RING-Dependent Tumor Suppression and G2/M Arrest Induced by the TRC8 Hereditary Kidney Cancer Gene

Oncogene (2007) 26, 2263–2271 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE RING-dependent tumor suppression and G2/M arrest induced by the TRC8 hereditary kidney cancer gene

A Brauweiler1, KL Lorick2, JP Lee1, YC Tsai2, D Chan1, AM Weissman2, HA Drabkin1 and RM Gemmill1

1Division of Medical Oncology, Department of Medicine, University of Colorado at Denver and Health Sciences Center and University of Colorado Cancer Center, Aurora, CO, USA and 2Lab of Protein Dynamics and Signaling, National Cancer Institute, Fort Detrick, Frederick, MD, USA

TRC8/RNF139 and von Hippel–Lindau (VHL) both membrane-associatedE3 ubiquitin (Ub) ligase disrupted encode E3 ubiquitin (Ub) ligases mutated in clear-cell by the translocation (Gemmill et al., 2002). The renal carcinomas (ccRCC). VHL, inactivated in nearly targeting subunit of a secondE3 ligase, encodedby 70% of ccRCCs, is a tumor suppressor encoding the the von Hippel–Lindau gene (VHL), is defective in most targeting subunit for a Ub ligase complex that down- sporadic clear-cell renal cell carcinomas (ccRCC) regulates hypoxia-inducible factor-a. TRC8/RNF139 is a (Kaelin, 2005). The best-known targets for the VHL putative tumor suppressor containing a sterol-sensing E3 ligase are the hypoxia-inducible transcription (HIF) domain and a RING-H2 motif essential for Ub ligase factors, HIF-1 and-2 a . The HIF system represents a activity. Here we report that human kidney cells are major pathway for sensing oxygen levels andis directly growth inhibited by TRC8. Inhibition is manifested by linkedto kidney tumorigenesis. G2/M arrest, decreased DNA synthesis and increased A novel oxygen-sensing mechanism involving lipid apoptosis and is dependent upon the Ub ligase activity of homeostasis was recently described in fission yeast the RING domain. Tumor formation in a nude mouse (Hughes et al., 2005; Todd et al., 2006). The sterol model is inhibited by TRC8 in a RING-dependent manner. regulatory element binding protein (SREBP) homolog Expression of TRC8 represses genes involved in choles- in Schizosaccharomyces pombe, Sre1, is activatedby terol and fatty acid biosynthesis that are transcriptionally oxygen limitation while mutations in Sre1 inhibit cell regulated by the sterol response element binding proteins growth under severe hypoxia (Hughes et al., 2005). In (SREBPs). Expression of activated SREBP-1a partially mammalian cells, the SREBP transcription factors are restores the growth of TRC8-inhibited cells. These data responsible for coordinately inducing cholesterol and suggest that TRC8 modulation of SREBP activity fatty acidbiosynthetic enzymes, such as 3-hydroxy-3- comprises a novel regulatory linkbetween growth control methylglutaryl coenzyme A reductase (HMGCR) (Hor- and the cholesterol/lipid homeostasis pathway. ton et al., 2003; Goldstein et al., 2006). Interestingly, Oncogene (2007) 26, 2263–2271. doi:10.1038/sj.onc.1210017; ccRCCs contain abundant lipid inclusions composed of publishedonline 2 October 2006 newly synthesizedcholesterol esters andother neutral lipids (Clayman et al., 1986; Gebhard et al., 1987), Keywords: ubiquitin ligase; SREBP; clear-cell renal suggesting that deregulation of lipid homeostasis is carcinoma; RING-H2 characteristic for these tumors. OverexpressedSREBPs have been foundin a variety of tumor types, including prostate (Ettinger et al., 2004), breast (Martel et al., 2006) andglioblastoma (Ma et al., 2005), andwe have identified similar changes in ccRCCs (Lee et al., Introduction submitted). Only 60–70% of ccRCCs have direct defects in the The TRC8/RNF139 gene was identified in a unique VHL pathway andsearches for mutations in other family with the constitutional translocation, components of the VHL complex have proven negative t(3;8)(3p14.2;q24.1), andhereditary renal andthyroid (Clifford et al., 2001). VHL loss alone appears cancer (Cohen et al., 1979; Li et al., 1993; Gemmill et al., insufficient for tumor development, suggesting that 1998). TRC8 encodes an endoplasmic reticulum (ER) alterations in additional genes and/or regulatory path- ways are necessary (Mandriota et al., 2002). One Correspondence: Professor RM Gemmill, Division of Medical candidate is TRC8 (Gemmill et al., 1998, 2002, 2005), Oncology, University of Colorado at Denver and Health Sciences a 60 kDa ER protein with 10 predicted transmembrane Center andUniversity of ColoradoCancer Center, 12801 E. 17th segments anda sterol-sensing domain(Gemmill et al., Avenue, Mail Stop 8117, PO Box 6511, Aurora 80045-0511, CO, USA. E-mail: [email protected] 1998), similar to those foundin SCAP andHMGCR Received19 June 2006; revised16 August 2006; accepted18 August 2006; that confer sterol-sensitivity to the SREBP/lipidhomeo- publishedonline 2 October 2006 static pathway (Goldstein et al., 2006). We have recently RING-dependent tumor suppression by TRC8 A Brauweiler et al 2264 discovered that TRC8 stability is sterol-regulated (Lee ectopic TRC8 expression (Figure 1a). Co-expression of et al., submitted). TRC8 also contains a RING-H2 enhancedgreen fluorescent protein (EGFP) indicated domain in the C-terminus that catalyses in vitro transfection efficiency. Ad-TRC8 infection (multiplicity ubiquitylation reactions (Lorick et al., 1999). In of infection (MOI) of 50, >90% EGFP positive) Drosophila, DTrc8 is a potent growth suppressor and resultedin a five-foldreductionin cell proliferation appears to function in conjunction with, or parallel to, after 48 h comparedto cells infectedwith Ad-EGFP DVhl (Gemmill et al., 2002). Here we demonstrate that alone (Figure 1b). Flow cytometry (SKRC-09 is shown TRC8 is a potent growth suppressor in mammalian in Figure 1c) revealeda large increase in the sub-G1 cells, inducing G2/M arrest, decreased DNA synthesis apoptotic fraction in Ad-TRC8-infected cells compared andincreasedapoptosis. These in vitro effects, along to controls (Figure 1d). These results support a tumor- with in vivo inhibition of tumorigenicity, are dependent suppressive function for TRC8 through inhibition of cell upon the Ub ligase activity of the RING domain. TRC8 proliferation and induction of cell death. inhibits the expression of SREBP-regulatedgenes and To investigate these effects further, we utilizedhuman ectopic expression of activatednuclear SREBP embryonic kidney (HEK293) FlpIn TRex cells (293FI) (nSREBP) partially restores growth to TRC8-inhibited in which TRC8 was placedunder the control of a cells. Our results suggest that TRC8 is a tumor tetracycline-inducible promoter (Figure 2a). Consistent suppressor implicatedin a novel regulatory relationship with the adenovirus results, TRC8 suppressed the linking the cholesterol/lipidbiosynthetic pathway with proliferation of 293FI cells comparedto doxycycline cellular growth control. (dox)-treated control cells (Figure 2b). Flow cytometry demonstrated that TRC8 induced apoptosis and in- creasedcell populations with 4 N DNA content, consistent with G2/M arrest (Figure 2c). Bromodeox- Results yuridine (BrdU) incorporation, indicative of DNA synthesis, was suppressedby TRC8 (Figure 2d).These TRC8 inhibits proliferation of transformed kidney cells data indicate that in 293FI cells, TRC8 blocks Adenovirus infection of three ccRCC cell lines (SKRC- proliferation by inhibiting DNA synthesis, increasing 02, -09 and-17) was usedto assess short-term effects of apoptosis andinducingcell-cycle arrest.

