RAB7 Controls Melanoma Progression by Exploiting a Lineage-Specific Wiring of the Endolysosomal Pathway

Cancer Cell

Article RAB7 Controls Melanoma Progression by Exploiting a Lineage-Specific Wiring of the Endolysosomal Pathway Direna Alonso-Curbelo,1 Erica Riveiro-Falkenbach,1 Eva Pe´rez-Guijarro,1 Metehan Cifdaloz,1 Panagiotis Karras,1 Lisa Osterloh,1 Diego Megı´as,2 Estela Can˜o´n,1 Tonantzin G. Calvo,1 David Olmeda,1 Gonzalo Go´mez-Lo´pez,3 Osvaldo Gran˜a,3 Vı´ctor Javier Sa´nchez-Are´valo Lobo,4 David G. Pisano,3 Hao-Wei Wang,5 Pablo Ortiz-Romero,6 Damia` Tormo,1,8 Keith Hoek,7 Jose´ L. Rodrı´guez-Peralto,6 Johanna A. Joyce,5 and Marı´a S. Soengas1,* 1Melanoma

Laboratory, Molecular Oncology Programme Microscopy Unit, Biotechnology Programme 3Bioinformatics Unit, Structural Biology and Biocomputing Programme 4Epithelial Carcinogenesis Laboratory, Molecular Pathology Programme Spanish National Cancer Research Centre (CNIO), Madrid 28029, Spain 5Cancer Biology and Genetics Program, Memorial Sloan Kettering Cancer Center, New York, NY 10065, USA 6Instituto de Investigacio ´ n i+12, Hospital 12 de Octubre, Universidad Complutense, Madrid 28041, Spain 7Department of Dermatology, University Hospital of Zurich, Zurich 8091, Switzerland 8Present address: BiOncoTech Therapeutics, Valencia 46980, Spain *Correspondence: [email protected] http://dx.doi.org/10.1016/j.ccr.2014.04.030 2Confocal

SUMMARY

Although common cancer hallmarks are well established, lineage-restricted oncogenes remain less understood. Here, we report an inherent dependency of melanoma cells on the small GTPase RAB7, identified within a lysosomal gene cluster that distinguishes this malignancy from over 35 tumor types. Analyses in human cells, clinical specimens, and mouse models demonstrated that RAB7 is an early-induced melanoma driver whose levels can be tuned to favor tumor invasion, ultimately defining metastatic risk. Importantly, RAB7 levels and function were independent of MITF, the best-characterized melanocyte lineage-specific transcription factor. Instead, we describe the neuroectodermal master modulator SOX10 and the oncogene MYC as RAB7 regulators. These results reveal a unique wiring of the lysosomal pathway that melanomas exploit to foster tumor progression.

INTRODUCTION Most solid tumors that have been scrutinized in detail contain a vast array of (epi)genetic alterations (Garraway and Lander, 2013). Therefore, the identification of tumor drivers represents a main challenge in oncology. This is remarkably daunting in cutaneous melanoma, where over 85,000 coding mutations have been reported (Lawrence et al., 2013). Gene discovery in melanoma is further hampered by the inherently dynamic behavior of these tumor cells (Javelaud et al., 2011; Roesch

et al., 2010). In particular, gene expression changes linked to epithelial-to-mesenchymal-like transition have been described whereby melanomas deregulate proliferation, lineage specification, and embryonic differentiation genes to acquire invasive properties (Alonso et al., 2007; Caramel et al., 2013; Gupta et al., 2005; Widmer et al., 2012). However, although expression analyses, spheroid cultures, and mouse xenografts support melanoma cell plasticity (Caramel et al., 2013; Pinner et al., 2009; Widmer et al., 2012), the underlying mechanisms are still not fully understood.

Significance Mechanisms underlying lineage specificity can help explain why particular genes drive malignancy in only certain tumor types. The discovery of oncogenic dependencies is particularly challenging in melanoma, a disease with a strikingly high number of mutations and epigenetic alterations. Here, we show that despite their complex genotype, melanoma cells retain a developmental memory that reflects a unique wiring of vesicle trafficking pathways. Specifically, we identify a melanomaenriched endolysosomal gene cluster that revealed unanticipated roles of RAB7 in controlling the proliferative and invasive potential of these aggressive tumors. Our data illustrate how lysosomal-associated degradation, otherwise a rather universal feature of eukaryotic cells, can be hijacked in a tumor-type- and stage-dependent manner. Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc. 61

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

B

Esophagus

A

C

D

E

F

Nonmelanoma

G

(legend on next page)

62 Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc.

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

A less explored strategy to identify drivers of tumor progression is to focus on lineage-specific traits, instead of on genes or pathways that may be deregulated across multiple cancer types (Garraway and Lander, 2013). The underlying hypothesis is that malignant cells may potentiate or rewire signaling cascades that are specifically required for the survival and/or physiological functions of their normal cell precursors (Garraway and Sellers, 2006). In melanoma, for example, studies of the microphthalmia-associated transcription factor (MITF) have illustrated how cancer cells can hijack lineage specification genes to promote tumor development (Tsao et al., 2012). Thus, protumorigenic targets of MITF include a variety of factors involved in the control of cell death, DNA replication, repair, and mitosis, as well as microRNA production, membrane trafficking, and mitochondrial metabolism, among others (Cheli et al., 2010; Haq et al., 2013a; Vazquez et al., 2013). Nevertheless, MITF is a dynamically regulated rheostat, favoring or dampening malignant features of melanoma cells in a context-dependent manner (Koludrovic and Davidson, 2013). Moreover, although MITF is amplified in a subset of melanomas, it can also be inhibited posttranscriptionally (Bell and Levy, 2011; Javelaud et al., 2011; Thurber et al., 2011). Consequently, MITF and its targets, such as the trafficking modulator RAB27A or the antiapoptotic BCL2A1 protein, are overexpressed (and required) only by a fraction of melanomas (Akavia et al., 2010; Haq et al., 2013b). It is unclear whether there are alternative lineage-restricted drivers common to a broader spectrum of melanomas. Here, we have performed a comprehensive computational, genetic, and functional characterization of melanoma-enriched pathways to identify lineage-specific tumor drivers that define the inherent plasticity of this aggressive malignancy. RESULTS Selective Enrichment of Lysosomal-Associated Traits in Melanoma To identify signaling cascades unique to melanoma, gene set enrichment analysis (GSEA) was performed on transcriptional data sets (Barretina et al., 2012; Scherf et al., 2000; Shankavaram et al., 2007) covering over 35 different tumor types. As expected, pigmentation-related Gene Ontology (GO) gene sets were enriched in melanoma cells (false discovery rate [FDR] < 3.04 3 10
5; Table S1 available online). However, the highest enrichment in melanoma was found for the GO lysosome, vacu-

