Abstract
Over 60% of human melanoma tumors bear a mutation in the BRAF gene. The most frequent mutation is a substitution at codon 600 (V600E), leading to a constitutively active BRAF and overactivation of the MAPK pathway. Patients harboring mutated BRAF respond to kinase inhibitors such as vemurafenib. However, these responses are transient, and relapses are frequent. Melanoma cells are efficiently lysed by activated natural killer (NK) cells. Melanoma cells express several stress-induced ligands that are recognized by activating NK-cell receptors. We have investigated the effect of vemurafenib on the immunogenicity of seven BRAF-mutated melanoma cells to NK cells and on their growth and sensitivity to NK-cell–mediated lysis. We showed that vemurafenib treatment modulated expression of ligands for two activating NK receptors, increasing expression of B7-H6, a ligand for NKp30, and decreasing expression of MICA and ULBP2, ligands for NKG2D. Vemurafenib also increased expression of HLA class I and HLA-E molecules, likely leading to higher engagement of inhibitory receptors (KIRs and NKG2A, respectively), and decreased lysis of vemurafenib-treated melanoma cell lines by cytokine-activated NK cells. Finally, we showed that whereas batimastat (a broad-spectrum matrix metalloprotease inhibitor) increased cell surface ULBP2 by reducing its shedding, vemurafenib lowered soluble ULBP2, indicating that BRAF signal inhibition diminished expression of both cell-surface and soluble forms of NKG2D ligands. Vemurafenib, inhibiting BRAF signaling, shifted the balance of activatory and inhibitory NK ligands on melanoma cells and displayed immunoregulatory effects on NK-cell functional activities. Cancer Immunol Res; 5(7); 582–93. ©2017 AACR.
Introduction
Metastatic melanoma is an aggressive type of skin cancer with a high metastatic potential. For more than 20 years, the two therapies approved by the FDA were high-dose IL2 and dacarbazine (DTIC), each associated with response rates of only 10% to 20%, low percentages of complete responses, and no improved overall survival. A search for mutations in components of the MAPK pathway in various common cancers revealed that 60% of melanomas (and 7% to 8% of all cancers) carry an activating mutation in the BRAF gene encoding the serine–threonine protein kinase BRAF. Most (90%) reported BRAF mutations result in a substitution of glutamic acid for valine (c1799 T>A) at amino acid 600 (the V600E mutation; ref. 1). This mutation constitutively activates the kinase and the downstream signal transduction in the MAPK pathway (2), including MEK and subsequently ERK. The study of this genetic change has established the MAPK pathway as a potential drug target in melanoma, leading development of BRAF and MEK inhibitors.
Vemurafenib is a specific inhibitor of the V600E variant of BRAF that inhibits tumor outgrowth of BRAF-mutated melanoma cells. Vemurafenib inhibits the kinase activity of BRAF, abrogates downstream signaling, and blocks in vitro proliferation of cells (3). A randomized phase III trial with vemurafenib in metastatic melanoma patients with a BRAF mutation resulted in complete or partial tumor regression in 48% of patients compared with 5% with dacarbazine (4).
However, these responses lack durability and most tumors develop resistance to the drug. Such resistance to BRAF (and MEK) inhibitors limits efficacy of these treatments. Combination of MAPK inhibitors with immunotherapy is emerging as an alternative treatment for metastatic melanoma tumors.
Natural killer (NK) cells are part of the immune response in melanoma patients (5, 6). NK cells are potent cytolytic effectors in innate and adaptive antitumor immune responses. NK-derived IFNγ primes Th 1 responses (7). NK cells are efficiently expanded in vitro, used in autologous and allogeneic settings, and thus are candidate reagents for adoptive cell therapy. In humans, CD3−CD56+ NK cells represent 5% to 20% of circulating lymphocytes. NK-cell activation depends on a balance of activating and inhibitory signals that determines whether the target will be susceptible to NK-mediated lysis (8). Three natural cytotoxicity receptors (NCR) are involved in NK-cell activation: NKp46 and NKp30, expressed by resting NK cells, and NKp44, induced after stimulation by cytokines. Activation of NK cells induces other receptors, as well. NKG2D, expressed by peripheral NK cells, binds MHC class I polypeptide-related chains A and B (MICA/B) and UL16-binding proteins (ULBP1-6). NKG2D ligands are induced on the membrane of stressed cells and thus promote many of NK functions in vitro. NK cells require DNAM-1 for the elimination of tumor cells that are resistant to NK-cell–mediated cytotoxicity due to the absence of other NK-cell–activating ligands (9, 10). Engagement with inhibitory receptors depends on expression of class I HLAs. Classical HLA-A, -B, and -C alleles are recognized by the killer immunoglobulin receptors (KIR), whereas nonclassical HLA-E ligand binds the NKG2A receptor (8, 11). Accordingly, NK cells can kill target cells that express little or no HLA class I, a characteristic frequently seen in tumors, particularly melanoma cells. Furthermore, NK cells infiltrate various tumors, including melanoma (12), which leads to a favorable clinical outcome (13).
