The MEL-XY3 cell line has already been described (24). The MEL-XY13 cell line was obtained from a lymph node amelanotic metastasis of an 82-year-old male patient. Both cell lines are HLA-A*0201-positive and have the BRAF V600E mutation, and c-kit (exons 11 and 17) and Nras (exons 2 and 3) sequencing revealed no additional mutations. Both cell lines were grown in melanoma medium (MM) (25) plus 10% fetal bovine serum (FBS) (Natocor, Carlos Paz, Córdoba, Argentina) at 37°C in air:CO₂ (95:5%) humid incubator.

MEL-XY3_SUR and MEL-XY13_SUR were generated by exposing cancer cells to 10 µM PLX4032, 1 µM GDC-0973 or combined treatment for 5 weeks. Media were changed twice a week. PLX4032 and GDC-0973 were provided by Genentech (South San Francisco, CA, USA).

DNA synthesis was assessed by measuring ³[H]-labeled thymidine incorporation. Ten thousand cells/well were seeded in 96-well plates in 200 µl of MM. When indicated, cells were incubated overnight and PLX4032 and/or GDC-0973 were subsequently added for different periods. After performing a 2-h pulse at 37°C with 1 µCi/ml ³[H]-labeled thymidine (Perkin-Elmer, Boston, MA, USA), the cells were harvested with a NuncCell Harvester 8 (Nalge Nunc International Corp., Rochester, NY, USA) and the incorporated radioactivity was determined with a liquid scintillation counter (Wallac 1214 RackBeta; Pharmacia, Turku, Finland).

Cells were seeded in 96-well flat-bottomed plates in triplicate. Twenty-four hours later, serial dilutions of PLX4032 and/or GDC-0973 were added. After incubation for 72 h, 100 µl of 1 mg/ml 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Sigma-Aldrich, St. Louis, MO, USA) diluted in MM were added to each well and incubation was carried out for 90 min. The supernatant was discarded and the crystal products were resuspended with isopropyl alcohol and incubated for 1 h at 30°C. Colorimetric evaluation was performed with a spectrophotometer at 570 nm. The inhibition of proliferation induced by the drugs was shown as the percentage of the growth of the untreated control cells. IC₅₀ was determined using GraphPad Prism 5.0 software.

Total RNA was purified using TRIzol reagent (Ambion, Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Quantification was performed with a NanoDrop 2000 (Thermo Fisher Scientific, Inc., Wilmington, DE, USA). One microgram of RNA was reverse-transcribed using SuperScript™ II Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA). The product was used in subsequent RT-qPCR using KAPA SYBR FAST Universal Master Mix (Applied Biosystems, Carlsbad, CA, USA) with the primers listed in Table I. Relative expression levels were determined by the ΔΔCq method (26), using β-actin gene expression to normalize all samples. Α melting curve analysis and gel electrophoresis assessment were used to confirm the specificity of PCR reactions.

Cells were detached, centrifuged, washed with cold phosphate-buffered saline (PBS) and resuspended in RIPA lysis buffer and protease inhibitor cocktail (Sigma-Aldrich). Phosphatase inhibitor cocktail 2 (Sigma-Aldrich) and NaF 10 mM were added for phospho-protein determinations. Cell extracts were run on a 10% SDS-PAGE gel and transferred to a nitrocellulose membrane (Immobilon, 0.45-µm pore size, HATF 304 FO; Millipore, Bedford, MA, USA). After blocking in 3% BSA with PBS, the blots were incubated with mouse anti-human CD271 mAb (clone C40-1457; BD Pharmingen, San Jose, CA, USA), rabbit anti-ERK1 polyclonal Ab (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), rabbit anti-p-ERK polyclonal Ab (Cell Signaling Technology, Inc., Beverly, MA, USA) or mouse anti-β-actin (clone AC-74; Sigma-Aldrich). After being washed, blots were further incubated with 1/2,500 alkaline phosphatase-AffiniPure F(ab')₂ fragment goat anti-mouse IgG (H+L) (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) and developed with nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT-BCIP) (Promega, Madison, WI, USA). Alternatively, blots were incubated with peroxidase anti-rabbit antibody (Vector Laboratories, Burlingame, CA, USA), developed with SuperSignal West Dura Extended Duration substrate (Thermo Fisher Scientific, Rockford, IL, USA), and chemoluminescence was measured in an ImageQuant LAS 4000 instrument (GE Healthcare, Uppsala, Sweden).