ole, and lytic vacuole gene sets (FDR < 1.0 3 10
8; Table S1). In particular, a cluster of lysosome-associated factors was coexpressed within all 55 melanoma cell lines of the Cancer Cell Line Encyclopedia (CCLE) in a distinct manner from >750 examples of other tumor types (see heatmaps in Figure 1A, the corresponding enrichment plots in Figure 1B, and a ranked list of melanoma-enriched genes in Figure S1A). Lysosomes share common precursors with melanosomes (Raposo and Marks, 2007). However, additional GSEA also confirmed a specific enrichment of ‘‘lysosome-only’’ genes in melanoma (including proteases, lipases, acid ceramidases, acid phosphatases, and other lytic enzymes), with no link to melanosome biogenesis or function (Figure S1B). Consistent with these results, representative melanoma cell lines (n = 8) were found to have an overall higher lysosomal-associated proteolytic activity compared to cells of other tumor types (n = 7), as reflected by the cleavage of the DQ-Green BSA probe (Figure S1C). Moreover, melanoma cell lines were found to be highly sensitive to the lysosomotropic agent chloroquine under conditions in which neither the viability nor proliferation rate of other nonmelanoma cell lines was significantly affected (Figure 1C; Figures S1D–S1F). Importantly, chloroquine effectively inhibited global lysosomal function in all cell lines tested, as defined by the accumulation of the autophagic flux markers LC3-II, p62, and NBR1 (Figures S1G and S1H). Together, these data support a specific wiring of lysosomalassociated degradative pathways in melanoma cells, which is required for their maintenance. Preferential Accumulation of RAB7 in Melanoma Cells and Tumor Specimens Histological and functional analyses were then performed across different tumor types to identify putative lineage-specific cancer drivers within the melanoma-enriched lysosomal signature. We selected the RAB7A GTPase for validation based on the following criteria. (1) RAB7A (hereafter referred to as RAB7 for simplicity) showed the highest enrichment in melanoma (Figure 1D), exceeding other RABs such as RAB27A (Figure S1I), known to modulate the proliferation of a subset of melanoma cells (Akavia et al., 2010). (2) RAB7 has key roles in lysosome biogenesis and lysosomal-associated degradation of cytoplasmic vesicles (Bucci et al., 2000; Zerial and McBride, 2001; Zhang et al., 2009), processes also found by GSEA to be upregulated in melanoma (Table S1). (3) RAB7 maps within a genomic region frequently amplified in melanoma (see representative array

Figure 1. Melanoma-Selective Overexpression of a Lysosomal Gene Cluster Containing the Small GTPase RAB7 (A) GSEA heatmap showing a selective enrichment of the Lysosome Gene Ontology gene set (GO:0005764) in melanoma cell lines compared to nonmelanoma tumor cell lines in the CCLE. (B) Enrichment plot for the lysosome GO gene set described in (A) in melanoma versus nonmelanoma tumor cell lines (GSEA FDR = 0.018). (C) Effect of chloroquine (20 mM, 48 hr) on the viability of the indicated tumor cell lines. Shown is an example of two independent experiments performed in triplicate. Data are represented as mean ± SEM. ***p < 0.001. (D) Box plots showing the relative RAB7A mRNA levels across the different tumor types, extracted from CCLE_Expression_Entrez_2012-10-18.res, with genecentric robust multiarray analysis-normalized mRNA expression data. The number of cell lines of each tumor type analyzed is indicated in parentheses. (E) Western blot analysis showing the relative levels of RAB7, MITF, and RAB27A protein in the indicated melanoma (blue) and nonmelanoma (black) tumor cell lines. (F) Quantification of RAB7 protein expression by immunohistochemical staining on multitumor tissue microarrays containing the indicated human cancer types (n = 121). (G) Representative immunohistochemical staining of RAB7 protein (pink) in human biopsies from the indicated cancer types. Scale bars represent 200 mm in lowmagnification images and 20 mm in the insets. See also Figure S1 and Tables S1 and S2.

Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc. 63

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

A

B

C

D

Nonmelanoma

E Nonmelanoma

F

G

(legend on next page)

64 Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc.

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

comparative genomic hybridization [CGH] data in Figure S1J and additional information in Table S2), consistent with computational algorithms that suggested a putative driver role in this disease (Akavia et al., 2010). (4) There was no previous indication that RAB7 could be regulated or acting in a tumor-type-specific manner (Wang et al., 2011; Zhang et al., 2009). Moreover, the specific roles of RAB7 in cancer are unclear, as it can support protumorigenic (Wang et al., 2012; Williams and Coppolino, 2011) or suppressive effects (Edinger et al., 2003; Steffan and Cardelli, 2010; Steffan et al., 2014) depending on the cellular context. Western blot and tissue microarray (TMA) protein expression analyses across 17 different cancer types confirmed a selective enrichment of RAB7 in melanoma cell lines (Figure 1E) and tumor specimens (Figures 1F and 1G; p < 0.001, n = 121), even in cases with negligible levels of MITF (Figure 1E). RAB7 levels were significantly elevated in melanoma cells as compared to normal skin melanocytes (see Figure S1K for representative examples of costaining of RAB7 and the S100 melanocytic marker in normal skin and primary melanoma human specimens). Together, these results suggest that RAB7 could represent an MITF-independent lineage-specific melanoma driver. Lineage-Specific Dependency of Melanoma Cells on RAB7 Melanoma-restricted roles of RAB7 were investigated by stable transduction of three independent short hairpin interfering RNAs (shRNAs) in a panel of cell lines from melanoma (n = 8) and from a selection of other tumor types (n = 8). Melanoma cells responded with a significant inhibition of cell proliferation (Figures 2A–2D; Figures S2A and S2B) and colony-forming ability (Figure 2E). In contrast, RAB7 downregulation had minimal or moderate effects on the proliferative capacities of the other tumor cell lines (Figures 2A–2E; Figures S2A and S2B). Controls for RAB7 shRNA specificity included the overexpression of the well-characterized dominant-negative RAB7 T22N mutant (Figure 2E; Figures S2C– S2E) and of wild-type RAB7, the latter increasing melanoma cell proliferation (Figures S2C and S2D). Importantly, the inhibitory effects of RAB7 shRNA and the dominant-negative RAB7 mutant translated into a significantly reduced tumorigenic potential in vivo (i.e., xenograft models) of both BRAFV600Eand NRASQ61R-mutated melanoma cell lines (see examples in Figures 2F and 2G). Of note, the selective dependency of melanoma cells on RAB7 was independent of their pigmentation status and basal MITF levels (Figures 1E and 2C–2G and results not shown), and was not recapitulated by other RAB proteins with shared functions in melanosome transport (Figure S2E).

To define whether the high expression of RAB7 in melanoma cells could be traced to normal melanocytes, human skin was processed for the isolation of melanocytes, keratinocytes, and fibroblasts. Primary cultured melanocytes had intrinsically higher levels of RAB7 (Figure S2F), and were more dependent on this protein than their genetically matched nonmelanocytic counterparts (Figures S2G–S2I). Still, melanoma cells were more sensitive than melanocytes to RAB7 depletion, responding with a more marked inhibition of proliferation and acquisition of senescence-like features (Figures S2G–S2K). These results suggest that melanoma cells may exploit (and depend on) proliferative roles of RAB7 present in normal precursors but acquire additional signals that impose a strict dependency on this GTPase. Invasive and Metastatic Potential of Melanoma Cells Controlled by RAB7 Aberrant vesicle trafficking can modulate cell shape, cell motility, and metastasis (Goldenring, 2013). Therefore, we determined whether the inhibition of melanoma cell proliferation was associated with cytoskeletal changes and alterations in cell motility. As shown in Figure 3A, RAB7 depletion induced marked morphological changes in all melanoma cell lines tested. In particular, prominent cellular vacuolization and increased dendricity were characteristic of NRAS- and BRAF-mutated cases, respectively (Figure 3A; see also Movie S1). RAB7 downregulation reduced cell-cell contacts (Figure 3A), increased cell scattering (Figure S3A), and induced the disassembly of stress fibers and cortical F-actin (see phalloidin staining of cells grown in monolayers in Figure S3B). Cytoskeletal alterations of shRAB7 melanoma cells were also evident in single-cell analysis of cell polarization (Figure S3C), which also identified decreased focal adhesions (paxillin staining; Figure S3C). These marked phenotypic changes driven by RAB7 downregulation were not observed in normal melanocytes and fibroblasts or in tumor cells from seven other cancer types (Figure 3A). Counterintuitively, although inhibition of RAB7 expression reduced the proliferation and colony formation of melanoma cells, the phenotypic changes described above suggested enhanced rather than reduced cell motility upon RAB7 downregulation. Indeed, RAB7 shRNA increased the migration capacity of melanoma cells in scratch assays (Figure S3D), increased the invasive potential of moderately metastatic melanoma cells in Matrigel invasion assays (Figure 3B), and favored their lung extravasation ability after tail vein injection in immunosuppressed mice (Figure 3C). Furthermore, melanoma cell lines with the highest invasive and metastatic potential in vitro and

