Inducible Nitric Oxide Synthase Drives mTOR Pathway Activation and Proliferation of Human Melanoma by Reversible Nitrosylation of TSC2

Melanoma incidence is increasing rapidly, and although surgical removal of early-stage disease is often curative, the mean survival time following metastasis is rarely beyond one year due to chemotherapeutically resistant disease (1). Thus, it is critical to identify new molecular targets for melanoma therapy.

Inducible nitric oxide (NO) synthase (iNOS) is constitutively overexpressed in many cancers, including melanoma and gastric (2), breast (3), and head and neck (4) carcinomas. In melanoma, iNOS expression correlates strongly with poor clinical outcome (5, 6). On balance, preclinical data support a protumor role for iNOS/NO (7) by a variety of mechanisms (8–10). In both in vitro and in vivo models, iNOS and NO have been variously shown to enhance carcinogenesis and tumor progression, stimulate angiogenesis, support tumor growth, and promote metastasis (11, 12). Thus, there is substantial interest in the molecular mechanisms through which iNOS and NO enhance tumor growth and aggressive behavior.

The PI3K–AKT–mTOR oncogenic signaling pathway (13) is one potential target of iNOS/NO. mTOR is activated by the small GTPase Ras homolog enriched in brain (Rheb; refs. 14, 15), which is the target of the GTPase-activating domain of the tuberous sclerosis complex 2 (TSC2) protein. TSC2 is a tumor suppressor gene, which, when complexed with TSC1, inactivates Rheb and thus inhibits activation of mTOR and downstream pathway members such as p70 ribosomal S6 kinase (P70S6K) and eukaryotic initiation factor 4E-binding protein 1 (4EBP1).

Limited experimental evidence supports the hypothesis that iNOS/NO-driven mTOR activation can contribute to oncogenic signaling. Some studies have suggested that NO can activate mTOR signaling in noncancer model systems, including murine macrophages (16) and vascular smooth muscle cells (17). However, as yet there have been no studies assessing the ability of cancer-expressed iNOS and endogenously produced NO in physiologic concentrations to activate the mTOR pathway.

In the current study, we assessed the effect of iNOS inhibition with small-molecule antagonists or RNA interference-mediated gene silencing (knockdown) on human melanoma in the chick chorioallantoic membrane (CAM) model, which has been widely used to study cancer growth and angiogenesis in vivo (18). Our data show that in human melanoma, mTOR pathway activation is dependent on iNOS expression/activity, and support a model where iNOS modulates mTOR activity by NO-mediated posttranslational modification of the upstream regulator TSC2.

Authenticated human melanoma lines, A375 and MeWo, were obtained from American Type Culture Collection. The line mel624 was obtained from the laboratory of E.A Grimm, UT MD Anderson Cancer Center (Houston, TX) and authenticated by the MD Anderson Characterized Cell Lines Core Facility. For CAM assays, 2 × 10⁶ log-phase, A375 cells were implanted in 25 μL Matrigel on the CAM of 10-day-old chick embryos (SPAFAS) as described (19, 20). Implanted cells were treated with L-N-(1-Imonoethyl) Lysine (L-NIL) applied topically onto the upper CAM, or transfected 24 hours earlier with iNOS-targeting, scrambled, or nontargeting control siRNA. After 6 days, tumor growth was evaluated macroscopically, and tumor tissue with a minimal amount of surrounding CAM was minced, dissociated into single-cell suspension with type 1A collagenase, and live tumor cells counted by Trypan blue exclusion. Alternatively, excised CAM tissues were homogenized with passive lysis buffer (Boston Bioproducts) for Western blot analysis, or fixed in 10% zinc formalin, embedded in paraffin, cut in 5-μm sections, and analyzed by hematoxylin and eosin (H&E) staining and immunohistochemistry (IHC).

A375, Mel624, and/or MeWo cells were transfected with Lipofectamine according to the manufacturer's (Invitrogen) protocol. At 24 hours after transfection, cells were replated for use in the indicated assays. Plasmid expressing the full-length murine GFP- iNOS sequence was a generous gift from Ignacio Rodriguez-Crespo (Complutense University of Madrid, Madrid, Spain) and has been previously described (21). pMSCV-puro GFP vector (Addgene) was used as a control. Human iNOS-targeting siRNA Stealth RNA duplexes and nontargeting control siRNA (catalog number 46-2002) were from Invitrogen.

