VHL suppresses RAPTOR and inhibits mTORC1 signaling in clear cell renal cell carcinoma

Inactivation of the tumor suppressor von Hippel–Lindau (VHL) gene is a key event in hereditary and sporadic clear cell renal cell carcinomas (ccRCC). The mechanistic target of rapamycin (mTOR) signaling pathway is a fundamental regulator of cell growth and proliferation, and hyperactivation of mTOR signaling is a common finding in VHL-dependent ccRCC. Deregulation of mTOR signaling correlates with tumor progression and poor outcome in patients with ccRCC. Here, we report that the regulatory-associated protein of mTOR (RAPTOR) is strikingly repressed by VHL. VHL interacts with RAPTOR and increases RAPTOR degradation by ubiquitination, thereby inhibiting mTORC1 signaling. Consistent with hyperactivation of mTORC1 signaling in VHL-deficient ccRCC, we observed that loss of vhl-1 function in C. elegans increased mTORC1 activity, supporting an evolutionary conserved mechanism. Our work reveals important new mechanistic insight into deregulation of mTORC1 signaling in ccRCC and links VHL directly to the control of RAPTOR/mTORC1. This may represent a novel mechanism whereby loss of VHL affects organ integrity and tumor behavior.

Renal cell carcinoma (RCC) causes more than 140,000 deaths per year worldwide 1 . RCC have several histologic subtypes, with the most common one (> 80%) being clear cell RCC (ccRCC). Bi-allelic inactivation of the von Hippel Lindau (VHL) gene is a hallmark event that arises in the majority of sporadic ccRCC cases implicating the VHL gene as the most important renal tumor suppressor gene in general 2,3 . Mutations of VHL are also causative for the inherited autosomal dominant von Hippel-Lindau syndrome manifested by a variety of benign and malignant tumors including ccRCC.
The VHL protein functions as the substrate recognition subunit of an E3 ubiquitin ligase complex targeting the hypoxia inducible transcription factor (HIF) α subunits for proteasomal degradation under normal oxygen levels. Genetic VHL inactivation leads to constitutive HIFα accumulation and consequently formation of HIF heterodimers, which induce transcription programs promoting bioenergetic adaptation to hypoxia. Reprogramming of numerous cellular systems by HIF contributes to the pathogenesis of ccRCC by altering metabolism, angiogenesis, extracellular matrix, invasion and apoptosis-resistance. Accumulating evidence indicates that VHL inactivation is not sufficient to cause renal tumor formation. Additional genetic events and cellular alterations are required as second hits for malignant transformation. Comprehensive genomic analyses of ccRCC identified epigenetic control and PI3K-mechanistic Target of Rapamycin (mTOR) pathways as major determinants in ccRCC pathogenesis 4,5 .
In ccRCC the mTOR pathway is commonly hyperactivated 6,7 . Dysregulation of mTORC1 signaling plays a key role in the oncogenesis and progression of ccRCC, and hyperactivation of mTOR correlates with poor outcome in ccRCC patients. Hence, mTOR inhibitors (such as everolimus and temsirolimus) have been approved for treatment of advanced RCCs, but therapy resistance develops in most patients 8 .
The mTOR kinase is a highly conserved and fundamental regulator of cell growth, metabolism, and proliferation in all eukaryotes 9 . mTOR constitutes the catalytic subunit of two distinct complexes, mTORC1 and mTORC2. mTORC1 contains three core components: mTOR, mLST8 and its unique defining subunit, the regulatoryassoiated protein of mTOR (RAPTOR). Diverse extra-and intra-cellular signals activate mTORC1, including nutrient availability and growth factors, while hypoxia and low cellular energy levels inhibit mTORC1 activity. The primary role of mTORC1 is to initiate biosynthesis cascades for proteins, lipids and nucleotides to support cell growth, while also suppressing catabolic pathways like autophagy 10  www.nature.com/scientificreports/ The mechanisms underlying increased mTOR signaling activity in ccRCC have remained unclear. Genomic studies have found that ~ 26% of ccRCC harbor mutations in a number of PI3K-AKT-mTORC1 pathway genes 2,5 . Genetic alterations thus likely contribute to mTORC1 activation in ccRCC. Further integrated molecular studies of ccRCC revealed high levels of AKT-mTORC1 signaling also without an associated genetic alteration 11 suggesting that additional molecular players and upstream signals are involved. Notably, VHL was recently found to directly suppress AKT activity, and in VHL-deficient cells AKT was activated promoting cell survival and tumorigenesis 12 . HIF is generally thought to inhibit mTORC1 in hypoxia by activating the expression of its downstream target gene REDD1, which subsequently activates the mTORC1 repressor tuberous sclerosis complex 13 . However, mTOR can evade inhibition by REDD1 14 . On the other hand HIF2α increases mTORC1 activity under low amino acid availability by increasing the expression of the SLC7A5 amino acid carrier 15 . It was also shown that suppression of the mTOR inhibitor DEPTOR in VHL-deficient ccRCC accelerated tumor cell proliferation 16 .