Figure 2. Lineage-Specific Dependency of Melanoma Cells on RAB7 (A) Western blot analysis showing RAB7 downregulation with three different RAB7 shRNAs (1–3) in the indicated melanoma (blue) and nonmelanoma (black) cell lines. Noninfected cells (-) or cells expressing scramble shRNA control (Ctrl) are also shown as reference controls. (B) Bright-field images of cell populations generated as in (A) and imaged at day 6 posttransduction. Scale bars represent 50 mm. (C) RAB7 and b-actin immunoblots of total cell extracts isolated from the indicated tumor cell lines, noninfected (-) or infected shRNA control (Ctrl), or RAB7 shRNA3 (RAB7)-expressing lentiviruses. (D) Impact of RAB7 downregulation on cell proliferation of the indicated cell populations. Cells were plated at equal cell numbers at day 6 postinfection and cell number was quantified during consecutive days. Data correspond to the mean ± SEM of three experiments performed in triplicate. (E) Colony-formation assays with the indicated tumor cell lines expressing RAB7 shRNA3 (shRAB7), RAB7 T22N dominant-negative mutant (RAB7DN), or their corresponding controls. Scale bars represent 5 mm. (F and G) Growth after subcutaneous implantation in nude mice of the indicated melanoma cell lines expressing RAB7 shRNA3 (shRAB7) (F), RAB7 T22N dominant-negative mutant (RAB7 T22N) (G), or their corresponding controls. Results represent the mean ± SEM of ten mice per condition. ***p < 0.001, **p < 0.01, *p < 0.05. See also Figure S2.

Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc. 65

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

Nonmelanoma

A

B

C

H

D

E

G

F

(legend on next page)

66 Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc.

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

in vivo were found to have lower basal levels of RAB7 (Figure 3D). In a meta-analysis performed on six independently generated melanoma transcriptomic data sets (Widmer et al., 2012), RAB7 mRNA levels correlated negatively with a defined proinvasive melanoma gene signature (Hoek et al., 2008) but positively with proproliferative gene sets (Figure S3E). Therefore, RAB7 has different dose- and tumor-type-dependent roles in cancer cell proliferation and invasion. Global Impact of RAB7-Dependent Vesicular Trafficking on the Melanoma Proteome Next, we sought to dissect the effector mechanisms by which RAB7 levels could modulate the proliferative and invasive capacities of melanoma cells. First, we assessed autophagosome maturation, as this is a prototypic function of RAB7 (Wang et al., 2011; Zhang et al., 2009) and defective autophagy can compromise melanoma cell proliferation per se (Checinska and Soengas, 2011). Focal accumulation of the autophagosome marker LC3 confirmed impaired autophagic degradation in RAB7 shRNA-treated melanoma cells (Figure S3F, left panels). Nonetheless, a series of differences with canonical degradative programs was identified. (1) LC3-positive vesicles accumulating upon RAB7 downregulation were unusually large and clearly distinct from the ‘‘classical’’ compact LC3 foci induced by treatment with rapamycin, a standard autophagy inducer (Figure S3F). (2) The formation of these large ring-shaped LC3 vesicles was independent of ATG7, a key autophagy regulator (Figure S3G). (3) Instead, LC3 was loaded into pre-existing late endosomes (matured from RAB5-decorated vesicles) that accumulated when RAB7 was depleted (Figures S3H and S3I; Movie S2). (4) The LC3-recruiting endosomes were identified as macropinosomes, because they originated from membrane ruffles (not shown) and were filled with fluid-phase endocytic tracers (Figure S3J). Together, these observations point to the existence of an ATG7-independent route for lysosomal degradation of autophagosome-endosome hybrids in melanoma cells, which depends on RAB7 for its proper execution. Of note, normal

melanocytes, which exhibited milder phenotypic changes upon RAB7 downregulation, displayed an intrinsically silent macropinocytic activity (Figure S3K). RNA sequencing (RNA-seq) was performed at early time points after transduction of shRNA-expressing vectors to broadly dissect the molecular changes induced by deregulating RAB7-mediated trafficking in representative melanoma cells. HCT116 colon cancer cells were also analyzed as a nonmelanoma reference. Notably, RAB7 depletion resulted in alterations in the transcriptome of HCT116, but these changes were either less significant or even opposite the effects in melanoma cells (see GSEA results in Table S3). By contrast, depletion of RAB7 in melanoma cells led to a significant downregulation of multiple gene sets involved in cell-cycle progression (Figure 3E; Figure S3L; Table S3). The E2F1 signaling cascade (Figure 3F) was particularly relevant, considering its requirement for melanoma cell proliferation (Halaban et al., 2000). Protein validation analyses demonstrated that RAB7 shRNA induced the downregulation of the E2F1 cofactor TFDP1, the E2F1/TFDP1 targets CDC2 and CDC6, and various MCM proteins (Figure 3G and results not shown). Other key mitosis and cell-division-related pathways were also identified (e.g., involving aurora kinases or PLK1-associated programs; see Figure S3L and Table S3), which interestingly have not previously been linked to RAB7. Of note, the downregulation of cell-cycle factors (see TFDP1 in Figure S3M, top panel) and the phenotypic changes induced by RAB7 shRNA (Figure S3M, bottom panel and Figure S3N) were rescued by transducing cells with a nontargetable RAB7 construct, confirming the specificity of the depletion system used. RNA-seq also provided mechanistic insight into the proinvasive phenotype induced by RAB7 shRNA in melanoma cells (see examples of pathways involved in cell motility, tumor invasion, and extracellular matrix remodeling in Figures 3E and 3F, Figure S3L, and Table S3). Western blot analyses validated CEACAM1, a clinically relevant prometastatic melanoma gene (Thies et al., 2002), as a RAB7-controlled factor (Figure 3G).

Figure 3. RAB7 Levels Inversely Modulate Cell Proliferation and Invasion-Related Pathways of Melanoma Cells (A) Micrographs showing the selective impact of RAB7 downregulation by RAB7 shRNA3 (shRAB7) on the morphology of melanoma cells versus the indicated nonmelanoma cell types. NRAS- and BRAF-mutated melanoma cells are labeled with an asterisk or a number sign (#), respectively. Scale bars represent 20 mm. (B) Invasiveness of SK-Mel-28 expressing scramble shRNA control (shControl) or two different RAB7 shRNA (shRAB7) constructs, evaluated by Matrigel invasion assay (48 hr). Data are presented as the average ± SEM of three independent experiments performed in duplicate. **p < 0.01. (C) Representative histological analysis of lung sections of nonobese diabetic/severe combined immunodeficiency mice 48 hr after intravenous implantation of mCherry-labeled SK-Mel-28 expressing shControl or RAB7 shRNA3 (shRAB7). Arrows point to tumor cells in the lung parenchyma, stained in brown with an antiCherry antibody. An average of 12 ± 2 cells were found extravasated per 10 mm2 lung area in shControl versus 28 ± 3 in shRAB7 (p = 0.04, n = 5 mice/group). (D) Inverse correlation between RAB7 protein levels and the invasive capacity of the indicated melanoma cell lines, evaluated by western blotting and Matrigel invasion assays, respectively. The percentage of invading cells is presented as the average ± SEM of three independent Matrigel invasion experiments. Relative RAB7 protein levels are presented as the average ± SD of two independent western blots quantified with ImageJ software (NIH) and using a b-actin blot as the loading control. (E) Pie chart representing the distribution of significantly down- and upregulated genes (FDR < 0.05) upon transduction of RAB7 shRNA3 in UACC-62 melanoma cells. (F) GSEA plots showing representative examples of gene modules significantly downregulated (left: E2F pathway; FDR = 5.48E-04) or upregulated (right: membrane trafficking; FDR = 0.016) upon RAB7 depletion (RAB7 KD) in UACC-62 melanoma cells. The complete GSEA data for both UACC-62 melanoma and HCT116 nonmelanoma cell lines are shown in Table S3. (G) Opposing (and melanoma cell-specific) effect of RAB7 downregulation on the levels of factors involved in cell proliferation (i.e., TFDP1, CDC2, CDC6) and invasion (i.e., CEACAM1), shown by immunoblotting in SK-Mel-28 and UACC-62 melanoma cell lines. HCT116 colon cancer cells (black) are included as a nonmelanoma reference. (H) Immunoblot analyses of the indicated proteins in the conditioned media (CM) or in total cell lysates isolated from nonmelanoma (black) and melanoma (blue) cell lines, either noninfected (-) or infected to express scrambled shRNA (Ctrl) or RAB7 shRNA3 (RAB7). Additional information is included in Figure S3, Movies S1 and S2, and Table S3.

Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc. 67

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

A

B

Visceral

(Noninvasive)

C

E

F

D

G 5 YEAR FOLLOW-UP 1.39–4.60 1.18–4.00

10 YEAR FOLLOW-UP 1.41–4.19 1.19–3.59

(legend on next page)

68 Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc.

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

Furthermore, proteomic analyses of conditioned media, guided by the deregulation of membrane trafficking and protein secretion pathways revealed by RNA-seq (Figure 3F; Table S3), demonstrated that RAB7 downregulation ultimately promotes the secretion of fibronectin and multiple lysosomal cathepsins (CTS) (Figure 3H and data not shown), all with well-known roles in tumor metastasis (Mason and Joyce, 2011). Although cathepsin secretion has been associated with RAB7 downregulation (Steffan et al., 2014), these effects are likely cell-typedependent, as they were not found in the nonmelanoma HCT116 cell line (even in the case of the secretion of CTS-X, which is equally expressed in all cell types analyzed; Figures 3G and 3H). Immunofluorescence analyses demonstrated that the increased protein secretion of RAB7-depleted melanoma cells was a direct consequence of changes in the fate and distribution of endocytic vesicles. CTS-B accumulated in RAB5decorated early endosomes that, instead of being transferred to perinuclear lysosomes for degradation, relocalized toward the plasma membrane when RAB7 was reduced (Figure S3O). Together, these data provide mechanistic insight into the broad effector mechanisms whereby RAB7 influences melanoma cell proliferation and invasion, and further underscore a cancertype-dependent impact of deregulated vesicle trafficking. RAB7 Is an Early-Induced Melanoma Driver Tuned Down at Invasive Stages of Tumor Progression In Vivo The dual roles of RAB7 found in cultured melanoma cells (i.e., a requirement for proliferation but an enhancement of invasive phenotypes when tuned down) predicted that this protein would be regulated in a dynamic manner in vivo, with a selective downmodulation favoring the switch to metastatic stages. To validate our findings, RAB7 expression was analyzed in a comprehensive series of TMAs containing human benign intradermal nevi and biopsies collected from radial growth phase (RGP), vertical growth phase (VGP), and metastatic melanomas (n = 152 cases). Melanoma specimens showed significantly higher levels of RAB7 compared to surrounding unaffected normal skin and benign nevi (Figures 4A and 4B), with the highest enrichment in the early-stage RGP specimens (p < 0.001). High expression of RAB7 positively correlated with high Cyclin D1 levels (Figures S4A and S4B), further supporting a role for this GTPase in melanoma cell proliferation. However, RAB7 levels did not remain constant, as a significant reduction of RAB7 levels was observed at the RGP-to-VGP transition (Figure 4B), the stage at which mel-

anoma cells are known to acquire invasive properties. Single-cell confocal analyses of RAB7 levels in whole-tissue sections of primary melanomas further revealed decreasing RAB7 expression toward the dermal-invading front of primary tumors (see a representative example in Figures 4C and 4D). Similarly, human melanoma cells grown in mice also showed a decreasing gradient of RAB7 expression toward the deeper areas of the xenograft tumors (see arrows in Figure S4C). The inverse correlation between RAB7 expression and depth of melanoma invasion (Breslow scoring) was confirmed in an independent cohort of primary human melanomas (Figure 4E; p < 1.0 3 10
3, n = 116). Importantly, no RAB7-negative tumor was identified, and even the ‘‘low-expressing’’ melanomas retained RAB7 at levels invariably higher than the surrounding stromal counterparts (marked with asterisks in Figure 4A), in agreement with a lineage-specific dependence of melanomas on this factor. To assess the physiological relevance of the downmodulation of RAB7 in melanoma metastasis, retrospective 10-year followup analyses were performed using clinically annotated TMAs containing a total of 112 primary melanoma specimens. Patients with reduced expression of RAB7 in primary tumors showed an increased risk for the development of metastasis (p = 0.001, Figure 4F; Cox hazard ratio of 2.52 at 10 years after diagnosis, Figure 4G) and a poor overall survival (Figures S4D and S4E). Low RAB7 levels were predictive of metastatic risk even when adjusted for Breslow depth (Figure 4G). Together, these data identify RAB7 as a dynamically regulated and physiologically relevant risk factor that might underlie the phenotypic plasticity of aggressive melanoma cells. Oncogene-Dependent Regulation of RAB7 by MYC Computational analyses of the RAB7 promoter revealed the presence of eight E boxes (Figure 5A, top panel) characteristically recognized by the oncogenic transcription factor MYC, well known for its involvement in melanoma progression (Polsky and Cordon-Cardo, 2003). This oncogene was also attractive to investigate further, because its depletion leads to melanoma cell vacuolization and dendricity in a manner similar to RAB7 (Zhuang et al., 2008). Histological analyses revealed a marked heterogeneity of MYC expression in melanocytic lesions. Nonetheless, MYC was enriched in melanocytic nests close to the epidermal/ dermal interface, where RAB7 also accumulated (Figure 5B). Chromatin immunoprecipitation demonstrated that the three most proximal E boxes upstream of the RAB7 transcriptional

Figure 4. Stage-Dependent Expression and Prognostic Value of RAB7 in Human Melanoma Biopsies (A) Representative IHC of RAB7 (pink) in the indicated human benign (a) and malignant melanocytic lesions (b–e). Lower panels (f–h) show high-magnification micrographs of the corresponding upper panels (a–e). Asterisks mark stromal cells. (B) Quantification of RAB7 IHC staining on TMAs containing the indicated human melanocytic lesions. The number of samples per category analyzed is indicated in parentheses at the bottom of the bar graph (total n = 152). ***p < 0.001. (C) Confocal immunofluorescence (IF)-based quantification of the mean RAB7 protein expression per melanoma cell in a representative whole-tissue VGP melanoma specimen. Blue color represents stromal cells. (D) Staining of RAB7 and Cyclin D1 by IHC in consecutive slides from the same VGP melanoma specimen shown in (C). (E) Inverse correlation between RAB7 protein levels and Breslow index or depth of invasion (p < 0.001). n indicates the total number of cases analyzed in each group. (F) Kaplan-Meier curves showing 10-year metastasis-free survival upon resection of primary melanomas. Patients were stratified as a function of high and low RAB7 protein levels in primary tumor specimens (n = 112, p = 0.001). (G) Cox regression models indicating the predictive value of RAB7 expression on 5- and 10-year metastasis-free survival, with or without adjusting for Breslow depth. HR, hazard ratios; 95% CI, 95% confidence intervals. See Figure S4 for additional information.

Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc. 69

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

A

D

E (AFU)

C

B

(AFU) F

G

(legend on next page)

70 Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc.

