High-fat diet-induced upregulation of exosomal phosphatidylcholine contributes to insulin resistance

Abstract

High-fat diet (HFD) decreases insulin sensitivity. How high-fat diet causes insulin resistance is largely unknown. Here, we show that lean mice become insulin resistant after being administered exosomes isolated from the feces of obese mice fed a HFD or from patients with type II diabetes. HFD altered the lipid composition of exosomes from predominantly phosphatidylethanolamine (PE) in exosomes from lean animals (L-Exo) to phosphatidylcholine (PC) in exosomes from obese animals (H-Exo). Mechanistically, we show that intestinal H-Exo is taken up by macrophages and hepatocytes, leading to inhibition of the insulin signaling pathway. Moreover, exosome-derived PC binds to and activates AhR, leading to inhibition of the expression of genes essential for activation of the insulin signaling pathway, including IRS-2, and its downstream genes PI3K and Akt. Together, our results reveal HFD-induced exosomes as potential contributors to the development of insulin resistance. Intestinal exosomes thus have potential as broad therapeutic targets.

Fig. 1. High-fat diet (HFD) alters the lipid composition of intestinal epithelial cell-released exosomes.

a, b Lipids were extracted from fecal exosomes (CD63⁺A33⁺) released from intestinal epithelial cells of mice fed either a regular chow diet (RCD) or a high-fat diet (HFD) for 6 months (n = 10/group). Pie charts representing the percentage of each lipid, as determined using triple-quadrupole mass spectrometry (a). PE and PC concentration (nmol) shown at the right panel (b). Filled triangle—L-Exo and filled rectangle—H-Exo. c, d Triple-quadrupole mass spectrometry of exosomal lipids harvested from mice feces after 12 months of their respective diets (n = 10/group). Pie charts representing the percentage of each lipid (c). PE and PC concentration (nmols) shown at the right panel (d). e, f Pie charts (e) representing the percentage of lipid subtypes derived from the fecal exosomes (CD63⁺A33⁺) of healthy (Healthy-Exo, n = 5) or patients with type 2 diabetes (T2D-Exo, n = 7). PE and PC concentration (nmols) determined by triple-quadrupole MS shown at the right panel (f). Hollow triangle—Healthy-Exo and hollow rectangle—T2D-Exo. Data are represented as mean ± S.D. Student’s t test (two-tailed). Source data are provided as a Source Data file.

Fig. 2. CD63⁺A33⁺ exosomes derived from HFD mice induce insulin resistance in mice fed an HFD.

a Glucose (GTT) and insulin tolerance tests (ITT) for C57BL/6 mice (n = 10) after receiving adoptive transfer of mouse CD63⁺A33⁺ fecal exosomes for 14 days while mice were fed HFD. Filled circle—PBS, filled triangle—L-Exo, and filled rectangle—H-Exo. b Glucose infusion rate (GIR) during the clamp assay (n = 4/group). c Blood glucose levels during the clamp assay (n = 4/group). d Plasma insulin levels at basal (−10 min) and during the clamp assay (120 min). e, f Hepatic glucose production and % suppression of hepatic glucose production (n = 4/group). g Whole-body glycolysis rate determined by the clamp assay (n = 4/group). h Glucose uptake by brown adipose (BAT), white adipose tissue (WAT), and muscle tissue after the clamp assay (n = 4/group). i Plasma free-fatty acids at basal and during the clamp assay (n = 4/group). j–m GTT (upper panels) and ITT (lower panels) for either C57BL/6 or C57BL/6 germ-free mice after receiving adoptive transfer of human exosomes (j), mouse and human exosomes (k), mouse exosomes with or without antibiotic treatment (l), and nanoparticles from L-Exo or H-Exo with added (PC+) or depleted PC (PC−) for 14 days while mice were fed HFD (n = 5/group). Data are represented as the mean ± SD. One-way ANOVA with a Tukey post hoc test. * < 0.05; ** < 0.01, and *** < 0.001. Statistical significances were shown between the PBS and H-Exo group or as otherwise indicated. Source data are provided as a Source Data file.