This study reveals new mechanistic insights into deregulation of mTORC1 in ccRCC. Combining cellular models for ccRCC and the C. elegans system we identify an additional layer of interaction of VHL and the PI3K-mTORC1 pathway and directly link VHL to control of RAPTOR, the essential scaffolding protein of mTORC1.

Results
VHL interacts with the mTORC1 subunit RAPTOR. mTORC1 is frequently hyperactivated in ccRCC and accelerates tumor progression. mTORC1 consists of the three core components mTOR, RAPTOR, and LST8. To test whether VHL can physically interact with one or more of mTORC1 core proteins, HEK293T cells were co-transfected with tagged expression constructs of VHL, RAPTOR, mTOR, and LST8 respectively. Immunoblotting analysis showed association of VHL with RAPTOR, mTOR (Fig. 1a,b), and LST8 ( Supplementary  Fig. 1a). Binding of VHL to the inhibitory subunits DEPTOR (DEP domain-containing mTOR-interacting protein) and PRAS40 (proline-rich Akt substrate of 40 kDa) was not observed (Supplementary. Figure 1b). As we noticed constant reduction of RAPTOR in the presence of VHL in our cell lysates by immunoblotting, we evaluated RAPTOR protein levels in different ccRCC cell lines. Interestingly, RAPTOR expression was increased in VHL-deficient ccRCC cell lines compared to human renal proximal tubular epithelial cells (RPTECs) (Fig. 1c). Consistent with these observations, bioinformatic analysis using the online web portal UALCAN (http:// ualcan. path. uab. edu) revealed a significant upregulation of RAPTOR protein in tumor tissues of VHL-dependent ccRCCs when compared to normal tissues ( Fig. 1d) according to the CPTAC mass-spectrometry-based proteomic tumor dataset 17 . These findings led us to investigate the interaction between VHL and RAPTOR in more detail.
The VHL gene encodes two biologically active isoforms, full length VHL consisting of 213 amino acid residues with a molecular mass of 30 kDa (VHL30) and an internally translated form corresponding to amino acid residues 54-213 with a molecular mass of 19 kDa (VHL19) (Supplementary Fig. 1c) 18 . Both isoforms act similarly to promote HIFα degradation and so appear to retain tumor suppressor activity, yet have isoformspecific functions 19,20 . HEK293T cells were transiently transfected with Myc-tagged RAPTOR and Flag-tagged VHL30 or VHL19. Following immunoprecipitation with Flag-antibody, immunoblotting analysis revealed that the full length VHL30 efficiently immunoprecipitates RAPTOR, while the shorter VHL19 is able to immunoprecipitate only a very small amount of RAPTOR protein, despite being expressed at similar levels to VHL30 (Fig. 1e). A similar experiment was conducted with the N-terminal fragment of VHL consisting of amino acid residues 1-53 (VHL(AA1-53)). Again, immunoprecipitation of VHL30 led to co-precipitation of RAPTOR, Figure 1. VHL interacts and co-localizes with RAPTOR. (a) VHL binds RAPTOR. HEK293T cells were co-transfected with Flag.VHL and Myc.RAPTOR, or Flag.Luciferase as control. Cell lysates were used for detecting RAPTOR expression by anti-Myc antibody and for immunoprecipitation (IP) by anti-Flag antibody. The lower panel shows the IP of Flag-tagged proteins. Co-immunoprecipitation of RAPTOR was detected by anti-Myc (middle panel). kD, kilodalton. Full-length blots are presented in Supplementary Fig. 4. (b) VHL interacts with mTOR. Flag.mTOR and V5.VHL were transiently co-expressed in HEK293T cells. V5.Luciferase was used as control. After immunoprecipitation (IP) with anti-V5 antibody, the immobilized mTOR was detected by Western blot (WB) analysis using anti-Flag antibody in the precipitate containing VHL, but not Luciferase (middle panel). Full-length blots are presented in Supplementary Fig. 4. (c) RAPTOR expression in upregulated in VHL-deficient ccRCC cell lines. Cell lysates of RCC4, 786-O, A-498 and RPTEC cells were analyzed by Western blot using anti-RAPTOR and anti-mTOR antibodies. Equal concentrations of total protein were determined by Bradford assay. β-actin and γ-tubulin levels were used as a loading control. Full-length blots are presented in Supplementary Fig. 4. (d) RAPTOR expression data from the UALCAN database comparing ccRCC to normal tissue, ***p ≤ 0.001. (e) VHL30 but not VHL19 interacts with RAPTOR. Myc.RAPTOR and Flag.VHL30 or Flag.VHL19 were transiently