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

start site were bound by MYC (Figure 5A, bottom panel). When MYC was depleted by validated small interfering RNA (siRNA) (Doe et al., 2012), RAB7 expression was inhibited by >60% (Figure 5C). Lineage-Specific but MITF-Independent Regulation of RAB7 by SOX10 We next interrogated lineage-restricted transcription factors that may confer specificity to the melanoma-enriched expression of RAB7 found by GSEA. Although the best-characterized melanocyte developmental pathways involve MITF, melanoma cells with low or undetectable MITF retained high levels of RAB7 in culture (Figure 1E; Figure S5A) and in tumor biopsies (Figure 5D; see quantification in Figure 5E and additional information in Figure S5B). Indeed, downregulation of MITF by validated siRNAs had no measurable impact on RAB7 levels in different melanoma cell lines (Figure 5F) or at different time points (Figure S5C) in which the expression of other known MITF targets, such as RAB27A, was effectively blocked. Of note, depletion of TFEB, a transcription factor of the MITF/TFE family that controls the expression of various autophagy-associated factors (Settembre et al., 2011), had no significant effect on RAB7 (although it did exert a modest reduction [30%] of RAB27A expression; Figure 5G). Having excluded a role for MITF in the control of RAB7 mRNA or protein levels, we next interrogated the contribution of PAX3 and SOX10, transcription factors acting early in the specification of the melanocytic lineage. No significant changes in RAB7 levels were found upon depletion of PAX3 (Figure 6A). However, siRNA against SOX10 effectively reduced RAB7 mRNA and protein (Figures 6A–6D). Of note, SOX10 expression mimicked that of RAB7, being retained in MITF-negative melanoma cells and expressed at lower levels in highly invasive cell lines (Figure 6E). A significant positive correlation was found between RAB7 and SOX10 mRNAs in frozen human melanoma specimens (Figure 6F; r = 0.6, n = 15), as further supported by analysis of protein expression in consecutive sections of benign and malignant melanocytic biopsies (Figure 6G). Thus, gradients of SOX10 were identified within human melanocytic lesions, although less marked than for RAB7 (likely reflecting additional mechanisms of regulation of both factors). Nonetheless, intraepidermal nests

in nevi or primary melanomas with strong RAB7 expression were highly positive for SOX10 (Figure 6G, b and c). Parallel expression of SOX10 and RAB7 was also found in metastatic melanomas (see Figure 6G, d and e for examples of specimens in which these proteins are coordinately expressed at low or high levels, respectively). Together, our results identify two levels of regulation of RAB7, by the oncogenic transcription factor MYC and by the melanocytic lineage specifier SOX10. These data illustrate how endolysosomal trafficking can be controlled in a lineage-specific manner in cancer, distinguishing melanoma from other tumor types. DISCUSSION Here, we have demonstrated that melanoma cells exploit a lineage-specific wiring of the endolysosomal pathway to drive tumor progression. Specifically, we have shown that the small GTPase RAB7 is selectively upregulated in melanoma as part of a lysosomal-associated signature that distinguishes this malignancy from over 35 cancer types. This induction occurs at early stages of melanoma development and sustains melanoma cell proliferation. However, expression studies in clinical biopsies combined with functional studies in cultured cells revealed that this GTPase can be partially downregulated in highly invasive melanoma cells (see model in Figure 7). This pattern of expression reflects MITF-independent roles of RAB7 regulating the fate of endolysosomal traffic in melanoma cells, modulating their transcriptome, proteome, and secretome and ultimately influencing the proliferation and invasion of these aggressive cells. The physiological relevance of these data was demonstrated in 10-year follow-up studies of patient prognosis, which support RAB7 as a risk factor for melanoma metastasis and overall poor survival. Lysosomes are essential organelles in normal and tumor cells (de Duve, 2005). Similarly, RAB7 has been described to modulate endosomal maturation and autophagosome resolution in various cell types (Wang et al., 2011; Zhang et al., 2009). Why then are endo/lysosomal pathways (and, more specifically, RAB7) wired (and required) in such a specific manner in melanoma? An attractive scenario to account for the coregulation

Figure 5. Control of RAB7 Transactivation by MYC, but Not by MITF or TEFB (A) The top panel shows eight putative Myc-binding site E boxes (E1–E8) flanking the transcriptional start site (TSS) of the RAB7 gene as indicated. The bottom panels summarize quantitative (q)ChIP data for the MYC antibody (blue bars), presented as the mean ± SEM of three independent experiments with respect to the IgG control (gray bars). Acetylcholine receptor and nucleolin (ACHR and NCL, respectively) promoter amplicons were used as negative and positive controls for MYC binding, respectively. (B) IHC for RAB7 and MYC in intradermal nevi (a and b) and VGP melanoma (c and d). The staining was optimized for optimal visualization of melanocytic versus stromal compartments (the latter are marked with asterisks). T marks the tumoral compartment. Areas contained in the dashed boxes are shown at high magnification on the right. (C) Downregulation of RAB7 protein expression shown by immunoblotting 72 hr after transfection of the indicated concentrations of MYC siRNA. Basal RAB7 levels in control (Ctrl) siRNA-transduced cells are included as a reference. (D) Confocal microphotographs of melanoma biopsies stained for RAB7 (red) and MITF (green). In the merged figure, nuclei appear counterstained with 40 ,6-diamidino-2-phenylindole. Shown are lymph node (LN) metastasis and primary melanoma specimens coexpressing both factors (left and middle) and an example of a metastasis positive for RAB7 only (right). (E) Confocal-based quantification of the mean immunofluorescence signal per cell for RAB7 and MITF in the same patient biopsies shown in (D). AFU, arbitrary fluorescence units. (F) Immunoblots showing relative levels of MITF, RAB27A, RAB7, and b-actin (loading control) protein levels in the indicated melanoma cell lines expressing control or MITF siRNA. Here and in (G), data are shown at 72 hr posttransfection. (G) RT-PCR analyses of RAB7, RAB27A, TFEB, and 18S (loading control) mRNA levels in the indicated melanoma cell lines expressing control or TFEB siRNAs. See Figure S5 for additional information.

Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc. 71

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

A

B

D

E

C

F

G

(legend on next page)

72 Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc.

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

Figure 7. Model Summarizing Key Aspects of the Regulation and Function of RAB7 Described in This Study RAB7 was found to be selectively enriched in melanoma, expressed in an MITF-independent manner and controlled by the lineage-specific transcription factor SOX10 (1). A second level of regulation was found linked to MYC, a wellknown oncogenic transcription factor activated at early stages of melanoma development (2). Importantly, RAB7 levels were not constant during melanoma progression, being selectively downmodulated (but never abrogated) to favor invasive phenotypes (3), and ultimately increasing the risk for metastatic development (4). RAB7-dependent vesicles are depicted as red dots. See the text for additional details.

