Targeting the MAPK pathway by combined inhibition of BRAF and MEK has increased overall survival in advanced BRAF-mutant melanoma in both therapeutic and adjuvant clinical settings. However, a significant proportion of tumors develop acquired resistance, leading to treatment failure. We have previously shown p63 to be an important inhibitor of p53-induced apoptosis in melanoma following genotoxic drug exposure. Here, we investigated the role of p63 in acquired resistance to MAPK inhibition and show that p63 isoforms are upregulated in melanoma cell lines chronically exposed to BRAF and MEK inhibition, with consequent increased resistance to apoptosis. This p63 upregulation was the result of its reduced degradation by the E3 ubiquitin ligase FBXW7. FBXW7 was itself regulated by MDM2, and in therapy-resistant melanoma cell lines, nuclear accumulation of MDM2 caused downregulation of FBXW7 and consequent upregulation of p63. Consistent with this, both FBXW7-inactivating mutations and MDM2 upregulation were found in melanoma clinical samples. Treatment of MAPK inhibitor–resistant melanoma cells with MDM2 inhibitor Nutlin-3A restored FBXW7 expression and p63 degradation in a dose-dependent manner and sensitized these cells to apoptosis. Collectively, these data provide a compelling rationale for future investigation of Nutlin-3A as an approach to abrogate acquired resistance of melanoma to MAPK inhibitor targeted therapy.
Upregulation of p63, an unreported mechanism of MAPK inhibitor resistance in melanoma, can be abrogated by treatment with the MDM2 inhibitor Nutlin-3A, which may serve as a strategy to overcome resistance.
Advanced melanoma is an aggressive cancer. Discovery of the oncogenic mutation BRAFV600E led to development of targeted BRAFi, which have significantly prolonged overall survival and progression-free survival in advanced melanoma (1). Combined BRAF and MEK targeted therapy increases the rate and durability of treatment responses, and this combination therapy is now used in both therapeutic and adjuvant clinical settings (2). However, despite initially impressive clinical responses, most patients with advanced melanoma treated with MAPK inhibitors develop acquired resistance within 6 to 7 months (3). Resistance is an important clinical challenge, and understanding the mechanisms of resistance is currently an area of intense research activity.
TP63 is critical for the development and maintenance of squamous epithelia. It is mainly expressed in the basal layers of stratified epithelial tissues where it is considered a cell lineage–specific marker (4). TP63 has been linked to chemosensitivity in a number of tumors and is a biomarker for poor prognosis in breast cancer and follicular cell lymphoma (5, 6). We previously provided the first evidence supporting an oncogenic role for p63 in melanoma. We reported upregulation of p63 mRNA and protein in melanoma and induction of p63 nuclear stabilization and partial relocation to the mitochondria in response to DNA-damaging agents. Consequently, p63 has an antiapoptotic role in melanoma, contributing to chemoresistance and its expression correlates with melanoma-specific death (7).
Multiple ligases degrade p63 in keratinocytes. This includes E3 ubiquitin ligases (Nedd4, Itch, and its homolog WWP1) and Ubc9, a SUMO-conjugating enzyme. Pirh2 (p53-induced RING-H2) also modulates keratinocyte differentiation by targeting both TA and ΔNp63 for proteasomal degradation (8–16).
FBXW7 is a E3 ubiquitin ligase that contributes to p63 degradation in keratinocytes (17) and functions as a potent tumor suppressor with many oncoprotein targets. FBXW7 is also mutated in about 8% of melanomas and is inactivated in about 40% of melanoma cell lines (18). In keratinocytes, MDM2 and FBXW7 cooperatively control the stability of p63 protein during keratinocyte differentiation and UV irradiation. MDM2 binds with ΔNp63 and transports it into the cytoplasm where its phosphodegron domain is recognized by FBXW7 (17). HDAC inhibitors (HDACi) reduce ΔNp63 protein levels via FBXW7 E3 ubiquitin ligase (19). FBXW7 is alternatively spliced to produce three C-terminal isoforms, alpha, beta, and gamma (20–24). The cellular localization signal is in the C-terminal variable region, whereas the N-terminal region includes seven WD40 domains, which recognizes and binds with its target protein, and is conserved between different isoforms. Several inactivating mutations are reported in this WD40 region, which selectively loses the capacity to target specific protein in human tumors (25–28).
MDM2 and MDM4 are potent oncogenes and are commonly overexpressed in cancer, including melanoma. They are well-known inhibitors of p53 and in normal tissues maintain p53 at reduced intracellular levels by mediating its rapid degradation and subsequently inhibiting its transcriptional activity (29–32). P53 mutations are common in many UV-associated skin cancers (33), but are less common in melanoma, present in up to 19% as a point mutation, and in the remaining wild type (wt) tumors, p53 is often inactivated by other mechanisms (34), such as MDM2 amplification.
