenced (shared opposite cluster) (Figure 5F). These genes were also mainly linked to immune responses, interferon-beta signaling, and cytokine production (Figure 5H; Supplementary Table 4). These results demonstrate that both genes significantly influence immune- and inflammation-related gene programs, but in opposing directions. Notably, the number of genes with opposing regulation was roughly 3 times greater than those with shared regulation.

Chrebp-exclusive upregulated DEGs were also associated with immune system responses such as nuclear factor κB (NF-κB) signaling, regulation of leukocyte differentiation, and cytokine production. On the other hand, Chrebp-exclusive downregulated DEGs were primarily associated with fatty acid and lipid metabolism (Figure 5H; Supplementary Table 4). Albeit numerically smaller than Chrebp-exclusive genes, downregulated Bmal1-exclusive genes were enriched in pathways related to innate immune response and chemotaxis, whereas Bmal1-exclusive upregulated DEGs were associated with circadian rhythms, steroid metabolism, and lipid metabolism (Figure 5H; Supplementary Table 4). Further analysis of Bmal1-associated genes revealed several indirect regulators of energy metabolism (eg, Igf1, Lbh, Rorc, and Foxa1) in addition to identifying Abca1 and Abca8b, key cholesterol efflux proteins (Figure 5F and G).

To further investigate an interaction between BMAL1 and CHREBP, we performed ChipPCR analysis of established Chrebp targets (e.g., Txnip, Pklr, Klf10) at the highest time of BMAL1 binding (ZT 6).²¹ In Hep-Bmal1KO mice, loss of BMAL1 was associated with increased CHREBP occupancy at these promoters, while this pattern was unaffected by the presence of MASH. These findings suggest that hepatic Bmal1 deficiency enhances CHREBP binding to its target genes, indicating a perturbation of CHREBP-regulated transcriptional programs (Figure 6A–C). Notably, the transcriptomic perturbation, evidenced by the number of DEGs following gene knockdown, was more pronounced in Chrebp-silenced samples, suggesting that Chrebp has a stronger influence on liver transcriptome regulation. Moreover, we identified several Chrebp-regulated gene programs associated with lipid metabolism. The significant downregulation of genes involved in lipid biosynthesis is consistent with the observed reduction in triglyceride levels in Chrebp-silenced samples, in line with previous studies on Chrebp overexpression or knockdown.22, 23, 24

Figure 6.

Chrebp ChipPCR in Hep-Bmal1WT and Hep-Bmal1KO livers. (A–C) Analysis of selected Chrebp targets in livers of Hep-Bmal1WT or Hep-Bmal1KO mice fed with NC or cdHFD. N = 5 per group performed at ZT6.

Although Bmal1 silencing had a relatively modest impact on the overall transcriptome, it notably upregulated several cholesterol-related genes, particularly Abca1 and Abca8b, suggesting a compensatory response to elevated cholesterol levels. This highlights the selective role of Bmal1 in liver cholesterol metabolism. In vitro data, supported by in vivo diurnal transcriptome analyses, indicate that Bmal1 deletion disrupts multiple aspects of cholesterol regulation at the transcriptional level. This includes reduced cholesterol uptake (Ldlr, reduced expression), impaired intracellular cholesterol transport (loss of Npc1 rhythmicity), diminished bile acid secretion (loss of Cyp7a1 rhythmicity), and increased cholesterol secretion (elevated Abcg5 and Abcg8 expression). Although such changes were found at the transcriptional level, hepatic cholesterol measurements—by 2 independent methods—showed increased hepatic cholesterol in the livers of Hep-Bmal1KO fed with cdHFD. In addition, our in vitro assay, where Bmal1 expression was silenced in murine hepatocytes treated with palmitate, yielded increased cholesterol. Taken together, our findings suggest a selective role of Bmal1 in regulating cholesterol metabolism.

Patients With Metabolic Dysfunction–Associated Steatohepatitis Show an Altered Internal Circadian Phase

To translate our findings to the human situation, we evaluated hepatic cholesterol content and circadian clock gene expression in liver biopsies from a patient cohort living with obesity, and either metabolic dysfunction-associated steatotic liver (MASL, simple steatosis) or MASH with a body mass index ≥40 kg/m² or ≥35 with at least 1 cardiometabolic comorbidity (hypertension, type 2 diabetes mellitus, dyslipidemia, or obstructive sleep apnea syndrome) (Supplementary Table 5). To account for temporal effects on clock gene expression, we developed an internal phase index by multiplying PER3 expression with NR1D2, then dividing the result by BMAL1 expression. A higher ratio implies an increased expression of PER3 and NR1D2, whereas a decreased ratio suggests increased BMAL1 expression (Figure 7). Importantly, our measurements cannot be used to infer internal phase, but they can be used to indicate a potential circadian disruption condition. Indeed, we identified that patients with MASH had a higher ratio compared with patients with MASL (eg, simple steatosis), thus suggesting altered internal phasing in MASH (Figure 8A). Importantly, both MASL and MASH samples were collected across the day and not restricted to a given phase, thus reducing a temporal bias in our sample collection. However, no differences in total hepatic cholesterol between MASH and MASL samples were observed (Figure 8A). Considering the entire cohort (MASL and MASH), hepatic cholesterol was positively correlated with serum high-density lipoprotein levels. Internal phasing had a borderline significant trend with disease score (NAFLD Activity Score; P = .07) (Figure 8B).