of the lysosomal-related genes found here enriched in melanoma is that they are coordinately involved in functions that are unique to this tumor type. For example, various lysosomal factors including RAB7 are well known for participating in melanosome maturation (Gomez et al., 2001). Nevertheless, our data revealed an enrichment of ‘‘lysosome-only’’ genes in melanoma cells that transcends cellular pigmentation. Moreover, these cells showed a particular dependency on lytic activities of the lysosome that rely on RAB7 function. In particular, the inherently active influx of ATG7-independent autophagosome-macropinosome hybrids found here in melanoma cells may impose a requirement for RAB7 that cannot be compensated by other RAB members with more restricted roles in membrane traf-

ficking. By modulating the fate of these vesicles (i.e., from lysosomal targeting to recycling or secretion), RAB7 may control the levels and/or localization of other key protumorigenic proteins, exemplified here by lysosomal cathepsins and other prometastatic factors. Finding that RAB7 is controlled by SOX10 and MYC in an MITF-independent manner has important basic and translational implications. Thus, these results separate RAB7 from key modulators of proliferation and survival such as RAB27A, BCL2A1, or PGC1a, which are strictly controlled by MITF (Cheli et al., 2010; Haq et al., 2013a, 2013b). Moreover, SOX10 is not inhibited by mechanisms that downregulate MITF, some of which, including BRAF mutations, are relatively frequent in malignant melanomas (Haq et al., 2013a; Vazquez et al., 2013). This may ensure a ‘‘developmental memory’’ in the expression of RAB7. Nevertheless, RAB7 does not lead to the apoptotic collapse recently described for SOX10 depletion (Shakhova et al., 2012), suggesting that there are nonoverlapping functions for both factors. The direct regulation of RAB7 by MYC is also relevant, as this oncogene can be modulated by a variety of signaling cascades (Dang, 2012), opening the possibility of a broader oncogenic regulation of RAB7 in melanoma. In this context, it is tempting to speculate that the downregulation of RAB7 (and/or its upstream drivers) in the invasive front of aggressive melanomas is modulated by epithelial-to-mesenchymal-like mechanisms, such as those recently described to underlie the transcriptional switch associated with prometastatic phenotypes (Caramel et al., 2013). The spatiotemporal regulation of RAB7 is central to understanding its role in melanomagenesis and metastasis. Studies limited to either primary lesions or to late stages of the disease would have missed the downregulation of this protein occurring at the RGP-to-VGP transition, predicting metastatic risk. It would therefore be interesting to determine whether the levels of RAB7 fluctuate during the development of other tumor types. Further lineage-specific analyses of RAB7 may also be relevant in hematopoietic and lymphoid tumors, which intriguingly showed a systematic downregulation of the endolysosomal gene cluster found here upregulated in melanoma (Figure 1A). Whether lineage specificity may account for modest effects of RAB7 depletion on the viability of normal murine T cells (Roy et al., 2013), or even for suppressive roles in murine hematopoietic lines (Edinger et al., 2003) or prostate cancer (Steffan et al., 2014), deserves further investigation. Although this study focused on sporadic cancers, the protumorigenic roles herein described

Figure 6. Lineage-Specific Modulation of RAB7 by SOX10 (A) Quantitative RT-PCR analyses of RAB7, PAX3, and SOX10 expression (relative to 18S mRNA) in UACC-62 melanoma cells expressing the indicated siRNAs. Here and in (B)–(D), data are shown at 72 hr posttransfection. Error bars in (A) and (C) correspond to the SEM of two independent experiments performed in triplicate. (B) RAB7, SOX10, and MITF mRNA levels in total cell extracts isolated from the indicated melanoma cell lines expressing control or SOX10 siRNA. 18S mRNA levels are included as a loading reference. (C) RAB7 and MITF mRNA levels (relative to 18S mRNA levels) determined by qRT-PCR in the indicated melanoma cell lines transduced with siC or siSOX10. (D) Impact of siSOX10 on RAB7 and MITF protein levels monitored by immunoblotting in the indicated cell lines. b-actin is shown as the loading control. (E) Immunoblots of total cell extracts isolated from the indicated melanoma cell lines and probed for RAB7, SOX10, MITF, and a-tubulin (loading control). Highly invasive melanoma cell lines (as shown in Figure 3D) are marked in green. (F) Positive correlation between relative mRNA levels of RAB7 versus SOX10 (estimated with respect to HPRT mRNA) in human melanoma biopsies (Pearson correlation coefficient, r = 0.6). (G) Representative IHC for RAB7 (upper panels, pink) and SOX10 (lower panels, brown) in normal skin (a) and the indicated benign (b) or malignant (c–e) melanocytic lesions. Stromal cells can be distinguished by negative SOX10 staining. Scale bars represent 200 mm.

Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc. 73

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

for RAB7 might offer a mechanistic framework to explain the development of melanomas in a subset of patients with Charcot-Marie-Tooth 2B, a hereditary disease associated with activating mutations in the RAB7 gene (Greene et al., 1980; McCray et al., 2010). In summary, despite the massive number of genetic and epigenetic alterations that melanoma cells accumulate during tumor progression, these cells wire endolysosomal programs (via RAB7) in a rather uniform and unique manner. This lineage-specific wiring is exploited by melanoma cells to favor progression, but it also serves as both a vulnerability and a potential marker for patient prognosis. EXPERIMENTAL PROCEDURES Gene Expression Analyses in Cultured Cells and Tissue Specimens Details of the human cancer cell lines used in this study, isolation of skin melanocytes and fibroblasts, CGH analysis, RNA sequencing, protein immunoblotting and immunofluorescence in cultured cells, and histological analyses in benign and malignant melanocytic biopsies (including single-cell confocal analysis of gene expression) are described in Supplemental Experimental Procedures. Human tumor biopsies were obtained from the i+12 Biobank (RD09/ 0076/00118) of the Hospital 12 de Octubre and the Spanish Hospital Biobank Network, under appropriate ethical protocols by their Clinical Investigation Ethical Committees. Gene Set Enrichment Analysis in Multitumor Data Sets GSEA was applied to previously reported data sets from the NCI-60 panel (Scherf et al., 2000; Shankavaram et al., 2007) and the Cancer Cell Line Encyclopedia (Barretina et al., 2012). GO terms (biological process, cellular component, and molecular function) from levels 3–19 were also evaluated. Genes were ranked using the t statistic. After Kolmogorov-Smirnoff testing, those pathways showing false discovery rates <0.25 were considered enriched in the classes under comparison. Estimation of the Prognostic Value of RAB7 Expression in Human Melanomas Complete 10-year follow-up survival data were available for 112 patients (15 cases of RGP and 97 cases of VGP). Immunohistochemistry (IHC) scoring was performed blind by two pathologists, and linkage to clinical data was performed only after all analyses were completed. The specimens were classified as low-intensity or high-intensity RAB7 staining. Metastasis-free survival and overall survival were determined with a time-to-event analysis carried out by computing time to these occurrences from the date of melanoma diagnosis. For both metastasis and death, patients who did not experience the event at the end of follow-up were considered to be right-censored observations. Time-to-event distributions were estimated using the Kaplan-Meier method, and log-rank homogeneity tests were carried out. To assess the possible prognostic effect of Breslow index, American Joint Committee on Cancer stage, and RAB7 labeling, Cox proportional hazards models were fitted for each event to compute both crude (univariate) and adjusted (multivariate) hazard ratios of each covariate with their corresponding 95% confidence limits. We performed an automatic backward elimination procedure to keep in models only variables with corresponding p values <0.2. For Breslow index, we separated melanomas <1.5 mm from those R1.5 mm. Protein Secretion Assays For the analysis of changes in protein secretion by RAB7 depletion, cells were transduced with lentiviruses coding for control or RAB7 shRNA. Seven days after lentiviral infection, 2 3 106 cells were incubated in serum-free DMEM for 18 hr. The corresponding conditioned medium was harvested, clarified by centrifugation, filtered through a 0.45 mm filter, and then concentrated in Amicon Ultra-15 centrifugal filter devices with Ultracel-3 3 kDa NMWL membrane (Millipore) by centrifugation at 4,000 3 g for 5 hr in a swinging bucket rotor. Ten microliters of concentrated conditioned media was subjected to protein immunoblotting.

74 Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc.

