Animals and intervention

All animal experimental procedures were approved by Peking University Animal Care and Use Committee (LA2020508). Animals were housed conventionally with free access to food and water on a 12-h light/dark cycle.

Seven-week-old male db/db (BKS-Leprem2Cd479/Gpt) mice (GemPharmatech, Nanjing, China), a NAFLD model [12, 13], were adaptively fed for one week, and weight-matched animals were randomly assigned to three groups (n = 9 per group). Dapagliflozin (1 mg·kg−1·d−1, HY-10450; MedChemExpress, Monmouth Junction, NJ, USA), canagliflozin (10 mg·kg−1·d−1, HY-10451; MedChemExpress) or vehicle were administered intragastrically daily for 8 weeks. The doses of dapagliflozin and canagliflozin were referred to the previous studies [14,15,16,17]. Vehicle-treated male db/m littermates were used as normal control (n = 10).

Six-week-old male C57BL/6J mice (GemPharmatech) were randomized into chow diet (CD) feeding (n = 6) and western diet (WD) feeding (n = 45). Mice were fed for 20 weeks on a CD (fat 10%, carbohydrate 70% and protein 20%; Research Diets, New Brunswick, NJ, USA) and on a high-fat, high-fructose and high-cholesterol diet (palm oil 40%, fructose 20% and cholesterol 2%; Research Diets), respectively [12, 13]. WD-fed mice weighing over 40 g were chosen for further experiments (n = 20) [18, 19], and divided into three groups (n = 6−7 per group). For the weight-matched WD-fed mice, dapagliflozin (1 mg·kg−1·d−1), canagliflozin (10 mg·kg−1·d−1) or vehicle were administered intragastrically daily for 8 weeks. Vehicle-treated CD-fed mice served as normal control (n = 6).

Since global homozygous disruption of Pfkfb3 results in embryonic lethality before E8 [20], global heterozygous Pfkfb3 knockout (Pfkfb3+/−) mice were usually used for the experiments [21,22,23,24]. Pfkfb3+/− mice were generated on a C57BL/6J genetic background by GemPharmatech. Briefly, CRISPR/Cas9 technology was used to modify Pfkfb3 gene and caused a knockout of the 223-bp coding sequence in exon2-exon3 of Pfkfb3. Six-week-old male Pfkfb3+/− mice and littermate Pfkfb3+/+ mice (n = 6 per group) were fed on a WD for 12 weeks that were used for inducing metabolic inflammation [25].

Six-week-old male C57BL/6J mice were fed on a WD for 8 weeks that were used for inducing metabolic inflammation [25]. WD-fed mice were divided into two groups (n = 6 per group) and were treated for 4 weeks with the PFKFB3-specific inhibitor 3-(3-pyridinyl)-1-(4-pyridinyl)-2-propen-1-one (3PO; 50 mg/kg, 4 times every week, HY-19824; MedChemExpress) or vehicle.

Histological staining

For H&E, Masson and Sirius red staining, liver tissues were fixed in a 4% (w/v) paraformaldehyde solution overnight, and then embedded in paraffin and cut into 5-µm sections. For oil red O staining, liver tissues were cryopreserved in O.C.T. compound (Tissue-Tek; Torrance, CA, USA) and prepared into 10-µm sections. Sections were subjected to staining by using H&E, Masson, Sirius red and oil red O staining kits (all from Solarbio, Beijing, China) following the manufacturer’s instructions. The stained sections were analyzed using a digital scanner (NanoZoomer-SQ; Hamamatsu, Hamamatsu City, Japan). The kits are summarized in Supplementary Table S1.

Immunofluorescent staining

Liver frozen sections were prepared in the same way as for oil red O staining. Sections were blocked with goat serum and incubated with primary antibodies at 4 °C overnight. Subsequently, sections were incubated with secondary antibodies for 1 h at room temperature. Finally, sections were stained with DAPI. The fluorescence was visualized with a Leica TCS SP8 confocal fluorescence microscope (Leica Microsystems, Wetzlar, Germany). The specific information of all antibodies is listed in Supplementary Table S2.