Cells were incubated with human-specific CD271 (clone C40-1457; BD Pharmingen), MART-1 (2A9) (27), gp100 (clone Hmb45; Dako, Glostrup, Denmark) and HLA-A2-PE (clone BB7.2; BD Pharmingen) antibodies for 1 h at 4°C. Before the use of MART-1 and gp100 antibodies a permeabilization step with 0.1% saponin was performed. Washed cells were resuspended in PBS and stored on ice or labeled with RPE conjugated anti-mouse polyclonal Ab (1:50; Dako). Ag expression was assessed using a FACSCalibur flow cytometer (BD Biosciences, San Diego, CA, USA). Analysis was performed with the FlowJo 7.6 software. Isotype-matched mouse antibodies were used as controls.

For the carboxyfluorescein succinimidyl ester (CFSE) assay, cells were detached, washed with PBS and labeled with 10 µM CFSE according to the manufacturer's indications (Invitrogen), and then plated for 96 h in the presence or absence of PLX4032. Cells were then detached and analyzed with a FACSCalibur flow cytometer.

For cell cycle analysis, cells were detached and fixed with 70% ethanol ON at 4°C. Cells were stained with propidium iodide solution (50 µg/ml with 100 µg/ml RNase A in PBS) for 30 min at 37°C. Cells were analyzed on a FACSCalibur flow cytometer using the FlowJo 7.6 software for data analysis.

Magnetic-activated cell sorting (MACS) to separate the CD271⁺ and CD271⁻ subpopulations was performed according to the manufacturer's instructions using anti-human-specific CD271 (clone C40-1457; BD Pharmingen), anti-mouse IgG microbeads and MS columns (Miltenyi Biotec, Bergisch Gladbach, Germany). Sorting verification was performed by flow cytometry.

Cells were either left untreated or were treated with 10 µM PLX4032 for 72 h or 5 weeks. Senescence-associated β-galactosidase (SA-β-Gal) staining was performed according to the chromogenic assay described by Debacq-Chainiaux et al (28).

Cell lysates were 2-fold serial diluted for five dilutions (from undiluted to 1:16 dilutions) and arrayed on nitrocellulose-coated slides. Samples were probed with antibodies by a tyramide-based signal amplification approach and visualized by DAB colorimetric reaction. Slides were scanned on a flatbed scanner to produce 16-bit tiff images. Spots from tiff images were identified and the density was quantified by an Array-Pro Analyzer. Relative protein levels for each sample were determined by interpolation of each dilution curve from the standard curve (supercurve) of the slide (antibody). All data points were normalized for protein loading and transformed into a linear value and normalized linear values were transformed to log₂ values.

NOD/LtSz-scid IL2Rγ^null mice (NSG) were subcutaneously injected with 500 MEL-XY3 parental or MEL-XY3_SUR-PLX cells in a suspension of Matrigel (BD Matrigel™ Basement Membrane Matrix, BD Biosciences) with PBS in a 1:1 ratio. Tumor growth was monitored three times a week for the entire period of the experiment. Tumor size was assessed using a caliper to calculate tumor volume (V) by applying the formula: V = [length (mm)] × [width (mm)]²/2. All animal procedures were approved by the Institutional Animal Care Board of the Leloir Institute (Buenos Aires, Argentina).

To determine lysis, effector CD8⁺ lymphocytes (E) specific for the HLA-A*0201 restricted MART-1 (M27:AAGIGILTV) or gp100 (G154:KTWGQYWQV) antigens (29) and target MEL-XY3 (control, MEL-XY3_SUR-PLX or MEL-XY3_SUR-GDC) cells (T) were incubated overnight at different E:T ratios (1:1, 5:1 and 10:1) in AIM V medium (Invitrogen, Grand Island, NY, USA). Cells were recovered and plated in quadruplicate for a clonogenic assay. For this, the cells were resuspended in MM + 10% FBS with 1.5% methylcellulose (Sigma-Aldrich), plated at 1,000 cells/well on MW24 plates over a 0.5% agar MM + 10% FBS underlayer and incubated for 14 days. Colonies ≥30 cells were counted using an inverted Olympus microscope.