Functional Assays of Tumorigenicity, Metastatic Potential, Lysosome-Related Vesicle Trafficking, Expression Analyses, Quantitative Chromatin Immunoprecipitation, and Microscopy Functional inactivation of RAB7 (by shRNA, siRNA, and the dominant-negative mutant RAB7 T22N), analyses of cell proliferation, colony formation, wound-healing capacity, Matrigel invasion, lung colonization and cytoskeletal alterations (CYTOOChips), and xenograft tumor models are described in Supplemental Experimental Procedures. All experiments with mice were performed in accordance with protocols approved by the Institutional Ethics Committee of the Centro Nacional de Investigaciones Oncolo´gicas (CNIO) and the Instituto de Salud Carlos III. Fluid-phase endocytosis was visualized by uptake of Lucifer yellow (Sigma) and 70 kDa rhodamine-dextran (Invitrogen), as summarized in Supplemental Experimental Procedures. Details of lysosomal activity assays, drug treatments, microscopy, sh/siRNA-mediated silencing, RNA sequencing, and autophagy are also provided in Supplemental Experimental Procedures. Statistical Analyses Statistical tests performed to assess the significance of the computational, histological, and functional data indicated above are described in Supplemental Experimental Procedures. ACCESSION NUMBERS The RNA-sequencing data discussed have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE42735. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, five figures, three tables, and two movies and can be found with this article online at http://dx.doi.org/10.1016/j.ccr.2014.04.030. ACKNOWLEDGMENTS The authors thank their colleagues in the CNIO Melanoma Group, particularly A´ngel Colmenar for excellent technical assistance; Lionel Larribere, David Sa´enz, and Francisco Real for their help and support; Jose´ A. Esteban, F. Javier Carmona, Ashani Weeraratna, Xose´ Bustelo, and Charles Sherr for critical reading of the manuscript; Beatriz Pe´rez-Go´mez for help with the survival analyses; the i+12 Biobank of the Hospital 12 de Octubre for providing melanoma patient samples; Corine Bertolotto for providing MITF siRNA constructs; and Orlando Domı´nguez, Marta Can˜amero, Luis Lombardı´a, and Juan Cruz Cigudosa for their technical assistance. M.S.S. is funded by Projects SAF2011-28317 and Consolider RNAREG from the Spanish Ministry of Economy and Innovation, R01CA125017 from the NIH, and a Team Science Award by the Melanoma Research Foundation. J.L.R.-P. and P.O.-R. are funded by grants FIS 11/025685 and FIS 11/1759, respectively, from the Spanish Ministry of Health. J.L.R.-P. was also supported by grant FMM-2008-106 of Fundacio´n Mutua Madrilen˜a, and P.O.-R. by Red Tematica de Investigacion Cooperativa en Cancer. D.A.-C. and E.P.-G. are recipients of Scientists in Training predoctoral fellowships from the Spanish Ministry of Science and Innovation. M.C. and P.K. are funded by predoctoral fellowships of Fundacio´n La Caixa. E.R.-F. is the recipient of a postdoctoral fellowship from Fundacio´n Cientı´fica de la Asociacio´n Espan˜ola Contra el Ca´ncer, and J.A.J. and H.-W.W. are funded by the American Cancer Society (RSG-12-076-01-LIB). Received: July 17, 2013 Revised: March 13, 2014 Accepted: April 28, 2014 Published: June 26, 2014 REFERENCES Akavia, U.D., Litvin, O., Kim, J., Sanchez-Garcia, F., Kotliar, D., Causton, H.C., Pochanard, P., Mozes, E., Garraway, L.A., and Pe’er, D. (2010). An integrated approach to uncover drivers of cancer. Cell 143, 1005–1017.

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

Alonso, S.R., Tracey, L., Ortiz, P., Pe´rez-Go´mez, B., Palacios, J., Polla´n, M., Linares, J., Serrano, S., Sa´ez-Castillo, A.I., Sa´nchez, L., et al. (2007). A highthroughput study in melanoma identifies epithelial-mesenchymal transition as a major determinant of metastasis. Cancer Res. 67, 3450–3460. Barretina, J., Caponigro, G., Stransky, N., Venkatesan, K., Margolin, A.A., Kim, S., Wilson, C.J., Leha´r, J., Kryukov, G.V., Sonkin, D., et al. (2012). The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 483, 603–607.

Javelaud, D., Alexaki, V.I., Pierrat, M.J., Hoek, K.S., Dennler, S., Van Kempen, L., Bertolotto, C., Ballotti, R., Saule, S., Delmas, V., and Mauviel, A. (2011). GLI2 and M-MITF transcription factors control exclusive gene expression programs and inversely regulate invasion in human melanoma cells. Pigment Cell Melanoma Res. 24, 932–943. Koludrovic, D., and Davidson, I. (2013). MITF, the Janus transcription factor of melanoma. Future Oncol. 9, 235–244.

Bell, R.E., and Levy, C. (2011). The three M’s: melanoma, microphthalmiaassociated transcription factor and microRNA. Pigment Cell Melanoma Res. 24, 1088–1106.

Lawrence, M.S., Stojanov, P., Polak, P., Kryukov, G.V., Cibulskis, K., Sivachenko, A., Carter, S.L., Stewart, C., Mermel, C.H., Roberts, S.A., et al. (2013). Mutational heterogeneity in cancer and the search for new cancerassociated genes. Nature 499, 214–218.

Bucci, C., Thomsen, P., Nicoziani, P., McCarthy, J., and van Deurs, B. (2000). Rab7: a key to lysosome biogenesis. Mol. Biol. Cell 11, 467–480.

Mason, S.D., and Joyce, J.A. (2011). Proteolytic networks in cancer. Trends Cell Biol. 21, 228–237.

Caramel, J., Papadogeorgakis, E., Hill, L., Browne, G.J., Richard, G., Wierinckx, A., Saldanha, G., Osborne, J., Hutchinson, P., Tse, G., et al. (2013). A switch in the expression of embryonic EMT-inducers drives the development of malignant melanoma. Cancer Cell 24, 466–480.

McCray, B.A., Skordalakes, E., and Taylor, J.P. (2010). Disease mutations in Rab7 result in unregulated nucleotide exchange and inappropriate activation. Hum. Mol. Genet. 19, 1033–1047.

Checinska, A., and Soengas, M.S. (2011). The gluttonous side of malignant melanoma: basic and clinical implications of macroautophagy. Pigment Cell Melanoma Res. 24, 1116–1132.

Pinner, S., Jordan, P., Sharrock, K., Bazley, L., Collinson, L., Marais, R., Bonvin, E., Goding, C., and Sahai, E. (2009). Intravital imaging reveals transient changes in pigment production and Brn2 expression during metastatic melanoma dissemination. Cancer Res. 69, 7969–7977.

Cheli, Y., Ohanna, M., Ballotti, R., and Bertolotto, C. (2010). Fifteen-year quest for microphthalmia-associated transcription factor target genes. Pigment Cell Melanoma Res. 23, 27–40.

Polsky, D., and Cordon-Cardo, C. (2003). Oncogenes in melanoma. Oncogene 22, 3087–3091.

