Fatty acid mobilization from adipose tissue is mediated by CD36 posttranslational modifications and intracellular trafficking

Abstract

The mechanism controlling long-chain fatty acid (LCFA) mobilization from adipose tissue is not well understood. Here, we investigated how the LCFA transporter CD36 regulates this process. By using tissue-specific KO mouse models, we showed that CD36 in adipocytes and endothelial cells mediated both LCFA deposition into and release from adipose tissue. We demonstrated the role of adipocytic and endothelial CD36 in promoting tumor growth and chemoresistance conferred by adipose tissue-derived LCFAs. We showed that dynamic cysteine S-acylation of CD36 in adipocytes, endothelial cells, and cancer cells mediated intercellular LCFA transport. We demonstrated that lipolysis induction in adipocytes triggered CD36 deacylation and deglycosylation, as well as its dissociation from interacting proteins, prohibitin-1 (PHB) and annexin 2 (ANX2). Our data indicate that lipolysis triggers caveolar endocytosis and translocation of CD36 from the cell membrane to lipid droplets. This study suggests a mechanism for both outside-in and inside-out cellular LCFA transport regulated by CD36 S-acylation and its interactions with PHB and ANX2.

Keywords: Adipose tissue; Cancer; Cell Biology; Metabolism; Obesity.

Figure 1. CD36 in adipocytes and the endothelium mediates LCFA transport in adipose tissue.

(A) Steady-state plasma concentration of free fatty acids (FFAs), higher in CD36 EC-KO and Ad-KO mice. n = 10 mice. (B) Body composition measured with EchoMRI and body weight (BW) revealed reduced adiposity in CD36-KO mice (n = 10 mice, 1-way ANOVA). (C) Reduced BODIPY-C₁₆ uptake by adipocytes in CD36 Ad-KO and EC-KO mice. Visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), and brown adipose tissue (BAT) were recovered 180 minutes after i.v. BODIPY-C₁₆ injection and green fluorescence was imaged in cell suspension upon tissue digestion with collagenase. Scale bar: 50 μm; blue: DNA. (D) Western blotting confirmed KO of CD36 in 3T3-L1 cells. Images: lipid droplet formation in control and CD36-KO adipocytes differentiated for 5 days. Western blotting on proteins extracted from control and CD36-KO bEND.3 cells demonstrated loss of CD36, but not of PHB and ANX2, immunoblotted for as loading controls. Arrow: glycosylated CD36. Arrowhead: nonglycosylated CD36. Graphs: Western quantification in AU. Scale bar: 50 μm for all panels. (E) QBT assay demonstrating that LCFA uptake by CD36-null 3T3-L1 adipocytes and bEND.3 cells was inefficient compared with nondifferentiated (ND) cells. n = 5 wells. (F) Concentration of FFA in culture medium 3 hours after lipolysis induction in WT and CD36-null 3T3-L1 adipocytes (n = 10 wells, Student’s t test). (G) Relative increase in plasma concentration of FFA 15 minutes after isoproterenol injection observed in WT but not CD36-KO mice (n = 10 mice, 1-way ANOVA). In all panels, data are shown as mean ± SEM; *P < 0.01.

Figure 2. CD36 in cancer cells is not required but can promote LCFA transport.

