Reports

The phosphatidylinositol 3′-OH kinase (PI3K)/AKT path-way is a key regulator of cell survival, proliferation and growth and is commonly activated in human cancers by receptor tyrosine kinase signaling (1, 2). To identify new regulators of PI3K/AKT signaling, we performed an RNA interference (RNAi) screen using lentiviruses that express short hairpin RNAs (shRNAs) (3) targeting genes coding for all known G-proteins and lipid/protein kinases. Because AKT phosphorylation is a faithful indicator of PI3K activity, we used an immunofluorescent approach to quantitatively measure phosphorylation of AKT at serine 473 (S473) in HeLa cells, which display robust growth factor induced AKT phosphorylation and lack mutations in core PI3K signaling pathway genes (Fig. 1A). We screened a collection of 7,450 shRNAs in an arrayed format in triplicate to identify kinases or GTPases whose depletion could alter PI3K-dependent AKT phosphorylation.

To identify shRNAs that altered AKT phosphorylation (i.e., “hits”), the phospho-AKT signal in each screened well was translated to a z-score (4). shRNAs targeting genes that positively regulate AKT phosphorylation—RICTOR, PIK3CA,

PIK3CB, AKT1, and AKT3—all diminished AKT phosphoryla-tion and scored as “pAKT low” hits (Fig. 1B and tables S1 to S3). Conversely, shRNAs targeting genes whose protein products inhibit AKT signaling—such as RAPTOR and RHEB—elevated phospho-AKT levels and scored as “pAKT high” hits. Thus, our screen successfully identified known regulators of PI3K/AKT signaling as well as a number of genes not previously implicated in regulation of the pathway (ta-ble S4). We prioritized hits for genes that had no reported link to PI3K/AKT signaling (table S5) and searched oncogenomic da-tabases to identify those with somatic mutations reported in human cancers (5, 6). This left us with a list of 29 genes (table S6), 26 of which were “pAKT low” hits. Next, we grouped the 26 “pAKT low” hits by their anno-tated functions from the litera-ture and found that the best represented functional group was the RAB GTPases, which are well known to regulate intracel-lular protein trafficking (Fig. 1C). This suggested to us that dysregulated protein trafficking

by mutant RAB proteins could play a role in oncogenesis. Of the five RAB proteins that appeared in our list in Fig.

1C, we pursued RAB35 because, in addition to meeting all of our criteria already mentioned, it is a well-studied GTPase for which reagents and tools are readily available. RAB35 is a regulator of cytoskeletal organization and trafficking at the recycling endosome (7–10). Endomembrane trafficking by RAB35 is regulated by the DENND1 family of RAB35 guanine exchange factors (GEFs) (8, 11–13) and by the TBC1D10 family of RAB35 GTPase activating proteins (GAPs) (14–16). Because RAB35 was a robust hit in our screen (fig. S1A), has not been implicated as a regulator of PI3K/AKT signaling, and somatic mutations have been re-ported in human cancers that map to residues conserved in other small GTPases (Fig. 1D and table S7), we prioritized RAB35 for further study.

Next, we confirmed that depletion of RAB35 with the same shRNAs that scored in the original screen suppressed AKT phosphorylation, and to a similar extent as depletion of the mTORC2 component RICTOR (Fig. 1E). Furthermore, additional shRAB35 hairpins not present in the original

Identification of an oncogenic RAB protein Douglas B. Wheeler,1,2* Roberto Zoncu,3 David E. Root,4 David M. Sabatini,4,5,6,7,8† Charles L. Sawyers1,8† 1Human Oncology and Pathogenesis Program, Memorial Sloan-Kettering Cancer Center, New York, NY 10065, USA. 2Weill-Cornell/Rockefeller University/Sloan-Kettering Tri-Institutional MD-PhD Program, New York, NY 10021, USA. 3Department of Molecular and Cell Biology, University of California at Berkeley, Berkeley, CA 94720, USA. 4Broad Institute of MIT and Harvard, Cambridge, MA 02142, USA. 5Whitehead Institute for Biomedical Research, Cambridge, MA 02142, USA. 6Department of Biology, Massachusetts Institute of Technology (MIT), Cambridge, MA 02142, USA. 7David H. Koch Institute for Integrative Cancer Research at MIT, Cambridge, MA 02142, USA. 8Howard Hughes Medical Institute, Chevy Chase, MD 20815, USA.

*Present address: Dana-Farber Cancer Institute, Boston, MA 02215, USA and Broad Institute of MIT and Harvard, Cambridge, MA 02142, USA.

†Corresponding author. E-mail: sawyersc@mskcc.org (C.L.S.) or sabatini@wi.mit.edu (D.M.S.)

In an shRNA screen for genes that affect AKT phosphorylation, we identified the RAB35 small GTPase—a protein previously implicated in endomembrane trafficking—as a regulator of the PI3K pathway. Depletion of RAB35 suppresses AKT phosphorylation in response to growth factors, whereas expression of a dominant active GTPase-deficient mutant of RAB35 constitutively activates the PI3K/AKT pathway. RAB35 functions downstream of growth factor receptors and upstream of PDK1 and mTORC2 and co-purifies with PI3K in immunoprecipitation assays. Two somatic RAB35 mutations found in human tumors generate alleles that constitutively activate PI3K/AKT signaling, suppress apoptosis, and transform cells in a PI3K-dependent manner. Furthermore, oncogenic RAB35 is sufficient to drive PDGFRα to LAMP2-positive endomembranes in the absence of ligand, suggesting there may be latent oncogenic potential in dysregulated endomembrane trafficking.

/ www.sciencemag.org/content/early/recent / 3 September 2015 / Page 1 / 10.1126/science.aaa4903

screen also suppressed AKT phosphorylation as well as the AKT substrate FOXO1/3A (fig. S1B). Thus, multiple inde-pendent shRNAs targeting RAB35 consistently impair AKT phosphorylation in HeLa cells.

