Glucose-Mediated N-Glycosylation of SCAP Is Essential for SREBP-1 Activation and Tumor Growth

Article

Glucose-Mediated N-glycosylation of SCAP Is Essential for SREBP-1 Activation and Tumor Growth

Graphical Abstract Authors Chunming Cheng, Peng Ru, Feng Geng, ..., Jianjie Ma, Arnab Chakravarti, Deliang Guo

In Brief Cheng et al. show that glucose controls N-glycosylation of SCAP, which is essential for SREBP-dependent lipogenesis, and that activated growth factor receptors increase glucose intake and SREBP activation. Inhibition of SCAP N-glycosylation inhibits the growth of glioblastoma cells with hyperactived EGFR.

Highlights d SCAP acts as a glucose-responsive protein, linking fuel supply to SREBP activation d N-glycosylation of SCAP facilitates SCAP/SREBP trafﬁcking from ER to the Golgi d EGFR activates SCAP/SREBP by promoting glucose import and SCAP N-glycosylation d Targeting SCAP N-glycosylation is a promising strategy to treat malignancies

Cheng et al., 2015, Cancer Cell 28, 569–581 November 9, 2015 ª2015 Elsevier Inc. http://dx.doi.org/10.1016/j.ccell.2015.09.021 Cancer Cell Article

Glucose-Mediated N-glycosylation of SCAP Is Essential for SREBP-1 Activation and Tumor Growth

Chunming Cheng,1 Peng Ru,1 Feng Geng,1 Junfeng Liu,1,2 Ji Young Yoo,3 Xiaoning Wu,1 Xiang Cheng,1 Vanessa Euthine,4 Peng Hu,1 Jeffrey Yunhua Guo,1 Etienne Lefai,4 Balveen Kaur,3 Axel Nohturfft,5 Jianjie Ma,6 Arnab Chakravarti,1 and Deliang Guo1,* 1Department of Radiation Oncology, The Ohio State University Comprehensive Cancer Center, Columbus, OH 43210, USA 2College of Life Science and Technology, Beijing University of Chemical Technology, Beijing 100029, PR China 3Department of Neurosurgery, The Ohio State University Comprehensive Cancer Center, Columbus, OH 43210, USA 4CarMeN Laboratory, INSERM U1060, INRA 1397, Faculte´ de Me´ decine Lyon Sud BP 12, Universite´ de Lyon, 69921 Oullins Cedex, France 5Vascular Biology Research Centre, St. George’s University of London, London SW17 0RE, UK 6Department of Surgery, Davis Heart and Lung Research Institute, The Ohio State University Medical Center, Columbus, OH 43210, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.ccell.2015.09.021

SUMMARY

Tumorigenesis is associated with increased glucose consumption and lipogenesis, but how these pathways are interlinked is unclear. Here, we delineate a pathway in which EGFR signaling, by increasing glucose uptake, promotes N-glycosylation of sterol regulatory element-binding protein (SREBP) cleavage-activating protein (SCAP) and consequent activation of SREBP-1, an ER-bound transcription factor with central roles in lipid metabolism. Glycosylation stabilizes SCAP and reduces its association with Insig-1, allowing movement of SCAP/SREBP to the Golgi and consequent proteolytic activation of SREBP. Xenograft studies reveal that blocking SCAP N-glycosylation ameliorates EGFRvIII-driven glioblastoma growth. Thus, SCAP acts as key glucose-responsive protein linking oncogenic signaling and fuel availability to SREBP-dependent lipogenesis. Targeting SCAP N-glycosylation may provide a promising means of treating malignancies and metabolic diseases.

INTRODUCTION scribed from different transcriptional start sites, mainly regulate the expression of genes required for fatty acid synthesis. Elevated lipogenesis is a common patho-physiological charac- SREBP-2 is encoded by SREBF2 and is responsible for the teristic of cancer and metabolic diseases (Guo et al., 2013; synthesis of cholesterol (Goldstein et al., 2006; Horton et al., Menendez and Lupu, 2007; Moon et al., 2012; Ru et al., 2013; 2002, 2003). Recent evidence shows that the nuclear form of Tang et al., 2011). In these processes, a critical regulatory role SREBP-1 is highly upregulated in a variety of malignancies (Et- is played by sterol regulatory element-binding proteins tinger et al., 2004; Guo et al., 2009b; Li et al., 2014). Targeting (SREBPs), a family of transcription factors that control the SREBP-1 has become a promising therapeutic strategy to treat expression of genes important for the uptake and synthesis of cancer and other metabolic syndromes (Grifﬁths et al., 2013; cholesterol, fatty acids, and phospholipids (Goldstein et al., Guo et al., 2014; Kamisuki et al., 2009; Tang et al., 2011). 2006; Jeon and Osborne, 2012; Nohturfft and Zhang, 2009). SREBP-1 and SREBP-2 are synthesized as inactive precur- There are two mammalian SREBP genes, SREBF1 and SREBF2. sors bound to the membrane of the ER through two trans- SREBP-1a and -1c, which are encoded by SREBF1 with different membrane domains (Goldstein et al., 2006). SREBP activation N terminus (20 amino acids) owing to their mRNAs being tran- requires proteolytic release of an N-terminal fragment that

Signiﬁcance

Twenty years ago, Nobel Laureates Brown & Goldstein discovered SREBPs as key proteins that regulate lipid metabolism. They revealed that sterols modulate Insig interaction with SCAP/SREBP to inhibit SREBP function under physiological conditions. Our study identifies glucose as an indispensible component of SREBP function. Without glucose, reduction of sterols is insufficient to initiate SCAP/SREBP trafficking and SREBP activation. Glucose-mediated N-glycosylation of SCAP facilitates SCAP/SREBP movement from the ER to the Golgi and the subsequent lipogenic function of SREBP. We show that SCAP acts as a key glucose-responsible protein to integrate oncogenic signaling and fuel availability for control of lipid metabolism and tumor growth.

Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. 569 constitutes a basic helix-loop-helix transcription factor (Gold- the expression of SREBP-1 or SREBP-2 regulated genes stein et al., 2006; Wang et al., 1994). Currently, post-translational involved in lipid metabolism (Figure S1C). Using confocal fluo- activation of SREBPs is best understood in the context of cellular rescence imaging, we found that stimulation of U87 cells with cholesterol homeostasis (Radhakrishnan et al., 2008; Sun et al., glucose promoted nuclear translocation of transgenic GFP- 2007). When cholesterol level is high, it binds to SREBP-cleav- SREBP-1 (Figures 1D and S1D). Translocation of endogenous age activating protein (SCAP), inducing a conformational change SREBP-1 into the nucleus in response to glucose stimulation that promotes binding to ER-anchored insulin induced gene pro- was demonstrated by immunofluorescence (Figures 1E and tein (Insig), thus preventing Golgi transport and activation of S1E). In contrast, under glucose-deprived conditions, even with SREBPs (Adams et al., 2004; Sun et al., 2007; Yang et al., removal of sterols, SREBP-1 was still retained in the ER mem- 2002). When the cholesterol concentration drops, the SCAP/ brane as shown by its co-staining with protein disulfide isom- SREBP complex dissociates from Insig, allowing vesicular trans- erase (PDI), an ER membrane protein (Figures 1E and S1E) port to the Golgi where SREBPs are exposed to proteases that (Uehara et al., 2006). These data suggest that glucose is required release the transcriptionally active N-terminal fragment (Gold- for SREBP activation under conditions of sterol deprivation. stein et al., 2006; Nohturfft et al., 2000; Sun et al., 2007). SREBP stability, transport to the Golgi, and cleavage require Glucose is a major resource for de novo lipid synthesis. In can- formation of a complex between SREBP and SCAP (Nohturfft cer cells, elevated glucose consumption is often accompanied et al., 2000; Rawson et al., 1999; Sakai et al., 1998). We found by increased lipogenesis (Guo et al., 2013; Menendez and that glucose-induced cleavages of SREBP-1 and SREBP-2 Lupu, 2007), and the link between glucose supply and SREBP-1 were accompanied by an increase of SCAP protein levels in activation seems common in both physiological and patho- GBM cells (Figures 1F, S1A, and S1B). Similar results were physiological conditions (Guillet-Deniau et al., 2004; Hasty observed in various other cancer cell lines in which both et al., 2000; Horton et al., 1998; Kaplan et al., 2008). In a previous SREBP-1 and -2 were activated (see Figure S1F). HEK293T cells study, we showed that epidermal growth factor receptor (EGFR) expressing GFP-SCAP and full length FLAG-SREBP-1a, -1c, via PI3K/Akt signaling activated SREBP-1 in glioblastoma (GBM) or HA-SREBP-2 displayed glucose-dependent upregulation of cells (Guo et al., 2009b, 2011). While a number of studies have GFP-SCAP and activation of all three SREBP isoforms, which demonstrated that elevated EGFR signaling is coupled with were detected by increased N-terminal fragment of epitope- enhanced glucose uptake and lipogenesis in tumorigenesis (Ba- tagged SREBP in the nuclear fraction under conditions of bic et al., 2013; Cloughesy et al., 2014; Guo et al., 2009a, 2009b), glucose supplement (Figures 1G and S1G). We found that the molecular mechanisms that underlie the crosstalk between knock down of SCAP reduced glucose-mediated activation of the altered glucose and lipid metabolism in tumorigenesis SREBP-1 and SREBP-2 (Figure 1H). remain largely unknown. In this study, we test the hypothesis Because liver X receptor (LXR), a member of the nuclear re- that SCAP acts as a key glucose-responsive protein to integrate ceptor family of transcription factors, promotes the expression oncogenic signaling and fuel availability for modulation of of SREBP-1c upon activation (Repa et al., 2000), and glucose SREBP-dependent lipogenesis. is reported to activate LXR in hepatocytes (Mitro et al., 2007), we sought to examine whether LXR possibly involves in RESULTS glucose-activated SCAP/SREBP signaling. As shown in Figures S1A–S1C, our data show that both the protein and mRNA levels Glucose Activates SREBPs via Upregulation of SCAP of ATP-binding cassette activating proteins ABCA1 and ABCG1, To investigate how glucose and sterol signaling are integrated which are major downstream targets of LXR (Zelcer and Tonto- during SREBP activation, human GBM U87 cells that have noz, 2006), were not significantly upregulated by glucose sup- been shown to contain a high SREBP-1 activity (Guo et al., plement in U87 cells within 12 hr, demonstrating that LXR was 2009b, 2011) were grown in the absence or presence of glucose not strongly activated by glucose in GBM cells. In contrast, and sterols (cholesterol/25-hydroxycholesterol) and SREBP pro- SCAP protein and SREBP-1 and SREBP-2 cleavage were cessing was analyzed by western blot and immunofluorescence enhanced by glucose stimulation (Figures S1A and S1B). microscopy. We found that in the absence of glucose, only the Together, these data suggest that LXR was not involved in the ER-bound SREBP-1 precursor could be clearly detected, even glucose-mediated SREBP signaling pathway. in cells that were deprived of sterols by removal of serum (Fig- Collectively, these data demonstrate that glucose availability ure 1A). While glucose supplement modestly activated SREBP-1 constrains SREBP activity by controlling the levels of SCAP. in the presence of sterols (as assessed by the appearance of the N-terminal cleavage fragment of SREBP-1), maximal SREBP-1 Glucose Promotes SCAP N-glycosylation and Enhances and SREBP-2 cleavage required both glucose supplement and Its Stability low sterols in the absence or presence of serum (Figures 1A To explore the mechanisms of how glucose enhances SCAP and S1A). Moreover, glucose activated SREBP-1 and SREBP-2 protein levels and activates SREBP, the intermediate metabolite, in a dose- and time-dependent manner and increased the protein pyruvate or lactate of the glycolysis pathway, or N-acetylglucos- levels of fatty acid synthase (FASN) and low-density lipoprotein amine (GlcNAc) of the hexosamine biosynthetic pathway (HBP) receptor (LDLR), the genes encoding both of which are down- (Figure 2A) was added to GBM U87 cells in glucose- and ste- stream targets for SREBP-1 (Figures 1B, 1C, S1A, and S1B) (Ben- rol-free medium, respectively. The data showed that GlcNAc nett et al., 1995; Yokoyama et al., 1993). Furthermore, real-time was as effective as glucose at enhancing SCAP protein levels quantitative (q)PCR analysis showed that glucose stimulation and promoting SREBP-1 cleavage, while pyruvate and lactate enhanced both SREBP-1a and -1c expression and activated had no effect (Figure 2B). Upon exposure of cells to an inhibitor

570 Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. - AB+ - + Glucose 0 0.5 1 5 25 mM Glucose C0 2 4 8 12 24 hr Glucose + + - - Sterols -P -P - P SREBP-1 SREBP-1 SREBP-1 -N -N -N FASN FASN FASN LDLR LDLR LDLR -P - P -P SREBP-2 SREBP-2 SREBP-2 -C -C - C

β-actin β-actin β-actin

D E SREBP-1 PDI DAPI Merge - Glucose + Glucose p < 0.0001 p < 0.0001 - Glucose + Glucose

- - + + Glucose F G SCAP - + Glucose PDI SCAP - + - + - + Glucose + + + + + + GFP-SCAP - P SREBP-1 PDI GFP-SCAP - N - PDI P SREBP-2 SREBP-1 N - C Lamin A FASN Lamin A Nuclear form LDLR β-actin

Figure 1. Glucose Activates SREBPs via Upregulation of SCAP (A) Western blot analysis of U87 cells cultured in serum-free media with or without glucose (5 mM) in the presence or absence of sterols (10 mg/ml 25-hydroxycholesterol and 10 mg/ml cholesterol) for 12 hr. (B and C) Western blot analysis of U87 cells cultured in serum-free media with different dose of glucose for 12 hr (B) or stimulated with 5 mM glucose at indicated times (C). (D and E) Confocal microscopy images show GFP-SREBP-1 (green), which was derived from a cDNA lacking the exon 1 of SREBP-1 (encodes the identical aa sequence between SREBP-1a and -1c) (D) or endogenous SREBP-1 (E, red) subcellular localization in U87 cells cultured in serum-free media with or without glucose (Gluc) for 12 hr. The nuclei were stained by DAPI (blue). The scale bars represent 10 mm. The nucleus intensity of GFP-SREBP-1 (D) or endogenous SREBP-1 (red) (E) was quantified over 30 cells by using ImageJ. The red lines in the quantification graphs show mean ± SEM (n = 30). The significance was determined by an unpaired Student’s t test. (F) Western blot analysis of membrane and nuclear extracts from U87 cells cultured in serum-free media with or without glucose (5 mM) for 12 hr. (G and H) Western blot analysis of membrane, nuclear extracts, or total cell lysates from HEK293T cells co-transfected with GFP-SCAP or GFP vector and full length (FL) FLAG-SREBP-1a, -1c, or HA-SREBP-2 plasmids (G) or from U87 cells after knock down of SCAP in comparison with shRNA control (shCtrl) in serum- free media with or without glucose (5 mM) for 12 hr (H) (precursor of SREBPs, P; N terminus of SREBP-1, N; and C terminus of SREBP-2, C). See also Figure S1. of HBP (azaserine, AZA) or to an inhibitor of N-glycan synthesis and SREBP-1 and SREBP-2 cleavages in U87 cells (Figure S2F). (tunicamycin, Tuni), both SCAP protein levels and SREBP-1 These data suggest that the effects of glucose on SCAP cleavage were reduced, while an inhibitor of O-glycosylation and SREBP were likely mediated by an induction of protein (BADGP) had no effect (Figures 2C and S2A–S2D). Immunofluo- N-glycosylation. rescence imaging showed that glucose-induced Golgi and Nohturfft et al. (1998) showed that SCAP protein carries nuclear translocation of SREBP-1 were largely blocked by AZA three N-linked oligosaccharides at asparagine (N) positions and Tuni (Figures 2D and S2E). As expected, the addition of N263, N590, and N641 and mutations of one or two of these GlcNAc restored AZA treatment-reduced SCAP protein levels asparagines had no apparent effect on the function of SCAP.

Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. 571 A BCD

E F

H I J K

Figure 2. Glucose Promotes SCAP N-glycosylation and Enhances Its Stability (A) Schematic model shows the glucose metabolism divided into glycolysis, de novo lipid synthesis, oxidative pentose phosphate pathway (PPP), and hexosamine synthesis pathway (HBP) for glycosylation modification. The key enzymes controlling HBP and glycosylation modifications and their specific inhibitors are shown in the scheme (pyruvate, Pyr; lactate, Lac; TCA; fructose-6-phosphate, Fruc-6-P; GlcNAc; AZA; Tuni; fructose-6-phosphate amidotransferase, GFAT; dolichyl-phosphate, UDP-GlcNAc, GlcNAc-1-P transferase, DPAGT1; and O-glycosylation transferase, OGT). (B and C) Western blot analysis of U87 cell lysates, which were derived from cells cultured in serum-free media with or without glucose (5 mM), pyruvate (10 mM), lactate (10 mM), or GlcNAc (20 mM) (B) or in combination with AZA (100 mM), Tuni (1 mg/ml), or BADGP (1 mM) for 12 hr (C). (D) Confocal microscopy images of SREBP-1 subcellular localization in relation with the Golgi protein marker Giantin and nuclear DAPI staining in U87 cells with same treatment procedure as (C). The scale bars represent 10 mm. (E) Western blot analysis of cell membrane fractions for protease-protected N-glycosylation fragment (aa 540–707) of SCAP (upper) or total GFP-SCAP protein levels (lower) from HEK293T cells transiently transfected with GFP-SCAP for 24 hr and then cultured in serum-free media with or without glucose (5 mM) or GlcNAc (20 mM) for another 12 hr. The numbers on the left side of the blot indicate the number of N-glycosylation on SCAP protein (upper) (Nohturfft et al., 1998). For details, please see Supplemental Information. (F and G) Western blot analysis of SCAP N-glycosylation (upper) or GFP-SCAP protein levels (lower) from HEK293T cells transiently transfected with GFP-SCAP for 24 hr and then cultured in serum-free media with or without GlcNAc (20 mM) or glucose (5 mM) in the presence or absence of AZA (100 mM) (F) or with or without glucose in combination with Tuni (1 mg/ml) (G) for 12 hr. (H) Western blot analysis of cell membrane fractions from U87 cells cultured in serum-free media with or without glucose (5 mM) in combination with Tuni(1mg/ml) and protease inhibitor E64D (10 mg/ml)/Pepstatin A (10 mg/ml)/Leupeptin (10 mg/ml) or proteasome inhibitor MG132 (50 mM) for 12 hr. (I) Western blot analysis of cell membrane fractions from U87 cells cultured in serum-free media with or without glucose (5 mM) in the presence or absence of MG132 (50 mM) for 6 hr. (J and K) Western blot analysis of cell membrane fractions from HEK293T cells transfected with indicated amounts of plasmid expressing the wild-type GFP- SCAP (NNN) or mutant GFP-SCAP (QQQ) for 24 hr (J), then treated with or without MG132 (50 mM) for another 4 hr (K). The levels of GFP-SCAP-NNN or -QQQ were quantified by ImageJ and normalized with PDI. The results are shown as mean ± SD (n = 3). The significance was determined by an unpaired Student’s t test (*p < 0.01). See also Figure S2.

572 Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. However, a version of SCAP with all three asparagines mutated Glucose-Mediated Glycosylation Promotes SCAP to glutamine (NNN to QQQ) could not be detected, suggesting Trafficking to the Golgi Leading to SREBP Activation that N-glycosylation is important for the stability of SCAP (Noh- We wondered whether N-glycosylation altered the association of turfft et al., 1998). Therefore, we asked whether glucose might SCAP with Insig and thereby promoted SCAP/SREBP trafficking control the levels of SCAP by affecting its glycosylation. Further- to the Golgi in the absence of sterols. To address this, HEK293T more, the role for N-glycosylation in SCAP/SREBP function has cells were co-transfected with Myc-Insig-1 and GFP-SCAP or not been extensively explored. GFP-vector to determine their interaction in response to glucose We adopted the approach of Nohturfft et al. (1998) and exam- and Tuni treatment. We found that glucose treatment reduced ined the effects of glucose on SCAP N-glycosylation in HEK293T the association of GFP-SCAP with Myc-Insig-1 compared with cells and CHO cells. SCAP contains a luminal region (amino acid, a glucose-deprived condition (Figure 3A). Further, blocking aa, 540–707) with two N-glycosylation sites that is protected N-glycosylation by Tuni restored the binding of GFP-SCAP to from proteolysis when intact membranes are treated with trypsin Myc-Insig-1 (Figure 3A). These data suggest that N-glycosyla- (Nohturfft et al., 1998). This luminal fragment has a molecular tion modification by glucose reduced the binding of SCAP to weight of 30 kDa that is small enough to allow the resolution Insig-1. To test if this affected the SCAP/SREBP trafficking to of individual glycosylated variants of SCAP by SDS-PAGE (Fig- the Golgi, the subcellular distribution of GFP-SCAP was exam- ures 2E, upper, and S2G). Glucose deprivation for 12 hr resulted ined by confocal microscopy. The data show that glucose stim- in two weaker bands that correspond to SCAP fragments car- ulation promoted the GFP-SCAP trafficking to the Golgi, which rying one or two oligosaccharides in HEK293T cells expressing was blocked by inhibition of N-glycosylation with Tuni (Figures GFP-SCAP (Figure 2E, upper), indicating that the level of SCAP 3B and S3A). N-glycosylation was decreased under a low glucose condition Data with transient expression of GFP-SCAP-QQQ mutant that was correlated with reduced total GFP-SCAP protein (Fig- and Myc-Insig-1 in HEK293T cells further supported the notion ures 2E, lower, and S2H). Removal of the oligosaccharides by that N-glycosylation reduces SCAP binding to Insig-1 and sub- treatment with endoglycosidase (PNGase) reduced the apparent sequent SCAP/SREBP trafficking to the Golgi. We found that molecular weight to the unglycosylated form of SCAP (Nohturfft GFP-SCAP-QQQ displayed stronger association with Myc- et al., 1998). Notably, in the presence of glucose, high amounts Insig-1 compared with GFP-SCAP-NNN (Figure 3C). While of fully glycosylated forms of SCAP and total GFP-SCAP GFP-SCAP-NNN promoted trafficking of FLAG-SREBP-1c to protein were detected (Figures 2E, S2G, and S2H), which were the Golgi and subsequent translocation to the nucleus, the associated with elevated SREBP activation (Figures 1F, 1G, GFP-SCAP-QQQ mutant failed to do so (Figures 3D and S3B). and S1G). Supplementing the culture medium with GlcNAc The contribution of individual N-glycosylation sites to SCAP enhanced SCAP glycosylation and increased the total GFP- function and trafficking was further investigated in HEK293T and SCAP protein to a level similar to that found upon glucose treat- U87 cells expressing the GFP-SCAP protein harboring single, ment (Figure 2E). Furthermore, AZA or Tuni treatment blocked double, or triple mutations. Our data showed that none of the sin- glucose-promoted SCAP N-glycosylation and reduced GFP- gle or double mutations in SCAP impaired the trafficking and acti- SCAP levels in HEK293 cells and endogenous SCAP levels in vation of FLAG-SERBP-1c in response to glucose stimulation CHO cells (Figures 2F, 2G, S2G, and S2H), and the addition of (Figures 3E–3G, S3C, and S3D). These data were consistent GlcNAc restored AZA treatment-mediated reduction of SCAP with the early studies by the Brown and Goldstein Laboratory, N-glycosylation and total GFP-SCAP protein to levels similar to which demonstrated that mutations of one or two N-glycosylation cells exposed to glucose supplement alone (Figure 2F). In sites produced negligible effects on SREBP-2 activation (Nohturfft GBM cells treated with Tuni, the endogenous SCAP protein level et al., 1998). Interestingly, we found that the triple mutant GFP- was reduced (Figures 2H and S2C). Co-treatment with MG132, a SCAP-QQQ was unable to traffic to the Golgi and failed to activate proteasome inhibitor, but not protease inhibitors (combination of epitope-tagged SREBP-1a, -1c, and SREBP-2 (Figures 3F, 3G, E64d, pepstatin A, and leupeptin), restored the SCAP protein to a S3C, and S3E–S3G). Thus, even though maintenance of a single level similar to that when cells were exposed to glucose alone glycosylation site on SCAP appeared to be sufficient for its biolog- (Figures 2H and 2I). Taken together, these data demonstrate ical function (Nohturfft et al., 1998), complete abolition of all glyco- that glucose deprivation-led reduction of SCAP protein levels sylation sites on SCAP disrupted the exit pathway for the SCAP/ is caused by proteasome-dependent degradation. SREBP complex from ER to the Golgi. To examine the effect of N-glycosylation on SCAP stability, we Elegant studies by Brown and Goldstein Laboratory estab- generated a mutant form of SCAP by replacing all three aspara- lished that sterols are critical factors in the regulation of SCAP gines (N263, N590, and N641) with glutamines (NNN to QQQ). trafficking and SREBP activation (Goldstein et al., 2006). We con- We added a GFP tag to the N-terminal end of SCAP and used ducted similar studies and found that the incubation of cells with a strong CMV promoter to drive its expression in HEK293T cells. sterols could block glucose-mediated SCAP trafficking and This enabled us to produce a measurable amount of GFP-SCAP- SREBP-1 nuclear translocation (Figures S3H and S3I). Overex- QQQ protein, overcoming the problem Nohturfft et al. (1998) pressing Myc-Insig-1 or increasing sterol concentration reduced faced. Compared with the wild-type protein (GFP-SCAP-NNN), glucose-mediated SREBP-1 activation (Figures S3J and S3K). lower levels of GFP-SCAP-QQQ were detected (Figure 2J), sug- While these data are consistent with the early work of Brown gesting that the unglycosylated SCAP proteins are less stable and Goldstein Laboratory, the present findings suggest that and presumably are more susceptible to proteasomal degrada- glucose is indispensable for SCAP/SREBP trafficking and function. Indeed, treatment of cells with MG132 significantly restored tion. Our data support that N-glycosylation relieves SCAP asso- GFP-SCAP-QQQ protein levels (Figure 2K). ciation with Insig, leading to SCAP/SREBP trafficking to the Golgi

Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. 573 AB C GFP-SCAP Giantin Merge

- - - -Gluc + Tuni IP: GFP Myc-Insig-1 + - + + Glucose IP: GFP Myc-Insig1 WB GFP-SCAP

WB GFP-SCAP + Gluc GFP GFP GFP-SCAP

GFP-SCAP + Gluc Input GFP +Tuni Myc-Insig-1 Input GFP Myc-Insig1

D E Flag- GFP-SCAP SREBP-1c DAPI Merge

NNN GFP-SCAP p < 0.0001 - + - + - + - + - + - + - + - + -+PNGase 2 1 Protease-protected fragment of SCAP QQQ 0

F G

NNN QNN NQN NNQ QNQ QQN NQQ QQQ GFP-SCAP

PDI GFP-SCAP Flag-SREBP-1c N

Giantin GFP

Lamin A Merge

Figure 3. N-glycosylation Promotes SCAP Trafficking to the Golgi Leading to SREBP-1 Activation (A) Total (Input) or immunoprecipitated (IP) lysates from HEK293T cells transfected with GFP-SCAP or GFP vector together with Myc-Insig-1 without or with glucose (5 mM) stimulation in the presence or absence of Tuni (1 mg/ml) for 12 hr were immunoblotted using the indicated antibodies. (B) Confocal microscopy images show GFP-SCAP trafficking related to the Golgi marker Giantin in response to glucose (5 mM) in the presence or absence of Tuni (1 mg/ml) in U87 cells for 12 hr. The scale bars represent 10 mm. (C) Immunoblot analysis of total or immunoprecipitated lysates from HEK293T cells transfected with wild-type GFP-SCAP (NNN) or mutant GFP-SCAP (QQQ) together with Myc-Insig-1 in response to glucose (5 mM) for 12 hr using the indicated antibodies. (D) Confocal microscopy images show SCAP and SREBP-1 trafficking in U87 cells transfected with wild-type GFP-SCAP (NNN) or mutant GFP-SCAP (QQQ) together with FLAG-SREBP-1c (full length, FL) in response to glucose (5 mM) for 12 hr. The scale bars represent 10 mm. The nucleus intensity of FLAG-SREBP-1c was quantified over 30 cells by using ImageJ. The results are shown as mean ± SEM (n = 30). The significance was determined by an unpaired Student’s t test. (E) Western blot analysis of the N-glycosylation levels of SCAP mutant in HEK293T cells transfected with different mutant of GFP-SCAP in comparison with wild- type (NNN). (F) Confocal microscopy images show the trafficking of different SCAP mutants in U87 cells transfected with wild-type (NNN) or different mutant GFP-SCAP under glucose (5 mM) treatment in serum-free media. The scale bars represent 10 mm. (G) Western blot analysis of membranes and nuclear extracts from HEK293T cells transfected with wild-type or different mutant GFP-SCAP together with FLAG- SREBP-1c (full length) in response to glucose (5 mM) stimulation in serum-free media. See also Figure S3. and subsequent SREBP activation; and sterols could antagonize Based on our observation that SREBP activation in sterol- glucose function on SCAP trafficking and SREBP activation. deprived conditions was fully dependent on glucose (Figure 1), we tested whether EGFR-mediated SREBP-1 activation also EGFR Signaling Activates SREBP-1 by Enhancing the required glucose. As shown in Figure 4A, EGF stimulation was Glucose Uptake, SCAP Protein Levels, and N- unable to promote the nuclear translocation of SREBP-1 in the glycosylation absence of glucose and the addition of glucose could restore EGFR signaling has been shown to enhance glucose uptake and EGF-mediated SREBP-1 nuclear translocation. Western blotting activate SREBP-1 (Babic et al., 2013; Guo et al., 2009a, 2009b). showed that EGF stimulation had no effect on SREBP-1 cleavage

574 Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. Figure 4. EGFR Signaling via Promoting ABSREBP-1 DAPI Merge Glucose Uptake Upregulates SCAP and - - + + EGF Activates SREBP-1 - + - + Glucose (A and B) Confocal microscopy images of -P SREBP-1 subcellular distribution (A) or western Glucose SREBP-1 -N blot analysis of SREBP-1 cleavage (B) in U87/ EGFR cells in response to EGF (50 ng/ml) in the p-EGFR Y1086 presence or absence of glucose (5 mM) for 12 hr. m EGFR The scale bars represent 20 m. The SREBP-1 nucleus intensity was quantified over 30 cells by p-Akt Thr308 using ImageJ. The results are shown as mean ± +Glucose - SEM (n = 30). The significance was determined by p-Akt Ser473 an unpaired Student’s t test. Akt (C) Glucose uptake analysis of U87/EGFR cells in response to EGF (50 ng/ml) by using 14C-glucose p-S6 Ser235/236 isotope. The results are shown as mean ± SD (n = 3). The significance was determined by an S6

- Glucose/EGF unpaired student t test. β-actin (D) Western blot analysis of cell membrane and nuclear extracts from U87/EGFR cells in response to EGF (50 ng/ml) stimulation in the presence or absence of glucose (5 mM) for 12 hr. (E) Western blot analysis of U87/EGFR cells after knock down of SCAP in comparison with shRNA +Glucose/EGF control (shCtrl) in response to EGF (50 ng/ml) p < 0.0001 E stimulation in the presence of glucose (5 mM) p < 0.0001 for 12 hr. --+ + EGF + + + + Glucose See also Figure S4. SCAP

PDI SREBP-1 P S4C), suggesting that the activation C D SREBP-1 N of the glucose-SCAP-SREBP signaling Lamin A pathway could be induced by various 100 p < 0.05 - - + + EGF - + - + Glucose other growth factors. We then knocked 75 p-EGFR Tyr1086 SCAP down the expression of SCAP in U87/

cells/minute) 50 5 EGFR EGFR cells in order to examine whether 25 PDI Glucose uptake the effect of EGF on SREBP-1 activation 0 p-Akt Thr308 (CPM/10 SREBP-1 N was mediated by the glucose-SCAP

