Gab2-Mediated Signaling Promotes Melanoma Metastasis

Tumor Specimens and Cell Lines

Sixty-four metastatic melanoma frozen tissue samples were obtained from the tissue banks of the Department of Pathology at Columbia University and the Department of Cancer Genetics at University of Münster, Fachklinik Hornheide, Germany. Tumors were examined histopathologically and those with at least 90% neoplastic tissue were selected. In some samples where tumor cells were admixed with a prominent lymphocytic infiltrate, cells were microdissected using the PixCell II laser capture microdissection system (Arcturus, Mountain View, CA). Twenty-three cell lines derived from primary and metastatic melanomas were obtained from the American Type Culture Collection (Rockville, MD) and the Wistar Institute (Philadelphia, PA) and cultured according to their instructions. A normal human melanocyte cell line was generated from neonatal foreskin as described10 and maintained in melanocyte growth medium-4 with supplements and growth factors (Lonza Inc., Allendale, NJ). The protocol was approved by Columbia University’s Institutional Review Board.

BAC Array CGH

Arrays with ∼19,000 RPCI-11 BAC clones providing an average resolution of 150 kb, were used.11 BAC array CGH was performed as described previously.12 Briefly, 2 μg of reference and test sample genomic DNA were fluorescently labeled using the BioArray CGH labeling system (Enzo Life Sciences, Inc., New York, NY). The DNA was denatured in the presence of random primer at 99°C for 10 minutes, cooled to 4°C, labeled by adding dNTP-cyanine 3 mix (or dNTP-cyanine 5) and Klenow, and incubated overnight at 37°C. Hybridization to the 19K arrays was performed at 55°C for 16 hours using a GeneTAC hybridization station (Genomic Solutions, Inc., Ann Arbor, MI). The hybridized slides were scanned with a Genepix 4200A scanner (Molecular Devices, Inc., Sunnyvale, CA) and image analysis was performed with ImaGene software (Biodiscovery, Inc., Elsegundo, CA). Log₂ ratios were computed and normalized using subgrid loess. Clones were ordered by chromosomal position according to the University of California-Santa Cruz human genome sequence build 36.1 (http://genome.ucsc.edu).

SNP Array

SNPs were genotyped and the copy numbers were estimated using 250K arrays (Affymetrix, Santa Clara, CA). Genomic DNA was cleaved with the restriction enzyme, StyI, ligated with linkers, followed by polymerase chain reaction (PCR) amplification. The PCR products were purified and digested with DnaseI to a size ranging from 250 to 2000 bp. Fragmented PCR products were then labeled with biotin and hybridized to the array. Arrays were washed on the Affymetrix fluidics stations. The bound DNA was then fluorescently labeled using streptavidin-phycoerythrin conjugates and scanned using the Gene Chip Scanner 3000 (Affymetrix). DChip software was used for SNP array data analysis as described previously.13 Data were normalized to baseline array with median signal intensity at the probe intensity level using the invariant set normalization method. A model-based (PM-MM) method was used to obtain the signal values for each SNP in each array. To infer the DNA copy number from raw signal data, we used the Hidden-Markov model based on the assumption of diploidy for normal samples. Mapping information of SNP locations and cytogenetic band were based on curation of Affymetrix and University of California-Santa Cruz. A cutoff of >2.8 copies in more than three consecutive SNPs were defined as amplification.

Quantitative Real-Time PCR

cDNA was synthesized using the Superscript first strand synthesis kit as per instructions of the manufacturer (Invitrogen, Carlsbad, CA). One μg total RNA, 10 mmol/L dNTP mix, and 50 μmol/L oligo(dT)₂₀ were incubated at 65°C for 5 minutes, placed on ice for 1 minute, and cDNA synthesis mix was added. The reaction was incubated at 50°C for 50 minutes and terminated at 85°C for 5 minutes. Real-time PCR was performed with 100 ng of input RNA per reaction containing 1× TaqMan universal PCR Master mix and the appropriate TaqMan probe on a 7300 Real-Time PCR system (Applied Biosystems, Foster City, CA). Gab2 (Hs00373045) and β-Actin (4326315E) TaqMan probes were purchased from Applied Biosystems. All samples were prepared in triplicates on each plate and at least three plates were analyzed. Relative mRNA levels were determined using the comparative C_T method.