co-expressed in HEK293T cells. Flag.MO-25 was used as control. After immunoprecipitation (IP) with anti-Flag antibody, the immobilized RAPTOR was detected by Western blot (WB) analysis using anti-Myc antibody in the precipitate containing VHL30, but not VHL19 (middle panel). Full-length blots are presented in Supplementary Fig. 4. (f) Endogenous RAPTOR interacts with VHL. Endogenous VHL was immunoprecipitated (IP) from HeLa cell lysates by anti-VHL antibody (lower panel). Endogenous RAPTOR co-precipitating with VHL was detected by Western blot analysis using anti-RAPTOR antibody (middle panel). Full-length blots are presented in Supplementary Fig. 4. (g) Co-localization of VHL and RAPTOR. HeLa cells were transfected with GFP-tagged RAPTOR (green, upper panel) and RFP-tagged VHL (red, middle panel). Co-expression of VHL and RAPTOR revealed co-localization in peri-nuclear foci (lower panel). Fluorescent confocal microscopy images of HeLa cells. Hoechst (blue) was used for nuclei staining. Scale bar 10 µm. www.nature.com/scientificreports/ while VHL(AA1-53) did not associate with RAPTOR ( Supplementary Fig. 1d). Together, these findings show that the VHL30 N-terminal tail, although necessary, is not sufficient for RAPTOR binding. To further confirm the interaction between VHL and RAPTOR we tested the association of endogenous proteins. HeLa cell extracts were subjected to immunoprecipitation with anti-VHL antibodies and immunoblotting analysis with a RAPTORspecific antibody revealed the binding of endogenous RAPTOR to VHL (Fig. 1f). RAPTOR accumulates in small granules in the cytoplasm that have previously been identified as Golgi and the endoplasmic reticulum (ER) 21,22 ( Fig. 1g). VHL is predominantly cytoplasmic and can shuttle to the nucleus. In the cytoplasm, VHL proteins were shown to display ER localization 23 . Co-transfection of VHL and RAPTOR revealed a partial co-localization in cytoplasmic foci near the nucleus potentially relating to the ER (Fig. 1g).

VHL regulates RAPTOR protein abundance.
Our experiments so far suggest that VHL may have an impact on mTORC1 and most notably RAPTOR. Following this hypothesis we analyzed whether VHL alters RAPTOR protein levels under different cellular conditions. As observed before, overexpressed VHL significantly reduced the levels of RAPTOR in HEK293T cells (Fig. 2a). To confirm the effect of VHL on RAPTOR protein abundance endogenous RAPTOR protein levels were also measured. Overexpression of VHL resulted in reduced levels of endogenous RAPTOR (Fig. 2b). Since only full length VHL30 but not the short form VHL19 interacted with RAPTOR, we tested whether the VHL protein isoforms differ in their impact on RAPTOR protein abundance. Consistently, VHL30 markedly diminished RAPTOR protein levels, while VHL19 only had a weak effect (Fig. 2c). Given the prominent regulation of RAPTOR by VHL, RAPTOR protein levels in 786-O ccRCC cells were investigated. 786-O cells are a well-established cell model for ccRCC and have been previously used to show that mTOR activation drives tumorigenesis in ccRCC 7 . We re-introduced a retrovirus encoding full length VHL cDNA (VHL30) and observed downregulation of RAPTOR protein abundance as compared with VHL-deficient 786-O tumor cells (Fig. 2d). Together, these observations indicate that VHL suppresses RAPTOR protein abundance. The risk for developing ccRCC seems to be linked to VHL genotype: type 1 mutations (large deletions and premature stops) and type 2B missense mutations are associated with a high risk, while type 2A missense mutations have lower risk 24 . To understand the basis for the suppression of RAPTOR by VHL we tested the ability of the most frequent VHL type 2A (Y98H and Y112H) and type 2B (Y98N and Y112N) representative mutants 25 to regulate RAPTOR. The type 2A mutations Y98 H and Y112 H retained the ability to reduce RAPTOR protein levels to the same extent as wildtype VHL (Fig. 2e). Likewise, the type 2B VHL(Y98N) mutant form still suppressed RAPTOR protein levels, whereas overexpression of VHL(Y112N) mutant failed to reduce RAPTOR abundance (Fig. 2e). Mutations in residues Y98 and Y112 have been shown to affect the interaction with the key cellular substrate HIFα 25 , but we observed similar binding of VHL type 2B mutations to RAPTOR ( Supplementary Fig. 2).
