To find out if AOAH prevents MASLD, we fed co-housed Aoah^+/+ and Aoah^-/- mice NC or a HFD plus fructose and glucose in the drinking water, for 28 weeks (Figure 1B; Liu et al., 2018). Aoah^-/- mice fed either NC or HFD gained more weight than did Aoah^+/+ control mice (Figure 1C). When they were fed either NC or HFD, the livers of Aoah^-/- mice were heavier than those of Aoah^+/+ control mice (Figure 1D). Histological examination and Oil Red O staining revealed that Aoah^-/- mouse livers accumulated more lipid droplets than did the livers of Aoah^+/+ mice (Figure 1E). When we scored MASLD severity based on steatosis, hepatocyte ballooning degeneration, and inflammation, Aoah^-/- mice developed more severe MASLD than did Aoah^+/+ mice whether the mice were fed NC or HFD (Sheka et al., 2020; Figure 1E). When the mice were fed either NC or HFD, Aoah^-/- mouse livers contained more triacylglycerol (TAG) than did Aoah^+/+ mouse livers, while livers from both mouse strains contained a similar amount of total cholesterol (TCHO, Figure 1F). When the mice were fed the HFD, Aoah^-/- mice had higher serum levels of TAG, TCHO, low-density lipoprotein (LDL), and free fatty acids than did Aoah^+/+ mice (Figure 1G). Collectively, these findings were evidence that AOAH reduced hepatic TAG accumulation when mice were fed either NC or HFD.

Aoah^-/- mice fed HFD had significantly higher serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels than did control Aoah^+/+ mice, suggesting that Aoah^-/- mice experienced more severe liver inflammation and tissue damage (Figure 2A). To assess liver inflammation, we measured pro- and anti-inflammatory cytokine expression. When mice were fed the HFD, Aoah^-/- mouse livers produced more pro-inflammatory Il6, Ifng, and anti-inflammatory Il10 mRNA than did Aoah^+/+ mouse livers, suggesting greater inflammation (Figure 2B). We also found that Aoah^-/- livers had more Timp1 (a pro-fibrosis gene) mRNA and less Mmp2 (an anti-fibrosis gene) mRNA than did Aoah^+/+ mouse livers (Kisseleva and Brenner, 2021), indicating that Aoah^-/- mouse livers may be developing more severe fibrosis, although we did not detect fibrosis with Masson staining (Figure 2C). We analyzed the myeloid cells in the liver (Daemen et al., 2021) and found that when the mice were fed the HFD, Aoah^-/- mouse livers contained more neutrophils, monocytes, and lipid-associated macrophages (hepatic LAMs) (Remmerie et al., 2020; Su, 2002) than did Aoah^+/+ mouse livers (Figure 2D and E). Collectively, when the mice were fed an HFD, the livers of Aoah^-/- mice developed significantly greater inflammatory responses and tissue damage.