Fluorescent in Situ Hybridization

RP11-653J20 and RP11-444N24 BAC clones were obtained from Invitrogen. DNA was prepared from BAC clones using standard methods and labeled by nick-translation using spectrum red or spectrum green dUTP fluorochromes. Spectrum red- or spectrum green-labeled centromeric probes were used to enumerate chromosome numbers (Vysis, Downers Grove, IL). Hybridization signals were scored on at least 20 metaphase spreads of cell lines or 200 interphase nuclei from tissue sections on 4,6-diamidino-2-phenylindole (DAPI)-counterstained slides.

Immunofluorescence and Automated Quantitative Analysis (AQUA) of Tissue Microarrays

For protein expression, tissue microarrays containing specimens obtained from melanocytic nevi (n = 18), primary melanomas (n = 15), and metastases (n = 15) were used. Immunofluorescence staining of tissue microarrays was performed as described previously14 using an anti-Gab2 antibody (Upstate, Charlottesville, VA). The AQUA software linked to the fluorescence microscopy system was used as described previously14 for quantification of the Gab2 protein within the tumor region of each tissue microarray core.

Immunoblotting and Antibodies

Cells were lysed with the PhosphoSafe extraction reagent (Novagen, San Diego, CA) containing protease and phosphatase inhibitors. Protein quantity was determined by Bradford assay (Bio-Rad Laboratories, Inc., Hercules, CA). Proteins were separated by sodium dodecyl sulfate gel electrophoresis under reducing conditions and then transferred to 0.45-μm nitrocellulose membranes by electroblotting. The membranes were blocked with 5% nonfat powdered milk in phosphate-buffered saline and then probed with primary antibodies at a 1:1000 dilution in TBST with 5% milk. The membranes were then washed with TBST and probed with horseradish peroxidase-conjugated secondary antibodies at 1:2000 dilutions in TBST with 5% milk. After washing in TBST, the membranes were developed with the ECL Western blotting detection system (Pierce Biotech, Rockford, IL). The films were scanned using a Molecular Dynamics personal densitometer SI and the protein band density was quantified using the ImageQuant software (Molecular Dynamics, Sunnyvale, CA). The following commercial antibodies were used in this study: Gab2 (Upstate); Gab1, phospho-Akt (Ser-473), total Akt, phospho-PDK1, total PDK1 (Cell Signaling, Danvers, MA); and actin (Santa Cruz Biotechnology, Santa Cruz, CA).

Gene Knockdown Using siRNA

Small interfering RNA against Gab2 and control (scrambled) RNAs were purchased from Dharmacon (Lafayette, CO). Cells (1.5 × 10⁵) were transfected with 300 pmol/L of siRNA using oligofectamine reagent in Opti-MEM I reduced serum medium (Invitrogen). Four hours later, the medium was replaced with culture medium supplemented with 10% fetal bovine serum. Forty-eight hours later the cells were harvested for immunoblotting.

Generation of Gab2-Expressing Stable Cell Lines

Primary melanoma cell lines (WM793, WM278, WM1552, and WM3862) expressing either Gab2 or vector were generated using retroviral transduction as described previously.15,16 Gab2 full-length cDNA, obtained from Origene (Rockville, MD), was subcloned into pBABE-puro retroviral vector (Addgene, Cambridge, MA). Phoenix-Eco (ecotropic) packaging cells were transiently transfected with 20 μg of pBABE-puro retroviral vector by the CaPO₄ co-precipitation method. The packaged viruses released by the Phoenix-Eco cells were used for infecting AM12 (amphotropic) packaging cells to generate stable clones of virus-producing cells and to achieve high viral titers.15 Forty-eight hours after infection, the cells were split into puromycin selection medium (2 μg/ml) and surviving colonies were selected. Expression levels of Gab2 were evaluated by Western blot analysis of AM12 clones. Viral supernatant was then collected from these producer cells throughout a 24- to 48-hour period, filtered through a 0.45-μm filter to remove cell debris, and used for infecting the melanoma cells. Lysates from infected cells were subjected to Western blot analysis using an antibody against Gab2.