Fig. 3. Uptake of CD63⁺A33⁺ exosomes by liver cells depends on exosomal lipid composition.

a Confocal images of GFP-positive exosomes detected in mouse liver after injection of GFP-MC38 epithelial cells into the colon. DAPI was used to stain the nucleus. Each data point was measured in triplicate. b Flow cytometry analysis of PKH-26-labeled exosome uptake by hepatocytes (albumin⁺) and Kupffer cells (F4/80⁺). Filled triangle—L-Exo and filled rectangle—H-Exo. c PKH26-labeled exosomes visualized by confocal microscopy in hepatocytes/albumin⁺/green (yellow arrow) and Kupffer cells/F4/80/purple (white arrow) (Ieft). The percentages of total exosome uptake per cell type are summarized on the right. Scale bar as indicated. Each data point was measured in triplicate. d Flow cytometry analysis of Kupffer cells cultured with PKH26-labeled H-Exo for 16 h in the presence or absence of endocytosis inhibitors (indicated). Percentage of PKH26⁺ Kupffer cells summarized below. e, f PKH-26-labeled nanoparticles were cultured with primary hepatocytes (upper panels, e) and Kupffer cells (lower panels, f). Cells were analyzed by flow cytometry and the percentage of PKH-26-positive cells were assessed after treatment with each nanoparticle summarized at the right. Circle—nanoparticles made up of total lipids of L-Exo; rectangle—H-Exo lipids; upward triangle—H-Exo lipid-depleted PC and downward triangle—L-Exo PC supplemented. Source data are provided as a Source Data file. Data are represented as the mean ± SD. Student’s t test (two-tailed) or one-way ANOVA with a Tukey post hoc test. ** < 0.01; *** < 0.001, and **** < 0.0001.

Fig. 4. Crosstalk between hepatocytes and macrophages contributes to insulin resistance.

a Plasma and WAT cytokine array of mice that received CD63⁺A33⁺ exosomes (L-Exo or H-Exo) for 14 days. b Fold change in H-Exo vs. L-Exo-induced plasma cytokine expression for all cytokines showing greater than two-fold change. Red bars show cytokines/factors known to be involved in insulin resistance. c TNF-α (left) and IL-6 (right) upregulation following treatment with H-Exo were confirmed by ELISA in plasma. Filled circle—PBS, filled triangle—L-Exo, and filled rectangle—H-Exo. d ITT performed on C57BL/6 mice that received exosomes via adoptive transfer for 14 days followed with or without macrophage depletion. Filled rectangle—macrophage-depleted mice treated with H-Exo; filled diamond—mice without macrophage depletion treated with PBS and circle—mice without macrophage depletion treated with H-Exo. e Glucose uptake assay performed on hepatocytes cultured with different concentrations of H-Exo (as indicated in the figure). f Glucose uptake assay performed on mouse hepatocytes supplemented with supernatant derived from macrophages cultured with nanoparticles derived from H-Exo total lipids (H-Exo Nano) and PC (34:2). g Supernatants from H-Exo-treated macrophages (monocytes+ 5 × 10⁶) were preneutralized with anti-TNF-α and/or anti-IL-6 antibodies. Glucose uptake by hepatocytes cultured in the presence of preneutralized supernatant was estimated. Data are represented as the mean ± SD. One-way ANOVA with a Tukey post hoc test. * < 0.05; **** < 0.0001. Source data are provided as a Source Data file.

Fig. 5. HFD-induced CD63⁺A33⁺ exosomal lipids contribute to insulin resistance in an AhR-dependent manner.

