Roxadustat was provided by Zhejiang Jianfeng Pharmaceutical Co., Ltd. (Jinhua, China). Bovine serum albumin (BSA) was purchased from Sigma Aldrich (Missouri, United States). Hematoxylin and eosin (H&E) staining solution, 4% polyformaldehyde, phosphate buffer saline (PBS), and BCA Protein Assay Kit were purchased from Biosharp (Hefei, China). Total RNApure reagent (Trizol) was purchased from Beijing Zomen Biotechnology Co., Ltd. (Beijing, China). HiScript II Q Select RT SuperMix and AceQ SYBR qPCR Master Mix were purchased from Vazyme (Nanjing, China). The dihydroethidium (DHE) staining kit was purchased from Beyotime (Shanghai, China). A cocktail of protease inhibitors, PMSF, and enhanced chemiluminescence (ECL) kit were purchased from Millipore (Darmstadt, Germany). Bromphenol blue, triton X-100, and sodium dodecyl sulfate (SDS) were purchased from Solarbio (Beijing, China). Mouse anti-fatty acid synthase (FASN) and CD68 monoclonal antibodies were purchased from Santa Cruz Biotechnology (CA, United States). Mouse anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH), rabbit anti-superoxidase dismutase 1 (SOD1), SOD2, α-Tubulin, peroxisome proliferator-activated receptor α (PPARα), and interleukin-1β (IL-1β) polyclonal antibodies were purchased from Abclonal (Wuhan, China). Rabbit anti-HIF-1α, carnitine palmitoyltransferase 1A (CPT1A), and cytochrome P450 2E1 (CYP2E1) polyclonal antibodies were purchased from Affinity Biosciences (OH, United States). Mouse anti-tumor necrosis factor α (TNF-α), rabbit anti-sterol-responsive element-binding protein 1c (SREBP1c), carbohydrate response element-binding protein α (ChREBPα) polyclonal antibodies, HRP-conjugated goat anti-rabbit IgG (H + L), and mouse IgG (H + L) were purchased from Proteintech Group Inc. (IL, United States). All other chemical reagents were analytical grade.

HepG2 and RAW264.7 cells were purchased from ATCC (VA, United States), and cultured in complete MEM or 1640 medium containing 10% fetal bovine serum (FBS, AusGeneX, Australia) and 50 μg/ml streptomycin/penicillin (UT, United States), in a humidified incubator with 5% CO₂ at 37°C. Before treatment, cells were incubated in a serum-free medium.

Homo HIF-1α siRNA and the corresponding scrambled siRNA were purchased from RiboBio Biotechnology (Guangzhou, China). HepG2 cells were cultured in a 6-well plate at a density of 5 × 10⁵ cells/well in a serum-free Opti-MEM. HIF-1α or control siRNA (40 nM/well) were transfected into cells using Lipofectamine RNAiMAX Transfection Reagent (Invitrogen, CA, United States). After 24 h transfection, HepG2 cells received indicated treatment (Wang et al., 2020).

The eight-week-old male C57BL/6J mice were purchased from GemPharmatech (Nanjing, China). Mice were maintained in a chamber with constant temperature (22 ± 2°C) and humidity (55 ± 2%) for a 12-h light/dark cycle.

The chronic ALD mouse model was constructed as described (Bertola et al., 2013). In brief, mice were divided into four groups (8 mice/group); all mice were fed with the Lieber–DeCarli control diet for the first 5 days. Then, mice in control groups were fed with the Lieber–DeCarli control diet (ethanol free) for 8 weeks plus intragastric (i.g.) administration of a single maltose dextrin solution (9 g/kg body weight, equal calorie to ethanol); mice in ALD groups were fed with the Lieber–DeCarli diet (contain 5% ethanol) for 8 weeks plus i.g. administration of single binge ethanol (5 g/kg body weight). Mice were euthanized after 9 h of the single binge ethanol or maltose solution administration.

The acute ALD mouse model was constructed as follows (Chang et al., 2015): mice in the control groups were fed normal chow for 3 days, and then received i.g. administration of maltose dextrin solution (9 g/kg body weight). In model groups, mice were fed an HFD (60% kcal; CAT#D12492) for 3 days and received i.g. administration of 31.25% (vol/vol) ethanol solution (5 g/kg body weight) on the last day. All mice were euthanized after 9 h of ethanol or maltose dextrin solution administration, followed by a collection of blood and tissue samples.

To determine the role of roxadustat in ALD, mice in the vehicle group received intraperitoneal (i.p.) injection of PBS every day; mice in roxadustat groups received i.p. injection of roxadustat solution (10 mg/kg body weight or 25 mg/kg body weight for chronic or acute ALD mouse model, respectively) every day. The selection for doses of roxadustat is described as follows: previous studies used a serious range of roxadustat for in vivo experiments, mainly from 10 to 60 mg/kg body weight (Beck et al., 2017; Deguchi et al., 2020; Kabei et al., 2020). To verify if different doses of roxadustat have protective effects in acute and chronic ALD mouse models, we chose a low dose of roxadustat for the chronic model, while a middle dose for the acute model, which is based on the standard of animal ethics that use as fewer animals as possible in in vivo experiment.

After treatment, cells or 30 mg liver tissues were lysed or grated with lysis buffer. The BCA Protein Assay Kit was used to determine protein concentration. The same amount of protein (60 μg) from each sample was used to determine the protein expression of ACC1, SREBP1c, FASN, IL-1β, PPARα, CPT1A, CYP2E1, TNF-α, ChREBPα, HIF-1α, GAPDH, and α-Tubulin using Western blot (Yin et al., 2020). The signals were detected by ChemiScope 3000 mini (Qinxiang, Shanghai, China), and the band density was quantified using Photoshop software.

Liver CD68 expression was determined using immunohistochemical staining. Images were obtained using a ZEISS Scope A1 fluorescence microscope, and quantity analysis of CD68 positive cells was determined by Photoshop software.

After treatment, Trizol was used to extract total RNA from 20 mg liver tissues or RAW264.7 cells. cDNA was synthesized with 1 μg RNA from each sample with HiScript II Q Select RT SuperMix (gDNA wiper). RT-PCR was applied with the primers listed in Table 1. mRNA expression was normalized by β-actin mRNA in the corresponding samples.

A piece of the liver was fixed in 4% paraformaldehyde overnight, and then dehydrated with an auto dehydrator (Leica, Wetzlar, Germany). After being embedded in paraffin, the tissue samples were cut into 5 μm sections, and then conducted with H&E staining. Frozen liver tissues embedded in OCT were cut into 5 μm sections for the determination of lipid accumulation or ROS levels by Oil Red O or DHE staining, respectively (Wang et al., 2021). Images were obtained using a ZEISS Scope A1 fluorescence microscope. DHE fluorescence intensity was quantified using ImageJ software.

All data were generated from at least three independent experiments. GraphPad Prism 8.0 was used for data statistical analysis. All data were shown as means ± SEM. Data were analyzed by one-way ANOVA followed by Bartlett’s test, and the difference was considered significant at p < 0.05.

To investigate the role of roxadustat on ALD, we first constructed a chronic ALD mouse model by feeding the Lieber–DeCarli liquid diet plus a single alcohol gavage, and mice received roxadustat treatment for 8 weeks (Figure 1A). During the experiment, we monitored the pathology changes in the liver with an ultrasound scanner (Mathiesen et al., 2002; Pandit et al., 2019). The B-mode ultrasound images indicated the Lieber–DeCarli liquid diet caused hepatic steatosis after 2 weeks of feeding, evidenced by enhanced brightness. In contrast, lipid accumulation was attenuated along with roxadustat treatment (Supplementary Figure S1). In addition, we found long-term exposure to alcohol resulted in an overall larger and whiter liver, as well as an elevated ratio of liver weight to body weight, which was improved by roxadustat treatment (Figures 1B,C). Consistent with morphological changes in the liver, H&E staining revealed significant pathological morphological changes in chronic ALD mouse liver (Figure 1D). However, roxadustat improved the damage on liver tissues. Serum transaminase activities are indicators of liver damage, which can be produced by injured hepatocytes (Marin et al., 2017). Our results showed that serum ALT, AST, and ALP activities were enhanced in chronic ALD mouse serum while being attenuated by roxadustat, especially the AST and ALP activities (Figure 1E). The aforementioned results indicated that roxadustat inhibits the development of chronic ALD.

The results of B-mode ultrasound and H&E staining indicated that roxadustat can improve Lieber–DeCarli liquid diet-induced hepatic steatosis. We further conducted Oil Red O staining and found that lipid accumulation was enhanced in the chronic ALD group while being alleviated by roxadustat (Figure 2A). Compared to the model group, liver triglyceride levels were also attenuated by roxadustat treatment (Figure 2B). DGAT and FASN are key enzymes that catalyze the final reaction of triglyceride synthesis (Chitraju et al., 2017). Our results showed that DGAT and FASN mRNA levels were increased under the mediation of alcohol (Figure 2C). However, roxadustat treatment reduced alcohol-enhanced DGAT1 and FASN expression. PPARα is a nuclear receptor that regulates the expression of various genes involved in mitochondrial β-oxidation (You and Arteel, 2019). Alcohol intake affects mitochondrial oxidation and inhibits PPARα signaling. As shown in Figure 2D, PPARα protein expression was reduced in ALD mouse liver, which was enhanced in the roxadustat treatment group. CPT1A is a classical target gene of PPARα, a key regulator for fatty acid β-oxidation (Rakhshandehroo et al., 2010). Consistent with the results of PPARα, we also found that roxadustat treatment enhanced CPT1A protein levels (Figure 2D).

To further investigate the mechanism of roxadustat on lipid accumulation, we treated HepG2 cells with roxadustat and found that roxadustat reduced FASN protein expression in a dose-dependent manner (Figure 2E). Furthermore, we transfected cells with HIF-1α siRNA to knockdown HIF-1α levels. As shown in Figure 2F, roxadustat inhibited FASN expression in siNC HepG2 cells while having little effect on siHIF-1α HepG2 cells, indicating roxadustat regulates FASN levels depending on HIF-1α expression. In addition, Oil Red O staining results showed that roxadustat inhibited lipid accumulation in control cells, but not in HIF-1α knockdown cells (Figure 2G). In contrast to the in vivo results, we found roxadustat had little effect on PPARα protein levels neither in control cells nor in HIF-1α knockdown cells (Figure 2F). Taken together, the abovementioned results demonstrated that roxadustat inhibits lipid accumulation both in vivo and in vitro.

Alcohol and its metabolic derivatives can act as harmful stimuli to the body, leading to an inflammatory response. Inflammation maintains the homeostasis of the body, but can also cause collateral damage to normal tissues (Xu et al., 2017). Alcohol impairs intestinal barrier function and increases LPS flux to the portal vein. Excess LPS binds to toll-like receptor four to activate macrophages, causing inflammatory cytokine secretion (Xu et al., 2017). In this study, we found pro-inflammatory cytokines expression, such as IL-1β and TNF-α, was increased in liver tissues of ALD mice, which was significantly inhibited by roxadustat (Figure 3A). In addition, we conducted immunohistochemical staining with CD68 antibody (a marker for monocyte) and found that levels of CD68⁺ cells were enhanced in ALD mouse liver. Consistent with the results of inflammatory cytokines, roxadustat reduced the infiltration of monocytes (Figure 3B). Furthermore, we treated RAW264.7 cells with LPS to induce inflammation in the presence or absence of roxadustat. As shown in Figure 3C, LPS-induced IL-1β and IL-6 mRNA levels were largely attenuated by roxadustat treatment. The above results suggest that roxadustat inhibits inflammatory response both in vivo and in vitro.

To further investigate if roxadustat has protective effect on HFD plus acute alcohol-caused liver injury, we conducted an acute ALD mouse model by feeding mice an HFD for 3 days plus a single gavage of ethanol (Figure 4A). After 9 h of ethanol treatment, the mice were sacrificed. Mouse liver of the model group showed a tendency to be much whiter and larger, with an enhanced ratio of liver weight to body weight (Figures 4B,C). In contrast, roxadustat attenuated acute alcohol-caused changes in the liver and slightly decreased liver weight/body weight. Moreover, fat vacuoles were found in acute ALD mouse liver, which were improved by roxadustat (Figure 4D). Consistent with the results of the chronic ALD mouse model, we showed acute alcohol also enhanced serum ALT, AST, and ALP activities (Figure 4E). However, AST and ALP activities were greatly inhibited by roxadustat treatment, indicating the hepatoprotective function of roxadustat.

The results of H&E staining in Figure 4D showed that roxadustat can also inhibit acute alcohol-induced lipid accumulation in the liver. We further conducted Oil Red O staining of liver frozen sections and found that roxadustat treatment significantly reduced lipid accumulation in acute ALD model mouse liver (Figure 5A). The results of liver triglyceride levels further confirmed that lipid levels were attenuated by roxadustat (Figure 5B). The abovementioned results indicated that roxadustat can inhibit acute alcohol-induced hepatic lipid accumulation. The liver generates fatty acids from non-lipid precursors via de novo lipogenesis. Multiple enzymes participate in this process; ACC-1 converts acetyl coenzyme A to malonyl coenzyme A and FASN synthesizes saturated fatty acids from malonyl coenzyme A (Huang et al., 2010). Compared to the control group, we found mRNA levels of DGAT1, FASN, and ACC1, and protein levels of ACC1 and FASN were induced in acute ALD mouse liver. However, the expression of the aforementioned lipogenesis-related genes was inhibited by roxadustat (Figures 5C,D). SREBP1c, a key transcription factor, is a master regulator of lipogenesis by activating genes related to fatty acid and triglyceride synthesis (Li et al., 2011). Our results showed that acute alcohol-induced both precursor and mature forms of SREBP1c protein expression were reduced by roxadustat treatment (Figure 5E). However, we found that roxadustat had little effect on acute alcohol-reduced PPARα protein expression (Figure 5F), which is consistent with the results of Figure 2D. ChREBPα is emerging as a critical driver of the lipid metabolism (Wang et al., 2015; Sanchez-Gurmaches et al., 2018). Our results indicated that ChREBPα mRNA and protein expression were enhanced in acute ALD mouse liver, which was reduced by roxadustat treatment (Figures 5C,G). Taken together, the above results suggested that roxadustat can attenuate the development of acute ALD through downregulation of lipid synthesis-related genes expression.

The alcohol metabolism generates large amounts of free radicals and ROS, which can cause oxidative stress and mitochondrial damage to lead further damage and apoptosis of hepatocytes. To determine if roxadustat can regulate oxidative stress in ALD mouse models, we conducted DHE staining of liver sections. As shown in Figure 6A, we found that ROS levels were enhanced in both chronic and acute ALD mouse liver. However, roxadustat decreased ROS accumulation in liver tissues, evidenced by reduced density of red fluorescence. CYP2E1 is an alcohol-inducible enzyme that contributes to ethanol metabolism. It can cause oxidative stress, depletion of the antioxidant system, and liver damage due to massive rupture of hepatocyte mitochondrial membranes (Albano, 2008). We found protein expression of CYP2E1 was largely enhanced in ALD mouse liver. Conversely, roxadustat reduced alcohol-enhanced CYP2E1 levels in those two mouse models with HIF-1α activation (Figures 6B,C). SOD1 and SOD2 are critical antioxidant enzymes. Our results showed that roxadustat had little effect on SOD2 protein expression. However, inhibited SOD1 levels were enhanced by roxadustat in both acute and chronic ALD mouse liver tissues (Figures 6B,C). The aforementioned results showed that roxadustat ameliorates alcohol-induced ROS accumulation by decreasing CYP2E1 and enhancing SOD1 expression.