Figure 7.

Defining the internal timing. (A) Graphic representation of the diurnal gene expression of BMAL1 (black curve) and PER3-NR1D2 (purple curve). (B) Presentation of the internal time calculation obtained from the graph described in (A).

Figure 8.

Altered circadian phase in livers of patients with MASH compared with simple steatosis. (A) NAFLD Activity Score, internal timing expression, and hepatic cholesterol levels of patients with MASL and MASH are depicted. (B) Correlation of clinical data with hepatic cholesterol and internal timing. N = 14 for MASL and N = 24 for MASH.

Therefore, the human findings suggest a correlation between disrupted circadian rhythms, namely altered internal phasing, with MASL-to-MASH progression.

Discussion

Our research highlights the role of the hepatocyte circadian clock. Under physiological conditions, the absence of the circadian clock in hepatocytes led to a reorganization of the hepatic transcriptome, impacting thousands of genes with alterations in mesor, phase, and/or amplitude. It further decreased the expression of several metabolic-associated TFs, suggesting a possible interaction between BMAL1 and other TFs. The protective function of the hepatocyte clock became apparent when mice were subjected to cdHFD, as Hep-Bmal1KO mice demonstrated exacerbated MASH development. At the transcriptional level, the absence of the hepatocyte circadian clock significantly modified cholesterol metabolism, influencing several genes implicated in cholesterol uptake, transport, and degradation. Predictive bioinformatic analyses identified a potential interaction between Chrebp and Bmal1. Subsequent in vitro mechanistic studies revealed that Bmal1 and Chrebp share an antagonistic relationship, with these TFs modulating multiple metabolic pathways in opposite directions. A novel aspect of our findings revealed that the hepatic Bmal1, under our experimental conditions, selectively regulates cholesterol metabolism. Importantly, our investigation in a cohort of obese patients with MASL/MASH had an altered internal phasing, suggesting disrupted circadian rhythms in MASH livers, but did not recapitulate the hepatic cholesterol phenotype observed in experimental models.

Under NC conditions, the livers of Hep-Bmal1KO mice exhibited significant disruption in several clock genes, including Bmal1, Clock, Cry1, Cry2, Nr1d1, and Nr1d2. However, it is noteworthy that approximately 50% of the transcriptome remained rhythmic in both Hep-Bmal1KO and Hep-Bmal1WT mice. This observation must be reconciled with the fact that Hep-Bmal1KO mice display rhythmic locomotor and feeding behaviors.¹⁵ Given that the liver’s circadian clock is particularly sensitive to feeding regimens,²⁵ the presence of such consistent rhythms is not unexpected. Comprehensive circadian analyses revealed a pronounced rewiring effect. Among the numerous identified processes, the absence of Bmal1 in hepatocytes resulted in diminished expression of genes linked to lipid and glycerolipid metabolism. Additionally, phase effects, characterized by advances or delays, were noted in lipid and cholesterol metabolism, suggesting a complex regulation of these pathways. However, a minor effect on the lipidome was observed between Hep-Bmal1WT and Hep-Bmal1KO fed with ad libitum NC. A notable finding, however, was the substantial downregulation of several TFs. Although rhythmicity prevailed in most instances, several TFs exhibited a marked reduction in mesor, indicating Bmal1’s regulatory role in the diurnal regulation of these TFs.

Extensive research on the effects of Chrebp regulating energy metabolism has been performed. Global Chrebp knockout increased plasma glucose due to reduced insulin sensitivity, increased hepatic glycogen levels, and reduced lipogenesis, events which were enhanced when mice were fed a high-starch diet.²⁶ In the leptin-deficient model (ob/ob), global Chrebp knockout decreased hepatic fatty acid synthesis, normalized triglyceride levels, and reduced body weight.²⁷ Similar findings were observed in ob/ob mice with hepatocyte-targeted adenovirus-silenced Chrebp, including a reduction in hepatic cholesterol levels.²⁸ Global Chrebp knockout mice fed with a Western diet had reduced hepatic lipid and cholesterol levels, decreased beta-oxidation and ketogenesis, and increased intestinal lipid absorption.²⁴ Lipid accumulation in hepatocyte-specific Chrebp knockout mice was similar to that in control mice when they consumed regular chow or a high-fat diet (HFD). However, these knockout mice exhibited lower lipid levels when fed a carbohydrate-rich diet. Despite being protected from carbohydrate-induced hepatic steatosis, the hepatocyte-specific Chrebp knockout mice showed impaired glucose tolerance.²⁹ Conversely, hepatic Chrebp-overexpressing mice fed an HFD had more pronounced hepatic steatosis despite having improved insulin signaling and glucose tolerance compared with controls.²²

