The hepatocyte IKK:NF-κB axis promotes liver steatosis by stimulating de novo lipogenesis and cholesterol synthesis

Objective Obesity-related chronic inflammation plays an important role in the development of Metabolic Associated Fatty Liver Disease (MAFLD). Although the contribution of the pro-inflammatory NF-κB signaling pathway to the progression from simple steatosis to non-alcoholic steatohepatitis (NASH) is well-established, its role as an initiator of hepatic steatosis and the underlying mechanism remains unclear. Here, we investigated the hypothesis that the hepatocytic NF-κB signaling pathway acts as a metabolic regulator, thereby promoting hepatic steatosis development. Methods A murine model expressing a constitutively active form of IKKβ in hepatocytes (Hep-IKKβca) was used to activate hepatocyte NF-κB. In addition, IKKβca was also expressed in hepatocyte A20-deficient mice (IKKβca;A20LKO). A20 is an NF-κB-target gene that inhibits the activation of the NF-κB signaling pathway upstream of IKKβ. These mouse models were fed a sucrose-rich diet for 8 weeks. Hepatic lipid levels were measured and using [1–13C]-acetate de novo lipogenesis and cholesterol synthesis rate were determined. Gene expression analyses and immunoblotting were used to study the lipogenesis and cholesterol synthesis pathways. Results Hepatocytic NF-κB activation by expressing IKKβca in hepatocytes resulted in hepatic steatosis without inflammation. Ablation of hepatocyte A20 in Hep-IKKβca mice (IKKβca;A20LKO mice) exacerbated hepatic steatosis, characterized by macrovesicular accumulation of triglycerides and cholesterol, and increased plasma cholesterol levels. Both De novo lipogenesis (DNL) and cholesterol synthesis were found elevated in IKKβca;A20LKO mice. Phosphorylation of AMP-activated kinase (AMPK) - a suppressor in lipogenesis and cholesterol synthesis - was decreased in IKKβca;A20LKO mice. This was paralleled by elevated protein levels of hydroxymethylglutaryl-CoA synthase 1 (HMGCS1) and reduced phosphorylation of HMG-CoA reductase (HMGCR) both key enzymes in the cholesterol synthesis pathway. Whereas inflammation was not observed in young IKKβca;A20LKO mice sustained hepatic NF-κB activation resulted in liver inflammation, together with elevated hepatic and plasma cholesterol levels in middle-aged mice. Conclusions The hepatocytic IKK:NF-κB axis is a metabolic regulator by controlling DNL and cholesterol synthesis, independent of its central role in inflammation. The IKK:NF-κB axis controls the phosphorylation levels of AMPK and HMGCR and the protein levels of HMGCS1. Chronic IKK-mediated NF-κB activation may contribute to the initiation of hepatic steatosis and cardiovascular disease risk in MAFLD patients.


INTRODUCTION
Non-alcoholic fatty liver disease (NAFLD), renamed as Metabolic Associated Fatty Liver Disease (MAFLD) [1], is becoming the most common liver disease globally [1]. MAFLD frequently coexists with obesity, type 2 diabetes, and metabolic syndrome and increases the risk of cardiovascular disease [1e3]. Hepatic lipid accumulation in the absence of liver damage and inflammation is the predominant characteristic of benign MAFLD, also referred to as simple steatosis. Simple steatosis can progress to non-alcoholic steatohepatitis (NASH), characterized by steatosis, liver inflammation, and hepatocellular damage. At this stage of MAFLD, the liver is predisposed to develop liver fibrosis, cirrhosis, and hepatocellular carcinoma (HCC) [4]. Compelling evidence indicates a critical role for the activation of the nuclear factor kB (NF-kB) of transcription factors in the progression from simple steatosis to NASH and the other MAFLD-related liver complications, but its contribution to the initiation of hepatic steatosis remains undefined [5e7]. The NF-kB family consists of five members, NF-kB1 (p105/p50), NF-kB2 (100/p52), RelA (p65), Relb and cRel [8]. Activation of the trimeric IkB kinase (IKK) complex is an important step in the canonical NF-kB activation pathway [9]. The IKK complex, composed of NF-kB essential modulator (NEMO) and the subunits IKKa and IKKb, phosphorylates the inhibitory IkB proteins, including IkBa, leading to proteasomal degradation of these inhibitors [10,11]. This enables the NF-kB subunits to translocate from the cytoplasm to the nucleus, where they activate the transcription of an array of genes involved in many different pathways, including inflammation, cell proliferation, and apoptosis. The gene A20 (also called TNF alpha-induced protein 3, TNFAIP3), encoding a ubiquitin-editing protein, is one of the few NF-kB target genes that act upstream of IKK as a negative feedback loop to terminate NF-kB signaling and protects cells from apoptosis [12]. The NF-kB pathway is chronically activated in NASH and its activation in obese individuals is often associated with higher circulating levels of pro-inflammatory cytokines such as TNF, IL-6, and IL-1b [13,14]. Several studies have indicated that hepatic NF-kB activation is an important player in the progression and initiation of MAFLD. Inhibition of NF-kB activity by hepatocytic silencing of the NF-kB subunit p65 has been shown to protect mice from diet-induced steatosis [15]. Chronic activation of hepatocyte NF-kB in a diet-induced MAFLD model or by overexpressing a constitutively active IKKb mutant (Ikk2ca) in hepatocytes exacerbates hepatic steatosis without concomitant hepatocellular damage or lobular inflammation [16,17]. Furthermore, increased expression of A20 has been identified in steatotic human liver samples [18]. Although these studies suggest a causal relationship between hepatic NF-kB activation and MAFLD development, the contribution of NF-kB activation to the initiation of hepatic steatosis and the underlying mechanism remains unclear.
