Nutrient-sensing nuclear receptors coordinate autophagy

Abstract

Autophagy is an evolutionarily conserved catabolic process that recycles nutrients upon starvation and maintains cellular energy homeostasis. Its acute regulation by nutrient-sensing signalling pathways is well described, but its longer-term transcriptional regulation is not. The nuclear receptors peroxisome proliferator-activated receptor-α (PPARα) and farnesoid X receptor (FXR) are activated in the fasted and fed liver, respectively. Here we show that both PPARα and FXR regulate hepatic autophagy in mice. Pharmacological activation of PPARα reverses the normal suppression of autophagy in the fed state, inducing autophagic lipid degradation, or lipophagy. This response is lost in PPARα knockout (Ppara(-/-), also known as Nr1c1(-/-)) mice, which are partially defective in the induction of autophagy by fasting. Pharmacological activation of the bile acid receptor FXR strongly suppresses the induction of autophagy in the fasting state, and this response is absent in FXR knockout (Fxr(-/-), also known as Nr1h4(-/-)) mice, which show a partial defect in suppression of hepatic autophagy in the fed state. PPARα and FXR compete for binding to shared sites in autophagic gene promoters, with opposite transcriptional outputs. These results reveal complementary, interlocking mechanisms for regulation of autophagy by nutrient status.

Extended Data Figure 1. PPARα or FXR agonist affects autophagic flux in murine hepatocytes

a, Autophagic flux was assessed by LC3-Immunoblot analysis in AML12 cells treated with indicated dose of Wy or co-treated with Wy and bafilomycin A1 (BafA1). b and e, Biochemical determination of autophagy (p62/SQSTM1 immunoblot) in AML12 cells treated with indicated doses of Wy-14,643 (Wy) or GW4064 for 24 h. GW4064-treated cells were starved for 2 h. c, Primary hepatocytes prepared from GFP-LC3^Tg/+ mice were treated with indicated doses of Torin1 or GW7647. GFP-LC3 cleavage was assessed by GFP-Immunoblot analysis. d, LC3-Immunoblot in AML12 cells treated with indicated doses of GW4064 or co-treated with GW4064 and Torin1. All drug treatments were done in complete media for 24 h otherwise indicated in each panel. β-actin is a loading control. f, Quantification of autophagic flux shown in Fig. 1a. Numbers of autolysosomes (ALs) induced by 2 h starvation + Veh were set as 100%. Numbers of autolysosomes = RFP positive vesicles – GFP positive vesicles. 30 cells were counted per condition (**P < 0.01 vs No starvation + Veh). Data represent mean±s.e.m. Statistics by two-tailed t-test.

Extended Data Figure 2. Nutrient availability regulates the expression of autophagy-related genes

a and b, Biochemical determination of autophagy (LC3 immunoblot, LC3-I/II: non-lipidated and PE conjugated forms of MAP1LC3, respectively) or mTORC1 activity (p-S6^S240/244 immunoblot) in AML12 cells treated with indicated time points of 100 µM of Wy-14,643 (Wy) or 10 µM of GW4064. GW4064-treated cells were starved in HBSS medium for 1 h. β-actin is a loading control. c, Bile acids suppress autophagosome formation. AML12 cells were treated with indicated doses of each bile acid for 24 h, followed by 1 h starvation in HBSS medium. d, Immunoblot analysis of LC3-I/II and β-actin in AML12 cells treated with indicated dose of GW7647 or GW4064 for 24 h. GW4064-treated cells were starved in HBSS medium for 1 h. e and f, Fed ad libitum, 24 h fasted, or 24 h refed after 24 h fasting WT mice were sacrificed to collect livers at 6:00 PM. e, Immunoblot analysis of LC3-I/II, β-actin, phospho-4E–BP1 (p-4E–BP1^T37/46), and total 4E–BP1 (4E–BP1) in liver samples. 50 µg of proteins from liver homogenates were separated by SDS-PAGE and probed with the indicated antibodies. β-actin is a loading control. A pooled sample was loaded onto the gel in duplicates (n=5 per group). f, Hepatic expression levels of autophagy-related genes, PPARα target gene, and FXR target genes affected by nutrient availability. (n=5 per group, *P < 0.05, **P < 0.01 vs Fed WT mice). g, Hepatic expression levels of PPARα or FXR target gene (Acox1 and Cyp7a1, respectively) were determined by qPCR analysis. Fed or fasted WT mice were orally gavaged with vehicle (Veh), GW7647, or GW4064 twice a day. (n=5 per group, *P < 0.05, **P < 0.01 vs Fed WT mice). Panels in f and g, data represent mean±s.e.m. Statistics by two-tailed t-test.

