A375, 624.38 and 397 melanoma cell lines, kindly provided by Dr F. M. Marincola and Dr M. Bettinotti (NIH, Bethesda, MD, USA), were used. All cell lines were cultured in RPMI-1640 medium, supplemented with 3 mM L-glutamine (both from Invitrogen-Gibco, Paisley, UK), 2% penicillin/streptomycin and 10% fetal bovine serum (FBS). The cell cultures were incubated at 37°C in a humidified 5% CO₂ atmosphere. Dabrafenib and trametinib were purchased from Selleck (Munich, Germany).

A375 (50 cells/cm²), 624.38 (50 cells/cm²) and 397 (250 cells/cm²) melanoma cells were plated in 24-multi-wells, previously coated with 1% gelatin, with complete medium containing dabrafenib (30 nM) or not containing the drug. The medium was replenished every 2 days. After 7 days, the cells were fixed in 4% paraformaldehyde (PFA) and stained with 0.15% crystal violet. Plates were imaged by scanner and colonies were imaged on a Leica DMI6000 inverted microscope (Leica, Mannheim, Germany). The observed number of colonies was determined from 5 independent areas using ImageJ software (http://rsbweb.nih.gov/ij/).

Dabrafenib-resistant melanoma cells were selected by growing all 3 melanoma cell lines in medium containing increased concentrations of dabrafenib for 4 weeks. A375 and 397 melanoma cell lines resistant to 30 nM dabrafenib were selected, while for the 624.38 cell line melanoma cells resistant to 100 nM dabrafenib were selected.

Cells (1,500/well) were seeded in triplicate in 96-well plates. After 24 h, the cells were treated with different drug concentrations. The monitoring of the number of cells following 6 days of drug treatment was performed using CyQUANT cell proliferation assay kit, according to the manufacturer's procedure (Invitrogen, Paisley, UK). Dose-response data were analyzed by GraphPad Prism version 5.00 for Windows (GraphPad Software) to determine the IC₅₀ values.

Cells were harvested in phosphate-buffered saline (PBS) containing 2 mM EDTA, washed once with PBS and fixed in ethanol at 96°C. After washing in PBS, 1×10⁶ cells were incubated with 5 µg/ml propidium iodide (PI) (Sigma Chemical Co., St. Louis, MO, USA) plus 25 µl RNase (1 mg/ml), overnight at 4°C in the dark. Stained nuclei were analyzed using FACSAria II (Becton-Dickinson, Franklin Lakes, NJ, USA). Data were analyzed using a ModFit III (Verity, Topsham, ME, USA) cell cycle analysis programme.

Genomic DNA was isolated from melanoma cell lines by standard methods and quantified using the Qubit^® fluorometer (Life Technologies, Gent, Belgium). For library construction, DNA (10 ng) was amplified using the Ion Torrent AmpliSeq Hotspot V2/CHPv2 Cancer Panel (Life Technologies-ThermoFisher Scientific, Waltham, MA, USA). An amplicon library was generated for sequencing ~2,800 mutations in the 50 most common oncogenes and tumor-suppressor genes. The amplicons were digested and amplified using the Ion AmpliSeq™ Library kit 2.0 (Life Technologies), according to the manufacturer's instructions. Finally, the template was loaded on an Ion 316™ chip and sequenced on a PGM™ sequencer with the Ion PGM™ sequencing 200 kit v2 (Life Technologies), according to the manufacturer's protocols. Only mutations reported in the Human Gene Mutation Database (HGMD) at: http://www.hgmd.cf.ac.uk/ac/index.php and in the Catalogue Of Somatic Mutations In Cancer (COSMIC) at: http://www.sanger.ac.uk/genetics/CGP/cosmic/ were taken into account; silent or intronic mutations were not considered. All significant gene mutations identified on the Ion AmpliSeq Cancer Hotspot Panel were confirmed by Sanger sequencing of gene-specific amplicons (primer pairs provided on request). Briefly, polymerase chain reaction (PCR) was performed on 20 ng of genomic DNA in a Veriti^® 96-Well Fast Thermal Cycler (Life Technologies-ThermoFisher Scientific). All PCR-amplified products were directly sequenced using an automated fluorescence-cycle sequencer (ABI3130; Life Technologies-ThermoFisher Scientific). Sequencing analysis was conducted in duplicate and in both directions (forward and reverse) for all evaluated samples.

