Increased adipocyte O2 consumption triggers HIF-1α, causing inflammation and insulin resistance in obesity

Abstract

Adipose tissue hypoxia and inflammation have been causally implicated in obesity-induced insulin resistance. Here, we report that, early in the course of high-fat diet (HFD) feeding and obesity, adipocyte respiration becomes uncoupled, leading to increased oxygen consumption and a state of relative adipocyte hypoxia. These events are sufficient to trigger HIF-1α induction, setting off the chronic adipose tissue inflammatory response characteristic of obesity. At the molecular level, these events involve saturated fatty acid stimulation of the adenine nucleotide translocase 2 (ANT2), an inner mitochondrial membrane protein, which leads to the uncoupled respiratory state. Genetic or pharmacologic inhibition of either ANT2 or HIF-1α can prevent or reverse these pathophysiologic events, restoring a state of insulin sensitivity and glucose tolerance. These results reveal the sequential series of events in obesity-induced inflammation and insulin resistance.

Figure 1

Increased adipocyte oxygen consumption by FFA-induced uncoupled respiration contributes to adipose tissue hypoxia in obesity. (A) Whole-mount immunohistochemistry analysis of eWAT from mice fed normal chow diet (NCD) or HFD for 3 days. Green, pimonidazole adduct; Red, perilipin; Blue, PECAM-1. (B) Western blot analysis of HIF-1α in eWAT. (C-D) Q-PCR analysis of VEGF expression (C) or lactate concentration (D) in eWAT from mice fed NCD or HFD. D1, 1 day HFD; D3, 3 day HFD; D7, 7 day HFD; W10, 10 week HFD. n=8 per group. (E) Relative oxygen consumption rate in the presence or absence of oligomycin in primary adipocytes from NCD/lean or HFD/obese mice. n=4 per group. (F) FFA level in eWAT of NCD or HFD mice. n=5-7 per group. *, P<0.05; **, P<0.01. (G) Oxygen consumption rate in 3T3-L1 adipocytes before and after BSA, KLA (Kdo2), or each of FFAs indicated (50 μM) in the presence or absence of an inhibitor of ANT, carboxyatractyloside (CAtr). *, P<0.05; **, P<0.01. (H) Palmitate-induced respiration is not inhibited by Etomoxir (Etom) or oligomycin (Oligo). (I) ATP production as calculated from Seahorse data as described previously (Sridharan et al., 2008). (J) Palmitate induces hypoxia (pimonidazole adducts) in cultured 3T3-L1 adipocytes. Detailed protocols are described in the Extended Experimental Procedures. Blue, DAPI; Green, hypoxia probe. (K) Western blot analysis of HIF-1α protein in 3T3-L1 adipocytes with or without palmitate (100 μM) and/or CAtr (1 μM). (L) Acute injection of CAtr reduces adipocyte hypoxia in mice fed a HFD for 3 days. Green, hypoxia probe; Red, BODIPY. All data are mean+/-SEM. See also Figure S1.

Figure 2

Effects of ANT inhibition in obesity. (A) Expression of different isoforms of ANT and UCP genes in isolated primary eWAT adipocytes of chow or HFD mice at 1 day (left) or 15 weeks (right). *, P<0.05; **, P<0.01. (B) Effect of ANT1 or ANT2 knock down in 3T3-L1 adipocytes. 3T3-L1 adipocytes were transfected with scrambled siRNA (Sc) or siRNAs against ANT1(A1) or ANT2 (A2), or both (A1+2), and oxygen consumption rate was measured using Clark type electrode. *, P<0.05. (C) In vivo ANT2 knockdown decreases adipose tissue hypoxia. Green staining (middle panel) represents pimonidazole adduct-positive area. *, P<0.05; **, P<0.01. (D-I) Effect of chronic CAtr treatment on hypoxia (D), body weight (E), blood oxygenation (F), eWAT and liver weight (G), glucose tolerance (H) and insulin tolerance (I) at day 5-9. The graph shows body weight at day 5. *, P<0.05; **, P<0.01; § , P<0.05 0.4 mg/kg CAtr vs control; ; #, P<0.05 1.0 mg/kg CAtr vs control. See also Figure S2.