Migration and Invasion Assays

Cell migration was assessed by adding 1 × 10⁵ cells into the upper chamber of a motility chamber system containing 8-μm pores (modified Boyden chamber assay; BD Biosciences, San Jose, CA). Invasion assays were performed by seeding 1 × 10⁵ cells into Biocoat Matrigel invasion chambers (BD Biosciences) in serum-free medium, and by adding 15% fetal bovine serum to the lower wells as chemoattractant. For both assays, the cells were incubated for 24 hours, the filters were stained with crystal violet, and the number of cells that penetrated through the filter was counted under a light microscope at ×20 magnification. For each membrane, the mean number of cells in three randomly selected fields was determined. For each assay, three independent experiments were performed in which the bars represent the mean ± SD.

In Vivo Tumorigenicity and Metastasis Assays

Tumor growth was evaluated by subcutaneous injection of 1 × 10⁶ vector or Gab2-expressing stable cell lines into the flanks of 7-week-old female severe combined immunodeficient (SCID) mice (n = 4 in each group). Tumor length and width were measured using calipers. Mice were sacrificed when tumor surface area reached 1.5 cm² and tumors were harvested and examined histologically. For experimental metastasis assays, vector or Gab2-expressing stable cell lines were injected into SCID mice via the tail veins (1 × 10⁶ cells per mouse, n = 12 in each group). Mice were sacrificed 3 weeks after inoculation, the lungs were harvested and examined histologically. Animal experiments were performed under a protocol approved by the Columbia University’s Institutional Committee for Care and Use of Animals.

Statistical Analysis

Statistical analyses were performed with a t-test in which P < 0.05 was considered as statistically significant. Differences between the groups in the experimental metastasis assays were analyzed using Fisher’s exact test.

Increased DNA Copy Numbers of Gab2 in Metastatic Melanoma

To identify genes critical for melanoma pathogenesis, we used BAC array CGH and SNP arrays to search for genome-wide copy number alterations in 64 metastatic melanoma samples and 20 melanoma cell lines. We found several distinct amplicons on chromosome 11q. Among them, increased DNA copy numbers of the cyclin D1 locus on 11q13.2, previously described for melanoma, was present in 5 of the 84 (6%) samples (1 of 64 tumors, 5 of 20 cell lines). More importantly, we noted a distinct amplicon, independent of cyclin D1, in 9 of the 84 (11%) samples (7 of 64 tumors, 2 of 20 cell lines), in which consistent focal amplifications with Gab2 gene at the center of the amplicon were observed (Figure 1, A, B, and D). These data were further confirmed by fluorescence in situ hybridization analysis (Figure 1C). Six of the nine cases were available for evaluation of Gab2 mRNA expression by qRT-PCR, which showed consistent up-regulation in all six cases compared with a human melanocyte cell line and nonamplified tumor samples (Figure 1E). Identification of increased DNA copy numbers of Gab2 among metastatic melanomas along with gene amplification in a subset of tumors resulted in further analysis of Gab2 in melanoma.

Gab2 Is Overexpressed in Metastatic Melanoma

We next examined the expression profile of Gab2 in a normal human melanocyte cell line and a panel of melanoma cell lines by Western blot analysis. Gab2 protein was expressed at low levels in the melanocyte cell line and in primary melanoma cells whereas high-level Gab2 expression was observed in metastatic cell lines (Figure 2). To confirm the above results and to investigate the expression profile of Gab2 during clinical melanoma progression, we extended our analysis by studying an independent series of melanocytic tumor samples, including melanocytic nevi, primary and metastatic melanomas. For these studies, we used the automated quantitative analysis (AQUA) system, which allows for rapid analysis of immunofluorescence on tissue microarrays, reduces the human variability of scoring by eye, and results in a continuous range of protein expression rather than ordinal (0, +1, +2, and + 3) scores. It also allows for quantification of the protein of interest within user-defined subcellular compartments within the tumor region of each tissue microarray spot. Specifically, we performed AQUA analysis of Gab2 protein expression on 18 melanocytic nevi, 15 primary and 15 metastatic melanoma samples each represented by two tissue microarray spots. Results showed that Gab2 protein is expressed at significantly higher levels in metastatic melanomas compared with melanocytic nevi (P = 0.0044) and primary melanomas (P = 0.0137), identifying Gab2 as a molecular marker of neoplastic progression (Figure 3).