Liver lipid, cytokine and chemokine analyses

For each mouse, liver sample from the same region of left lobe was collected for the following detection. Liver tissues (100 mg) were homogenized in radioimmunoprecipitation assay lysis buffer (Applygen, Beijing, China). Biochemical assay kits for triglyceride (TG) and total cholesterol (TC) (both from Applygen), and specific ELISA kits for mouse TNF-α, IL-1β, IL-6, CCL-2, IL-10 and TGF-β (all from Invitrogen, Carlsbad, CA, USA) were used to detect the lipids, cytokines and chemokines in liver tissue lysates in accordance with the manufacturer’s instructions. The specific information of all kits is listed in Supplementary Table S1.

Culture and intervention of primary mouse bone marrow-derived macrophages (BMDMs)

To obtain murine BMDMs, the femurs of C57BL/6 mice aged 6−10 weeks were dissected, the bone ends were removed, and bone marrow was flushed from femurs with sterile phosphate buffered saline (PBS). Bone marrow cells were collected by centrifugation in DMEM (Gibco, Grand Island, NY, USA) at 400 × g for 4 min. Subsequently, cells were suspended in a complete DMEM medium containing 10% fetal bovine serum (FBS; Gibco), 20 ng/mL recombinant macrophage colony-stimulating factor (M-CSF; PeproTech, Rocky Hill, NJ, USA) and 1% penicillin-streptomycin (Gibco) in petri dish at a density of 1 ×106 cell/mL, and the day was marked as day 0. On day 3, half of the medium was replaced with fresh medium. By day 7, cells were deemed as naïve macrophages and prepared for the experiments. BMDMs were cultured for 12 h with dapagliflozin (10, 20 and 40 μmol/L) or canagliflozin (10, 20 and 40 μmol/L) or vehicle in the presence or absence of LPS (100 ng/mL; Sigma-Aldrich, St. Louis, MO, USA) + IFN-γ (50 ng/mL; PeproTech).

Culture and intervention of primary human peripheral blood mononuclear cell (PBMC)-derived macrophages

PBMCs were collected from patients with type 2 diabetes. Written informed consent was obtained from blood donors in accordance with ethical approval from the Ethics Committee of Peking University Third Hospital (M2022039).

PBMCs were separated using Ficoll gradient centrifugation. Briefly, blood was diluted at 1:1 in PBS. Ficoll solution (5 mL; Sigma-Aldrich) was added to a 15-mL centrifuge tube and the diluted blood (10 mL) was softly added on the top of Ficoll solution. The tube was centrifuged at 700−800 × g for 20−30 min at 4 °C (no-brake). PBMCs were obtained from the interphase and rinsed twice with PBS. Subsequently, PBMCs were purified by using magnetic bead sorting with MojoSort™ human CD14 selection kits (Biolegend, San Diego, CA, USA). The kits are listed in Supplementary Table S1.

Purified PBMCs were suspended in RPMI medium (Gibco) with 25 ng/mL M-CSF, 10% FBS and 1% penicillin-streptomycin at a density of 1 ×106 cell/mL, and pre-incubated for 6 d. On day 3, half of the medium was replaced by fresh medium. The PBMC-derived macrophages were cultured for 12 h on day 7 with 20 μmol/L dapagliflozin, canagliflozin or vehicle in the presence or absence of LPS (100 ng/mL) + IFN-γ (50 ng/mL).

Culture and intervention of mouse macrophage and hepatocyte cell lines

Mouse macrophage cell line RAW264.7 cells (Procell, Wuhan, China) were maintained in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin, and 2.5 ×105 cells were placed in each well of 6-well plates for seeding. After a 12-h pre-incubation, RAW264.7 cells were cultured for 12 h with 20 μmol/L dapagliflozin, canagliflozin or vehicle in the presence or absence of LPS (100 ng/mL).

Mouse hepatocyte cell line AML-12 cells (Procell) were incubated in DMEM/F12 (Gibco) containing 10% FBS, 10 μg/mL insulin, 5.5 μg/mL transferrin, 40 ng/mL dexamethasone, 5 ng/mL selenium and 1% penicillin-streptomycin (all from Gibco), and 2 ×105 cells were placed in each well of 6-well plates for seeding. After a 24-h pre-incubation, AML-12 cells were cultured for 24 h with 20 μmol/L dapagliflozin, canagliflozin or vehicle in the presence or absence of palmitic acid (PA; 100 μmol/L; Sigma-Aldrich) + oil acid (OA; 200 μmol/L; Sigma-Aldrich) that were used for inducing an in vitro model of liver lipid accumulation.