To ensure that the effect of RAB35 knockdown on AKT signaling was not cell-type specific, we depleted RAB35 pro-tein in human HEK-293E cells and in mouse NIH-3T3 cells (Fig. 2, A and B). Knockdowns of RAB35 with two different shRNAs diminished AKT phosphorylation at S473 in re-sponse to serum stimulation in both human and mouse cells, and did so more potently than RICTOR. Furthermore, the PI3K-dependent phosphorylation of FOXO1/3A and the SGK1 substrate NDRG-1 was also decreased in cells depleted of RAB35. We expanded this analysis to 7 additional cancer cell lines with different mutational backgrounds (oncogenic PI3KCA or KRAS, deleted PTEN, mutated NF2 or TP53) and generally found reduced AKT phosphorylation after RAB35 depletion, with some modest differences likely attributable to distinct hairpins in distinct cellular contexts (fig. S2, A to C). Furthermore, AKT phosphorylation at the PDK1 phos-phorylation site T308 was consistently reduced by RAB35 knockdown, suggesting that RAB35 may function upstream of PDK1. Collectively these data indicate that RAB35 is broadly necessary for the efficient activation of PI3K/AKT signaling in response to serum stimulation.

To explore whether RAB35 is sufficient to activate the PI3K/AKT pathway, we generated cell lines that stably ex-pressed either wild-type RAB35 (RAB35wt) or the dominant active GTPase-deficient, GTP-bound RAB35 Q67L mutant (RAB35Q67L). Cells expressing the wild-type RHEB GTPase (RHEBwt), which regulates mTORC1, served as a control. Stable expression of either RHEBwt or RAB35wt did not alter phosphorylation of AKT or ERK in response to growth fac-tor signaling (Fig. 2C and fig. S3). However, expression of the active RAB35Q67L mutant rendered AKT phosphorylation (at both T308 and S473) constitutively elevated and refrac-tory to growth factor deprivation but did not impact ERK phosphorylation. FOXO1/3A phosphorylation was also con-sistently elevated in serum-deprived cells expressing RAB35Q67L. Therefore, the expression of GTP-bound RAB35 is sufficient to activate PI3K/AKT signaling in cells even in the absence of growth factors.

The fact that RAB35Q67L results in constitutive phosphor-ylation of T308 as well as S473 suggests that RAB35 may function upstream of the T308 kinase PDK1 (17, 18). Howev-er, attribution of RAB35 function to PDK1 versus mTORC2 activation is challenging because the phosphorylation states of T308 and S473 are interdependent (19–21). To address this issue, we took advantage of cells that stably express al-leles of murine AKT1 (mAkt1) in which either T308 or S473 were mutated to phosphomimetic aspartate residues (22) (fig. S4, A and B). As expected, mTOR blockade with the TORC1/2 inhibitor Torin1 or by RICTOR depletion (resulting in loss of mTORC2 function) suppressed phosphorylation at both sites in mAkt1wt expressing cells (fig. S4C). In

mAkt1S473D expressing cells, which lack a regulated mTORC2 site, these same interventions either had no impact on T308 phosphorylation (RICTOR depletion) or enhanced T308 phosphorylation (Torin1).

Having confirmed the ability of this system to distin-guish direct effects on T308 versus S473, we examined the impact of RAB35 knockdown. Unlike RICTOR or Torin1, depletion of RAB35 decreased T308 phosphorylation in mAkt1wt and in mAkt1S473D expressing cells. RAB35 depletion also decreased S473 phosphorylation in mAkt1wt and in mAKTT308D expressing cells (fig. S4C). These data suggest that RAB35 signals to AKT by acting upstream of both PDK1 and mTORC2.

We reasoned that if RAB35 acts upstream of both PDK1 and mTORC2, it may be controlling their common regula-tor, PI3K. Although we could not detect an interaction be-tween endogenous RAB35 and PI3K, we found that stably expressed RAB35—but not RHEB—co-purified with PI3K when the kinase was immunoprecipitated by its p85 subunit (Fig. 2D). Interestingly, more FLAG-RAB35 co-purified with PI3K in immunoprecipitates from cells expressing constitu-tively GTP-bound RAB35Q67L. Conversely, we found that the same experiment performed using lysates from cells stably expressing a constitutively GDP-bound allele of RAB35 (RAB35S22N) did not recover any FLAG-RAB35. Thus, the physical association of RAB35 and PI3K is nucleotide de-pendent. We also asked whether RAB35 is necessary for PI3K kinase activity in vitro. As expected, PI3K immuno-purified from cells depleted of PI3Kα had significantly re-duced in vitro kinase activity toward PIP2 (fig. S5A). PI3K purified from cells depleted of RAB35 had an approximately 50% reduction in in vitro PI3K kinase activity. Further, we found that depletion of RAB35 did not reduce the amount of p110α recovered in p85 immunoprecipitates, which suggest-ed to us that RAB35 depletion does not disrupt the p85-p110α interaction (fig. S5B). Thus, RAB35 co-purifies with PI3K in a nucleotide-dependent manner and may be re-quired for optimal PI3K enzymatic function.

Given that RAB35 functions upstream of PI3K, we inves-tigated whether RAB35 acts downstream of any particular growth factor receptor. Depletion of RICTOR, PI3Kα, or RAB35 in HeLa cells reduced S473 phosphorylation of AKT in response to treatment with either insulin-like growth factor (IGF-I), epidermal growth factor (EGF), platelet-derived growth factor (PDGF-AA) or vascular endothelial growth factor (VEGF) (fig. S6A). Interestingly, RAB35 deple-tion did not dramatically alter phosphorylation of ERK or tyrosine phosphorylation of IGFR, EGFR, PDGFR or VEGFR (fig. S6, B and C). Collectively, these data suggest a model whereby RAB35 functions upstream of PI3K and down-stream of growth factor receptor tyrosine kinases to activate AKT.

One criterion for prioritization of genes on our hit list (table S6) was evidence for mutation in human cancer data-bases (table S7). For RAB35, missense mutations have been

/ sciencemag.org/content/early/recent / 3 September 2015 / Page 2 / 10.1126/science.aaa4903

reported at two residues that are conserved in RAS-like GTPases, A151T and F161L. We noticed that the A151T and F161L mutations in RAB35 were strikingly similar to two well documented KRAS mutations (A146T and F156L) that have been previously identified in human tumor samples (Fig. 3A) (23, 24). While they are not “canonical” activating mutations, stable expression of either of these two KRAS mutants was sufficient to activate ERK signaling and trans-form NIH-3T3 cells in vitro. We therefore asked if the two similar mutations in RAB35 might also be gain-of-function mutations that activate RAB35 signaling.

