Blood sample

The study participants included patients with OSA, patients with OSA combined with NAFLD, and healthy individuals (n = 20 each). The inclusion criteria were as follows: (1) age between 20 years and 60 years; (2) diagnosed with OSA or NAFLD; and (3) first diagnosed disease. The exclusion criteria were as follows: (1) liver diseases; (2) respiratory diseases; and (3) infections, malignant tumors, hematological diseases, hypertension, diabetes, and other systemic diseases. Briefly, 2 mL of peripheral blood was collected, rapidly transferred into a tube of ethylenediamine tetraacetic acid, and mixed by vortexing or using a pipette tip. Within 1 h, the samples were centrifuged at 8,200 × g (9,400 rpm) for 10 min at 4 °C. One milliliter of the supernatant was pipetted and transferred to a clean 1.5 mL centrifuge tube. It was centrifuged at 16,000 × g (13,200 rpm) for 10 min at 4 °C. The supernatant was carefully pipetted into a new centrifuge tube. The samples were stored in a -80 °C refrigerator. All participants provided their informed consent for participation before enrollment. The study was approved by the Ethics Committee of Yan’an Hospital Affiliated with the Kunming Medical University.

Animal grouping

Ninety 4-week-old C57BL/6J male mice (SPF, Shanghai Slack Experimental Animal Center, China) and miR-421−/− mice (Linmei Biological Technology Co., Ltd. Hefei, China) were raised in a pathogen-free environment, alternating light and dark cycle for 12 h (8:00 am–8:00 pm). The NAFLD model was constructed by feeding the mice a high-fat diet (32.1% carbohydrates, 16.5% protein, and 51.4% fat) for 6 weeks. A normal diet consisted of 32.1%, 16.5%, and 51.4% carbohydrates, protein, and fat, respectively. This experiment was conducted per the Ethics Committee of Yan’an Hospital Affiliated with the Kunming Medical University. The animal experiment guidelines and research plan were approved by the Animal Experiment Committee of Yan’an Hospital Affiliated with the Kunming Medical University.

In Group 1 (n = 6), the mice were divided into Control, Control-50, Control-100, OSA-50, and OSA-100 types. After 1 week of adaptation, healthy mice were intraperitoneally injected with exosomes from healthy or OSA mice (50–100 µg/mouse). The injection was administered once every 3 days. Six weeks later, peripheral blood was collected; liver tissues were collected after euthanasia.

In Group 2 (n = 6), the mice were divided into Healthy-Control, Healthy-OSA, Healthy-OSA + NAFLD, miR-421−/−-Control, miR-421−/−-OSA, and miR-421−/−-OSA + NAFLD types. Healthy and miR-421 gene knockout mice were used to construct the OSA and OSA + NAFLD models, respectively. Liver tissues were collected 6 weeks after euthanasia.

In Group 3 (n = 6), the mice were divided into Control, OSA, and OSA + miR-421−/− types. The exosomes from healthy, OSA, and miR-421−/− OSA mice were used to treat the healthy mice (100 µg/mouse). The injection was administered once every 3 days. Liver tissues were collected 6 weeks after euthanasia.

Exosome collection and labeling

ExoQuick and ExoQuick-TC Kits (System Biosciences, Mountain View, CA) were used for exosome collection. First, 500 µL of fresh human/mouse serum or 10 mL of cell culture medium was mixed with the reagent and separated according to the manufacturer’s instructions. The exosomes were labeled using PKH26 (Sigma-Aldrich, St. Louis, USA). Second, these exosomes (50 µL) and PKH26 (1 µL) were added to Reagent C and incubated for 5 min at 25 ℃. Third, 1% bovine serum albumin (BSA) (150 µL) was added to neutralize the excess PKH26. The exosomes were purified using an exosomal spin column (Invitrogen, Carlsbad, CA, USA). The morphology of separated exosomes was observed under an electron microscope. The particle size distribution of the exosomes was evaluated using a particle size analyzer (NanoSight LM10, Malvern, UK). Western blotting was conducted to detect the expression of exosomal marker proteins CD9, CD63, and CD81, and exosome-negative marker protein Calnexin.