Co-culture of mouse macrophage and hepatocyte cell lines

Transwell chamber with 0.4-μm pores (Corning, New York, NY, USA) was used to establish the co-culture system. RAW264.7 cells were seeded in top chamber and pre-incubated for 12 h with 20 μmol/L dapagliflozin, canagliflozin or vehicle in the presence or absence of LPS (100 ng/mL). AML-12 cells were seeded in bottom chamber and pre-incubated for 24 h with PA (100 μmol/L) + OA (200 μmol/L). Two types of cells were pre-incubated separately. Before co-culturing, cells were rinsed with PBS and new culture media without the above agents were replaced. After a 24-h incubation in the co-culture systems, AML-12 cells were harvested for subsequent analysis.

ELISA and chemical assays

Specific ELISA kits for mouse TNF-α, IL-1β, IL-6, CCL-2 and IL-10, and human TNF-α, IL-6 and CCL-2 (all from Invitrogen) were used to measure the cytokines and chemokines in cell-free culture supernatants according to the manufacturer’s instructions. Nitric oxide was measured by using a chemical assay kit (Beyotime, Shanghai, China) in accordance with the manufacturer’s instructions. The specific information of all kits is summarized in Supplementary Table S1.

Flow cytometry

Cells were collected, rinsed in cold PBS containing 1% FBS, and centrifuged at 400 × g for 5 min at 4 °C. The supernatant was discarded and the cells were resuspended in PBS, followed by addition of required antibodies against macrophage markers in accordance with the manufacturer’s recommendation. The antibodies are summarized in Supplementary Table S2. After vortexed for 5 s, cells were incubated for 30 min at 4 °C in the dark. Cells were rinsed twice with PBS, centrifuged at 400 × g for 5 min, and then resuspended in 100 μL PBS. The cells were quickly analyzed by a flow cytometer (CytoFlex; Beckman Coulter, Miami, FL, USA). Data were analyzed by using CytExpert 2.4 software (Beckman Coulter) and FlowJo 7.6.1 software (FlowJo LLC, Ashland, OR, USA).

RNA-seq and data analysis

To examine the effect of SGLT2i on macrophage transcriptome, BMDMs were cultured for 12 h with 20 μmol/L dapagliflozin, canagliflozin or vehicle in the presence or absence of LPS (100 ng/mL) and IFN-γ (50 ng/mL). Four sets of BMDMs were harvested for extracting total RNA with TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA) in accordance with the manufacturer’s instructions. PCR was used to amplify cDNA fragments following reverse transcription. Illumina Genome Analyzer II system was utilized for RNA-seq (MetWare Biotechnology Co., Ltd., Wuhan, China). Gene expression was quantified with fragments per kilobase of transcript per million mapped reads. Differentially expressed genes (DEGs) were determined using R software by identifying genes with P-value < 0.01 and fold change >1.5 or <0.666. Enriched biological processes were analyzed by using Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis.

Seahorse analysis

Extracellular acidification rate (ECAR) and oxygen consumption rate (OCR), indicative of glycolysis and respiration respectively, were monitored using an XF96 Extracellular Flux Analyzer (Agilent Technologies, Santa Clara, CA, USA) with Seahorse XF glycolytic rate assay kit and Seahorse XF cell Mito stress test kit (both from Agilent Technologies), respectively. The kits are listed in Supplementary Table S1. Briefly, BMDMs were suspended in phenol red-free DMEM media supplemented with 10 mmol/L glucose, 2 mmol/L glutamine and 1 mmol/L pyruvate. Cells were placed into a 96-well Fluxpak (Agilent Technologies) at a density of 1 ×104 cells per well, and then cultured for 12 h with 20 μmol/L dapagliflozin, canagliflozin or vehicle in the presence or absence of LPS (100 ng/mL) + IFN-γ (50 ng/mL). Mitochondrial respiratory parameters and glycolytic parameters were assessed by OCR (pmol/min) and ECAR (mpH/min), respectively using the injection solutions: oligomycin, carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone, rotenone/antimycin A, and 2-deoxy-D-glucose.

Glucose uptake assay

BMDMs were equilibrated in PBS with 1% FBS for 30 min. Cells were cultured for 30 min at 37 °C in PBS containing 100 μmol/L 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino]-2-deoxy-D-glucose (Invitrogen) with 20 μmol/L dapagliflozin or canagliflozin or vehicle in the presence or absence of LPS (100 ng/mL) + IFN-γ (50 ng/mL). The cells were washed twice by PBS and the flow cytometry analysis was carried out using a flow cytometer (CytoFlex).

Metabolite detection by LC-MS/MS

Intracellular metabolite detection was conducted using LC-MS/MS as previously described [26]. ACQUITY UPLC-I-Class was coupled to Xevo TQ-S micro-ESI tandem mass system (Waters, Milford, MA, USA) for metabolite separation and detection. A reversed-phase chromatography method with an Agilent Poroshell 120 SB-Aq column (2.1 mm × 100 mm, 2.7 μm) was used for compound separation at 30 °C. Mobile phase used ultrapure water containing 5 mmol/L ammonium acetate and acetonitrile in A phase and B phase, respectively. Liquid chromatography gradient was listed as follows: the starting condition was 100% A phase, then the proportion of B phase gradually increased to 50% from 0.5 min to 4 min, and flushing ratio was subsequently stable with 100% B phase from 4 min to 6 min. At 6 min, the proportion of mobile phase returned to the initial proportion, and balanced for 2 min. The flow rate was 0.3 mL/min and inject volume was 2 μL for electrospray ionization (ESI)+ and 5 μL for ESI−.

Triple quadrupole mass spectrometer was equipped with ESI ion source, and the relevant parameters were listed as follows: capillary voltage, 3.0 kV for ESI+ mode and 2.0 kV for ESI− mode; source temperature, 100 °C; temperature of desolvention gas, 400 °C; flow rate of cone gas, 100 L/h; flow rate of desolvention gas, 1200 L/h. The parameters of mass spectrum acquisition channels for each compound were shown in Supplementary Table S3 (the positive and negative modes were used to collect different compounds for each sample). The multiple reaction monitoring parameters for compounds are listed in Supplementary Table S3.

Pfkfb3 knockdown and overexpression

For Pfkfb3 knockdown, RAW264.7 cells were transfected for 24 h with 50 nmol/L Pfkfb3 small interfering RNAs (siRNAs) (si-Pfkfb3-A and si-Pfkfb3-B) or negative control siRNA (Genechem Co., Ltd., Shanghai, China) using Lipofectamine™ RNAiMAX (Thermo Fisher Scientific). The siRNA sequences against Pfkfb3 are summarized in Supplementary Table S4. For Pfkfb3 overexpression, RAW264.7 cells were transfected for 24 h with 1 μg Pfkfb3 overexpression plasmid or empty vector (Genechem Co., Ltd.) using Lipofectamine™ 3000 (Thermo Fisher Scientific).

Quantitative real-time PCR

TRIzol reagent was utilized to isolate total RNA, and the RevertAid First Strand cDNA Synthesis kit (Thermo Fisher Scientific) was employed to synthesize the cDNA. Quantitative real-time PCR was performed using THUNDERBIRD SYBR qPCR Mix (Toyobo Co., Ltd., Osaka, Japan) on the QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific). The relative expression of target genes was normalized to Actb and calculated by the 2−ΔΔCt method. The primer sequences are listed in Supplementary Table S5.

Protein quantitation and Western blot

Protein quantitation was performed by using bicinchoninic acid assay (Thermo Fisher Scientific). The denatured proteins (20 μg) were separated by 10% (w/v) SDS-PAGE electrophoresis and then transferred to nitrocellulose membranes. The membranes were incubated overnight with primary antibody at 4 °C, followed by a 1-h incubation with secondary antibody at room temperature. Protein bands were visualized by using an Odyssey 290 infrared imaging system (LI-COR Biosciences, Lincoln, NE, USA). Images were analyzed by Image J software (National Institutes of Health, Bethesda, MD, USA). β-Actin was used as a loading control.

Statistical analysis

If data were Gaussian distributed, they are presented as mean ± SD and were analyzed by ANOVA followed by the post hoc Tukey-Kramer test, or by unpaired Student’s t test (two-tailed), as appropriate. If data were not Gaussian distributed, they are presented as median (interquartile range) and were analyzed by Kruskal–Wallis test followed by the Dunn multiple comparisons test. P < 0.05 was considered statistically significant. Statistical analysis was carried out using GraphPad Prism 9.0 (GraphPad Software, San Diego, CA, USA).