EGF p-Akt Ser473 Lamin A axis. As shown in Figure 4E, knock down Control Akt of SCAP completely abolished EGF- mediated activation of SREBP-1. β-actin Inhibition of the HBP pathway by AZA or suppression of N-glycosylation by Tuni blocked EGF-mediated nuclear in glucose-free media, even though the EGFR-PI3K-Akt-mTOR translocation of SREBP-1, whereas the O-glycosylation inhibitor signaling pathway was strongly activated as evidenced by the BADGP treatment failed to do so (Figure 5A). Western blotting upregulation of p-EGFR, p-Akt, and p-S6 (Figures 4B and S4A). showed that AZA or Tuni treatment blocked EGFR signaling- In contrast, EGF stimulation in the presence of glucose promoted dependent increase in SREBP-1 cleavage, whereas no changes the cleavage of SREBP-1 (Figures 4B and S4A). were observed by BADGP treatment (Figure 5B). Moreover, Using 14C-glucose radioisotope measurements, we found that EGF-mediated upregulation of SCAP was reduced by AZA or EGF stimulation led to the significant increase of glucose uptake Tuni treatment, but not by the BADGP treatment (Figure 5B). in U87/EGFR cells (Figure 4C). While EGF stimulation did not We incubated U87/EGFR cells with 14C-glucose and measured appear to affect the SCAP protein level in the absence of isotope labeled lipid products and found that AZA or Tuni treat- glucose, the additive effect of glucose and EGF-dependent ment significantly reduced EGFR signaling-dependent increase enhancement of SCAP expression was observed (Figures 4D in de novo lipid synthesis (Figure 5C). These results demonstrate and S4A). Moreover, the EGF-induced changes in SCAP expres- that N-glycosylation was a key mediator for EGFR signaling- sion were correlated with SREBP-1 activation, as reflected by mediated increase in SCAP level and subsequent activation of the increased appearance of the SREBP-1 N-terminal fragment SREBP-1. in the nuclear fraction (Figures 4D and S4A). Furthermore, the in- We then assessed the effect of EGF stimulation on SCAP crease in glucose uptake by insulin in HepG2 cells also upregu- glycosylation. Stimulation of cells with EGF led to the increase lated SCAP protein and activated SREBP-1 (Figures S4B and of the total and the N-glycosylated GFP-SCAP protein when

Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. 575 A B C

Figure 5. EGFR Signaling Activates SREBP-1 via Upregulation of SCAP N-glycosylation (A and B) Confocal microscopy images of SREBP-1 translocation (A) or western blot analysis of SCAP levels and SREBP-1 cleavage (B) in U87/EGFR cells cultured in serum-free media in response to EGF (50 ng/ml) in the presence or absence of AZA (100 mM), Tuni (1 mg/ml), or BADGP (1 mM) with glucose (5 mM) for 12 hr. The scale bars represent 20 mm. The SREBP-1 nucleus intensity was quantified over 30 cells by using ImageJ. The results are shown as mean ± SEM (n = 30). The significance was determined by an unpaired Student’s t test. (C) De novo lipid synthesis analysis in U87/EGFR cells cultured in serum-free media treated with or without EGF (50 ng/ml) in combination with AZA (100 mM) or Tuni (1 mg/ml) for 12 hr, then adding 0.5 mCi 14C-labeled glucose for 2 hr, followed by lipid extraction and analysis by scintillation counter. The results are presented as mean ± SD (n = 3). The significance was determined by an unpaired student t test (*p < 0.05 and **p < 0.01). (D) Western blot analysis of total protein (upper) and N-glycosylation levels (lower) of GFP-SCAP and SREBP-1 nuclear form (upper) in LN229/EGFR cells transiently expressing GFP-SCAP after the stimulation of EGF (50 ng/ml) in the presence or absence of Tuni (1 mg/ml) for 12 hr. (E) Western blot analysis of U87/EGFR cells transfected with GFP, wild-type GFP-SCAP (NNN), or triple mutant GFP-SCAP (QQQ) after the stimulation of EGF (50 ng/ml) in serum-free media for 12 hr. compared with control and both were blocked by Tuni (Fig- Impairment of SCAP N-glycosylation Suppresses GBM ure 5D). In correlation, EGF stimulation also upregulated Tumor Growth SREBP-1 activity, which was inhibited by Tuni (Figure 5D, up- To determine the effect of EGFR on SCAP/SREBP-1 function in per). Similar to the studies in Figure 3G, we measured the effect tumorigenesis, we employed a cell culture model with stable of EGF on SREBP-1 function in U87/EGFR cells expressing GFP- expression of EGFRvIII, a constitutively active form of EGFR SCAP-NNN or GFP-SCAP-QQQ. We found that cells expressing with enhanced activation of PI3K-Akt signaling that leads to GFP-SCAP-NNN showed enhanced SREBP-1 activation in aggressive tumorigenesis (Guo et al., 2009b; Huang et al., response to EGF stimulation, whereas cells expressing GFP- 1997). Western blot revealed that U87/EGFRvIII cells displayed SCAP-QQQ showed reduced SREBP-1 activation even below elevated SCAP protein levels and enhanced SREBP-1 activation the levels observed with cells expressing GFP alone (as control) associated with upregulated p-Akt and p-S6 levels (Figure 6A). (Figure 5E). Interestingly, biochemical studies showed that the Based on this observation, we asked whether EGFRvIII-medi- other signaling components for EGFR, e.g., p-EGFR, p-AKT, ated tumorigenesis is mediated by enhanced SCAP expression and p-S6, were not affected by overexpression of either GFP- and its glycosylation. SCAP-NNN or GFP-SCAP-QQQ in U87/EGFR cells (Figure 5E). We established a stable U87/EGFRvIII cell line with knock down Taken together, our data show that EGFR-dependent SREBP-1 of SCAP (U87/EGFRvIII-shSCAP). Western blotting showed that activation is mediated by upregulation of SCAP and its N-glyco- knock down of SCAP reduced SREBP-1 activation (Figure 6B) sylation through enhancing glucose uptake. and which significantly reduced tumor growth in mouse flanks

576 Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. ABC D

E F

G H

Figure 6. Impairment of SCAP N-glycosylation Suppresses Tumor Growth and Significantly Prolongs Overall Survival of GBM-Bearing Mice (A) Western blot analysis of SCAP levels and SREBP-1 cleavage in U87 versus U87-EGFRvIII cells. (B–D) The effects of shRNA-mediated knock down of SCAP in U87/EGFRvIII cells analyzed by western blot (B), subcutaneous tumor growth (1 3 106 cells/ mouse), which are shown as mean ± SEM (n = 8) (C), or overall survival of intracranial tumor bearing-mouse (1 3 105 cells/mouse) assessed by Kaplan-Meier analysis (D). (E–H) The effects of wild-type GFP-SCAP (NNN) or mutant GFP-SCAP (QQQ) compared with the control GFP expression in U87/EGFRvIII cells analyzed by western blot (E), subcutaneous tumor growth (1 3 106 cells/mouse), which are shown as mean ± SEM (n = 5) (F), luminescence imaging of intracranial tumor (1 3 105 cells/mouse) at day 14 after implantation, which are shown as mean ± SEM (n = 7) (G), or GBM bearing-mouse overall survival assessed by Kaplan-Meier analysis (H). The significance was determined by an unpaired Student’s t test (*p < 0.05 and **p < 0.01). The survival was analyzed by log rank test. compared with the short hairpin (sh)RNA control group (Figure 6C). moted GBM tumor growth compared with the GFP control, To assess the potential of these modified GBM cells to form ortho- whereas GFP-SCAP-QQQ expression significantly reduced the topic tumors, shSCAP or shControl cells were implanted into GBM tumor growth. mouse brain. A Kaplan-Meier plot showed that reducing the Finally, to determine the effect of SCAP N-glycosylation on SCAP level significantly prolonged the overall survival of GBM- GBM intracranial tumor growth, we used a U87/EGFRvIII cell bearing mice (Figure 6D). line with stable expression of luciferase, which allows the visual- We next examined whether the impairment of SCAP N-glyco- ization of a tumor in mouse brain using luminescence imaging sylation had any effect on GBM tumorigenesis. For this purpose, (Wojton et al., 2013). The data showed that the expression of GFP-SCAP-NNN, GFP-SCAP-QQQ, or GFP (as control) was GFP-SCAP-NNN promoted tumor growth, whereas the expres- transiently expressed in U87/EGFRvIII cells. The data showed sion of GFP-SCAP-QQQ significantly reduced tumor growth that the expression of GFP-SCAP-NNN enhanced, whereas (Figure 6G). Moreover, overexpression of GFP-SCAP-NNN GFP-SCAP-QQQ reduced SREBP-1 activation, compared with reduced the overall survival of GBM-bearing mice (Figure 6H). cells transfected with GFP (Figure 6E), suggesting that GFP- Inversely, impairment of SCAP N-glycosylation significantly pro- SCAP-QQQ had a dominant negative effect on SREBP-1 activa- longed the overall survival of mice (Figure 6H). Collectively, these tion. Next, these transfected cells were implanted into mouse data demonstrate that N-glycosylation on SCAP is a critical fac- flanks. As shown in Figure 6F, GFP-SCAP-NNN expression pro- tor for EGFRvIII-induced GBM tumorigenesis.

Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. 577 Figure 7. Schematic Diagram Illustrating EGFR Signaling via Enhancing Glucose Uptake Promotes SCAP N-glycosylation Leading to SREBP-1 Activation (A) Signaling cascade for EGFR stimulation and glucose uptake to SCAP N-glycosylation and SREBP-1 activation. (B) Without glucose, unglycosylated SCAP binds to Insig, leading to retention of SREBP in the ER. Under this condition, changes in sterol level are insufficient to trigger exit of the SCAP/SREBP complex from the ER. In the presence of glucose, N-glycosylation of SCAP leads to increased stability of SCAP and its dissociation from Insig. Thus, the SCAP/SREBP complex can traffic to the Golgi, resulting in SREBP cleavage and its nuclear function. The sterols block glucose- mediated SCAP/SREBP trafficking and the subsequent SREBP activation. The elevated EGFR signaling causes increased glucose uptake and SCAP/SREBP activation to promote tumor growth.

DISCUSSION negative regulators for SCAP/SREBP trafficking (Goldstein et al., 2006), and an 5% decrease of cholesterol in the ER membrane Malignant cells, in general, have high rates of de novo lipid syn- is sufficient to trigger the activation of SREBP (Radhakrishnan thesis (Currie et al., 2013; Guo et al., 2013; Menendez and Lupu, et al., 2008). Such a tight regulation mechanism necessitates 2007). Since solid tumors often reside in an environment with the participation of additional cellular factors to coordinate fluctuant nutrient supply (Ackerman and Simon, 2014; Gullino with sterols for the fine tuning of SREBP function in normal and et al., 1967; Hirayama et al., 2009), they must develop adaptive patho-physiological conditions. Our study unravels the impor- mechanisms to preserve energy for maintenance of tumor cell tant function of glucose in controlling lipid metabolism in tumor- survival under conditions of low glucose supply. Proper gauging igenesis, linking EGFR signaling to glucose uptake and SCAP/ of glucose levels and lipid synthesis is not only critical for tumor SREBP activation (Figure 7). growth and survival, but also for normal cell function under stress Activation of SREBP relies on SCAP-escorted trafficking from conditions. In this study, we demonstrate that SCAP acts as a ER to the Golgi (Goldstein et al., 2006). Sterols serve as negative key glucose-responsive protein to orchestrate fuel availability regulators enhancing SCAP binding to Insig to retain the SCAP/ and sterol levels for control of SREBP function in lipid synthesis. SREBP complex in ER membrane (Goldstein et al., 2006). In this We show that glucose controls N-glycosylation of SCAP, which study, we show that in glucose-deficient conditions, removal of is indispensable for SCAP/SREBP trafficking from ER to the sterols was unable to relieve SCAP binding to Insig-1 and trigger Golgi and for the nuclear activation of SREBPs (Figure 7). This exit of SCAP/SREBP from ER. A supplement of glucose pro- finding renovates our current understanding of the regulation moted SCAP N-glycosylation and SCAP/SREBP trafficking to of SREBP function in physiology and patho-physiology. Elegant the Golgi and consequent SREBP activation. We show that the studies by Brown and Goldstein demonstrated that sterols act as interaction between SCAP and Insig-1 is reduced when SCAP

578 Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. is glycosylated. Studies by Brown and Goldstein have shown SCAP N-glycosylation-SREBP-1 activation metabolic pathway that Insig is a sterol sensor that negatively regulates SCAP inter- enhances lipogenesis and promotes rapid tumor growth. Once action with SREBP (Goldstein et al., 2006). Our data are consis- the glucose level drops, SCAP loses N-glycosylation and discon- tent with their idea, as we show that even in the presence of nects oncogenic signaling with SREBP-1, leading to reduced glucose, sterol still exerts its control of SCAP/SREBP trafficking lipogenesis and thereby preserving limited energy sources and subsequent SREBP activation. Thus, glucose functions as a for tumor cell survival. These data demonstrate the plasticity prerequisite activator for control of SCAP trafficking and SREBP and survival capability of tumor cells in facing harsh nutrient activation. Since sterol can still block SCAP/SREBP signaling microenvironment. complex, future studies will be required to examine how glucose Complete understanding of the molecular mechanisms of communicates with sterol in orchestrating the crosstalk between SCAP/SREBP trafficking and its relationship to fuel supply and energy supply and lipid metabolism in physiology and cancer growth signaling may help design effective therapeutic strate- biology. gies to target lipid metabolism in cancer and other metabolic There are three N-glycosylation sites on SCAP, one located syndromes. in loop 1 (N263) and two in loop 7 (N590 and N641) and both loops reside in the inner lumen of ER (Nohturfft et al., 1998). EXPERIMENTAL PROCEDURES Interestingly, mutation in any one or two sites of N-glycosylation was unable to affect SCAP trafficking and SREBP function. More detailed procedures were described in the Supplemental Information. This observation was consistent with a previous report by Plasmids Nohturfft et al. (1998), who demonstrated that, as long as one GFP-SCAP wild-type plasmid is a gift from Dr. Peter Espenshade. Original N-glycosylation site is present on SCAP, its control of plasmids for SCAP mutants were provided by Drs. Nohturfft, Brown, and Gold- SREBP-2 is unperturbed. Here, we tested the hypothesis that stein (Nohturfft et al., 1998). The 2 3 HA-SREBP-2 (full length) plasmid is a gift the presence of any single glycosylation site on SCAP is suffi- from Dr. John Shyy. Adenovirus expressing GFP-SREBP-1 (full length) was cient for SCAP/SREBP trafficking and activation. We found that produced and amplified as described (Dif et al., 2006). even though presence of a single glycosylation site on SCAP Cell Culture and Transfection appeared to be sufficient for its biological function, complete U87, U87EGFR, U87EGFRvIII (Guo et al., 2009b), HEK293T, U87EGFRvIII-luc, abolition of all glycosylation sites on SCAP disrupted the exit and other cancer cell lines were maintained in Dulbecco’s modified Eagle’s pathway for the SCAP/SREBP complex from ER to the Golgi. medium (DMEM; Cellgro) supplemented with 5% HyClone fetal bovine serum This further supports the notion that specific recognition of (FBS; Thermo Scientific). All cell cultures were supplemented with 1% peni- the glycosylation motif may provide a mechanism for modula- cillin/streptomycin/glutamine and cells incubated at 5% CO2 at 37 C. Trans- tion of the Insig/SCAP/SREBP interaction. Clearly, identification fection of plasmids was performed using X-tremeGENE HP DNA Transfection Reagent (Roche) following the manufacturer’s instructions. of the binding motif for Insig or SCAP for N-linked oligosaccha- ride will require further studies. Co-immunoprecipitation Our previous studies revealed that SREBP-1 is activated by Co-immunoprecipitation assay was performed as described previously (Yang the EGFR-PI3K-Akt signaling pathway, but mTORC1 seems et al., 2002). not involved (Guo et al., 2009b, 2011). Studies by other investi- gators reported that PI3K-Akt-mTORC1 signaling regulates Preparation of Cell Membrane Fractions and Nuclear Extracts Cell membrane and nuclear fractions were isolated as described previously SREBP-1 activation (Du et al., 2006; Porstmann et al., 2005; (Nohturfft et al., 1998). Yecies et al., 2011). Therefore, the mechanisms that underlie the oncogenic signaling to SREBP-1 function remain unclear. Detection of SCAP N-glycosylation In the present study, we show that in the absence of glucose, The detection of SCAP N-glycosylation was performed according to the EGF cannot activate SREBP-1, even though other downstream method described previously (Nohturfft et al., 1998). signaling components (e.g., p-Akt, p-S6) remain upregulated. Xenograft Mouse Model and Mouse Survival This suggests that EGFR-mediated control of SREBP-1 is fully GBM cells (1 3 106 cells) were implanted into the flank of athymic nu/nu female dependent on glucose. We also examined mTORC1 activity by mice (6–8 weeks old) subcutaneously. Mice were sacrificed by euthanasia checking its downstream effectors p-S6 and total S6 levels in when tumor size reached the limitation and tumors were isolated and weighed. the absence or presence of serum. The data show that the levels For intracranial xenograft models, 1 3 105 GBM cells in 4 ml of PBS were of p-S6 and total S6 were not significantly changed by glucose stereotactically implanted into mouse brain. Mice were then observed until withdrawal within a 12 hr time window. In contrast, the levels they became moribund, at which point they were sacrificed. All animal proce- of SCAP and SREBP-1 and SREBP-2 activation were reduced. dures were approved by the Subcommittee on Research Animal Care at The Ohio State University Medical Center. Moreover, in the absence of glucose, even a growth factor (EGF or insulin) strongly activated mTORC1, but failed to activate Mouse Luminescence Imaging SCAP/SREBP signaling. Taken together, these data demon- Mice implanted with GBM cells expressing luciferase were injected with lucif- strate that glucose, not mTORC1, plays a critical role in regu- erin (Perkin Elmer) solution (15 mg/ml in PBS and dose of 150 mg/kg) by an lating SCAP/SREBP signaling. intraperitoneal route for about 5–15 min. Mice were placed into a clear Plexi- Our study revealed that EGFR signaling via promoting glucose glas anesthesia box (2.0%–3.0% isoflurane) that allowed unimpeded visual monitoring of the animals. Animals were then placed on non-fluorescent black uptake enhances SCAP N-glycosylation and its protein levels to paper on the imaging platform of an IVIS Lumina II to reduce background activate SREBP-1. This finding provides a plausible explanation noise. The imaging chamber was continuously infused with 1%–1.5% of for the conundrum faced by the tumor cells: Under rich nutrient isoflurane and the imaging platform was heated at 37C to keep the environment, oncogenic signaling via hijacking glucose import- mice warm. Animals were imaged 10 min after luciferin injection to ensure

Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. 579 consistent photon flux. The imaging experiments were conducted at The Ohio Goldstein, J.L., DeBose-Boyd, R.A., and Brown, M.S. (2006). Protein sensors State University (OSU) Small Animal Imaging Core. for membrane sterols. Cell 124, 35–46. Griffiths, B., Lewis, C.A., Bensaad, K., Ros, S., Zhang, Q., Ferber, E.C., Konisti, Statistical Analysis S., Peck, B., Miess, H., East, P., et al. (2013). Sterol regulatory element binding Statistical analyses were performed by Excel or GraphPad Prism5. All data are protein-dependent regulation of lipid synthesis supports cell survival and analyzed by two-tailed t test as well as by ANOVA as appropriate and p < 0.05 tumor growth. Cancer Metab. 1,3. was considered statistically significant. Guillet-Deniau, I., Pichard, A.L., Kone´ , A., Esnous, C., Nieruchalski, M., Girard, J., and Prip-Buus, C. (2004). Glucose induces de novo lipogenesis in rat mus- SUPPLEMENTAL INFORMATION cle satellite cells through a sterol-regulatory-element-binding-protein-1c- dependent pathway. J. Cell Sci. 117, 1937–1944. Supplemental Information includes Supplemental Experimental Procedures and four figures and can be found with this article online at http://dx.doi.org/ Gullino, P.M., Grantham, F.H., and Courtney, A.H. (1967). Glucose consump- 10.1016/j.ccell.2015.09.021. tion by transplanted tumors in vivo. Cancer Res. 27, 1031–1040. Guo, D., Hildebrandt, I.J., Prins, R.M., Soto, H., Mazzotta, M.M., Dang, J., AUTHOR CONTRIBUTIONS Czernin, J., Shyy, J.Y., Watson, A.D., Phelps, M., et al. (2009a). The AMPK agonist AICAR inhibits the growth of EGFRvIII-expressing glioblastomas by D.G. conceived the ideas. C.C. and D.G. designed the experiments. C.C., P.R., inhibiting lipogenesis. Proc. Natl. Acad. Sci. USA 106, 12932–12937. F.G., J.L., J.Y.Y., X.W., X.C., V.E., P.H., J.Y.G., E.L., and D.G. performed the Guo, D., Prins, R.M., Dang, J., Kuga, D., Iwanami, A., Soto, H., Lin, K.Y., experiments. C.C., B.K., A.C., and D.G. analyzed the data. C.C., A.N., J.M., Huang, T.T., Akhavan, D., Hock, M.B., et al. (2009b). EGFR signaling through and D.G. wrote the manuscript, and all authors reviewed and approved the an Akt-SREBP-1-dependent, rapamycin-resistant pathway sensitizes glio- manuscript for publication. blastomas to antilipogenic therapy. Sci. Signal. 2, ra82.

ACKNOWLEDGMENTS Guo, D., Reinitz, F., Youssef, M., Hong, C., Nathanson, D., Akhavan, D., Kuga, D., Amzajerdi, A.N., Soto, H., Zhu, S., et al. (2011). An LXR agonist promotes We are grateful to Drs. Mike S. Brown and Joseph L. Goldstein for their careful glioblastoma cell death through inhibition of an EGFR/AKT/SREBP-1/LDLR- 1 reading of the manuscript and helpful comments. We thank Dr. Paul Mischel dependent pathway. Cancer Discov. , 442–456. for GBM cell lines. We thank Dr. S. Jaharul Haque for helpful comments. Guo, D., Bell, E.H., and Chakravarti, A. (2013). Lipid metabolism emerges as a This work was supported by NIH grants NS072838 and NS079701 to D.G. promising target for malignant glioma therapy. CNS Oncol. 2, 289–299. and AG28614 to J.M. and American Cancer Society Research Scholar Grant Guo, D., Bell, E.H., Mischel, P., and Chakravarti, A. (2014). Targeting SREBP- RSG-14-228-01-CSM to D.G. We also appreciate the support from the Ohio 1-driven lipid metabolism to treat cancer. Curr. Pharm. Des. 20, 2619–2626. State Neuroscience Core (P30 NS038526) and OSUCCC start-up funds. Hasty, A.H., Shimano, H., Yahagi, N., Amemiya-Kudo, M., Perrey, S., Received: March 12, 2015 Yoshikawa, T., Osuga, J., Okazaki, H., Tamura, Y., Iizuka, Y., et al. (2000). Revised: August 4, 2015 Sterol regulatory element-binding protein-1 is regulated by glucose at the tran- Accepted: September 27, 2015 scriptional level. J. Biol. Chem. 275, 31069–31077. Published: November 9, 2015 Hirayama, A., Kami, K., Sugimoto, M., Sugawara, M., Toki, N., Onozuka, H., Kinoshita, T., Saito, N., Ochiai, A., Tomita, M., et al. (2009). Quantitative metab- REFERENCES olome profiling of colon and stomach cancer microenvironment by capillary electrophoresis time-of-flight mass spectrometry. Cancer Res. 69, 4918– Ackerman, D., and Simon, M.C. (2014). Hypoxia, lipids, and cancer: surviving 4925. the harsh tumor microenvironment. Trends Cell Biol. 24, 472–478. Horton, J.D., Bashmakov, Y., Shimomura, I., and Shimano, H. (1998). Adams, C.M., Reitz, J., De Brabander, J.K., Feramisco, J.D., Li, L., Brown, Regulation of sterol regulatory element binding proteins in livers of fasted M.S., and Goldstein, J.L. (2004). Cholesterol and 25-hydroxycholesterol inhibit and refed mice. Proc. Natl. Acad. Sci. USA 95, 5987–5992. activation of SREBPs by different mechanisms, both involving SCAP and Insigs. J. Biol. Chem. 279, 52772–52780. Horton, J.D., Goldstein, J.L., and Brown, M.S. (2002). SREBPs: activators of the complete program of cholesterol and fatty acid synthesis in the liver. Babic, I., Anderson, E.S., Tanaka, K., Guo, D., Masui, K., Li, B., Zhu, S., Gu, Y., J. Clin. Invest. 109, 1125–1131. Villa, G.R., Akhavan, D., et al. (2013). EGFR mutation-induced alternative splicing of Max contributes to growth of glycolytic tumors in brain cancer. Horton, J.D., Shah, N.A., Warrington, J.A., Anderson, N.N., Park, S.W., Brown, Cell Metab. 17, 1000–1008. M.S., and Goldstein, J.L. (2003). Combined analysis of oligonucleotide micro- Bennett, M.K., Lopez, J.M., Sanchez, H.B., and Osborne, T.F. (1995). Sterol array data from transgenic and knockout mice identifies direct SREBP target 100 regulation of fatty acid synthase promoter. Coordinate feedback regulation genes. Proc. Natl. Acad. Sci. USA , 12027–12032. of two major lipid pathways. J. Biol. Chem. 270, 25578–25583. Huang, H.S., Nagane, M., Klingbeil, C.K., Lin, H., Nishikawa, R., Ji, X.D., Cloughesy, T.F., Cavenee, W.K., and Mischel, P.S. (2014). Glioblastoma: from Huang, C.M., Gill, G.N., Wiley, H.S., and Cavenee, W.K. (1997). The enhanced molecular pathology to targeted treatment. Annu. Rev. Pathol. 9, 1–25. tumorigenic activity of a mutant epidermal growth factor receptor common in human cancers is mediated by threshold levels of constitutive tyrosine phos- Currie, E., Schulze, A., Zechner, R., Walther, T.C., and Farese, R.V., Jr. (2013). phorylation and unattenuated signaling. J. Biol. Chem. 272, 2927–2935. Cellular fatty acid metabolism and cancer. Cell Metab. 18, 153–161. Dif, N., Euthine, V., Gonnet, E., Laville, M., Vidal, H., and Lefai, E. (2006). Insulin Jeon, T.I., and Osborne, T.F. (2012). SREBPs: metabolic integrators in physi- 23 activates human sterol-regulatory-element-binding protein-1c (SREBP-1c) ology and metabolism. Trends Endocrinol. Metab. , 65–72. promoter through SRE motifs. Biochem. J. 400, 179–188. Kamisuki, S., Mao, Q., Abu-Elheiga, L., Gu, Z., Kugimiya, A., Kwon, Y., Du, X., Kristiana, I., Wong, J., and Brown, A.J. (2006). Involvement of Akt in ER- Shinohara, T., Kawazoe, Y., Sato, S., Asakura, K., et al. (2009). A small mole- to-Golgi transport of SCAP/SREBP: a link between a key cell proliferative cule that blocks fat synthesis by inhibiting the activation of SREBP. Chem. Biol. pathway and membrane synthesis. Mol. Biol. Cell 17, 2735–2745. 16, 882–892. Ettinger, S.L., Sobel, R., Whitmore, T.G., Akbari, M., Bradley, D.R., Gleave, Kaplan, M., Kerry, R., Aviram, M., and Hayek, T. (2008). High glucose concen- M.E., and Nelson, C.C. (2004). Dysregulation of sterol response element-bind- tration increases macrophage cholesterol biosynthesis in diabetes through ing proteins and downstream effectors in prostate cancer during progression activation of the sterol regulatory element binding protein 1 (SREBP1): inhibi- to androgen independence. Cancer Res. 64, 2212–2221. tory effect of insulin. J. Cardiovasc. Pharmacol. 52, 324–332.