(A) Immunofluorescence (IF) on paraffin sections of E0771 tumor grafts showing mainly intracellular CD36 and PHB expression in cancer cells; in intratumoral adipocytes (a) colocalization is at the surface (yellow arrow). Blue: nuclei. (B) IF showing that CD36 and PHB (arrows) are intracellular in cultured E0771 cells. Blue: nuclei. (C) Western blotting on extracts from murine and human cell lines demonstrated low expression of CD36 in cancer cells. Arrow: glycosylated CD36. Arrowhead: nonglycosylated CD36. NS, nonspecific band. (D) Western blotting demonstrated expression of CD36 in MCF7 cancer cells comparable to that in 3T3-L1 adipocytes. Arrow: glycosylated CD36. Arrowhead: nonglycosylated CD36. ANX2 immunoblotting: loading control. (E) Western blotting confirming CD36 KO by CRISPR/Cas9 in MCF7 cells transduced with sgCD36. NS, nonspecific band. (F) ³H CPM in indicated cell cultures after 30 minutes exposure to 75 μM ³H-palmitate demonstrated that ³H-palmitate uptake was inhibited by SSO and CD36 KO. n = 5 independent wells. Data are shown as mean ± SEM; *P < 0.01, (1-way ANOVA). (G) 4T1.2 cells preinduced to undergo lipogenesis were untreated or treated with BODIPY-FL-C₁₆ for 10 minutes and imaged for LCFA uptake (arrow). (H) Intercellular fatty acid transfer from 3T3-L1 adipocytes (not plotted) preloaded with BODIPY-FL-C₁₆ (green) to cocultured RFP⁺ 4T1 cells detected by flow cytometry with 530 nm (BODIPY) and 610 nm (RFP) lasers. The histogram shows the difference in BODIPY-FL-C₁₆ uptake for double-positive (BODIPY-FL-C₁₆⁺/RFP⁺) 4T1 cells cocultured with WT versus CD36-KO adipocytes. Scale bar: 50 μm.

Figure 3. CD36 in adipocytes and ECs mediates LCFA transfer to tumors.

(A) Experiment schematic: i.v.-administered ³H-palmitate was deposited into adipocytes and cleared from circulation, after which tumor cells were grafted and ³H content in tumors was measured by liquid scintillation counting. (B) ³H counts per minute (CPM) in 100 mg of E0771 tumor tissue recovered from mice treated as described in A, indicating reduced ³H-palmitate transfer to cancer cells in CD36 Ad-KO (left graph) and EC-KO (right graph) mice compared with CD36⁺ (WT) littermates. n = 10 tissue probes (Student’s t test). (C) Reduced E0771 graft growth in CD36 Ad-KO females compared with CD36⁺ (WT) littermates. n = 10 (Student’s t test). Images of representative tumors are shown in the right. (D) E0771 graft growth, reduced by cisplatin treatment, was further reduced in CD36 Ad-KO females compared with CD36⁺ (WT) littermates. n = 5 (1-way ANOVA). Images: representative tumors. (E) Images of RM1 tumors 3 weeks after s.c. grafting into male mice. Graphs: quantification of final tumor volume, being lower in CD36 EC-KO compared with CD36⁺ (WT) littermates. n = 8–10 (Student’s t test). (F) IF analysis showing that tumors in CD36 EC-KO mice lacked intratumoral adipocytes (a) positive for perilipin-1 (PLN1), lacked HADHA expression, and had increased GLUT1 expression in cancer cells. Data are shown as mean ± SEM; *P < 0.05. Scale bar: 50 μm.

Figure 4. Acylation/deacylation of CD36 mediates LCFA transport.

(A) Western blotting on extracts from 4T1.2 cells nontransduced or transduced with WT CD36 or CD36 mutant lacking S-acylated cysteines. Arrow: glycosylated CD36, also expressed in 3T3-L1 adipocytes. NS, nonspecific band. (B) 4T1.2 cells transduced with WT CD36 or CD36 mutant lacking S-acylated cysteines grown in 2D were preinduced to undergo lipogenesis as in Figure 2G and then treated with BODIPY-FL-C₁₆ for 10 minutes and imaged to visualize uptake by lipid droplets (arrows). Scale bar: 50 μm. (C) Intercellular fatty acid transfer from 3T3-L1 adipocytes (not plotted) preloaded with BODIPY-FL-C₁₆ (green) to adjacent RFP⁺ 4T1 cells detected by flow cytometry with 530 nm (BODIPY) and 610 nm (RFP) lasers. Histograms at the bottom are provided to compare double-positive (BODIPY-FL-C₁₆⁺/RFP⁺) population in 4T1 cells expressing WT versus mutant CD36. (D) 3T3-L1 adipocytes treated with 300 μM palmitic acid for 0 to 120 minutes analyzed by acyl biotin exchange (ABE) assay (Supplemental Figure 4A). Extracted proteins were alkylated with hydroxylamine (HA) where indicated, de–S-acylated, biotinylated, and removed with streptavidin beads. Remaining proteins were immunoblotted with indicated antibodies. Note progressive accumulation of nonacylated glycosylated (arrow) and nonglycosylated (arrowhead) CD36 upon LCFA treatment. PHB immunoblot indicates constant PHB acylation and equal loading. Quantification is on the right. (E) bEND.3 endothelial cells treated with 300 μM palmitic acid for 0 to 120 minutes analyzed by ABE assay. Extracted proteins were alkylated (HA), de–S-acylated, biotinylated, and removed with streptavidin beads. The remaining proteins were immunoblotted with indicated antibodies. Note accumulation of nonacylated glycosylated (arrow) and nonglycosylated (arrowhead) CD36 upon LCFA treatment. ANX2 immunoblot indicates constant ANX2 acylation and equal loading. Quantification is on the right.