Cells were grown at 80% confluence, fixed in 4% PFA and stained with 0,15% crystal violet. Cell images were captured using an inverted microscope (Leica, DMI6000). Cells were stained following incubation for 30 min at +4°C with the following antibodies (5 µl/ml): mouse anti-human CD90 PE-Cy5, mesenchymal marker (BD Biosciences, Franklin Lakes, NJ, USA), human E-cadherin (CD324) APC, human CD133/2 (293C3)-PE (both from Miltenyi Biotec, GmbH, Bergisch Gladbach, Germany), mouse anti-human CD20 FITC and mouse anti-human CD44 FITC (both from BD Biosciences). After being washed with PBS, the samples were analyzed by FACS. The fold-change of the mean fluorescence intensity (MFI) values was used to show the differences in surface marker expression between drug-sensitive and -resistant melanoma cells. It was calculated from 3 independent experiments.

In order to induce EMT, all 3 melanoma cell lines (A375, 624.38 and 397) were treated with 2 ng/ml TGF-β1 (Abcam, Milan, Italy) for 48 h and used for further investigations.

Total RNA (300 ng) from the melanoma cell lines was converted to cDNA using M-MLV reverse transcriptase (Invitrogen, Life Technologies, Monza, MB, Italy). Primers used for E-cadherin were: forward, 5′-ACCACCTCCACAGCCACCGT-3′ and reverse, 5′-GTCCAGTTGGCACTCGCCCC-3′; β-actin was used as an internal control (forward 5′-TTCTACAATGAGCTGCGTGTG-3′ and reverse 5′-GGGGTGTTGAAGGTCTCAAA-3′). Experiments were performed in triplicate. qRT-PCR was carried out on an ABI PRISM 7900HT sequence detection system), using PowerUp SYBR-Green Master Mix (both from Applied Biosystems, Life Technologies, Grand Island, CA, USA).

Cells were seeded to confluency in 6-well plates, and a wound was made in the center of the cell monolayer using a sterile 200-µl pipette tip. The monolayer was rinsed 3 times with PBS and placed in RPMI-1640 medium. Phase contrast images were captured after 24 h and a digital image of the scar was captured at a magnification of ×10 (Leica, DMI6000). All experiments were performed in triplicates.

Cells were grown in RPMI on glass coverslips (22×22 mm) and after 24 h, fixed in 4% PFA and permeabilized in PBS containing 0.1% of Triton X-100. After blocking with BSA 5% in PBS for 20 min at RT, cells were incubated with the following primary antibodies: vimentin (ab8069; 1:150; Abcam) and Oct4 (Oct4 C-10 antibody; 1:500; Santa Cruz Biotechnology, Inc., TX, USA) at 4°C overnight. After extensive washing, fluorescent secondary antibodies goat anti-rabbit (ab96885; Abcam), goat anti-mouse FITC (ImmunoReagents, Inc., Raleigh, NC, USA), diluted 1/500 in PBS were added for 90 min at +4°C. The cells were then counterstained with 4,6-diamidino-2-phenylindole (DAPI) (1:100 in PBS) for 7 min at +4°C, to stain the nuclei. The coverslips were washed again and mounted on slides using fluorescence mounting medium (Dako Diagnostics, Mississauga, ON, USA). Fluorescent images were captured using a Nikon upright microscope. The same immunofluorescence protocol was used for 3D human tissues (14), which were successively stained with hematoxylin and eosin (H&E). Briefly, 3D human tissues from a differentiated and full-thickness skin reconstruction model of A375 melanoma cells were purchased from MatTek (Ashland, MA, USA). The 3D tissues were fed through the basolateral (bottom) surface, and incubated in duplicate with serum-free medium containing 0.2% dimethyl sulfoxide (DMSO) as a control or dabrafenib plus trametinib. The medium was replenished every other day, and the 3D tissues were collected on days 0, 9 and 12 from the beginning of treatment and fixed with 4% PFA. Tissues were paraffin-embedded, serial-sectioned, deparaffinized in xylene and rehydrated through graded decreasing concentrations of alcohol and used for immunofluorescence studies and H&E staining.