MDM2 and MDM4 amplification, although mutually exclusive, are often involved in mechanisms of wt p53 inactivation in melanoma (35–38), and targeting MDM2 is a recognized therapeutic approach in wt TP53 cancers. Nutlin-3A, a small-molecule MDM2 inhibitor, prevents its association with p53, leading to wt p53 accumulation and reactivation (39). Nutlin-3A also stabilizes p73 and stimulates p53-independent proapoptotic activities (40). Nutlin-3A is currently used as an antitumor agent both in vitro and in vivo, targeting solid tumors including melanoma, breast cancer, prostate cancer, retinoblastoma, and hematologic malignancies (41). Nutlin-3A treatment downregulates IGF1R protein levels by stabilizing MDM2 and increases migratory capacity of melanoma cells via ERK activation, although it reduces their invasive potential by destabilizing MAPK/PI3K-balanced activation, leading to disrupted IGF1R–MMP2 activation (42).
Here, we investigate a possible role for p63 involvement in acquired resistance to MAPK inhibition, characterizing p63 expression and mechanisms of p63 regulation upon acquisition of MAPK inhibitor resistance. We also provide a rationale for the use of Nutlin-3A as an approach to reducing acquired resistance associated with therapeutic MAPK inhibition in melanoma.
Materials and Methods
HaCaT, HCT116, and HEK293T cells were cultured in DMEM medium supplemented with 10% FCS and 1% pen/strep antibiotic solution. All melanoma cell lines were cultured in RPMI-1640 medium supplemented with 10% FCS and in antibiotic-free conditions. Establishment of vemurafenib- or trametinib-resistant melanoma cells was performed on BRAFV600E-mutated cell lines seeded and treated with increasing doses of PLX4032 or trametinib for at least 1 month. Media with fresh drug were replaced every other day, and cells were split when necessary. Afterward PLX4032 drug dose was maintained to 3 μmol/L and trametinib for 50 nmol/L. Cells were used in further experiments only after the resistance was confirmed. Cell lines LM16, LM17, and LM36 (sensitive and PLX-resistant) kindly provided by Dr. Monica Rodolfo were established in another laboratory following similar procedure.
Protein extraction and Western blotting
Equal amount of protein lysates was resuspended in 10X NuPage reducing agent (500 mmol/L DTT, Invitrogen #NP0009) and 4X NuPage lithium dodecyl sulfate sample buffer (Invitrogen #NP0007). Samples were mixed and heated at 70°C for 10 minutes. Protein samples were resolved using 10% Bis-Tris gels. The fast transfer was performed using the BioRad Trans-Blot Turbo RTA Mini PVDF Transfer Kit (#1704274) with Trans-Blot Turbo system at 25 A and 2.5 V for 10 minutes. Upon completion of protein transfer, the membrane was washed with dH2O and left for blocking for 1 hour with 5% BSA (Sigma). The membrane was incubated with primary antibody at +4°C for overnight. The next day, after 3 × 10 minutes phosphate buffered saline Tween-20 0.1% (PBST) washings, the membrane was incubated with anti-mouse antibody (DAKO) diluted 1:5,000 (vol/vol) or anti-rabbit antibody (DAKO) diluted 1:10,000 (vol/vol) in PBST supplemented with 1% BSA for 1 hour at room temperature. The membranes were washed 3 times with PBS-Tween 0.1%, and bound secondary antibody was detected by enhanced chemiluminescence with ECL Plus (GE Healthcare) according to the manufacturer's protocol. For cytoplasmic and nuclear protein extraction, cell pellet was resuspended in 300 μL hypotonic buffer. Later, it was mixed with 25 μL 10% NP40 (detergent) reagent and vortexed for 20 seconds, followed by 10-minute centrifugation at 3,000 rpm at 4°C. Then carefully supernatant was collected in separate tube labeled as cytoplasmic fraction. The nuclear pellet was gently washed twice with ice-cold PBS to avoid the contamination from cytoplasmic fraction. Nuclear pellet was resuspended in 50 μL complete cell extraction buffer (#FNN0011, Thermo Fisher), followed by 30-minute incubation on ice with vortexing at every 10-minute intervals. The cells debris was removed by 30-minute centrifugation at full speed at 4°C. The supernatant (nuclear fraction) was collected into a clean microcentrifuge tube. These both fractions were analyzed on SDS-PAGE as previously described. Antibodies used were as follows: p63 (#SFI-6, DSC Diagnostics Germany), p63 (#M731729-2, Dako, Agilent Technologies LDA UK Limited), p63 (#H137, Santa Cruz Biotechnology), GAPDH (Generon), MDM2 (#N20, Santa Cruz Biotechnology), and FBXW7 (#109617, Abcam).
In vitro ubiquitination assay
In vitro ubiquitination assay performed with Abcam (ab139469) kit using purified recombinant FBXW7 protein. As per the kit protocol, two control reactions (without ATP or MDM2) and final reactions with recombinant FBXW7 mixed together. The reactions were incubated at 37°C for 90 minutes and terminated with 2x SDS-PAGE gel loading buffer, followed by heating at 95°C. The reaction products were subsequently analyzed by SDS-PAGE and chemiluminescent Western blot.
