Identification of MET and SRC Activation in Melanoma Cell Lines Showing Primary Resistance to PLX4032^1,2

Cells and Cellular Assays

The short-term melanoma cell lines LM4–LM41 have previously been described [15]; LM42 and LM43 were derived from visceral metastases and were similarly generated and characterized. The cell line LM17R was generated by treating the parental cell line LM17 with PLX4032 (3.2 µM) for 96 hours, allowing the few surviving cells to regrow, and repeating treatment for 11 times. MTT assays were used to evaluate the inhibition of cell growth at 72 hours, adding drugs 24 hours after cell plating. The bioluminescent ToxiLight bioassay kit (Lonza, Valais, Switzerland) was used to measure the release of adenylate kinase (AK) from dying cells. Caspase 3 activation was measured using the Active Caspase 3 Apoptosis Kit (Becton Dickinson, Franklin Lakers, NJ). The analysis of the cell cycle was performed by determining the DNA content distribution after propidium iodide staining using a FACSCalibur and ModFit LT v3.1 software. Silencing of v-raf-1 murine leukemia viral oncogene homolog 1 (CRAF) and met proto-oncogene (MET) was obtained using SMART pool small interfering RNA (siRNA; L-003601 and L-003156; Dharmacon, Lafayette, CO) and Lipofectamine 2000 (Gibco, Grand Island, NY). A scrambled control was used (D-001810-10). Invasion assays were performed as previously described [16] on cells exposed for 24 hours to the inhibitors. Scratch wound assays were set on confluent cell monolayer in six-well plates. The monolayer was scratched using a sterile pipette tip, rinsed to remove detached cells, and treated with inhibitors for 72 hours. Matrix metalloproteinase 2 and 9 (MMP-2/-9) activity was assessed using 10% SDS-PAGE gelatin substrate zymography (Invitrogen, Carlsbad, CA) in serum-free conditioned medium after concentration with Amicon Ultra 10K (Millipore, Billerica, MA). Anti-human β₁-integrin antibody (552828; Becton Dickinson) was used with APC-conjugated anti-rat immunoglobulin G ( Jackson ImmunoResearch, Plymouth, PA) and analyzing staining by FACS analysis. Fluorescent in situ hybridization (FISH) analysis was performed using the probe kit D7S522/CEP7 according to the manufacturer's protocol (Abbott Vysis, Abbott Park, IL).

Genetic Analysis

Copy numbers of BRAF, microphthalmia-associated transcription factor (MITF), MET, cyclin D1 (CCND1), and β-catenin (CTNNB1) genes in melanoma samples were determined by quantitative real-time polymerase chain reaction (PCR) analysis using TaqMan Copy Number Assays from Applied Biosystems (Branchburg, NJ). In particular, the copy number of BRAF gene was evaluated by targeting intron 13 (Hs04958893_cn) and intron 16 (Hs05004157_cn), whereas a single assay was used for MITF (Hs02258756_cn), MET (Hs00305306_cn), CCND1 (Hs01425024_cn), and CTNNB1 (Hs02393264_cn). TaqMan copy number reference assay RNase P was used as endogenous reference gene. DNA isolated from blood samples of healthy donors was used as control. PCRs were performed in quadruplicate and run on the ABI Prism 7900HT machine. Results were analyzed using the Copy Caller software version 1.1 and copy numbers 4 or higher were considered gene amplifications. The methylation status of the PTEN promoter was determined after bisulfite conversion using the EZ DNA Methylation-Gold Kit (Zymo Research, Irvine, CA) by performing PCR analysis using previously reported primers and protocols with minor modifications [17]. Multiplex ligation-dependent probe amplification (MLPA) SALSA kits P005, P006, and P007 were used to profile changes in chromosomal regions as detailed by the manufacturer (MRC-Holland, Amsterdam, the Netherlands). Results were analyzed by Coffalyser v 9.4 software by normalizing to three samples of normal DNA. The resulting values were categorized as homozygous loss (≤0.3), loss of heterozygosity (≤0.6), gain (≥1.3), and amplification (≥2).

