A total of 20 C57BL/6J male mice aged 7 weeks were purchased from B&K Experimental Animal Corporation Ltd (Shanghai, China). All of them were housed in Zhejiang Chinese Medical University Animal Facility with specific pathogen-free conditions (five mice/cage; 12 h light–dark cycle; 22 ± 1 °C; 60%-65% humidity). They were cared for in according to the Guide for the Care and Use of Laboratory Animals.

After a week of habituation, mice were randomized into alcohol-fed group (AF) and pair-fed control group (PF). Mice in the control group took the control Lieber-DeCarli liquid diet, with 47% energy from carbohydrate, 18% energy from protein and 35% energy from fat. In the alcohol-fed group, equicaloric maltose dextrin was substituted with 95% ethanol to maintain the isoenergy intake between two groups. And energy percentage of alcohol in total energy was gradually increased from 0% in the initial 1–3th days, to 5.5% in the 4–5th days, to 11% in the 6–7th days, to 22% in the second week, to 27% of total energy in the third week, to 32% of total energy in the fourth week finally. It is a mature method in our lab to induce the mice model of alcoholic fatty liver disease. The detail composition of the diet was shown in Table 1. Dietary intake and body weight were recorded twice in each week.

At the end of 4 weeks’ intervention, all mice were sacrificed under anesthesia after fasting for 12 h. Blood and liver tissues were collected for further analysis, including the following biochemical analysis, histological assessment, qRT-PCR and lipidomics analysis.

Blood sample was centrifuged with 2000 rpm for 15 min at 4 °C, and plasma was separated and then kept at − 80 °C until use. The concentrations of aspartate aminotransferase (AST), alanine aminotransferase (ALT), total cholesterol (TC) and triacylglycerol (TAG) in plasma were determined with an automatic analyzer (Model XE-2100, Sysmex Kobe, Japan).

Liver tissue samples were homogenized in methanol and centrifuged for 15 min at 3000 rpm. The supernatant was separated for detecting hepatic TC and TG concentrations with enzymatic colorimetric methods using commercially available kits (Nanjing Jiancheng Bioengineering Institute, Jiangsu, China). Another liver sample was homogenized in saline and then centrifuged. The supernatants were separated for investigating the concentrations of superoxide dismutase (SOD) and malondialdehyde (MDA) with commercial kits (Lianke Biotechnology CO., Ltd., Zhejiang, China).

Liver specimens were fixed in 10% formalin, then dehydrated and embedded in paraffin. The frozen paraffin blocks were sectioned and subsequently stained with Haematoxylin and Eosin (H&E) for histological examination. In addition, another liver samples were embedded in Tissue-Tek OCT, then sectioned in 4–5 μm thick sections, and stained with Oil Red O (Sigma-Aldrich, St. Louis, MO, USA) for the evaluation of fat deposition. Finally, they were viewed under a light microscope (100× and 200×) and analyzed via Image J software.

Total RNA was purified from liver tissue using Trizol reagent (Thermo, Waltham, MA, USA), according to the manufacturers’ instructions. Then cDNA was synthesized with Prime Script RT Reagent kit (Takara Biotechnology Co., Ltd, Dalian). Quantitative real-time PCR was performed in 7500 Fast Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) using SYBR Premix Ex Taq TM (Tli RNaseH Plus) kit (Takara Biotechnology Co., Ltd, Dalian), as described by the manufacturer. The sequences of primers for real-time PCR of IL-6 were 5′-CCGGAGAGGAGACTTCACAG-3′ (forward) and 5′-CAGAATTGCCATTGCACAAC-3′ (reverse); for real-time PCR of TNF-α were 5′-CCCTCACACTCACAAACCAC-3′ (forward) and 5′-ACAAGGTACAACCCATCGGC-3′ (reverse). The levels of gene expression in control group were set to 1.0, and results from the other group were shown as relative expression ratios to the control group.