Figure 1 TRC8 inhibits growth of RCC cell lines in vitro.(a) Adenoviral constructs expressing EGFP alone (EGFP) or EGFP with TRC8 were introduced into three ccRCC cell lines (SKRC-02, -09 and -17) using an MOI of 50. Microscopic examination 24 h post- infection confirmed>90% expressedEGFP (not shown). Lysates (48 h) were analysedby Western blot for TRC8 expression using anti-HA antibody Y-11. Tubulin provided a loading control. (b) Cells were platedin triplicate andinfectedwith control or TRC8- encoding adenoviruses (MOI ¼ 50). After 48 h, total viable cells were countedusing trypan-blue exclusion. Cell numbers were normalizedto the EGFP controls. ( c and d) Parallel cultures were analysedby flow cytometry 48 h post-infection. ( c) Cell cycle profile of TRC8-Ad-infected SKRC-09 cells compared to EGFP-Ad-infected controls. (d) Quantitation of apoptotic sub-G1 population (percent of total).

Oncogene RING-dependent tumor suppression by TRC8 A Brauweiler et al 2265 et al., 2003). PuriﬁedGST fusion proteins (Figure 3c) were assayedin reactions containing 32P-labelledUb, E1, ATP andUbe2d2(formerly UbcH5b). Reaction products were analysed by sodium dodecyl sulfate– polyacrylamide gel electrophoresis (SDS-PAGE) and autoradiography (Figure 3b). Wild-type GST-TRC8 generatedprominent ubiquitylation productscontaining Ub multimers up to 3 andsmall amounts of higher molecular weight adducts (lane 2). The positive control GST-AO7 (lane 7), a RING domain-containing protein with well-deﬁned Ub ligase activity, yielded abundant higher molecular weight products (Lorick et al., 1999). In contrast, mutation of the Zn2 þ -coordinating Cys residues in the CAIC and CPMC mutations resulted in nearly total ablation of ubiquitylation activity (lanes 3 and 6). Thus, as expected, direct disruption of the Zn2 þ coordinating sites eliminated ubiquitylation activity for the TRC8 RING domain. The remaining mutations affected amino acids outside the Zn2 þ binding sites and showeddifferential effects on ubiquitylation. LRK (lane 5) lost all activity while SAR (lane 4) retained about 30% of wild-type activity.