Extended Data Figure 3. Pharmacologic activation of PPARα or FXR in fed or fasted mouse liver

a and c, Hepatic expression levels of autophagy-related genes (LC3a, LC3b and Atg12), PPARα target gene (Acox1), and FXR target gene (SHP) were determined by qPCR analysis. Fed or fasted WT PPARα^−/−, or FXR^−/− mice were orally treated with Veh, GW7647, or GW4064 twice a day. (n=5 per group, *P < 0.05, **P < 0.01 vs Fed WT mice treated with Veh). b and d, Immunoblot analysis of LC3-I/II, β-actin, phospho-S6 (p-S6^S240/244), and total S6 (S6) in liver samples. Fed or fasted WT mice were orally gavaged with Veh, GW7647, or GW4064 twice a day. A pooled sample was loaded onto the gel in duplicates. (n=5 per group). β-actin is a loading control. e, Representative confocal images (out of 9 tissue section per condition) of GFP-LC3 puncta (GFP color: autophagosomes) and DAPI (blue color: DNA) staining in livers. Fed or fasted bigenic PPARα^−/−; GFP-LC3^Tg/+, or FXR^−/−; GFP-LC3^Tg/+ mice were orally gavaged with Veh, GW7647, or GW4064 twice a day. Liver samples were fixed and cryosections were analyzed by confocal microscopy. Scale bar, 50 µm. f, Co-localization of BODIPY 493/503 (green) with LC3 (red) in AML12 cells treated with Veh or 1 µM of GW4064 for 24 h and simultaneously cultured with or without 125 µM oleate in complete medium. GW4064-treated cells were starved for 2 h. DNA was stained with DAPI (blue). Scale bar, 20 µm. Quantification of Lipophagic vacuoles shown in Fig. 3b and Extended Data Fig. 3f. 30 cells were counted per condition (**P < 0.01). g, Measuring β-hydroxybutyrate. AML12 cells were transiently transfected with siControl, siAtg5, or siAtg7 for 24 h followed by indicated drug treatments for 48 h with or without 250 µM oleate (Veh: 0.1% DMSO, Wy; 10 µM Wy-14,643). Released β-hydroxybutyrate in the medium was determined. (**P < 0.01 vs siControl treated with Veh; ^#P < 0.01 vs siControl treated with Wy; ^##P < 0.01 vs siControl treated with Oleate + Wy). h, Serum β-hydroxybutyrate were normalized with liver weights. Fed or 24 h fasted control littermates (Atg7^F/F) and hepatocyte-specific Atg7^F/F null (Alb-Cre/+; Atg7^F/F) mice were treated with Veh or GW7647 twice a day. (n=4 per group, *P < 0.05, **P < 0.01 vs Fed Atg7^F/F mice treated with Veh; ^#P < 0.01 vs Fasted Atg7^F/F mice treated with Veh; ^##P < 0.01 vs Fasted Atg7^F/F mice treated with GW7647). Panels in a, c, g, f and h, data represent mean±s.e.m. Statistics by two-tailed t-test.

Extended Data Figure 4. PPARα activation or loss of FXR induces autophagy in liver

Magnification of representative TEM images (our of 30 cells per group) of livers. a–c, Fed or fasted WT PPARα^−/− or FXR^−/− mice were orally gavaged with Veh, GW7647, or GW4064 twice a day. Lipophagy (yellow arrowhead), autophagosome (blue arrowhead), autolysosome (red arrowhead), microautophagy (black arrowhead), and multivesicular bodies (MVBs, purple arrowhead). Scale bar, 0.5 µm.

Extended Data Figure 5. Expression profiles of autophagy-related genes by PPARα or FXR activation in liver

a–c, Hepatic expression levels of autophagy-related genes were determined by qPCR analysis in WT (a, b) or PPARα^−/− or FXR^−/− (c) mice. 11 genes shown in panel a were induced by PPARα activation, but not affected by FXR activation. 4 genes shown in panel b were suppressed by FXR activation, but not affected by PPARα activation. (panel a and b, n=5 per group, *P < 0.05, **P < 0.01 vs Fed WT mice treated with Veh; ^#P < 0.05, ^##P < 0.01 Fasted WT mice treated with Veh). Altered expression levels of 13 genes shown in Fig. 4a were lost in PPARα^−/− or FXR^−/− mice in panel c. (panel c, n=5 per group, *P < 0.05, **P < 0.01 vs Fed PPARα^−/− or FXR^−/− mice treated with Veh). Fed or fasted WT PPARα^−/− or FXR^−/− mice were orally gavaged with Veh, GW7647 or GW4064 twice a day. Panels in a–c, data represent mean±s.e.m. Statistics by two-tailed t-test.

Extended Data Figure 6. Cistromic analysis of PPARα and FXR in mouse liver

a De novo motif analysis of PPARα bound genomic regions. Top PPARα peak regions (+/− 150 bp from peak summits, ranked by enrichment fold) were subjected to de novo motif discovery by MEME. The best motif discovered by MEME (top, E value=4.7e-227) highly resembles the PPARG::RXRA motif from JASPAR(bottom, ID: MA0065.2) as a direct repeat 1 motif (DR1). b, Venn diagram depicting increasing PPARα cistrome upon PPARα agonism in vivo. PPARα highly confident (HC) binding peaks: peaks of WT mice treated with Veh or GW7647 subtracted from peaks of PPARα^−/− mice treated with Veh or GW7647, respectively. c, Venn diagram showing overlapping binding peaks between PPARα ChIP-seq and FXR ChIP-seq from WT mice treated with its synthetic agonist GW7647 or GW4064. d, Autophagy-related genes of PPARα and FXR cistrome. Within 20 kb from transcription start site (TSS), PPARα ChIP-seq showed that 7738 genes of total 28661 genes (mm9) have HC peaks in WT mice treated with GW7647 (FDR <0.0001, enrichment over PPARα^−/− >10), and that 124 genes out of 230 autophagy-related genes (HADb: Human Autophagy Database, autophagy.lu/) have at least one PPARα peak. FXR ChIP-seq showed that 3835 genes have peaks in WT mice treated with GW4064, and 61 genes out of 231 autophagy-related genes have at least one FXR peak. e, PPARα ChIP-qPCR for known PPARα target genes. (n=4 per group, *P < 0.05, **P < 0.01 vs Fed WT mice treated with Veh). Data represent mean±s.e.m. Statistics by two-tailed t-test.

Extended Data Figure 7. PPARα ChIP-seq profiles at loci of autophagy-related genes

Fed WT or PPARα^−/− mice were orally gavaged with Veh or GW7647 twice a day. Mouse livers were taken out 6 h after the last injection of drugs to perform PPARα ChIP-seq and ChIP-qPCR. a, Representative ChIP-seq reads for PPARα aligned to the autophagy-related genes (LC3a, LC3b, Gabarapl1, Bnip3, Atg12, Pex14, Sesn2, Atg7, and Prkaa2). b, PPARα ChIP-qPCR for autophagy-related genes shown in panel a. (n=4 per group, *P < 0.05, **P < 0.01 vs Fed WT mice treated with Veh). Data represent mean±s.e.m. Statistics by two-tailed t-test.

Extended Data Figure 8. PPARα/FXR genomic competition for DR1 in Acox1 gene and autophagy-related genes

a, Representative ChIP-seq reads for FXR and PPARα aligned to Acox1 and LC3b genes. The peaks in the box contain DR1 motif. Fed WT mice were orally gavaged with Veh or GW764 twice a day. (n=4 per group). b, PPARα or FXR ChIP-qPCR in livers. Fed or fasted WT mice were orally gavaged with Veh, GW7647, or GW4064 twice a day. (n=3 per group, *P < 0.05, **P < 0.01 vs Fed WT mice treated with Veh; ^##P < 0.01 vs Fasted WT mice treated with Veh). c, Cell-based luciferase reporter assays. AML12 cells were transiently transfected with 3 × PPRE luciferase reporter construct (3 × PPRE-Luc) and CMX-β-gal in a combination of expression plasmids of PPARα, FXR or both, followed by drug treatment for 20 h (Veh, 0.1% DMSO; Wy, 10 µM Wy 14,643; GW4064, 1 µM GW4064). Normalized values (luciferase activity / β-galactosidase activity) of Veh-treated cells transfected with empty plasmid were set as fold 1. (*P < 0.05, **P < 0.01 vs Empty treated with Veh; ^##P < 0.01). d, Functional role of DR1 motif in the regulatory region of mouse LC3a and LC3b for PPARα or FXR activity. Cell-based luciferase reporter assays were performed in AML12 cells by transiently transfecting 3 tandem copies of mouse LC3a/LC3b DR1 luciferase reporter construct (3 × LC3a/LC3b DR1 WT-Luc) or mutated version (3 × LC3a/LC3b DR1 mutant-Luc) and CMX-β-gal in a combination of expression plasmids of PPARα, FXR or both, followed by drug treatment for 20 h (Veh, 0.1% DMSO; Wy, 10 µM Wy-14,643; GW4064, 1 µM GW4064). Normalized values (luciferase activity / β-galactosidase activity) of Veh-treated cells transfected with empty plasmid were set as fold 1. (**P < 0.01 vs Empty treated with Veh; ^#P < 0.05, ^##P < 0.01). e, Cell-based luciferase reporter assays were performed in AML12 cells by transiently transfecting siControl, siNCoR, siSMRT, or siSHP along with 3 tandem copies of mouse LC3b DR1 luciferase reporter construct (3 × LC3b DR1 WT-Luc), expression plasmids of PPARα and FXR, and CMX-β-gal followed by drug treatment for 24 h (Veh, 0.1% DMSO; Wy, 10 µM Wy 14,643; GW4064, 1 µM GW4064). Normalized values (luciferase activity / β-galactosidase activity) of Veh-treated cells transfected with siControl were set as fold 1. (**P < 0.01 vs siControl treated with Veh; ^##P < 0.01). Panels in b-e, data represent mean±s.e.m. Statistics by two-tailed t-test.

Extended Data Figure 9. PPARα or FXR activation controls recruitments of coregulators and epigenetic marks in the enhancer regions of LC3a and LC3b genes

Fed or fasted WT mice were orally gavaged with Veh, GW7647, or GW4064 twice a day. Hepatic ChIP-qPCR analysis with indicated antibodies (p300, NCoR1, SMRT, acetyl-H4, and H3K27me3) was used to determine recruitments of coregulators and subsequent alterations of epigenetic marks induced by PPARα/FXR genomic competition for DR1 found in the enhancer region of LC3a and LC3b genes. (n=3 per group, *P < 0.05, **P < 0.01 vs Fed WT mice treated with Veh; ^##P < 0.01 vs Fasted WT mice treated with Veh). Data represent mean±s.e.m. Statistics by two-tailed t-test.