Liver samples were weighed and then cut into small sections (n = 6 per group). Sections from the same place of each liver sample were prepared for lipidomics analysis with ExionLC UHPLC system coupled to QTrap 6500 + mass spectrometer (SCIEX, USA). A mixture of 10 mg liver sample and 400 μL water was homogenized for 60 s, and then ground with a mixer mill at 45 Hz for 4 min, followed by ultrasonic treatment for 5 min in ice-water bath. Then a mixture of 10 μL homogenate, 190 μL water and 480 μL extract solution (methyl tert-butyl ether: methanol = 5:1) containing internal standard was vortexed vigorously and centrifuged at 3,000 rpm for 15 min at 4 °C. After that, 250 μL of supernatant was transferred into a fresh tube and dried in vacuum at 4 °C. The dried samples were dissolved in 100 μL of solution (DCM: methanol: water = 60: 30: 4.5), and then centrifuged with 12,000 rpm for 15 min at 4 °C. The supernatant was transferred into a fresh glass vial for further analysis. In addition, the quality control (QC) sample was prepared by pooling 10 μL of supernatants from each sample together.

2 μL of the sample solution was injected into the liquid chromatographic column (Acquity HSS T3 Column, 1.8 μm, 2.1 mm × 100 mm). The mobile phase A consisted of 60% acetonitrile, 40% water and 10 mmol/L ammonium formate. The mobile phase B consisted of 90% isopropanol, 10% acetonitrile and 10 mmol/L ammonium formate. The elution program was as follows: 20% B in 0–1 min, 20–60% B in 1–4 min, 60–98% B in 4–15 min, 98% B in 15–16 min, 20% B in 16.1–18 min. The flow rate was 0.3 mL/min, and temperature of column and auto-sampler were set at 40 °C and 6 °C, respectively. During sequence analysis, QC sample was injected after every six samples to detect the reproducibility of sample and the stability of analytical platform.

Sciex QTrap 6500 + MS was conducted in both positive-ion and negative-ion modes. Primary ion source parameters were as follows: Ion-spray voltage, +5500/−4500 V; Curtain gas, 40 psi; Temperature, 350 °C; Ion source gas, 1: 50 psi; Ion source gas, 2: 50 psi.

The raw data were imported into Skyline 20.1 to quantify the target compounds. The features would be excluded if they were present in less than 50% of the quality control samples, or if they showed a coefficient of variation of more than 20% in quality control samples. Features with a peak area less than 3000 in 80% of all samples were regarded as non-detected. The internal standards used in this method were listed in Additional file 1: Table S1. The absolute content of each lipid was calculated according to the peaks area and actual concentration of the internal standard in each lipid class.

Biochemical data were presented as means ± standard deviations (SD). The statistical significance between two groups were tested with Student’s t-test. Difference of P-value less than 0.05 was considered statistically significant. The analysis was performed with SPSS 23.0 software.

For data of lipidomics metabolites, the dataset containing sample name, peak number and normalized peak area were imported to SIMCA-P version 16.0.2 for multivariate analysis. Firstly, values of features were log10-transformed and subjected to Unit Variance Scaling. Then principal component analysis (PCA), an unsupervised multivariate pattern recognition analysis, was conducted to visualize the clustering of the samples from same group and the QC samples. Secondly, a supervised multivariate orthogonal partial least squares discriminant analysis (OPLS-DA) was conducted to discriminate the separation of lipid profiling between two groups. The quality of the fitted model was assessed with R2Y and Q2Y parameters, which mean the goodness of fit and prediction, respectively. Then a 200 times permutation test was conducted to further check the robustness and predictive ability of the obtained OPLS-DA model. The value of variable influence on the projection (VIP) of the first principal component was obtained from OPLS-DA. The values summarize the contribution of each variable to the model. Only the metabolites with a VIP value higher than 1 and P < 0.05 (student’s t-test) were considered as significantly different. Furthermore, correlations between lipids and ALD indicators were investigated with the Pearson’s correlation test.