To study the role of the pro-inflammatory NF-kB signaling pathway in the initiation of hepatic steatosis, we used mice that expressed a constitutively active form of IKKb exclusively in hepatocytes with or without hepatocytic A20. This study identified a hepatocyte-specific role for the NF-kB signaling pathway in the initiation of MAFLD, distinct from its role as a key regulator of inflammation. We showed that IKK-mediated activation of hepatocyte NF-kB induces de novo lipogenesis (DNL) and cholesterol synthesis, leading to hepatic accumulation of triglycerides and cholesterol, accompanied by high plasma cholesterol levels. We further report that the IKK:NF-kB axis controls the phosphorylation levels of 5 0 AMP-activated protein kinase (AMPK) and hydroxymethylglutaryl-CoA reductase (HMGCR) and the protein levels of hydroxymethylglutaryl-CoA synthase (HMGCS1). Thus, the hepatic NF-kB signaling pathway acts as an initiator of hepatic steatosis and cholesterol synthesis, hence may also contribute to cardiovascular disease risk in MAFLD patients.

MATERIAL AND METHODS
2.1. Mice B6(Cg)-Gt(ROSA)26Sor tm4(Ikbkb)/Rsky /J mice were obtained from Jackson Laboratory, California, USA (#008242) [17,19,20]. To express the constitutively active form of IKKb specifically in hepatocytes, these mice were crossed with Alb-Cre-transgenic mice, which were purchased from Jackson Laboratory (#003574) to obtain mice homozygous for the Ikk2ca transgene and hemizygous for the Alb-Cre allele (Hep-IKKbca). Previously described mice carrying the floxed A20 allele [20] were crossed with Hep-IKKbca mice to generate mice expressing IKKbca specifically in hepatocytes in a hepatocyte A20deficient background (IKKbca;A20 LKO ). All experiments were performed on male mice and all strains were maintained on a C57BL/6J genetic background. Mice were housed individually in climatecontrolled rooms with a 12-hour light/12-hour dark cycle and fed ad libitum with a standard laboratory show diet (RM1, Special Diet Services, UK) or a carbohydrate-rich diet (33.5% sucrose, 29.9% corn starch, 3.3% Lodex 10 (Research Diets D12450Bi), as indicated. The mice received a carbohydrate-rich diet at an age of 12 weeks. In all experiments, littermates without the Alb-Cre allele were used as wildtype controls (WT). Mice were sacrificed after a 4-hour fast, and tissues were snap-frozen in liquid nitrogen and stored at À80 C until further analysis. Blood was collected by cardiac puncture, and plasma was collected after centrifugation at 1,000 g for 10 min at 4 C. Certain cohorts of mice were anesthetized by i.p. injection and subjected to gallbladder cannulation for 30 min as described previously [21]. During bile collection, body temperature was stabilized using an incubator. Bile was stored at À80 C until further analysis. All animal studies were approved by the Dutch Central Committee for Animal Experiments and the Animal Care and Use Committee of the University of Groningen, The Netherlands. Studies with the liver-specific A20 deficient mice were performed in a pathogen-free animal facility of the VIB Inflammation Research Center. Hepatocyte A20-deficient mice were fed on the carbohydrate-rich diet (33.5% sucrose, 29.9% corn starch, 3.3% Lodex 10 (Research Diets D12450Bi) for 8 weeks, and all experiments were performed according to institutional, national, and European animal regulations.

Glucose tolerance test and insulin levels
A glucose tolerance test (GTT) was performed as previously described [22]. Briefly, mice were fasted overnight and injected intraperitoneal with a low dose of glucose (100 mg/kg). Blood glucose concentrations were measured at indicated time points in a blood drop collected by tail-tip bleeding. At the end of the experiment, an orbital puncture was performed to obtain a blood sample (Crystalchem, #90010/90020) for insulin measurements.

Gene expression analysis
Frozen liver tissue (AE100 mg) was homogenized in QIAzol Lysis Reagent (Qiagen), and total RNA was isolated according to the manufacturer's instructions. Two micrograms of total RNA were used for cDNA synthesis, according to the manufacturer (Invitrogen, #28025013) protocols. Twenty nanograms of cDNA were used as a template for quantitative real-time PCR (qRT-PCR) analysis using SYBR Green Master (Roche) and qRT-PCR primers (Supplemental Table 1). QRT-PCR data were analyzed using QuanStudioÔ Real-Time PCR software using the deltaedelta Ct method, and expression was normalized to Ppia expression and expressed as fold change compared to WT mice. Total RNA used for RNA sequence analysis was isolated from frozen liver tissue using the RNAeasy plus kit (Qiagen, #74134) according to the manufacturer's instructions. Quantity and quality were assessed using nanodrop and gel electrophoreses, respectively. RNA samples were used for library preparation and sequencing, performed and analyzed by Novogene Co. Ltd. Europe. The Genome Reference Consortium Mouse Build 38 Organism (GRCm38) genome and gene model Original Article annotation files were downloaded from the genome website browser (NCBI/UCSC/Ensembl) directly. Paired-end clean reads were aligned to the reference genome using HISAT2 software. HTSeq was used to count the read numbers mapped to each gene, including known and novel genes. Reads per kilobase of exon model per million mapped reads of each gene were calculated on the length of the gene and read count mapped to this gene (Supplemental Table 2). Differential expression analysis between two groups (n ¼ 6 per group) was performed using the DESeq2 R package. DESeq2 provides statistical routines for determining differential expression in digital gene expression data using a model based on the negative binomial distribution. The resulting P-values were adjusted using Benjamini and Hochberg's approach for controlling the False Discovery Rate (FDR). Genes with an adjusted P-value <0.05 found by DESeq2 were assigned as differentially expressed. RNA sequencing data were subsequently used to construct heatmaps and perform GSEA. Heatmaps were constructed based on the z-score calculated from the raw FPKM scores of individual genes per mouse. Gene Set Enrichment Analysis (GSEA) for beta-oxidation was performed using the Gene Ontology ID: GO:0006635 dataset and RNA expression data were ranked according to the log2FC score of gene expression. Enrichment score (ES) and normalized ES (NES) were determined and GSEA was performed using GSEA 4.1.0 software (Broad Institute and UC San Diego).