580 Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. Li, C., Yang, W., Zhang, J., Zheng, X., Yao, Y., Tu, K., and Liu, Q. (2014). Ru, P., Williams, T.M., Chakravarti, A., and Guo, D. (2013). Tumor metabolism SREBP-1 has a prognostic role and contributes to invasion and metastasis of malignant gliomas. Cancers (Basel) 5, 1469–1484. 15 in human hepatocellular carcinoma. Int. J. Mol. Sci. , 7124–7138. Sakai, J., Nohturfft, A., Goldstein, J.L., and Brown, M.S. (1998). Cleavage of Menendez, J.A., and Lupu, R. (2007). Fatty acid synthase and the lipogenic sterol regulatory element-binding proteins (SREBPs) at site-1 requires interac- phenotype in cancer pathogenesis. Nat. Rev. Cancer 7, 763–777. tion with SREBP cleavage-activating protein. Evidence from in vivo competi- tion studies. J. Biol. Chem. 273, 5785–5793. Mitro, N., Mak, P.A., Vargas, L., Godio, C., Hampton, E., Molteni, V., Kreusch, A., and Saez, E. (2007). The nuclear receptor LXR is a glucose sensor. Nature Sun, L.P., Seemann, J., Goldstein, J.L., and Brown, M.S. (2007). Sterol-regu- 445, 219–223. lated transport of SREBPs from endoplasmic reticulum to Golgi: Insig renders sorting signal in Scap inaccessible to COPII proteins. Proc. Natl. Acad. Sci. Moon, Y.A., Liang, G., Xie, X., Frank-Kamenetsky, M., Fitzgerald, K., USA 104, 6519–6526. Koteliansky, V., Brown, M.S., Goldstein, J.L., and Horton, J.D. (2012). The Scap/SREBP pathway is essential for developing diabetic fatty liver and car- Tang, J.J., Li, J.G., Qi, W., Qiu, W.W., Li, P.S., Li, B.L., and Song, B.L. (2011). bohydrate-induced hypertriglyceridemia in animals. Cell Metab. 15, 240–246. Inhibition of SREBP by a small molecule, betulin, improves hyperlipidemia and insulin resistance and reduces atherosclerotic plaques. Cell Metab. 13, 44–56. Nohturfft, A., and Zhang, S.C. (2009). Coordination of lipid metabolism in mem- Uehara, T., Nakamura, T., Yao, D., Shi, Z.Q., Gu, Z., Ma, Y., Masliah, E., brane biogenesis. Annu. Rev. Cell Dev. Biol. 25, 539–566. Nomura, Y., and Lipton, S.A. (2006). S-nitrosylated protein-disulphide isom- Nohturfft, A., Brown, M.S., and Goldstein, J.L. (1998). Topology of SREBP erase links protein misfolding to neurodegeneration. Nature 441, 513–517. cleavage-activating protein, a polytopic membrane protein with a sterol- Wang, X., Sato, R., Brown, M.S., Hua, X., and Goldstein, J.L. (1994). SREBP-1, sensing domain. J. Biol. Chem. 273, 17243–17250. a membrane-bound transcription factor released by sterol-regulated proteol- Nohturfft, A., Yabe, D., Goldstein, J.L., Brown, M.S., and Espenshade, P.J. ysis. Cell 77, 53–62. (2000). Regulated step in cholesterol feedback localized to budding of SCAP Wojton, J., Chu, Z., Mathsyaraja, H., Meisen, W.H., Denton, N., Kwon, C.H., from ER membranes. Cell 102, 315–323. Chow, L.M., Palascak, M., Franco, R., Bourdeau, T., et al. (2013). Systemic Porstmann, T., Griffiths, B., Chung, Y.L., Delpuech, O., Griffiths, J.R., delivery of SapC-DOPS has antiangiogenic and antitumor effects against glio- Downward, J., and Schulze, A. (2005). PKB/Akt induces transcription of blastoma. Mol. Ther. 21, 1517–1525. enzymes involved in cholesterol and fatty acid biosynthesis via activation of Yang, T., Espenshade, P.J., Wright, M.E., Yabe, D., Gong, Y., Aebersold, R., 24 SREBP. Oncogene , 6465–6481. Goldstein, J.L., and Brown, M.S. (2002). Crucial step in cholesterol homeosta- Radhakrishnan, A., Goldstein, J.L., McDonald, J.G., and Brown, M.S. (2008). sis: sterols promote binding of SCAP to INSIG-1, a membrane protein that Switch-like control of SREBP-2 transport triggered by small changes in ER facilitates retention of SREBPs in ER. Cell 110, 489–500. cholesterol: a delicate balance. Cell Metab. 8, 512–521. Yecies, J.L., Zhang, H.H., Menon, S., Liu, S., Yecies, D., Lipovsky, A.I., Rawson, R.B., DeBose-Boyd, R., Goldstein, J.L., and Brown, M.S. (1999). Gorgun, C., Kwiatkowski, D.J., Hotamisligil, G.S., Lee, C.H., and Manning, Failure to cleave sterol regulatory element-binding proteins (SREBPs) causes B.D. (2011). Akt stimulates hepatic SREBP1c and lipogenesis through parallel cholesterol auxotrophy in Chinese hamster ovary cells with genetic absence of mTORC1-dependent and independent pathways. Cell Metab. 14, 21–32. SREBP cleavage-activating protein. J. Biol. Chem. 274, 28549–28556. Yokoyama, C., Wang, X., Briggs, M.R., Admon, A., Wu, J., Hua, X., Goldstein, Repa, J.J., Liang, G., Ou, J., Bashmakov, Y., Lobaccaro, J.M., Shimomura, I., J.L., and Brown, M.S. (1993). SREBP-1, a basic-helix-loop-helix-leucine Shan, B., Brown, M.S., Goldstein, J.L., and Mangelsdorf, D.J. (2000). zipper protein that controls transcription of the low density lipoprotein receptor 75 Regulation of mouse sterol regulatory element-binding protein-1c gene gene. Cell , 187–197. (SREBP-1c) by oxysterol receptors, LXRalpha and LXRbeta. Genes Dev. 14, Zelcer, N., and Tontonoz, P. (2006). Liver X receptors as integrators of meta- 2819–2830. bolic and inflammatory signaling. J. Clin. Invest. 116, 607–614.

Cancer Cell 28, 569–581, November 9, 2015 ª2015 Elsevier Inc. 581