VHL impairs mTORC1 signaling. To further investigate the regulation of RAPTOR by VHL, HeLa cells were infected with lenti-VHL shRNA or lenti-control shRNA. shRNA-mediated VHL knockdown efficiency was verified by qPCR and, as expected, depletion of VHL increased HIF1α ( Supplementary Fig. 3a,b). In VHLdepleted cells RAPTOR protein levels were strongly upregulated (Fig. 3a). We also tested whether the changes in RAPTOR levels were due to altered transcription. However, knockdown of VHL in HeLa cells and re-expression of VHL in 786-O cells did not substantially change RPTOR mRNA expression (Fig. 3b,c). These results indicate that VHL modulates RAPTOR at the post-transcriptional level. mTORC1 directly phosphorylates and activates a set of well-characterized targets, most notably p70 S6 kinase (p70 S6K). Upon shRNA-mediated knockdown of VHL an increase in p70 S6K phosphorylation was observed (Fig. 3d). Immunoblotting analysis of VHL-deficient RCC4 cells lines confirmed the activation of mTORC1 HEK293T cells were co-transfected with Myc.RAPTOR and Flag.VHL. Cell lysates were analyzed by immunoblotting using Myc-and Flag-specific antibodies. Protein levels of RAPTOR were quantified using LabImage 1D software and normalized to β-actin protein levels. The lower panel shows quantification of relative RAPTOR protein levels from three independent experiments. Data are presented as mean ± SEM. p < 0.05 by t test. kD, kilodalton. Full-length blots are presented in Supplementary Fig. 4. (b) VHL diminished endogenous RAPTOR levels. HEK293T cells were transiently co-transfected with Flag.VHL and serum starved overnight. Endogenous RAPTOR was detected by a specific anti-RAPTOR antibody. Lower panel shows the quantification of RAPTOR levels from three independent experiments, p < 0.05 (t test). Full-length blots are presented in Supplementary Fig. 4. (c) Full length VHL30 but not VHL19 regulates RAPTOR protein abundance. Immunoblotting analysis of HEK293T cells co-transfected with expression vectors for RAPTOR and control protein (Luciferase), VHL30 or VHL19. Lower panel shows the quantification of relative RAPTOR levels from three independent experiments, p < 0.05; n.s., not significant (t test). Full-length blots are presented in Supplementary Fig. 4 www.nature.com/scientificreports/ signaling and re-expression of VHL significantly decreased phosphorylation of p70 S6K and its target ribosomal protein S6 (Fig. 3e). Consistently, overexpression of VHL in HEK293T cells significantly reduced the phosphorylation of p70 S6K (Fig. 3f). Cell proliferation rates were not affected by manipulating VHL in different cell lines ( Supplementary Fig. 3c) indicating that proliferation does not account for the VHL-dependent RAPTOR abundance and mTORC1 activation. Taken together, this experimental evidence supports the idea that VHL associates with RAPTOR and impairs mTORC1 signaling.
To test whether the interplay between VHL and RAPTOR in our system was HIF-dependent, ccRCC cells were treated with the HIF2α isoform specific inhibitor PT2385 26 .Treatment of RCC4 and 786-O cells with PT2385 significantly reduced HIF2α activity, as evidenced by decreased expression of the target gene PAI-1 16,27 , but did not alter RAPTOR protein levels or mRNA expression (Fig. 3g,h, and Supplementary Fig. 3d,e). HIF1α was inhibited by IDF-11774 28 , again no significant differences in RAPTOR protein expression were observed (Supplementary Fig. 3f.). Taken together, these results point to a HIF-independent link between VHL and RAPTOR.