Lentiviral vector pLKO.1 puro was used to deliver shRNAs against p63 and FBXW7 expression. For p63: GAAGAAACTCTACTGCCAAAT, GAATTCAACGAGGGACAGATT and for FBXW7: GCAACAACGACGCCGAATTAC, GGACTCCTCAGGAAACCAAGA and scramble: TGTTATACGATTTCCTGCACC sequences were cloned in pLKO.Puro vector according to Addgene protocol. pLenti-puro was a gift from Ie-Ming Shih (Johns Hopkins University School of Medicine, Baltimore, MD; Addgene plasmid #39481).
DNA mutation analysis
Paraffin-embedded tissues derived from each lesion were macrodissected, and DNA was isolated by using the QIAamp DNA micro Kit (Qiagen). BRAF exon 15 and NRAS exons 2 and 3 were amplified by previously optimized PCR primers and conditions. PCR products were purified by ExoSAP-IT (USB Corporation) and sequenced using Big Dye Terminator V3.1 Cycle Sequencing Kit on AB3500dx DNA Analyzer (Life Technologies). Sequence analysis was performed by Variant Reporter (Life Technologies).
Paraffin-embedded tissue sample materials representing different stages of melanoma development analyzed in this study were obtained from Barts and the London NHS Trust. The investigators obtained informed written consent from the patients. All patients' data collected and used in the article have been consented with an Institutional Review Board–approved protocol that allows comprehensive analysis of tumor samples and covered by an ethical approval (07/QO604/23 by the East London and The City HA Local Research Ethics committee 2, UK). Tissue samples were fixed in 10% formal saline solution before being embedded in paraffin wax. Note that 3 to 4 μm sections were mounted onto a positively charged slide, dewaxed in xylene twice for 3 minutes, and rehydrated in alcohol twice for 3 minutes before being placed in running water for 5 minutes. Antigen retrieval for p63 staining, slides were microwaved in Tris-EDTA unmasking solution pH9 for 35 min. Following antigen retrieval sections were placed in running water for 5 minutes before endogenous peroxidase activity was blocked by incubation with 3% hydrogen peroxide solution for 15 minutes. Sections were washed 3 times in wash buffer before Normal Horse Serum (Vector R.T.U. kit, Vector Labs) was applied for 20 minutes. The serum was then replaced by primary antibody diluted in antibody diluent to the appropriate concentration. Negative control sections were treated with antibody diluent without antibody. Sections were incubated at room temperature for 1 hour and then washed 3 times in wash buffer. Universal Biotinylated Secondary Antibody (Vector R.T.U. kit) was added for 30 minutes, and sections were washed in wash buffer twice and then incubated with Vectastain Elite ABC Reagent (Vector R.T.U. kit) for 30 minutes. Sections were washed 3 times with wash buffer and then incubated for 5 minutes with DAB solution before being washed in running water for 5 minutes. The clinical samples labeled for p63, FBXW7, and MDM2 were reviewed by three independent scorers, blinded to any clinical details of the patients.
Immunofluorescence staining of tissue sections
Paraffin-embedded tissue samples of patients with melanoma before and after acquisition of MAPKi resistance were kindly provided by Dr. Christian Posch. After hydration of paraffin-embedded tissue in a xylene-alcohol series, antigen retrieval was done in citric buffer and left to cool down to room temperature. Samples were washed, covered by blocking buffer in a humidity chamber, then incubated with primary antibodies (MART-1 from Signalway Ab and SFI-6 against p63 from DSC Diagnostics), diluted in blocking buffer, and incubated overnight at 4°C in a humidity chamber. The next day, excess primary antibody was removed with 3 washes and samples were incubated with the appropriate fluorophore-coupled secondary antibody, against the antigen of the primary antibody (1:800 in blocking buffer, 1.5 hours, room temperature). This was followed by 3 washes and the staining of the nuclei with 4′, 6-diamidino-2-phenylindole (DAPI; 1:1,000 in PBS, 5 minutes, room temperature). Excess DAPI was removed with one wash in PBS (5 minutes, room temperature). Coverslips were mounted on the slides with the sections using Immu-Mount (Thermo Fisher) to preserve the immunofluorescence staining. Images of stained samples were then acquired with an INCell 2200 automated microscope using both 4× and 20× objectives. 4× objective fields were taken with a 15% overlap to allow image stitching and to provide a section overview. All images have been contrasted based on a negative control.
Cells were seeded onto coverslips and fixed the following day in 4% formaldehyde for 10 minutes followed by two 5-minute washes in PBS. Cell membranes permeabilized in 0.3% Triton X-100 for 2 minutes at room temperature coverslips were then washed twice with PBS. Cells were incubated with 5% goat serum/PBS for 30 minutes. After removal of the serum, cells were incubated with primary antibody diluted in 5% goat serum/PBS overnight at 4°C. Cells were washed 3 times for 10 minutes in PBS before being incubated with secondary antibody dilution 1/500 in 5% PBS/goat serum) for 1 hour at room temperature in the dark. Cells were washed twice for 10 minutes in PBS and then incubated with DAPI for 10 minutes followed by two washes in PBS for 10 minutes. Coverslips were mounted onto a glass slide and stored at 4°C protected from light.