Histological analysis
For histological analysis, liver samples were fixed in 4% paraformaldehyde or snap-frozen in liquid nitrogen. Paraformaldehyde samples were embedded in paraffin and cut into sections of 4 mm for hematoxylin-eosin (H&E), F4/80, CD11B, B220, CD3, and Pico Siriusred staining (SR). Frozen samples were cut into sections of 5 mm and used for Oil-Red-O (ORO) staining (Sigma Aldrich, #O0625). Images of the H&E, ORO, and SR sections were obtained with the Leica DM3000 microscope with a mounted Leica DFC420 camera, while F4/80, CD11B, B220, and CD3 staining was imaged using a Hammatusu NanoZoomer (Hamamatsu Photonics, Almere, the Netherlands). Scoring of steatosis and lobular inflammation was done unbiasedly by a board-certified veterinary pathologist of the Dutch Molecular Pathology Center (UU, Utrecht, The Netherlands) according to a method described previously [23]. The immunohistochemically stained liver sections were analyzed as follows: the number of positive cells was counted in 5 (F4/80) or 10 (CD3, B220, CD11b) 40x fields for each case. The average number of positive cells per field was calculated for each case, and an average per group based on genotype was determined finally. Only cells with lymphocyte morphology were counted for CD3, whereas for B220, CD11b, and F4/80 all the positive cells were counted (for F4/80 most of the positive cells were represented by Kupffer cells). Antibodies used for immunohistochemical staining are indicated in Supplemental Table 3. 2.6. Cholesterol and triglyceride analysis in plasma and liver Lipids were extracted from liver homogenates (15% wt/vol in PBS) according to the Bligh & Dyer method [24] and subsequently dissolved in 2% TX100 to determine triglycerides and cholesterol concentration. Total cholesterol (TC) levels were determined using colorimetric assays (Roche, #11489232), and cholesterol standard FS (DiaSys Diagnostic Systems GmbH, #113009910030) was used as a reference. Triglyceride levels were measured using a Trig/GB kit (Roche, #1187771), and Precimat Glycerol standard (Roche, #16658800) was used as a reference.
2.7. De novo lipogenesis, chain elongation, fatty acid determination, and cholesterol synthesis [1e 13 C]-acetate (2% wt/vol) (99 atom %, Isotec/Millipore Sigma) was added to drinking water for five days. De novo lipogenesis (DNL) and fatty acyl chain elongation were determined as described previously [25,26]. Blood spots were collected every 12 h via tail bleeding on filter paper to determine the fractional cholesterol synthesis rates. TC was extracted from blood spots using ethanol/acetone (1:1 v/v). Cholesterol synthesis measurements were performed as previously described [26].

Fatty acid oxidation
Acylcarnitines concentrations were measured in liver homogenate samples, as previously described [27].
2.9. Fast-performance liquid chromatography (FPLC) and targeted proteomics Individual plasma samples were fractionated by FPLC, as previously described [28]. The samples were pooled and the following fractions 36e40, 41e45, 46e50, and 50e55 were used for targeted liquid chromatography-mass spectrometry analysis to determine Apolipoprotein B (ApoB) protein in FPLC fractions, as previously described [28].
2.10. Liver enzymes, ATP concentration, and plasma bile acids Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were determined in plasma using a clinical chemistry analyzer (Cobas 6000, Roche Diagnostics) with standard reagents (Roche Diagnostics). Plasma bile acid concentrations were determined according to the protocol provided by the manufacturer (Crystal Chem, #80471). ATP concentrations were determined in liver homogenates according to the manufacturer's protocol (Roche, #11699695001) and were normalized to total protein concentration (Biorad, #5000113 and #5000114).

Statistics
All data were presented as the mean AE SEM. Statistical analyses were performed using GraphPad version 8.4.0 Software, and an unpaired 2tailed Student's t-test was used to compare differences between any two groups. P-value of less than 0.05 was considered statistically significant: *P < 0.05, **P < 0.01, and ***P < 0.001.

Chronic activation of the hepatocytic NF-kB signaling pathway leads to steatosis
To evaluate the contribution of the pro-inflammatory NF-kB signaling pathway in the development of MAFLD, mice expressing a constitutively active IKKb mutant in hepatocytes (Hep-IKKbca) were generated. Transgenic mice carrying cDNA encoding Flag-tagged IKKb containing Relative liver mRNA expression of inflammatory markers (n ¼ 15e16). (H) Glucose tolerance test (GTT) and (I) plasma insulin levels at t ¼ 120 of the GTT in WT and Hep-IKKbca mice (n ¼ 7). Data are presented as mean AE SEM, *P < 0.05, **P < 0.01, ***P < 0.001 as determined by Student's t-test.