In ccRCC, HIF-mediated upregulation of the mTORC1 inhibitor REDD1 has been described 14 . Consistent with the literature, REDD1 levels were upregulated in VHL-deficient 786-O cells, but these cells still expressed RAPTOR and showed high mTORC1 activity ( Supplementary Fig. 3g) 12,14 indicating that VHL suppresses RAP-TOR in a REDD1-independent manner. VHL-1 impairs mTORC1 signaling in C. elegans. To gain further insight into the significance of the interaction between VHL and mTORC1 signaling pathways in vivo, we used the genetically tractable model C. elegans. Core components of the VHL and mTOR signaling pathways are highly evolutionary conserved in C. elegans and this model organism has provided comprehensive insights into their fundamental functions [29][30][31][32] .
First, we analyzed whether the direct association between VHL and RAPTOR is conserved between species. Co-immunoprecipitation experiments from HEK293T cells transiently transfected with DAF-15, the C. elegans homolog of human RAPTOR, and VHL-1 were performed. Indeed, DAF-15/RAPTOR co-precipitated with VHL-1 (Fig. 4a). To investigate the functional connection between VHL and mTORC1 signaling we analyzed canonical mTORC1 downstream targets. Monitoring the phosphorylation status of p70 S6 kinase has been previously utilized to characterize mTORC1 activity in C. elegans 33,34 . Inactivation of vhl-1 enhanced the phosphorylation level of p70 S6K compared to wild type (Fig. 4b) consistent with hyperactivation of mTORC1 in vhl-1 mutants. We next examined whether VHL-1 might regulate daf-15/Raptor transcription by performing quantitative PCR. The mRNA expression of daf-15/Raptor was not changed in vhl-1 and hif-1 mutants compared to wild type (Fig. 4c). Hence, the activation of the mTORC1 pathway in vhl-1 mutants was not due alteration of corresponding daf-15/Raptor mRNA levels. mTORC1 is essential for larval growth and development in C. elegans and previous studies have shown that loss of daf-15/Raptor causes developmental arrest at L3 larval-stage 35 . We inactivated daf-15/Raptor and vhl-1 together and found that a substantial portion of larvae (approx. 42%) progressed through development to adulthood, while all larvae from daf-15/Raptor deficient animals arrested at L3 stage (Fig. 4d, f, h). vhl-1 single mutants did not display growth defects (Fig. 4g). The observation that mutation of vhl-1 suppresses the developmental defects associated with loss of daf-15/Raptor suggests that the genes function in the same pathway. mTORC1 is a well-described inhibitor of autophagy, a cellular recycling pathway degrading cytoplasmic material.
LGG-1, the C. elegans homolog of mammalian LC3 has been widely used as an indicator of autophagy 36,37 . To monitor the function of vhl-1 in autophagy we used a reporter strain expressing mCherry-tagged LGG-1 in the intestine 38 . Upon induction of autophagy, mCherry::LGG-1 changes its diffuse cytoplasmic distribution pattern to punctate structures reflecting autophagosomal localization of LGG-1. As expected, mCherry::LGG-1 was diffusely localized in intestinal cells of fully fed wild type worms and relocalized into punctae after starvation, a potent inducer of intestinal autophagosome formation (Fig. 4i,j,l). Well-fed vhl-1 mutants also displayed a diffuse www.nature.com/scientificreports/ cytoplasmic LGG-1 distribution (Fig. 4k). Importantly, upon starvation we observed a significant decrease of mCherry:: LGG-1 positive foci in vhl-1 mutants compared to wild type ( Fig. 4i,l,m). Together, these results provide strong evidence for the evolutionary conservation of the link between VHL and mTORC1 signaling and indicate that VHL-1 serves as an inhibitor for mTORC1 in multiple species.

RAPTOR is a novel target of the VHL E3 ubiquitin ligase.
Our results indicate that VHL regulates RAPTOR protein abundance. Next, we sought to investigate the underlying mechanism. First, the decay kinetic of RAPTOR protein after treatment with the translation inhibitor cycloheximide was analyzed. Upon VHLoverexpression RAPTOR levels were significantly diminished with time by cycloheximide treatment, while RAPTOR abundance was unchanged under basal conditions (Fig. 5a). Previous work has shown that AMPK phosphorylates RAPTOR, induces binding to 14-3-3 and inhibits mTORC1 activity 39 . VHL also associated with AMPK ( Supplementary Fig. 1b) but did not interfere with AMPK-mediated RAPTOR phosphorylation (Fig. 5b). VHL functions as the substrate recognition subunit of an E3 ubiquitin ligase complex targeting HIFα and other proteins for proteasomal degradation. To test whether the regulation of RAPTOR by VHL depends on the function of the proteasome, RAPTOR levels following treatment with the proteasome inhibitor ALLN were analyzed. While overexpression of VHL strongly suppressed RAPTOR protein levels, inhibition of the proteasome by ALLN prevented VHL-mediated RAPTOR degradation (Fig. 5c). These data suggest that RAPTOR could be a specific target of the VHL ubiquitin ligase complex. Therefore, we next examined whether RAPTOR is ubiquitinated by VHL. Flag-tagged RAPTOR, V5-tagged VHL, and HA-ubiquitin were transfected into HEK293 cells. We detected RAPTOR polyubiquitination, and VHL overexpression further increased ubiquitination of RAPTOR (Fig. 5d). Together our data reveal a novel mechanism where VHL increases ubiquitination and degradation of RAPTOR and thereby limits mTORC1 signaling (Fig. 5e).