Growth suppression mediated by TRC8 is RING dependent Figure 2 TRC8 inhibits growth of HEK293 cells in vitro.(a) We subsequently testedthese mutations for growth effects HEK293FI cells were stably transfectedwith the pcDNA5/FRT/ in 293FI cells. Stable transfectants containing tetracycline- TO shuttle vector containing TRC8-HA or no insert (vector inducible RING mutants, including a deletion of the entire control) andselectedfor hygromycin resistance. Populations of resistant cells were treatedwith 1 mg/ml dox and samples harvested RING (DRH2), or the relatedRING protein, gp78, were periodically. Lysates were analysed for inducible TRC8 expression generatedandverified(Figure 4a–c). Viable cell counts using anti-HA antibodies and compared to uninduced cells (0) and were usedto compare growth over 3 daysbetween dox- induced vector controls. (b) Cells carrying TRC8 or empty vector induced control cells, and cells expressing TRC8wt, were platedin triplicate andthen inducedwith1 mg/ml dox 12 h DRH2 CAIC wt later. Cultures were harvestedperiodicallyandcountedfor viable TRC8 andTRC8 (Figure 4d). While TRC8 - cells using trypan-blue exclusion. (c) 293FI-TRC8 or empty vector expressing cells failedto proliferate, as expected,expression cells were induced with 1 mg/ml dox at time zero, then periodically of the CAIC mutation or DRH2 RING deletion yielded harvestedandanalysedby flow cytometry. Superimposedcell cycle proliferation rates indistinguishable from control cells. profiles are shown at 24, 48 and72 h post-induction with apoptotic, Comparisons among all the RING mutants (Figure 4e and G1 and G2/M populations indicated. (d) Parallel cultures of 293FI- TRC8 cells were induced with dox at time zero. BrdU (10 mM) was Table 1) showedthat the SAR mutation retainedmost of added for the final 5 h before harvest. Cells were fixed, stained with the growth inhibitory activity of wildtype, whereas CAIC, anti-BrdU antibody and analysed by flow cytometry to quantitate LRK andCPMC grew almost as well as uninducedcells. the proportion of BrdU-positive cells. These RING mutations thus lost nearly all growth inhibitory activity associatedwith wild-typeTRC8. Expression of gp78, an ER-associateddegradation Mutation of the TRC8 RING-H2 domain abrogates Ub (ERAD)-associatedRING protein (Fang et al., 2001) ligase activity relatedto TRC8 (40% similar over 200 amino acids) The C-terminus of TRC8 contains a RING-H2 domain resultedin no detectable growth inhibition, showing that (Gemmill et al., 1998) (Figure 3a) competent for E3 Ub this effect is specific for TRC8. Flow cytometry confirmed ligase activity (Lorick et al., 1999). RING domains that all mutants in Zn2 þ coordination sites (asterisks) were utilize Cys and His residues to coordinate two Zn2 þ ions unable to induce G2/M arrest or apoptosis (Figure 4f and in a cross-bracedconfiguration, creating a bindingsite Table 1). This result was extended to three ccRCC cell lines for Ub-conjugating enzymes (E2s) (Zheng et al., 2000). using adenovirus-mediated expression of TRC8CAIC To evaluate the role of this domain for TRC8 function, (Figure 4g). Despite similar expression levels, this RING we generateda series of alanine or serine substitution mutation was unable to inhibit growth of ccRCCs while mutations in highly conservedamino acidsandestab- wild-type TRC8 was highly effective. These results lishedtheir ubiquitylation activity in vitro. RING demonstrate that the known biological effects of TRC8 mutations were createdin glutathione S-transferase require the Ub ligase activity of its RING domain. (GST)-TRC8 fusion proteins encompassing amino acids 504–664. These mutations were chosen to either abrogate Zn2 þ coordination sites, and thus disrupt TRC8 activates ATM and Chk2 kinases RING folding, or to interfere with specific amino-acid G2/M arrest suggestedthat upstream checkpoint motifs predicted to form an E2 binding site (Katoh activators might be influencedby TRC8. Western blot

Oncogene RING-dependent tumor suppression by TRC8 A Brauweiler et al 2266

Figure 3 In vitro ubiquitylation activity of wild-type and RING mutant TRC8. (a) Clustal-W alignment of TRC8 RING domains (boxed) from Homo sapiens (Hs), Mus musculus (Mm) and Drosophila melanogaster (Dm) emphasizing the Zn2 þ -coordinating Cys and His residues (larger font). RING mutations are indicated by arrows. Amino-acid positions are from the H. sapiens sequence. (b) In vitro ubiquitylation reactions (see Supplementary Methods) were carried out using GST fusion proteins containing the indicated wild-type or mutant RING domains of TRC8. The AO7 RING domain fused to GST served as a positive control (lane 7) (Lorick et al., 1999). Radiolabelled Ub was incorporated into high molecular weight products in a reaction containing murine E1, Ube2d2, 32P-Ub, ATP and an ATP regeneration system. Labelled products were detected by SDS–PAGE and phosphorimaging. Background bands observed with GST alone (lane 1) are tentatively identified along the right side. (c) PurifiedGST fusion proteins were semi-quantifiedby Coomassie Blue staining and comparison to a bovine serum albumin standard.