We found previously that AOAH was mainly expressed in Kupffer cells in the liver (Shao et al., 2007). To confirm the previous results, we first consulted the single-cell RNA-seq analysis reported by Remmerie et al., 2020, who found that AOAH is expressed in Kupffer cells, monocytes, monocyte-derived cells, NK (circulating NK), and ILC1 (tissue resident NK) cells in mouse livers (Remmerie et al., 2020; Figure 3A). We used flow cytometry to sort mouse Kupffer cells (CD45⁺NK1.1^- F4/80^hiCD11b^mid), monocytes (CD45⁺NK1.1^- F4/80^midCD11b^hi), NK cells (CD45⁺SSC^lo NK1.1⁺) (including circulating NK and resident ILC1), and purified the hepatocytes. Using qPCR analysis, we found that AOAH mRNA was present in Kupffer cells, monocytes, and NK cells but not in hepatocytes, in keeping with previous findings (Shao et al., 2007; Figure 3B). Western blot analysisanalysis confirmed that Aoah^+/+ mouse livers but not Aoah^-/- mouse livers had AOAH protein (Figure 3C). When Aoah^+/+ mice aged or had an HFD, their hepatic AOAH expression increased (Figure 3D). We also studied the expression of AOAH in human liver cells by analyzing the single-cell RNA-seq data obtained by Ramachandran et al., 2019; AOAH was expressed in macrophages, monocytes, resident and circulating NK cells, and some T cells (Figure 3E). AOAH expression increased in liver macrophages and monocytes in MASLD patients (Figure 3E). As gut-derived LPS can be transported via the portal vein into the liver (Han et al., 2021), we hypothesized that AOAH prevents hepatic inflammation and fat accumulation by inactivating LPS in the gut, portal vein, and liver. We found that Aoah^-/- mouse feces, liver, and plasma had higher bioactive LPS levels when Aoah^-/- and Aoah^+/+ mice were fed either NC or HFD (Figure 3F). HFD increased gut permeability, but there was no permeability difference between Aoah^+/+ and Aoah^-/- mice that were fed either NC or HFD for 28 weeks, suggesting that the increased hepatic LPS levels in Aoah^-/- mouse livers were mainly caused by failure to inactivate LPS (Figure 3G). After the mice were fed HFD for 28 weeks, Aoah^-/- mouse livers increased fatty acid uptake gene Fabp3 mRNA and fatty acid synthesis gene Fasn mRNA, changes that may have contributed to lipid accumulation (Figure 3H). Thus, AOAH may prevent hepatic lipid accumulation by diminishing bioactive LPS in the liver.

We found that AOAH reduced hepatic lipid accumulation when mice were about 8 months old if they were fed either NC or HFD. To investigate the mechanism, we analyzed the livers of young mice. We co-housed 3–4-week-old Aoah^+/+ and Aoah^-/- mice for 3–4 more weeks before removing their livers for RNA-seq analysis. The expression of several fatty acid biosynthesis genes, such as Acacb (acetyl-coenzyme A carboxylase beta), Acss2 (acetyl-CoA synthetase 2), Pcx (pyruvate carboxylase), Acly (ATP citrate lyase), Fasn (fatty acid synthase), and Scd1 (stearoyl-coenzyme A desaturase 1), was significantly increased in Aoah^-/- mouse livers (Figure 4A). When we used gene set enrichment analysis (GSEA) to analyze deferentially expressed genes, we found that the fatty acid biosynthesis pathway was enriched (Figure 4B). We then did qPCR and confirmed the increases in mRNAs for FA biosynthesis genes described in Figure 4A as well as for Acaca (acetyl-coenzyme A carboxylase alpha, encoding ACC1, the first and key enzyme on FA synthesis pathway) in Aoah^-/- mouse livers compared with Aoah^+/+mouse livers (Figure 4C). In addition, mRNAs for enzymes involved in fatty acid oxidation (Acot2 [acyl-CoA thioesterase 2] and Ppara [peroxisome proliferator-activated receptor α]) (Bougarne et al., 2018; Moffat et al., 2014) decreased (Figure 4D), while Cd36 (fatty acid uptake; Chen et al., 2022) mRNA levels increased in Aoah^-/- mouse livers (Figure 4E). We confirmed that FASN and SCD1 protein levels also increased in Aoah^-/- mouse livers (Figure 4F). We found previously that LPS and other TLR agonists increase lipid accumulation in cultured macrophages by increasing expression of Acsl1 (acyl-CoA synthetase long-chain family member 1) and Dgat2 (diacylglycerol O-acyltransferase 2) and by reducing the production of Pnpla2 (Patatin-like phospholipase domain containing 2, Atgl) (Huang et al., 2014), yet the livers of Aoah^+/+ and Aoah^-/- mice had similar levels of Acsl1, Dgat2, and Pnpla2 mRNA, suggesting that AOAH does not regulate hepatic TAG metabolism (Figure 4—figure supplement 1). These results suggest that AOAH reduces liver fat accumulation by diminishing the expression of fatty acid synthesis and uptake genes and increasing that of fatty acid oxidation genes.

