AMPKα2 controls the anti-atherosclerotic effects of fish oils by modulating the SUMOylation of GPR120

Abstract

Consuming fish oils (FO) is linked to reduced risk of cardiovascular disease in certain populations. However, FO failed to exhibit therapeutic effects in some patients with cardiovascular disease. This study aimed to determine the possible reasons for the inconsistent effects of FO. AMP-activated protein kinase (AMPK) α2 is an important energy metabolic sensor, which was reported to involve in FO mediated regulation of lipid and glucose metabolism. In an in vivo study, FO administration significantly reduced the aortic lesions and inflammation in the Ldlr^-/- mouse model of atherosclerosis, but not in Ldlr^-/-/Prkaa2^-/-and Ldlr^-/-/Prkaa2^-/-Sm22Cre mice. Mechanistically, inactivation of AMPKα2 increased the SUMOylation of the fatty acid receptor GPR120 to block FO-induced internalization and binding to β-arrestin. In contrast, activation of AMPKα2 can phosphorylate the C-MYC at Serine 67 to inhibit its trans-localization into the nuclei and transcription of SUMO-conjugating E2 enzyme UBC9 and SUMO2/3 in vascular smooth muscle cells (VSMCs), which result in GPR120 SUMOylation. In human arteries, AMPKα2 levels were inversely correlated with UBC9 expression. In a cohort of patients with atherosclerosis, FO concentrations did not correlate with atherosclerotic severity, however, in a subgroup analysis a negative correlation between FO concentrations and atherosclerotic severity was found in patients with higher AMPKα2 levels. These data indicate that AMPKα2 is required for the anti-inflammatory and anti-atherosclerotic effects of FO.

Fig. 1. FO administration inhibited atherosclerotic plaque formation in LDLR^−/−, but not LDLR^−/−/AMPKα2^−/− mice.

LDLR^−/−/AMPKα2^−/− mice and their LDLR^−/− littermates were fed a western diet (containing 0.21% cholesterol) for 12 weeks with or without 5% FO administration. A, E, I Representative images of Oil Red O staining from each of four groups: LDLR^−/− (n = 11), LDLR^−/− + FO (n = 17), LDLR^−/−/AMPKα2^−/− (n = 16), and LDLR^−/−/AMPKα2^−/− + FO (n = 22). Scale bar, 200 μm. Quantification of aortic-lesion areas using Image-Pro plus software (B, F, J) and reduced plaque areas (C, G, K) between mice with and without FO treatment. D, H, L Percentages of lesion areas (ratio to lumen areas) of aortic roots were calculated for the four groups. p value vs LDLR^−/− by a two-sided Student’s t test. Data are presented as the mean ± s.e.m. WD western diet.

Fig. 2. AMPKα2 was required for the anti-inflammatory effects of FO in vivo.

Serum MCP-1 (A) and IL-6 (B) levels were measured in LDLR^−/− and LDLR^−/−/AMPKα2^−/− mice fed a western diet with or without 5% FO treatment (n = 6). Representative western blot (C) and quantification of MCP-1 (D) and IL-6 (E) and protein levels in aortic tissues from LDLR^−/− and LDLR^−/−/AMPKα2^−/− mice fed a western diet for 12 weeks, with or without 5% FO administration (n = 3), distinctly from loading controls. ELISAs for MCP-1 (F) and IL-6 (G) in aortic tissue lysates from LDLR^−/− and LDLR^−/−/AMPKα2^−/− mice with or without 5% FO administration (n = 3). H Representative immunofluorescence images for SM α-actin and CD68 in aortic tissues of LDLR^−/− and LDLR^−/−/AMPKα2^−/− mice, with or without 5% FO administration. I, J Quantification of CD68 area and ratio in aortic tissues of LDLR^−/− and LDLR^−/−/AMPKα2^−/− mice, with or without 5% FO administration (n = 4). Scale bars, 200 μm. The data shown represent one of two separate experiments. All values are expressed as means ± s.e.m. p value vs LDLR^−/− mice by two-sided Student’s t tests. FO fish oils, MCP-1 monocyte chemoattractant protein-1, IL-6 interlukin-6.