The putative interaction between Bmal1 and Chrebp has not yet been explored in depth. Predictive transcription factors linked to genes exhibiting reduced mesor under both control and cdHFD conditions have identified Chrebp as the most significantly associated TF factor in the livers of Hep-Bmal1KO mice. Our in vitro model of steatosis utilizing hepatocytes further illuminated this novel interaction. Transcriptome analysis indicated that Chrebp and Bmal1 likely interact through opposing regulatory pathways, which is evident from the fact that DEGs with opposing regulation are approximately 3 times higher than those with shared regulation. Silencing Chrebp in vitro resulted in a greater disruption, evidenced by nearly a 5-fold increase in DEGs compared with Bmal1-silenced hepatocytes. Consistent with earlier research,^22,26, 27, 28 Chrebp silencing diminished genes related to lipogenesis. Further confirmatory findings came from the in vitro palmitate model in which Chrebp-silenced hepatocytes had reduced triglyceride levels compared with control and Bmal1-silenced hepatocytes. Importantly, dual Bmal1 and Chrebp-silencing also resulted in reduced triglycerides, thus suggesting a Chrebp-dependent effect in line with the transcriptome predictions. Furthermore, the findings supporting the Chrebp-Bmal1 interaction revealed that CHREBP-ChIPPCR showed enhanced binding of CHREBP to classical CHREBP-regulated genes (Txnip, Pklr, Klf10) in Hep-Bma1KO livers under NC conditions, which remained unchanged by MASH. Therefore, our findings identified a novel interaction of Chrebp with Bmal1, thus suggesting that Chrebp effects in the liver can be time-of-day–dependent, a matter for future investigation.

Our research highlights the protective role of the hepatic clock during MASH development. Interestingly, at the transcriptional level, many lipid-associated genes had reduced mesor and amplitude in Hep-Bmal1KO mice, but these changes did not affect lipid levels across the genotypes. Our transcriptional findings agree with a previous report showing that adenovirus-dependent Bmal1 deletion leads to reduced expression of lipogenic genes via an AKT-dependent pathway. An important distinction is that these findings were performed in fasted mice that were refed, whereas our findings are derived from ad libitum–fed mice.³⁰ Moreover, primary hepatocytes from global Bmal1 knockout mice show reduced Chrebp expression and impaired lipogenesis in response to glucose overload.³⁰ In contrast, cholesterol-associated genes were affected, and hepatic cholesterol levels were higher in Hep-Bmal1KO mice. Poorer MASH scores and increased fibrosis were also noted in Hep-Bmal1KO mice. Our findings indicate a role for Bmal1 in regulating hepatic cholesterol metabolism, supported by in vitro experiments with Bmal1-silenced hepatocytes. The selective role of Bmal1 in regulating cholesterol metabolism is unclear from previous studies. For example, Hep-Bmal1WT and Hep-Bmal1KO fed with a chow diet exhibited decreased triglyceride levels only in the morning (ZT 0), and no cholesterol levels were measured.³¹ In addition, loss of Bmal1 in hepatocytes of HFD-fed mice leads to increased hepatic levels.³² In our experiments (ZT 6), no changes in hepatic lipids were noted, except for a few types of glycerophospholipids. A recent study revealed that the silencing of Bmal1 in hepatocytes enhances m6A mRNA methylation, contributing to the lipid disruption seen in Hep-Bmal1KO mice.³³ In an elegant study, global Bmal1 KO mice in an Apoe-null background exhibited elevated hepatic cholesterol and triglyceride levels and reduced cholesterol excretion in bile and feces. In Hep-Bmal1KO/ApoeKO mice, hepatic cholesterol and triglyceride levels were also elevated compared with Hep-Bmal1WT/ApoeKO mice. Overexpression of Bmal1 in the liver of Hep-Bmal1KO/ApoeKO mice increased cholesterol and bile acid levels in the bile. This effect was dependent on Bmal1 regulating the rhythmic expression of Abcg5 and Abcg8.³⁴ Notably, our findings support earlier research by Pan et al and reinforce Bmal1’s involvement in hepatic cholesterol regulation. However, Pan’s study was conducted in an Apoe-null background, whereas our research concentrated on the role of Bmal1 in hepatocytes. Importantly, the subtle changes observed in serum lipidomics suggest that the hepatic clock regulates cholesterol metabolism in a liver-autonomous manner, with minimal influence on systemic parameters. Some studies have suggested potential pathways for the protective role of the hepatic clock. For instance, a study has shown that the protective effects of the hepatocyte clock are mediated via transforming growth factor-β signaling that suppresses fibrotic gene expression in healthy conditions. Metabolic stress disrupts this rhythmic control, enhancing profibrotic pathways and promoting liver fibrosis in MASH.³⁵ Conversely, disruption of the hepatic clock has been shown to disrupt mitochondrial structure, leading to increased oxidative stress, which contributes to insulin resistance, increased hepatic lipid accumulation, and metabolic dysfunction. Hepatic Bmal1 restoration mitigates the deleterious effects caused by an HFD, thus showing a direct link between Bmal1 and mitochondrial regulation.³² Using our in vitro steatosis model, we could identify that hepatocytes treated with palmitate, a mimic for steatotic condition,^19,20 had increased cholesterol levels when Bmal1 was silenced compared with control and Chrebp-silenced groups. Transcriptome analysis also identified an upregulation of cholesterol efflux transporters (Abca1 and Abca8b), suggesting a compensatory mechanism due to the high cholesterol levels in vitro. Interestingly, the gene signature associated with cholesterol in vivo was associated with reduced mesor and/or amplitude for cholesterol uptake, intracellular transport, bile acid secretion, and higher cholesterol secretion. Collectively, both in vitro and in vivo evidence point towards a selective role of Bmal1 in regulating hepatic, but not serum cholesterol metabolism under our experimental conditions.