L-NIL (hydrochloride) and DEANONOate were purchased from Cayman Chemical. 2-(4-Carboxyphenyl)-4,4, 5, 5–tetramethylimidazoline-1-oxyl-3-oxide (PTIO), S-nitroso-penicillamine (SNAP), and dithiothreitol were from Sigma-Aldrich. Vemurafenib (PLX4032/RG7204) was purchased from Selleckchem. All chemicals were prepared freshly from powdered stock on the day of each experiment. All cell culture reagents were from Invitrogen.

Following SDS-PAGE and transfer to nitrocellulose, membranes were incubated overnight at 4°C with primary antibodies against iNOS (sc-651; Santa Cruz Biotechnology), mTOR (sc-2983; Santa Cruz Biotechnology), p-mTOR Ser 2448 (sc-2971; Santa Cruz Biotechnology), Phospho-4E-BP1 (Thr37/46; sc-9459; Santa Cruz Biotechnology), 4E-BP1 (sc-9977; Santa Cruz Biotechnology), Tuberin/TSC2 (D57A9; sc-3990; Santa Cruz Biotechnology), Hamartin/TSC1 (sc-4906), and Actin (sc-1616; Santa Cruz Biotechnology) followed by appropriate secondary antibody conjugated to horseradish peroxidase. Results were visualized by commercial chemiluminiscence kit (Millipore). Relative signal intensities for total and phospho-proteins were quantitated by densitometry using ImageJ software (NIH, Bethesda, MD), and normalized densitometry values were expressed as ratio of phospho- to total protein levels.

Cells were seeded into 6-well plates and incubated for 1 hour at 37°C with 100 or 300 μmol/L fresh SNAP, 100 μmol/L fresh DEANONOate, and 1,000 or 3,000 μmol/L L-NIL. Cells were washed with ice-cold PBS and lysed in NP-40 and X100 plus protease inhibitors for 10 minutes on ice, then cleared by centrifugation (10,000 g for 10 minutes at 4°C). For TSC2 immunoprecipitation, the antibody (Bethyl Laboratories cat A300-526A) was incubated with lysates on ice for 90 minutes before treatment with protein A beads and incubation at 4°C for 3 hours with gentle rotation. Beads were washed, recovered between each wash by centrifugation (1,000 g for 30 seconds at 4°C), and resuspended in sample buffer before immunoblot analysis.

Formalin-fixed, paraffin-embedded (FFPE) sections (5 μmol/L) of excised human melanoma tumors from CAM were immunostained for Ki-67 (sc-23900 Santa Cruz Biotechnology), S100 (ab14849, Abcam), or HMB45 (Ventana Medical System, Inc), p70 S6K (49D7; Cell Signaling Technology Cat # 2708), Phospho-4E-BP1 (Thr37/46; 236B4; Cell Signaling Technology Cat # 2855), 4E-BP1 (53H11; Cell Signaling Technology Cat # 9644), iNOS (Creative Biolabs), nitrotyrosine (EMD Millipore Corp.). Preimmune normal mouse immunoglobulin G (IgG; Vector Laboratories) and anti-vimentin antibody (BioGenex Laboratories) were used as negative and positive controls, respectively

For Ki-67 staining of CAM tissue, the fractional area occupied by the immunoreactive Ki-67 was quantified with the Bioquant image analysis system (BIOQUANT Image Analysis Corporation) as previously described (22). Random fields from seven independent sections were analyzed per sample and averaged.

A375 cells were transfected with iNOS-targeting or control siRNA as described above. Forty hours after transfection, cells were washed once with phosphate-free Dulbecco's Modified Eagle Medium (Invitrogen), and processed with the Rheb Activation Assay Kit (NewEast Biosciences) as previously described (23). For positive and negative controls, A375 cell lysate, was incubated with GTP-γS or GDP, respectively, for 90 minutes before immunoprecipitation.

The biotin switch assay was performed according to the method of Jaffrey and Snyder (24) with some modifications. Cell lysates were prepared in SDS sample buffer incubated for 16 hours at 37°C in the dark with either 100 μmol/L fresh SNAP, 1,000 μmol/L L-NIL, 300 μmol/L PTIO, and 1 hour with 10 mmol/L DTT room temperature or left untreated, after which they were subjected to biotin-switch assay using the S-Nitrosylated Protein Detection Kit (Cayman Chemical). After the biotin-switch assay, biotinilated proteins were pulled down with streptavidine-agarose beads (Thermo Scientific) and the eluate was resolved by SDS-PAGE and analyzed by immunoblotting.