analysis was usedto examine the activation state of each animal group receiveddoxcontinuously in the Chk2 andATM kinases in response to TRC8 induction. drinking water starting 1 day before tumor cell injection. Wild-type but not RING mutant TRC8 resulted in Dox-induced TRC8wt-expressing cells did not form any phosphorylation of Chk2 on Thr 68 (Figure 4h), a site tumors, but expression of the TRC8CAIC RING muta- essential for subsequent autophosphorylation in the tion permittedrobust tumor formation (Figure 5a). activation loop. The Thr 68 site is phosphorylatedby Measurement of tumor diameters (at day 34 post- activatedATM/ATR kinases, andwe foundthat TRC8 inoculation) revealedsimilar growth rates for tumors induction resulted in Ser1981 phosphorylation of ATM derived from uninduced TRC8wt- andTRC8 CAIC-trans- (Figure 4g, bottom panels). Here also, the RING fectedcells andfor tumors inducedto express the mutant TRC8CAIC did not alter phosphor-ATM levels. TRC8CAIC mutation (Figure 5b). These studies demon- These modiﬁcations are consistent with the G2/M arrest strate that wild-type TRC8 is a potent tumor suppressor observedin Figures 2c and4f. when expressedin HEK293 cells andthis effect is dependent upon a functional RING domain.

TRC8 inhibits tumor formation in a nude mouse model Xenograft studies were conducted to determine if TRC8 Ectopic SREBP can partly restore growth to TRC8- couldsuppress tumorigenicity of 293FI cells. Stable inhibited cells transfectants carrying inducible TRC8wt or TRC8CAIC We recently discovered that TRC8 downregulates were inoculatedsubcutaneously into nu/nu mice. Half of expression of SREBP target genes in HEK293 cells

Oncogene RING-dependent tumor suppression by TRC8 A Brauweiler et al 2267

Figure 4 TRC8 RING mutations lose growth inhibition. (a) HEK293FI cells expressing full-length wild-type or RING mutant TRC8 proteins were induced for 24 h with 1 mg/ml dox. Lysates were analysed by Western blot to verify induction and equal expression. Tubulin provided a loading control. (b) and( c) demonstrate dox induction for gp78 and TRC8DRH2, respectively, in equivalent 293FI cell lines. (d) Equal numbers of cells were platedin triplicate andviable cells countedafter 1 to 3 daysof growth in the presence of 1 mg/ ml dox. Data were normalized to the cell number at day 1. (e) Relative proliferation of viable cells was establishedfor each indicated cell line as in (d). Data are reported for the 3-day time point as the percent of proliferation compared to uninduced controls (set to 100%). (f) Cell cycle analysis for the samples in (e) determined by propidium iodide staining and flow cytometry. Asterisks (*) mark those mutations that affect Zn2 þ coordination sites. (g) Relative proliferation of RCC cell lines infectedfor 48 h with Adenovirus- expressing WT or CAIC mutant TRC8. Lysates were analysedby Western blot to verify expression using anti-TRC8 antibodies (Abnova Corp., Taipai, Taiwan). (h) 293FI cells were induced to express WT or CAIC mutant TRC8 for 48 h. Lysates were analysed using anti phospho-Chk2 antibody specific for phosphorylated Thr 68 and with anti-phospho-ATM antibody specific for phosphorylatedSer 1981. Total Chk2 andATM demonstratedunchangedprotein levels.

(Lee et al., submitted). Table 2 shows results of real-time nuclear form of SREBP1 (nSREBP1, amino acids 1–460) polymerase chain reaction (PCR) analysis for two was transfectedinto 293FI cells andcell proliferation SREBP target genes, HMGCR andfatty acidsynthase quantiﬁedwith andwithout TRC8 induction (FAS), verifying their suppression by TRC8 induction. (Figure 6a). TRC8 expression reduced cell number to As expression of SREBP target genes is often essential about 35% of uninduced control cells (Po0.001). for rapidcell growth, andoverexpression occurs in Although nSREBP1 hadno effect on uninducedcells, several tumor types (Ettinger et al., 2004; Ma et al., it doubled cell numbers for TRC8-inhibited cells, 2005; Lee et al., Submitted; Martel et al., 2006), we restoring growth to about 70% of controls (Po0.01). askedif SREBP replacement couldovercome TRC8- Despite this increase, expression of constitutively active induced growth inhibition. The constitutively active nSREBP did not fully restore growth, suggesting that