Circadian rhythm evaluation in patients with MASLD is challenging due to the difficulty of obtaining liver samples from the same individual at multiple time points throughout the day. However, in an elegant study by Johanns and colleagues, the authors successfully collected liver biopsies from healthy patients, patients with MASL, and patients with MASH throughout the day. Their analysis revealed time-of-day-dependent transcriptomic changes between morning (AM) and afternoon (PM) samples, regardless of disease status. They also observed that the AM-to-PM fold changes of common time-dependent DEGs were progressively attenuated in steatotic livers and further diminished in MASH livers, compared with healthy controls.³⁶ This is in line with the predictions of a gradual disruption of hepatic rhythms as MASLD progresses.⁶ Our findings in humans further support these results, as patients with MASH exhibit altered internal phasing compared with those with MASL, which suggests a further impairment of liver circadian rhythms in MASH compared with MASL. Importantly, our analyses may be subject to cofounding factors that were not controlled, such as meal timing or meal content, and should be interpreted as suggestive evidence of impaired rhythms in patients with MASH.

Our study has its limitations. For instance, our differential rhythm analysis was based on an uncorrected P value, which may result in higher false-positive rates. To address this, a more stringent P value threshold (< .01) was set to determine rhythmicity, with these genes later used for differential rhythm analysis. Our analyses did not include the zonation effect. Triacylglycerols, diacylglycerols, sphingolipids, and ceramides exhibited specific alterations and a shift from pericentral to periportal localization in MASH.³⁷ However, our earlier research did not find a distinct zonation effect regarding lipid accumulation in mice fed a cdHFD.¹⁴ This phenomenon is related to the strong induction of MASH by cdHFD, which can occur in just 2 weeks. However, it does not accurately reflect the progression of human MASH, limiting the clinical applicability of our results.^38,39 In fact, the hepatic cholesterol phenotype induced by Bmal1 loss was not replicated in the human cohort. This difference could be due to various factors, including differences in hepatic zonation, as biopsy samples likely vary in their representation of metabolic regions with distinct lipid profiles. One should also consider that Bmal1 knockout in the liver is an extreme condition that completely disables BMAL1 function, whereas human MASH livers still express BMAL1 (likely functional), despite disrupted rhythmic transcript expression. The analysis of total rather than specific cholesterol fractions, and the early disease stage with minimal fibrosis in all patients, may have precluded detectable cholesterol accumulation at the time of sampling. We could not make a full comparison because there were no healthy lean liver biopsy biopsies available, owing to ethical limitations. Although our target Bmal1 deletion in hepatocytes led to marked effects in the liver, the contribution of other nonhepatocyte cells in regulating cholesterol metabolism remains elusive and should be addressed using single-cell RNA sequencing in future experiments.

Taken altogether, our study offers a valuable overview of the role of the hepatocyte clock under NC and MASH conditions, providing a useful resource for future experiments. We discovered that impairing the circadian clock of hepatocytes impacts the expression of various metabolism-associated TFs, contributing to the exacerbated development and progression of MASH. We also identified a new connection between Bmal1 and Chrebp, which, although not directly affecting hepatic cholesterol metabolism, highlighted a potential circadian regulation of Chrebp in the liver. Importantly, these findings are linked to a specific role of Bmal1 in regulating hepatic cholesterol metabolism. Our experimental findings were supported by clinical data showing that human MASH livers show altered internal phasing compared with MASL livers. Collectively, our findings suggest that targeting the circadian clock can be a promising pathway to treat MASH in humans.