As Acaca, Fasn, and Scd1 are all target genes for sterol regulatory element-binding protein 1 (SREBP1, encoded by Srebf1 gene), a key transcription factor for fatty acid biosynthesis, we next analyzed SREBP1 expression in the liver (Horton et al., 2002; Shimano and Sato, 2017; Yokoyama et al., 1993). There are two isoforms of SREBP1, SREBP1a, and SREBP1c, and the liver predominantly expresses SREBP1c (Horton et al., 2002). The abundance of Srebf1a and Srebf1c mRNA increased in the livers of young Aoah^-/- mice (Figure 5A). SREBP1 is synthesized as a 125 kDa precursor (full length, flSREBP1) in the endoplasm reticulum, transferred to the Golgi apparatus, and cleaved sequentially by Site-1 protease and Site-2 protease to generate nuclear SREBP1 (nSREBP1, 68 kDa), which enters the nucleus and activates fatty acid biosynthesis gene transcription (Horton et al., 2002; Shimano and Sato, 2017). The livers mainly had the short-form (68 kDa) nSREBP1 protein, which was significantly more abundant in Aoah^-/- mice (Figure 5B). We separated liver cytosol and nuclei and found that the short-form SREBP1 was mainly present in nuclei and that Aoah^-/- mouse liver nuclei contained significantly more SREBP1 than did Aoah^+/+ mouse liver nuclei (Figure 5C).

When we analyzed the transcript profiles from co-housed 6–8-week-old Aoah^+/+ and Aoah^-/- mouse livers, we noticed that the expression of inflammatory genes for serum amyloid A1 (Saa1), Saa2, and Saa3 was significantly higher in Aoah^-/- mouse livers (Figure 5—figure supplement 1A). We isolated hepatocytes from Aoah^+/+ and Aoah^-/- mice and found that Aoah^-/- mouse hepatocytes expressed increased levels of Saa1, Saa2, Saa3, and inflammation regulatory Irak3 mRNA (Figure 5—figure supplement 1B). Purified hepatocytes from Aoah^-/- mice had increased levels of Acly, Acaca, Acacb, Fasn, and Cd36 mRNA (Figure 5—figure supplement 1C and D) and decreased levels of Acot2 and Ppara mRNA (Figure 5—figure supplement 1E), changes that may contribute to lipid accumulation as mice grow older. Thus, in young mice, even before hepatic lipid accumulation can be observed, Aoah^-/- mouse hepatocytes have altered expression of genes that may promote lipid storage.

We next used clodronate-liposomes to deplete Kupffer cells and found that clodronate-liposome treatment diminished liver Aoah mRNA, confirming that Kupffer cells are the major source of hepatic AOAH (Shao et al., 2007; Figure 5D). Notably, after clodronate liposome treatment, nSREBP1 levels increased in the liver significantly, resembling Aoah^-/- mice (Figure 5D). AKT-mTOR1-p70 S6-kinase (S6K) activation induces SREBP1c processing in hepatocytes (Jeon et al., 2023; Owen et al., 2012; Yecies et al., 2011). Livers from clodronate-liposome-treated mice had increased AKT and mTOR activation (Figure 5D), suggesting that when gut-derived LPS cannot be inactivated by AOAH in the liver, bioactive LPS stimulates the mTOR pathway and induces SREBP1 activation.