Oncogene RING-dependent tumor suppression by TRC8 A Brauweiler et al 2268 additional mechanisms are contributing to TRC8- but ATM inactivation leads to increased reactive oxygen mediated inhibition. Co-transfection of pSYNSRE-luc species (Ito et al., 2004), suggesting a connection to the (Figure 5b, top panel), a reporter plasmidsensitive to VHL/HIF system (Kaelin, 2005) and/or to the hypoxic SREBP activation, verifiedthat nSREBP protein activation of SREBPs (Hughes et al., 2005). The lack of (bottom panel) was active. proliferative alterations from overexpression of the TRC8-relatedRING protein, gp78 (Figure 4), supports the conclusion that TRC8 growth effects are specific. Subcutaneous tumor formation by HEK293 cells was Discussion completely blockedby induction of TRC8, while uninduced cells formed tumors efficiently. These results Our results show that TRC8 inhibits proliferation in support a tumor suppressor role for TRC8 in mamma- mammalian cells through mechanisms involving G2/M lian cells (Figure 5), consistent with our previous data in arrest andcell death.Ectopic TRC8 expression in RCC Drosophila where dTrc8 inhibited wing and eye imaginal lines induced cell death (Figure 1), whereas in HEK293 disc development (Gemmill et al., 2002, 2005). As the cells, a combination of G2/M arrest andcell deathwas current results have been obtainedonly in kidney- observed(Figure 2). DNA synthesis, as measuredby derived cell lines, we cannot exclude the possibility that BrdU incorporation, was also inhibited. Cell cycle TRC8 has cell-type-specific effects not observedhere. checkpoint kinases ATM andChk2 were activatedby The C-terminus of TRC8 is able to catalyse ubiqui- TRC8 expression (Figure 4g), consistent with G2/M tylation reactions in vitro in keeping with the presence of arrest (Figures 2c and4f). ATM is activatedby a highly conservedRING-H2 domain (Lorick et al., radiation-induced DNA damage or by reactive oxygen 1999) (Figure 3). Although it is likely that TRC8 species (Falck et al., 2001; Ito et al., 2004). ATM kinase generates K48-linkedpoly-Ub leadingto proteasome- phosphorylates Chk2 andactivates p53 andBRCA1, mediated degradation, this has not yet been established. inhibits CDC25A/C phosphatases andinhibits cell cycle It is also possible that different E2s and/or substrates progression (Matsuoka et al., 1998; Falck et al., 2001). cause TRC8 to generate alternative linkages (e.g., K63) How TRC8 induces ATM phosphorylation is unknown, with alternative fates (Ciechanover, 2006), such as protein transport, processing or activation as in the nuclear factor-kB pathway. 2 þ Table 1 Cell cycle analysis of TRC8 mutations in HEK293 cells As with other RING proteins, mutation of the Zn - coordinating Cys and His residues eliminates ubiquity- a b b b Cell line Apoptosis G2/M G1/S P-values lation activity of TRC8 (Lorick et al., 1999; Zheng et al., Uninduced 2.170.2 1870.3 79.970.4 Wildtype c 17.770.5 32.572.3 49.872.8 o0.0001 CAICd 2.170.2 17.770.5 80.270.7 Table 2 TRC8 inhibits HMGCR andFAS mRNA SARc 12.370.5 30.870.3 56.970.8 o0.0001 LRKd 1.470.1 18.470.2 80.270.3 Gene Uninduced a TRC8b Fold change CPMCd 1.570.1 18.370.2 80.270.3 TRC8 0.4 43.5 >100 Abbreviation: HEK ¼ human embryonic kidney cell line. aExperiments HMGCR 3.3 0.8 À4.3 performedin triplicate. bPercent of total cell population, 7s.d. cCells FAS 1.0 0.2 À5.3 expressing TRCSwt or TRC8SAR showedsignificantly increased apoptotic andG2/M populations andsignificantly decreasedG1/S Abbreviations: FAS ¼ fatty acidsynthase; HMGCR ¼ 3-hydroxy-3- (Po0.0001, one-way analysis of variance) comparedto uninduced methylglutaryl coenzyme A reductase. aMessage levels reportedas cells. dLoss-of-function RING mutations were not significantly percent of GAPDH. bThirty-six hours induction with 100 ng/ml different from uninduced cells. doxycycline.

Figure 5 TRC8 suppression of tumorigenicity is RING-dependent. (a) Four groups of 12 mice each were inoculatedsubcutaneously with 293FI cells, two groups with TRC8wt andtwo with TRC8 CAIC. One experimental group for each cell line was treatedcontinuously with dox in the cage water. Percent of mice with tumors in each treated group are reported; tumor numbers for the untreated groups were similar to the treatedTRC8 CAIC group (not shown). (b) Mean tumor diameters for all four groups were measured at 34 days post- inoculation. Similar results were obtained at 22 days with average diameters of 7 mm.

