Normalizing HIF-1α Signaling Improves Cellular Glucose Metabolism and Blocks the Pathological Pathways of Hyperglycemic Damage

Abstract

Intracellular metabolism of excess glucose induces mitochondrial dysfunction and diversion of glycolytic intermediates into branch pathways, leading to cell injury and inflammation. Hyperglycemia-driven overproduction of mitochondrial superoxide was thought to be the initiator of these biochemical changes, but accumulating evidence indicates that mitochondrial superoxide generation is dispensable for diabetic complications development. Here we tested the hypothesis that hypoxia inducible factor (HIF)-1α and related bioenergetic changes (Warburg effect) play an initiating role in glucotoxicity. By using human endothelial cells and macrophages, we demonstrate that high glucose (HG) induces HIF-1α activity and a switch from oxidative metabolism to glycolysis and its principal branches. HIF1-α silencing, the carbonyl-trapping and anti-glycating agent ʟ-carnosine, and the glyoxalase-1 inducer trans-resveratrol reversed HG-induced bioenergetics/biochemical changes and endothelial-monocyte cell inflammation, pointing to methylglyoxal (MGO) as the non-hypoxic stimulus for HIF1-α induction. Consistently, MGO mimicked the effects of HG on HIF-1α induction and was able to induce a switch from oxidative metabolism to glycolysis. Mechanistically, methylglyoxal causes HIF1-α stabilization by inhibiting prolyl 4-hydroxylase domain 2 enzyme activity through post-translational glycation. These findings introduce a paradigm shift in the pathogenesis and prevention of diabetic complications by identifying HIF-1α as essential mediator of glucotoxicity, targetable with carbonyl-trapping agents and glyoxalase-1 inducers.

Keywords: Warburg effect; carnosine; cellular energetics; diabetes; glycolysis; hyperglycemia; inflammation; methylglyoxal; prolyl 4-hydroxylase 2; trans-resveratrol.

Figure 1

HG increases HIF-1α nuclear translocation in endothelial cells and LPS-stimulated macrophages. Representative IF ((A), scale bar 20 µm) and Western blot (B) for nuclear HIF-1α in HUVEC exposed to HG (20 mM) vs. NG (5.5 mM) for 48 h, with or without the carbonyl trapping agent Car (20 mM) or the glyoxalase inducer Res (10 µM), and relative band densitometry analysis from three separate experiments. Western blot analysis for nuclear HIF-1α in U937 cells stimulated (or not, UNG = untreated NG) with LPS (10 ng/mL) (C), and HCAEC (D), after 48 h incubation with HG vs. NG with or without Car or Res, and relative band densitometry analysis from three separate experiments. Dose response curve (E) of nuclear HIF-1α protein levels (n = 3 separate experiments per condition, black line) and mRNA HIF-1α expression (n = 5 wells in duplicate per condition, blue line) in HUVEC exposed to varying concentrations of glucose ranging from 5,5 mM to 30 mM for 48 h, and representative Western blot image (right side of panel (E)). Time course of HIF-1α mRNA expression in HUVEC (F) and U937 cells stimulated (or not, UNG) with LPS (G), exposed to HG vs. NG for different times ranging from 0 to 72 h, with or without Car; n = 5 wells in duplicate per time point per condition. Bars represent mean ± SEM. Post hoc multiple comparison: *** p < 0.001, ** p < 0.01 or * p < 0.05 vs. NG; ††† p < 0.001, †† p < 0.01 or † p < 0.05 vs. HG.

Figure 2

HG-induced HIF-1α activity is associated with proinflammatory activation in endothelial cells and LPS-stimulated macrophages. HIF-1 activity, as assessed by dual-luciferase gene reporter assay, in HUVEC (A), and in U937 macrophages stimulated (or not, UNG = untreated normal glucose) with LPS (10 ng/mL) (B), after 48 h incubation with HG (20 mM) vs. NG (5.5 mM), with or without Car (20 mM); n = 5 wells in duplicate per condition. VCAM-1 (C) and MCP-1/CCL2 (D) mRNA levels in HUVEC, and IL-1β mRNA levels in U937 macrophages stimulated (or not, UNG) with LPS (E), exposed to HG vs. NG for 48 h, silenced for HIF-1α (si-HIF-1α), or treated with Car; n = 5 wells in duplicate per condition. IL-1β (F) and TNF-α (G) protein levels in the culture medium of U937 cells stimulated (or not, UNG) with LPS after 48 h incubation with HG vs. NG, silenced for HIF-1α, or treated with Car; n = 5 wells in duplicate per condition. Each dot represents the mean of two individual technical replicate and bars represent mean ± SEM. Post hoc multiple comparison: *** p < 0.001, ** p < 0.01 or * p < 0.05 vs. NG; ††† p < 0.001 or † p < 0.05 vs. HG.