Original Article 4 two serine to glutamate substitutes (S177E, 181E) in the kinase domain of IKKb [19] were crossed with Alb-cre transgenic mice to remove the loxP-flanked STOP cassette in order to express the Ikk2ca transgene in hepatocytes. The hepatic expression of IKKb was confirmed using quantitative RT-PCR (qRT-PCR) ( Figure 1A) and immunoblotting using an antibody directed against the Flag-tag ( Figure 1B). IKK-mediated NF-kB activation results in translocation of the subunit p65 from the cytoplasm to the nucleus to transactivate the expression of NF-kB-target genes [8]. Therefore, we analyzed the nuclear levels of hepatic p65 in wildtype (WT) and Hep-IKKbca mice. Nuclear p65 levels were increased in Hep-IKKbca mice compared to WT mice ( Figure 1C). In addition, cytosolic p65 levels were increased in Hep-IKKbca compared to WT mice ( Figure 1C), which conforms to previous observations [19]. In line with the increased nuclear p65 levels, the expression of the NF-kB targets genes: Traf-1, Ciap1, Nfkbia (Ikba), and A20 showed significant upregulation in Hep-IKKbca mice ( Figure 1D). Body and liver weight and hepatic lipid levels were not different between Hep-IKKbca and WT mice fed a standard chow diet ( Figure S1AeD). Histological examination of the livers did not identify liver steatosis, inflammation, and damage in Hep-IKKbca and WT mice ( Figure S1E). Altogether, these data indicate successful activation of the NF-kB signaling pathway in hepatocytes, and chronic activation of this pathway does not result in an overt liver phenotype. In contrast to our observations ( Figure S1D and E), hepatic IKKbca expression has previously been shown to result in increased hepatic triglyceride (TG) levels [17]. These phenotypic differences prompted us to stimulate hepatic de novo lipogenesis (DNL) by feeding the mice a carbohydrate-rich diet (33.5% sucrose, 29.9% corn starch, 3.3% Lodex 10) [29] for 8 weeks. Feeding the mice the carbohydrate-rich diet resulted in a significant increase in hepatic TG concentration in Hep-IKKbca mice compared to WT mice ( Figure 1E), which was confirmed by ORO staining ( Figure 1F). In addition, the hepatic cholesterol concentrations tended to be elevated in Hep-IKKbca mice ( Figure 1E). Although IKK-mediated NF-kB activation is generally strongly associated with inflammation, the hepatic mRNA levels of the cytokine Il1b and the macrophage marker CD68 were only mildly elevated in Hep-IKKbca mice, while no differences in the expression of other inflammatory markers, such as Tnf, Ila, Il6, Ccl2 were observed ( Figure 1G). Histological examination of the livers did not reveal a substantial difference in the number of inflammatory foci between WT and Hep-IKKbca mice ( Figure 1F, Table S4). Furthermore, immunohistochemical staining for the macrophage marker F4/80, monocytederived macrophage (MdM) marker CD11b, B-cell marker B220, and T-cell marker CD3 ( Figure S2A and B) revealed no difference in liver inflammation. The increased hepatic activity of IKK can lead to glucose intolerance and insulin resistance [30], but Hep-IKKbca mice did not display any glucose intolerance, as determined by GTT ( Figure 1H). Plasma insulin levels at the end of the GTT showed no difference between the groups ( Figure 1I). Furthermore, no differences in the expression of gluconeogenic genes between Hep-IKKbca and control mice ( Figure S2C) were observed. Overall, these data suggest that chronic activation of the NF-kB signaling pathway by hepatocytic expression of IKKbca sensitizes mice to develop hepatic steatosis without causing severe inflammation and glucose intolerance.

Steatosis in
Hep-IKKbca mice is not caused by altered expression of genes related to lipogenesis, cholesterol synthesis, or fatty acid oxidation Next, using RNA-seq analysis, we compared the transcriptomes of Hep-IKKbca livers with WT livers to elucidate the mechanism underlying the steatotic phenotype of Hep-IKKbca mice fed the carbohydrate-rich diet. In contrast to a previous study [17], we found no significant differences in the expression of lipogenic genes between WT and Hep-IKKbca mice (Figure 2A). In addition, cholesterogenic genes were not differentially expressed upon hepatocytic IKKbca expression ( Figure 2B). Pharmacological inhibition of IKKb has been shown to prevent hepatic steatosis and inflammation by improving boxidation [31]. To assess whether b-oxidation is affected in Hep-IKKbca mice, we analyzed the expression of genes related to fatty acid oxidation. No marked expression differences were observed between WT and Hep-IKKbca mice ( Figure S3A and B). In line with this observation, the content of acylcarnitines in livers of Hep-IKKbca mice was not affected ( Figure S3C). These findings indicate that hepatic lipid accumulation in Hep-IKKbca mice is not caused by impaired boxidation or altered expression of lipogenic and cholesterogenic genes.