a Representative gene expression heat map for the Affymetrix array of liver tissue from mice orally administered exosomes for 14 days. Induction of AhR expression highlighted in the red box. Elevated AhR expression was confirmed by qPCR (bar graph, right, n = 3). Filled triangle—L-Exo and filled rectangle—H-Exo. b Total AhR protein expression was confirmed by western blots in liver tissue. c Phosphorylated AhR (pAhR) protein expression in hepatocytes (FL83B cells) cultured with L-Exo, H-Exo, L-Exo^lipids, or H-Exo^lipids. Densitometry analysis summarized below. Filled circle—PBS, filled triangle—L-Exo, and filled rectangle—H-Exo. d FL83B cells were cultured with different concentrations (as indicated) of PC (34:2) for 16 h, and the resulting effects on AhR expression were determined by western blots. Ratio to β-actin shown in the middle as numbers. e Glucose uptake assay for FL83B cells cultured with varying concentrations of PC (34:2). f, g SPR sensogram showing the interaction of AhR recombinant protein with nanoparticles derived from total lipids of H-Exo (f) and PC (34:2) (g). h PC direct binding to AhR protein. i SPR was performed with AhR protein-coated onto an NTA chip and H-Exo^{lipid PC−} and PC (34:2)^lipid run over as the mobile phase. j AhR expression in the cytoplasm vs. nucleus of mouse hepatocytes cultured with L-Exo or H-Exo. Densitometry analysis of cytoplasmic (left) vs. nuclear (right) AhR protein expression following treatment with L-Exo or H-Exo summarized below. k GTT and ITT performed on AhR null (AhR^−/−) HFD-fed mice treated with CD63⁺A33⁺ exosomes (L-Exo or H-Exo) for 14 days. l, m Glucose uptake assay performed on AhR-knockout (AhRKO, l) and wild-type (m) FL83B cells. n GTT and ITT of AhR^−/− mice with re-expression of AhR in hepatocytes by adenovirus (5 × 10⁹ pfu) injected via tail vein. Mice were orally gavaged with H-Exo for 14 days while being fed the HFD. Student’s t test, one-tailed. Data represent the mean ± SD. Student t test (two-tailed) or one-way ANOVA with a Tukey post hoc test. * < 0.05 and ** < 0.01. Source data are provided as a Source Data file. Each data point was measured in triplicate.

Fig. 6. H-Exo impact on insulin signaling in liver, adipose, and muscle tissues.

a Gene expression heat map from the insulin-signaling PCR array (Qiagen) performed for liver, adipose, and muscle tissues derived from mice receiving 14 days of oral administration of CD63⁺A33⁺ exosomes. b Upregulated or downregulated genes in liver tissues derived from mice treated with PBS, L-Exo. and H-Exo (n = 3). Filled circle—PBS, filled triangle—L-Exo, and filled rectangle—H-Exo. c, d Western blots of pIRS-2 in liver, adipose, and muscle tissue extracts from C57BL/6 (c) and AhR^−/− mice (d). Ratio to β-actin shown in the middle. e, f Western blots of PI3K, pAKT and AKT in liver tissue extracts from C57BL/6 and AhR^−/− mice. Ratio to β-actin shown in the middle as numbers. g AhR was re-expressed in the liver via tail injection of Ad-AhR (adenovirus). Mice were treated with L-Exo or H-Exo for 14 days. pIRS-2 levels measured in liver lysates by western blot. Ratio to β-actin shown in the middle as numbers. h, i Glucose uptake measured in tissue extract-treated hepatocytes, adipocytes, and myocytes transfected with either control (h) or IRS2 (i) plasmid vectors (n = 3). j Scanning images of liver, WAT, and muscle for DIR-labeled exosomes. k Tissue-specific and H-Exo-dependent alteration in the expression of genes regulating insulin signaling. Expressions are shown in comparison with PBS-treated mice (n = 3). Circle—liver, rectangle—WAT, and downward triangle—muscle. Data are presented as the mean ± SD. One-way ANOVA with a Tukey post hoc or two-way ANOVA with a Tukey post hoc test. Source data are provided as a Source Data file.

Fig. 7. H-Exo induced liver steatosis and dyslipidemia in C57BL/6 and C57BL/6 germ-free mice, but not in AhR^−/− mice.

a H&E staining of liver tissue sections of mice receiving PBS, L-Exo, or H-Exo for 14 days. Blue arrows indicate steatosis. The scale bar is 20 μm. Data were repeated at least three times. b Plasma cholesterol and triglyceride levels after exosome treatment for 14 days (n = 5). Filled circle—PBS, filled triangle—L-Exo, and filled rectangle—H-Exo. c Plasma ALT and AST levels (n = 5). d Model: The intestinal epithelial cells of HFD-fed mice-release CD63⁺A33⁺ exosomes that induce insulin resistance and glucose intolerance via an AhR-mediated pathway. Data are presented as the mean ± SD. One-way ANOVA with a Tukey post hoc test. Source data are provided as a Source Data file.