Materials and Methods

Mouse Model and Experimental Conditions

To achieve hepatocyte-specific deletion of Bmal1, we used Alb^Cre mice (B6.Cg-Speer6-ps1Tg(Alb-cre)21Mgn/J, Jackson Laboratory, stock #003574) crossed with Bmal1^flx/flx mice (B6.129S4(Cg)-Bmal1-tm1Weit/J, Jackson Laboratory, stock #007668), as initially described.¹⁵ All mice were on a C57BL/6J background. This breeding generated hepatocyte-specific Bmal1 knockout mice (Alb^Cre/+/Bmal1^flx/flx) and their littermate controls (Alb^+/+/Bmal1^flx/flx) referred to as Hep-Bmal1KO and Hep-Bmal1WT, respectively. Genotyping was carried out using specific primers recommended by the Jackson Laboratory. Eight- to 10-week-old male and female mice were housed under a 12-hour light/12-hour dark cycle at 22 °C ± 2 °C, with free access to food and water, and acclimated to the experimental conditions for at least 1 week. Mice were fed either NC (27% protein, 14% fat, 59% carbohydrate, Altromin) or a cdHFD (9% protein, 60% fat, 2% cholesterol, and 31% carbohydrate, with 0.17% methionine, Ssniff E15673-94) for 2 weeks to induce MASH, as previously described.¹⁴ After the diet period, mice were euthanized every 4 hours, starting 2 hours after lights on (ZT 2), for tissue collection, which was immediately flash-frozen in liquid nitrogen and stored at −80°C. The study was ethically approved by the Animal Health and Care Committee of the Government of Schleswig–Holstein, following international animal welfare guidelines. Both male and female mice were used in balanced cohorts for the experiments, except group Hep-Bmal1WT fed with cdHFD at ZT14 (RNA sequencing), where it had one female sample. Data reporting follows Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.

Human Cohort

The study enrolled patients from 2 different medical centers, the Department of General, Visceral, Thoracic, Transplantation, and Pediatric Surgery, University Medical Center Schleswig-Holstein, Kiel; and the Department of Surgery, University Medical Centre Schleswig-Holstein, Campus Lübeck. All patients with obesity had undergone surgical liver biopsy during the bariatric intervention. The patients were aged >18 years and fit the criteria for obesity surgery according to current guidelines, namely, body mass index ≥40 kg/m² or ≥35 kg/m² and at least 1 cardiometabolic comorbidity (hypertension, type 2 diabetes mellitus, dyslipidemia, or obstructive sleep apnea syndrome). Preoperative evaluation included a detailed medical history, physical examination, and nutritional, metabolic, cardiorespiratory, and psychological assessment. Exclusion criteria were autoimmune, inflammatory, or infectious diseases, viral hepatitis, cancer, or known alcohol consumption (>20 g/day for women and >30 g/day for men). Groups were categorized according to liver histology. Clinical data and type of bariatric surgery are summarized in Supplementary Table 5. All patients provided written informed consent. The local ethics committee approved the present clinical investigations. The study was conducted in accordance with the Declaration of Helsinki. Liver samples were either snap-frozen directly in liquid nitrogen in the operating theatre or stored in phosphate-buffered saline (PBS) on ice for a short period for transport before freezing. Samples were stored at −80 °C until further processing.

Bulk Transcriptome Analyses

Transcriptome analysis was performed using the Bulk RNA barcoding and sequencing (BRB-seq). In this method, only the 3′ portion of the mRNA is sequenced, thus reducing the need to sequence the whole transcript, and it has a similar efficiency compared with the TruSeq method.⁴⁰ RNA from livers was extracted using Trizol according to the manufacturer's recommendation. High-quality RNA with ratios (260/280 and 260/230 ratios) higher than 1.8 were used. RNA integrity was assessed by gel electrophoresis. Similar RNA input was used to generate complementary DNA (cDNA) libraries using the MERCURIUS kit (Alithea Genomics). Libraries were sequenced on the Illumina NovaSeq 6000 platform at a depth of 8 million raw reads per sample. The sequencing reads were demultiplexed using the BRB-seq tools suite and aligned against mouse genome (mm10) using STAR, and count matrices were generated using HTSeq.⁴⁰ FeatureCounts was used to count the read numbers mapped to each gene.