Materials

The following antibodies were used: anti-pERK1/2 (M8159), anti-ERK (M5670), and anti-vinculin (V9131) from Sigma (St. Louis, MO); anti-AKT (610861) from Becton Dickinson; anti-pAKT (4051), anti-pSRC (2105), anti-pMET (3077), anti-phosphorylated signal transducer and activator of transcription 3 (STAT3; 9131), anti-pPaxillin (2541), and anti-pp130CAS (4011S) from Cell Signaling Technology (Danvers, MA); anti-Src (05-184), anti-p70 S6 kinase (p70^S6K; 05-781), anti-pp70 S6 kinase (04-392), and anti-Src homology 2 domain-containing transforming protein (SHC; 06-203) from Upstate Biotechnology (Lake Placid, NY); anti-CCND1 (M7155) from Dako (Glostrup, Denmark); anti-MET (sc-10), anti-STAT3 (sc-483), anti-CRAF (sc-133), anti-phosphorylated focal adhesion kinase (pFAK; sc-101679), anti-FAK (sc-932), anti-pSHC (sc-18074-R), and anti-actin (sc-166) from Santa Cruz Biotechnology (Santa Cruz, CA); anti-paxillin (P13520) from Transduction Laboratories (Lexington, KY); anti-p130CAS (ab33539) from Abcam (Cambridge, UK); anti-breast cancer resistance protein (BCRP; MON9041) and anti-multidrug resistance protein 4 (MRP4; MON9069) from Monosan (Valter Occhiena, Torino, Italy); anti-KIT (v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog; 566) from MBL (Woburn, MA); and peroxidase-conjugated secondary antibodies anti-mouse immunoglobulin (Becton Dickinson) and anti-rabbit immunoglobulin G (Amersham, Piscataway, NJ) were used. For anti-phosphorylated tyrosine (pTyr) immunoprecipitation and MALDI-TOF mass spectrometry analysis, samples were processed as previously described [18,19]. Only proteins identified in at least three separate experiments were considered.

PLX4032 was obtained by agreement with Plexxikon, Inc (Berkeley, CA). SU11274/Sugen, UO126, PHA-665752 (Sigma), BMS-354825/Dasatinib (LC Laboratories, Woburn, MA), JNJ-38877605, SGX-523 (Selleck, Houston, TX), and E804/Indirubin (Calbiochem, Gibbstown, NJ) were purchased. After dose-response tests, the drugs were used at the concentrations indicated.

Data Analysis

Fitted lines were generated using the four-parameter nonlinear regression with a sigmoidal dose response (variable slope), and the IC₅₀ values (growth-adjusted inhibitory concentration of 50%) for inhibition of cell growth at 72 hours of PLX4032 treatment were calculated using Prism v 5.0 software. Student's t test and one-way analysis of variance (ANOVA) followed by the Bonferroni correction were used to evaluate statistical significance. Drug interaction was evaluated as described elsewhere [20] with interaction index values greater than 1 indicating synergism. The reported data are representative of three independent experiments.

PLX4032 Growth Inhibitory Effects in BRAF^V600E-Mutated Melanoma Cells Are Not Associated with Other Common Melanoma Gene Alterations Including PTEN Loss

The growth inhibitory effect of PLX4032 was tested in a panel of 27 genetically characterized melanoma cell lines, including 20 lines that were heterozygous for the V600E BRAF mutation and 7 lines carrying wild-type BRAF gene. The effect of other genetic alterations, including mutations in CDKN2A, PTEN, and tumor protein p53 (TP53) and amplification of BRAF and MITF, on melanoma cell sensitivity to PLX4032 was considered. We found that PLX4032 inhibition of cell growth was strictly dependent on the presence of BRAF^V600E and independent of other gene alterations. In fact, 18 of 20 BRAF^V600E-mutated melanoma cell lines were sensitive to the compound, with IC₅₀ values ranging between 0.01 and 1 µM, whereas 2 cell lines displayed a poor sensitivity and showed IC₅₀ values that were approximately 10 µM. The different IC₅₀ values were not associated with the mutational profiles of the cell lines, including the amplification of the BRAF or MITF genes, or to the expression of KIT protein (Table 1).

Melanoma cell lines LM20 and LM38 showed primary resistance to PLX4032 lacked p16 and KIT protein expression but showed different gene alterations because LM20 cells harbored MITF amplification and mutated TP53, whereas LM38 lacked p14/ARF gene and PTEN expression because of gene methylation. PTEN deficiency has been hypothesized to promote melanoma cell proliferation and survival through AKT activation, which may decrease the dependency on ERK signaling. Moreover, PTEN loss has been detected in a melanoma tissue biopsy obtained from a patient relapsing on treatment with PLX4032 [13]. When response of melanoma cell lines to PLX4032 concentrations inhibiting cell growth was examined, we found that the drug produced an accumulation in the G₁ phase of cell cycle regardless of PTEN status (Figures 1 and W1). Growth inhibition was associated with apoptotic cell death, as documented by AK release and activation of caspase 3, at higher levels in PTEN-positive samples, indicating a role for PTEN in the induction of cell death in response to PLX4032 (Figure 1, A and B).