Annexin V binding assay
To induce apoptosis, cells were treated with different drugs for 24 hours. After treatment, adherent cells and any cells in supernatants were collected and pelleted before being washed in PBS. Cells were resuspended in 400 μL 1X Becton Dickinson Annexin V Binding Buffer (BD Pharmingen) and processed according to the manufacturer's instructions. Results were analyzed using the Becton Dickinson FACSDiva software.
p63 expression significantly correlates with BRAF and NRAS mutational status in melanoma
We have previously shown that expression of oncogenic p63 correlates with worse outcome in melanoma (7), but the association between p63 expression and BRAF/NRAS mutational status has not yet been established. A panel of 50 melanomas previously analyzed for p63 expression (Fig. 1A) was macrodissected and genotyped for BRAF (exon 15) and NRAS (exons 2–3) mutational status. We identified a significant correlation between p63 expression and mutation in BRAF and NRAS (Fig. 1B and Table 1), supporting a similar correlation in established melanoma cell lines at both gene (Supplementary Fig. S1B) and protein levels (Fig. 1C). Validation with an independent melanoma dataset from The Cancer Genome Atlas (TCGA; ref. 43) confirmed an association (although not statistically significant) between increased p63 expression and BRAF and NRAS mutational status (Supplementary Fig. S1A).
In order to test whether p63 expression affects MAPK activation, p63 isoforms were transiently expressed in p63-null melanoma cell lines (Mel 224 and LM1) and resulted in increased expression of phosphorylated MEK1/2 (Fig. 1D). Conversely silencing of endogenous p63 in a BRAFV600E melanoma cells (A375M) significantly reduced expression of phosphorylated MEK1/2 (Fig. 1E). Overexpression of p63 ΔNβ isoform in WM164 cells significantly increased EGFR expression and affected phosphorylation activity. EGFR upregulation, already reported in melanoma cell lines resistant to BRAF inhibition (44), resulted in increased phosphorylated MEK1/2 expression (Fig. 1F). EGFR inhibition abolished p63-driven effects on phospho-MEK1/2 expression, suggesting that p63 can influence MAPK signaling via EGFR in melanoma cells.
Development and characterization of MAPK inhibitor–resistant cell lines
Cellular models were established to investigate the consequences of p63 expression in MAPKi-resistant melanomas. Five BRAF inhibitor (BRAFi)–resistant (WM1158, WM793, A375M, WM278, and WM164) and 2 MEK inhibitor (MEKi)–resistant (A375M and WM164) melanoma cell lines were chronically treated 3 times a week for 10 to 12 weeks with 2 μmol/L PLX4032 (vemurafenib) and 50 nmol/L trametinib, respectively, until they acquired resistance. Major changes in cell shape, plasticity, and size were observed: WM793-PLX and WM278-PLX cells became very elongated and acquired phenotypic features of mesenchymal cells; conversely, WM1158-PLX cells transformed from mesenchymal-like to an epithelial-like morphology (Supplementary Fig. S1C). Trametinib-resistant cells increased their size and became very large and multinucleated. These morphological changes were driven by a cytoskeleton reorganization with significant changes in actin fibers polarization and angle dispersion (Supplementary Fig. S1D and S1E). To confirm the acquisition of BRAFi or MEKi resistance, cells were treated with vemurafenib (3 and 6 μmol/L) or with trametinib (50 nmol/L) for 24 hours. Expression of phosphorylated MEK and ERK was significantly decreased in BRAFi- and MEKi-sensitive cell lines, whereas high-dose vemurafenib or trametinib had no effect on MEK and ERK phosphorylation in MAPKi-resistant cells (Fig. 2A–C; Supplementary Fig. S1F and S1G).
Resistance to BRAFi was further tested by measuring cell viability of resistant and sensitive cell lines after 24 hours of increasing concentration of vemurafenib. All cell lines tested showed significantly increased cell survival with vemurafenib compared with their sensitive counterpart (Fig. 2D). Growth curves revealed significant lower cell proliferation in BRAFi- and MEKi-resistant cells (Fig. 2E; Supplementary Fig. S1H). High expression levels of CD271 have been associated with metastatic progression, enhanced survival, and resistance to MAPK inhibitors (45), and these cells also displayed enrichment of CD271 expression (Fig. 2F; Supplementary Fig. S1I) when compared with their parental nonresistant cell lines.
Upregulated p63 expression contributes to development of MAPKi resistance in melanoma
To investigate the impact of p63 expression in MAPKi-resistant melanoma cells, qPCR analysis was performed using specific primers designed to detect all isoforms of TAp63 and ΔNp63 isoforms (7). TP63 mRNA expression was compared with normal primary human melanocytes (NHEM), which confirmed lack of detectable TA and ΔN p63 expression. QPCR analysis of TA and ΔN p63 showed an overall significant upregulation of both isoforms in all our melanoma MAPKi-resistant cell lines (Fig. 3A; Supplementary Fig. S2A). Similar results were observed when p63 protein expression was measured: p63 isoforms were either upregulated in a cell-specific manner or their expression was only increased in MAPKi-resistant cells (Fig. 3B–D; Supplementary Fig. S2B). We further confirmed p63 upregulation at RNA and protein level in an independent panel of BRAFi-resistant melanoma cell lines (Fig. 3A and C).
In order to determine whether increased p63 expression is sufficient to affect cell survival, we used RNA interference specifically targeting p63 expression in melanoma cells. Western blot analysis showed that both shRNA sequences reduced endogenous expression of ΔNp63β protein in A375M cells compared with vector control (Fig. 3E). Cells depleted of p63 were subsequently exposed to vemurafenib demonstrating significant reduction in cell viability when compared with control cells (Fig. 3F). Similarly, BRAFi-resistant cells infected with p63 shRNAs showed significant decrease in cell viability upon exposure to increasing doses of vemurafenib (Fig. 3G; Supplementary Fig. S2C), demonstrating that reduction of p63 expression restores sensitivity to MAPKi treatment. Reactivation of p63 expression was also detected in melanoma samples from patients who developed resistance to combined BRAFi and MEKi (Fig. 3H; Supplementary S2D and S2E).
Proteasomal degradation influences p63 stability in melanoma cells
To examine regulation of p63 isoform stability in melanoma, established melanoma cell lines were treated with the peptide aldehyde inhibitor MG132 (2 μmol/L) for 8 hours. Western blotting analysis showed stabilization of endogenous p63 protein expression, suggesting that proteasomal degradation may be one possible mechanism controlling p63 protein stability in melanoma cells (Fig. 4A). Given the data supporting proteasome-mediated degradation of p63 in keratinocytes, we investigated E3-ubiquitin ligases in relation to p63 degradation in melanoma cells. Expression of ITCH is heterogeneous and does not always inversely correlate with p63 expression (Supplementary Fig. S3A), suggesting that its interaction with p63 in melanoma might only occur in a cell-type–specific manner. Moreover, Western blotting analysis showed no significant changes in ITCH expression in BRAFi-resistant melanoma cell lines despite consistent upregulation of p63 isoforms (Supplementary Fig. S3B). We therefore investigated an alternative E3 ubiquitin ligase, namely FBXW7. Protein analysis revealed FBXW7β is the main isoform expressed in several established melanoma cell lines regardless of BRAF mutational status (Supplementary Fig. S3C). Isoform-specific QPCR analysis (alpha, beta, and gamma FBXW7) showed significant reduction in FBXW7β isoforms in resistant cells (Fig. 4B; Supplementary Fig. S3D). Similarly, Western blotting analysis showed significant downregulation of FBXW7 protein in all MAPKi-resistant cells (Fig. 4C and D). Immunofluorescence staining of FBXW7 protein in WM1158 and WM793 cells revealed that in both parental and BRAFi-resistant cells, FBXW7β is predominantly localized to the nucleus (Fig. 4E; Supplementary Fig. S3E). Subcellular fractionation experiments confirmed this nuclear localization in all melanoma cell lines, and also revealed significantly reduced FBXW7β expression in BRAFi-resistant cells (Fig. 4F). Our data therefore demonstrate that melanoma expresses FBXW7β protein, which is mainly confined to the nucleus.
Downregulated FBXW7 mRNA and protein in MAPKi-resistant cells suggested a possible downstream effect of MAPK signaling on FBXW7. Pancreatic cancer samples with NRAS mutation displayed higher ERK activation leading to decreased FBXW7 expression, with ERK found to directly phosphorylate FBXW7 at the Thr205 residue and regulates protein stability via proteasomal degradation (46). To investigate this in melanoma, A375M and WM793 melanoma cells were treated with vemurafenib for 24 hours. In contrast to BRAFi-resistant cells, transient incubation of melanoma cells with vemurafenib did not show any significant reduction of endogenous expression of FBXW7 (Supplementary Fig. S3F), suggesting that FBXW7 protein stability in melanoma cells does not rely solely on ERK phosphorylation.
Using specific antibodies against FBXW7, immunoprecipitation experiments in few melanoma cell lines demonstrated direct protein interaction between endogenous ΔNp63 and FBXW7 (Fig. 4G). FBXW7 interaction occurs with all p63 isoforms (Fig. 4H). To further confirm the direct relationship between FBXW7 and p63 protein, FBXW7 shRNA knockdown experiments were performed: in A375M melanoma cells, p63 protein levels were significantly upregulated only in the cells with FBXW7 knockdown (Fig. 4I).
Expression and location of MDM2 influence p63/FBXW7 interaction in MAPKi-treated melanoma
In keratinocytes, MDM2 and FBXW7 cooperatively control stability of p63 protein following differentiation and UV irradiation. MDM2 transports ΔNp63 into the cytoplasm where, upon interaction with the phosphodegron motif of FBXW7, it undergoes proteasomal degradation (17). Interaction between FBXW7 and MDM2 has not been systematically assessed in any other cellular systems, and in order to test whether this model is also relevant in melanoma, we examined patterns of expression and localization of MDM2 and FBXW7. Primary melanomas showed strong nuclear reactivity of FBXW7 in both tumor and epidermal keratinocytes, whereas MDM2 distribution was cytoplasmic and nuclear with slight reactivity within basal layer of epidermis (Fig. 5A). To further analyze cytoplasmic and nuclear reactivity for MDM2 and FBXW7, 31 and 24 primary melanoma tumor nests respectively were analyzed in seven clinical samples. Significantly higher MDM2 reactivity was present in the cytoplasm compared with the nucleus, whereas the reverse trend was observed for FBXW7. This inverse correlation between MDM2 and FBXW7 expression was further observed at the gene level in a TCGA SKCM dataset of skin cutaneous melanoma (Supplementary Fig. S4A). To investigate the effect of resistance on MDM2 expression, MDM2 mRNA and protein levels were compared in MAPKi-sensitive and -resistant cell lines. MDM2 mRNA and protein was significantly elevated in most established MAPKi-resistant cell lines (Fig. 5B–D), whereas no significant changes were observed in most cell lines for MDM4 (Fig. 5C). MDM2 was mainly localized to the nucleus in immunofluorescence experiments and, in MAPKi-resistant cell lines, the intensity of fluorescence was increased significantly (Fig. 5E; Supplementary Fig. S4B and S4C). Subcellular fractionation experiments confirmed nuclear enrichment of MDM2 expression in most MAPKi-resistant melanoma cells (Fig. 5F and G).
To explore whether MDM2 expression and localization reflected a potential interaction with nuclear FBXW7, we undertook immunoprecipitation experiments in WM793, A375M melanoma cells, and HaCaT keratinocytes, which confirmed protein–protein interactions between MDM2 and FBXW7 in melanoma cell lines (Fig. 5H) and HaCaT keratinocytes. In vitro ubiquitylation experiment shows that MDM2 is able to increase FBXW7 ubiquitylation (Fig. 5I). To further investigate the consequence of this interaction, MDM2 was overexpressed in melanoma cells to test the effect on FBXW7 protein expression levels. Transfected cells were treated with the proteasome inhibitor MG132 (2 μmol/L) 40 hours after transfection and cells harvested after 8 hours of MG132 treatment. Western blotting revealed that exogenously overexpressed MDM2 reduced FBXW7 protein in the absence of MG132 (Fig. 5J; Supplementary Fig. S4D), and this downregulation was rescued by proteasome inhibition. The effect on FBXW7 was specific for MDM2 rather than MDM4, which, upon overexpression, failed to reduce endogenous FBXW7 (Supplementary Fig. S4E).
Nutlin-3A degrades p63 expression and sensitizes MAPKi-resistant melanoma cells to apoptosis
Having established that MAPKi resistance in melanoma cells results in p63 upregulation, we next investigated whether p63 inhibition might restore chemosensitivity in MAPKi-resistant cells. As few inhibitors of p63 expression are currently available, we adopted an indirect approach and tested the effects on p63 expression of molecules targeting other p63 regulators. Nutlin-3A, an MDM2-p53 interaction inhibitor, induces p53 reactivation. Nutlin-3A inhibits p73 interactions with MDM2 and MDM4, but does not have an inhibitory effect on TA and ΔNp63 (47). To test whether Nutlin-3A could affect p63 expression, MAPKi-sensitive and -resistant cells were treated with increasing doses of Nutlin-3A. There was a dose-dependent reduction in p63 protein expression by Western blotting in most cellular model tested. In A375M cells, treatment with Nutlin-3A reduced ΔNp63β protein expression in a dose-dependent manner (Supplementary Fig. S5A). Interestingly, Nutlin-3A was able to affect the stability of several p63 isoforms, as seen in the WM793 cells (Fig. 6A). In addition, Nutlin-3A treatment also reduced all p63 isoforms in WM793-PLX cells (Fig. 6B) and in A375M-TR (Fig. 6C). This is the first evidence that Nutlin-3A inhibits p63 expression in melanoma and N-TERT human immortalized keratinocytes (Supplementary Fig. S5B). Because no significant changes were detected at mRNA level in p63 TA and ΔN isoforms in either parental or resistant cells (Fig. 6D; Supplementary Fig. S5C), this suggests that Nutlin-3A only affects p63 protein stability. After Nutlin-3A treatment, most melanoma cell lines showed significant downregulation of ITCH protein (Supplementary Fig. S5D) but increased FBXW7 protein expression in a dose-dependent manner (Fig. 6E and F; Supplementary Fig. S5E). These observations provide further support for the hypothesis that p63 protein stability is specifically controlled by FBXW7 in melanoma cells. Similarly, Nutlin-3A induced 10- and 15-fold increased transcriptional expression of FBXW7 mRNA in A375M and WM793 cells, respectively, particularly the FBXW7β isoform (Fig. 6G; Supplementary Fig. S5F).
Previous evidence has revealed the existence of a p53-binding site in exon 1b of the FBXW7 promoter, and expression of endogenous FBXW7β, but not FBXW7α, was previously shown to be induced in a p53-dependent manner (48). In order to establish whether FBXW7 upregulation after Nutlin-3A treatment was p53-mediated, melanoma cells were transfected with specific siRNA against p53. In the absence of p53, Nutlin-3A had no effect on FBXW7 expression, suggesting that Nutlin-3A–mediated upregulation of FBXW7 may be dependent on p53 stabilization (Fig. 6H). This was further supported by the inability of Nutlin-3A to degrade p63 expression in a couple of p53-mutant melanoma cell lines (CHL-1 and SK-MEL-28; Fig. 6I; Supplementary Fig. S5G). We further investigated whether the effect of mdm2 inhibition in melanoma cells in combination with MAPK inhibitors could synergistically increase cell death. Melanoma cell lines were treated with vemurafenib, trametinib, or Nutlin-3A as well as RG7112 and AMG232 alone or in combination. Apoptosis induction was significantly higher with combined treatment for 24 hours (Fig. 6J; Supplementary Fig. S5H). Taken together, these data highlight the involvement of p63 upregulation in MAPKi resistance and suggest a potential therapeutical approach to prevent it by combining BRAF and/or MEK inhibition with Nutlin-3A or an alternative mdm2-inhibitory treatment.
We have previously reported p63 protein expression in more than 50% of melanomas irrespective of tumor stage (7). Here, we further analyzed mutational status of BRAF and NRAS in a subset of tumors previously scored for p63 expression establishing a statistically significant correlation between p63 expression and BRAF or NRAS mutation. Although our results raised the possibility of an association between p63 and MAPK, a direct link between the two signaling pathways has never been established.
p63 upregulation has been reported in another skin malignancy—cutaneous squamous cell carcinoma (49–53). In keratinocytes, tyrosine kinase receptor EGFR has been implicated in regulating ΔNp63 expression by several signaling mechanisms including via PI3K pathway (52). Here, we showed that in melanoma cells, ectopic overexpression of p63 isoforms in BRAF wt cell lines in conjunction with knockdown of p63 in BRAFV600E-mutated cell lines resulted in upregulation and downregulation, respectively, of the MAPK signaling cascade. Therefore, in melanoma, p63 expression may positively affect MAPK signaling via EGFR, although further experiments are needed to elucidate exact mechanisms by which p63 could regulate phosphorylation of MEK1/2 in melanoma.
In order to investigate the functional relevance of p63 expression on MAPKi resistance development, we established an in vitro model by generating BRAFi- and MEKi-resistant melanoma cell lines. To date, multiple MAPKi resistance mechanisms have been identified and include emergence of NRAS or MEK mutation, drug-resistant splice variant of BRAF, increased expression of MAP3K, COT, cyclin D1, AEBP1, and several growth factor receptors including IGFR and PDGFR (54–58). Once our cells acquired resistance, drastic changes in cell shape, plasticity, and size were observed (59). Chronically treated cells become more resistant to BRAF or MEK inhibition with reduced apoptosis induction compared with sensitive parental cell lines. Similar findings have been corroborated by other groups (60). Enrichment of CD271 was recently shown in BRAFi-resistant melanoma cells (61, 62). Analysis of our MAPKi-resistant cell lines confirmed significant upregulation of CD271 expression, although with variation across cell lines. Aberrant activation of the MAPK signaling significantly upregulated p63 mRNA and protein expression in our cellular models and in two of three clinical samples from patients with acquired MAPKi resistance. Although such observed p63 changes could be downstream effects of MAPKi, they may represent a driving mechanism for BRAFi resistance.
We previously showed that p63 has an antiapoptotic role in melanoma and confers resistance to chemotherapy (7). Depletion of p63 by RNA interference results in activation of p53-dependent mitochondrial apoptotic pathways, rendering melanoma cells sensitive to chemotherapies as well as to vemurafenib treatment. Consistent with these results, A375M cells depleted of p63 and treated with vemurafenib demonstrated significantly decreased cell survival. Furthermore, p63 depletion in BRAFi-resistant cells resensitized these cells to vemurafenib and enhanced apoptosis. These results suggest that survival role of p63 is enhanced in cells chronically treated with vemurafenib.
In order to understand the impact of p63 upregulation in MAPKi resistance, we explored regulation of p63 expression in melanoma cells. Treatment with the proteasome inhibitor MG132 stabilized p63 isoforms, suggesting p63 stability in melanoma is controlled by an E3 ubiquitin ligase. E3 ubiquitin ligases such as ITCH, FBXW7, WWP1, and Pirh2 have been extensively explored in keratinocytes and shown to regulate p63 protein and differentiation (11, 15, 16, 17). Only FBXW7β was significantly reduced when E3 ubiquitin ligases expressions were compared in MAPKi-sensitive and -resistant cells to identify potential regulators of p63 stability. In keratinocytes, MDM2 and FBXW7 can cooperatively control the stability of p63 protein upon keratinocytes differentiation and UV irradiation. MDM2 transports ΔNp63 into the cytoplasm where it interacts with FBXW7 via phosphodegron motif and degrades it via proteasomal degradation pathway (15). More recently was shown that HDACi also reduces ΔNp63 protein level via FBXW7 (19). Using immunoprecipitation experiments, we have demonstrated the endogenous interaction between p63 and FBXW7 in melanoma cells whereby stable depletion of FBXW7 resulted in upregulation of p63 protein.
P53 mutations are less common in melanoma compared with other skin cancers (34, 62), but MDM2 is often overexpressed in melanoma and recognized as one of the potential causes of p53 inactivity in melanoma cells (36, 63, 64). FBXW7 was found to be mutated in melanoma samples with frequent downregulation of its expression (18). Based on these observations, we hypothesized that FBXW7 downregulation could be either the direct result of the lack of its transcriptional activation by p53, which is degraded by high levels of MDM2 in melanoma, or because of a direct interaction between the two. No evidence for an interaction between MDM2 and FBXW7 has previously been reported for any cell type. We analyzed an independent cohort of melanoma clinical samples (TCGA SKCM) dataset for MDM2 and FBXW7 mRNA expression, and the data at the gene level matched our staining expression pattern, confirming the inverse relationship observed in our established cell lines. Further analysis of MDM2 expression revealed nuclear MDM2 enrichment in MAPKi-resistant cells, and immunoprecipitation experiments confirmed endogenous interaction of MDM2 with FBXW7.
To support our model, inhibition of MDM2 activity by Nutlin-3A resulted in increased FBXW7 expression and decreased p63 protein. Although it is possible that MDM2 overexpression could reduce FBXW7 protein via p53 downregulation, direct interaction between the two suggests that this is less likely. These results provide evidence of a previously unreported mechanism by which MDM2 can increase ubiquitylation activity and regulate FBXW7 protein stability in melanoma, ultimately affecting p63 expression.
Taken together, these data suggested that chronic treatment with MAPK inhibitors results in nuclear mdm2 enrichment with consequent FBXW7 protein downregulation. p63 accumulates as a result of failure of FBXW7-mediated degradation, and overexpression of p63 contributes to increased cell survival and resistance to MAPK inhibitors. Although complete therapeutic knockdown of p63 would be challenging, indirect pharmacologic inhibition of p63 function might reduce MAPKi resistance and potentially improve survival outcome. In this regard, we propose that use of nutlin-3A during treatment with MAPKi targeted therapy would synergistically induce apoptosis and suppress melanoma growth, thereby abrogating MAPKi resistance (65). Our data unveil a novel survival role of p63 in melanoma and provide an approach to abrogate or overcome acquired resistance of melanoma to MAPK inhibitor targeted therapy.
Disclosure of Potential Conflicts of Interest
C. Posch is consultant at Roche, Pierre Fabre, Novartis, BMS, and MSD. No potential conflicts of interest were disclosed by the other authors.
Conception and design: A. Patel, C.A. Harwood, D. Bergamaschi
Development of methodology: A. Patel, V. Mannella, L. Gammon, C.A. Harwood, D. Bergamaschi
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Patel, L. Fraile Garcia, V. Mannella, L. Gammon, T.-M. Borg, T. Maffucci, M. Scatolini, E. Vergani, M. Rodolfo, A. Maurichi, C. Posch, C.A. Harwood, D. Bergamaschi
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Patel, L. Fraile Garcia, V. Mannella, L. Gammon, T. Maffucci, G. Chiorino, R.N. Matin, C.A. Harwood, D. Bergamaschi
Writing, review, and/or revision of the manuscript: A. Patel, T.-M. Borg, A. Maurichi, C. Posch, R.N. Matin, C.A. Harwood, D. Bergamaschi
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Patel, T.-M. Borg, E. Vergani, C.A. Harwood
Study supervision: C.A. Harwood, D. Bergamaschi
We are grateful to Dr. Gary Warnes, Flow Cytometry Core Facility Manager, for technical advice, to Rebecca Carroll and Laura Neal for the assistance with the tissue samples immunostaining, and to Drs. Belen Martin and Jan Soetaert for the technical support with fluorescence microscopy. Funding support was provided by British Skin Foundation for A. Patel (6063s to D. Bergamaschi) and V. Mannella (016/s/16 to T. Maffucci), Barts and The London Charity (458/1494 to D. Bergamaschi and C.A. Harwood), EC Erasmus+ for L. Fraile Garcia (2017-SMP-P-211), and AIRC for E. Vergani (IG 17462 to M. Rodolfo).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.