Lipidomics Analysis of the Liver via Liquid Chromatography Coupled to Tandem Mass Spectrometry

Liver tissue, obtained from ZT6, from male and female mice was cryo-ground using a cell crusher device. Fifty microliters of water were added to 20 mg of tissue powder. Serum (20 μL) was obtained from ZT 10. Lipidomics analyses were performed as previously described.¹⁹ In brief, measurements were conducted with a Dionex Ultimate 3000 RS LC-system and an Orbitrap mass spectrometer, using high-quality solvents and a heated-electrospray ionization probe. Lipid profiling involved extracting lipids from tissue homogenate using a specific solvent mix (1 mL or 0.4 mL for tissue and serum, respectively). Dried extracts were reconstituted in 50 μL or 20 μL of methanol/isopropyl alcohol (1:1, v/v) for tissue and serum, respectively, followed by liquid chromatography coupled to tandem mass spectrometry analysis on an Accucore C30 RP column. Data were acquired with data-dependent mass spectrometry² scans, and lipids were identified using Compound Discoverer 3.3/3.5 and 2 additional databases.⁴¹ The area under the peak was normalized to the internal standard and, in the case of tissue, to sample weight. An extraction blank and quality control samples were used to ensure data quality and linearity for both tissue and serum samples. Raw data were log2 transformed. To compare all groups, 2-way analysis of variance (ANOVA) followed by Tukey’s honest significant difference, corrected by multiple comparisons was used. For pair-wise comparison, Welch Student’s t test was used. Validation of hepatic cholesterol levels was performed using a commercial kit (STA 384; Cell Biolabs). Two-way ANOVA followed by Tukey’s post-test was used to analyze the data. In both cases, Padj < .05 was deemed significant. For serum lipidomics, sample 108,025 was considered an outlier and excluded.

Hepatic Cholesterol Evaluation

Total hepatic cholesterol was processed according to the manufacturer’s instructions (Cell Biolabs, STA 384). In short, liver samples (approximately 10 mg) from male and female mice were homogenized in 0.2 mL of extraction buffer (chloroform:isopropanol:NP40, 7:11:0.1) using a bead-based homogenizer. Following centrifugation at 15,000 × g for 10 minutes at room temperature, the supernatant was carefully transferred to new tubes, ensuring avoidance of the pellet. Samples were incubated at 50 °C for 30 to 60 minutes and vacuum-dried at 45 °C for 30 to 60 minutes. Dried lipids were resuspended in 200 μL of 1× assay diluent. The cholesterol reaction reagent was freshly prepared by combining cholesterol oxidase (1:50), horseradish peroxidase (1:50), colorimetric probe (1:50), and cholesterol esterase (1:250) in 1× assay diluent. Cholesterol standards were created by serial dilution of a 250-μM cholesterol stock solution in 1× assay diluent. Fifty microliters of samples diluted 1:20 were added to a 96-well plate, followed by 50 μL of cholesterol reaction reagent. After incubation at 37 °C for 45 minutes in the dark, absorbance was measured at 555 nm, and cholesterol levels were determined using a standard curve. Hepatic cholesterol was normalized by the amount of tissue. Permutation test using the package lmPerm in R studio was used to compare human data.

Histologic Analyses

For hematoxylin and eosin staining, mouse liver samples were collected, rinsed in PBS, and fixed in 4% paraformaldehyde (Electron Microscopy Sciences) for 48 hours. Samples from male and female ZT 6, 10, 18, and 22 were used. Subsequently, the samples were dehydrated through a graded ethanol series. The livers were then embedded in paraffin and sectioned into 2- to 5-μm slices. Following dewaxing and rehydration, the slices were stained with hematoxylin and eosin for examination under a light microscope. Elastica van Gieson staining and Gömöri Silver Impregnation was used. A senior pathologist, who was blinded to the sample identities, assessed the sections and assigned scores using the NAFLD Activity Score and the Steatosis, Activity, and Fibrosis scoring systems. A Mann-Whitney test was used to compare the groups.

Rhythm Analyses

Genes containing a sum of reads among the groups lower than 100 were excluded from the analysis. The principal component analysis method was used to assess possible outliers. No sample exclusion was performed, and all 96 samples were considered for the downstream analysis. The remaining genes were log2 transformed in DESeq2 using the vs function.⁴² A total of 15,548 transcripts were considered for subsequent analyses. To analyze the diurnal patterns in gene expression, we employed 4 distinct algorithms: CircaN,⁴³ JTK cycle,⁴⁴ Metacycle,⁴⁵ and DryR.⁴⁶ CircaCompare was utilized for the comparative analysis of rhythmic parameters across different groups.⁴⁷ For the evaluation of mesor and amplitude, CircaCompare was applied to fit a sine curve without the necessity for rhythmicity criteria fulfillment. Phase comparisons were specifically conducted for genes exhibiting robust rhythmicity. The algorithms CircaN, JTK cycle, and Metacycle were run within a CircaN framework, using standard settings and assuming a fixed 24-hour period. We established a threshold for rhythmicity significance at P < .01 for each analytical method. Subsequently, rhythmic genes identified in at least 1 of these methods were aggregated into a unified list for each group, and rhythm parameters were compared utilizing the CircaCompare algorithm using a P < .05.

Enrichment Analysis

Gene lists were analyzed using the ClusterProfiler package in R environment, with the entire identified genes serving as the background dataset. For cases where identified biological processes exceeded 100 hits, the simplify function was applied with a cutoff value of 0.6 to streamline the results. In scenarios with fewer processes, only those involving more than 3 genes and exhibiting a P < .01 were considered for further analysis.