To test this possibility, we generated NIH-3T3 cells sta-bly expressing either RAB35wt, GTPase-deficient RAB35Q67L or the RAB35A151T and RAB35F161L alleles reported in human tumors. As expected from our earlier experiments in human cells, expression of RAB35wt did not activate AKT phosphor-ylation upon serum deprivation, whereas expression of the GTPase-deficient RAB35Q67L did (Fig. 3B). Remarkably, ex-pression of the two naturally occurring RAB35A151T and RAB35F161L mutants also elevated AKT phosphorylation lev-els during serum deprivation, indicating that both are gain-of-function alleles sufficient to activate PI3K/AKT signaling.

Because PI3K/AKT signaling inhibits apoptosis, we next examined whether cells expressing mutant RAB35 alleles are resistant to apoptosis triggered by growth factor with-drawal. 4 hours of serum deprivation of NIH-3T3 cells stably expressing RAB35wt was sufficient to elevate cleaved levels of the apoptotic markers PARP and Caspase3 (Fig. 3C). In comparison, cells stably expressing one of the three RAB35 mutants or oncogenic p110αH1047R had decreased levels of cleaved PARP and cleaved Caspase3 following serum with-drawal. Further, while cells expressing RAB35wt died in re-sponse to serum deprivation, cells expressing mutant alleles of RAB35 or p110αH1047R had significantly improved cell via-bility when deprived of growth factors (Fig. 3D). Taken to-gether, these data suggest that mutant RAB35 proteins can mitigate cell death in response to growth factor deprivation.

To determine if the mutant alleles of RAB35 are onco-genic, we examined their activity in a standard NIH-3T3 cell focus-formation assay—an in vitro model of density-independent growth that often correlates with tumorigenici-ty (25). Indeed, NIH-3T3 cells expressing oncogenic p110αH1047R, myristoylated AKT1 (AKT1myr) or any of the three mutant alleles of RAB35—but not RAB35wt—formed foci (Fig. 3E and fig. S7). Moreover, we found that when cultured in medium containing 250 nM of the pan-class I PI3K inhib-itor GDC-0941 (26), cells expressing RAB35 mutants and p110αH1047R were unable to form foci. Not surprisingly, PI3K blockade did not inhibit the growth of cells transduced with AKT1myr. Thus, the expression of the RAB35 mutants identi-fied in human cancers can transform NIH-3T3 cells in vitro in a PI3K-dependent manner.

Since RAB proteins regulate endomembrane trafficking, we considered whether our evidence implicating RAB35 in AKT activation, downstream of RTKs and upstream of PI3K,

is linked to this more established function of RAB proteins. The fact that many RTKs undergo ligand-stimulated inter-nalization and recycling through endosomes, while remain-ing competent to signal (27–31) suggested to us that RAB35 may function in this process. In considering this possibility, we noted highly elevated levels of tyrosine phosphorylation of the platelet-derived growth factor receptor (PDGFRα/β)—but not of IGF1R, FGFR, VEGFR or MET—in cells expressing the constitutively active RAB35Q67L allele (Fig. 4A and fig. S8). This increase in PDGFRα/β phosphorylation was mark-edly reduced by treatment of cells with crenolanib (CP-868596)—a PDGFR inhibitor that inhibits both PDGFRα and PDGFRβ with similar potency (32–34) (Fig. 4B and fig. S9). Remarkably, crenolanib treatment also diminished phosphorylation of AKT and the downstream substrate FOXO1/3A, suggesting that the RAB35Q67L allele activates AKT through PDGFR. Treatment with a more β-selective PDGFR inhibitor (CP-673451) only moderately affected AKT phosphorylation, implicating PDGFRα as the most likely RAB35Q67L target. The pan-PI3K inhibitor (GDC-0941) and the PI3Kα-specific inhibitor (BYL719) both suppressed AKT phosphorylation but not PDGFRα/β phosphorylation, providing further evidence that PDGFR/RAB35Q67L functions upstream of PI3K.

We postulated that RAB35Q67L might initiate PDGFR-mediated AKT activation by altering PDGFRα localization. To explore this possibility, we transfected cells that stably expressed either RAB35wt or RAB35Q67L with cDNA encoding for myc-tagged PDGFRα and looked at the localization of tagged PDGFRα in the presence or absence of the PDGFRα ligand PDGF-AA (fig. S10). As expected, in cells that ex-pressed RAB35wt, stimulation with PDGF-AA relocalized PDGFRα into punctate intracellular structures. Remarkably, PDGFRα was constitutively localized to punctate structures in RAB35Q67L expressing cells even in the absence of PDGF-AA stimulation. This effect was specific to PDGFRα because internalization of epidermal growth factor receptor (EGFR) was not altered by RAB35Q67L expression (fig. S11). We next asked if the punctate localization of PDGFRα in RAB35Q67L expressing cells was the same as that seen in the normal physiologic context of PDGFRα activation by PDGF-AA, by simultaneously visualizing PDGFRα-myc with the lysosomal protein LAMP2 or the early endosome marker EEA1. In both PDGF-AA ligand stimulated RAB35wt cells, and in unstimu-lated RAB35Q67L-expressing cells, transfected PDGFRα-myc colocalized with endogenous LAMP2 (Fig. 4, C and D, and fig. S12) and EEA1 (fig. S13). We also found that FLAG-tagged RAB35 is present in endomembranes that contain LAMP2 (fig. S14A) but, interestingly, not in those that con-tain EEA1 (fig. S14B). This suggests that PI3K activity down-stream of RAB35Q67L may originate on LAMP2-positive membranes. Finally, since it has been established that PI3K activity is required for the internalization and trafficking of PDGFR (35, 36), we asked whether the localization of PDG-FRα-myc in cells expressing RAB35Q67L was a result of the

/ sciencemag.org/content/early/recent / 3 September 2015 / Page 3 / 10.1126/science.aaa4903

increased PI3K signaling in these cells. Although PI3K was required for ligand-induced PDGFRα internalization in cells that express RAB35wt (fig. S15A), treatment of cells express-ing RAB35Q67L with a pan-PI3K inhibitor did not diminish the localization of PDGFRα-myc to LAMP2-positive com-partments (fig. S15B). Interestingly, PI3K inhibitor treat-ment resulted in changes in LAMP2 vesicle morphology that may merit further investigation (fig. S15, A and B). Together, these data establish that activated PDGFRα is normally traf-ficked through LAMP2-positive endomembranes and that constitutively active RAB35 likely drives unliganded PDG-FRα to this location where it is active and can signal to PI3K/AKT (Fig. 4E). Whether mutant RAB35 diverts newly synthesized PDGFRα from the cell membrane to LAMP2-positive endomembranes or prevents the exit of existing PDGFRα from LAMP2 compartments will require further study.