Dang, C.V. (2012). MYC on the path to cancer. Cell 149, 22–35. de Duve, C. (2005). The lysosome turns fifty. Nat. Cell Biol. 7, 847–849. Doe, M.R., Ascano, J.M., Kaur, M., and Cole, M.D. (2012). Myc posttranscriptionally induces HIF1 protein and target gene expression in normal and cancer cells. Cancer Res. 72, 949–957. Edinger, A.L., Cinalli, R.M., and Thompson, C.B. (2003). Rab7 prevents growth factor-independent survival by inhibiting cell-autonomous nutrient transporter expression. Dev. Cell 5, 571–582. Garraway, L.A., and Sellers, W.R. (2006). Lineage dependency and lineagesurvival oncogenes in human cancer. Nat. Rev. Cancer 6, 593–602. Garraway, L.A., and Lander, E.S. (2013). Lessons from the cancer genome. Cell 153, 17–37. Goldenring, J.R. (2013). A central role for vesicle trafficking in epithelial neoplasia: intracellular highways to carcinogenesis. Nat. Rev. Cancer 13, 813–820. Gomez, P.F., Luo, D., Hirosaki, K., Shinoda, K., Yamashita, T., Suzuki, J., Otsu, K., Ishikawa, K., and Jimbow, K. (2001). Identification of rab7 as a melanosome-associated protein involved in the intracellular transport of tyrosinaserelated protein 1. J. Invest. Dermatol. 117, 81–90. Greene, M.H., Mead, G.D., Reimer, R.R., Bergfeld, W.F., and Fraumeni, J.F., Jr. (1980). Malignant melanoma and Charcot-Marie-Tooth disease. Am. J. Med. Genet. 5, 69–71. Gupta, P.B., Kuperwasser, C., Brunet, J.P., Ramaswamy, S., Kuo, W.L., Gray, J.W., Naber, S.P., and Weinberg, R.A. (2005). The melanocyte differentiation program predisposes to metastasis after neoplastic transformation. Nat. Genet. 37, 1047–1054. Halaban, R., Cheng, E., Smicun, Y., and Germino, J. (2000). Deregulated E2F transcriptional activity in autonomously growing melanoma cells. J. Exp. Med. 191, 1005–1016. Haq, R., Shoag, J., Andreu-Perez, P., Yokoyama, S., Edelman, H., Rowe, G.C., Frederick, D.T., Hurley, A.D., Nellore, A., Kung, A.L., et al. (2013a). Oncogenic BRAF regulates oxidative metabolism via PGC1a and MITF. Cancer Cell 23, 302–315. Haq, R., Yokoyama, S., Hawryluk, E.B., Jo¨nsson, G.B., Frederick, D.T., McHenry, K., Porter, D., Tran, T.N., Love, K.T., Langer, R., et al. (2013b). BCL2A1 is a lineage-specific antiapoptotic melanoma oncogene that confers resistance to BRAF inhibition. Proc. Natl. Acad. Sci. USA 110, 4321–4326. Hoek, K.S., Eichhoff, O.M., Schlegel, N.C., Do¨bbeling, U., Kobert, N., Schaerer, L., Hemmi, S., and Dummer, R. (2008). In vivo switching of human melanoma cells between proliferative and invasive states. Cancer Res. 68, 650–656.

Raposo, G., and Marks, M.S. (2007). Melanosomes—dark organelles enlighten endosomal membrane transport. Nat. Rev. Mol. Cell Biol. 8, 786–797. Roesch, A., Fukunaga-Kalabis, M., Schmidt, E.C., Zabierowski, S.E., Brafford, P.A., Vultur, A., Basu, D., Gimotty, P., Vogt, T., and Herlyn, M. (2010). A temporarily distinct subpopulation of slow-cycling melanoma cells is required for continuous tumor growth. Cell 141, 583–594. Roy, S.G., Stevens, M.W., So, L., and Edinger, A.L. (2013). Reciprocal effects of rab7 deletion in activated and neglected T cells. Autophagy 9, 1009–1023. Scherf, U., Ross, D.T., Waltham, M., Smith, L.H., Lee, J.K., Tanabe, L., Kohn, K.W., Reinhold, W.C., Myers, T.G., Andrews, D.T., et al. (2000). A gene expression database for the molecular pharmacology of cancer. Nat. Genet. 24, 236–244. Settembre, C., Di Malta, C., Polito, V.A., Garcia Arencibia, M., Vetrini, F., Erdin, S., Erdin, S.U., Huynh, T., Medina, D., Colella, P., et al. (2011). TFEB links autophagy to lysosomal biogenesis. Science 332, 1429–1433. Shakhova, O., Zingg, D., Schaefer, S.M., Hari, L., Civenni, G., Blunschi, J., Claudinot, S., Okoniewski, M., Beermann, F., Mihic-Probst, D., et al. (2012). Sox10 promotes the formation and maintenance of giant congenital naevi and melanoma. Nat. Cell Biol. 14, 882–890. Shankavaram, U.T., Reinhold, W.C., Nishizuka, S., Major, S., Morita, D., Chary, K.K., Reimers, M.A., Scherf, U., Kahn, A., Dolginow, D., et al. (2007). Transcript and protein expression profiles of the NCI-60 cancer cell panel: an integromic microarray study. Mol. Cancer Ther. 6, 820–832. Steffan, J.J., and Cardelli, J.A. (2010). Thiazolidinediones induce Rab7-RILPMAPK-dependent juxtanuclear lysosome aggregation and reduce tumor cell invasion. Traffic 11, 274–286. Steffan, J.J., Dykes, S.S., Coleman, D.T., Adams, L.K., Rogers, D., Carroll, J.L., Williams, B.J., and Cardelli, J.A. (2014). Supporting a role for the GTPase Rab7 in prostate cancer progression. PLoS ONE 9, e87882. Thies, A., Moll, I., Berger, J., Wagener, C., Bru¨mmer, J., Schulze, H.J., Brunner, G., and Schumacher, U. (2002). CEACAM1 expression in cutaneous malignant melanoma predicts the development of metastatic disease. J. Clin. Oncol. 20, 2530–2536. Thurber, A.E., Douglas, G., Sturm, E.C., Zabierowski, S.E., Smit, D.J., Ramakrishnan, S.N., Hacker, E., Leonard, J.H., Herlyn, M., and Sturm, R.A. (2011). Inverse expression states of the BRN2 and MITF transcription factors in melanoma spheres and tumour xenografts regulate the NOTCH pathway. Oncogene 30, 3036–3048. Tsao, H., Chin, L., Garraway, L.A., and Fisher, D.E. (2012). Melanoma: from mutations to medicine. Genes Dev. 26, 1131–1155.

Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc. 75

Cancer Cell Selective Dependence of Melanoma Cells on RAB7

Vazquez, F., Lim, J.H., Chim, H., Bhalla, K., Girnun, G., Pierce, K., Clish, C.B., Granter, S.R., Widlund, H.R., Spiegelman, B.M., and Puigserver, P. (2013). PGC1a expression defines a subset of human melanoma tumors with increased mitochondrial capacity and resistance to oxidative stress. Cancer Cell 23, 287–301. Wang, T., Ming, Z., Xiaochun, W., and Hong, W. (2011). Rab7: role of its protein interaction cascades in endo-lysosomal traffic. Cell. Signal. 23, 516–521. Wang, T., Zhang, M., Ma, Z., Guo, K., Tergaonkar, V., Zeng, Q., and Hong, W. (2012). A role of Rab7 in stabilizing EGFR-Her2 and in sustaining Akt survival signal. J. Cell. Physiol. 227, 2788–2797. Widmer, D.S., Cheng, P.F., Eichhoff, O.M., Belloni, B.C., Zipser, M.C., Schlegel, N.C., Javelaud, D., Mauviel, A., Dummer, R., and Hoek, K.S. (2012). Systematic classification of melanoma cells by phenotype-specific gene expression mapping. Pigment Cell Melanoma Res. 25, 343–353.

76 Cancer Cell 26, 61–76, July 14, 2014 ª2014 Elsevier Inc.

Williams, K.C., and Coppolino, M.G. (2011). Phosphorylation of membrane type 1-matrix metalloproteinase (MT1-MMP) and its vesicle-associated membrane protein 7 (VAMP7)-dependent trafficking facilitate cell invasion and migration. J. Biol. Chem. 286, 43405–43416. Zerial, M., and McBride, H. (2001). Rab proteins as membrane organizers. Nat. Rev. Mol. Cell Biol. 2, 107–117. Zhang, M., Chen, L., Wang, S., and Wang, T. (2009). Rab7: roles in membrane trafficking and disease. Biosci. Rep. 29, 193–209. Zhuang, D., Mannava, S., Grachtchouk, V., Tang, W.H., Patil, S., Wawrzyniak, J.A., Berman, A.E., Giordano, T.J., Prochownik, E.V., Soengas, M.S., and Nikiforov, M.A. (2008). C-MYC overexpression is required for continuous suppression of oncogene-induced senescence in melanoma cells. Oncogene 27, 6623–6634.