IFN-γ secretion into the supernatant by cytotoxic T cells was determined with a BD OptEIA human IFN-γ ELISA kit (BD Biosciences), following the manufacturer's recommendations. A calibration curve was constructed for each experiment and the sample concentration was calculated by regression analysis. Controls for this experiment included CM cell lines HLA-A*0201-negative that do not express MART-1 or gp100 and the IIB-BRG breast cancer cell line HLA-A*0201-positive that does not express MART-1 or gp100 (negative controls) and CTLs alone.

Since clinical resistance to MAPKi almost always emerges after months of treatment, we sought to determine whether some cells remained viable after long-term treatment of BRAF-mutated cell lines with MAPKi and, if so, to characterize them. Accordingly, we first investigated the effect of PLX4032 and GDC-0973 on the MEL-XY3 CM cell line, heterozygous for the BRAF V600E mutation. MEL-XY3 cells were sensitive to PLX4032 (IC₅₀ 0.45 µM) (Fig. 1A) and to GDC-0973 (IC₅₀ 0.1 nM) (Fig. 1B). Inhibition of the MAPK pathway was also confirmed by inhibition of DNA synthesis (Fig. 1C) and of ERK phosphorylation (Fig. 1D). Though some DNA synthesis occurred within 0–2 h of inhibitor treatment, after 24 and 48 h it was totally abolished. ERK phosphorylation was inhibited 30 min after starting PLX4032 and GDC-0973 treatment. The effect of short-term treatment of MEL-XY3 with both PLX4032 and GDC-0973 was cytostatic, with cells starting to detach after a few days. We next exposed MEL-XY3 cells to 10 µM PLX4032 for 5 weeks; some cells remained attached and viable (Fig. 2A, left) and were detectable even after five months of treatment. We referred to this subpopulation of surviving cells as MEL-XY3_SUR-PLX. We analyzed the fate of MEL-XY3_SUR-PLX cells when the inhibitor was removed. After an initial period of several days, in which some cells presented abundant cytoplasmic vesicles and some detached from the plates (Fig. 2A, middle), MEL-XY3_SUR-PLX cells re-established vigorous growth (Fig. 2A, right), and generated cells with a similar phenotype to the parental cell line.

In order to determine whether such a quiescent phenotype could also be generated by other MAPKi, MEL-XY3 cells were maintained in the presence of 1 µM GDC-0973 for 5 weeks; a small, viable subpopulation with analogous characteristics to MEL-XY3_SUR-PLX cells was also observed, and was named MEL-XY3_SUR-GDC. After combined treatment with 10 µM PLX4032 and 1 µM GDC-0973 (MEL-XY3_SUR-PLX-GDC), surviving cells with a similar phenotype were observed. The status of the MAPK pathway in MEL-XY3_SUR-PLX and MEL-XY3_SUR-GDC cells while they were in the presence of the inhibitors was investigated. MEL-XY3_SUR-PLX cells presented lower levels of p-ERK than parental cells, although somewhat higher than MEL-XY3_SUR-GDC cells (Fig. 2B).

When we studied DNA synthesis at different time-points after drug removal, we observed that MEL-XY3_SUR-PLX cells started proliferation before MEL-XY3_SUR-GDC, while MEL-XY3_SUR-PLX-GDC cells were the last to resume growth (Fig. 2C). Notably, after a 5-week treatment with PLX4032 or GDC-0973, and allowing treated cells to regrow in the absence of the inhibitors, their sensitivity to PLX4032 or GDC-0973 remained unaltered with respect to the parental cells (Fig. 2D and E).

In order to evaluate whether the SUR phenotype also arose in other melanoma cell lines, we studied MAPKi treatment in the MEL-XY13 cell line. This cell line is heterozygous for the BRAF V600E mutation and sensitive to PLX4032 (0.36 µM) and GDC-0973 (1.25 nM). When MEL-XY13 cells were treated for 5 weeks with PLX4032 or GDC-0973, MEL-XY13_SUR-PLX and MEL-XY13_SUR-GDC, respectively, were obtained. Furthermore, when the drugs were removed, MEL-XY13_SUR cells restarted growth and their sensitivity to PLX4032 and GDC-0973 was equal to the parental cell line (Fig. 3A and B).