OSA model

In this study, chronic intermittent hypoxia (CIH) was used to construct an OSA model. The induction principle of OSA-related diseases, including metabolic dysfunction-associated steatotic liver disease, was based on OSA-induced chronic hypoxia [21,22,23]. OSA could decrease O2 inhalation directly. Therefore, OSA-induced hypoxia was simulated by directly controlling the inhaled O2 concentration, i.e., CIH was used to simulate OSA. The inhaled O2 concentration was controlled by placing the mice in a glass chamber (30 × 20 × 20 in, Oxycycler model A44XO, BioSpherix, Redfield, NY, USA). After 1 week of acclimatization, they received daily intermittent hypoxia treatment for 8 h. The chamber outlet was connected to an instrument detecting O2 concentration. Based on the outlet concentration, the online control system adjusted the air and nitrogen flow at the inlet, such that the average O2 concentration was 10 ± 1% during hypoxia. In the absence of light, the O2 concentration was maintained at 10% for 2 min. Subsequently, it was restored to normoxic conditions (21%) alternately for 8 h. The remaining time involved normoxia conditions. Intermittent hypoxia lasted for 6 weeks.

Hematoxylin and eosin staining

The liver specimens were fixed in 10% neutral formalin, sectioned, and deparaffinized. After adding hematoxylin, the specimens were incubated at 25 ℃ for 10 min. Subsequently, 0.5% eosin solution was added; the specimens were incubated at the same temperature for 3 min. Lesions of the colon cross-section were observed under a microscope (Olympus, Japan).

RNA fluorescence in situ hybridization

miR-421 localization in the liver tissues was detected using fluorescence in situ hybridization. First, the paraffin sections were deparaffinized, and a freshly diluted pepsin stock solution containing 3% citric acid was added. The sections were digested at 37 °C for 1 min. Second, 4% paraformaldehyde containing 0.1% diethyl pyrocarbonate (pH = 7.2–7.6) was added to the sections. Moreover, 20 µl of pre-hybridization solution was added; the sections were incubated for 2 h. Third, a fluorescein-labeled (red fluorescent) miR-421 probe (Shenggong Biological Engineering Co., Ltd., Shanghai, China) was added and incubated overnight at 37 °C. Finally, 4′,6-diamidino-2-phenylindole (DAPI) was added for nucleus staining. The sections were observed under a laser confocal microscope.

Isolating primary hepatocytes and KCs

The primary hepatocytes and macrophages were isolated [24, 25]. A collagenase/protease mixture (Vitacyte, Indianapolis, IN) was applied for the in-situ perfusion of mouse liver to enable digestion. Liver tissues were placed in a 150-mm petri dish and operated in a sterile environment. The petri dish contained 1% BSA, 15 mM Hanks’ Balanced Salt Solution (HBSS), and 1 g/L glucose (Gibco, Carlsbad, CA, USA). The liver parenchymal cells were shed by gentle shaking. The sample was passed through a filter (105 μm) and centrifuged (4 °C, 70 × g, 3 min). The pellet was used to separate the hepatocytes. Subsequently, the supernatant was used to separate the macrophages. The hepatocytes (purity > 99%) were collected using Percoll density gradient centrifugation (GE Healthcare, Marlborough, MA).

The supernatant and sample containing nonparenchymal cells were passed through the filter (297 μm) and centrifuged (4 °C, 500 × g, 6 min). The particles containing nonparenchymal cells were resuspended in HBSS and passed through a 70 μm cap filter. Furthermore, Avanti J-26XP centrifuge (Beckman-Coulter, Brea, CA), Optiprep density gradient centrifugation (Sigma-Aldrich), and countercurrent elutriation centrifugation were used to separate the macrophages from other nonparenchymal cells. The purity of macrophages was determined using flow cytometry (> 95%).

Hepatocyte culture and treatment

The hepatocytes were cultured in Dulbecco’s Modified Eagle Medium/F12 medium (Gibco, Carlsbad, CA, USA) with 10% fetal bovine serum (Gibco, Carlsbad, CA, USA) in a 5% CO2 incubator at 37 °C.

To induce NAFLD in vitro, a free fatty acid (FFA) mixture (oleate: palmitate = 2:1) (0.5 mmol/L concentration) was added to the culture medium and incubated for 24 h. Additionally, 20 µg/mL of exosomes were added to the medium to induce the hepatocytes.

To induce intermitted hypoxia in vitro [26], the O2 concentration in the incubator was adjusted by modifying the N2 content. Moreover, CO2 concentration was maintained at 5%. Every 30 min, the O2 concentration was changed from 1 to 21%. The cells were incubated for 6 days.

Macrophage culture and treatment

The PKH26-labeled exosomes were red fluorescent. Briefly, 20 µg/mL of exosomes were added to the culture medium to induce the macrophages. After 24 h of induction, the macrophages were fixed and DAPI (Sigma) was added for nucleus counterstaining. The internalization of macrophages in exosomes was observed under a confocal microscope (Zeiss LSM 700, Germany).