Here, we investigated the susceptibility of seven melanoma cell lines carrying BRAF mutations to lysis mediated by NK cells and the effect of treatment with vemurafenib on their NK immunogenicity.
Materials and Methods
Reagents and cell lines
Vemurafenib (PLX4032/RO5185426; Genentech) was obtained from Institut Roche (Roche Scientific Partnership). Vemurafenib was dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich). DMSO was used as vehicle control.
Human melanoma cell line MelS was obtained in our laboratory (6). Other cell lines were either purchased from the ATCC (A375 and RPMI-7951, hereinafter referred to as RPMI) or provided by other laboratories (SK28 by F. Faure; M14, M88, and M199 by H. Benlalam). Melanoma cells were maintained in DMEM or RPMI media (Life Technologies) supplemented with 10% FCS (Eurobio) and 1% penicillin–streptomycin (Life Technologies). Tumor cells were thawed every 2 months and cultured 1 week before use in experiments. Cell lines were periodically shown to be mycoplasma-free by PlasmoTest (InvivoGen) and were recently authenticated by STR DNA typing (DSMZ Authentication Service).
Sequencing
DNA purification was performed with the DNAeasy blood and tissue Kit (Qiagen). DNA concentration was measured with a NanoDrop device (Thermo Scientific). The Molecular BioPathology module of the translational research laboratory at the Gustave Roussy Institute (Villejuif, France) and the Department of Genetics and Molecular Biology at Cochin Hospital (Paris, France) performed sequencing of BRAF as well as targeted regions for 22 known genes in the cell lines, using Ion Torrent Next-Generation Sequencing (Life Technologies) and Ion AmpliSeq Colon and Lung Cancer Research Panel (v2).
Cell proliferation using the xCELLigence system and CFSE staining
The proliferation of melanoma cells was evaluated with the xCELLigence System (ACEA Biosciences). After background measurement, cells were seeded in E-plates in a volume of 100 μL/well at 1.3 × 104 cells/well (M14, A375, RPMI, and MelS) or 1 × 104 cells/well (SK28, M88, and M199). Vemurafenib was added 6 hours later in a volume of 50 μL/well. Cell proliferation was monitored every 15 minutes for up to 3 days by measuring impedance values, expressed as cell index (CI), normalized to the addition of the drug.
Alternatively, cell proliferation was monitored with carboxyfluorescein succinimidyl ester (CFSE) staining. Cells were labeled with CFSE dye and incubated with vemurafenib (10 μmol/L) or DMSO as control. Fluorescence (FITC) was measured at 0, 24, 48, or 72 hours by flow cytometry using a FACSCanto II cytometer.
Protein kinase phosphorylation analysis by NanoPro assay
Cells were lysed in Bicine/CHAPS lysis buffer (20 mmol/L Bicine, 0.6% CHAPS, 20 μL DMSO, and aqueous inhibitor). Lysates were incubated 20 minutes on ice, centrifuged at 14,000 rpm for 17 minutes at 4°C, and protein extracts were kept at –80°C before being loaded in small capillaries for phosphorylation analysis. ERK phosphorylation was measured by the NanoPro 1000 System (ProteinSimple) at the 3P5 proteomic facility (Université Paris Descartes, Institut Cochin, Paris). A total anti-ERK antibody (Cell Signaling Technology) was used to detect the presence of all ERK isoforms and phosphorylated species. The isoelectric point (pI) of each isoform having been validated, the anti-ERK antibody allows the identification of the peaks corresponding to each species. Data analysis was carried out with the Compass software v.2.4.7 (ProteinSimple). Isoforms and phosphorylated proteins were quantitated as peak area.