Since MEL-XY3_SUR cells were non-proliferating and different reports associate drug resistance with the CSC phenotype, we sought to determine whether they displayed some CSC characteristics. After treatment with 10 µM PLX4032 for 5 weeks, we analyzed the expression levels of putative CSC markers, such as CD271, ABCB5 and CD133 (prominin-1) (30–32). As compared to parental MEL-XY3, the mRNA levels for CD271 and ABCB5 in MEL-XY3_SUR-PLX cells increased 1,000- and 30-fold, respectively, whereas CD133 levels decreased (Fig. 4A). MEL-XY3_SUR-GDC cells obtained after 1 µM of GDC-0973 treatment (Fig. 4B) and MEL-XY3_SUR-PLX-GDC obtained after the combined treatment with both inhibitors (Fig. 4C) also had increased CD271 and ABCB5 mRNA levels, although the increase in CD271 was smaller. Similar results were obtained with MEL-XY13_SUR, with the exception that in MEL-XY13_SUR-GDC cells, CD271 mRNA levels did not change (Fig. 3C-E). Increased CD271 expression in MEL-XY3_SUR-PLX cells was confirmed at the protein level by flow cytometry (Fig. 4D) and by western blot analysis (Fig. 4E). In the flow cytometric analysis of MEL-XY3_SUR-PLX cells, two peaks were observed, demonstrating the presence of two cell subpopulations.

We also analyzed whether changes in CD271 ligand expression also took place in the MEL-XY3_SUR-PLX cells; all the mRNA levels of neurotrophins that bind CD271 (NGF, BDNF, NT-3, NT-4) increased 10- to −300-fold (Fig. 4F). Expression levels of the stem cell associated transcription factors Sox10, Sox2, Oct4 and Nanog were also determined, and a modest 2- to 4-fold increase was observed in MEL-XY3_SUR-PLX cells (Fig. 4G). We analyzed the expression time course of ABCB5, CD271 and CD133 in MEL-XY3 cells exposed to 10 µM PLX4032 at different time-points. ABCB5 expression rapidly increased and remained at high levels during the 5-week treatment. Instead, CD271 expression did not increase until the third week of treatment and CD133 started decreasing after 2 weeks of treatment (Fig. 4H). In order to determine whether CD271 and ABCB5 were uniformly expressed in the MEL-XY3_SUR-PLX cell population, CD271⁺ and CD271⁻ cells were separated by magnetic sorting. CD271⁻ predominated over CD271⁺ cells but both expressed ABCB5 (Fig. 4I), suggesting that the expression of both markers was not related. CD133 mRNA was not detected in CD271⁺ MEL-XY3_SUR-PLX.

It has been previously shown that PLX4032 and another mutant BRAF inhibitor, LGX818 (encorafenib), induce senescent characteristics in melanoma cells (33,34). It has also been shown that the expression of mutated BRAF in normal melanocytes determines oncogene-induced senescence (35). Therefore, we decided to investigate senescent characteristics in our SUR population. The cell cycle of MEL-XY3_SUR-PLX and MEL-XY3_SUR-GDC was analyzed by flow cytometry, and a decrease in the S phase and an increase in the proportion of cells in the G0/G1 phase were observed, consistent with the observations of Fig. 2 (Fig. 5A).

We then analyzed senescence-associated-β-galactosidase (SA-β-gal) activity in MEL-XY3_SUR-PLX. Although MEL-XY3 cells treated with 10 µM PLX4032 for 72 h did not stain positive for SA-β-gal, most MEL-XY3_SUR-PLX cells did (Fig. 5B), thereby suggesting senescence. To further investigate other senescence-related proteins, we performed a RPPA (Fig. 5C). In relation to the parental cell line, senescence-related proteins superoxide dismutase 2 (SOD2), insulin-like growth factor binding protein-5 (IGFBP5) and p16^INK4 were increased. On the other hand, E2F1, p53, plasminogen activator inhibitor-1 (PAI1) and p27^kip1 displayed only small changes. In MEL-XY3_SUR-PLX cells, there was also a general decrease in cell cycle-related proteins, with striking decreases in p-Rb, CDK1 and cyclin B1.

