LETTER Doi:10.1038/Nature14557

LETTER doi:10.1038/nature14557

The CREB coactivator CRTC2 controls hepatic lipid metabolism by regulating SREBP1

Jinbo Han1*, Erwei Li1*, Liqun Chen1, Yuanyuan Zhang1, Fangchao Wei1, Jieyuan Liu2, Haiteng Deng2 & Yiguo Wang1

Abnormal accumulation of triglycerides in the liver, caused in part synthesized as inactive precursors bound to the ER4,5. Upon sensing by increased de novo lipogenesis, results in non-alcoholic fatty liver insulin stimulation or sterol depletion, SREBP1 is transported to the disease and insulin resistance1,2. Sterol regulatory element-binding Golgi through COPII-mediated vesicle trafficking, released by a two- protein 1 (SREBP1), an important transcriptional regulator of step proteolytic cleavage, and then shuttled to the nucleus to induce the lipogenesis, is synthesized as an inactive precursor that binds to expression of genes involved in cholesterol and fatty acid synthesis3–5. the endoplasmic reticulum (ER). In response to insulin signalling, On the basis of these results, we checked whether CRTC2 modulates SREBP1 is transported from the ER to the Golgi in a COPII- nuclear SREBP protein levels. In Crtc22/2 mice fed with both regular dependent manner, processed by proteases in the Golgi, and then and high-fat diets, the active, nuclear-localized SREBP1 (nSREBP1) was shuttled to the nucleus to induce lipogenic gene expression3–5; significantly increased while the full-length SREBP1 (flSREBP1) was however, the mechanisms underlying enhanced SREBP1 activity slightly decreased (Fig. 1c), although total Srebp1c mRNA levels in insulin-resistant obesity and diabetes remain unclear. Here we remained unchanged (Extended Data Fig. 1b), suggesting that CRTC2 show in mice that CREB regulated transcription coactivator 2 mediates SREBP1 activity at the post-transcriptional level. By contrast, (CRTC2)6 functions as a mediator of mTOR7 signalling to modu- Crtc2 deficiency decreased insulin levels and did not modulate the activ- late COPII-dependent SREBP1 processing. CRTC2 competes with ity of SREBP2, a master regulator of cholesterol synthesis5,whichis Sec23A, a subunit of the COPII complex8, to interact with Sec31A, consistent with the lack of cholesterol accumulation (Extended Data another COPII subunit, thus disrupting SREBP1 transport. Fig. 1a). In addition, the abnormal accumulation of nSREBP1 and the During feeding, mTOR phosphorylates CRTC2 and attenuates enhanced lipogenic profile in Crtc22/2 mice were normalized to the level its inhibitory effect on COPII-dependent SREBP1 maturation. found in fed Crtc21/1 mice by adenovirus-mediated wild-type (WT) As hepatic overexpression of an mTOR-defective CRTC2 mutant CRTC2 overexpression or knockdown of Srebp1 (Fig. 1d and Extended in obese mice improved the lipogenic program and insulin sens- Data Fig. 2). Taken together, these results suggest that CRTC2 modulates itivity, these results demonstrate how the transcriptional coactiva- triglyceride synthesis via regulation of SREBP1 maturation. tor CRTC2 regulates mTOR-mediated lipid homeostasis in the fed state and in obesity. a RD HFD c RD HFD Crtc2+/+ Crtc2–/– Crtc2+/+ Crtc2–/– Insulin resistance, which is associated with hyperglycaemia and (kDa) hypertriglyceridaemia, is the central problem of type 2 diabetes1,2,9. Crtc2+/+ 100 flSREBP1 50 nSREBP1 Although the mechanisms that underlie enhanced glucose and trigly- 100 SCAP ceride levels remain elusive, hepatic lipogenesis and gluconeogenesis 50 pAKT –/– AKT are well known to contribute to the paradoxical effects of insulin Crtc2 50 pCRTC2 resistance1,2,9–11. Hepatic lipogenesis is regulated in a combinatorial 75 CRTC2 50 Tubulin manner by transcription factors including peroxisome proliferator- Crtc2–/– b Crtc2+/+ d +/+ activated receptor gamma (PPARc), liver X receptor (LXR), carbohyd- Crtc2–/– Crtc2 Δ 100 TAD GFP GFP WT ΔTAD /AA rate response element binding protein (ChREBP) and sterol regulatory * (kDa) element-binding protein-1c (SREBP-1c), while gluconeogenesis is 100 flSREBP1 50 nSREBP1 regulated by forkhead box protein O1 (FOXO1), PPARc coactiva- protein) –1 250 FASN tor-1a (PGC-1a), cAMP response element-binding protein (CREB) 250 ACACA and CREB regulated transcription coactivators (CRTCs) at the tran- 50 37 SCD1 5,6,10,11 100 SCAP scriptional level . CRTCs are extensively studied in glucose home- * 50 pAKT ostasis, whereas previous studies have suggested possible roles of AKT 50 HA-CRTC2 12–14 100 CRTC2 CRTCs in lipid homeostasis . These findings prompted us to invest- Δ

Triglycerides (mg g (mg Triglycerides 75 TAD igate whether CRTC2 directly affects lipid levels in the liver, where 0 ΔTAD/AA Tubulin CRTC2 is highly expressed. RD HFD 50 We measured hepatic lipid levels in Crtc21/1 and Crtc22/2 mice. Figure 1 | Accumulation of hepatic lipids and mature SREBP1 in Crtc22/2 Hepatic triglycerides, but not cholesterol, are increased by 50% in mice. a–c, Haematoxylin and eosin sections of liver (a), levels of hepatic Crtc22/2 mice compared to Crtc21/1 mice fed with a regular diet; triglycerides (b) and immunoblots showing hepatic amounts of full-length, the ratio is even higher under high-fat diet (HFD) feeding conditions inactive SREBP1 (flSREBP1) and cleaved, active SREBP1 (nSREBP1) in liver extracts (c) from Crtc21/1 and Crtc22/2 mice fed with a regular diet (RD) (Fig. 1a, b and Extended Data Fig. 1a). Transcript analysis revealed or high-fat diet (HFD) for 18 weeks. Scale bar, 50 mm. Data are shown as specific upregulation of SREBP1-target genes involved in triglyceride 2/2 mean 6 s.e.m. *P , 0.01, n 5 10. d, Effect of CRTC2 and its mutants synthesis in Crtc2 mice fed with both regular and high-fat diets (CRTC2(DTAD), amino acids 1–630; CRTC2(DTAD/AA), amino acids 1–630 (Extended Data Fig. 1b). SREBP1 (also called SREBF1) belongs to the with double alanine mutations at serine 171 and serine 275) on SREBP1 family of basic helix–loop–helix-leucine zipper transcription factors maturation and lipogenic protein levels (FASN, SCD1 and ACACA).