Peripheral blood mononuclear cell isolation and NK-cell immunoselection
Blood samples from healthy donors were obtained from Etablissement Français du Sang (EFS, Saint Louis Hospital, Paris, France). Peripheral blood mononuclear cells were isolated by Ficoll-Paque PLUS (GE Healthcare) density gradient centrifugation. NK cells were purified by negative immunoselection using the NK Cell Isolation Kit and MS Columns (Miltenyi Biotec), leading to >95% of CD3−CD56+ cells. NK cells (1 × 106/mL) were cultured in RPMI medium supplemented with 10% human serum AB (Biowest) and IL2 or IL15 (both at 10 ng/mL, Miltenyi Biotec) for 6 days.
Flow cytometry analyses
Multicolor flow cytometry assays were performed on a FACSCanto II cytometer (BD Biosciences) and analyzed with Diva 6.0 or FlowJo softwares. Melanoma cells treated with vemurafenib or DMSO were labeled with HLA-A, -B, -C (clone B9.12.1), HLA-DR (Beckman Coulter), HLA-E (Biolegend), CD112 (BD Pharmingen), CD155, MICA, MICB, ULBP1, ULBP2/5/6, and ULBP3 (R&D Systems) antibodies. B7-H6 antibody was provided by F. Vély. The Genesis software (Graz University of Technology) was used to display the results as heatmap, representing relative ratios of vemurafenib and DMSO conditions. In some experiments, melanoma cells were incubated with a broad spectrum matrix metalloproteases (MMP) inhibitor (batimastat, 5 μmol/L) before the staining with specific antibodies.
To evaluate the potential direct consequences of vemurafenib, NK cells were activated by cytokines and simultaneously treated with vemurafenib (10 μmol/L) or DMSO during 6 days. NK cells were then stained for the expression of CD45, CD3, CD56, NKp46, NKp30, NKp44, NKG2D, CD69 (BD Biosciences), KIR2DL1/DS1, KIR2DL2,3/DS2, KIR3DL1/DS1, CD16, NKG2A, CD57, HLA-DR (Beckman Coulter), NKG2C (R&D Systems), and DNAM1 (Miltenyi Biotec).
Additional information about antibodies used in this article is provided in Supplementary Table S1. The phenotype was analyzed on gated CD45+CD3−CD56+ cells.
CD107a degranulation and IFNγ secretion by NK cells
CD107a degranulation was evaluated by FACS analysis on gated CD45+CD3−CD56+ NK cells. NK cells were stimulated with K562 or melanoma cells (1:1 effector/target ratio) in the presence of the CD107a-FITC antibody in round-bottom plates in a final volume of 200 μL of medium containing monensin (a protein transport inhibitor). Melanoma cells were pretreated with vemurafenib (10 μmol/L) or DMSO for 48 hours. After 5 hours of coculture, cells were labeled for 30 minutes at 4°C with CD45, CD3, CD56, and CD16 antibodies, washed, and fixed. Cells were analyzed on a FACSCanto II flow cytometer. Baseline NK-cell CD107a degranulation was determined in the absence of target cells.
Cocultures for 24 hours were also conducted, with 5 × 104 NK cells and 5 × 104 K562 or vemurafenib-pretreated melanoma cells in order to measure the levels of IFNγ secreted by NK cells. Cell-free supernatants were harvested and an ELISA assay was carried out following the manufacturer's instructions (OptiEIA Set Human IFN-γ; BD Biosciences).
NK-cell–mediated lysis assay using the xCELLigence system
Background reading was measured in 96-well E-plates (ACEA Biosciences) containing medium (50 μL/well) and exposed to current flow on the xCELLigence instrument placed in a 37°C incubator. Cells were seeded in a volume of 100 μL/well and their adhesion monitored. Vemurafenib (10 μmol/L) or DMSO was added 6 hours later in a volume of 25 μL/well. Between 20 and 24 hours after the beginning of the experiment, cytokine-activated NK cells were added (at 1:1 ratio, in a volume of 25 μL/well). CI was measured every 15 minutes for 6 hours.
Results are expressed as percentage of lysis determined from normalized CI (nCI, to the addition of NK cells) with the RTCA Software, with the following equation: % of lysis = [nCI (no effector) – nCI (effector)]/nCI (no effector) × 100. Percentages of lysis were calculated taking as reference the CI from targets without NK cells (Supplementary Fig. S3A).
Soluble ligands of NKG2D
Melanoma cells were seeded and treated for 24 or 48 hours with vemurafenib (10 μmol/L), Batimastat (5 μmol/L), or DMSO as control, in 6-well plates in a final volume of 2 mL. Soluble forms of MICA or ULBP2 were quantified in cell-free supernatants by ELISA following the manufacturer's instructions (Human MICA or ULBP2 DuoSet ELISA; R&D Systems).