Lipopolysaccharides (LPS) were used to activate the NLRP3 inflammasomes and M1 polarization. Approximately 1 mg/mL of LPS (L2630, Sigma-Aldrich) was added to the medium and incubated for 4 h. The proteasome inhibitor MG-132 (S2619, Selleck, USA) was added to the medium to inhibit degradation. Subsequently, 50 µmol/L of 3-methyladenine (3-MA, S2767, Selleck) was added to inhibit autophagy.

Cell co-culture

Forty-eight-well type I collagen-coated plates were used to construct a co-culture system of hepatocytes and macrophages. Hepatocytes and macrophages were used in a ratio of 2:1 [27]. First, hepatocytes were added at a density of 4 × 105 cells/well and shaken gently every 15 min. Second, macrophages were added at a density of 2 × 105/well and shaken gently every 15 min. After four times, the medium was removed; fresh medium was added and cultured.

Cell transfection

The miR-421 mimic and corresponding negative control (NC) plasmids were obtained from GenePharma (Shanghai, China). The cells were transfected with 50 nM mimic and inhibitor using Lipofectamine™ 2000 (Invitrogen, Waltham, USA).

Dual luciferase report

The 3’-UTR sequence of wild-type (wt-) SIRT3 mRNA was amplified to the downstream site of the pGL4 luciferase vector (Promega, Madison, WI, USA). To generate the mutated (mut-) SIRT3 mRNA 3’-UTR, the rapid site-directed mutagenesis kit (D0206, Beyotime) was used. The macrophages were seeded in 24-well plates at a density of 3 × 104/well. After 24 h, 1 µg of wt- or mut-SIRT3 luciferase plasmid, 50 nM of miR-421 mimic or miR-421 NC, and 150 ng of Renilla luciferase plasmid (Beyotime) were transfected into the macrophages via LipofectamineTM2000. The cells were incubated at 37 °C for 36 h. Following the manufacturer’s protocol, a dual luciferase reporter gene detection kit (Promega, Madison, WI, USA) was used to detect luciferase activity. All data were normalized to detect Renilla luciferase activity.

Triglyceride and cholesterol measurement

The levels of triglyceride (TG) and total cholesterol (TC) in the liver tissues and cells were detected using an enzymatic assay kit (E1025-105, E1026-105, Applygen Technologies Inc., Beijing, China).

Oil red O staining

First, Oil red O (ORO) reagent (C0158M, Beyotime, Shanghai, China) was added to the liver Sect. (10 μm) and stained at 25 ℃ for 20 min. Second, the cells were stained using hematoxylin (C0107M, Beyotime) and incubated for 5 min for nucleus staining. The cells were treated using 10% formalin for fixation (15 min). They were covered using a washing solution and incubated for 20 min. After rinsing the washing solution, ORO reagent was added. The cells were incubated at 25 ℃ for 20 min. After washing with phosphate-buffered saline (PBS) for 20 s, hematoxylin was added for counterstaining.

Transmission electron microscopy

Autophagosomes in the cells were observed via transmission electron microscopy (TEM). The cells were fixed with 2.5% glutaraldehyde solution overnight at 4 °C and 1% osmic acid for 2 h. After washing thrice with PBS, the cells were embedded in Epon/Araldite, 1.5% uranyl acetate, and 2% lead citrate. Autophagosomes were observed and pictures were captured using TEM (JEM 1010, JEOL, MA, USA).

Immunofluorescence staining

First, the tissues and cell slides were fixed in paraformaldehyde overnight, dehydrated, and frozen with an embedding agent on a -20 °C freezing table. The frozen tissue blocks were smoothed and cut into 4-µm-thick slices. Second, the temperature was changed to 25 ℃. The samples were washed thrice with 1X PBS for 5 min each. The samples were soaked completely in 0.01 M citrate buffer (pH = 6.0), heated in a microwave oven over medium-low heat for repair, and maintained in a moderate boiling state for 5 min for the best effect. Third, permeabilization was conducted using 0.5% Triton X-100 (prepared in 1 X PBS) for 20 min at 25 ℃. F4/80 (ab111101, 1: 100), CD86 (ab119857, 1:100), and CD206 (ab300622, 1:100) were added to the samples. They were incubated overnight at 4 °C. Phycoerythrin/ fluorescein isothiocyanate-conjugated secondary antibody (Abcam) was added to the samples and incubated for 1 h. Finally, DAPI (Sigma) was added for nucleus counterstaining. Antifade-containing glycerol was used for mounting. Protein localization in the cells was observed under a laser confocal microscope (FV 1200, Faxitron, US).