To find out if excessive gut LPS increases liver LPS levels and promotes fatty acid synthesis gene expression, we orally gavaged Aoah^+/+ mice with LPS. We confirmed that orally gavaged (i.g.) LPS increased hepatic LPS levels in Aoah^+/+ mice (Figure 6A). Like Aoah^-/- mice, Aoah^+/+ mice that received gavaged LPS had increased levels of Srebf1a, Srebf1c, Pcx, Acaca, Acacb, Fasn, Scd1, and Cd36 mRNA in their livers (Figure 6B and C). LPS administered i.g. also increased nSREBP1 in Aoah^+/+ mouse livers (Figure 6D). Consistently, elevated AKT-mTOR-S6K activation was found in Aoah^-/- mouse livers, and when we gave Aoah^+/+ mice i.g. LPS, hepatic AKT-mTOR-S6K activity increased (Figure 6D). To find out whether LPS can directly stimulate hepatocytes to induce SREBP1 activation, we isolated primary hepatocytes and found that LPS stimulated mTOR activation and nSREBP1 upregulation (Figure 6E). Adding purified Kupffer cells to the hepatocyte culture did not further increase SREBP1 activation, suggesting that LPS directly acts on hepatocytes (Yu et al., 2021), at least in vitro. Blocking mTOR activation using torin1 prevented LPS-induced nSREBP1 upregulation (Figure 6E), suggesting that upon LPS stimulation SREBP1 activation depends upon the mTOR pathway. Collectively, these results confirm that excessive hepatic LPS derived from the Aoah^-/- mouse intestine induces hepatocyte mTOR activity, which increases nSREBP1 abundance; AOAH prevents hepatic lipid accumulation by inactivating gut-derived LPS.

C57BL/6J Aoah^-/- mice were produced at the University of Texas Southwestern Medical Center, Dallas, Texas, (Lu et al., 2003), transferred to the National Institutes of Health, Bethesda, Maryland, USA, and then provided to Fudan University, Shanghai, China. The mutated Aoah gene had been backcrossed to C57BL/6J mice for at least 10 generations. Aoah^+/+ and Aoah^-/- male mice were housed in a specific pathogen-free facility with 12 h light/dark cycle at 22°C at the Fudan University Experimental Animal Center (Shanghai, China). Mice were randomly assigned to each experimental group. Aoah^+/+ and Aoah^-/- male mice were co-housed for at least 3 weeks before the start and throughout the experiments. Co-housed Aoah^+/+ and Aoah^-/- mice were processed and analyzed blindly. All studies used protocols approved by the Institutional Animal Care and Use Committee (IACUC) of Fudan University (2023-DWYY-03JZS). All animal study protocols adhered to the Guide for the Care and Use of Laboratory Animals.

Co-housed Aoah^+/+ and Aoah^-/- male mice were fed either an NC or a high-fat calorie diet (D12492, Research Diets, USA) that contained protein:carbohydrate:fat (20:20:60, kcal%) plus high fructose (23.1 g/l; F3510, Sigma, USA) and glucose (18.9 g/l; G8270, Sigma) in the drinking water (HFD) for 28 weeks (Liu et al., 2018).

Total TAG, TCHO, LDL, HDL, AST, and ALT in mouse serum were measured at the Department of Laboratory Animal Science, Fudan University, using ADVIA Chemistry XPT.

After livers were excised and fixed in 4% paraformaldehyde for 18 h, they were sectioned and stained with H&E and Oil Red O. The samples were examined for steatosis, hepatocyte ballooning degeneration, lobular inflammation, and lipid droplets using a Nikon E200 microscope.

Histological analysis of the liver was performed based on the Sheka MASLD scoring criteria (Sheka et al., 2020). Liver steatosis is an infiltration of hepatic fat with minimal inflammation and is graded based on the fat percentage in hepatocytes: grade 0 (< 5%), grade 1 (5–33%), grade 2 (33–66%), and grade 3 (>66%). Inflammatory activity is manifested by two factors: grade 0 (no inflammation), grade 1 (<2 foci per 200× field), grade 2 (2–4 foci per 200× field), grade 3 (>4 foci per 200× field), and the presence of hepatocyte ballooning degeneration: no ballooned cells (grade 0), a few ballooned cells (grade 1), and many ballooned cells (grade 2).

Undiluted serum samples and liver homogenates (50 mg/ml) were run in duplicate alongside a standard curve of glycerol (triglyceride assay), cholesterol (cholesterol assay), or palmitic acid (free fatty acid assay) according to the manufacturer’s instructions. Triglyceride Determination Kit and Cholesterol Determination Kit were obtained from Applygen. Free Fatty Acid Quantitation Kit was purchased from Sigma-Aldrich.