Cells were seeded into 6-well plates at concentrations of 500/well or 1,000/well in triplicate. After 24 hours, L-NIL was added or cells were transiently transfected with siRNA as described above before changing to serum-containing medium. Cells were incubated for 10 to 14 days to form visible colonies. Colonies were stained with methylene blue, rinsed with water, air-dried and imaged and quantified by using ImageJ, and colony number was counted. Cell colonies were visually identified and colonies more than 50 cells counted (25)

FFPE tissue sections (5 μm) from CAM were processed for terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL). The ApoTag Fluorescein In Situ Apoptosis Detection Kit (Millipore) was used according to the manufacturer's instruction. Five different fields were counted per sample. TUNEL kit: S7110, EMD Millipore Corporation. Mounting Medium with propidium iodide is H-1300 (Vector Laboratories).

Cells were plated at a density of 1 × 10⁵ cells/5 mL medium in 60 cm-diameter dishes and L-NIL added 24 hours later in fresh medium. Adherent cells were harvested 48 hours later by trypsinization and centrifugation. After washing with PBS, propidium iodide was added and the DNA content of stained nuclei analyzed by flow cytometry with the FACSCalibur using FlowJo 7.6 software.

Associations between staining number and intensity levels of iNOS, nitrotyrosine, p-4EBP1, and p70S6K coexpression in clinical melanoma samples were assessed using Fisher exact tests. Overall survival was determined from stage III diagnosis to last known vital status. Patients alive at the last follow-up date were censored. Associations between iNOS expression levels and overall survival were assessed using univariate Cox proportional hazards models. Throughout the paper, error bars indicate SEM, * indicates P < 0.05 significance, and ** indicates P < 0.01 significance.

Nitrotyrosine is a stable NO end product and marker of cumulative iNOS activity (26). To assess the contribution of iNOS to human melanoma growth in a physiologically relevant setting, we established primary A375 melanoma tumors in the CAM assay and treated them with the iNOS-selective competitive antagonist L-NIL (27, 28) or gene knockdown with iNOS-targeting siRNA (29). iNOS knockdown decreased constitutive iNOS gene expression (Fig. 1A), and both siRNA and L-NIL markedly suppressed nitrotyrosine accumulation in A375 tumors grown on CAM (Fig. 1A and B).

Well-vascularized three-dimensional tumor nodules were present by 6 days after inoculation of melanoma cells into Matrigel upon the CAM. Immunohistochemical staining with the melanoma markers S100 (Fig. 2A; ref. 30), and HMB45 (data not shown) as well as conventional H&E sections (Fig. 2B) verified discrete human melanoma tumors on the background of chicken-derived CAM cells. iNOS knockdown (Fig. 2C) or L-NIL (Fig. 2D) suppressed both macroscopic tumor growth across several human melanoma cell lines (A375, Mel624, and MeWo), and the number of viable tumor cells (Fig. 2B and C and Supplementary Fig. S1A and S1B).

The clonogenic survival assay has been shown to correlate with tumor cell proliferative capacity in vivo (31). Both iNOS inhibition and knockdown suppressed in vitro clonogenic growth of A375 cells (Fig. 3A), consistent with the marked reduction in in vivo tumor growth. L-NIL treatment in vitro also arrested A375 cells at the G₂–M phase of the cell cycle (Fig. 3B; ref. 32). iNOS knockdown in vivo also inhibited by over 50% immunostaining for the proliferation marker Ki-67 (Fig. 3C; ref. 33). Furthermore, iNOS knockdown increased by 2-fold apoptosis in vivo as measured by TUNEL assay (Fig. 3D; ref. 34). Thus, the marked melanoma growth suppression observed with iNOS inhibition or silencing is the result of concomitant inhibition of proliferation and increased apoptotic cell death.

Because the mTOR pathway has been shown to be a key mediator of proliferation and survival for melanoma (35) and other tumors (36), we focused on this pathway in two human metastatic melanoma cell lines grown in CAM, A375 (bearing the clinically significant BRAFV600E mutation; Fig. 4A and B), and MeWo (wild-type BRAF; Fig. 4C and D). iNOS-targeting siRNA and iNOS inhibition with L-NIL coordinately downregulated expression of the activated (phosphorylated) form of multiple mTOR pathway members in a dose-dependent fashion (Fig. 4A and B).

Specificity of the iNOS-targeting siRNA is demonstrated by the similar effect of siRNA constructs targeting different regions of the iNOS gene (Fig. 4C), and the lack of effect of two different control constructs derived from scrambled iNOS-targeting sequences (Figs. 2C and 4A and C) We further confirmed specificity of iNOS-targeting siRNA by performing genetic rescue of siRNA-treated cells with a construct expressing GFP fused to murine iNOS (GFP-iNOS; Fig. 5A; ref. 21), which is incompletely homologous with the human ortholog and thus resistant to knockdown (Supplementary Fig. S2). GFP iNOS markedly upregulated iNOS expression and reversed siRNA-mediated inhibition of S6RP and 4EBP1 phosphorylation (Fig. 5B). We also noted modest, but reproducible upregulation of S6RP and 4EBP1 phosphorylation in cells transfected with the GFP-iNOS construct but not GFP alone, consistent with previously published data (37). Taken together, our results demonstrate that endogenous iNOS expression is required to maintain high levels of the constitutive mTOR pathway activation in two human melanoma and that inhibition of iNOS activity or gene expression markedly downregulates mTOR signaling in melanoma cells in a reversible fashion.

S-Nitrosylation is a reversible NO-mediated posttranslational modification of cysteine residues, which can affect protein stability or activity, and thus a candidate mediator of activation by iNOS. To examine whether this mechanism occurs in human melanoma cells, A375 cells were treated with SNAP (a donor of NO), PTIO (an NO scavenger), or L-NIL for 16 hours before harvest of cell lysates and biotin switch assay to detect nitrosylated proteins (38). Constitutively nitrosylated proteins biotinylated during the biotin switch reaction were observed in untreated and SNAP-treated lanes, with a marked reduction in biotinylation intensity in lysates from L-NIL- and PTIO-treated cells (Fig. 6A), demonstrating reversible nitrosylation of endogenous proteins in melanoma.

To assess nitrosylation of specific mTOR pathway proteins, we performed a streptavidin-agarose pulldown after the biotin switch reaction, followed by immunoblot analysis for specific proteins of interest. This method successfully detected nitrosylation of several control proteins previously known to be S-nitrosylated, including Bcl-2 (39) and AKT (Supplementary Fig. S3; ref. 40). We observed endogenous nitrosylation of TSC2, which was enhanced by treatment of the cell lysate with the NO donor SNAP. Reversibility of TSC2 nitrosylation was confirmed by loss of the SNO-TSC2 band after biotin switch with lysate from L-NIL- or PTIO-treated cells (Fig. 6B). However, total TSC2 levels were not affected, suggesting that nitrosylation does not affect TSC2 stability.

TSC1/TSC2 dimerization is a critical checkpoint for downstream Rheb activation required for mTOR pathway signaling. To further characterize the mechanism of NO-mediated control of TSC2 activity, we performed coimmunoprecipitation studies to determine the effect of NO on TSC1/TSC2 binding. A375 cells were treated with two structurally different NO donors DEANONOate and SNAP (Fig. 6C) or L-NIL. We immunoprecipitated cell lysates with anti-TSC2 antibody and performed immunoblotting of immunoprecipitate complexes and crude lysates (total lysate) with anti-TSC1, -TSC2, or -actin antibody. NO treatment decreased the binding of TSC1 to TSC2, whereas L-NIL enhanced binding. These experiments demonstrate that iNOS/NO modulates dimerization of TSC2 with its inhibitory partner TSC1. We next investigated the association of TSC2 nitrosylation with activation of downstream signaling, by assessing Rheb GTPase activity. iNOS-targeting siRNA almost completely inhibited Rheb activation (Fig. 6D), demonstrating control of Rheb activity by iNOS. Although iNOS and NO can potentially also control mTOR pathway activation by acting on AKT or on upstream pathways, which modulate AKT activation (41), we saw no evidence that iNOS inhibition downregulated AKT phosphorylation (Supplementary Fig. S4). We also saw little effect of the NO-dependent guanyl cyclase inhibitor ODQ, on mTOR pathway activation (Supplementary Fig. S5), suggesting that iNOS/NO regulation is independent of this mechanism. Thus, we infer that iNOS/NO indirectly controls Rheb activity by nitrosylation of the immediately upstream regulator TSC2.

We used IHC to examine expression of iNOS, nitrotyrosine, p-4EBP1, and p70S6K in vivo in tissue microarrays (TMA) from metastatic tumors from stage III melanoma patients (representative stains shown in Fig. 7A). Staining for iNOS was discrete and variable among tumors, whereas the NO biomarker nitrotyrosine showed more diffuse staining, consistent with the high diffusivity of NO. Antibodies to phospho-4EBP1 and p70S6K also showed the anticipated cytoplasmic staining.

To explore whether iNOS and nitrotyrosine staining levels correlated with mTOR pathway activation, we evaluated staining number (% of cells positively staining) and intensity for all markers in random fields from three TMA cores (6) and tested for pairwise associations between markers, and between staining intensity and overall and disease-free survival. Of 118 available patient samples, staining data were available for 112 (nitrotyrosine), 108 (iNOS), 103 (p-4EBP1), and 100 (p70S6K) patients. We found a significant correlation between iNOS and nitrotyrosine staining number (P = 0.0043) and intensity (P < 0.0001), consistent with the role of iNOS as the primary NO-producing enzyme in human melanoma cells. We also found statistically significant associations between phospho-4EBP1 and p70S6K number (P < 0.0001) and intensity (P < 0.0001) by Fisher exact test, presumably reflecting upregulation of both markers during the mTOR pathway activation. iNOS staining intensity and nitrotyrosine staining intensity were significantly correlated with p70S6K and nitrotyrosine intensity, respectively (Fig. 7B). Thus, we observed a positive association between iNOS upregulation and the mTOR pathway activation in clinical melanoma specimens similar to what we observed in human melanoma in the CAM assay

We then examined the effect of NO on mTOR signaling downstream of the BRAF inhibitor vemurafenib (PLX4032/RG7204; refs. 39, 40) because it is used clinically in treatment of melanoma. Cells were treated with vemurafenib and the exogenous NO donor DEANONOate, or a combination of both for 3 hours. Western blot analysis revealed that short-term treatment with vemurafenib alone markedly reduced phosphorylation of downstream mTOR pathway members S6RP and 4EBP1 in A375 lysates. We found that exogenously supplied NO reverses mTOR inhibition seen after vemurafenib treatment (Fig. 7C and D), suggesting that NO can act downstream of BRAF to bypass BRAF inhibition, and thus may serve as a potential mechanism of escape (41, 42).

In the present study, we used human melanoma grown in the in vivo CAM model to test the hypothesis that iNOS upregulation by cancer cells enhances tumor growth and survival by driving mTOR pathway activation, a critical mediator of oncogenic signaling. We found that iNOS expression in human melanoma leads to nitrosylation of TSC2, which is associated with impaired dimerization of TSC2 with its inhibitory partner TSC1 and enhanced activation of its target, the small GTPase Rheb, and subsequent activation of downstream members of the mTOR pathway. iNOS-mediated activation of mTOR signaling is oncologically significant, leading to enhanced proliferation as measured by Ki-67 immunostaining and cell-cycle analysis, decreased apoptosis, and macroscopic tumor growth on CAM. We also find evidence that iNOS-driven mTOR activation is clinically relevant because iNOS and nitrotyrosine expressions are positively correlated with S6K and phospho-4EBP1 upregulation by IHC of metastatic human melanoma samples.

In the present study, we functionally dissect the role of endogenous iNOS/NO expression with the selective iNOS inhibitor L-NIL and iNOS-targeting siRNA, thus showing that the upregulation of iNOS previously reported in melanoma and many other cancers is itself sufficient to drive mTOR pathway activation and tumor growth. This is consistent with prior studies showing that iNOS upregulation induces many of the hallmarks of mTOR activation (proliferation, survival, angiogenesis, treatment resistance) and is associated with poor prognosis in human melanoma patients. In fact, downregulation of mTOR activation by iNOS inhibition or knockdown was associated with a striking reduction in melanoma growth on the CAM, similar to that which we have previously reported in mouse xenograft studies (27). We also noted a robust effect of iNOS inhibition or knockdown on in vitro colony formation by human melanoma cells, suggesting that even in vitro iNOS may be required for optimum clonogenicity of founder cells. We also found that exogenously supplied NO reverses mTOR pathway inhibition seen after vemurafenib treatment, suggesting that NO can act downstream of BRAF to bypass BRAF inhibition, and thus is a potential mechanism of escape (40, 42).

Although NO is pleiotropic and can potentially regulate mTOR pathway activation through other mechanism, such as nitrosylation of EGF receptor and other molecules upstream of mTOR signaling (PMID 22878588), our study is also the first to propose nitrosylation of an mTOR pathway member as a biochemical mechanism linking cancer-related iNOS/NO overexpression with pathway activation. NO can alter signal transduction pathways via covalent modification of target proteins through nitration and nitrosylation. We show by biotin switch assay that TSC2, which in the activated state acts to suppress mTOR activation, by forming a Rheb-inactivating complex with TSC1, is constitutively nitrosylated in melanoma cells. Immunoprecipitation experiments demonstrate that iNOS/NO modulates dimerization of TSC2 with its inhibitory partner TSC1, a critical checkpoint for downstream Rheb activation, thus modulating physical association of TSC1 and TSC2 is a plausible and novel mechanism by which iNOS/NO can control mTOR pathway activation. TSC2 has many cysteines in functionally relevant areas of the protein that could potentially be nitrosylated, and ongoing work in our laboratory is focused on identifying specific cysteines that are targets for nitrosylation in melanoma cells.

A limitation of our study approach is the possibility that iNOS is abnormally dysregulated in the CAM system, and thus our results are not relevant to melanoma in human patients. However, melanoma lines used in this study constitutively express iNOS even in tissue culture, and iNOS expression is well documented in human melanoma cell lines and clinical specimens. Also the correlation between iNOS and NO expression levels and S6K and 4EBP1 we observed in human clinical metastatic melanoma specimens supports the relevance of this mechanism to clinical melanoma. Although other NOS isoforms (NOS1, NOS3) may also be expressed in cancer cells and play a role in their biology, we conclude that in our system iNOS is a key driver of NO-mediated signaling in human melanoma cells.

iNOS and other inflammatory molecules have been shown to play pivotal roles in carcinogenesis, tumor progression, proliferation, and apoptosis resistance of human cancers. Molecular therapy targeting cancer-related inflammation is an attractive approach to cancer treatment because iNOS and other inflammatory mediators are broadly overexpressed in a number of human cancers, largely dispensable for untransformed cells, and amenable to pharmacologic inhibition. We and others have described potent antitumor effects of iNOS inhibition, which are both direct (43, 44) and indirect (28), highlighting the dual role of iNOS as signaling molecule and inflammatory mediator. Our results provide a convincing biologic mechanism by which iNOS upregulation activates oncogenic signaling in melanoma cells through the posttranslational modification of TSC2, and set the stage for further investigation into specific mechanisms linking inflammation, redox signaling, and cancer biology.

M.A. Davies has commercial research grant from Glaxosmithkline, Genentech, Merck, Astrazeneca, Oncothyreon, and Myriad, and is a consultant/advisory board member of Genentech, Novartis, and GlaxoSmithKline. J.A Aguirre-Ghiso has commercial research support from ImClone - Eli Lilly and is a consultant/advisory board member of Novartis. A.G. Sikora has commercial research grant from Advaxis Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

Conception and design: E. Lopez-Rivera, P. Jayaraman, A.G. Sikora

Development of methodology: E. Lopez-Rivera, J.E. Chipuk, A.G. Sikora

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Lopez-Rivera, P. Jayaraman, F. Parikh, M.A. Davies, S. Ekmekcioglu, S. Izadmehr, Y. Estrada, A.G. Sikora

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Lopez-Rivera, M.A. Davies, S. Izadmehr, D.R. Milton, E.A. Grimm, J.A. Aguirre-Ghiso, A.G. Sikora

Writing, review, and/or revision of the manuscript: E. Lopez-Rivera, M.A. Davies, S. Ekmekcioglu, D.R. Milton, E.A. Grimm, J.A. Aguirre-Ghiso, A.G. Sikora

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Lopez-Rivera

The authors thank the helpful review and suggestions provided by Dr. Stuart Aaronson, Department of Oncological Sciences, Icahn School of Medicine at Mount Sinai.

This work was partially supported by grants from the NIH National Institute of Dental and Craniofacial Research (5R03DE021741-02; A.G. Sikora and J.A. Aguirre-Ghiso), the NIH National Cancer Institute (1K08CA154963-01A1; A.G. Sikora), The Dr. Miriam and Sheldon G. Adelson Medical Research Foundation (E.A. Grimm), the NIH Melanoma Specialized Program of Research Excellence (NIH P50 CA093459; E.A. Grimm and S. Ekmekcioglu), and Institutional Core Grant NIH CA16672 (MD Anderson Characterized Cell Line Core).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.