Extended Data Figure 10. Working model of the coordination of hepatic autophagy by nutrient sensing NRs, PPARα and FXR

a, Fed or fasted WT mice were orally gavaged with Veh, GW7647, or GW4064 twice a day. Hepatic CRTC2 ChIP-qPCR in the promoter and enhancer region of LC3a gene. (n=3 per group, **P < 0.01 vs Fed WT mice treated with Veh; ^##P < 0.01 vs Fasted WT mice treated with Veh). Data represent mean±s.e.m. Statistics by two-tailed t-test. b, Proposed model depicting transcriptionally activating or suppressive nutrient sensing NR, PPARα or FXR, respectively, which coordinates autophagy in liver. PPARα or FXR activation competes each other for binding to response elements found in autophagy-related genes.

Figure 1. Activation of PPARα or FXR controls autophagic flux in murine hepatocytes

Representative confocal images of AML12 cells transiently expressing mRFP-GFP-LC3 plasmids followed by treatment of vehicle (Veh), GW7647 (100 nM), or GW4064 (10 µM) for 24 h (out of 30 cells per condition). Cells were starved for 2 h (lower panels: Veh or GW4064-treated cells). DNA was counterstained with DAPI (blue). Scale bar, 20 µm. b and c, Immunoblotting of autophagy (LC3-I/II and p-Ulk1^S757) or mTORC1 activity (p-S6^S240/244) in AML12 cells treated with indicated doses of Wy-14,643 (Wy) or GW4064 for 24 h or 48 h. GW4064-treated cells were starved for 1 h.

Figure 2. Activation of PPARα or FXR controls autophagy in liver

a and b, Fed or fasted WT PPARα^−/−, or FXR^−/− mice treated with Veh, GW7647 or GW4064. LC3 Immunoblots of liver samples. Lanes are pooled (n=5 per group). c, Representative confocal images (out of 9 tissues sections per condition) of liver GFP-LC3 puncta (green: autophagosomes), DAPI (blue: DNA). Fed or fasted GFP-LC3^Tg/+, bigenic PPARα^−/−; GFP-LC3^Tg/+, or FXR^−/−; GFP-LC3^Tg/+ mice treated with Veh, GW7647, or GW4064. Fixed liver samples analyzed by confocal microscopy. Scale bar, 50 µm. GFP puncta per cell are quantified in graph. Data represent mean±s.e.m. of 9 tissue sections. (n=3 per group, **P < 0.01 vs Fed GFP-LC3^Tg/+ mice treated with Veh). Statistics by two-tailed t-test.

Figure 3. PPARα and FXR control autophagic vesicle formation in liver

a, Representative liver electron micrographs. Fed or fasted WT, PPARα^−/−, or FXR^−/− mice treated with Veh, GW7647 or GW4064. Red arrowhead (autolysosome). Scale bar, 0.5 µm. Autophagic vesicles (autophagosome/autolysosome) per cell quantified in graph. Data represent mean±s.e.m. for 30 cells per group (n=3 per group, *P < 0.05, **P < 0.01 vs Veh treated fed WT). Statistics by two-tailed t-test. b, Co-localization of BODIPY 493/503 (green) with LC3 (red) in AML12 cells treated with Veh or 10 µM Wy-14,643 (Wy) for 24 h and cultured with or without 125 µM oleate. DNA stained with DAPI (blue). Scale bar, 20 µm.

Figure 4. Transcriptional coordination of hepatic autophagy by nutrient sensing nuclear receptors in vivo

a, Hepatic autophagy gene expression. Fed or fasted WT mice treated with Veh, GW7647 or GW4064. (n=5 per group). b, PPARα and FXR ChIP-seq tracks for LC3a. Fed WT mice treated with Veh or GW764. (n=4 per group). Boxed peaks contain DR1 motif. c, PPARα/FXR binding to LC3a/b DR1 region determined by ChIP-qPCR. (n=3 per group). (panels in a and c, *P < 0.05, **P < 0.01 vs Veh treated fed WT; ^#P < 0.05, ^##P < 0.01 vs Veh treated fasted WT). In panel a and c, Data represent mean±s.e.m. Statistics by two-tailed t-test.