Quantitative reverse transcription polymerase chain reaction

Total RNAs from the tissues and cells were acquired using TRIzol (Sigma, St. Louis, MO, USA), and the purities were detected. For mRNA, reverse transcription (60 min at 42 °C and 5 min at 70 °C; stored at 4 °C) and quantitative polymerase chain reaction (40 cycles at 95 °C for 10 min, 94 °C for 15 s, 60 °C for 1 min, and 60 °C for 1 min; stored at 4 °C) were performed using the PrimeScript-RT kit (Takara, Japan) and SYBR Premix Ex Taq™ kit (Takara, Japan), respectively. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as the standardized reference. 2−ΔΔCt was used to analyze the relative expression of the target RNAs [28].

For miR-421, the total miRNA was extracted using the miRNeasy Mini kit (GE Healthcare, USA). cDNA was constructed using the TaqMan miRNA reverse transcription kit (DBI Bioscience, Germany). The TaqMan miRNA kit (DBI Bioscience, Germany) was applied to measure miRNA expression. miR-421 expression was normalized to U6 using the 2−ΔΔCt method.

The following primers were designed:

Induced nitric oxide synthase (iNOS): F: 5′-GCCACCAACAATGGCAACAT-3′, R: 5′-AGCAAAGAGGACTGTGGCTC-3′;

Interleukin (IL)-6: F: 5’-CACTTCACAAGTCGGAGGCT-3’, R: 5′-GCCACTCCTTCTGTGACTCC-3′;

Tumor necrosis factor alpha (TNF-α): F: 5’-CACTCTTCAGGGAACCAGGC-3’, R: 5′- GAGTTAGCTAGGCCACCCCA-3′;

Arginase 1 (Arg1): F: 5’-GTCCTTAAGCCGTTCCCTCG-3’, R: 5′-CGGCTGTGCATCATACAACG-3′;

Resistin-like molecule alpha 1 (Fizz1): F: 5’-AGGCCTGCATCAGTCTATGG-3’, R: 5′-TGGGTGATAAAACATGAAACCAAGT-3′;

miR-421: F: 5’-CTCACTCACATCAACAGACATTAATT-3’, R: 5′-GTGCAGGGTCCGAGGT-3′;

GAPDH: F: 5′- GCATCTTCTTGTGCAGTGCC-3′, R: 5′-ACTGTGCCGTTGAATTTGCC-3′; and.

U6: F: 5′- CTCGCTTCGGCAGCACATATACT-3′, R: 5′- ACGCTTCACGAATTTGCGTGTC-3′.

Western blot

Total proteins from the exosomes or cells were extracted and detected using the bicinchoninic acid kit. Briefly, 40 µg of total protein was separated using sodium dodecyl-sulfate polyacrylamide gel electrophoresis (120 V, 90 min) and transferred to polyvinylidene fluoride (PVDF) membranes (90 V, 90 min). For blocking, 5% non-fat milk was added to the PVDF membranes for 1 h. Anti-CD9 (1: 1,000, ab307085, Abcam, Cambridge, MA, USA), anti-CD63 (1: 1,000, ab217345, ab68418), anti-CD81 (1: 1,000, ab109201), anti-SIRT3 (1: 500, ab189860), anti-AMPK (1: 1,000, ab207442), anti-p-AMPK (1: 1,000, ab133448), anti-LC3II/I (1: 1,000, ab48394), anti-P62 (1: 1,000, ab240635), anti-NLRP3 (1: 1,000, ab263899), anti-apoptosis-associated speck-like protein containing a CARD domain (ASC) (1: 1,000, ab70627), anti-IL-1β (1: 1,000, ab283822), anti-iNOS (1:1,000, ab178945), anti-IL-6 (1: 1,000, ab233706), anti-TNF-α (1:1,000, ab183218), anti-Calnexin (1:1000, ab22595) and anti-GAPDH (1:1,000, ab8245) were added and incubated overnight at 4 °C. Moreover, the horseradish peroxidase-labeled goat-anti-rabbit secondary antibody (1:5,000 diluted) was incubated at 25 ℃ for 2 h. Protein blots were detected using Pierce™ ECL (Thermo Fisher, Waltham, USA) in ChemiDoc MP (Bio-Rad, California, USA).

Statistical analysis

Data are expressed as mean ± SD. All statistical analyses were performed using the one-way analysis of variance and Tukey’s multiple comparison tests (GraphPad Prism version 7.0). T test is conducted to analyze the differences between the groups. P-values < 0.05 indicated statistical significance.