In summary we demonstrate a previously unappreciated role for the small GTPase RAB35 in regulating growth factor signaling to PI3K/AKT. We also characterize two RAB35 missense mutations found in human tumors that confer gain of function in standard transformation assays, display-ing phenotypes similar to well characterized oncogenic KRAS alleles. Searches of existing human tumor sequencing databases reveals that RAB35 mutations are rare (table S7), and are therefore unlikely to appear on current lists of hu-man cancer genes that rely solely on statistical methods to distinguish between driver and passenger mutations. None-theless, the conserved location of the RAB35 mutations to analogous residues in KRAS and their oncogenic phenotype in functional assays provides strong evidence that these are true driver mutations and perhaps even clinically “actiona-ble” based on sensitivity to PI3K inhibition. Considering the broad role of RAB family proteins in endomembrane traf-ficking, it is intriguing to speculate that the oncogenic RAB35 phenotype reported here may be a consequence of dysregulated membrane trafficking. Our studies showing that the constitutively active RAB35Q67L allele is sufficient to drive PDGFRα to LAMP2-positive endomembranes, in a pat-tern that mirrors that seen with native PDGF-AA ligand, is consistent with this model. Alternatively, these mutations may be neomorphic, conferring a novel molecular function to RAB35 that results in signaling in a RAS-like manner. The fact that other RAB proteins (RAB1B, RAB39, RABL3, RASEF) emerged as hits in our screen and have rare mis-sense mutations reported in human tumors (table S6) begs a broader examination of this question.

REFERENCES AND NOTES 1. R. J. Shaw, L. C. Cantley, Ras, PI(3)K and mTOR signalling controls tumour cell

growth. Nature 441, 424–430 (2006). Medline doi:10.1038/nature04869 2. L. C. Cantley, The phosphoinositide 3-kinase pathway. Science 296, 1655–1657

(2002). Medline doi:10.1126/science.296.5573.1655 3. J. Moffat, D. A. Grueneberg, X. Yang, S. Y. Kim, A. M. Kloepfer, G. Hinkle, B. Piqani,

T. M. Eisenhaure, B. Luo, J. K. Grenier, A. E. Carpenter, S. Y. Foo, S. A. Stewart, B. R. Stockwell, N. Hacohen, W. C. Hahn, E. S. Lander, D. M. Sabatini, D. E. Root, A lentiviral RNAi library for human and mouse genes applied to an arrayed viral

high-content screen. Cell 124, 1283–1298 (2006). Medline doi:10.1016/j.cell.2006.01.040

4. See supplementary methods for a detailed explanation of z-score calculation and hit criteria.

5. E. Cerami, J. Gao, U. Dogrusoz, B. E. Gross, S. O. Sumer, B. A. Aksoy, A. Jacobsen, C. J. Byrne, M. L. Heuer, E. Larsson, Y. Antipin, B. Reva, A. P. Goldberg, C. Sander, N. Schultz, The cBio cancer genomics portal: An open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2, 401–404 (2012). Medline doi:10.1158/2159-8290.CD-12-0095

6. S. A. Forbes, N. Bindal, S. Bamford, C. Cole, C. Y. Kok, D. Beare, M. Jia, R. Shepherd, K. Leung, A. Menzies, J. W. Teague, P. J. Campbell, M. R. Stratton, P. A. Futreal, COSMIC: Mining complete cancer genomes in the Catalogue of Somatic Mutations in Cancer. Nucleic Acids Res. 39 (suppl. 1), D945–D950 (2011). Medline doi:10.1093/nar/gkq929

7. D. Dambournet, M. Machicoane, L. Chesneau, M. Sachse, M. Rocancourt, A. El Marjou, E. Formstecher, R. Salomon, B. Goud, A. Echard, Rab35 GTPase and OCRL phosphatase remodel lipids and F-actin for successful cytokinesis. Nat. Cell Biol. 13, 981–988 (2011). Medline doi:10.1038/ncb2279

8. P. D. Allaire, A. L. Marat, C. Dall’Armi, G. Di Paolo, P. S. McPherson, B. Ritter, The Connecdenn DENN domain: A GEF for Rab35 mediating cargo-specific exit from early endosomes. Mol. Cell 37, 370–382 (2010). Medline doi:10.1016/j.molcel.2009.12.037

9. C. E. Chua, Y. S. Lim, B. L. Tang, Rab35—A vesicular traffic-regulating small GTPase with actin modulating roles. FEBS Lett. 584, 1–6 (2010). Medline doi:10.1016/j.febslet.2009.11.051

10. J. Zhang, M. Fonovic, K. Suyama, M. Bogyo, M. P. Scott, Rab35 controls actin bundling by recruiting fascin as an effector protein. Science 325, 1250–1254 (2009). Medline doi:10.1126/science.1174921

11. A. L. Marat, M. S. Ioannou, P. S. McPherson, Connecdenn 3/DENND1C binds actin linking Rab35 activation to the actin cytoskeleton. Mol. Biol. Cell 23, 163–175 (2012). Medline doi:10.1091/mbc.E11-05-0474

12. A. L. Marat, P. S. McPherson, The connecdenn family, Rab35 guanine nucleotide exchange factors interfacing with the clathrin machinery. J. Biol. Chem. 285, 10627–10637 (2010). Medline doi:10.1074/jbc.M109.050930