We next investigated whether MEL-XY3_SUR-PLX and MEL-XY3_SUR-GDC quiescent cells coexisted in the cell line with proliferating cells before MAPKi treatment. If that were the case, it could be assumed that after drug removal and growth resumption, at least a portion of SUR cells would remain quiescent, thereby confirming that they are a ‘reservoir’ for proliferating cells. On the other hand, if quiescence is a transitory, variable state induced by chemotherapy, every surviving cell would leave behind its quiescence and start regrowth when the drugs were removed. We stained MEL-XY3_SUR-PLX cells with CFSE, and they remained in culture for 96 h in the absence or presence of the inhibitor; retention of the label was then analyzed. During those 96 h, MEL-XY3_SUR-PLX cells underwent two rounds of cell division without leaving behind a resting subpopulation. When a similar experiment was performed with the parental cell line, three rounds of cell division were achieved (Fig. 6A). This experiment suggests that tumor cell plasticity and phenotypic switching would account for the previously described acquisition of quiescence. When the same experiment was performed with the MEL-XY3_SUR-GDC cells, no cell division was observed after 96 h, which is consistent with the lagging DNA synthesis after drug removal (Fig. 2C).

To determine whether MEL-XY3_SUR-PLX cells retained their tumorigenic potential, we injected 500 cells into NSG mice and compared their growth rate with parental MEL-XY3 cells. Even though both cell types developed tumors, MEL-XY3_SUR-PLX grew faster than parental cells, although the difference was not statistically significant (Fig. 6B).

It has been reported that MDA expression increases after PLX4032 treatment, both in vitro and in patients (21,22). We confirmed increased mRNA levels of MDAs MART-1, gp100, TYR and Trp-2 when MEL-XY3 cells were treated with 10 µM PLX4032 for 72 h (Fig. 7Α). We found that MEL-XY3_SUR-PLX cells presented MDA mRNA levels similar to or even higher than those of parental cells (Fig. 7Β). Most importantly, MART-1 and gp100 protein levels were similar in MEL-XY3_SUR-PLX and MEL-XY3_SUR-GDC compared to those of parental MEL-XY3 cells (Fig. 7Β), thereby revealing persistent MDA synthesis. MITF mRNA expression was also analyzed in MEL-XY3_SUR-PLX cells but no significant changes in expression were found (Fig. 7Β).

It was therefore important to determine whether, although resistant to chemotherapy, MEL-XY3_SUR cells could be killed by immunological effectors. First, we established that MEL-XY3_SUR cells expressed, although at lower levels than the parental line, the HLA-A*0201 haplotype, in which MART-1 and gp100 peptides are presented (Fig. 7C). We then studied the susceptibility of MEL-XY3_SUR cells to two CTL clones, HLA-A*0201-restricted and specific for gp100 (G154) and MART-1 (M26) antigens (29). We had previously shown that MEL-XY3 cells are lysed by specific CTL (36). We now demonstrated that MEL-XY3_SUR-PLX cells were killed by specific CTL clones, as determined by clonogenic assays. At a 1:1 effector:target ratio, a 50% lytic effect was observed, as evidenced by the reduction of colony formation; at higher ratios almost no cells survived. MEL-XY3_SUR-GDC cells were also sensitive to CTL killing, although it is worth noting that MEL-XY3_SUR-GDC cells were less clonogenic than MEL-XY3_SUR-PLX cells (Fig. 8A). As a surrogate measure confirming that MEL-XY3_SUR cells were sensitive to CTL action, IFN-γ release after co-incubation of MEL-XY3_SUR cells with CTL clones was assessed. A strong cytokine release was triggered by MEL-XY3_SUR-PLX, indicating a potent interaction between CTL and tumor cells, even stronger than with parental cells. However, almost no CTL activation occurred after incubation with MEL-XY3_SUR-GDC cells, even though a lytic effect was observed (Fig. 8B). Similar results were obtained with the CTL clone specific for MART-1 (M26) (Fig. 8C and D). Equivalent results were obtained with the CTL clone specific for gp100 in MEL-XY13 and MEL-XY13_SUR-PLX cells (Fig. 8E and F).