Oncogene RING-dependent tumor suppression by TRC8 A Brauweiler et al 2269

Figure 6 ActivatedSREBP partially restores growth to TRC8 inhibitedcells. ( a) HEK293FI cells containing TRC8wt were transfected, in triplicate, with vector alone (pCMV7) or with pCMV7-nSREBP-1 expressing the nuclear, activated form of SREBP-1a (nSREBP). Cells were simultaneously induced to express TRC8 with dox treatment (1 mg/ml). Viable cell counts were performed48 h post-transfection/induction. Differences in growth between uninduced and induced cells (untransfected with nSREBP1a) were statistically significant (**Po0.001) as were differences in growth between uninduced and induced cells transfected with nSREBP1a (*Po0.01, one-way analysis of variance). Data are representative of four repetitions. (b) Cultures of 293FI cells containing TRC8wt or TRC8CAIC were co-transfectedwith the pSYNSRE-luc reporter construct with or without pCMV7-nSREBP1. Induction of luciferase activity verifiedthe nSREBP1a constructs. Protein expression of nSREBP1a was verifiedby Western blot; asterisk indicatesa background band indicative of even loading.

2000). Importantly, all the mutations that lost this This process depends upon pre-SREBP binding of activity also lost the biological effects (Figure 4d–g). SCAP along with a cholesterol-mediated interaction Moreover, the TRC8CAIC RING mutation lost the ability with INSIG. Interestingly, ccRCCs are characterizedby to inhibit tumor formation (Figure 5). These results lipidinclusions composedprimarily of newly synthe- support the hypothesis that the ubiquitylation activity of sizedcholesterol esters andother neutral lipids(Clay- the RING domain is essential for TRC8 function. man et al., 1986; Gebhard et al., 1987), suggesting that The LRK mutation, which affects three non-Zn2 þ - this lipidhomeostasis pathway is deregulatedin the coordinating residues adjacent to the third Zn-binding most common type of kidney tumor. Moreover, the half-site (Figure 3a), eliminatedubiquitylation activity mevalonate pathway initiatedby HMGCR is necessary while the SAR mutation retainednearly 30% of wild- for isoprenoidbiosynthesis andprenylation of small type activity. These sites were chosen for mutagenesis on GTPase proteins such as Ras andRap. Thus, inhibition the basis of crystallographic andnuclear magnetic of the SREBP pathway by TRC8 couldhave important resonance (NMR) structural data suggesting their consequences for proliferation. importance for interactions with E2 Ub-conjugating We thus testedthe ability of activated,nSREBP1a to enzymes (Zheng et al., 2000; Katoh et al., 2003). Our restore growth to TRC8-inhibitedcells (Figure 6). preliminary data suggest that the TRC8 RING domain Transfection of nSREBP1a was able to restore some can utilize a number of E2 enzymes for ubiquitylation, growth to HEK293 cells. These observations suggest at least in vitro (K. Lorick, unpublishedobservations). that TRC8 Ub ligase andtumor suppressor activities As the ubiquitylation assays in Figure 3 testedonly may form one connection between growth control and Ube2d2 (UbcH5b), it is possible that the LRK and SAR lipidhomeostasis. However, growth restoration was not mutations retain the capability to utilize other E2’s. This complete, suggesting that while SREBP destabilization may explain why SAR, with strongly reduced Ube2d2- has an important role, additional mechanisms are likely dependent activity, nevertheless retained nearly wild- important for TRC8-mediated inhibitory effects. Sur- type G2/M andapoptotic effects. prisingly, the SREBP pathway has recently been We recently discovered that TRC8 induction reduced identiﬁed as the major oxygen-sensing mechanism in the expression of SREBP target genes (Lee et al., ﬁssion yeast (Hughes et al., 2005; Todd et al., 2006). In submitted, and Table 2). In mammalian cells, cholester- preliminary experiments, we foundthat severe hypoxia ol andlipidhomeostasis is regulatedby an ER complex induced some SREBP target genes and that ectopic including Insulin-Induced Gene (INSIG), SREBP TRC8 couldblock this (H Drabkin, unpublished Cleavage Activating Protein (SCAP) andSREBPs observations). These observations support a connection (Goldstein et al., 2006). Sterol starvation induces between SREBP-mediated lipid biosynthesis and hypox- cholesterol andfatty acidbiosynthetic enzymes by ia responses in mammalian cells. The major mechanism activating the latent transcription factor activity of the of renal cancer development involves mutation of VHL SREBPs through a unique mechanism of cholesterol- andderegulation of HIF- a (Kaelin, 2004). Thus, we sensitive transport andintra-membrane cleavage in the suggest that hypoxia may deregulate fatty acid and Golgi (Horton et al., 2002, 2003; Goldstein et al., 2006). sterol biosynthesis andthat by modulating the SREBP