13. S. Yoshimura, A. Gerondopoulos, A. Linford, D. J. Rigden, F. A. Barr, Family-wide characterization of the DENN domain Rab GDP-GTP exchange factors. J. Cell Biol. 191, 367–381 (2010). Medline doi:10.1083/jcb.201008051

14. C. Hsu, Y. Morohashi, S. Yoshimura, N. Manrique-Hoyos, S. Jung, M. A. Lauterbach, M. Bakhti, M. Grønborg, W. Möbius, J. Rhee, F. A. Barr, M. Simons, Regulation of exosome secretion by Rab35 and its GTPase-activating proteins TBC1D10A-C. J. Cell Biol. 189, 223–232 (2010). Medline doi:10.1083/jcb.200911018

15. G. Patino-Lopez, X. Dong, K. Ben-Aissa, K. M. Bernot, T. Itoh, M. Fukuda, M. J. Kruhlak, L. E. Samelson, S. Shaw, Rab35 and its GAP EPI64C in T cells regulate receptor recycling and immunological synapse formation. J. Biol. Chem. 283, 18323–18330 (2008). Medline doi:10.1074/jbc.M800056200

16. M. Chaineau, M. S. Ioannou, P. S. McPherson, Rab35: GEFs, GAPs and effectors. Traffic 14, 1109–1117 (2013). Medline

17. J. R. Bayascas, D. R. Alessi, Regulation of Akt/PKB Ser473 phosphorylation. Mol. Cell 18, 143–145 (2005). Medline doi:10.1016/j.molcel.2005.03.020

18. A. Mora, D. Komander, D. M. van Aalten, D. R. Alessi, PDK1, the master regulator of AGC kinase signal transduction. Semin. Cell Dev. Biol. 15, 161–170 (2004). Medline doi:10.1016/j.semcdb.2003.12.022

19. R. M. Biondi, A. Kieloch, R. A. Currie, M. Deak, D. R. Alessi, The PIF-binding pocket in PDK1 is essential for activation of S6K and SGK, but not PKB. EMBO J. 20, 4380–4390 (2001). Medline doi:10.1093/emboj/20.16.4380

20. M. P. Scheid, P. A. Marignani, J. R. Woodgett, Multiple phosphoinositide 3-kinase-dependent steps in activation of protein kinase B. Mol. Cell. Biol. 22, 6247–6260 (2002). Medline doi:10.1128/MCB.22.17.6247-6260.2002

21. D. D. Sarbassov, D. A. Guertin, S. M. Ali, D. M. Sabatini, Phosphorylation and regulation of Akt/PKB by the rictor-mTOR complex. Science 307, 1098–1101 (2005). Medline doi:10.1126/science.1106148

22. D. D. Sarbassov, S. M. Ali, S. Sengupta, J. H. Sheen, P. P. Hsu, A. F. Bagley, A. L. Markhard, D. M. Sabatini, Prolonged rapamycin treatment inhibits mTORC2 assembly and Akt/PKB. Mol. Cell 22, 159–168 (2006). Medline doi:10.1016/j.molcel.2006.03.029

23. J. W. Tyner, H. Erickson, M. W. Deininger, S. G. Willis, C. A. Eide, R. L. Levine, M. C. Heinrich, N. Gattermann, D. G. Gilliland, B. J. Druker, M. M. Loriaux, High-

/ sciencemag.org/content/early/recent / 3 September 2015 / Page 4 / 10.1126/science.aaa4903

throughput sequencing screen reveals novel, transforming RAS mutations in myeloid leukemia patients. Blood 113, 1749–1755 (2009). Medline doi:10.1182/blood-2008-04-152157

24. M. Janakiraman, E. Vakiani, Z. Zeng, C. A. Pratilas, B. S. Taylor, D. Chitale, E. Halilovic, M. Wilson, K. Huberman, J. C. Ricarte Filho, Y. Persaud, D. A. Levine, J. A. Fagin, S. C. Jhanwar, J. M. Mariadason, A. Lash, M. Ladanyi, L. B. Saltz, A. Heguy, P. B. Paty, D. B. Solit, Genomic and biological characterization of exon 4 KRAS mutations in human cancer. Cancer Res. 70, 5901–5911 (2010). Medline doi:10.1158/0008-5472.CAN-10-0192

25. G. J. Clark, A. D. Cox, S. M. Graham, C. J. Der, Biological assays for Ras transformation. Methods Enzymol. 255, 395–412 (1995). Medline doi:10.1016/S0076-6879(95)55042-9

26. A. J. Folkes, K. Ahmadi, W. K. Alderton, S. Alix, S. J. Baker, G. Box, I. S. Chuckowree, P. A. Clarke, P. Depledge, S. A. Eccles, L. S. Friedman, A. Hayes, T. C. Hancox, A. Kugendradas, L. Lensun, P. Moore, A. G. Olivero, J. Pang, S. Patel, G. H. Pergl-Wilson, F. I. Raynaud, A. Robson, N. Saghir, L. Salphati, S. Sohal, M. H. Ultsch, M. Valenti, H. J. Wallweber, N. C. Wan, C. Wiesmann, P. Workman, A. Zhyvoloup, M. J. Zvelebil, S. J. Shuttleworth, The identification of 2-(1H-indazol-4-yl)-6-(4-methanesulfonyl-piperazin-1-ylmethyl)-4-morpholin-4-yl-thieno[3,2-d]pyrimidine (GDC-0941) as a potent, selective, orally bioavailable inhibitor of class I PI3 kinase for the treatment of cancer. J. Med. Chem. 51, 5522–5532 (2008). Medline doi:10.1021/jm800295d

27. M. D. Chamberlain, J. C. Oberg, L. A. Furber, S. F. Poland, A. D. Hawrysh, S. M. Knafelc, H. M. McBride, D. H. Anderson, Deregulation of Rab5 and Rab4 proteins in p85R274A-expressing cells alters PDGFR trafficking. Cell. Signal. 22, 1562–1575 (2010). Medline doi:10.1016/j.cellsig.2010.05.025

28. A. V. Vieira, C. Lamaze, S. L. Schmid, Control of EGF receptor signaling by clathrin-mediated endocytosis. Science 274, 2086–2089 (1996). Medline doi:10.1126/science.274.5295.2086