Gab2 Promotes Migration and Invasion of Melanoma Cells

To investigate the role of Gab2 in tumor progression, we used siRNA technology to down-regulate Gab2 in three metastatic melanoma cell lines that overexpress Gab2; Mewo, Ht-144, and A2058. Gab2 silencing did not affect Gab1 expression (Figure 4A). As shown in Figure 4B, Gab2 knockdown decreased the migratory potential of tumor cells and led to a significant decrease in their invasive capacity (P < 0.05), suggesting that continuous Gab2 expression is required for in vitro invasion of metastatic melanoma cells. We next determined whether Gab2 overexpression could enhance migration and invasiveness of primary melanoma cells. For these experiments, we used WM793, WM278, WM1552, and WM3862 cells that are derived from primary melanomas and have low or undetectable levels of Gab2. All of these cell lines except WM3862 cells harbor the BRAF^V600E mutation. Forced expression of Gab2 in these cells led to a significant enhancement of migration and invasion (P < 0.05) (Figure 4, C–E). Taken together, these studies show that Gab2 is capable of enhancing motility and invasive capabilities of primary melanoma cells.

Gab2 Accelerates Tumorigenic Potential and Induces Metastasis in Vivo

To study the effects of Gab2 on tumor growth in vivo, we injected WM3862 primary melanoma cells with forced Gab2 expression or vector alone subcutaneously into SCID mice. The mice were sacrificed when tumor surface area reached 1.5 cm². All mice injected with Gab2-overexpressing cells reached the tumor burden threshold between days 16 to 21, whereas mice injected with vector alone reached this threshold between days 26 to 46. The survival curves in Figure 5A indicate that tumor growth was significantly increased in response to forced Gab2 expression suggesting a critical role for Gab2 in promoting tumor growth (Figure 5, A and B; P < 0.05). Intriguingly, WM3862 cells expressing vector alone led to pigmented tumors in SCID mice, whereas WM3862 cells overexpressing Gab2 were amelanotic (see Supplemental Figure 1 at http://ajp.amjpathol.org). We then tested whether Gab2 activation confers a metastatic behavior in Gab2-overexpressing WM3862 cells in an experimental metastasis assay. Vector or Gab2-overexpressing cell lines were injected into SCID mice via the tail veins and animals were assessed for lung metastases 3 weeks after inoculation. One mouse in the Gab2-overexpressing group died prematurely and could not be evaluated. We found tumor cells invading the lung parenchyma in 7 of 11 (63%) mice injected with Gab2-overexpressing WM3862 cells, whereas none of the mice injected with WM3862 cells expressing vector alone showed any evidence of lung metastases on multiple sections (Figure 5, C and D; P < 0.05). These results suggest that Gab2 increases the intrinsic ability of melanoma cells to metastasize in vivo.

PI3K-Akt Signaling Is Crucial for Gab2-Mediated Metastatic Phenotype

To gain mechanistic insight into signaling pathways that may contribute to Gab2-mediated invasive phenotype in melanoma cells, we examined AKT activation. Silencing of Gab2 in a Gab2-overexpressing metastatic melanoma cell line, Mewo, resulted in a marked decrease in Akt phosphorylation (Figure 6A). Furthermore, in a reciprocal experiment, forced expression of Gab2 into four primary melanoma cell lines with low endogenous Gab2 levels (WM793, WM278, WM1552, WM 3862; Figure 2) increased Akt activation in the absence of exogenous growth factors (Figure 6B), indicating that Gab2 activation leads to stimulation of Akt in melanoma cells. Because PDK1 is a downstream effector of PI3K that is recruited to the plasma membrane by phosphatidylinositol 3-phosphate (PIP₃) and phosphorylates the activation loop of Akt and is critical for regulating cell migration, we examined its activation in cells with ectopic Gab2 expression. All melanoma cells with Gab2 overexpression also showed increased PDK1 phosphorylation (Figure 6B), suggesting that Gab2 activation in melanoma cells may be resulting in Akt phosphorylation by signaling through PDK1.

To explore the possibility that Gab2 mediates the invasive phenotype via PI3K-Akt signaling, we examined the effect of the PI3K inhibitor, LY294002, on in vitro cell migration and invasion. Inhibition of PI3K-Akt signaling in Gab2-overexpressing melanoma cells decreased migration and invasive capability significantly (P < 0.05, Figure 6C). Collectively, these data suggest that Gab2-mediated signaling results in Akt stimulation and promotes an invasive and metastatic phenotype in melanoma cells via PI3K/PDK1/Akt signaling.