Statistical analyses
Analyses of significance were performed with the Wilcoxon rank test, a nonparametric paired t test, using the GraphPad Prism 7.0 software (*, P < 0.05 and **, P < 0.01).
Results
Mutation profile of melanoma cell lines
Seven melanoma cell lines bearing V600E or V600K mutations were studied. The genotyping of melanoma cell lines by Next Generation Sequencing revealed differences in their mutational status (Table 1). Four melanoma cell lines had a homozygous BRAFV600E mutation: SK28, M14, M88, and A375. Three cell lines had a heterozygous mutation: RPMI and M199 were BRAFV600E/V mutated, MelS displayed a BRAFV600K/V mutation. All these cell lines harbored various oncogenic mutations, listed in Table 1. All studied cell lines except MelS and A375 had a mutated P53. In particular, SK28, RPMI, and MelS had a PTEN mutation, SK28 cell line had an EGFR mutation, MelS had a KRAS mutation, and M14 cells had a PIK3CA mutation.
Mutational profile of the melanoma cell lines
Vemurafenib treatment reduced cell proliferation in BRAFV600E melanoma cells
We assessed the effect of vemurafenib on melanoma cell proliferation using the xCELLigence system, a device that measures the impedance of adherent cells to indicate cell growth and spreading (6). Cells were allowed to adhere before the addition of different concentrations of vemurafenib (0.4–50 μmol/L). BRAF-mutated melanoma cell lines displayed a dose-dependent growth inhibition after treatment with vemurafenib (Fig. 1A). The cell lines bearing BRAFV600E/E mutations (SK28, M14, M88, and A375) were more sensitive to vemurafenib, leading to a growth inhibition of 50% to 90% at 48 hours, than were the heterozygous MelS and RPMI cell lines, which exhibited a growth inhibition of 10% to 15% in the same conditions. (Fig. 1A). From these results, we identified 10 μmol/L as the concentration of vemurafenib to use for subsequent experiments. In the CFSE assay at 48 and 72 hours of treatment, vemurafenib inhibited fluorescence intensity shading, indicating reduced proliferation. Proliferation of RPMI was less affected by the drug (Fig. 1B). We analyzed three melanoma cell lines bearing a nonmutated BRAF gene (HS944, MelC, and M113) as controls (Supplementary Table S3; Supplementary Fig. S5). Vemurafenib did not alter their proliferation (CFSE assay; Supplementary Fig. S5A).
Effect of vemurafenib on the proliferation of melanoma cells. Homozygous BRAFV600E (top: SK28, M14, M88, and A375) and heterozygous BRAFV600E (bottom: RPMI, M199, and MelS) cell lines. A, Proliferation of cells treated with different vemurafenib concentrations was monitored dynamically by xCELLigence system: vemurafenib was added after the phase of adhesion of 6 hours. B, CFSE staining in cells treated with vemurafenib (10 μmol/L) or DMSO: Histograms show fluorescence at 0, 24, 48, and 72 hours after staining. Representative results of three independent experiments.
ERK activation reduced in melanoma cell lines bearing a BRAFV600E mutation
To study BRAF downstream signaling, we assessed ERK phosphorylation using the NanoPro Assay. This technique allows simultaneous detection of phosphorylated and nonphosphorylated species of each ERK isoform. BRAFV600E cell lines in basal conditions expressed diphosphorylated forms of ERK (ppERK1 and ppERK2; Fig. 2A). ERK1/2 phosphorylation was higher in homozygous BRAFV600E SK28, M14, and A375 cell lines than in heterozygous cell lines. There was more ppERK1 than ppERK2 in all cell lines (Fig. 2B). Vemurafenib treatment (30 minutes) induced loss of phosphorylation of ERK in all cell lines (Fig. 2), regardless of their mutational status. We quantified the percentage of ppERK1 or ppERK2 (Fig. 2B). Patterns of ERK activation in basal conditions varied across cell lines, and cell lines with a homozygous BRAF mutation tended to show higher rates of ERK phosphorylation. Vemurafenib abolished ERK activation in cell lines bearing any BRAFV600E mutation. In addition, vemurafenib increased ERK1/ERK2 ratio in SK28 and M14 cell lines (Supplementary Fig. S1).
Analysis of ERK activation of melanoma cell lines treated with vemurafenib. Representative electrophoretograms of ERK1/2. NanoPro Assay was used to quantitatively measure ERK phosphorylation in unstimulated melanoma cell lines and after treatment with vemurafenib (10 μmol/L, 30 minutes). A, Plots show phosphorylated and nonphosphorylated ERK species as indicated, and protein levels are represented by the peaks. B, Electrophoretograms were quantified for ERK1/2 and phosphorylated ERK1/2 using Compass software to measure the AUC.