Chromatin Immunoprecipitation–X Enrichment Analysis Version 3

TF prediction was performed using the online application ChIP-X Enrichment Analysis Version 3 (ChEA3; https://maayanlab.cloud/chea3/)¹⁶ using DRGs as input. Metabolic DRGs were selected using the Kyoto Encyclopedia of Genes and Genomes term “Metabolic Pathways.” Mean library rank was used. TFs were filtered for a score < 400.

In Vitro Experiments

AML-12 cells, obtained from ATCC Biobank (catalog no. CRL-2254), were cultured in Gibco Dulbecco’s Modified Eagle Medium supplemented with Nutrient Mixture F12 (ThermoFisher), containing 1% penicillin–streptomycin, 1% insulin–transferrin–selenium (ThermoFisher), 10% non–heat-inactivated fetal bovine serum (FBS; ThermoFisher), and 10 nM dexamethasone (Sigma-Aldrich). Cells were incubated at 37 °C in a humidified atmosphere with 5% CO₂. To mimic a steatosis condition, cells were cultivated in media without dexamethasone, and FBS was replaced by charcoal/dextrane-treated FBS (Hyclone). One hundred thousand cells were seeded per well in 12-well plates and 24 hours later received palmitate-BSA (0.25 mM, 7:1 Cayman Chemical). On the following day, cells were transfected with esiRNA against Bmal1 (100 nM), Chrebp (100 nM), or both genes (200 nM) in addition to the negative control esiRNA (200 nM Egfp, Euphoria Biotech) using lipofectamine 3000 according to the manufacturer's recommendation (Thermofisher). Twenty-four hours after esiRNA addition, cells were collected and processed for transcriptome analysis (BRB-seq—see above) or flow cytometry, as described next.

Dual triglyceride and cholesterol labeling was performed as described before.¹⁹ In brief, cells were detached from the wells using Trypsin-EDTA (ThermoFisher), pelleted by centrifugation (500 × g per 5 min) and incubated sequentially with Zombie NIR dye (1:500 dilution, Biolegend), Bodipy-Cholesterol (1 μM, MedChem Express), and AdipoRed (0.3 μL per mL, Lonza) at room temperature. Each incubation step was combined with centrifugation (1000 × g per 3 minutes) and washing with PBS. Cells were then fixed in 4% paraformaldehyde for 30 minutes at room temperature and stored for flow cytometry analysis. Samples were assessed in Cytek Aurora at the Cell Analysis Core Facility (CAnaCore facility at the University of Lübeck). Data were acquired using predefined templates with adjustments for optimal event rates and detector gains. Spectral unmixing was applied using the manufacturer’s software with single-stained controls to resolve fluorescent populations. At least 10,000 events were captured for each sample. Gating strategies were implemented using FlowJO V10 software, with each gate clearly defined in terms of fluorescence.

Differentially Expressed Genes and Clustering Strategy

DESeq2⁴² was used to analyze DEGs based on raw count data. Genes with counts lower than 25 were excluded. Pairwise comparisons, such as BSA vs palmitate, were performed using the Wald test with standard parameters (Padj < .1). To identify genes regulated by Bmal1 and Chrebp, the likelihood ratio test was conducted in DESeq2. Unsupervised clustering was then applied using the complete method and Euclidean distance method. Bmal1- and Chrebp-specific target genes were further refined by applying a log2 fold change threshold (0.32) and ensuring that each gene was exclusively expressed in only 1 group. The resulting clusters were then subjected to functional enrichment analysis using clusterProfiler.

Quantitative Polymerase Chain Reaction

RNA purity was confirmed by obtaining 280/260 and 260/230 absorbance ratios greater than 1.8 using a spectrophotometer. Up to 2 μg of the total RNA was reverse transcribed with random hexamer primers using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). qPCR was performed using the Go Taq qPCR Master Mix (Promega) using 50 ng of cDNA. The following amplification program was used on a Bio-Rad CFX96 cycler (Bio-Rad): 5 minutes at 94 °C, 45 cycles of 15 seconds at 94 °C, 15 seconds at 60 °C, and 20 seconds at 72 °C, and final extension for 5 minutes at 72 °C. After amplification, a melt curve was generated to verify product specificity by heating the product from 65 °C to 95 °C at 0.5 °C/s. For human gene expression, the TaqMan method with specific genes and probes (IDT) was employed. The qPCR conditions included an initial activation step of 5 minutes at 95 °C, followed by 15 seconds at 95 °C, and then 45 cycles of 60 seconds at 59 °C, and 30 seconds at 72 °C. All genes and probes are detailed in Table 1. Relative expression ratios for each transcript were calculated based on individual primer efficiencies using the Pfaffl method.⁴⁸

Table 1.