Modulation of MAPK and AKT Signaling by PLX4032 Treatment

To define the cellular response that was associated with PLX4032 sensitivity, we examined the effect of treatment on downstream signaling pathways that regulate cell growth and survival. PLX4032 treatment strongly reduced the levels of pERK and pAKT in most drug-sensitive cell lines, independently of PTEN status. In addition, down-regulation of p70^S6K, which is activated downstream of the mammalian target of rapamycin signaling, was detectable in most lines, and CCND1 expression was downregulated in all drug-sensitive cell lines, consistently with an accumulation in the G₁ phase of the cell cycle. In contrast, pAKT, pERK, pp70^S6K, and cyclin D1 levels were not affected by the treatment in the resistant LM20 and LM38 cells, in keeping with the poor antiproliferative and cytotoxic effects (Figure 1C).

A resistant cell line (LM17R) was generated by repeated drug exposure from the cell line LM17, which showed extensive cell death after PLX4032 treatment. LM17R showed reduced sensitivity to the antiproliferative effect of PLX4032, diminished AK release, caspase 3 activation, and G₁ block of the cell cycle, as well as unresponsiveness of pERK, pAKT, and CCND1 (Figure 2). Sequence analysis confirmed the presence of the heterozygous V600E BRAF mutation and excluded the presence of secondary mutations in exons 11 and 15 and in RAS gene; in addition, the same number of copies of the BRAF gene as the parental LM17 cells was detected.

To assess whether the MAPK pathway can be modulated downstream of mutated BRAF in resistant cells, we tested whether MEK inhibition affected pERK levels and cell proliferation. Treatment with the MEK1/2 inhibitor UO126 reduced pERK signal and inhibited proliferation in LM20 and LM38 as well as in LM17R cells compared with that in LM17 (Figure 3, A and B), indicating that these cell lines retained the susceptibility to MEK inhibition.

A shift in signaling from BRAF to CRAF after BRAF inhibition has been described in melanoma cells, with CRAF mediating ERK activation [21]. Therefore, we silenced CRAF in LM38 cells using specific siRNA to test whether the sensitivity to PLX4032 increased by reducing CRAF levels. The CRAF siRNA downregulated CRAF protein levels without affecting pERK levels and cell sensitivity to PLX4032. Similar results were obtained also in LM17R cells (Figure 3, C and D).

Molecular Characterization of Melanoma Cell Lines Showing Resistance to PLX4032

To identify new potential markers that are associated with PLX4032 resistance and candidate genes, the MLPA analysis was used to genetically characterize the resistant melanoma cell lines. Several probes showed values indicating gene gain or loss (Figure W2). Amplification of CCND1 at 11q13 and of CTNNB1 at 3p21 was detected in LM20 cells, whereas the LM38 line showed a different pattern of alterations, including MET amplification at 7q31. MET, CCND1, and CTNNB1 gene amplifications in LM38 and in LM20 were confirmed by FISH analysis (Figure 4A and data not shown) and by using quantitative PCR assessing gene copy number (Table W1). MLPA analysis showed no difference in the pattern of alterations between LM17R and LM17, indicating that the acquisition of resistance to PLX4032 was not associated to gain or loss of the tested genes.

To further explore the mechanisms of PLX4032 resistance, a proteomic-multiplexed analysis of pTyr signaling and antibody validation was used to screen pTyr proteins that were modulated by treatment in PLX4032-sensitive and -resistant melanoma cells. We observed a high degree of heterogeneity in the pTyr profiles in the different cell lines (Figure W3). To identify the most abundant phosphorylated proteins in LM20 and LM38 cell lines, protein bands from anti-pTyr immunoprecipitates of cell lysates were resolved in SDS-PAGE, excised from preparative silver-stained gel, and processed for MALDI-TOF mass spectrometry analysis. The identified proteins indicated that pTyr-based cell signaling was activated in the v-src sarcoma viral oncogene homolog (SRC)/FAK axis in LM20 cells, whereas it was prevalently activated in the MET axis in LM38 cells (Figure 4C). These data were consistent with MET gene amplification in LM38 cells and CTNNB1 amplification in LM20 cells for the role of SRC activity in regulating CTNNB1 signaling. Immunoblot analysis confirmed the presence of the phosphorylated MET receptor in LM38 cells, whereas the phosphorylated form of STAT3, which is activated downstream of SRC, was detectable in LM20 cells. The MET and STAT3 proteins were present but not phosphorylated in the other cell line. In particular, high levels of non-tyrosine-phosphorylated STAT3 were detected in LM38 cells, and both lines showed high pSRC levels, which were not reduced by PLX4032 treatment (Figure 4B).