Cells were lysed in lysis buffer [1% Nonidet P-40, 150 mmol/l NaCl, 10 mmol/l Tris (pH 7.4), 1 mmol/l EDTA, 1 mmol/l EGTA (pH 8.0), 0.2 mmol/l sodium orthovanadate, 1X inhibitor cocktail) (Sigma-Aldrich, St. Louis, MO, USA) for 30 min at 4°C with constant agitation. Insoluble material was removed by centrifugation (16,000 × g at 4°C) for 15 min and the total protein concentration was determined in the supernatant by Bradford assay.

Western blotting was performed according to standard procedures. Rabbit monoclonal antibodies against Snail (diluted 1:1,000), rabbit polyclonal against Twist1 (diluted 1:1,000), and β-actin (diluted 1:1,000) were used. All the antibodies were purchased from Cell Signaling Technology (Danvers, MA, USA). Detection was achieved by HRP-conjugated anti-rabbit (1:10,000; Cell Signaling Technology) antibodies. Immune complexes were visualized by an enhanced chemiluminescence system (ECL Advance™; Amersham Pharmacia Biotech, Piscataway, NJ, USA). β-actin was used as a loading control. The image analysis was performed using ImageJ software (http://rsbweb.nih.gov/ij/).

Values are displayed as the mean ± SEM of measurements of at least 3 independently performed experiments to avoid possible variation of cell cultures. Student's t-test was employed, and P<0.05 was considered to be statistically significant.

In order to characterize the phenotype of melanoma cells resistant to B-RAF^V600E inhibition, we generated melanoma cells resistant to dabrafenib (a B-RAFi) using 3 melanoma cell lines, A375, 397 and 624.38, all carrying a B-RAF^V600E mutation and sensitive to dabrafenib.

As shown in Fig. 1A, cells plated at a low density in the presence of 30 nM dabrafenib were able to form only a few sizeable colonies, after 7 days, comparable to the ones observed in the untreated cells, as expected.

We selected and expanded the few cells able to proliferate and form colonies under drug treatment, by growing them in medium containing increasing concentrations of dabrafenib, as described in Materials and methods. The resulting dabrafenib-resistant melanoma cells are indicated as A375_B-RAFi (R), 397_B-RAFi (R) and 624.38_B-RAFi (R). Since the resistance to B-RAF^V600E inhibition in melanoma is reversible and adaptive (15), the resistant cells were characterized, after their expansion, under drug selective pressure.

In order to confirm that A375_B-RAFi (R), 397_B-RAFi (R) and 624.38_B-RAFi (R) cells were resistant to dabrafenib treatment, the IC₅₀ value was assessed, using the survival colorimetric assay. The A375 and the derived dabrafenib-resistant melanoma cells, A375_B-RAFi (R), were treated at increasing concentrations of B-RAFi (0, 0.1, 1, 5, 25 and 100 nM) for 6 days. Dose-response analysis revealed that dabrafenib-resistant cells showed an IC₅₀ value of 110.5 µM compared to the 9.5 nM value determined for the dabrafenib-sensitive A375 counterparts (Fig. 1B). Similarly, a dabrafenib IC₅₀ value of 280 µM was determined for the 397_B-RAFi (R) cells, while a 0.006 nM value was assessed for the 397 drug-sensitive cells (Fig. 1B). Finally, 624.38 and 624.38_B-RAFi (R) cells were treated at increasing concentrations of B-RAFi (0, 5, 25, 50, 100 and 500 nM) for 6 days. A dabrafenib IC₅₀ value of 0.1 µM was determined for the 624.38_B-RAFi (R) cells, while a 12 nM value was assessed for the 624.38 drug-sensitive cells (Fig. 1B).

The effect on dabrafenib on the cell cycle profile was also examined by FACS analysis. Cell cycle profiles with PI revealed a marked difference between sensitive and resistant melanoma cells to dabrafenib. A significant accumulation of cells in the G0/G1 phase (91.9 vs. 45.1%, dabrafenib-treated vs. untreated, respectively) with a concomitant decrease in the number of cells in the G2/M phase (0.69 vs. 11.8%, dabrafenib-treated vs. untreated, respectively) was detected in the A375 cells when treated with the B-RAF inhibitor. Conversely, no significant G0/G1 arrest in response to dabrafenib in the A375_B-RAFi (R) cells was observed (Fig. 2A). Similarly, an accumulation of cells in the G0/G1 phase with a decreased number of cells in the G2/M phase was observed for both 397 and 624.38 melanoma cells, when treated with the B-RAFi, while no significant cell cycle profile difference was detected for the resistant 397_ B-RAFi (R) and 624.38_B-RAFi (R) melanoma cell lines, upon B-RAF inhibition (data not shown).