Figure 5. CD36 deacylation induced by lipolysis.

(A) Concentration of FFA in culture medium 5 hours after lipolysis induction in 3T3-L1 adipocytes untreated or treated with an inhibitor of S-acylation, ML211 (30 nM). Data are shown as mean ± SEM; *P < 0.05 (Student’s t test). (B) ABE assay (Supplemental Figure 4A) on 3T3-L1 1adipocytes untreated or induced to undergo lipolysis for indicated time (minutes). Note that IP of S-acylated CD36, specifically observed upon hydroxylamine (HA) treatment, was partly inhibited by lipolysis. Immunoblotting for calnexin and ANX2 from the same extracts indicated constant ANX2 acylation and equal loading. Quantification is on the right. (C) CD36 metabolic labeling (Supplemental Figure 4B). After incubation of live 3T3-L1 adipocytes with alkynylated LCFA analog (0.1 mM 17-ODA, 12 hours), de novo S-acylation of CD36 was detected by IP with anti-CD36 antibodies, subsequent click chemistry with IRDye800-azide probe (IRDye800-N3), and SDS-PAGE. Note that IP of S-acylated CD36, specifically observed upon HA treatment, was inhibited by lipolysis induction (IBMX/forskolin/isoproterenol). Arrow: glycosylated CD36. Arrowhead: nonglycosylated CD36. CD36-IR800 quantification is on the right. (D) Probing of the IP blot from C with PHB and ANX2 antibodies, and subsequently with CD36 antibodies, demonstrated a decrease in PHB and ANX2 association with CD36 concomitant with a decrease of CD36 acylation upon lipolysis induction.

Figure 6. Lipolysis induces CD36 trafficking to lipid droplets.

(A) Immunofluorescence analysis of 3T3L1 adipocytes with antibodies against indicated proteins and secondary antibodies conjugated with green, red, and blue fluorophores. Note that in control adipocytes, CD36 was concentrated on the cell membrane along with PHB and ANX2 (white arrows), whereas 30-minute treatment with lipolysis-inducing agents induced CD36 dissociation from PHB (red arrows) and ANX2 (green arrows), internalization (blue arrows), and localization to lipid droplets (ld). n, nuclei. (B) Immunofluorescence analysis of 3T3L1 adipocytes with antibodies against indicated proteins and secondary antibodies conjugated with green and red; nuclei (n) are stained blue. Note that in control adipocytes, CD36 was concentrated on the cell membrane along with caveolin-1, whereas 30-minute treatment with lipolysis-inducing agents induced translocation of CD36 and caveolin-1 to ld.

Figure 7. A model of CD36-mediated outside-in and inside-out LCFA transport.

Postprandially, insulin signaling activates extracellular LCFA uptake mediated by acylated CD36 at the cell surface in complex with PHB and ANX2. Deacylation of CD36 enables caveolar endocytosis and lipid droplet trafficking of LCFA-bound CD36. In lipolytic conditions, CD36 deacylation, caveolar endocytosis, and lipid droplet trafficking enable CD36 loading with LCFA released from lipid droplets and their plasma membrane trafficking and release via exocytosis.