29. J. C. Chow, G. Condorelli, R. J. Smith, Insulin-like growth factor-I receptor internalization regulates signaling via the Shc/mitogen-activated protein kinase pathway, but not the insulin receptor substrate-1 pathway. J. Biol. Chem. 273, 4672–4680 (1998). Medline doi:10.1074/jbc.273.8.4672

30. C. L. Howe, J. S. Valletta, A. S. Rusnak, W. C. Mobley, NGF signaling from clathrin-coated vesicles: Evidence that signaling endosomes serve as a platform for the Ras-MAPK pathway. Neuron 32, 801–814 (2001). Medline doi:10.1016/S0896-6273(01)00526-8

31. S. Kermorgant, D. Zicha, P. J. Parker, PKC controls HGF-dependent c-Met traffic, signalling and cell migration. EMBO J. 23, 3721–3734 (2004). Medline doi:10.1038/sj.emboj.7600396

32. N. L. Lewis, L. D. Lewis, J. P. Eder, N. J. Reddy, F. Guo, K. J. Pierce, A. J. Olszanski, R. B. Cohen, Phase I study of the safety, tolerability, and pharmacokinetics of oral CP-868,596, a highly specific platelet-derived growth factor receptor tyrosine kinase inhibitor in patients with advanced cancers. J. Clin. Oncol. 27, 5262–5269 (2009). Medline doi:10.1200/JCO.2009.21.8487

33. M. C. Heinrich, D. Griffith, A. McKinley, J. Patterson, A. Presnell, A. Ramachandran, M. Debiec-Rychter, Crenolanib inhibits the drug-resistant PDGFRA D842V mutation associated with imatinib-resistant gastrointestinal stromal tumors. Clin. Cancer Res. 18, 4375–4384 (2012). Medline doi:10.1158/1078-0432.CCR-12-0625

34. J. Dai et al., Large-scale analysis of PDGFRA mutations in melanomas and evaluation of their sensitivity to tyrosine kinase inhibitors imatinib and crenolanib. Clin. Cancer Res. 19, 6935–6942 (2013). Medline doi:10.1158/1078-0432.CCR-13-1266

35. M. Joly, A. Kazlauskas, F. S. Fay, S. Corvera, Disruption of PDGF receptor trafficking by mutation of its PI-3 kinase binding sites. Science 263, 684–687 (1994). Medline doi:10.1126/science.8303278

36. M. Joly, A. Kazlauskas, S. Corvera, Phosphatidylinositol 3-kinase activity is required at a postendocytic step in platelet-derived growth factor receptor trafficking. J. Biol. Chem. 270, 13225–13230 (1995). Medline doi:10.1074/jbc.270.22.13225

37. T. R. Peterson, M. Laplante, C. C. Thoreen, Y. Sancak, S. A. Kang, W. M. Kuehl, N. S. Gray, D. M. Sabatini, DEPTOR is an mTOR inhibitor frequently overexpressed in multiple myeloma cells and required for their survival. Cell 137, 873–886 (2009). Medline doi:10.1016/j.cell.2009.03.046

38. J.-H. Zhang, T. D. Y. Chung, K. R. Oldenburg, A simple statistical parameter for use in evaluation and validation of high throughput screening assays. J. Biomol. Screen. 4, 67–73 (1999). Medline doi:10.1177/108705719900400206

39. J. Gao, B. A. Aksoy, U. Dogrusoz, G. Dresdner, B. Gross, S. O. Sumer, Y. Sun, A. Jacobsen, R. Sinha, E. Larsson, E. Cerami, C. Sander, N. Schultz, Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal. 6, pl1–pl1 (2013). Medline

40. S. A. Forbes, G. Tang, N. Bindal, S. Bamford, E. Dawson, C. Cole, C. Y. Kok, M. Jia, R. Ewing, A. Menzies, J. W. Teague, M. R. Stratton, P. A. Futreal, COSMIC (the Catalogue of Somatic Mutations in Cancer): A resource to investigate acquired mutations in human cancer. Nucleic Acids Res. 38 (suppl. 1), D652–D657 (2010). Medline doi:10.1093/nar/gkp995

41. S. A. Forbes et al., The Catalogue of Somatic Mutations in Cancer (COSMIC). Curr. Protoc. Hum. Genet. 10, 10.11 (2008).

42. S. Bamford, E. Dawson, S. Forbes, J. Clements, R. Pettett, A. Dogan, A. Flanagan, J. Teague, P. A. Futreal, M. R. Stratton, R. Wooster, The COSMIC (Catalogue of Somatic Mutations in Cancer) database and website. Br. J. Cancer 91, 355–358 (2004). Medline

43. D. Sonkin, M. Hassan, D. J. Murphy, T. V. Tatarinova, Tumor suppressors status in Cancer Cell Line Encyclopedia. Mol. Oncol. 7, 791–798 (2013). Medline

44. L. Marum, Cancer Cell Line Encyclopedia launched by Novartis and Broad Institute. Future Med Chem 4, 947 (2012). Medline

45. J. Barretina, G. Caponigro, N. Stransky, K. Venkatesan, A. A. Margolin, S. Kim, C. J. Wilson, J. Lehár, G. V. Kryukov, D. Sonkin, A. Reddy, M. Liu, L. Murray, M. F. Berger, J. E. Monahan, P. Morais, J. Meltzer, A. Korejwa, J. Jané-Valbuena, F. A. Mapa, J. Thibault, E. Bric-Furlong, P. Raman, A. Shipway, I. H. Engels, J. Cheng, G. K. Yu, J. Yu, P. Aspesi Jr., M. de Silva, K. Jagtap, M. D. Jones, L. Wang, C. Hatton, E. Palescandolo, S. Gupta, S. Mahan, C. Sougnez, R. C. Onofrio, T. Liefeld, L. MacConaill, W. Winckler, M. Reich, N. Li, J. P. Mesirov, S. B. Gabriel, G. Getz, K. Ardlie, V. Chan, V. E. Myer, B. L. Weber, J. Porter, M. Warmuth, P. Finan, J. L. Harris, M. Meyerson, T. R. Golub, M. P. Morrissey, W. R. Sellers, R. Schlegel, L. A. Garraway, The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 483, 603–607 (2012). Medline doi:10.1038/nature11003