Figure 3

The glycolytic side-product MGO increases HIF-1α nuclear translocation in endothelial cells and LPS-stimulated macrophages but does not affect mRNA and protein levels of PHD2. Western blot analysis for HIF-1α in nuclear extracts from HUVEC (A) and U937 cells stimulated (or not, UCtr = untreated control) with LPS (10 ng/mL) (B), treated or untreated (Ctr) with the reactive dicarbonyl compound MGO (200 µM) for 48 h, in the presence or absence of the carbonyl trapping agent Car (20 mM), and relative band densitometry analysis from three separate experiments. PHD2 mRNA ((C), n = 5) and protein levels in total extracts (D), from HUVEC treated or untreated (Ctr) with MGO (200 µM) for 48 h, and relative band densitometry analysis from three separate experiments. Each dot in (C) represents the mean of two individual technical replicate. Bars represent mean ± SEM. Post hoc multiple comparison: *** p < 0.001 or ** p < 0.01 vs. Ctr; ††† p < 0.001 or † p < 0.05 vs. MGO.

Figure 4

The glycolytic side-product MGO activates nuclear translocation of HIF-1α by inducing post-translational glycation and inhibition of PHD2 activity. IP of MGO-protein adducts (MGO adds) (A) or PHD2 (B) on HUVEC treated (MGO) or not (Ctr) with 200 µM MGO for 6 h, using a specific antibody for MGO modified proteins (A) or a specific anti-PHD2 antibody (B), respectively. Mouse IgG were used as control. Total cell lysates (Input) and immunoprecipitates were immunoblotted for PHD2 and MGO protein adducts (A,B). In the presence of oxygen, HIF-1α is rapidly hydroxylated by PHD2 to generate (Pro-OH) HIF-1α, which is degraded through the ubiquitin-proteasome pathway; proteasome inhibition with MG132 led to Pro-OH HIF-1α increase, and adding MGO prevented (Pro-OH) HIF-1α increase and led to HIF-1α nuclear translocation (C). Western blot analysis for (Pro-OH) HIF-1α in total extracts (D), and non-hydroxylated HIF-1α in nuclear extracts (E) from HUVEC, treated or untreated (UCtr) with the proteasome inhibitor MG132 (10µM) for 6 h, with or without MGO, in the presence or absence of Car. Bars represent mean ± SEM. Post hoc multiple comparison: *** p < 0.001 or * p < 0.05 vs. Ctr; ††† p < 0.001 or †† p < 0.01 vs. MGO.

Figure 5

The glycolytic side-product MGO modulates HIF-1α target genes related to cellular glucose metabolism. Western blot analysis of the glycolytic enzymes HK2 (A), PKM2 (B) and LDHA (C), and of the inhibitor of pyruvate dehydrogenase activity PDK1 (D) in total cell extracts from HUVEC treated (or untreated, Ctr = control) with the reactive dicarbonyl compound MGO (200 µM) for 48 h, silenced for HIF-1α (si-HIF-1α), or in the presence of the carbonyl trapping agent Car (20 mM), and relative band densitometry analysis from three separate experiments. mRNA levels of the mitochondrial biogenesis marker PGC-1α (E) in HUVEC exposed to MGO for 48 h, silenced for HIF-1α, or treated with Car; each dot represents the mean of two technical replicates of 5 wells per condition. Representative IF (F, scale bar 20 µm) for MT-CO1 in HUVEC treated or untreated (Ctr) with MGO for 48 h, in the presence or absence of Car. Bars represent mean ± SEM. Post hoc multiple comparison: *** p < 0.001 or * p < 0.05 vs. Ctr; ††† p < 0.001, †† p < 0.01 or † p < 0.05 vs. MGO.

Figure 6

HG and the glycolytic side product MGO induce cellular energetic changes that resemble the Warburg effect and are associated with the activation of the alternative pathological pathways of glucose metabolism in HG conditions. Time course of oxygen consumption (A,B) and lactate production at 48 h (C,D) by HUVEC exposed to HG (20 mM) vs. NG (5.5 mM) (A,C), or treated with MGO (200 µM) vs. untreated cells (Ctr) (B,D), with or without the carbonyl trapping agent Car (20 mM); n = 3 separate experiments in duplicate per condition. Hexosamine (E), polyol (F) and AGE (G) pathways activation in HUVEC exposed to HG vs. NG for 48 h, with or without Car, as determined by measuring the levels of GFPT1 (E), D-sorbitol (F), and AGEs (G); each dot represents the mean of two technical replicates of 5 wells per condition. Bars represent mean ± SEM. Post hoc multiple comparison: *** p < 0.001, ** p < 0.01 or * p < 0.05 vs. NG or Ctr (as appropriate); ††† p < 0.001, †† p < 0.01 or † p < 0.05 vs. HG or MGO (as appropriate).