Fig. 3. AMPKα2 was required for the anti-inflammatory effects of DHA in cultured VSMCs.

A Representative western blot (A) and quantification for the expression of MCP-1 (B) and IL-6 (C) in PA (300 μM)-treated primary cultured WT and AMPKα2^−/− VSMCs with DHA (0, 10, 25, 50, 100 μM). The data shown represent one of at least two separate experiments (n = 2), distinctly from loading controls. ELISAs for MCP-1 (D) and IL-6 (E) in culture supernatants from WT or AMPKα2 deficient-VSMCs treated with or without PA (300 μM) and DHA (0, 10, 25, 50, 100 μM). The data shown represent one of three separate experiments (n = 3). DHA treatment group analysis by one-way ANOVA test. VSMC vascular smooth muscle cell, WT wild type, DHA docosahexaenoic acid, MCP-1 monocyte chemoattractant protein-1, IL-6 interlukin-6.

Fig. 4. Lack of AMPKα2 blocks DHA-induced trans-localization of GPR120 to plasma membrane and interaction with β-arrestin 2.

A Representative Quantification of GPR120 mRNA expression in WT and AMPKα2^−/− VSMCs (n = 4). B, C Western blotting and quantification of GPR120 expression in primary WT and AMPKα2^−/− VSMCs (n = 2). D–F Western blotting and quantification of IL-6 (D) and MCP-1 (E) expression in WT and AMPKα2^−/− VSMCs transfected with Ctr-siRNA or Ffar4-siRNA and treated with PA (300 μM) and/or DHA (50 μM). G Representative immunofluorescence images showing GPR120 localization in WT and AMPKα2^−/− VSMCs, before and after a 1-h DHA treatment. Scale bar, 25 μm. H–K Representative western blots (H, J) and quantification (I, K) of GPR120 expression in cytoplasmic and membrane fractions before and after a 1-h DHA treatment (n = 4). L, M Western blot and quantification of β-arrestin 2 expression in WT and AMPKα2^−/− VSMCs (n = 3). N, O IP assays for studying GPR120–β-arrestin 2 interactions in WT and AMPKα2^−/− VSMCs. All data shown represent one of two separate experiments. Quantification of western blots was performed using Image-Pro plus software, distinctly from loading controls. Data are presented as the mean ± s.e.m. p value by two-sided Student’s t tests. Ctr-siRNA control siRNA, Ffar4-siRNA siRNA for mouse Ffar4, IB immunoblot, IP immunoprecipitation.

Fig. 5. GPR120 SUMOylation at K32 blocked the effects of DHA in AMPKα2^−/− cells.

A, B IP assays showed endogenous GPR120 and SUMOylated GPR120 in WT and AMPKα2^−/− VSMCs by western blotting of whole cell extracts, prepared in the presence of isopeptidase inhibitors (C, D) IP assays for whole cell extracts from 293T cells co-transfected with either the WT GPR120-tGFP or K32R/GPR120-tGFP vector and the SUMO2-HA vector. E, F Western blotting (E) and quantification (F) of tGFP expression in membrane fractions from AMPKα2^−/− VSMCs transfected with GPR120-tGFP or K32R/GPR120-tGFP and treated for 1 h with DHA. Scale bars, 25 μm. G–I Western blotting (G) and quantification of IL-6 (H) and MCP-1 (I) expression in AMPKα2^−/− VSMCs transfected with GPR120-tGFP or K32R/GPR120-tGFP and treated with PA (300 μM) and DHA (50 μM) for 1 h. All IP and western blotting data shown represent one of two separate experiments, distinctly from loading controls. Quantification of western blot results was performed using Image-Pro plus software. Data are presented as the mean ± s.e.m. p value by two-sided Student’s t tests. K32R lysine 32 mutated to arginine, PA palmitic acid, DHA docosahexaenoic acid.

Fig. 6. Increases in the expression of UBC9 and SUMO2/3 promotes the GPR120 SUMOylation in AMPKα2^−/− VSMCs by activating the C-MYC S67 phosphorylation.