46. Y. Sancak, T. R. Peterson, Y. D. Shaul, R. A. Lindquist, C. C. Thoreen, L. Bar-Peled, D. M. Sabatini, The Rag GTPases bind raptor and mediate amino acid signaling to mTORC1. Science 320, 1496–1501 (2008). Medline doi:10.1126/science.1157535

47. S. A. Stewart, D. M. Dykxhoorn, D. Palliser, H. Mizuno, E. Y. Yu, D. S. An, D. M. Sabatini, I. S. Chen, W. C. Hahn, P. A. Sharp, R. A. Weinberg, C. D. Novina, Lentivirus-delivered stable gene silencing by RNAi in primary cells. RNA 9, 493–501 (2003). Medline doi:10.1261/rna.2192803

48. S. M. Ali, D. M. Sabatini, Structure of S6 kinase 1 determines whether raptor-mTOR or rictor-mTOR phosphorylates its hydrophobic motif site. J. Biol. Chem. 280, 19445–19448 (2005). Medline doi:10.1074/jbc.C500125200

49. J. J. Zhao, O. V. Gjoerup, R. R. Subramanian, Y. Cheng, W. Chen, T. M. Roberts, W. C. Hahn, Human mammary epithelial cell transformation through the activation of phosphatidylinositol 3-kinase. Cancer Cell 3, 483–495 (2003). Medline doi:10.1016/S1535-6108(03)00088-6

50. A. D. Cox, C. J. Der, Biological assays for cellular transformation. Methods Enzymol. 238, 277–294 (1994). Medline doi:10.1016/0076-6879(94)38026-0

51. R. Zoncu, L. Bar-Peled, A. Efeyan, S. Wang, Y. Sancak, D. M. Sabatini, mTORC1 senses lysosomal amino acids through an inside-out mechanism that requires the vacuolar H+-ATPase. Science 334, 678–683 (2011). Medline doi:10.1126/science.1207056

52. E. M. M. Manders, F. J. Verbeek, J. A. Aten, Measurement of co-localization of objects in dual-colour confocal images. J. Microsc. 169, 375–382 (1993). doi:10.1111/j.1365-2818.1993.tb03313.x

53. J. H. McDonald, K. W. Dunn, Statistical tests for measures of colocalization in biological microscopy. J. Microsc. 252, 295–302 (2013). Medline doi:10.1111/jmi.12093

54. S. E. Maxwell, H. D. Delaney, Designing Experiments and Analyzing Data: A Model Comparison Perspective (Lawrence Erlbaum Associates, Mahwah, NJ, ed. 2, 2004).

ACKNOWLEDGMENTS

We thank all of the members of the Sabatini and Sawyers laboratories for critical readings of this manuscript and helpful discussions. We also thank N. D. Novikov, M. A. Domser, D. J. Spencer, T. R. Peterson, C. Thoreen, S. Kinkel, M. J. Evans, R. Bose, P. Watson, O. S. Andersen, S. C. Blanchard, T. E. McGraw, N. Rosen, and A. Smogorzewska for helpful discussions. We thank Serena Silver for assistance with the RNAi screens, and Vitaly Boyko and Yevgeniy Romin of the MSKCC

/ sciencemag.org/content/early/recent / 3 September 2015 / Page 5 / 10.1126/science.aaa4903

Molecular Cytology Core Facility for assistance with confocal microscopy. We also thank Andrius Kazlauskas for helpful discussions about PDGFR biology. Lastly, we thank the late Alan Hall for helpful input about GTPase biology. Funds for the RNAi screen were provided by the Starr Cancer Consortium Award #I2-A117. D.B.W. was supported by the Department of Defense Prostate Cancer Research Program via the Office of the Congressionally Directed Medical Research Programs Predoctoral Prostate Cancer Training Award PC094483; and N.I.H. Medical Scientist Training Program grant GM07739. R.Z. was supported by the Glenn Foundation for Medical Research, the NIH Director’s New Innovator Award 1DP2CA195761-01 and the Pew-Stewart Scholarship for Cancer Research. This work was partially supported by NIH grants CA103866 and AI47389 (to D.M.S) and CA155169 and CA092629 (to C.L.S.). Data from the RNAi screen described in this manuscript can be found in the supplementary materials.

SUPPLEMENTARY MATERIALS www.sciencemag.org/cgi/content/full/science.aaa4903/DC1 Materials and Methods Figs. S1 to S15 Tables S1 to S7 References (37–54) 14 December 2014; accepted 24 August 2015 Published online 3 September 2015 10.1126/science.aaa4903

/ sciencemag.org/content/early/recent / 3 September 2015 / Page 6 / 10.1126/science.aaa4903

Fig. 1. An arrayed lentiviral loss-of-function RNA interference screen identifies known and novel regulators of the PI3K/AKT axis. (A) An immunofluorescent assay for phospho-S473 AKT quantitatively reflects PI3K activity. HeLa cells grown in 384-well plates were serum starved overnight, treated with PIK-90 and insulin as indicated, then fixed and processed for immunofluorescence with a whole cell stain (red) and an anti-phospho-S473 AKT monoclonal antibody (green) (top). Signals were imaged and quantified with a LiCor Odyssey near-infrared scanner, and the phospho-AKT signal intensity was divided by the whole-cell stain signal intensity to calculate a normalized phospho-AKT value for each well (bottom). (B) Averaged (n = 3) phospho-AKT z-scores of screened shRNAs. HeLa cells were plated in 384-well plates, transduced with lentiviral reagents expressing control shRNAs or shRNAs targeting human kinases and GTPases, then processed as in (A). Phospho-AKT z-scores for each shRNA were then calculated, averaged, and arranged as indicated. (C) Functional grouping of the 26 genes whose knockdown lowered AKT phosphorylation, are not known to regulate PI3K/AKT signaling, and are mutated in human cancers. (D) Representation of somatic mutations in RAB35 in cancer found in oncogenomic databases. Mutations predicted by the cBioPortal for Cancer Genomics to have a medium effect in protein function are noted in black, while mutations predicted to have a high impact on protein function are noted in red. Table S7 contains a full list of somatic mutations of RAB35 found in cancers. (E) shRNAs targeting RAB35 deplete RAB35 protein levels and inhibit AKT phosphorylation. HeLa cells were transduced with lentiviruses expressing the indicated shRNAs, starved overnight and treated with serum as indicated, lysed and cell lysates were analyzed by immunoblotting. Red asterisks indicate hit shRNAs from screen.