Discussion
Hyperactivation of mTORC1 signaling in VHL-deficient ccRCC is well described in the literature 6,7 although the functional inter-connection between VHL and mTORC1 and their mutual regulation are still ill defined. Integrating biochemical assays with molecular renal cancer cell analyses we show here that VHL binds to the key subunits of mTORC1, limits RAPTOR protein abundance, and suppresses mTORC1 signaling. Importantly, our C. elegans analyses demonstrate that the link between VHL and RAPTOR/mTORC1 is conserved from nematodes to human. We propose a new mechanism by which VHL regulates mTORC1 signaling: VHL-mediated ubiquitinylation and degradation of RAPTOR may result in impaired mTORC1 activity. Our investigation identified RAPTOR as a new VHL substrate in ccRCC. The endogenous levels of RAPTOR were elevated in VHL-deficient ccRCC cells and VHL shRNA cells (Figs. 2d, 3a). The in vivo significance of our findings was corroborated by increased RAPTOR expression in ccRCC tumor samples (Fig. 1d). Conversely, transient over-expression of VHL resulted in a concomitant decrease in RAPTOR protein (Fig. 2a-c). The near complete depletion of RAPTOR following treatment with cycloheximide within 6 h strongly suggests that VHL regulation occurs at the post-translational level (Fig. 5a), while mRNA expression of RPTOR was not affected in different cell lines (Fig. 3b,c). Moreover the regulation of RAPTOR by VHL was HIFα independent as pharmacological inhibition of HIFα did not alter RAPTOR abundance in ccRCC cells (Fig. 3g,h and Supplementary  Fig. 3d,e,f). Inhibiting the proteasome partially rescued RAPTOR levels (Fig. 5c). That RAPTOR is ubiquitylated by the VHL-associated E3 ligase is demonstrated by our finding that overexpression of VHL increases its ubiquitylation (Fig. 5d). Our data demonstrate VHL's ability to recognize and ubiquitinylate RAPTOR and  www.nature.com/scientificreports/ consequently to target it for proteasomal degradation. Of note, RAPTOR level has been shown previously to correlate with mTORC1 activity 40,41 . VHL has been reported to regulate additional oncogenic substrates besides the well-established HIFs such as ZHX2 42 , AKT 12 and SFMBT1 43 . Hence the function of VHL appears to be multi-faced, including both E3 ligase-dependent and -independent functions. Post-translational modification of mTOR pathway members by ubiquitinylation has been shown to modulate mTOR signaling activity. mTOR, the essential kinase of both mTORC1 and mTORC2 complexes, is targeted for ubiquitinylation and degradation by binding to the tumor suppressor FBXW7 44 . In colorectal carcinoma, downregulation of the E3 ligase FBX8 correlated with enhanced mTOR activity and might thereby promote invasion and metastasis 45 . RICTOR, the integral component of mTORC2, is also degraded through an FBXW7-mediated ubiquitination/proteasome mechanism 46 . RAPTOR protein is reportedly modified by ubiquitylation [47][48][49] , but the mechanism how the ubiquitin pathway governs RAPTOR/mTORC1 activation remains unclear. The DDB1-CUL4 ubiquitin ligase complex has been shown to interact with RAPTOR, but it seems to impact upon mTORC1 signaling indirectly through a non-degradative mechanism by affecting the assembly of the mTORC1 complex 50 . In this study we uncovered a novel VHL-mediated regulation of mTORC1 by targeted RAPTOR ubiquitination and degradation.