RNA from livers or isolated hepatocytes was purified using TRNzol Universal Reagent (Tiangen) and reversely transcribed (Tiangen). The primers used for qPCR are listed in Supplementary file 1. Actb was used as an internal control, and the relative gene expression was calculated using the ΔΔCt quantification method.

After mice were exsanguinated, 2 ml PBS containing collagenase (5 mg/ml, type IV, Sigma) were injected into the liver via the inferior vena cava. The liver was cut into small pieces and treated with collagenase (0.5 mg/ml) for 10 min at 37°C. The liver pieces were mashed using syringe plungers. The cells were passed through a 70 μm strainer (WHB scientific) and then centrifuged at 50 × g for 3 min, three times, to pellet hepatocytes. The immune cells in the supernatant were pelleted (500 × g for 15 min) and then isolated on a 40% Percoll step gradient (Cytiva). The cells were then stained and analyzed using flow cytometry (Shao et al., 2007).

Liver cells were collected by centrifugation and then incubated with Fc blocking antibody (purified anti-mouse CD16/32, BioLegend, AB_2262724) on ice for 15 min. After the cells were stained with fluorescence-conjugated antibodies for 30 min on ice, they were washed and subjected to FACS (BD, FACSCelesta). The FACS data were analyzed using FlowJo software (TreeStar, Inc). All antibodies used for flow cytometry were anti-mouse antigens. Anti-mouse antibodies used for flow cytometry were anti-CD45-BV785 (Clone 30-F11, BioLegend, AB_2564590), anti-CD11b-FITC (Clone M1/70, BioLegend, AB_312788), anti-F4/80-BV421 (Clone BM8, BioLegend, AB_2563102), anti-Ly6G-PerCP-Cy5.5 (Clone 1A8, BioLegend, AB_1877271), anti-Ly6C-APC-Cy7 (Clone HK1.4, BioLegend, AB_10643867), anti-MHC II-PE-Cy7 (Clone M5/114.15.2, BioLegend, AB_2069376), anti-VSIG4-APC (Clone NLA14, eBioscience, AB_2637428), and anti-Tim4-PE (Clone RMT4-54, BioLegend, AB_1227807).

Primary hepatocytes were isolated from adult mice using a two-step collagenase perfusion method (Charni-Natan and Goldstein, 2020). In brief, the peritoneal cavity was opened and the liver was perfused in situ via the portal vein at 37°C with 20 ml PBS followed by 20 ml DMEM containing 10 mg collagenase (type IV, Sigma). The liver was then removed and gently minced, and the released cells were dispersed in DMEM containing 5% fetal bovine serum and 1% penicillin/streptomycin. The solution containing the mixed cells and debris was passed through a 100 µm cell strainer, and then centrifuged at 50 × g for 3 min, twice. The hepatocytes were pelleted and then isolated on a 90% Percoll step gradient (Cytiva). Hepatocytes were then resuspended in a TRNzol Universal Reagent (Tiangen) to measure mRNA.