/ sciencemag.org/content/early/recent / 3 September 2015 / Page 7 / 10.1126/science.aaa4903

Fig. 2. RAB35 is necessary and sufficient for serum-induced PI3K/AKT signaling and associates with PI3Kα. (A and B) Depletion of RAB35 inhibits serum-induced activation of AKT. (A) HeLa and HEK-293E cells were transduced with lentiviral reagents expressing the indicated shRNAs. Cells were serum starved overnight, treated as indicated with fetal bovine serum (FBS), lysed, and cell lysates were then analyzed by immunoblotting. (B) Murine NIH-3T3 cells were serum starved then treated as indicated with bovine calf serum (BCS) and analyzed as in (A). (C) GTPase-deficient RAB35Q67L activates PI3K/AKT signaling. HeLa and HEK-293E cells were stably transduced with lentiviruses expressing FLAG-tagged RHEBwt, RAB35wt or RAB35Q67L. Cells were serum starved overnight, treated as indicated and analyzed as in (A). (D) RAB35 associates with PI3K in a nucleotide-dependent manner. Cells stably expressing either FLAG-tagged RHEBwt, RAB35wt, RAB35Q67L or RAB35S22N were lysed and immunoprecipitations were performed using the indicated antibodies. Immunoprecipitates and total cell lysates were then analyzed by immunoblotting with the indicated antibodies.

/ sciencemag.org/content/early/recent / 3 September 2015 / Page 8 / 10.1126/science.aaa4903

Fig. 3. RAB35 mutants identified in human tumors activate PI3K signaling, suppress apoptosis and are transforming in vitro. (A) Two mutations in RAB35 that code for amino acid changes A151T and F161L were identified in the MSKCC cBio and COSMIC datasets. Alignment of RAB35 with KRAS was performed using ClustalW2. (B) Mutant RAB35 alleles from human tumors can activate PI3K/AKT signaling. NIH-3T3 mouse fibroblasts stably expressing the indicated RAB35 alleles were serum starved for 4 hours, left unstimulated or treated with bovine calf serum (BCS), lysed, and cell lysates were analyzed by immunoblotting for the indicated proteins. (C and D) Expression of RAB35 mutants suppresses apoptosis. (C) NIH-3T3 cells stably expressing the indicated proteins were plated then treated as indicated for 4 hours with or without serum-containing medium. Cells were then lysed and the cell lysates were immunoblotted for the indicated proteins. (D) NIH-3T3 cells were treated as in (C), trypsinized and the number of viable cells was counted. Cell counts for each cell line were normalized to the number of viable cells for each cell line from non serum starved conditions. Graph bars represent mean normalized cell count, error bars represent standard deviations (N = 3), and asterisks indicate statistical significance (p < 0.05) as determined by a student’s t test. (E) RAB35 mutants can transform cells in vitro in a PI3K-dependent manner. NIH-3T3 cells stably expressing the indicated RAB35 alleles, HA-p110αH1047R, or myristoylated FLAG-AKT1myr were cultured for 4 weeks in medium containing 2% BCS with either DMSO or 250 nM GDC-0941 (a pan-PI3K inhibitor), then fixed, stained with crystal violet, and imaged.

/ sciencemag.org/content/early/recent / 3 September 2015 / Page 9 / 10.1126/science.aaa4903

Fig. 4. Expression of dominant active RAB35Q67L activates PI3K signaling via PDGFRα. (A) Stable expression of dominant active RAB35Q67L activates the platelet-derived growth factor receptor (PDGFR). HEK-293E cells stably expressing the indicated FLAG-tagged GTPases were serum starved, treated as indicated, lysed, and analyzed by immunoblotting. (B) Pharmacologic inhibition of PDGFRα inhibits PI3K/AKT activity in cells expressing dominant active RAB35Q67L. Cells stably expressing RAB35Q67L were treated with either DMSO, a pan-PI3K inhibitor (GDC-0941, 0.5 μM), a PI3Kα-specific inhibitor (BYL719, 1 μM), a PDGFRβ inhibitor (CP-673451, 1 μM), or a PDGFRα/β inhibitor (Crenolanib, CP-868596, 1 μM) in serum-free media overnight, then lysed and analyzed as in (A). (C) PDGFRα is trafficked to LAMP2-positive endomembranes after stimulation with PDGF. HEK-293E cells stably expressing FLAG-RAB35wt were transfected with cDNA expressing PDGFRα-myc, serum starved or serum starved then treated with 100 ng/mL PDGF-AA for 15 min, and processed for immunofluorescence with anti-myc and anti-LAMP2 antibodies. (D) Expression of RAB35Q67L traffics PDGFRα to LAMP2-positive endomembranes in the absence of PDGF. HEK-293E cells stably expressing FLAG-RAB35Q67L were transfected with cDNA coding for PDGFRα-myc, then treated and processed for immunofluorescence as in (C). Scale bars in (C) and (D) represent 10 μm. (E) Cartoon depicting the trafficking of PDGFRα to LAMP2-positive endomembranes in cells expressing RAB35Q67L. In cells expressing RAB35wt, PDGFRα is activated at the cell membrane by PDGF-AA ligand, then internalized to EEA1-positive and LAMP2-positive endomembranes, where liganded PDGFRα drives PI3K/AKT signaling. Dashed lines reflect the fact that endomembrane -based signaling may not be as prominent as plasma membrane signaling in normal states (left). In cells expressing GTPase-deficient, GTP-bound RAB35Q67L, unliganded PDGFRα is internalized to EEA1- and LAMP2-positive compartments, where it drives constitutive activation of PI3K/AKT signaling (right).

/ sciencemag.org/content/early/recent / 3 September 2015 / Page 10 / 10.1126/science.aaa4903