1MOE Key Laboratory of Bioinformatics, Tsinghua-Peking Center for Life Sciences, School of Life Sciences, Tsinghua University, Beijing 100084, China. 2Proteomics Facility, School of Life Sciences, Tsinghua University, Beijing 100084, China. *These authors contributed equally to this work.

Shuttling of CRTC2 between the cytoplasm and nucleus is con- To establish that CRTC2 and Sec31A interact directly, we performed trolled mainly by the phosphorylation status of serine at sites 171 His-tag pull-down assays in vitro. Wild-type CRTC2 effectively atte- and 275 (refs 6, 15). To investigate whether the cellular localization nuated the Sec23A–Sec31A interaction and bound less tightly to of CRTC2 affects SREBP1 maturation, we made two CRTC2 mutants: Sec31A than Sec23A did, while CRTC2(W143A) neither bound DTAD (amino acids 1–630), which lacks the transactivation domain Sec31A nor disrupted Sec23A–Sec31A (Fig. 2d). We further per- but still shuttles between the cytoplasm and nucleus; and DTAD/AA, formed an in vitro budding assay to examine whether CRTC2 regulates which is confined to the nucleus because of serine-to-alanine muta- COPII-dependent SREBP1 transport. As expected, the wild-type but tions at positions 171 and 275 (Extended Data Fig. 3a, b). Similar to not the Sec31A-interaction-defective mutant CRTC2 decreased wild-type CRTC2, the DTAD mutant also blocks SREBP1 maturation SREBP1c budding from the ER (Fig. 2e and Extended Data Fig. 5a), in Crtc22/2 mice without affecting the levels of pAKT or circulating which was further confirmed in mice by in vivo testing of nuclear insulin (Fig. 1d and Extended Data Fig. 3c, d), while DTAD/AA was SREBP1 levels and lipogenic gene expression (Fig. 2f and Extended not able to suppress the processing of SREBP1, suggesting that Data Fig. 5b, c). Thus, CRTC2 negatively regulates SREBP1 processing SREBP1 processing depends on cytosolic but not nuclear CRTC2. by competing with Sec23A for binding to Sec31A. On the basis of our previous results that CRTC2 is peripherally It is well established that the processing and nuclear activity of associated with the ER15,16, we hypothesized that CRTC2 may bind SREBP1 is induced in response to insulin signalling and nutrient sig- to proteins involved in the regulation of SREBP1 transport. Indeed, nalling3,5,18. We hypothesized that the CRTC2–Sec31 interaction may CRTC2 binds to Sec31A, a subunit of the COPII complex8, as iden- modulate hormonal and nutritional activation of SREBP1. In fact, both tified by mass spectrometry and confirmed by co-immunoprecipita- insulin and amino acid stimulation attenuated the CRTC2–Sec31A tion and co-immunostaining assays (Fig. 2a and Extended Data Fig. interaction with concomitant enhancement of the Sec23A–Sec31A 4a–d). Further analysis showed that Trp143 of CRTC2 is important for association (Extended Data Fig. 6a, b). As shown in Extended Data the CRTC2–Sec31A interaction, since a tryptophan-to-alanine mutant Fig. 6a, b, the regulatory effect of CRTC2 on the Sec23–Sec31A inter- (W143A) abolished the interaction between CRTC2 and Sec31A, action was abolished in the presence of torin1, an inhibitor of mTOR without perturbing its subcellular localization and transcriptional that controls lipid metabolism and SREBP1 activation7,18–21. To exam- activity (Fig. 2b and Extended Data Fig. 4e–g). Conversely, ine whether mTOR directly phosphorylates CRTC2, Sec23A or CRTC2 associates with a carboxy-terminal polypeptide of Sec31A Sec31A and modulates the association of this complex, we identified (Fig. 2c). Since the C terminus of Sec31A also mediates its interaction their phosphorylation sites by mass spectrometry analysis (Extended with Sec23A (ref. 17), we proposed that CRTC2 may block Sec23A– Data Fig. 6c and data not shown for Sec23A, Sec31A). The conserved Sec31A binding. Supporting this notion, increased amounts of CRTC2 serine site at position 136 (Ser136) of CRTC2, which occurs in attenuate Sec23A binding to Sec31A, and vice versa (Extended the context of a classic mTOR substrate motif (S/T-P)22,was Data Fig. 4h). However, Sec23A, with higher affinity to Sec31A, phosphorylated without rapamycin treatment and dephosphorylated was more potent at disrupting the CRTC2–Sec31A interaction. in the presence of rapamycin (Figs 2b and 3a and Extended Data

a IP IP c Flag–Sec31A HA–CRTC2 d Sec31A WT WD40 1220 + N WD40 340 – C 1220 + IgG CRTC2 Input IgG CRTC2 Input CRTC2 Sec23A 150 Sec31A CΔ420 800 – (kDa) CRTC2 WT W143A 75 IP: HA Input His-CRTC2: – – – – + + + + + + CRL-2189 Hepatocyte His-Sec23A: – + – + –- – + –- – + His/HA-Sec31A(C): – –- + + –- + + –- + + Merge 420 420 Sec31A CRTC2 DAPI Δ Δ 100 CRTC2 Sec31A: Con WT N C C Con WT N C C CRTC2: 75 Sec23A

L-2189 –++++ –++++ 150 (kDa) CR 100 Sec31A(C) IP: anti-HA IB: anti-His Merge 75 Sec31A CRTC2 DAPI 50 100 CRTC2 37 75 Sec23A (kDa) * CRTC2 Sec31A(C) Hepatocyte 75 Tubulin Silver staining IP: Flag Input 50 b e f Full reaction Crtc2+/+ Crtc2–/– Flag– CRTC2: CRTC2 GFP GFP WT W143A Con WT W143ACon WT W143A (kDa) 150 Sec31A (kDa) 100 flSREBP1