3.3. Hepatic A20 deficiency exacerbates steatosis in Hep-IKKbca mice In agreement with the study of Lu et al. [17], hepatic IKKbca expression induced the expression of numerous genes of the NF-kB pathway, including the NF-kB subunits Relb and NF-kB1 ( Figure 3A). Moreover, the NF-kB target gene A20, which acts upstream of IKK to terminate NF-kB activation [32], was significantly increased ( Figure 3A). This increase was confirmed by qRT-PCR analysis; we found that the A20 mRNA levels were 13-fold higher ( Figure 3B). An increase in A20 mRNA levels was also seen in Hep-IKKbca mice fed a standard chow diet ( Figure 1D), suggesting that the NF-kB-mediated A20 transactivation is diet-independent. We hypothesized that this marked increase in A20 mRNA levels dampens liver inflammation in Hep-IKKbca mice. To examine the role of A20 in preventing liver inflammation in Hep-IKKbca mice, we ablated A20 in hepatocytes of Hep-IKKbca mice (IKKbca;A20 LKO mice) by crossing the Hep-IKKbca mice with A20 floxed mice [20]. Although we successfully deleted hepatic A20 in Hep-IKKbca mice (Figure 3BeD), IKKbca;A20 LKO mice fed the carbohydrate-rich diet showed no marked liver inflammation compared to WT mice, as determined by qRT-PCR analysis ( Figure 3E) and histological and immunohistochemical characterization of liver tissues (Table S5, Figure S4A and B). However, the number of CD11b cells (MdM marker) was significantly increased in liver sections of IKKbca;A20 LKO mice ( Figure S4A and B). A study showed a positive correlation of the number of CD11b cells in livers with MAFLD progression [33]. In line with these results, levels of livers TG and cholesterol concentrations were significantly elevated in IKKb ca;A20 LKO mice compared to WT ( Figure 4A). This hepatic lipid accumulation ( Figure 4A and B) was accompanied by an increase in liver weight in IKKbca;A20 LKO mice, whereas body weight did not differ between the mice ( Figure 4C). The TG and cholesterol concentrations in IKKbca;A20 LKO livers were 2.6-fold and 2.1-fold higher ( Figure 4A) versus 1.8-fold and 1.5-fold higher in Hep-IKKbca livers compared to WT littermates, respectively ( Figure 1E). Hepatic A20-deficient mice (A20 LKO ) without transgenic expression of IKKbca did not display liver steatosis when fed the carbohydrate-rich diet, as no difference in liver TG concentrations were evident ( Figure S4C). Hepatic loss of A20 did not increase NF-kB activity, illustrated by the similar induction of the expression of NF-kB genes, including the NF-kB target genes Nfkbia (Ikba), Nfkbie (Ikbe), Nfkbib (Ikbb) ( Figure S4D). In addition, no differences in the expression of gluconeogenic genes ( Figure S5A), and glucose and insulin response were seen between IKKbca; A20 LKO and control mice ( Figure 4D and E). Taken together, these data strongly suggest that constitutive activation of the IKK complex in hepatocytes increases the susceptibility to develop steatosis, which is aggravated by hepatocyte A20 deficiency, likely due to a further increase in IKK complex activity but not by a further increase in NF-kB activity.

The IKK:NF-kB axis controls lipogenesis and cholesterol synthesis at the post-translational level
To understand mechanistically the hepatic steatosis development upon chronic IKK-mediated NF-kB activation, we quantified hepatic DNL and cholesterol synthesis by allowing incorporation of [1e 13 C]-acetate into fatty acids and cholesterol for five days, followed by mass isotopomer distribution analysis (MIDA) [34], in IKKbca;A20 LKO mice and WT mice. De novo synthesis of palmitoleic (C16:1) and oleic acid (C18:1), the primary fatty acids incorporated in TG, cholesterol esters, and phospholipids were increased in IKKbca;A20 LKO mice compared to WT mice ( Figure 5A). The pools of palmitic acid (C16:0), palmitoleic (C16:1), and oleic acid (C18:1) were elevated in IKKbca;A20 LKO mice ( Figure 5B). Furthermore, cholesterol synthesis in IKKbca;A20 LKO mice was 2-fold higher compared to WT mice ( Figure 5C). This increase in cholesterol synthesis was not only reflected by higher hepatic cholesterol levels ( Figure 4A) but was also associated with elevated total plasma cholesterol levels ( Figure 5D). The latter was mainly due to an increase in low-density lipoprotein (LDL) cholesterol levels, as determined by FPLC analysis ( Figure 5E) and targeted proteomics analysis of ApoB protein content in the FPLC fractions ( Figure 5F). No difference in plasma TG was seen between WT and IKKbca;A20 LKO mice ( Figure 5D). Despite the marked increase in lipogenesis and cholesterol synthesis in IKKbca; A20 LKO mice, the expression of genes related to these pathways (e.g., Acly, Acaca, Hmgcs1, Hmgcr, Sqle) were not different between WT and IKKbca;A20 LKO mice ( Figure S5B and C). Furthermore, the elevation in plasma cholesterol levels in IKKbca;A20 LKO mice could not be explained by a strong protein reduction in any of the well-known lipoprotein