Our data support the notion that VHL30 and VHL19 have different functional specializations. We connect full length VHL30 with RAPTOR and mTORC1 pathway regulation while VHL19 lacks this interaction (Fig. 2c). Both isoforms act as tumor suppressor inhibiting cancer development when a wild type copy is reintroduced in ccRCC, but isoform-specific VHL functions are emerging from the literature 51 . VHL30 has been shown to interact with p53 and p14ARF 51,52 to control cell cycle and apoptosis. Moreover, VHL30 co-localizes predominantly with cytoplasmic microtubules and alters microtubule dynamics 53 . Our findings show that VHL30 executes RAPTOR regulating functions, suggesting that VHL30 may contribute independently to tumor suppression in specific contexts. Moreover, we found that VHL mutations with altered binding capacity to HIFα 25,54 differentially regulated RAPTOR protein levels. Particularly, the VHL mutant Y112N did not lead to RAPTOR destabilization, in contrast to the VHL mutants Y98N, Y98H and Y112H which reduced RAPTOR protein comparable to wild type VHL (Fig. 2e). The mutational hot spot residues Y98 and Y112 make important contributions to HIFα binding 55,56 and have been shown to affect substrate interactions to different extents with type 2B mutations (Y98N and Y112N) causing more severe defects and reduced ubiquitylation activity 25 . The binding of other ubiquitinylation targets to the same key surface of VHL may be affected likewise. In our study VHL mutants still associated with RAPTOR ( Supplementary Fig. 2) albeit overexpression of VHL mutants could mask differences in binding affinity and activity. This remains to be determined in a more rigorous approach. Nevertheless, the effect of VHL on RAPTOR/mTORC1 complex stability suggests an additional mechanism for tumorigenesis in VHL-dependent ccRCC and part of the phenotypic variability observed in VHL disease may be due differential impact of VHL mutations on oncogenic pathways.
Remarkably, the mechanism of RAPTOR/mTORC1 inhibition by VHL is conserved in evolution between nematodes and mammals. Consistent with the hyperactivation of the mTORC1 pathway in VHL-deficient ccRCC cells, vhl-1 C. elegans mutants displayed increased phosphorylation of ribosomal S6 kinase, which represents the main activity of the mTORC1 pathway (Fig. 4b). The primary role of mTORC1 is to promote growth-related processes and development as well as to inhibit autophagy. In fact developmental defects of Raptor/daf-15 deficient worms were recovered by knockdown of vhl-1 (Fig. 4d-h) and starvation-induced autophagy was markedly suppressed by inactivation of vhl-1 (Fig. 4i,l,m). Together, our C. elegans findings support the inhibitory function of VHL in mTORC1 pathway regulation. The nematode C. elegans has been widely used to elucidate the molecular function of mTOR signaling proteins and their hierarchical order in the pathways 29 . C. elegans overcomes many problems of VHL-studies in cell culture and mouse models and offers a powerful tool in studying molecular and genetic aspects of the VHL-mTOR network 32 . Exploiting the genetic screening possible in C. elegans might provide further insight how mTOR influences phenotype and disease progression with VHL, and potentially reveal novel concepts to treat VHL-driven cancer.
Studies during the last decade have highlighted that mTORC1 signaling is hyperactivated in ccRCC. mTORC1 hyperactivation correlates with clinico-pathological parameters and poor outcome in ccRCC patients. The induction of mTORC1 signaling in ccRCC may occur at several levels and by multiple mechanisms: First, genomic studies have identified genetic alterations activating the mTORC1 pathway including MTOR, PTEN, AKT, and PIK3CA in ~ 26% of ccRCC cases 2 . Second, HIF-mediated aberrant expression of mTORC1 upstream genes including the mTOR inhibitor DEPTOR 16 , REDD1 13,14 and the amino acid carrier SLC7A5 15 has been shown to modulate mTORC1 signaling activity. Next, essential HIF-independent mechanisms have also been described recently. VHL can directly bind and inhibit AKT kinase activity. In VHL-deficient cells AKT was activated, promoting cell survival and tumorigenesis 12 . In this study we describe a novel VHL mechanism for regulation of mTORC1 signaling in vitro and in vivo by limiting RAPTOR protein abundance. The relative contribution of these pathways to mTORC1 activation in ccRCC remains to be determined.

Material and methods
Reagents and plasmids. Cycloheximide was obtained from Sigma, ALLN from Calbiochem, MG-132 from Enzo Life Sciences, DMOG from Frontier Scientific. PT2385 was purchased from Abcam and IDF-11774 from Hycultec. All reagents were used at concentrations as indicated.