Primers and Probes for qPCR

Gene name	Access number	Sequence forward (5′–3′)	Sequence reverse (5′–3′)	Probe (5′–3′)
Bmal1 (mouse)	NM_007489.4	ATC AGC GAC TTC ATG TCT CC	CTC CCT TGC ATT CTT GAT CC
Chrebp (mouse)	NM_021455.5	CCT GCA TCG ATC ACA GGT CA	AGA CCA GCT TGC CAC TGT AAG
Eef1a1 (mouse)	NM_010106.2	TGC CCC AGG ACA CAG AGAC TTC A	AAT TCA CCA ACA CCA GCA GCA A
BMAL1 (human)	NM_001178.6	GCC ACC AAT CCA TAC ACA GA	AGT ATC TTC CCT CGG TCA CA	/5HEX/AA ACA CCT C/ZEN/A TTC TCA GGG CAG CA/3IABkFQ/
PER3 (human)	NM_001289862.2	TCA TCA CCC TAC AGC TCC TAT C	GCT TGT GCT TCC CTT TCC T	/5TEX615/TC AGC AAG AAA GCA GGA GCA AAG C/3IAbRQSp/
NR1D2 (human)	NM_005126.5	GGT AAT CCC AAG AAT GGT GAT CT	CAC AGT AGA ACC ATG CCA CTA A	/5Cy5/AT CGA GTG C/TAO/A CCT GGG ATG ACA AA/3IAbRQSp/
HPRT (human)	NM_000194.3	CGA GAT GTG ATG AAG GAG ATG G	GTA ATC CAG CAG GTC AGC AA	/56-FAM/AC AGA GGG C/ZEN/T ACA ATG TGA TGG CC/3IABkFQ/

qPCR, quantitative polymerase chain reaction.

Carbohydrate-Responsive Element-Binding Protein Chromatin Immunoprecipitation–Polymerase Chain Reaction

Chromatin immunoprecipitation (ChIP) was performed as described previously.⁴⁹ In short, small pieces of frozen liver tissue were pulverized in liquid nitrogen and underwent cross-linking in 1% formaldehyde for 10 minutes, followed by quenching with 1/20 volume of 2.5 M glycine solution. Nuclear extracts were prepared by homogenizing in 20 mM HEPES, 0.25 M sucrose, 3 mM MgCl2, 0.2% IGEPAL CA-630, 3 mM β-mercaptoethanol, complete protease inhibitor tablet (Roche). Chromatin fragmentation was performed by sonication in 50 mM HEPES, 1% sodium dodecyl sulfate, and 10 mM EDTA, using a Bioruptor (Diagenode) for 20 cycles of 30 seconds at the highest level. Cross-linked chromatin was immunoprecipitated in 50 mM HEPES/NaOH at pH 7.5, 155 mM NaCl, 1.1% Triton X-100, 0.11% Na-deoxycholate, 1 mM EDTA, and complete protease inhibitor tablet, using 3 μg anti-ChREBP (Novus Bio NB400-135) antibody overnight. Input sample was taken before adding the antibody. Antibodies were precipitated with a 33% slurry of precleaned protein A Sepharose beads in 0.5% BSA/PBS for 2 hours. Crosslinking was reversed overnight at 65 °C, and DNA was isolated using phenol/chloroform/isoamyl alcohol extraction. Enrichment of genomic sites in input and ChREBP ChIP was determined by qPCR, normalized to a site near the Ins gene, and evaluated using standard curves for each primer pair. Data were normalized to a site near the Ins gene (−0.3 kb) that is not bound by ChREBP and percent of input was calculated by relating ChIP results to the abundance of each site in the respective input chromatin.

The following primers were used: mIns −0.3 kb Fw: CTTCAGCCCAGTTGACCAAT; mIns −0.3 kb Rv: AGGGAGGAGGAAAGCAGAAC; mTxnip −0.2 kb Fw: CCGAACAACAACCATTTTCC, mTxnip −0.2 kb Rv: CGTGCACAGTTCTCCCATT; mPklr +0.23 kb Fw: CTCTGCAGACAGGCCAAAG; mPklr +0.23 kb Rv: TGCCAATGGAAGCCTTGTA; mKlf10 +1.3 kb Fw: GCATGTGAACAAAGCGTGAT; mKlf10 +1.3 kb Rv: TGCTCAGGAAGTAGGGGAAA. Two-way ANOVA followed by Bonferroni post-test was used to analyze the data. Padj < .05 was deemed significant.

Correlation Analyses

Hepatic cholesterol levels, gene expression, and clinical data were analyzed for correlations (Spearman) using the corrplot package. Correlations were deemed significant with a P < .05.

Statistical Analyses

Samples were only excluded upon technical failure. Analyses were performed in RStudio (v. 4.2.1) or Prism software (v. 10.3.1). Specific statistical tests are described in each subsection.

Declaration of Generative AI and AI-Assisted Technologies in the Writing Process

During the preparation of this work, the authors used Grammarly to improve readability and language. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.