A, B Western blotting and quantification of UBC9 and GPR120 expression in cytoplasmic and membrane fractions from AMPKα2^−/− VSMCs transfected with Ctr-siRNA or Ube2i-siRNA, with or without DHA (50 μM) treatment (n = 4). p value versus Ctr-siRNA by one-way ANOVA tests. C–E Western blotting and quantification of UBC9, IL-6, and MCP-1 expression in AMPKα2^−/− VSMCs transfected with Ctr-siRNA or Ube2i-siRNA before or after PA (300 μM) and DHA (50 μM) treatment for 1 h (n = 2). F, G Western blotting and quantification of c-myc, UBC9, and SUMO2/3 expression in AMPKα2^−/− VSMCs transfected with Ctr-siRNA or Myc-siRNA (n = 4). H–K Western blotting (H) and quantification of tGFP (I), UBC9 (J), and SUMO2/3 (K) expression in 293T cells transfected with WT C-myc (WT), S64A/C-myc (S64A), or S67A/C-myc (S67A) vectors and treated for 2 hrs with AICAR (1 mM) (n = 2). L Extracts from 293T cells transfected with either WT, S64A, or S67A c-myc constructs were prepared to assess the interactions between the c-myc-tGFP fusion protein and AMPKα2 or pAMPKα. M Recombinant AMPK and recombinanted c-myc-tGFP protein were incubated in the kinase buffer with ATP. Immunoprecipitation was done using anti-tGFP antibody. N The phosphorylation of serine at amino acids 67 in the peptide PTPPLSPSRRSG by LC-MS/MS followed processed using Proteome Discoverer 1.3. O Representative images of immunostaining showing AMPKα2 and UBC9 expression and localization in human artery tissues from control (n = 4) and CAD patients (n = 8). Scale bar, 400 μM. P Semi-quantification analysis of immunostaining using Image-Pro plus software. Correlational analyses of AMPKα2 expression and the expression of UBC9 (Q) in human artery tissues. All data shown represent one of three separate experiments. Quantification of western blots was performed using Image-Pro plus software, distinctly from loading controls. Data are presented as the mean ± s.e.m. p value by two-sided Student’s t tests. AICAR 5-aminoimidazole-4-carboxamide 1-α-d-ribofuranoside, Ube2i-siRNA siRNA against mouse Ube2i, Myc-siRNA siRNA against mouse Myc, CAD coronary artery disease.

Fig. 7. AMPKα2 expression determined the protective effect of FO in CAD patients.

Plasma DHA (A), EPA (B), or AMPKα2 (C) in platelet versus CAD severity in 349 patients based on vessel-lesion numbers: 0 no lesion vessel represent non-CAD, 1 single vessel represents low severity, 2 two vessels represent medium severity, and 3 three vessels represent high severity. Data are presented as the median ± s.e.m. and were analyzed by one-way ANOVA with the Kruskal–Wallis test for multiple comparisons. Plasma DHA (D) and EPA (F) concentrations in CAD patients and vessel-legion numbers versus AMPKα2 activation in platelet-rich plasma. A grayscale value of <8.2 pg/mL for IP assay obtained AMPKα2 activation indicates low AMPKα2 activity and a value of ≥8.2 indicates high AMPKα2 activity. p value via one-way ANOVA with the Kruskal–Wallis test for multiple comparisons. Scatter plots showing correlations between plasma DHA (E) or EPA (G) concentrations in CAD patients with high AMPKα2 activation (n = 135). Spearman’s correlation-coefficient test was used to calculate r and p values. DHA docosahexaenoic acid, EPA eicosapentaenoic acid. H A Schematic figure showed that AMPKα2 mediated the anti-inflammation of FO. In physiological condition, AMPKα2 can phosphates C-MYC at 67 serine to block its translocation into the nuclei, then represses the transcription of UBC9 and SUMO2/3 and GPR120 SUMOylation. Pathologically, inactivation of AMPKα2 leads to the nuclei trans-localization of C-MYC to enhance the expression of UBC9 and SUMO2/3 and GPR120 SUMOylation in VSMCs.