receptors, LDLR, LRP1, and SR-BI ( Figure S6A and B). The de novo synthesis of palmitoleic (C16:1) also showed a significant increase in Hep-IKKbca mice, whereas the de novo synthesis of oleic acid (C18:1) and cholesterol synthesis tended to be elevated ( Figure S7A and B). This modest increase in DNL and cholesterol synthesis is in line with the less severe steatotic phenotype of Hep-IKKbca mice compared to IKKbca;A20 LKO mice. AMPK acts as a central player in controlling hepatic lipogenesis and cholesterol synthesis [35,36], and phosphorylation of AMPK inhibits lipogenesis and cholesterol synthesis [37e39]. In line with elevated DNL (Figure 5A), we found a 2-fold decrease in the phosphorylation levels of AMPK in livers of IKKbca;A20 LKO mice compared to WT mice ( Figure 6A and B). However, this did not affect the phosphorylation status of acetyl-CoA carboxylase 1 (ACC1) in IKKbca;A20 LKO mice or the protein levels of fatty acid synthase (FAS), both well-known downstream targets of AMPK ( Figure 6AÀC). Moreover, ATP citrate lyase (ACLY) protein levels À a key player in lipogenesis À were not affected upon IKK-mediated NF-kB activation in IKKbca;A20 LKO mice ( Figure 6A and C), indicating that IKK did not control lipogenesis by promoting the protein stability of ACLY, as shown previously [40]. However, the protein expression of 3-hydroxy-3-methylglutaryl-CoA synthase 1 (HMGCS1) was increased AE5-fold in livers of IKKb ca;A20 LKO mice ( Figure 6A and C), and the phosphorylation status of 3hydroxy-3-methylglutaryl-CoA reductase (HMGCR) at Serine 871 was significantly decreased in livers of IKKbca;A20 LKO mice compared to WT mice ( Figure 6A and D). The reduced levels of phosphorylated AMPK and HMGCR in IKKbca;A20 LKO livers could not be explained by reduced ATP levels ( Figure 6E). Phosphorylated AMPK has been shown to phosphorylate HMGCR, thus inhibiting its activity [35,39]. This suggests that the increased HMGCR activity and the elevated HMGCS1 levels observed in the current study may explain the increased cholesterol synthesis in IKKbca;A20 LKO mice.
3.5. Chronic hepatocytic activation of the IKK:NF-kB axis leads to liver inflammation but does not cause other MAFLD-associated liver complications in middle-aged mice Steatosis can progress to severe liver complications, such as hepatitis, fibrosis, cirrhosis, and HCC. Compelling evidence exists for a significant role of the NF-kB signaling pathway in all these pathologies [17,41]. To investigate the contribution of chronic hepatic IKKmediated NF-kB activation to MAFLD progression, we aged WT and IKKbca;A20 LKO mice to one year of age. Mice started receiving the carbohydrate-rich diet at the age of 12 weeks. We found no differences in body and liver weight, fasted (6 h) blood glucose levels ( Figure S8AeD), and both groups developed macrovesicular steatosis ( Figure 7A, Table S6). Although liver TG levels were not different between the groups, hepatic cholesterol concentrations were significantly increased in IKKbca;A20 LKO mice compared to WT mice ( Figure 7B). Similar to young mice, plasma cholesterol levels were elevated in 1-year-old IKKbca;A20 LKO mice, but no differences in plasma TG levels were seen ( Figure 7C). This increase in hepatic and plasma cholesterol levels was paralleled with a significant increase in HMGCS1 and HMGCR protein levels in IKKbca;A20 LKO compared to WT mice ( Figure 7D and E). The mRNA expression of the ATP-binding cassette (ABC) transporters G5 (Abcg5) and G8 (Abcg8), which form a heterodimer to facilitate biliary cholesterol secretion, were not affected in IKKbca;A20 LKO mice ( Figure S8E), suggesting that the elevated hepatic cholesterol levels in IKKbca;A20 LKO mice are not due to impaired biliary cholesterol secretion. The plasma bile acid concentrations between the two groups showed no significant difference ( Figure S8F), which indicates that bile acid metabolism is not affected upon chronic IKK-mediated NF-kB activation. The expression of proinflammatory cytokines, such as Tnf, Il1a, Il1b, Il6 ( Figure 7F) in livers of middle-aged IKKbca;A20 LKO mice were increased compared to control livers. The mRNA levels of the fibrotic markers, Col1a1 and Timp1, were also elevated ( Figure 7F), but no significant signs of liver fibrosis were observed by Sirius-red staining of liver sections of WT and IKKbca;A20 LKO mice ( Figure S8G and H). The plasma levels of the liver enzymes ALT and AST ( Figure S8I and J) were not different between the two groups, indicating that chronic IKK-mediated NF-kB activation does not lead to impaired liver function or liver damage. Histological examination did not identify any other severe liver complication, such as liver cirrhosis or cancer. Overall, these results indicate that hepatic activation of the NF-kB signaling pathway not only promotes the progression of liver steatosis, likely by increased cholesterol synthesis but also drives liver inflammation at a later stage of MAFLD.