Small pieces of liver were lysed with RIPA buffer (Biyotime) containing 1 mM PMSF (Biyotime) and a proteinase inhibitor mixture (Selleck). A commercial cytosol and nucleus Protein Extraction Kit (P0027, Beyotime) was used to separate cytosolic and nuclear proteins in the liver. The following antibodies were used for western analysis: anti-SREBP1 (SC-17755, Santa Cruz, AB_628283), anti-AOAH (HPA021666, Sigma-Aldrich, AB_3678739), anti-Lamin B1 (12987-1-AP, Proteintech, AB_2136290), anti-SCD1 (ab39969, Abcam, AB_945374), anti-FASN (SC-48357, Santa Cruz, AB_627584), anti-Phospho-mTOR (Ser2448) (5536, Cell Signaling Technology, AB_10691552), anti-Phospho-p70 S6 Ribosomal Protein (Thr389) (9234, Cell Signaling Technology, AB_2269803), anti-Phospho-AKT (Ser473) (4060, Cell Signaling Technology, AB_2315049), anti-mTOR (2983, Cell Signaling Technology, AB_2105622), anti-p70 S6 Ribosomal Protein (2217, Cell Signaling Technology, AB_331355), anti-AKT (4691, Cell Signaling Technology, AB_915783), and anti-α-Tublin (HRP-66031; Proteintech, AB_2687491). Anti-mouse IgG (7076S; Cell Signaling Technology, AB_330924) and anti-rabbit IgG (7074S, Cell Signaling Technology, AB_2099233) were used as secondary antibodies. ECL substrate (Bio-Rad Diagnostic) was used to detect proteins in western blotting, and the blot bands were quantified using ImageJ.

Fresh feces were collected and resuspended in endotoxin-free PBS (0.1 g/ml) and centrifuged at 800 × g for 5 min. The supernatant was heated at 70°C for 10 min. Mice were bled from the eye socket; 5 μl of 0.5 M EDTA was used as an anticoagulant. Livers were homogenized in PBS, centrifuged, and the supernatant was obtained. In some experiments, 200 µg of LPS was resuspended in 200 µl of PBS and then the suspension was slowly administered into the esophagus of mice using a gavage needle. Twenty-four hours later, the livers were collected for analysis. All the samples were collected for TLR4-stimulating activity using a cell-based colorimetric endotoxin detection kit (HEK-Blue LPS Detection Kit2, InvivoGen). Diluted samples were added to human embryonic kidney (HEK-293) cells that expressed hTLR4 and an NF-κB-inducible secreted embryonic alkaline phosphatase reporter gene. After 18 h incubation, cell culture media were applied to QUANTI-Blue medium to measure alkaline phosphatase activity. A preparation of E. coli 055:B5 LPS, standardized to FDA-approved control standard endotoxin, which was included in the kit, was used to quantitate TLR4-stimulating activity. Plates were read at a wavelength of 620 nm (Tecan).

After mice were fasted for 18 h, they were orally gavaged with fluorescein isothiocyanate (FITC)-conjugated 4 kDa dextran (50 mg per 100 g body weight) (46944, Sigma-Aldrich). Four hours after gavage, blood was collected from the facial vein and the serum was used for FITC fluorescence measurements (excitation, 490 nm; emission, 520 nm).

Total RNA was isolated using TRNzol from co-housed 6–8-week-old Aoah^+/+ and Aoah^-/- mouse livers. The libraries were sequenced on an Ilumina Novaseq 6000 platform and 150 bp paired-end reads were generated. Differential expression analysis was performed using the DESeq2. Q value <0.05 and foldchange >1.5 were set as the threshold for significantly differential gene expression. GSEA was performed using GSEA software. RNA-seq analysis was conducted by Shanghai OE Biotech. Co, Ltd., China, and the results were deposited at PRJNA1022016.

AOAH expression in mouse hepatic cells was analyzed based on single-cell RNA-seq data using the Liver Cell Atlas at https://www.livercellatlas.org/ (Remmerie et al., 2020). Liver single-cell RNA sequencing data from MASLD-cirrhosis patients (n=5) and healthy controls (n=5) were obtained from GSE136103 (Ramachandran et al., 2019). R package Seurat (version 5.1.0) was utilized for clustering and cell-type identification.

Kupffer cells were depleted by injecting 200 μl Clodronate-liposomes (5 mg/ml, Liposoma BV) i.v. and PBS-liposomes were used as controls. Two days after injection, Kupffer cell deletion was confirmed by staining F4/80⁺ cells in cryostat liver sections.

Data are presented as mean ± SEM. Differences between groups were analyzed using Mann–Whitney test. To compare kinetic difference, two-way ANOVA was used. The statistical significance was set at p<0.05. *p<0.05, **p<0.01, and ***p<0.001.