Finally, in order to ascertain whether the resistance to dabrafenib in these melanoma cells was associated with the reactivation of the MAPK signaling pathway (16), we assessed whether the treatment of these cells with MEK inhibitors compromised the viability of dabrafenib-resistant cells. As shown in Fig. 2B, these cells were sensitive to trametinib, a MEK inhibitor, supporting the association of resistance to dabrafenib with the reactivation of MAPK signaling.

To assess the occurrence of discrepancies in mutation patterns during the development of the resistance status, dabrafenib-sensitive and -resistant melanoma cell lines were analyzed using the high throughput Ion Torrent sequencing technology, as described in Materials and methods. For the mutational screening the Ion AmpliSeq Cancer Hotspot Panel v2, containing >200 primer pairs producing short (average of ~150 bp) amplicons which target hotspot regions of 50 main oncogenes and tumor-suppressor genes involved in tumorigenesis (Table I) was used.

All variants detected by NGS were confirmed through PCR-based Sanger sequencing. As shown in Table II, no difference in mutation patterns was observed between the sensitive and resistant cell lines, strongly suggesting that development of resistance was independent of the acquisition of genetic alterations in the 50 most common oncogenes and tumor-suppressor genes examined (Table I). Moreover, mutations in the CDKN2A and TP53 genes represented the main sequence variants associated with the BRAF^V600E mutation in our cell line series (Table II), further confirming the role of such crucial genes in controlling cell proliferation and survival in melanoma genesis.

In order to characterize the dabrafenib-resistant melanoma cells, their morphology and phenotype were examined. All cells were grown at 70–80% confluence, stained with crystal violet and imaged using a light microscope, as described in Materials and methods.

All dabrafenib-resistant cells showed a morphological change compared to the A375, 397 and 624.38 cells, respectively (Fig. 3A). A bigger difference in the inter-cellular space was found between 624.38_B-RAFi (R) and 624.38 cells, while it was not highly evident between the other examined cell lines. These data revealed that melanoma resistance to dabrafenib was also associated with cell morphological changes.

Since it has been reported that EMT-like cancer cells exhibit the ability to survive chemotherapy and radiation therapy (17,18), we examined the expression of mesenchymal CD90 (19) and E-cadherin epithelial markers in the dabrafenib-resistant melanoma cells, and compared the obtained values with the ones determined for the dabrafenib-sensitive cells, by flow cytometry, as described in Materials and methods.

The expression level of E-cadherin was decreased in the A375_B-RAFi (R) and 397_B-RAFi (R) dabrafenib-resistant melanoma cells compared to the A375 and 397 dabrafenib-sensitive counterparts by ~2.1- and 2.2-fold, respectively, while it was slightly decreased in the 624.38_B-RAFi (R) cells compared to the 624.38 cells by ~1.6-fold. On the contrary, the CD90 marker expression level was increased by ~2.7- 1.4- and 2.3-fold in the A375_B-RAFi (R) vs. A375, 397_B-RAFi (R) vs. 397, and 624.38_B-RAFi (R) vs. 624.38 cells, respectively (Fig. 3B).

We conclude that melanoma cells, under dabrafenib exposure, exhibited morphological changes with a more emphasized mesenchymal phenotype.

Although melanoma cells originate from melanocytes, not epithelial cells but neuroendocrine cells originating from the neural crest, their ability to switch phenotype from proliferative to more invasive states, in a process similar to classical epithelial-mesenchymal transition (EMT), has been reported (20–22). In order to investigate whether the EMT was associated with the resistance to dabrafenib, EMT was induced in dabrafenib-sensitive melanoma cells by stimulation with TGF-β1 (23), as described in Materials and methods, and then the cells were treated with dabrafenib (5 nM).

As shown in Fig. 4A after treatment with TGF-β1 for 48 h the cells underwent EMT and they were more sensitive to dabrafenib compared to the non-TGF-β1-induced melanoma cells, suggesting that the observed drug resistance was not linked to the ‘classical’ TGF-β1-induced EMT.

In order to shed more light on the phenotypic features of dabrafenib-resistant melanoma cells, we examined the network reorganization of vimentin filaments by immunofluorescence. We found that vimentin was mainly detected in the cytoplasm of the dabrafenib-sensitive melanoma cells, while it was observed in the peri-nuclear region in the dabrafenib-resistant cells (Fig. 4B), as expected in EMT.

Since downregulation of E-cadherin is regarded as a bona fide classical EMT inducer, we examined the E-cadherin transcriptional level in the dabrafenib-sensitive and -resistant melanoma cells, by RT-PCR, as described in Materials and methods. We observed that the E-cadherin level was unvaried between the dabrafenib-resistant and -sensitive cells, suggesting that EMT was not associated with the resistance to dabrafenib. However, a change in the E-cadherin protein distribution was observed between dabrafenib-resistant and -sensitive melanoma cells. In particular, E-cadherin was translocated from the membrane into the cytoplasm and nuclei of all drug-resistant cells (Fig. 4C), as expected in EMT.

Furthermore, all 3 melanoma cell lines resistant to dabrafenib exhibited unvaried expression levels of Snail, while Twist expression decreased in the 397_B-RAFi (R) vs. 397, and 624.38_B-RAFi (R) vs. 624.38 cells and was unvaried in the A375_B-RAFi (R) vs. A375 cells (Fig. 4D).

Collectively, these data revealed that melanoma cells, under drug exposure, underwent an epithelial-to-mesenchymal-like phenotypic transition, which was not a result of the activation of the TGF-β1-induced EMT pathway.

Since cells undergoing EMT gain motility (24), we also investigated the cell motility of both dabrafenib-sensitive and -resistant melanoma cells, by wound healing assay, as described in Materials and methods.

As shown in Fig. 5, wound size analyses performed at 24 h revealed enhanced motility in the dabrafenib-resistant melanoma cells compared to their sensitive counterparts, further supporting an epithelial-to-mesenchymal-like phenotypic transition of melanoma cells under drug exposure.

It has been reported that EMT can produce a breast cancer stem cell (BCSC) phenotype in breast cancer, which explains the disease progression and relapse (25). In the present study, we investigated the expression of various cancer stem cell-like markers, such as CD20, CD133 and CD44 by FACS analysis and Oct4 expression in all melanoma cell lines by immunofluorescence.

We found ~2-fold increase in the expression level of CD20 in the A375_B-RAFi (R) and 624.38_B-RAFi (R) vs. the A375 and 624.38 cells, respectively, while it was more pronounced in the 397_B-RAFi (R) vs. the 397 cells with an increase in ~3.8-fold (Fig. 6A). CD133 marker expression was increased in the A375_B-RAFi (R) and 397_B-RAFi (R) vs. the A375 and 397 cells, respectively, while it was unmodified in the 624.38_B-RAFi (R) vs. 624.38 cells. Finally, the expression level of CD44 was upregulated only in the 397_B-RAFi (R) vs. 397 cells, while it was unaltered in all the remaining cells (Fig. 6A).

Furthermore, we examined Oct4 expression by immunofluorescence, as described in Materials and methods. Oct4 is one among the regulatory core transcription factors involved in the maintenance of stemness in mesenchymal stem cells (26) as well as a factor induced by EMT (27). Oct4 protein was mainly detected at the cytoplasmic level in the melanoma cells sensitive to B-RAF inhibition, while it was located at the peri-nuclear site and in the nucleus of all 3 dabrafenib-resistant melanoma cell lines (Fig. 6B), suggesting Oct4 activation in all the resistant cancer cells (28,29).

In a 3-dimensional (3D) skin reconstruction model of A375 melanoma cells, we previously observed individual proliferating melanoma cells infiltrating the dermis under BRAFi/MEKi drug treatment (14), where MEKi used was trametinib drug. Vimentin and Oct4 expression was also analyzed in these 3D tissues by immunofluorescence. We observed that the melanoma cells infiltrating the dermis expressed vimentin and Oct4 in the peri-nuclear sites. Oct4 was also detected in the nuclei. These findings support the mesenchymal and stem-like phenotype of these cells (Fig. 6C).