DISCUSSION
Hepatic activation of the NF-kB signaling pathway has been observed in many stages of MAFLD [42], but the role of this pathway in the initiation of liver steatosis remains ill-defined. In this study, we show that increased activity of IKK, an upstream activator of the proinflammatory NF-kB signaling pathway, sensitizes mice to develop Bodyweight, liver weight, and liver to body weight ratios of WT and IKKbca;A20 LKO (n ¼ 15e16). (D) Glucose tolerance test (GTT) and (E) plasma insulin levels at t ¼ 120 of the GTT in WT and IKKbca;A20 LKO mice (n ¼ 7). Data are presented as mean AE SEM, **P < 0.01, ***P < 0.001 as determined by Student's t-test. (F) ApoB levels in pooled FPLC fractionated plasma depicted in (E) was determined by targeted proteomics, indicated as a percentage relative to the total ApoB levels in these fractions. BDL ¼ below detection limit. Data are presented as mean AE SEM, **P < 0.01, ***P < 0.001 as determined by Student's t-test. simple steatosis. This phenotype was exacerbated by the hepatocytic loss of A20, an upstream inhibitor of the IKK complex, as IKKb ca;A20 LKO mice exhibit hepatic accumulation of both TG and cholesterol. In addition, plasma levels of atherogenic LDL cholesterol were increased upon hepatocytic A20 ablation in Hep-IKKbca mice. We demonstrate that the IKK:NF-kB axis controls DNL and cholesterol synthesis post-translationally by regulating the phosphorylation levels of AMPK and HMGCR and the protein levels of HMGCS1. Furthermore, in middle-aged mice, prolonged activation of the NF-kB signaling pathway also drives the severity of MAFLD, illustrated by low-grade liver inflammation. Our data demonstrate a hepatocyte-specific role for the IKK:NF-kB axis in MAFLD distinct from its role as a key regulator of inflammation. Thus, we suggest that the NF-kB signaling pathway is a key player not only in the progression but also in the initiation of MAFLD. A hallmark of obesity is low-grade systemic inflammation, caused primarily by the secretion of inflammatory cytokines by enlarged adipose tissues [5,43]. Besides adipose tissues, microbiota-derived lipopolysaccharide (LPS) also participates in the onset of low-grade inflammation in obese individuals [44]. Diets including highfructose diets have a strong effect on the composition of the gut microbiota, resulting in increased gut permeability and systemic inflammation [45]. This low-grade systemic inflammation can activate the hepatic NF-kB through the IKK complex, which has been associated with the development of systemic and hepatic insulin resistance, MAFLD progression, and cardiovascular diseases [43e 46]. Our study suggests that the IKK complex is also engaged in the induction of simple steatosis and high plasma LDL cholesterol levels, a factor causally related to atherosclerotic cardiovascular disease (ASCVD) [47]. A part of our results is in line with a previous study showing that IKK-mediated NF-kB activation results in hepatic TG accumulation without liver inflammation and fibrosis [17]. Although both studies used an identical transgenic Ikk2ca mouse model, we found that liver TG and cholesterol were both elevated, whereas the hepatocyte-expressing IKKbca mice in the previous study by Lu et al. displayed only a modest accumulation of TG. Lu et al. explained this phenotype by an increased expression of lipogenic genes, including fatty acid synthase (Fasn) and stearol-CoA desaturase-1 (Scd1). However, we found no evidence for a significant change in the expression of genes associated with fatty acid or cholesterol metabolism. Instead, we showed that the IKK:NF-kB axis regulates these two pathways at a post-translational level. These phenotypic differences can likely be explained by the use of different diets, including the different chow diets because we observed Hep- IKKbca mice only developed steatosis on a carbohydrate-rich diet. Sugars, one of the main sources of carbohydrates in the diet, are associated with chronic inflammation and stimulate DNL and induce steatosis in mice and humans [48,49]. Supporting these findings, our WT control mice also developed liver steatosis after prolonged carbohydrate-rich diet feeding for AE40 weeks ( Figure 7A). Furthermore, the importance of additional metabolic stresses to reveal metabolic differences is distinctly illustrated in HMGCR Ser871Ala mutant knock-in (KI) mice [50]. HMGCR is a downstream target of AMPK; phosphorylated active AMPK inhibits HMGCR activity by increasing its phosphorylation state at Ser871 [35,39]. HMGCR Ser871Ala mutant is insensitive for AMPK-mediated repression but the effect of this mutation on cholesterol synthesis was only seen when the mice were challenged with a carbohydrate-rich diet. Similar Representative Immunoblot and (E) quantification of the hepatic levels of the indicated proteins in IKKbca;A20 LKO and WT mice (n ¼ 5). (F) Relative hepatic mRNA levels of proinflammatory and fibrotic markers in WT and IKKbca;A20 LKO mice as determined by qRT-PCR analysis (n ¼ 7e12). Data are presented as mean AE SEM, *P < 0.05, **P < 0.01, ***P < 0.001 as determined by Student's t-test.