Myc.RAPTOR was a gift from David Sabatinini (Addgene plasmid # 1859) 40 and RAPTOR cDNA was fused by standard cloning techniques to a pcDNA6 vector encoding an N-terminal GFP or Flag-tag (Invitrogen). pcDNA3-Flag mTOR wt was a gift from Jie Chen (Addgene plasmid # 26603) 57 . VHL30 was cloned into pcDNA6 vector with N-terminal Flag-, GFP-and V5-tag, and C-terminal RFP-tag. VHL(AA1-53) and VHL19 were generated by standard cloning techniques and fused to the appropriate vectors. Full-length human VHL(Y98N), VHL(Y112N), Antibodies. Antibodies used in this study included antibody to Flag (Sigma), antibody to Myc (Clone 9E10 Santa Cruz Biotechnology), mouse antibody to GFP (B-2, Santa Cruz Biotechnology), rabbit antibody to GFP (MBL), antibody to RAPTOR (Cell Signaling), rabbit antibody to VHL (Santa Cruz Biotechnology FL-181 and Cell Signaling), antibody to β-actin (Sigma), antibody to γ-tubulin (Sigma), rabbit antibody to V5 (Millipore), and mouse antibody to HA Clone12CA5 (Roche). The antibodies to mTOR, p70 S6 kinase, phospho-p70 S6 Kinase, phospho-S6 ribosomal protein, phospho-RAPTOR S792, REDD1 and phospho-drosophila p70 S6 (Thr398) kinase were all obtained from Cell Signaling, the antibody against HIF1α from BD Transduction Laboratories and the antibody against HIF2α from Abcam. For immunoprecipitation experiments, cells were transfected with the indicated plasmids, washed with PBS, lysed in IP buffer and after centrifugation (13.000 rpm, 15 min, 4 °C followed by 43,000 rpm, 30 min, 4 °C) the supernatants were incubated with anti-FLAG M2 affinity beads (Sigma) or V5 beads (Abcam) overnight at 4 °C. For GFP-tagged protein, lysates were incubated with GPF antibody (MBL) overnight, followed by incubation with Protein A beads (GE Healthcare) for 2 h at 4 °C. All the precipitates were washed extensively with IP buffer and subjected to SDS-PAGE and immunoblotting analysis with anti-Flag, anti-myc, anti-V5 or anti-GFP antibodies.
Immunoprecipitation experiments from endogenous proteins were previously described 63 . Briefly, for detection of the interaction between VHL and RAPTOR, after lysis with IP-buffer four dishes of confluent HeLa cells were pooled for each condition. After centrifugation (13.000 rpm, 15 min, 4 °C followed by 43,000 rpm, 30 min, 4 °C) the supernatants were incubated with anti-VHL antibody (Santa Cruz Biotechnology) or normal rabbit IgG (Santa Cruz Biotechnology) overnight. After addition and incubation with Protein A beads (GE Healthcare) for 2 h, 4 °C, the precipitates were washed and analyzed by Western blot with antibodies against RAPTOR and VHL.
For the quantification of total amounts of the overexpressed and endogenous proteins, cells were split in parallel, lysed in the buffer described above and analyzed by immunoblotting. The blots were scanned and bands were quantified using LabImage 1D software. Data were expressed as means or as means ± SEM. The statistical analysis was performed using SigmaPlot 11.0 software.
To analyze the turnover of RAPTOR, 24 h after transfection cells were treated with cycloheximide in DMEM for the indicated time points followed by lysis in IP-buffer as described in reference 60 . Proteins were fractionated by SDS/PAGE, and protein levels were analyzed by Western blot.
Ubiquitination assays were performed as previously described 59 . Briefly, HEK 293 T cells were transfected with the plasmids as indicated along with the HA-tagged ubiquitin construct and treated with MG-132 (5 µM, overnight). 24 h after transfection, the cells were washed with PBS and lysed in RIPA buffer (1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 50 mM NaF, 2 mM EDTA, 13.7 mM Na2HPO4, 6.3 mM NaH2PO4). The lysates were clarified by ultracentrifugation and incubated with Flag-M2 affinity beads at 4 °C for 2 h. After washing extensively with RIPA buffer, precipitates were subjected to SDS-PAGE and immunoblotting analysis with anti-Flag, anti-V5 and anti-HA antibodies.
shRNA stable polyclonal cell line. To generate a HeLa cell line for tetracycline-inducible knockdown of VHL, a lentivirus-based transduction system (pLVTH) was used as previously described 64 . The VHL shRNA targeting sequence was 5' ACA CAG GAG CGC ATT GCA CAT3' , the control targeting sequence was 5′ GTA CGC GGA ATA CTT CGA 3' . www.nature.com/scientificreports/