-Cytosol CRTC2 -ATP/GTP 75 t 50 nSREBP1 Tubulin –––– WT W143A 50 (° ) 100 SCAP

Inpu C H. sapiens 30 30 0 30 30 30 CRTC1 S PPADTS W RR (kDa) 50 pAKT M. musculus CRTC1 S PPADTS W RR 100 HA–SREBP1c 50 AKT D. rerio CRTC S PPPDTS W RR 50 ERGIC-53 HA–CRTC2 H. sapiens CRTC2 S PPPESS W RR GRP94 75 CRTC2 M. musculus CRTC2 S PPPESG W RR 75 GRP78 50 Tubulin Figure 2 | CRTC2 attenuates SREBP1 processing by competing with c, Deletion analysis of regions in Sec31A required for the CRTC2–Sec31A Sec23A for binding to Sec31A. a, Co-immunoprecipitation (co-IP) of interaction. Interaction-competent Sec31A polypeptides are indicated by (1) endogenous CRTC2 with Sec31A (top) and co-immunostaining of CRTC2 and in each schematic. d, In vitro pull-down assay showing the binding ability of Sec31A (bottom) in mouse CRL-2189 cells and primary hepatocytes. Scale bars, His-tagged full-length CRTC2, CRTC2(W143A) mutant and Sec23A to the C 10 mm. DAPI, 4,6-diamidino-2-phenylindole. b, Top: immunoblots showing terminus of His/HA–Sec31A (amino acids 701–1220). The asterisk shows an the relative association of Sec31A with wild-type CRTC2 (WT) or CRTC2 with unspecific protein band. e, f, Effect of wild-type (WT) and Sec31A-binding- a tryptophan-to-alanine mutation at position 143 (W143A) in transfected defective (W143A) CRTC2 on HA-tagged SREBP1c transport from the ER to HEK293T cells by co-immunoprecipitation assay. Con, mock transfection. the Golgi in an in vitro budding assay (e) and SREBP1 processing in fed mice (f). Bottom: amino acid sequence alignment of vertebrate CRTC orthologues with The constitutive budding protein (ERGIC-53) and ER lumen proteins (GRP94 the conserved Sec31A binding site (W) and mTOR phospho-site (S) boxed. and GRP78) are indicated as controls (e).

IP: IgG Sec31A a y + y + y + y + y + y + c Figure 3 | mTOR phosphorylates CRTC2 and 13 10 9 8 6 4 promotes COPII-dependent SREBP1 activation. HI DS SP YS PA Y L (p)S PP PE SGWR 124 144 CRTC2: a, Identification of CRTC2 Ser136 phosphoryl- b + b + b + b + b + b + b + WT S136AWT S136A 2 4 6 8 10 12 13 Insulin: –+ –+ –+ –+ ation by liquid chromatography mass spectrometry y + 8 75 HA–CRTC2 CRTC2 (LC-MS/MS) analysis. The MS/MS spectrum of b IP 75 Sec23A 150 Sec31A the phosphorylated peptide HIDSSPYSPAYL(p) (kDa) HA–CRTC2 SPPPESGWR is shown. The labelled peaks show WT S136AS6K 75 [M+2H]2+-P 75 Sec23A the masses of y or b ions of the phosphorylated b + mTOR: – + – + – + 2 b + pCRTC2 150 Sec31A b + + 13 -P 5 b 75 HA–pCRTC2 peptide. b, In vitro kinase assay showing the b + y + 8 y +-P y +-P pS6K 3 4 y + 10 13 75 (S136) b + 7 + Autoradiography 7 y -P Input phosphorylation of His-tagged CRTC2 and S6K by b + y + 9 b + b +b + 50 pAKT 4 6 10 11 12 (kDa) CRTC2 Relative abundance Relative AKT truncated mTOR. c, Immunoblots of co- 75 S6K 50 (m/z) 400 600 800 1,0001,200 1,400 Coomassie blue 50 Tubulin immunoprecipitation assay showing the effect of –/– wild-type and the mTOR-defective mutant Crtc2+/+ Crtc2 d e Crtc2+/+ Crtc2–/– CRTC2(S136A) on Sec31A–Sec23A interaction in CRTC2: GFP GFP WT S136A mouse primary hepatocytes in response to 100 nM (kDa) FastedRefedInsulinFastedRefedInsulin insulin stimulation for 30 min. d, Regulatory 75 CRTC2 fed sted fed a effect of CRTC2 on Sec31A–Sec23A binding by Sec23A FastedRe FastedRefed FastedRefed F Re 1 1 IP 75 (kDa) fasting, refeeding or insulin treatment in Crtc2 / 150 Sec31A flSREBP1 100 and Crtc22/2 mice. Mice were fasted for 3 h then CRTC2 50 nSREBP1 75 refed for 1 h or treated with intraperitoneal 75 Sec23A 100 SCAP pS6K 21 150 Sec31A 50 administration of insulin (0.1 U kg )for1h. pCRTC2 S6K 75 (S136) 50 e, Effect of wild type and the S136A mutant of pAKT 50 pS6K 50 CRTC2 on SREBP1 maturation in fasted (3 h) and

Input AKT 50 S6K 50 pCRTC2 refed (1 h after 3 h fasting) mice. 50 pAKT 75 (S136) AKT HA–CRTC2 50 75 CRTC2 50 Tubulin 50 Tubulin