To define whether PLX4032 resistance was mediated by the increased expression of ABC transporters, we assessed protein expression of ABCB1/Gp170, ABCC1/MRP1, ABCC2/MRP2, ABCC4/MRP4, and ABCG2/BCRP in the resistant melanoma cell lines. Differential expression was observed for BCRP and MRP4 (Figure W4). However, BCRP overexpression did not result in resistance to PLX4032 as shown by using a mutant BRAF isogenic model system [22]. In addition, topotecan, a well-known MRP4 substrate, displayed a similar effect in LM17 and LM17R cells despite increased MRP4 levels (data not shown). Thus, PLX4032 resistance is not determined by ABC transporters.

MET and SRC as Additional Targets for Combined Treatment with PLX4032

On the basis of the results of molecular profiling, MET and SRC represented new candidate targets expressed at high levels and activated in LM38 and LM20 melanoma cells intrinsically resistant to PLX4032. We thus tested the effect of combining PLX4032 with drugs that inhibited MET and SRC kinases.

The MET inhibitor SU11274 significantly inhibited the proliferation of most of the melanoma cell lines that were examined, including PLX4032-resistant lines, with IC₅₀ values of approximately 10 µM (data not shown). The combined treatment with SU11274 and PLX4032 produced a synergistic interaction when tested in LM38 cells (interaction index = 2.5), and growth inhibition was associated with an accumulation of cells in G₁ and AK release in the absence of caspase 3 activation (Figure 5A and not shown). The potentiating effect that was obtained by the concomitant inhibition was evident also when other MET inhibitors were tested (Figure 5B). After the cotreatment with SU11274 and PLX4032, pERK and pAKT were not downregulated; in contrast, we found a strong down-regulation of MET signaling through pFAK and pSHC (Figure 5C).

Because MET is involved in tumor invasion, we evaluated the effects of the combined treatment on the ability of melanoma cells to invade Matrigel and migrate in vitro. LM38 melanoma cells were highly responsive to the MET ligand hepatocyte growth factor (HGF), as the addiction of HGF determined a significant increase in the number of cells that migrated through the Matrigel layer (not shown), further confirming the role of MET signaling in mediating the invasive capacity in these cells. Indeed, blocking MET signaling by treatment with SU11274 alone or in combination with PLX4032 strongly inhibited Matrigel invasion. Notably, a moderate effect was observed after treatment with PLX4032, indicating that BRAF inhibition, although not affecting cell growth, may alter the invasive activity of melanoma cells, even in the presence of exogenous HGF (Figure 5D). Moreover, LM38 cells produced HGF (data not shown), thus suggesting that an autocrine loop contribute to MET pathway constitutive activation. In addition, the combined drugs downregulated the expression of β₁-integrin, the receptor for extracellular matrix laminin that is involved in adhesive and invasive cellular processes (Figure 5E). Scratch wound assays showed that the combination of PLX4032 with SU11274 prevented wound closure, whereas the single drugs impaired wound healing to a limited extent, confirming the effect of the combination on cell migration (Figure 5F).

To confirm that MET inhibition can cooperate with BRAF inhibition siRNA silencing of MET was tested. A synergic effect on cell proliferation was detected (interaction index = 1.36), and down-regulation of MET and SHC signal was shown, whereas pERK and pAKT levels were maintained (Figure 6, A and B).

To assess the functional relevance of the SRC pathway in LM20 cells, the BMS-354825 multikinase inhibitor targeting SRC family kinases was used. When tested in the panel of melanoma cell lines, BMS-354825 displayed a poor inhibitory effect on cell growth, and its antiproliferative effect was not related to the expression of KIT protein, which is one of the kinases targeted by the compound (not shown). BMS-354825 showed a weak inhibitory effect on cell growth in LM20 cells, whereas the combination of BMS-354825 with PLX4032 displayed significant antiproliferative and cytotoxic effects (interaction index = 2.1). Another SRC inhibitor, E804, exerted an additive effect with PLX4032, further corroborating the role of SRC signaling in LM20 cells (Figure 7A). Treatment with BMS-354825 downregulated the levels of phosphorylated SRC protein and of the downstream targets paxillin and p130CAS; in addition, BMS-354825 reduced pFAK levels. In contrast, no effect was detectable on pERK and pAKT levels also with this drug combination, suggesting that it is not a necessary requirement to impair cell proliferation (Figure 7B and data not shown). The combined treatment with PLX4032 and BMS-354825 decreased MMP-2 production by LM20 melanoma cells, which was measured using gelatin-gel zymography (Figure 7C), and reduced the expression of β₁-integrin (Figure 7D).