Figure 3

HIF-1α KO mice exhibit improved glucose tolerance and insulin sensitivity on HFD. (A) GTT with mice fed NCD. (B) GTT with mice fed HFD for 10 weeks. (C) Plasma insulin level during GTT in panel B. (D) ITT in mice fed HFD for 11 weeks. (E) Hyperinsulinemic euglycemic clamp performed on mice fed a HFD for 15 weeks. (F) Glucose uptake assays using primary adipocytes from WT or HAKO mice fed HFD for 15 weeks. All data represent mean+/-SEM. See also Figure S3.

Figure 4

HAKO mice exhibit reduced macrophage infiltration and inflammatory gene expression with reduced chemokine and LTB4 production. (A) IHC analysis of ATMs from WT or HAKO mice fed a HFD for 15 weeks. Red, neutral lipids (BODIPY); green, macrophage (CD11b). n=5 or 6 per group. (B) Flow cytometry analysis of eWAT SVCs from 15 week HFD mice. n=4 (NCD) or 7 (HFD) per group. (C) mRNA levels of inflammatory genes in adipose tissue were measured by RT-PCR analysis. n=6. (D) mRNA levels of class II MHC genes in primary adipocytes were measured by RT-PCR. n=7 per group. (E) Plasma levels of adipokines (n=7) as measured by ELISA or western blots. (F) Chemoattractive activity of ACM harvested from HAKO adipocytes is reduced compared with WT adipocytes (15 week HFD). n=4 per group. (G) FLAP mRNA level in adipocyte and SVC fractions of eWAT from mice fed NCD or HFD (15 weeks). n=7 or 9 per group. (H) LTB4 release from isolated adipocytes of NCD WT, HFD WT or HFD HAKO mice. n=3 or 7 per group. (I) In vivo monocyte tracking (n=4). All data represent mean+/-SEM. See also Figure S4.

Figure 5

HAKO mice fed a HFD show reduced NO production and Akt nitrosylation, and lower lactate production in adipocytes. (A) mRNA levels of iNOS and arginase 1 in adipocyte and eWAT SVCs from WT and HAKO mice. n=4 or 7 per group. (B) Nitrite level in eWAT of WT and HAKO mice. n=4 or 5 per group. (C) HFD-induced nitrosylation of Akt and total protein in adipose tissue is reduced by HAKO. (D) Akt phosphorylation in eWAT from WT or HAKO mice before and after insulin injection. (E) PDK expression. n=4 or 7 per group. (F) Fasting hepatic glucose production (HGP) is lower in HAKO mice on HFD. (G) Adipose tissue lactate content is decreased in HAKO mice. n=5 per group. All data represent mean+/-SEM.

Figure 6

H2AKO impairs glucose and insulin tolerance with increased adipose tissue inflammation. (A) mRNA levels of HIF-1α and HIF-2α in adipose tissue of HFD WT or H2AKO mice. n=6. (B) Body weight of WT and H2AKO mice fed NCD (n=10) or HFD (n=36). (C) eWAT mass on NCD and HFD (15 weeks; n=5 per group). (D) GTT in NCD mice. (E) GTT performed in 10 week HFD mice. (F) Plasma insulin level during GTT in panel E. (G) ITT test with mice on 11 week HFD. (H) Flow cytometry analysis of eWAT SVCs. n=4 or 9 per group. (I) RT-PCR analysis of inflammatory gene expression in primary adipocytes from HFD WT or HFD H2AKO mice. n=6 per group. All data represent mean+/-SEM. See also Figure S5.

Figure 7

DHAKO mice exhibit similar metabolic changes to HAKO mice. (A) mRNA levels of HIF-1α and HIF-2α in adipose tissue of HFD WT or DHAKO mice. (B) B.W. of WT and DHAKO mice on HFD. n=18 or 20 per group. (C) eWAT mass on NCD and HFD (15 weeks). n=7 or 10 per group. *, P<0.05. (D) Trichrome (blue) and H&E stainings of eWAT from mice fed a HFD for 15 weeks. (E-F) GTT in mice fed a NCD (E) or HFD for 10 weeks (F). *, P<0.05. (G) Plasma insulin level during GTT in panel F. (H) ITT in mice fed a HFD for 11 weeks. *, P<0.05; ***, P<0.001. (I) Hyperinsulinemic euglycemic clamp with mice fed a HFD for 15 weeks. *, P<0.05. (J) Flow cytometry analysis of eWAT SVCs. n=7 or 10 per group. *, P<0.05. (K) Nitrite level in eWAT of WT and DHAKO mice on chow or HFD. n=4 or 5 per group. *, P<0.05. All data represent mean+/-SEM.