Oncogene RING-dependent tumor suppression by TRC8 A Brauweiler et al 2270 homeostatic pathway, TRC8 may link lipidstatus and on an FC500 flow cytometer (Beckman Coulter, Fullerton, hypoxia responses with growth control. CA, USA). Results were analysedusing the CXP software package. For BrdU incorporation, cells were incubated with BrdU (Sigma-Aldrich, St Louis, MO, USA), washed, fixed, permeabilized and stained with anti-BrdU antibody (BD Materials and methods Biosciences, San Jose, CA, USA). Secondary was Alexa 488-conjugatedF(ab 0)2 fragments of goat anti-mouse Expression constructs immunoglobulin G. TRC8-HA in GFPC1 (Gemmill et al., 2002) was transferredto the pAdTrack-CMV vector using KpnI. The CAIC mutation was createdin full-length TRC8 (without the HA-tag) by site- Tumorigenicity assay directed mutagenesis (QuikChange kit, Stratagene, La Jolla, Four groups of 12 mice each were inoculatedsubcutaneously CA, USA). Sequence confirmedpAdTrackclones were with 5 Â 106 293FI cells containing either dox-inducible electroporatedinto pAdEasy-containingBJ5183 cells, selected TRC8wt or TRC8CAIC. Two groups (one group per line) were for recombination andtransfectedinto HEK293 cells for virus pretreatedfor 24 h with 1 mg/ml doxin water containing 1% production. TRC8-HA insert was transferred to pcDNA5/ sucrose, which was changedevery 2–3 days.Examinations for FRT/TO using KpnI andverifiedby sequence analysis. Site- tumor take were performedthree times a week andtumor directed mutagenesis also generated all RING mutations in diameters were measured at 22 and 34 days. both full-length TRC8-HA andin GST-TRC8 513À664,aC- terminal fusion described previously along with the GST-AO7 RING fusion (Lorick et al., 1999). Full-length gp78 insert was Supplemental methods transferredto pcDNA5/FRT/TO from pCIneo-gp78 (Fang Detailed descriptions of antibodies, real-time PCR primers et al., 2001) using NotI. (Ohira et al., 2003), in vitro ubiquitylation reactions (Lorick et al., 1999) andcell culturing are providedinSupplemental Methods. Generation of TET-on FlpIn TRex cell lines HEK293FI lines carrying TET-inducible constructs were generatedby co-transfection of pcDNA5/FRT/TO with Acknowledgements pOG44 encoding Flp-recombinase (1:10 pcDNA5:pOG44) using FuGENE6 (Roche Diagnostics, Indianapolis, IN, We thank H Schulze for expression plasmids, A Kraft for USA). After 2 days, recombinant cells were selected in 50 mg/ vectors and293FI cells. This investigation usedthe DNA ml hygromycin B (Invitrogen, Carlsbad, CA, USA) and 5 mg/ Sequencing, Flow Cytometry andBiostatistics Cores of the ml blasticidin. Dox-inducible expression of HA-tagged pro- University of Colorado Cancer Center. We thank J Jacobsen, teins was verifiedfor each construct by Western blot. C Korch, K Helm, M Ashton, C Childs and C Zeng for technical assistance. KLL, YCT andAMW are supportedby Flow cytometric analysis and BrdU labelling the Intramural Research Program of the Center for Cancer Harvested cells were washed and resuspended in propidium Research, NCI, NIH. HAD andRMG are supportedby NIH iodide with 0.3% saponin. Cell cycle analysis was carried out RO1 CA076035.

References

Ciechanover A. (2006). The ubiquitin proteolytic system: from from the endoplasmic reticulum. Proc Natl Acad Sci USA a vague idea, through basic mechanisms, and onto human 98: 14422–14427. diseases and drug targeting. Neurology 66: S7–S19. GebhardRL, Clayman RV, Prigge WF, Figenshau R, Clayman RV, Bilhartz LE, Buja LM, Spady DK, Dietschy Staley NA, Reesey C et al. (1987). Abnormal cholesterol JM. (1986). Renal cell carcinoma in the Wistar–Lewis rat: a metabolism in renal clear cell carcinoma. J Lipid Res 28: model for studying the mechanisms of cholesterol acquisi- 1177–1184. tion by a tumor in vivo. Cancer Res 46: 2958–2963. Gemmill RM, Bemis LT, Lee JP, Sozen MA, Baron A, Zeng C CliffordSC, Astuti D, Hooper L, Maxwell PH, Ratcliffe PJ, et al. (2002). The TRC8 hereditary kidney cancer gene Maher ER. (2001). The pVHL-associatedSCF ubiquitin suppresses growth andfunctions with VHL in a common ligase complex: molecular genetic analysis of elongin B and pathway. Oncogene 21: 3507–3516. C, Rbx1 andHIF-1alpha in renal cell carcinoma. Oncogene Gemmill RM, Lee JP, Chamovitz DA, Segal D, Hooper JE, 20: 5067–5074. Drabkin HA. (2005). Growth suppression induced by the Cohen AJ, Li FP, Berg S, Marchetto DJ, Tsai S, Jacobs SC TRC8 hereditary kidney cancer gene is dependent upon et al. (1979). Hereditary renal-cell carcinoma associated with JAB1/CSN5. Oncogene 24: 3503–3511. a chromosomal translocation. N Engl J Med 301: 592–595. Gemmill RM, West JD, Boldog F, Tanaka N, Robinson LJ, Ettinger SL, Sobel R, Whitmore TG, Akbari M, Bradley DR, Smith DI et al. (1998). The hereditary renal cell carcinoma Gleave ME et al. (2004). Dysregulation of sterol response 3;8 translocation fuses FHIT to a patched-related gene, element-binding proteins and downstream effectors in TRC8. Proc Natl Acad Sci USA 95: 9572–9577. prostate cancer during progression to androgen indepen- Goldstein JL, DeBose-Boyd RA, Brown MS. (2006). Protein dence. Cancer Res 64: 2212–2221. sensors for membrane sterols. Cell 124: 35–46. Falck J, MailandN, Syljuasen RG, Bartek J, Lukas J. (2001). Horton JD, Goldstein JL, Brown MS. (2002). SREBPs: The ATM-Chk2-Cdc25A checkpoint pathway guards activators of the complete program of cholesterol andfatty against radioresistant DNA synthesis. Nature 410: 842–847. acidsynthesis in the liver. J Clin Invest 109: 1125–1131. Fang S, Ferrone M, Yang C, Jensen JP, Tiwari S, Weissman Horton JD, Shah NA, Warrington JA, Anderson NN, Park AM. (2001). The tumor autocrine motility factor receptor, SW, Brown MS et al. (2003). Combinedanalysis of gp78, is a ubiquitin protein ligase implicated in degradation oligonucleotide microarray data from transgenic and