Fig. 6c). In addition, a co-immunoprecipitation assay revealed that and triglycerides were all increased in HFD-fed, db/db and ob/ob mice mTOR interacts with CRTC2, and an in vitro kinase assay showed (Fig. 4a and Extended Data Fig. 8). Meanwhile, the Sec31A–Sec23A that mTOR directly phosphorylates CRTC2 at Ser136 (Fig. 3b and interaction and the level of phospho-CRTC2(Ser136) were enhanced Extended Data Fig. 6d). To confirm the phosphorylation of CRTC2 after HFD feeding. Accordingly, the association of CRTC2–Sec31A at Ser136 in vivo, we raised a polyclonal antibody against the phos- was blocked, although the protein level of COPII subunits remained pho-CRTC2 (Ser136) peptide and found that both insulin and stable (Fig. 4a and Extended Data Fig. 8c). Since the CRTC2(S136A) amino acids stimulate CRTC2 Ser136 phosphorylation in an mutant blocks SREBP1 processing, we asked whether this mTOR- mTOR-dependent manner, suggesting that CRTC2 is a bona fide defective mutant was able to reduce hepatic hypertriglyceride levels substrate of mTOR (Fig. 3c and Extended Data Fig. 6a, b). by inhibiting SREBP1 activation in obesity. To exclude a gluconeo- Phosphorylation of CRTC2 Ser136 is much more sensitive to torin1 genic effect of CRTC2(S136A), we used an mTOR-defective CRTC2 than to rapamycin treatment (Extended Data Fig. 6e). Although mutant without the transactivation domain (CRTC2(DTAD/S136A)) mTOR-defective CRTC2(S136A) had similar cellular localization to restore SREBP1 processing in HFD-fed mice. CRTC2(DTAD/ and nuclear activity on Cre-luc to wild-type CRTC2 (Extended S136A) did not affect mouse body weight, fat mass, food intake, energy Data Fig. 6f, g), it reduced the insulin-stimulated enhancement of expenditure and liver function measured by alanine aminotransferase Sec23A–Sec31A association (Fig. 3c). (ALT) and aspartate aminotransferase (AST) activity, but it reduced Since CRTC2 was able to be phosphorylated in primary hepatocytes SREBP1 processing, the lipogenesis rate and gluconeogenic capacity, by mTOR, we next checked whether CRTC2 could be phosphorylated as well as improving hepatic steatosis and insulin sensitivity (Fig. 4b–e via fasting–refeeding transition in mice, thereby modulating the and Extended Data Fig. 9a–f). To investigate further the role of Sec23A–Sec31A interaction and SREBP1 activation. Refeeding led to CRTC2(DTAD/S136A) in insulin sensitivity, we performed hyper- CRTC2 phosphorylation and disrupted its association with Sec31A, insulinaemic-euglycaemic clamp studies. The steady-state glucose infu- which became available for Sec23A binding, thus enhancing the sion rate almost doubled during the clamp studies in CRTC2(DTAD/ Sec23A–Sec31A interaction; insulin had a similar but weaker effect S136A) mice, reflecting enhanced whole-body insulin responsiveness, (Fig. 3d). However, the dynamic regulation of Sec31A–Sec23A by and was accompanied by an increase in the glucose disposal rate CRTC2 was lost in Crtc22/2 mice (Fig. 3d). In addition, wild-type (Fig. 4f, g). The insulin-stimulated glucose disposal rate, which prim- CRTC2 normalized nSREBP1 in Crtc22/2 mice during both fasting arily reflects skeletal muscle insulin sensitivity, and the insulin- and refeeding to the level in Crtc21/1 mice, and was phosphorylated induced suppression of plasma free fatty acid levels, which indicates during refeeding (Fig. 3e). The CRTC2(S136A) mutant had a stronger white adipose tissue insulin sensitivity, were both slightly increased in inhibitory effect on SREBP1 processing and lipogenic gene expression the presence of hepatic CRTC2(DTAD/S136A), suggesting a possible during refeeding owing to deficient phosphorylation by mTOR (Fig. 3e role of inter-organ communication in orchestrating systemic insulin and Extended Data Fig. 7a, b). Furthermore, CRTC2(S136A) did not sensitivity. However, the insulin sensitivity in liver was markedly affect the outcome of torin1 treatment on SREBP1 activation, hepatic improved as judged from insulin-induced suppression of hepatic glu- triglyceride levels, lipin1 expression or cellular localization (Extended cose production and further confirmed by pAKT levels (Fig. 4h and Data Fig. 7c–e), suggesting that both CRTC2 and lipin1, as mTOR Extended Data Fig. 9g–k). downstream mediators, regulate SREBP1 activation at different Previous studies showed that nuclear CRTC2, as a transcriptional steps18. Taken together, these results demonstrate that mTOR modu- coactivator, plays an important part in gluconeogenesis during fasting lates COPII-dependent SREBP1 processing via Ser136 phosphoryla- or ER stress via its binding to different partners6. Here, our results tion of CRTC2. demonstrate that cytosolic CRTC2, as a critical mediator of mTOR, Considering that lipogenesis is chronically enhanced in obesity and modulates COPII activity, which leads to SREBP1 processing and diabetes1,2,5,11, we tested whether the mTOR–CRTC2 axis is altered in enhancement of de novo lipogenesis (Fig. 4i), thereby contributing this setting. Consistent with previous studies23–25, hepatic nuclear to hepatic lipid levels and insulin resistance along with potential altera- SREBP1, mTOR activity, and levels of branched-chain amino acids tions in free fatty acid supply from either dietary or adipose tissue

a HFD (W) c 1.2 d 60 e Figure 4 | Dysregulated SREBP1 signalling in (kDa) 0 6 12 18 obese mice is improved by overexpression of 75 CRTC2 mTOR-defective CRTC2. a, Immunoblots 0.8 40 GFP IP 75 Sec23A * showing co-immunoprecipitation of CRTC2 and 150 Sec31A * protein) Sec23A with Sec31A in livers of mice fed with a CRTC2 –1 75 0.4 20 HFD for different times (W, weeks). b–e, Effect

Triglycerides Triglycerides CRTC2 75 Sec23A of the mTOR-defective mutant CRTC2(DTAD/ (mg g (mg Δ 150 ( TAD/S136A) Sec31A liver in H-Palmitate

2 S136A) on SREBP1 maturation (b), relative 100 flSREBP1 0 0 synthesis rate of hepatic palmitate (c), levels of 50 nSREBP1 hepatictriglycerides(d), and hepaticsteatosis shown pCRTC2 GFP CRTC2(ΔTAD/S136A) 75 (S136) by haematoxylin and eosin staining of liver (e)in pmTOR Input 250 f 30 g 40 i mice fed with a HFD for 18 weeks. f–h,Glucose

250 mTOR )

) * infusion rate (GIR; f), glucose disposal rate –1 50 pS6K –1 * Insulin Amino acids 30 (GDR; g) and percentage suppression of hepatic S6K

50 min

min 20 glucose production (HGP; h) during –1 pAKT –1 50 mTOR AKT 20 hyperinsulinaemic-euglycaemic clamp studies. 50 i, mTOR modulates COPII-dependent SREBP1 50 Tubulin 10 10 activation via Ser136 phosphorylation of CRTC2. CRTC2 b kg (mg GIR GFP (ΔTAD/S136A) kg (mg GDR CRTC2 During feeding, Sec31A dissociates from CRTC2, (kDa) 0 0 which becomes phosphorylated at Ser136 by 100 flSREBP1 Sec31 nSREBP1 mTOR, and interacts instead with Sec23A, thus 50 promoting COPII-dependent transport and 250 FASN h 60 S136 ACACA * Sec31 P processing of SREBP1 in the Golgi. In obesity, 250 CRTC2 37 SCD1 Sec23 enhanced phosphorylation of CRTC2 by mTOR 40 100 SCAP contributes at least in part to increased SREBP1 50 pAKT activation and hepatic lipogenesis. Scale bar, 50 mm. AKT nSREBP1 Data are shown as mean 6 s.e.m. *P , 0.01, n 5 8 50 20 SREBP1 CRTC2 Lipogenesis (c, d), n 5 6(f–h). 75 ΔTAD Nu

