|Home | About | Journals | Submit | Contact Us | Français|
Activating mutations in NRAS and BRAF are found frequently in cutaneous melanomas. Because concurrent mutations of both BRAF and RAS are extremely rare, it is thought that transformation by RAS and BRAF occurs through a common mechanism. Also, there is evidence for a relationship of synthetic lethality between NRAS and BRAF oncogenes that leads to selection against cells with a hyperactive mitogen-activated protein kinase (MAPK) pathway. However, it is not known whether the hyperactivation of the MAPK pathway by overexpression of either oncogene alone could also inhibit melanoma tumorigenesis. Here, we show that in melanoma cells with oncogenic BRAF (mBRAF), high levels of mBRAF induce hyperactivation of ERK and senescence-like phenotype and trigger autophagy by inhibiting the mammalian target of rapamycin complex signaling. Growth inhibition and cell death caused by high mBRAF levels are partially rescued by downregulation of BRAF protein or inhibition of autophagy, but not by inhibition of the MAPK or apoptotic pathways. In nude mice, growth of mBRAF-overexpressing tumors is inhibited. Quantitative immunohistochemical analysis of human melanomas and cell lines showed a significant positive correlation between the levels of BRAF protein and autophagy marker light chain 3. Our data suggest that high oncogenic BRAF levels trigger autophagy, which may have a role in melanoma tumor progression.
Activating mutations in RAS and the RAS effector BRAF, which activate the mitogen-activated protein kinase (MAPK) signaling pathway, are found in a majority of melanomas (Demunter et al., 2001; Davies et al., 2002; Solit et al., 2006). As the concurrent mutations of both BRAF and RAS are extremely rare, it is thought that transformation by RAS and BRAF occurs through a common mechanism involving the MEK-ERK signaling cascade (Sensi et al., 2006). However, the specific role of oncogenic BRAF in melanoma remains to be understood (Karreth et al., 2009). For example, it has been shown that BRAF is not required for MAPK activation in RAS-transformed melanocytes (Wellbrock et al., 2004; Dumaz et al., 2006). Although melanoma cell lines with activating mutation in BRAF are more sensitive in vitro to MEK inhibition than RAS mutant cell lines (Solit et al., 2006), there was no correlation in vivo between the activation status of BRAF and the proportion of phosphor-ERK-positive cells or clinical course of the disease (Uribe et al., 2006). Moreover, sorafenib, a multikinase inhibitor that also inhibits BRAF kinase, has little or no antitumor activity in advanced melanoma patients as a single agent (Eisen et al., 2006).
Paradoxically, hyperactivation of RAS or BRAF and high levels of MAPK activities are also known to induce senescence and inhibit cell proliferation (Serrano et al., 1997; Braig and Schmitt, 2006). In both the in vitro and in vivo model systems, activated BRAF has been shown to induce only nonmalignant changes in melanocytes, consistent with the hypothesis that BRAF induces a biphasic cellular response, i.e., an initial proliferative burst followed by senescence (Michaloglou et al., 2005; Dankort et al., 2007). Thus, a majority of benign melanocytic nevi, which harbor activating BRAF mutations, also exhibit hallmarks of senescence (Michaloglou et al., 2005), suggesting that there may be an upper threshold for MAPK activation above which senescence is reactivated even in tumor cells. Thus, in human melanoma cells that carry mutant NRASQ61R, expression of BRAFV600E activates senescence, suggesting a relationship of synthetic lethality between the NRAS and BRAF oncogenes that results in selection against cells with a hyperactive MAPK pathway (Petti et al., 2006). Although it is known that additional genetic changes are required for cellular escape from BRAF-induced senescence (Michaloglou et al., 2005; Patton et al., 2005; Dankort et al., 2009) and that activated BRAF is required for tumor growth and maintenance (Hoeflich et al., 2006), the relationship between the BRAF oncogene dosage and melanoma tumor progression and its responsiveness to treatment with BRAF kinase inhibitors is not known.
Here, we show that high mBRAF expression not only inhibits cell growth and triggers autophagy by inhibiting mTOR signaling, but also inhibits tumor growth in vivo by hemorrhagic necrosis. In human melanoma cell lines and tumor tissues, expression levels of BRAF protein showed a positive correlation with a marker of autophagy. These findings suggest a role for BRAF dosage-dependent autophagy in melanoma development and progression.
A survey of a representative panel of primary and metastatic human melanoma cell lines with wild-type or mutant RAS or BRAF alleles showed that although melanoma cell lines have nearly three- to fivefold more BRAF protein and RAF kinase activity compared with cultured melanocytes, there was no clear correlation between the BRAF mutation status or BRAF expression levels and the RAF kinase activity or pERK1/2 (see Supplementary Data, Supplementary Figure S1 online). For example, metastatic melanoma cell lines 451Lu harboring the mutant BRAF allele and SK-MEL-2 carrying a mutation in RAS have similar levels of BRAF protein, RAF kinase activity, and pERK. To investigate the effect of hyperstimulation of the MAPK pathway in melanoma cells, we transfected oncogenic BRAFV600E (mBRAF) expression plasmid into melanoma cell lines 451Lu and SK-MEL-2 and generated stable cell lines expressing nearly twofold more BRAF protein compared with vector-transfected cells (Figure 1a, left and Figure 2a). The two high-mBRAF 451Lu stable clones, mBRAF2 and mBRAF3, isolated from two separate transfections, showed a decreased rate of proliferation despite having increased pERK levels (Figure 1b, left; Supplementary Figure S2a and S2b). Similarly, in a stable clone of 451Lu transfected with an inducible expression plasmid ER-BRAFV600E, a dose-dependent inhibition of growth was observed when mBRAF expression was increased (up to ~20%) by treatment with tamoxifen (Figure 1a and b, right, and Supplementary Figure S2c). Tamoxifen treatment had no effect on the growth of the parental 451Lu cells. In a clonogenic survival assay, 451Lu-mBRAF clones produced 50% fewer colonies than vector control cells (Figure 1c). To rule out the possibility that the stable clones represent clonally expanded, rare, and slow-growing G418-resistant variants of parental 451Lu melanoma cells, we studied the effect of transient overexpression of mBRAF in 451Lu cells as well as WM35 (a primary melanoma cell line that carries the mutant BRAF allele). Both WM35 and 451Lu cells showed significant growth inhibition similar to the stable clones (Supplementary Figure S3a). Transient expression of mBRAF in additional melanoma cells lines (SK-MEL-28, Hs294T, and WM115) also resulted in variable growth inhibition (Supplementary Figure S4), confirming that high mBRAF levels inhibit melanoma cell proliferation. Interestingly, overexpression of BRAF and high pERK levels did not inhibit the growth of melanoma cell line SK-MEL-2, which carries wild-type BRAF and mutant RAS gene (Figure 2a–c), suggesting that hyperactivation of the oncogenic BRAF and MAPK pathway in melanoma cells with activated BRAF, but not activated RAS, results in growth inhibition.
As activated BRAF induces senescence in melanocytes, we asked whether high mBRAF levels can reactivate senescence in melanoma cells. We used senescence-associated-β-galactosidase, p16 expression, and acetylated histone H3 (AcH3K9) as markers of senescence. In 451Lu-mBRAF, we observed a threefold increase in senescence-associated-β-galactosidase stained-positive cells (Figure 3a), upregulation of p16 protein and a decrease in acetylated histone H3 (Figure 3b). Similar increase in senescence-associated-β-galactosidase-positive cells was seen in SK-MEL-2 mBRAF stable clone, and also in WM35 and 451Lu cells transiently transfected with mBRAF (Supplementary Figure S3b). Together, these data suggest that overexpression of mBRAF (re)activates senescence in both mutant RAS and mutant BRAF melanoma cells, whereas high BRAF levels only inhibit the growth of cells expressing endogenous BRAFV600E.
To understand the mechanisms of growth inhibition by high mBRAF, we evaluated cell cycle distribution and TUNEL staining. In 451Lu-mBRAF cells, although there seems to be a small increase in the S-phase, there was no change in the proportion of cells in the G1 or G2/M phases of the cell cycle or the number of TUNEL-positive cells between vector and 451Lu-mBRAF clones (Figure 3c). Addition of 5 μM fmk-ZVAD, a pan-caspase inhibitor, to the culture medium did not rescue growth of 451Lu-mBRAF clones (Figure 3d). Treatment of vector control 451Lu cells with the same concentration of fmk-ZVAD modestly increased their survival (Figure 3d) and was sufficient to rescue, partially but significantly, tumor necrosis factor-α-induced apoptosis (Supplementary Figure S5). These data show that cell cycle arrest and apoptosis are not primary mechanisms of growth inhibition caused by high mBRAF levels.
We asked whether autophagy, an alternative form of cell death, could account for the growth inhibition caused by mBRAF overexpression. First, immunofluorescence staining for the autophagy marker microtubule-associated protein light chain 3 (LC3, a mammalian homolog of Atg8, a yeast protein involved in autophagy) showed punctate staining of LC3-positive autophagic vesicles in mBRAF3 (Figure 4a, top). The specificity of LC3 staining was verified by a similar pattern of accumulation of transfected green fluorescent protein (GFP)–LC3 fusion protein (Kabeya et al., 2000) in vesicles in 451Lu-mBRAF2 cells but not in control cells. Treatment with brefeldin A, an inducer endoplasmic stress and autophagy (Turcotte et al., 2008), also produced a similar pattern of GFP–LC3 accumulation in vector control 451Lu cells (Figure 4a, lower) showing the prevalence of autophagosomes in mBRAF cells. Biochemically, the autophagosome-associated lipidated form of LC3 (LC3-II) can be distinguished from free LC3 (LC3-I) by immunoblotting (Kabeya et al., 2000). Consistent with the morphological findings, a higher amount of lipidated LC3-II was seen in 451Lu-mBRAF clones (Figure 4a). Staining with the fluorescent dye monodansyl cadavarine, which labels large acidic autophagosome lysosomes (Supplementary Figure S6a), and transmission electron microscopy that showed characteristic double-membrane autophagic vesicles (Figure 4a) and numerous vacuoles (Supplementary Figure S6b) supported our data on activation of autophagy in melanoma cells overexpressing mBRAF.
Activation of autophagy in 451Lu-mBRAF cells raised the possibility that autophagic cell death might contribute to growth inhibition by high mBRAF. To test this, we treated control and mBRAF cells with 3-methyladenine (3-MA), an inhibitor of autophagy (Petiot et al., 2000), and evaluated the proportion of dead cells. In untreated 451Lu-mBRAF clones, there were 2–3 times more dead cells compared with vector-transfected 451Lu cells. Treatment with 3-MA for 48 hours significantly reduced the number of dead cells in the mBRAF clone, whereas 3-MA treatment had no effect on control cells (Figure 4d). A reduction in the amount of lipidated LC3-II in 3-MA-treated cells verified inhibition of autophagy by 3-MA (Supplementary Figure S6c). Conversely, the MTT assay showed that treatment with 3-MA resulted in a small, but significant, increase in growth of mBRAF cells (Supplementary Figure S6d).
As mammalian target of rapamycin (mTOR) is known to regulate autophagy, we tested whether hyperactivation of BRAF modulates mTOR activity. Western blot analysis of components of the mTOR complexes (Figure 4e) showed an increase in phosphorylation of mTOR at Ser2481 and an increase in Rictor, which are associated with the rapamycin-insensitive mammalian target of rapamycin complex-2 (mTORC2). There was no marked change in phosphorylation at Ser2448 and the expression of Raptor, which, however, are associated with mTORC1 that negatively regulates autophagy (Copp et al., 2009). Treatment of 451Lu vector control cells with autophagy-inducer brefeldin A also caused a similar increase in Rictor (lane 4, Figure 4e). Decreased signaling by mTORC1 in mBRAF cells is also indicated by reduced phospho-4E-BP1 (the translational repressor of eukaryotic elongation factor-binding protein) (Figure 4f). Similarly, phosphorylation of Akt at Thr308 and tuberin (TSC2) at Thr1462, both components upstream of mTORC1, is also decreased in mBRAF cells (Figure 4f). These data suggest that hyperactivation of BRAF inhibits Akt-mTORC1 signaling and thereby attenuates the negative regulation of autophagy by mTORC1. Thus, high levels of oncogenic BRAF appear to inhibit growth and induce autophagy-associated cell death by modulating the mTOR signaling pathway that is reported to be activated in the majority of melanomas (Karbowniczek et al., 2009).
To test whether pharmacological inhibition of mTOR by rapamycin also induces autophagy and inhibits melanoma cell growth, we treated 451Lu cells with rapamycin (0.1–2 μM) and measured cell proliferation (by MTT assay) and induction of autophagy (by GFP-LC3 staining). As shown in Supplementary Figure S7, treatment with 2 μM rapamycin caused significant growth inhibition and vesicular accumulation of GFP-LC3, showing that inhibition of the rapamycin-sensitive mTORC1 leads to induction of autophagy in 451Lu melanoma cells.
Next, we asked whether autophagy-associated growth inhibition by high mBRAF can be reversed by reducing BRAF levels. For this, we used BRAF small hairpin RNA (shRNA) lentiviral transduction followed by selection in puromycin. First, we titrated the virus and used a titer to achieve ~20% decrease in BRAF protein levels (Figure 5a). This level of BRAF knockdown in control 451Lu cells resulted in a small but significant decrease in growth. In contrast, a similar knockdown of BRAF in 451Lu-mBRAF cells resulted in a significant increase in growth (Figure 5b), suggesting that a high level of activated BRAF is required for growth suppression. Knockdown of BRAF also decreased senescence as indicated by the p16 levels (Figure 5c).
Infection with higher titers of BRAF shRNA lentivirus, which caused a further decrease (40–50%) in BRAF levels, significantly inhibited proliferation of both control and 451Lu-mBRAF cells (Supplementary Figure S8a and b), as expected, showing that a threshold level of BRAF is required for survival of melanoma cells, whereas high BRAF levels are growth inhibitory.
Next, we tested whether sustained high activation of BRAF downstream effector MEK is also required for growth suppression by high mBRAF. We treated cells with MEK inhibitor U0126 at a concentration (0.5 μM) that inhibits the growth of control 451Lu cells by 30% after a 7-day treatment. A similar treatment, however, did not significantly affect the survival or growth of 451Lu-mBRAF cells (Figure 5d). Western blot analysis did not reveal a marked difference in the degree of inhibition of ERK1/2 phosphorylation by U0126 in vector control compared with mBRAF cells (Supplementary Figure S9). Together, these data suggest that a high level of oncogenic BRAF, but not sustained activation of the MEK-ERK pathway, suppresses melanoma cell growth.
To investigate the relevance of autophagy triggered by high mBRAF to melanoma growth in vivo, we used the mouse xenograft model. We injected vector control and 451Lu-mBRAF cells (1 × 106) into flanks of 4- to 6-week-old athymic nude mice and measured the kinetics of tumor growth. As shown in Figure 6a, mBRAF tumors grew significantly slower, and by day 40, the tumors were up to 45% smaller than the vector control tumors (two-tailed Student’s t-test, P = 0.03). Kaplan–Meier analysis (percent animals surviving with a tumor volume of ≥1000 mm3) showed significant difference in survival of control and 451Lu-BRAF-injected mice (Figure 6a; P = 0.0035 for mBRAF2 and P = 0.0003 for mBRAF3).
Interestingly, after 3–4 weeks of growth, most mBRAF tumors appeared red (hemorrhagic), became progressively necrotic, and showed a concave surface contour (Figure 6b and Supplementary Figure S10). At day 60, when all mice were killed, 7 of 10 mBRAF2 and 8 of 9 mBRAF3 mice showed varying levels of necrosis as scored by the surface diameter of the tumor-showing necrosis (Carswell et al., 1975), whereas none (0/9) of the vector control tumor-bearing mice showed necrosis (Figure 6b and table below). Histochemical staining of the tumor sections showed robust angiotropic growth of cells in control tumors, whereas considerable engorgement and congestion of blood vessels accompanied by hemorrhagic necrosis could be seen in mBRAF tumors with ghost cells (cell shadows) making up the bulk of the tumor (Figure 6c). Immunohistochemical analysis of LC3 showed higher expression in mBRAF tumor cells compared with vector control tumors (Figure 6c). Electron microscopy of residual mBRAF tumors showed characteristic ultrastructural features of necrosis (Figure 6d), such as irregularly condensed chromatin, breakdown of plasma membrane and organelles, and clear cytoplasm with vacuoles (Romanko et al., 2004). These data suggest that activation of mBRAF induces autophagy and inhibits tumor growth in this in vivo mouse xenograft model.
To test whether in human melanomas expression levels of BRAF correlate with expression levels of autophagy marker LC3, we performed automated quantitative immunohistochemistry analysis (AQUA) using a tissue array consisting of primary (n = 16) lymph node (n = 12) and distant metastases of melanoma (n = 12), and a representative set of human cutaneous melanoma cell lines (n =5). The range of AQUA signal intensities for BRAF and LC3 are shown in Table 1 and representative staining of two primary melanoma tumor cores is shown in Figure 7. Core 36 represents a tumor with low BRAF (AQUA score: 12.96) and LC3 (37.87) expression, and Core 39 represents a tumor with high BRAF (35.86) and LC3 (51.91) expression. Spearman’s correlation analysis of intensity values of the three markers showed that BRAF and LC3 are significantly correlated among all melanoma samples (r = 0.591; P = 0.002). In contrast, expression of MAP2, a neuronal differentiation marker that we previously showed to be expressed in cutaneous melanomas (Fang et al., 2001), showed no correlation with BRAF levels (Table 1). These data suggest that high levels of oncogenic BRAF correlate with expression of autophagic marker in human melanoma, implicating BRAF dosage-dependent autophagy as a determinant of melanoma tumor progression.
In this study, we show that overexpression of BRAFV600E in melanoma cells induces not only senescence-like features but also autophagy-associated growth inhibition both in vitro and in vivo. Oncogene-induced senescence and growth arrest in vivo are considered barriers to tumorigenesis (Prieur and Peeper, 2008) and the molecular details of oncogene-induced senescence and the mechanisms of its evasion have been extensively investigated (Wajapeyee et al., 2008). The effect of different oncogene expression levels on cell fate has been investigated (Chen et al., 2008). Using a regulated expression model system, it was proposed that a multistep Ras tumorigenesis model requires at least three distinct oncogenic events––low level of oncogenic activation that stimulates proliferation, spontaneous upregulation of activated Ras levels that induce senescence and, finally, evasion of senescence and tumor progression (Sarkisian et al., 2007). Distinct threshold levels of Myc in vivo also seem to determine its biological output, suggesting that low-level deregulated Myc may be a more efficient initiator of oncogenesis than overexpressed Myc (Murphy et al., 2008). Similarly, an in vivo mouse model of BRAFV600E expression regulated by its own promoter also suggested that the distinct biological outcomes might be due to differences in the level of oncogene expression or activation (Dankort et al., 2007). Our data show that overexpression of oncogenic BRAF even in highly aggressive metastatic melanoma cells harboring endogenous activated BRAF can alter their growth.
The differences in biological responses to oncogene dosage levels are highlighted by our observation that even among melanoma cell lines, the effects of overexpression of BRAFV600E are dependent on the endogenous activated oncogene, RAS versus. BRAF. Thus, although metastatic melanoma SK-MEL-2 cells that carry a mutation in RAS, but not in BRAF, are also susceptible to induction of senescence by mBRAF, they do not undergo proliferative arrest. This is in contrast to the observations that expression of activated N-RAS in metastatic melanoma cells harboring BRAF mutation causes growth inhibition because of cell cycle arrest in the G0/G1 phase and activation of senescence-associated markers (Petti et al., 2006). Although the mutually exclusive occurrence of mutations in RAS and BRAF genes in melanoma is thought to indicate a common mechanism of action involving the downstream MEK–ERK pathway, BRAF appears to be dispensable for the RAS-activated MAPK pathway. In RAS-transformed melanocytes and melanoma cells harboring oncogenic RAS, depletion of BRAF did not block MEK–ERK signaling or cell cycle progression (Wellbrock et al., 2004). In addition, although overexpression of oncogenic BRAF induces senescence, it has been suggested that it is the inactivation by the inhibitory NF1 feedback loop rather than hyperactivation of the RAS pathway that acts a senescence trigger (Courtois-Cox et al., 2006). Thus, activated RAS and BRAF appear to activate common as well as distinct signaling mechanisms.
The observation that a pan-caspase inhibitor failed to rescue proliferation of mBRAF clones, together with the absence of cell cycle arrest and TUNEL-positive cells, prompted us to explore alternative cell death mechanisms. Autophagy is activated as a cell survival mechanism in response to nutrient or growth factor starvation. However, prolonged or extensive autophagy is known to contribute to cell death (Qu et al., 2003; Galluzzi et al., 2008). Although it has been documented that several oncogenes inhibit autophagy and tumor suppressors induce autophagy, the role of autophagy in tumor development and progression remains controversial (Maiuri et al., 2008). Induction of autophagy by high BRAFV600E may seem contradictory to the proposed role of oncogenes, but it is consistent with the complex relationship between autophagy and tumorigenesis (Maiuri et al., 2008). In NIH3T3 cells, overexpression of mutant H-Ras/Val12 has been shown to have a role in the regulation of nutrient starvation-induced autophagy through MAPK-independent pathways and through the class I PI3 kinase signaling pathway (Furuta et al., 2004). However, it must be noted that in colon cancer HT29 cells, overexpression of Ras/Val12 stimulates, rather than inhibits, autophagy (Pattingre et al., 2003). Thus, regulation of autophagy by oncogene over-expression appears to be cell type-specific. We show that high levels of BRAFV600E not only induce autophagy-associated growth inhibition but also correlate with higher expression of autophagy marker LC3 in human melanoma tumors, suggesting a tumor-suppressive role for BRAF-induced autophagy in melanoma.
Autophagy is negatively regulated by mTOR signaling (Lum et al., 2005). There are two distinct mTOR complexes, mTORC1 and mTORC2, that consist of distinct mTOR-binding proteins and also have different substrate specificities (Corradetti and Guan, 2006). We show that overexpression of mBRAF can downregulate mTOR signaling, specifically of mTORC1, which negatively regulates autophagy. On the basis of the recently published observations and our data that show an increase in the phosphorylation of mTOR on Ser2481 and an increase in Rictor, which are associated with mTORC2 (Copp et al., 2009), we suggest that cross talk between BRAF and mTOR signaling has a role in determining the fates of neoplastic melanocytes. This is supported by the observation that BRAF kinase activity is inhibited by Ras homolog enriched in the brain (Rheb) that binds and regulates mTOR kinase signaling, which is reported to be upregulated in the majority of melanomas (Karbowniczek et al., 2004, 2009; Long et al., 2005). Recently, GOLPH3, a Golgi-located protein that modulates mTOR signaling and rapamycin sensitivity in cancer, including melanoma, has been shown to be a previously unknown oncogene supporting a role for mTOR signaling in melanoma cell survival (Scott et al., 2009).
Although it has been proposed that the inhibition of autophagy could be used as the treatment of cancer, the role of autophagy in tumor progression and the benefit of induction of autophagy in cancer therapy remain to be settled (Høyer-Hansen and Jäättelä, 2008). Our data on oncogenic BRAF dosage-dependent induction of autophagy and growth inhibition in both primary and metastatic cells raises the possibility that in melanomas with activated BRAF, autophagy can serve not only as a prognostic indicator but also as an alternative cell death pathway that can be exploited for treatment.
Melanoma cell lines WM115, SK-MEL-2, −28, −31, and human embryonal carcinoma cell lines, NT2/D1 and HEK293T, were purchased from American Type Culture Collection (Manassas, VA). WM35 and 451Lu melanoma cells were provided by Dr M Herlyn (Wistar Institute, Philadelphia, PA) and grown as described (Fang et al., 2001). Neonatal foreskin melanocytes were isolated (Fang et al., 2001) and cultured in phorbol 12-myristate 13-acetate-free medium supplemented with growth supplement (Cascade Biologics, Portland, OR).
BRAF expression plasmids pMCEFplink, pMCEFBRAFV600E, pEF-BRAFV600E, and pEFplink were from Dr R Marais (Institute of Cancer Research, London, UK) and pLXSP3-G BRAFVE:ERT2 was from Dr M McMahon (CRI, University of California, San Francisco, CA). LC3 cDNA plasmid was from Dr N Mizushima (Tokyo University, Tokyo, Japan).
Anti-Raf-B, anti-MAPLC3 (Santa Cruz Biotech, Santa Cruz, CA), anti-p44/42 MAPK, anti-phospho-p44/42 MAPK (Thr202/Tyr204), anti-mTOR, anti-phospho-mTOR (Ser2448), anti-phospho-mTOR (Ser2481), anti-Raptor, anti-Rictor, anti-phospho-AKT (Thr308), anti-phospho-AKT(ser434), anti-AKT, anti-phospho-4E-BP1(Ser65), antiphospho-4E-BP1(Thr37/46), anti-4E-BP1, anti-phospho-TSC2(Thr1462), anti-TSC2 (Cell Signaling Technology, Beverly, MA), anti-p16 (Chemicon, Temecula, CA), anti-LC3 (Novus Biologicals, Littleton, CO), anti-acetyl-histone H3 (Lys9) (Upstate, Temecula, CA), and anti-β-actin antibodies were used. Horseradish peroxidase-conjugated goat anti-mouse IgG and horseradish peroxidase-conjugated donkey anti-rabbit IgG were from GE Health Care (Piscataway, NJ), and goat anti-mouse IgG Alexa 488 was from Molecular Probes (Carlsbad, CA).
Transient transfection was performed using Lipofectamine Plus (Invitrogen, Carlsbad, CA) or NHEM-Neo Nucleofector kit (Amaxa, Gaithersburg, MD) according to the manufacturer’s instructions. For stable clones, transfected 451Lu and SK-MEL-2 melanoma cells were selected and maintained in G418 (1 mg ml−1). The 451Lu ER-BRAFV600E stable clones were established by transfecting with pLXSP3-G BRAFVE:ERT2 plasmid, selected and maintained in 1 μg ml−1 of puromycin. SK-MEL-2 mBRAF and 451Lu-ER-BRAF stable cells represent a mixture of separate clones. In ER-BRAFV600E cells, BRAF was induced by adding 4-hydroxytamoxifen (Sigma, St Louis, MO).
BRAF knockdown was performed using human pLKO.1 lentiviral shRNA plasmids from the RNAi Consortium collection (Open Biosystem, Huntsville, AL). Lentiviruses were produced by transfect-ing scrambled control or BRAF shRNA with viral packing (pSVG) and envelope (pCMV ΔR8.2) plasmids into HEK293T cell lines. Supernatants were harvested and concentrated, and viral copy number was calculated using the quantitative reverse transcription–PCR titration kit (Clontech, Mountain View, CA). Cells were infected with an equal number of scrambled or BRAF viral particles and transduced cells were selected in medium containing 1 μg ml−1 Puromycin.
Cell growth was determined using MTT assays using 1 × 104 cells plated in a 96-well plate. MTT dye (5 mg ml−1, Sigma) was added and viable cell number (OD540) was measured. Cells (1 × 105) were cultured and on days 1, 3, and 5 were fixed in methanol and stained with 0.5% crystal violet solution in 25% methanol. Crystal violet-stained cells were solubilized in 0.1 M sodium citrate in 50% ethanol and measured (OD540). The number of dead cells was measured using the CytoTox-Glo cytotoxicity assay kit (Promega, Madison, WI) that measures dead cell protease activity. Cells (1 × 104) were plated in triplicate wells, treated with autophagy inhibitor 3-MA (Sigma), and after 24 hours cytotoxicity assay reagent (50 μl) was added and luminescence measured. For clonogenic assay, 5 × 103 cells were cultured for 14 days, colonies were fixed, stained with 0.1% crystal violet, washed, and the colonies were counted. For senescence-associated-β-galactosidase-staining, cells plated in triplicate for 72 hours were fixed, incubated with X-gal (5-bromo-4-chloro-3-indolyl-βD-galactopyranoside) (Cell Signaling) at 37 °C overnight, observed under a microscope for the development of blue color, and photographed. Cells stained blue were counted and percent stained cells were estimated.
Apoptotic cells were detected using the Apo-Direct apoptosis detection kit (Phoenix Flow System, San Diego, CA). Cells were fixed and the DNA breaks in the cells undergoing apoptosis were labeled with TdT and FITC-dUTP and scanned by flow cytometry. Propidium iodide staining was performed by incubating the cells with propidium iodide and cell cycle was analyzed by flow cytometry.
Autophagic vacuoles were identified using GFP–LC3 plasmid transfection (Kabeya et al., 2000). Cells on glass coverslips were transfected with pEGFP-LC3 plasmid for 24 hours, fixed in paraformaldehyde, washed, and mounted and images were captured using the Bio-Rad Radiance 2100 MP Rainbow (Hercules, CA) confocal microscope.
Cells cultured under the indicated conditions were lysed in 1% Triton X-100 in phosphate-buffered saline, or in RIPA buffer containing sodium vanadate (Sigma) and a cocktail of protease inhibitors (Roche Diagnostics, Indianapolis, IN). Protein was estimated using the BCA protein assay kit (Pierce Biotechnology, Rockford, IL), separated by SDS–PAGE, and transferred to a Polyscreen membrane (Perkin Elmer Life Sciences, Boston, MA). Membranes were blocked with 5% non-fat dry milk and incubated overnight at 4 °C with primary antibodies. The blots were incubated in appropriate horseradish peroxidase-conjugated secondary antibody and proteins were detected by chemiluminescence (Amersham Biosciences, Piscataway, NJ). β-Actin was used as a control to monitor protein loading variability.
Cells on coverslips were fixed in paraformaldehyde, permeabilized with cold methanol, incubated with LC3 antibody (Santa Cruz), and washed and the bound antibody was visualized using Alexa Fluor 488 (green)-conjugated secondary antibody. Nuclei were stained with propidium iodide and images were acquired using a confocal microscope. Immunohistochemistry was performed on sections of formalin-fixed paraffin-embedded mouse tumors and human melanoma tumors using the Histostain-plus kit (Zymed, San Francisco, CA) according to the manufacturer’s manual. Antigen retrieval was performed by heating in a microwave for 20 minutes in citrate buffer, pH 6.0. Endogenous peroxidase quenching was performed using 3% H2O2. Tissue sections were incubated in primary antibody, washed and incubated in biotinylated secondary antibody followed by horseradish peroxidase-conjugated streptavidin and chromogen AEC, counterstained with hematoxylin, and observed on a Nikon phase contrast microscope (Melville, NY).
Cells on coverslips were washed with phosphate-buffered saline, fixed with 2.5% glutaraldehyde, and processed for transmission electron microscopy. Ultrathin sections were observed and photographed using the Jeol 100 electron microscope (Tokyo, Japan) at the UW Electron Microscope Facility.
Athymic Ncr-nu/nu mice (4–6 weeks, male and female) were purchased from the National Cancer Institute (NCI, Frederick, MD). 451Lu mBRAF and vector control cells (1 × 106) were injected subcutaneously into the right flanks. Tumor size was measured every third day using digital calipers and photographed. All animal experiments were approved by and carried out according to IACUC guidelines.
Human melanoma tissue arrays (Mini Test Array, NCI, USA) consisting of primary lymph node and distant metastatic melanomas and melanoma cell lines were used to quantify the expression of BRAF, LC3, and MAP2 by AQUA as described (Warren et al., 2009). Anti-S100 protein (Biocare Medical, Concord, CA) and Alexa Fluor 555 tagged secondary antibody was used as the melanoma tumor mask. BRAF, LC3, and MAP2 were visualized with Alexa Fluor 647-tagged secondary antibodies. Nuclei were stained with 4,6-diamidino-2-phenylindole. Details of the image acquisition and pixel quantitation analysis are described in Supplementary Data.
Data were analyzed using Prizm 4.0 (GraphPad Software, La Jolla, CA). Mouse tumor data were analyzed using two-tailed t-test and all other data were analyzed using one-way analysis of variance. AQUA data were analyzed using Pearson’s correlation analysis.
This study was supported in part by the National Institutes of Health Grant R21CA125091 to VS.