Oncogene RING-dependent tumor suppression by TRC8 A Brauweiler et al 2271 knockout mice identifies direct SREBP target genes. Proc Ma D, Nutt CL, Shanehsaz P, Peng X, Louis DN, Kaetzel Natl Acad Sci USA 100: 12027–12032. DM. (2005). Autocrine platelet-derived growth factor- Hughes AL, Todd BL, Espenshade PJ. (2005). SREBP dependent gene expression in glioblastoma cells is mediated pathway responds to sterols and functions as an oxygen largely by activation of the transcription factor sterol sensor in fission yeast. Cell 120: 831–842. regulatory element binding protein and is associated with Ito K, Hirao A, Arai F, Matsuoka S, Takubo K, Hamaguchi I alteredgenotype andpatient survival in human brain et al. (2004). Regulation of oxidative stress by ATM is tumors. Cancer Res 65: 5523–5534. requiredfor self-renewal of haematopoietic stem cells. Mandriota SJ, Turner KJ, Davies DR, Murray PG, Morgan NV, Nature 431: 997–1002. Sowter HM et al. (2002). HIF activation identifies early Kaelin Jr WG. (2004). The von Hippel–Lindau tumor lesions in VHL kidneys: evidence for site-specific tumor suppressor gene andkidneycancer. Clin Cancer Res 10: suppressor function in the nephron. Cancer Cell 1: 459–468. 6290S–6295S. Martel PM, Bingham CM, McGraw CJ, Baker CL, Kaelin Jr WG. (2005). The von Hippel–Lindau protein, HIF Morganelli PM, Meng ML et al. (2006). S14 protein in hydroxylation, and oxygen sensing. Biochem Biophys Res breast cancer cells: direct evidence of regulation by Commun 338: 627–638. SREBP-1c, superinduction with progestin, and effects on Katoh S, Hong C, Tsunoda Y, Murata K, Takai R, Minami E cell growth. Exp Cell Res 312: 278–288. et al. (2003). High precision NMR structure andfunction of Matsuoka S, Huang M, Elledge SJ. (1998). Linkage of ATM the RING-H2 finger domain of EL5, a rice protein whose to cell cycle regulation by the Chk2 protein kinase. Science expression is increasedupon exposure to pathogen-derived 282: 1893–1897. oligosaccharides. J Biol Chem 278: 15341–15348. Ohira T, Gemmill RM, Ferguson K, Kusy S, Roche J, Li FP, Decker HJ, Zbar B, Stanton Jr VP, Kovacs G, Brambilla E et al. (2003). WNT7a induces E-cadherin in Seizinger BR et al. (1993). Clinical andgenetic studiesof lung cancer cells. Proc Natl Acad Sci USA 100: 10429–10434. renal cell carcinomas in a family with a constitutional Todd BL, Stewart EV, Burg JS, Hughes AL, Espenshade PJ. chromosome 3;8 translocation. Genetics of familial renal (2006). Sterol regulatory element binding protein is a carcinoma. Ann Intern Med 118: 106–111. principal regulator of anaerobic gene expression in fission Lorick KL, Jensen JP, Fang S, Ong AM, Hatakeyama S, yeast. Mol Cell Biol 26: 2817–2831. Weissman AM. (1999). RING fingers mediate ubiquitin- Zheng N, Wang P, Jeffrey PD, Pavletich NP. (2000). Structure conjugating enzyme (E2)-dependent ubiquitination. Proc of a c-Cbl-UbcH7 complex: RING domain function in Natl Acad Sci USA 96: 11364–11369. ubiquitin–protein ligases. Cell 102: 533–539.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene