Activation of GPR3-β-arrestin2-PKM2 pathway in Kupffer cells stimulates glycolysis and inhibits obesity and liver pathogenesis

Abstract

Kupffer cells are liver resident macrophages and play critical role in fatty liver disease, yet the underlying mechanisms remain unclear. Here, we show that activation of G-protein coupled receptor 3 (GPR3) in Kupffer cells stimulates glycolysis and protects mice from obesity and fatty liver disease. GPR3 activation induces a rapid increase in glycolysis via formation of complexes between β-arrestin2 and key glycolytic enzymes as well as sustained increase in glycolysis through transcription of glycolytic genes. In mice, GPR3 activation in Kupffer cells results in enhanced glycolysis, reduced inflammation and inhibition of high-fat diet induced obesity and liver pathogenesis. In human fatty liver biopsies, GPR3 activation increases expression of glycolytic genes and reduces expression of inflammatory genes in a population of disease-associated macrophages. These findings identify GPR3 activation as a pivotal mechanism for metabolic reprogramming of Kupffer cells and as a potential approach for treating fatty liver disease.

Fig. 1. DPI stimulates both rapid and sustained increases in glycolysis in macrophages.

a, b. The short-term effects of DPI on ECAR (a) and OCR (b) in ImKCs. ECAR and OCR were measured by Seahorse analyzer in ImKCs for 20 min, then for another 120 min following the addition of different concentrations of DPI (5, 50, or 500 nM), and then for another 40 min following the addition of 2-deoxylglucose (2-DG) (a) or rotenone plus antimycin A (Rot/AA) (b). Shown are representative data as the mean ± sd from three independent experiments. c, d The long-term effects of DPI on ECAR in ImKCs. ImKCs were seeded and incubated with or without DPI (50 and 500 nM) for 24 h. ECAR values were then measured under the basal conditions with sequential addition of 15 mM glucose, 2 µM oligomycin, and 50 mM 2-DG (c). Specific parameters for glycolysis, glycolytic capacity, and glycolytic reserve were calculated and data are presented as the mean ± sd from three independent experiments (n = 18 biological replicates) (d). e Select metabolite levels. ImKCs were treated with DPI for 6 h, and select metabolites in the glycolytic pathway and TCA cycle was quantified by LC–MS. Data are presented as the mean ± sd (n = 4 biological replicates). P values were calculated by the two-sided student’s t-test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are provided as a Source Data file.

Fig. 2. DPI stimulates glycolysis through GPR3 and β-arrestin2.

a DPI-stimulated glycolysis is independent of NOX activity. Wildtype (WT) and p47phox^−/− BMDMs were treated with DMSO or DPI (50 and 500 nM) for 24 h and ECAR was measured by Seahorse analyzer. Data are presented as the mean ± sd from three independent experiments (n = 15 biological replicates). b The effect of DPI on glucose uptake in WT and p47phox^−/− BMDMs. BMDMs were treated with DMSO or DPI for 24 h in the presence of the fluorescent glucose analog 2-NBDG. The mean fluorescence intensity (MFI) of 2-NBDG in cells was measured by flow cytometry and normalized to DMSO controls. Data are presented as the mean ± sd (n = 3 independent experiments). c ECAR in WT BMDMs without or with DPI in the absence or the presence of the NOX inhibitor apocynin (100 μM) for 24 h. Data are presented as the mean ± sd from three independent experiments (n = 12 biological replicates). d, e DPI-stimulated glycolysis requires GPR3. ECAR (d) and glucose uptake (e) were measured in WT and Gpr3^−/− BMDMs without or with DPI treatment as described in (a, b). n = 15 biological replicates in (d) from three independent experiments and n = 4 independent experiments in (e). f ECAR in ImKCs treated with DMSO, DPI (500 nM), or S1P (3 mM) for 24 h. Data are presented as the mean ± sd from three independent experiments (n = 12 biological replicates). g DPI induces β-arrestin2 translocation to cytoplasm membrane. ImKCs were transfected with the Arrb2-GFP fusion gene and stimulated with DMSO, DPI (50 nM), or S1P (3 mM). The GFP signal was captured with a TIRF microscope at indicated time points. Shown are representative data from three independent experiments. Scale bar: 5 μm. h, i DPI-stimulated glycolysis requires β-arrestin2. ECAR (h) and glucose uptake (i) were measured in WT and Arrb2^−/− ImKCs treated with DMSO or DPI. n = 15 biological replicates in (h) from three independent experiments and n = 4 independent experiments in (i). P values were calculated by the two-sided student’s t-test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are provided as a Source Data file.

Fig. 3. DPI stimulates a rapid increase in glycolytic activity through the formation of GPR3- β-arrestin2-GAPDH-PKM2 enzymatic super complex.

a Co-IP of β-arrestin2 with ERK1/2, enolase, GAPDH, and PKM2. ImKCs were transfected with β-arrestin2 and then treated with or without 50 nM DPI for 6 h. Cell lysates were immunoprecipitated with anti-β-arrestin2 and the precipitates were analyzed by Western blotting for the indicated proteins. Shown are representative data from one of the three experiments. b DPI-stimulated glycolysis requires PKM2. BMDMs were prepared from wildtype and Pkm2^−/− mice, seeded, and incubated with or without DPI (50 and 500 nM) for 24 h, and ECAR was measured by a Seahorse analyzer. Data are presented as the mean ± sd from three independent experiments (n = 15 biological replicates). c WT and Pkm2^−/− BMDMs were seeded and incubated with or without DPI (50 and 500 nM) for 24 h in the presence of 2-NBDG to measure glucose uptake. Data are presented as the mean ± sd (n = 4 independent experiments). d, e DPI stimulates enzymatic activities of PKM2 and GAPDH. Wildtype and Arrb2^−/− ImKCs were treated with vehicle (black line) or 500 nM DPI (red line) for 6 h and the enzymatic activities of PKM2 (d) and GAPDH (e) were measured by colorimetric assay kits (Biovision). Data are presented as the mean ± sd from n = 3 independent experiments. P values were calculated by the two-sided student’s t-test. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001. Source data are provided as a Source Data file.

Fig. 4. DPI stimulates a sustained increase in glycolytic activity through nuclear translocation of PKM2 and transcriptional activation of c-Myc.

a DPI-induced transcription of glycolytic genes requires PKM2. WT and Pkm2^−/− BMDMs were treated with DPI (50 and 500 nM) or vehicle for 24 h. The transcript levels of Pkm2, Ldha, Hk2, and c-Myc were measured by real-time qPCR. Data were collected from two independent experiments with three biological replicates per group. Transcriptional level was normalized to β-actin first and then to DMSO control. Data are presented as the mean ± sd. b, c Induction of dimeric PKM2 by DPI. ImKCs were treated with vehicle or DPI (50 and 500 nM) for 1 h. Cell lysates were run on native PAGE gel and analyzed by Western blotting. Shown are representative data (b) and summarized data (c) quantified by ImageJ from three independent experiments. d DPI induces nuclear translocation of PKM2 by Western blotting. ImKCs were treated with vehicle or DPI (50 nM) for 6 h. Proteins from cytosolic and nuclear fractions were isolated and analyzed by anti-PKM2 Western blotting. Shown are representative data, and the numbers are average expression levels of cytosolic and nuclear PKM2 from three independent experiments. e DPI induces nuclear translocation of PKM2 by confocal microscopy. ImKCs and human primary KCs (AcceGen) were treated with vehicle or DPI (50 nM) for 24 h, stained with anti-PKM2 (green) and DAPI (red), followed by confocal imaging. Shown are representative images from two independent experiments. The boxed areas are enlarged and scale bars are indicated. f DPI stimulates transactivation of c-Myc. c-Myc luciferase reporter plasmid was transfected into WT and Pkm2^−/− BMDMs. Transfected cells were treated with vehicle or DPI (50 and 500 nM) for 6 h, and luciferase activities were measured. Data are presented as the mean ± sd from two independent experiments (n = 5 biological replicates). P values were calculated by the two-sided student’s t-test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are provided as a Source Data file.

Fig. 5. DPI inhibits HFD-induced obesity and liver pathogenesis through PKM2 expression in Kupffer cells.

a–f DPI prevents weight gain in mice fed with HFD. Male B6 mice at 5 weeks of age were fed with HFD or normal chow (NC) for a total of 8 weeks (week 8). Three weeks after HFD (arrow), the mice were given either vehicle (Veh) or DPI (2 mg/kg) every 5 days for a total of 6 doses. The body weight (a) and food consumption (b) were monitored weekly. Data are presented as the mean ± sd from three independent experiments (n = 5–10 mice per group per experiment). c The weights of eWAT and iWAT at week 8. d Fasting glucose tolerance assay. At week 7 plus 3 days, mice were starved overnight (12–16 h) with water ad libitum. Glucose (1 g/kg) was injected i.p. and blood glucose levels were monitored at the indicated time. AUC (right panel) was calculated for statistics. e Serum AST and ALT activities in mice at week 8. f Comparison of H&E staining of liver sections at week 8. Shown are representative H&E staining from one mouse per group from (a). g–i Effect of DPI on KC-specific Pkm2^−/− mice fed with HFD. Experimental procedures with male KC-specific Pkm2^−/− mice and control Pkm2^f/f mice were described in (a). Body weights were monitored weekly (g). Data are presented as the mean ± sd from two independent experiments with n = 5 mice per group per experiment. P values were calculated by the two-sided student’s t-test and shown in Supplementary Fig. 9a. Mice fed with NC are used as controls. Statistical P-values between groups were shown in Supplementary Fig. 9a. Comparison of H&E staining of liver sections at week 8 (h). Shown are representative H&E staining from one mouse per group. Black arrows in f and h point to lipid droplets. Scale bar: 100 μm. F4/80⁺ Liver macrophages were purified from mice in g at week 8 using anti-F4/80 microbeads. ECAR was measured directly by Seahorse analyzer (i). Data are presented as the mean ± sd (n = 5). Each dot represents one mouse in (c–e) and (i). P values were calculated by the two-sided student’s t-test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant. Source data are provided as a Source Data file.

Fig. 6. DPI stimulates the expression of glycolytic genes but suppresses the expression of inflammatory genes in KCs from HFD-fed mice.

a Comparison of gene expression in KCs isolated from mice fed with NC or HFD. Single-cell suspensions were prepared from mice from Fig. 5a after 8 weeks on HFD (6 mice per group), stained with anti-F4/80, anti-CD11b, and anti-Gr-1. F4/80⁺CD11b⁺Gr1^low macrophages were purified by cell sorting followed by RNAseq. Shown are differentially expressed genes among the three groups. b Functional enrichment analysis of differentially expressed genes based on comparison of KCs from HFD-fed and NC-fed mice or from HFD-fed mice treated with DPI or vehicle. c GSEA of gene expression profiles of KCs either from HFD and NC mice or from HFD mice treated with DPI or vehicle. Orange and blue graphs in b and c indicate up- and down-regulated pathways as labeled. d Macrophage polarization index analysis based on the expression profile in (a) above using the online software MacSpectrum (https://macspectrum.uconn.edu). M1-type polarization is expressed as positive scores, whereas M2-type polarization is expressed as negative scores. e, f Purified F4/80⁺ liver macrophages from DPI-treated or vehicle-treated mouse livers were incubated with or without 100 ng/mL LPS for 6 h. Cytokines TNF-α, IL-1β, and IL-6 in the culture supernatant were quantified by ELISA (e) and intracellular ROS was assayed by CM-H2DCFDA staining (f). Each dot in e represents one mouse (n = 5) from one experiment. Data are presented as the mean ± sd. Representative histograms of ROS levels in f were shown. P values were calculated by the two-sided student’s t-test. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001. Source data are provided as a Source Data file.

Fig. 7. DPI upregulates glycolysis and suppresses inflammatory responses of Kupffer cells from patients with NAFLD.

a, b scRNAseq analysis of the liver macrophage populations. A total of 5497 macrophages based on the expression of CD14 and CD68 (clusters 5, 8, and 12 in Supplementary Fig. 10c) were subjected to clustering analysis by tSNE. A total of 7 clusters were identified (a). Each cluster was annotated based on the expression of typical markers or specific functional markers, as shown by dot plot (b). c Trajectory inference of the liver macrophages by slingshot. d, e GO enrichment analysis of differentially expressed genes (DEGs) between C3 and C0 (d) or C3 and C1/2 (e). f, g Comparison of gene expression changes induced by DPI in primary KCs isolated from NAFLD liver biopsies. CD14⁺ KCs were sorted from single cell suspensions of NAFLD human liver biopsies (n = 2) and treated with DMSO or DPI (500 nM) for 24 h, followed by RNAseq to quantify gene expression. Shown are the expression changes of glycolytic genes and DAM markers (f) and GO enrichment analysis of DEGs induced by DPI in KCs (g). Orange and blue graphs in d, e, and g indicate upregulated and downregulated pathways as labeled. P values in d, e, and g were computed by Fisher’s Exact test. Source data are provided as a Source Data file.