/S136A ER cl HGP suppression (%) suppression HGP 50 Tubulin 0 lipolysis1,2. Our findings expand the role of the transcription coactiva- 14. Altarejos, J. Y. et al. The Creb1 coactivator Crtc1 is required for energy balance and fertility. Nature Med. 14, 1112–1117 (2008). tor CRTC2 to include lipid metabolism, and provide insight into how 15. Wang, Y. et al. Inositol-1,4,5-trisphosphate receptor regulates hepatic SREBP1 activity is enhanced in obesity and diabetes. Considering the gluconeogenesis in fasting and diabetes. Nature 485, 128–132 (2012). isoform diversity of COPII subunits and CRTCs6,8, as well as the envir- 16. Wang, Y., Vera, L., Fischer, W. H. & Montminy, M. The CREB coactivator CRTC2 links onmental cues that activate mTOR7, mTOR–CRTC signalling may hepatic ER stress and fasting gluconeogenesis. Nature 460, 534–537 (2009). 17. Bi, X., Mancias, J. D. & Goldberg, J. Insights into COPII coat nucleation from the have important functions in regulating other cargo transport processes structure of Sec23NSar1 complexed with the active fragment of Sec31. Dev. Cell in different settings. Therefore, it will be interesting to explore the 13, 635–645 (2007). selectivity and specificity of this signalling axis in the future. 18. Peterson, T. R. et al. mTOR complex 1 regulates lipin 1 localization to control the SREBP pathway. Cell 146, 408–420 (2011). Online Content Methods, along with any additional Extended Data display items 19. Li, S., Brown, M. S. & Goldstein, J. L. Bifurcation of insulin signaling pathway in rat and Source Data, are available in the online version of the paper; references unique liver: mTORC1 required for stimulation of lipogenesis, but not inhibition of to these sections appear only in the online paper. gluconeogenesis. Proc. Natl Acad. Sci. USA 107, 3441–3446 (2010). 20. Porstmann, T. et al. SREBP activity is regulated by mTORC1 and contributes to Akt- Received 9 June 2014; accepted 28 April 2015. dependent cell growth. Cell Metab. 8, 224–236 (2008). 21. Yecies, J. L. et al. Akt stimulates hepatic SREBP1c and lipogenesis through parallel Published online 6 July 2015. mTORC1-dependent and independent pathways. Cell Metab. 14, 21–32 (2011). 22. Hsu, P. P. et al. The mTOR-regulated phosphoproteome reveals a mechanism 1. Samuel, V. T. & Shulman, G. I. Mechanisms for insulin resistance: common threads of mTORC1-mediated inhibition of growth factor signaling. Science 332, and missing links. Cell 148, 852–871 (2012). 1317–1322 (2011). 2. Perry, R. J., Samuel, V. T., Petersen, K. F. & Shulman, G. I. The role of hepatic lipids in 23. Khamzina, L., Veilleux, A., Bergeron, S. & Marette, A. Increased activation of the hepatic insulin resistance and type 2 diabetes. Nature 510, 84–91 (2014). mammalian target of rapamycin pathway in liver and skeletal muscle of obese 3. Yellaturu, C. R. et al. Insulin enhances post-translational processing of nascent rats: possible involvement in obesity-linked insulin resistance. Endocrinology 146, SREBP-1c by promoting its phosphorylation and association with COPII vesicles. J. 1473–1481 (2005). Biol. Chem. 284, 7518–7532 (2009). 24. Um, S. H. et al. Absence of S6K1 protects against age- and diet-induced obesity 4. Goldstein, J. L., DeBose-Boyd, R. A. & Brown, M. S. Protein sensors for membrane while enhancing insulin sensitivity. Nature 431, 200–205 (2004). sterols. Cell 124, 35–46 (2006). 25. Shimomura, I. et al. Decreased IRS-2 and increased SREBP-1c lead to mixed 5. Horton, J. D., Goldstein, J. L. & Brown, M. S. SREBPs: activators of the complete insulin resistance and sensitivity in livers of lipodystrophic and ob/ob mice. Mol. program of cholesterol and fatty acid synthesis in the liver. J. Clin. Invest. 109, Cell 6, 77–86 (2000). 1125–1131 (2002). 6. Altarejos, J. Y. & Montminy, M. CREB and the CRTC co-activators: sensors for Acknowledgements We thank M. Montminy, W. E. Balch, L. Yu, S. Fu, P. Li, J. Wu and hormonal and metabolic signals. Nature Rev. Mol. Cell Biol. 12, 141–151 (2011). D. Li for discussion, and X. Liu and P. Li for technical help with lipogenesis analyses and 7. Lamming, D. W. & Sabatini, D. M. A Central role for mTOR in lipid homeostasis. Cell clamp studies, respectively. This work was supported by grants from 1000 Talents Metab. 18, 465–469 (2013). Program for Young Scholars and the National Science Foundation of China 8. Zanetti, G., Pahuja, K. B., Studer, S., Shim, S. & Schekman, R. COPII and the (31322027, 31321003 and 31471127) to Y.W. regulation of protein sorting in mammals. Nature Cell Biol. 14, 20–28 (2012). 9. Brown, M. S. & Goldstein, J. L. Selective versus total insulin resistance: a pathogenic Author Contributions J.H., E.L. and Y.W. designed the study and analysed the data. J.H., paradox. Cell Metab. 7, 95–96 (2008). E.L., L.C., Y.Z., F.W. and Y.W. performed the experiments. J.L. and H.D. carried out mass 10. Lin, H. V. & Accili, D. Hormonal regulation of hepatic glucose production in health spectrometry analysis. J.H. and Y.W. wrote the paper. All authors reviewed and and disease. Cell Metab. 14, 9–19 (2011). commented on the manuscript. 11. Strable, M. S. & Ntambi, J. M. Genetic control of de novo lipogenesis: role in diet- induced obesity. Crit. Rev. Biochem. Mol. Biol. 45, 199–214 (2010). Author Information Reprints and permissions information is available at 12. Wang, Y. et al. Targeted disruption of the CREB coactivator Crtc2 increases insulin www.nature.com/reprints. The authors declare no competing financial interests. sensitivity. Proc. Natl Acad. Sci. USA 107, 3087–3092 (2010). Readers are welcome to comment on the online version of the paper. Correspondence 13. Song, Y. et al. CRTC3 links catecholamine signalling to energy balance. Nature 468, and requests for materials should be addressed to Y.W. 933–939 (2010). ([email protected]).

METHODS cell) or Cre-luc-transfected HEK293T cells were exposed to forskolin (10 mM) Mouse strains. Mice were housed in colony cages with a 12 h light/dark cycle in a for 4 h. All cell lines were routinely tested for mycoplasma presence using a temperature-controlled environment. For high-fat diet feeding experiments, regu- PCR detection kit (Sigma, MP0035). lar diet (Research Diets, D12450B) was replaced with a diet containing 60 kcal% fat Immunoblot, immunoprecipitation and immunostaining. Assays were per- (Research Diets, D12492). Crtc22/2 mice have been described previously12. formed as described15,16. CRTC2, pCREB, CREB, AKT, pAKT (Thr308), tubulin, Heterozygous Crtc21/2 mice were backcrossed with C57BL/6J for ten generations HA and Flag antibodies were as previously described15,16. Other antibodies were and then intercrossed to obtain Crtc22/2 mice confirmed by PCR-based genotyp- purchased as follows: rabbit polyclonal anti-Sec31A (A302-336A), Bethyl ing. All the other mice were purchased from Jackson Labs (Bar Harbour, ME). All Laboratories; mouse monoclonal anti-Sec31A (612351), BD Biosciences; anti- animal experiments were approved by the Animal Care and Use Committee at KDEL (ab12223) and anti-Sec12 (ab3422), Abcam; anti-ERGIC-53 (GTX63674), Tsinghua University. GeneTex; anti-His (D291-3), MBL International; anti-Sar1 (07-692), Millipore; Plasmids and adenoviruses. Plasmids containing human Sec13, Sec31A and anti-SREBP1 (SC-13551, SC-367), anti-Myc (SC-40), Santa Cruz; anti-SCAP Sec23A were provided by W. E. Balch (The Scripps Research Institute). HA-tagged (13102), anti-ACACA (3662), anti-FASN (3180), anti-SCD1 (2438), anti-Sec23A S6K (8984) and Myc-tagged mTOR (1861) plasmids were from Addgene. Myc- (8162), anti-S6K (2708), anti-pS6K (9234), anti-mTOR (2983) and anti-pmTOR DDK-tagged lipin1 (RC207138) plasmid was purchased from OriGene. (5536), anti-lipin1 (5195), Cell Signaling Technology; anti-Sec24A (15958-1-AP) Adenoviruses (1 3 108 plaque forming units (pfu) GFP, CRTC2, CRTC2(W143A), and anti-SREBP2 (14508-1-AP), Proteintech Group Inc.; anti-Sec13 (NBP2- CRTC2(S136A), CRTC2(DTAD) (amino acids 1–630), CRTC2(DTAD/AA) 20278), Novus Biologicals. The phospho-S136 CRTC2 antibody was generated (S171A, S275A), CRTC2(DTAD/S136A), Srebp1 RNAi, or unspecific RNAi) were and purified by Beijing Prorevo Biotech Co., Ltd. 1/1 2/2 delivered to male C57BL/6J, HFD-fed, Crtc2 ,orCrtc2 mice by tail vein In vitro budding assay. Mouse liver cytosol was harvested as described28. In brief, injection. Mice were injected with adenovirus on day 0 and killed on day 7. Wild- mouse livers were perfused with 0.9% (w/v) NaCl through the portal vein at room type CRTC2, Crtc2 RNAi, GFP and unspecific RNAi adenoviruses have been temperature. The perfused livers were disrupted in ice-cold buffer (20 mM HEPES described previously15,16. CRTC2(W143A), CRTC2(S136A), CRTC2(DTAD), pH 7.4, 150 mM KOAc, 5 mM MgOAc, 250 mM sorbitol) followed by 10 strokes in CRTC2(DTAD/AA) and CRTC2(DTAD/S136A) adenoviruses were constructed a 50 ml Dounce homogenizer. Homogenates were centrifuged at 1,000g for from mouse Crtc2 cDNA. Srebp1 RNAi adenovirus was constructed using the 10 min. Supernatants were sequentially centrifuged at 20,000g for 20 min, sequence 59-GGGATCAAAGAGGAGCCAGTGC-39. All expressed constructs 186,000g for 1 h, and 186,000g for 45 min. The final supernatant was dialysed, used in this study were confirmed by sequencing. divided into multiple aliquots and stored at 280 uC. In vivo analysis and histology. Triglyceride (Sigma, TR0100) and cholesterol In vitro budding assays were carried out as reported28,29. HEK293T cells were co- (BioVision, K603-100) levels in liver and plasma, plasma insulin (Mercodia, transfected with HA–SREBP1c and Flag–SCAP (provided by P. Li, Tsinghua 10-1247-01), plasma alanine aminotransferase (ALT, BioVision, K752-100), University) for 48 h. Cells were then cultured with lipid-deficient medium for plasma aspartate aminotransferase (AST, BioVision, K753-100), hepatic another 3 h, harvested and permeabilized by 5 min incubation in buffer (20 mM branched-chain amino acids (BCAA, Sigma, MAK003) and plasma free fatty acid HEPES pH 7.4, 150 mM KOAc, 5 mM MgOAc, 250 mM sorbitol) with 40 mgml21 levels (BioVision, K612-100) were measured according to the manufacturer’s digitonin. The permeabilized cells were washed with the same buffer and used at instructions. Blood glucose values were determined using a LifeScan automatic 40 mg protein per reaction. The budding reaction (4 mg ml21 mouse liver cytosol, glucometer. Insulin tolerance tests, glucose tolerance tests and pyruvate 1 mM ATP, 40 mM creatine phosphate, 0.2 mg ml21 creatine phosphokinase, 12 tolerance tests were conducted as previously reported . De novo lipogenesis 0.1 mM GTP, and 100 ng His–CRTC2 or His–CRTC2 (W143A) fusion proteins 26 experiments were performed as previously reported . Pieces of liver tissue purified from Escherichia coli) was assembled on ice, incubated for 30 min at 30 uC were sent to the Metabolomics Center at Tsinghua University to determine the and then put on ice. The donor membranes were removed by centrifugation at 2 H2O incorporated palmitate levels by liquid chromatography and mass spectro- 12,000g for 20 min at 4 uC. The supernatant fraction was centrifuged at 4 uCby metry. Hyperinsulinaemic-euglycaemic clamps were performed as previously 25 min 55,000 rpm using a TLA100 rotor in a Beckman Optima TLX ultracentri- 27 described . Three days after adenovirus administration, mice were implanted fuge. The collected vesicles were analysed by SDS–PAGE. with dual jugular catheters for 4 days for use in clamp studies. Food intake and In vitro kinase assay. His-tagged CRTC2, CRTC2(S136A) and S6K fusion pro- energy expenditure were simultaneously measured for individually housed mice teins were purified from E. coli. The kinase assay was performed as reported18. The with a PhenoMaster system (TSE Systems). Relative fat mass was measured with reaction system (20 ml), containing 150 ng fusion protein, 20 ng truncated mTOR an EchoMRI analyser. For histology, mouse tissues were fixed in 4% paraformalde- (Millipore, 14-770) in reaction buffer (25 mM HEPES pH 7.4, 50 mM KCl, 5 mM hyde and paraffin embedded. Sections (8 mm) were used for haematoxylin and 32 MgCl2,and5mMMnCl2), 50 mM cold ATP and 2 mCi [c- P]ATP, was incubated eosin staining. for 30 min at 30 uC. Reactions were stopped by adding 5 ml sample buffer, then Quantitative PCR. Total RNA from mouse liver was extracted using RNeasy boiled for 10 min and analysed by SDS–PAGE followed by autoradiography. kits (Qiagen). cDNA was obtained by the iScript cDNA synthesis kit (Bio-Rad). Mass spectrometry (MS). To identify CRTC2-interacting proteins, the ER frac- RNA levels were measured with the LightCycler 480 II (Roche) as previously 15,16 tion of CRL-2189 cells was isolated according to the manufacturer’s instructions described . The following primers were used for qPCR: Acaca-forward (Sigma-Aldrich, ER0100). Immunoprecipitates of endogenous CRTC2 from the GGATGACAGGCTTGCAGCTAT, Acaca-reverse TTTGTGCAACTAGGAAC ER fraction were prepared for MS studies as previously reported15,16 and analysed by GTAAGTCG; Acox1-forward CTGCCAAGGGACTCCAGAGCAGCT, Acox1- electrospray ionization tandem MS on a Thermo LTQ Orbitrap instrument. To reverse GACATGGACACATCCACCATGCAG; Actin-forward GTCCACCCC identify the mTOR phospho-site(s) in CRTC2, HEK293T cells were transfected by GGGGAAGGTGA, Actin-reverse AGGCCTCAGACCTGGGCCATT; Apoa4- Flag–CRTC2 and treated with or without 100 nM rapamycin for 1 h. Immuno- forward GCCCCATGCCAACAAAGTAA, Apoa4-reverse CCTTGATCGTGG precipitates of Flag–CRTC2 were used to determine phospho-site(s) by MS. TCTGCATG; Chrebp1-forward CTCAGGGTATGCAACCCAGGTG, Chrebp1- Statistical analyses. Age- and weight-matched male mice were randomly assigned reverse GACAGGGGTTGTTGTCTCTGGC; Fasn-forward TGGTGGTGTGG for the experiments. The animal numbers used for all experiments are outlined in ACATGGTCACAGA, Fasn-reverse CCGAAGCTGGGGGTCCATTGTGTG; the corresponding figure legends. No animals were excluded from statistical ana- Gpat1-forward GCCCTTCGTGGGAAGGTGCTGCTA, Gpat1-reverse CCGTC lyses, and the investigators were not blinded in the studies. All studies were TCGCCAGCCATCCTCTGTG; Mttp-forward GAGCGGCTATACAAGCTCA performed on at least three independent occasions. Results are reported as mean 6 CGTAC, Mttp-reverse TCACCATCAGGATTCCTCCACAGT; PPARa-forward s.e.m. Comparison of different groups was carried out using two-tailed unpaired TCTCCACGTTCCAGCCCTTCCTCA, PPARa-reverse TTCACATGCGTGA Student’s t-test. Differences were considered statistically significant at P , 0.05. ACTCCGTAGTG; PPARc-forward TCCGTGATGGAAGACCACTCGCAT, No statistical methods were used to predetermine sample size. PPARc-reverse CAGCAACCATTGGGTCAGCTCTTG; Scd1-forward CTGTA CGGGATCATACTGGTTCCC, Scd1-reverse CAGCCGAGCCTTGTAAGTTC 26. Zhao, X. et al. Regulation of lipogenesis by cyclin-dependent kinase 8-mediated TGTG; Srebp1c-forward GGAGCCATGGATTGCACATT, Srebp1c-reverse GG control of SREBP-1. J. Clin. Invest. 122, 2417–2427 (2012). CCCGGGAAGTCACTGT; Srebp2-forward GATGAGCTGACTCTCGGGGA 27. Li, P. et al. Adipocyte NCoR knockout decreases PPARc phosphorylation and CATC, Srebp2-reverse GTGGGGTAGGAGAGACTTTGACCT. enhances PPAR activity and insulin sensitivity. Cell 147, 815–826 (2011). 28. Nohturfft, A., Yabe, D., Goldstein, J. L., Brown, M. S. & Espenshade, P. J. Regulated Cell culture and luciferase assay. CRL-2189 and HEK293T (ATCC) cells were –1 step in cholesterol feedback localized to budding of SCAP from ER membranes. cultured in DMEM containing 10% FBS (HyClone), 100 mg ml penicillin–strep- Cell 102, 315–323 (2000). tomycin. Mouse primary hepatocytes were isolated and cultured as previously 29. Schindler, A. J. & Schekman, R. In vitro reconstitution of ER-stress induced ATF6 described15,16. For reporter studies, Ad-Cre-luc-infected hepatocytes (1 pfu per transport in COPII vesicles. Proc. Natl Acad. Sci. USA 106, 17775–17780 (2009).

Extended Data Figure 1 | Profiling of cholesterol, gene expression, hepatic amounts of full-length, inactive SREBP2 (flSREBP2) and cleaved, active protein and circulating insulin levels in Crtc21/1 and Crtc22/2 mice. SREBP2 (nSREBP2), phospho-CREB (pCREB), CREB, and subunits of the a–d, Hepatic cholesterol levels (a), qPCR results for expression of genes COPII complex (Sec12, Sar1, Sec23A, Sec24A, Sec13 and Sec31A) (d)in involved in lipogenic regulation, lipid transport, fatty acid oxidation and Crtc21/1 and Crtc22/2 mice fed with a regular diet (RD) or a high-fat diet triglyceride synthesis (b), plasma insulin level (c), and immunoblots showing (HFD) for 18 weeks. Data are shown as mean 6 s.e.m. *P , 0.01, n 5 10.

Extended Data Figure 2 | Validation of Srebp1 knockdown in mice. a–c, Hepatic triglycerides (TGs) (a), immunoblots (b) and qPCR results for lipogenic gene expression (c) showing the effect of Srebp1 RNAi in liver extracts of fed mice. Data are shown as mean 6 s.e.m. *P , 0.01, n 5 10. US, unspecific.

Extended Data Figure 3 | Effect of CRTC2 and its mutants on primary hepatocytes treated with or without 10 mM FSK for 4 h. Data are gluconeogenic and lipogenic gene expression. a, Cellular localization of shown as mean 6 s.e.m. *P , 0.01, n 5 6. c, d, Effect of CRTC2 and its mutants CRTC2 and its mutants CRTC2(DTAD) (amino acids 1–630) and on lipogenic gene (Fasn, Scd1, Acaca) expression (c) and plasma insulin level CRTC2(DTAD/AA) (amino acids 1–630 with double alanine mutations at (d) in fed mice. Data are shown as mean 6 s.e.m. *P , 0.01, n 5 8. NS, no Ser171 and Ser275) in mouse primary hepatocytes. FSK, forskolin. Scale bar, significant statistical difference. 10 mm. b, Effect of CRTC2 and its mutants on Cre-luc activity in mouse

Extended Data Figure 4 | Association of CRTC2 with Sec31A. a, Immuno- e, Deletion analysis of regions in CRTC2 required for the CRTC2–Sec31A staining showing relative co-localization of CRTC2 with an endoplasmic interaction. Interaction-competent CRTC2 peptides are indicated by (1)in reticulum (ER) marker (KDEL) in CRL-2189 cells. Scale bar, 10 mm. each schematic. f, g, Cellular localization of the tryptophan-to-alanine b, c, Strategy to purify CRTC2-interacting proteins (b), and the peptides mutant of CRTC2 (W143A) and its effect on Cre-luc activity in HEK293T cells. identified from Sec31A and Sec13 (c) by MS analysis of immunoprecipitates Scale bars, 10 mm. Data are shown as mean 6 s.e.m. *P , 0.01, n 5 6. NS, prepared with anti-CRTC2 antibody from CRL-2189 ER fractions. d, Co- no significant statistical difference. h, Co-immunoprecipitation assay showing immunoprecipitation assay showing amounts of Flag-tagged Sec13 or Sec31A amounts of Flag-tagged CRTC2 and YFP-tagged Sec23A recovered from recovered from immunoprecipitates of endogenous CRTC2 in HEK293T cells. immunoprecipitates of HA-tagged Sec31A in HEK293T cells.

Extended Data Figure 5 | Modulation of COPII-dependent SREBP1 activity and Sec31A-interaction-defective (W143A) CRTC2 on lipogenic gene by CRTC2. a, Immunostaining showing the effect of Crtc2 RNAi on the expression (b) and plasma insulin level (c) in fed mice. Data are shown as cellular localization of Sec31A. Scale bars, 10 mm. b, c, Effect of wild-type (WT) mean 6 s.e.m. *P , 0.01, n 5 8. NS, no significant statistical difference.

Extended Data Figure 6 | Characterization of CRTC2 phosphorylation anti-Flag from HEK293T cells treated with 100 nM rapamycin for 1 h (Rap1) site(s) by mTOR. a, Immunoblots showing co-immunoprecipitation of or not (Rap2). Serine 136 was phosphorylated (Yes) in the absence of Rap CRTC2 and Sec23A with Sec31A in mouse primary hepatocytes in response to treatment (Rap2) and dephosphorylated (No) in the presence of Rap (Rap1). insulin and/or torin1 treatment. Mouse primary hepatocytes were incubated d, Co-immunoprecipitation assay showing the association between Flag-tagged with 250 nM torin1 or control vehicle for 1 h before 30 min insulin (100 nM) CRTC2 and Myc-tagged mTOR in HEK293T cells. e, Effect of the mTOR stimulation. Phospho-S6K (pS6K), total S6K, phospho-AKT (pAKT), total inhibitors Rap and torin1 on CRTC2 phosphorylation. Mouse primary AKT and phospho-CRTC2 (Ser136) levels are also indicated. b, Immunoblots hepatocytes were pretreated with vehicle (Veh), 100 nM Rap, or 250 nM torin1 showing co-immunoprecipitation of CRTC2 and Sec23A with Sec31A in for 1 h before 100 nM insulin stimulation for 30 min. f, g, Cellular localization mouse primary hepatocytes in response to amino acids and/or torin1 of the phosphorylation-defective CRTC2 mutant (S136A) (f) and its effect treatment. Mouse primary hepatocytes incubated with amino-acid-free MEM on Cre-luc activity (g) in mouse primary hepatocytes. Scale bars, 10 mm. Data for 3 h were exposed to 250 nM torin1 or control vehicle for another 1 h, then are shown as mean 6 s.e.m. *P , 0.01, n 5 6. NS, no significant statistical treated with amino acids for 30 min. c, Phospho-peptides of Flag-tagged difference. CRTC2 identified by MS analysis of immunoprecipitates prepared with

Extended Data Figure 7 | Effect of CRTC2(S136A) on SREBP1 maturation, Torin1 (20 mg kg21) was intraperitoneally injected 6 h before livers were lipin1 localization and circulating insulin level. a, b, Effect of wild-type or harvested. For lipin1 localization, mouse primary hepatocytes were treated CRTC2(S136A) on lipogenic gene expression in liver (a) and plasma insulin with vehicle (Torin12) or 250 nM torin1 (Torin11) for 4 h. Scale bars, 10 mm. level (b) of fasted (3 h) and refed (1 h after 3 h fasting) mice. c–e, Effect of Data are shown as mean 6 s.e.m. *P , 0.01, n 5 8. NS, no significant CRTC2(S136A) and torin1 treatment on SREBP1 maturation (c), hepatic statistical difference. triglycerides (d) and lipin1 localization in mouse primary hepatocytes (e).

Extended Data Figure 8 | Enhanced SREBP1 activation, triglyceride levels subunits in HFD-fed mice (c). d–g, Hepatic triglyceride amounts and and branched-chain amino acid levels in obese mice. a–c, Immunoblots branched-chain amino acid (BCAA) levels in liver extracts from lean, showing relative amounts and/or phosphorylation status of SREBP1, SREBP2, db/db, ob/ob and HFD-fed mice in the fed state. Data are shown as SCAP, mTOR, S6K, CRTC2, AKT and COPII subunits in fed lean and db/db mean 6 s.e.m. *P , 0.01, n 5 10. mice (a), ob/ob mice (b), and relative amounts of SREBP2, SCAP and COPII

Extended Data Figure 9 | Improved insulin sensitivity in HFD-fed mice in lipogenic gene expression (c); glucose tolerance (d); insulin tolerance (e); the presence of CRTC2(DTAD/S136A). a–k, Effect of the mTOR-defective pyruvate tolerance (f); hepatic glucose production (HGP; g); insulin-stimulated mutant CRTC2(DTAD/S136A) on metabolic parameters (a), including body glucose disposal rate (IS-DGR; h); percentage of free fatty acid (FFA) weight, relative fat mass, food intake, plasma alanine aminotransferase (ALT) suppression (i); pAKT level in skeletal muscle (j); and pAKT level in epididymal and aspartate aminotransferase (AST) activity, plasma cholesterol, plasma white adipose tissue (k) from mice fed on a HFD for 18 weeks. Data are triglycerides, plasma insulin and blood glucose; energy expenditure (b); shown as mean 6 s.e.m. *P , 0.01, **P , 0.05, n 5 8(a–f), n 5 6(g–i).