to our model, this metabolic stress led to elevated hepatic and plasma cholesterol levels, accompanied by increased hepatic TG levels caused by enhanced de novo lipogenesis. The interaction between hepatic IKK activation and carbohydrates in the development of steatosis is substantiated by the observation that genetic ablation of the Toll-like receptor 4 (TLR4), a TLR that activates the IKK:NF-kB axis, protects mice from fructose-induced liver steatosis by an undefined mechanism [51,52]. Similarly, selective inhibition of tumor necrosis factor receptor 1 (TNFR1), another activator of the NF-kB signaling pathway, reduces liver steatosis in mice fed a high-fat diet supplemented with fructose and sucrose in the drinking water [53]. TNFR1 inhibition results in reduced expression of lipogenic genes due to decreased sterol regulatory element-binding protein 1 (SREBP1) expression and activation. We found that IKK-mediated NF-kB activation controls lipogenesis and cholesterol synthesis at a posttranslation level, a type of regulation that has also been reported in a recent study [40] that reported that IKKb augments lipid metabolism in liver tumors by promoting the stability of ACLY and FAS proteins, both key enzymes in lipogenesis. IKKb stabilizes ACLY and FAS through phosphorylation of the deubiquitinase USP30, promoting deubiquitination of ACLY and FAS by USP30. We observed no changes in the protein levels of ACLY and FAS, indicating that under our conditions the IKK:NF-kB axis controls lipogenesis via an alternative manner. We showed that phosphorylation of AMPK, a key regulator of lipogenesis and cholesterol synthesis, was decreased in IKKb ca;A20 LKO mice ( Figure 6A). Phosphorylated AMPK inhibits DNL [37,54]. Increasing the phosphorylation level of AMPK by polyphenols, metformin, or expressing an active form of AMPK attenuates sugarinduced steatosis development [37,53e55], supporting the role of AMPK in carbohydrate-induced steatosis development. Although our data imply an essential role of AMPK activity, the downstream targets of AMPK, such as ACC1 and FAS, were not affected. This is an interesting observation as these proteins were also not affected in mouse livers expressing a constitutively active AMPK mutant (y1D136A) [37]. AMPK y1D136A mutant mice were protected against hepatic TG accumulation following a carbohydrate-rich diet, which was explained by decreased DNL due to increased phosphorylation of ACC1 [37]. Although in vitro data encouraged this model no in vivo evidence showed that ACC1 phosphorylation was correlated with decreased hepatic lipid accumulation [37]. In addition to ACC1, other proteins of the lipogenesis pathway may be post-translationally modified [40] and controlled by either AMPK or IKK activity under specific metabolic conditions (e.g., carbohydrate-rich diet). As discussed above, HMGCR is a downstream target of AMPK and together with the elevated enzyme levels of HMGCS1, our data suggest that IKK-mediated NF-kB activation promotes cholesterol synthesis by increasing flux through the mevalonate pathway, also known as the HMG-CoA reductase pathway. This is an important finding because a recent study has shown an association of the activity of this pathway with the severity of MAFLD in humans [2]. Similar to our mouse model, decreased phosphorylation levels of AMPK and HMGCR were correlated with elevated cholesterol synthesis accompanied by high hepatic and plasma cholesterol. Presently, it remains unknown whether IKK directly regulates the mevalonate pathway or indirectly via AMPK. Several AMPK activators have been reported to investigate this but several effects induced by these activators could be explained by offtarget effects or by AMPK activation in other organs/cells, which limit the use of AMPK activators to study the direct effect of AMPK activity on metabolic pathways [37,56]. The use of genetic mouse models, such as transgenic and tissue-specific knockout mice, can overcome this issue and are valuable to decipher the mechanism by which IKK regulates the mevalonate pathway. It is interesting to note that chronic hepatic activation of AMPK by expressing AMPK y1D136A mutant only decreases liver TG levels but not cholesterol levels following a carbohydrate-rich diet; it suggests that at certain metabolic conditions, HMGCR activity is not only controlled by AMPK [37], which may indicate that IKK directly controls the mevalonate pathway. It remains poorly understood whether the metabolic effects are mediated by IKK or NF-kB activity, but because the NF-kB-mediated gene transactivation in our model is not further increased upon hepatocytic A20-deficiency and A20 ablation alone does not result in steatosis, we speculate an essential role for the IKK complex in controlling DNL and cholesterol synthesis [40]. A20 acts upstream of IKK and despite the expression of a constitutively active IKKb mutant, hepatocytic loss of A20 in Hep-IKKbca mice aggravates steatosis, which implies that A20 can mediate IKK functioning and its downstream effects via multiple mechanisms under specific conditions. A20 has shown modulating IKK activation via different ways using different proteins and post-translation modifications, such as linear ubiquitin chain assembly complex (LUBAC) and linear polyubiquitination [57]. In addition, other post-translational modifications of IKK components have also been reported [58]. Therefore, we speculate that stabilization and/or additional post-translational modifications of the IKK complex, which may be modulated by A20, control DNL and cholesterol synthesis in hepatocytes. Furthermore, in contrast to IKBKB (IKKb), A20 has been identified as a susceptibility locus for human immune and inflammatory diseases, such as rheumatoid arthritis, type 1 diabetes, coronary artery disease, and Crohn's disease [59]. However, associations of any of these gene polymorphisms with hepatic steatosis remain to be determined. It has been shown that AMPK can inhibit NF-kB activity and, subsequently, inflammation in different immune cells [60]. Inflammation can also suppress AMPK activity in muscle, heart, adipose tissue, and macrophages by different unknown mechanisms [56,61]. For example, in muscle, inflammation induces protein phosphatase 2 (PP2C) expression and inactivates AMPK in a TNFR1-dependent manner [62]. TNF upregulates PP2C, likely through the IKK:NF-kB axis, we documented no difference in hepatic expression of PP2C between control and IKKbca;A20 LKO mice (Table S2). However, the role of other members of the serine/threonine protein phosphatases, such as Ppmih and Ppmin, which were both significantly upregulated (>2-fold) in IKKbca;A20 LKO mice (Table S2), in regulating AMPK activity remains unknown. Additional research is therefore required to provide better mechanistic insights by which the pro-inflammatory NF-kB signaling pathway in hepatocytes may control AMPK activity and the mevalonate pathway and, subsequently, MAFLD severity in patients [2]. Overall, we conclude that the hepatocytic IKK:NF-kB axis plays a critical role in the initiation and progression of MAFLD by controlling lipogenesis and cholesterol synthesis, thus it might contribute to cardiovascular disease risk in MAFLD patients. Although further research is needed, our data suggest that IKK-mediated NF-kB activation regulates the phosphorylation of AMPK and HMGCR and the protein expression of HMGCS1. A better understanding of the underlying mechanism could offer potential targets to prevent and treat MAFLD at several stages of its development, thereby mitigating cardiovascular risk.