Activation and function of IL2-activated NK cells in response to vemurafenib
We determined the effect of vemurafenib on NK-cell activation by cytokines (Fig. 3A; Supplementary Fig. S2) by assessing the expression of specific receptors (Supplementary Table S2) and their degranulation capacities. Immunoselected NK cells were activated in the presence of vemurafenib or DMSO for 6 days. Vemurafenib altered the phenotype of IL2-activated NK cells, reducing the expression of activating receptors (NKG2D, NKp44, and DNAM-1) by 10% to 30%, and increasing NKp46 expression by 10% (Fig. 3A). These alterations did not change the degranulation potential of IL2-activated NK cells against K562 targets (Fig. 3B).
Effect of vemurafenib treatment on NK-cell activation by tumor cells. Phenotypic analyses (n = 7) of NK cells activated for 6 days with IL2 in the presence of vemurafenib (10 μmol/L) or DMSO (A), and their degranulation potential (n = 6) in response to K562 targets (B). C and D, Functional capacities of IL2-activated NK cells (n = 6) stimulated with vemurafenib (10 μmol/L) or vehicle (DMSO) pretreated melanoma targets. C, IL2-activated NK cells were incubated for 5 hours with pretreated melanoma cells, and degranulation of NK cells was assessed by measure of CD107a externalization. D, IL2-activated NK cells were stimulated for 24 hours with pretreated melanoma cells, supernatants were harvested, and secretion of IFNγ by NK cells was determined by ELISA.
We then asked whether vemurafenib treatment of melanoma cells interferes with NK-cell activation. We assessed the degranulation capacities of IL2-activated NK cells stimulated by tumor cells pretreated with vemurafenib or DMSO (Fig. 3C). We found no correlation between degranulation potential of NK cells toward the melanoma cell lines, regardless of hetero- or homozygous mutations. In all but one melanoma cell line (RPMI), treatment with vemurafenib decreased degranulation percentages of IL2-activated NK cells. In addition, we studied the secretion of IFNγ by NK cells stimulated for 24 hours by targets pretreated by vemurafenib or vehicle. We found less IFNγ concentrations produced by IL2-activated NK cells stimulated by any vemurafenib-treated cell line except RPMI and MelS (Fig. 3D). Similar results were obtained with IL15-activated NK cells, although the secreted amounts of IFNγ were lower than those produced by IL2-activated NK cells (Supplementary Fig. S2C and S2D).
Vemurafenib did not directly affect cytokine-activated NK-cell degranulation. However, vemurafenib treatment of melanoma targets led to a decrease in both NK-cell degranulation and secretion of IFNγ.
Lysis of melanoma cells treated by vemurafenib
We determined the susceptibility of BRAF-mutated targets treated with vemurafenib to cytokine-activated NK-cell–mediated lysis with the xCELLigence system (Fig. 4; Supplementary Fig. S3). This device allows monitoring of cell proliferation and death due to the activity of cytotoxic effectors (6). After adhesion, tumor cells were treated overnight with the drug before the addition of activated NK cells. This procedure allowed the evaluation of how vemurafenib modulates the susceptibility of melanoma cells to NK-cell lysis. Cytotoxicity was monitored for 6 hours and percentages of lysis measured. Results were analyzed, comparing cell lysis to cell growth without effectors. The two lysis curves (DMSO and vemurafenib) were compared for each experiment.
Effect of vemurafenib treatment on melanoma cell lysis. Targets were seeded for adhesion and treated with vemurafenib or DMSO overnight before the addition of IL2-activated NK cells (n = 7). CI was monitored by the xCELLigence system for 6 hours. Growth curves were normalized at the time of NK-cell addition. Results are expressed as percentages of melanoma cell lysis calculated in reference to CI without NK cells.
Each melanoma cell line exhibited a unique sensitivity to lysis by cytokine-activated NK cells. We found no correlation with BRAF mutation status. Gathering mean values from seven independent experiments, we found that vemurafenib-treated cells were less susceptible to lysis by IL2-activated NK cells. Lines M14, M199, and MelS were the least susceptible, lines SK28, M88, and A375 were moderately susceptible, and the RPMI line was still efficiently lysed by NK cells (Fig. 4). These data suggest that vemurafenib may modulate the NK/melanoma target interactions and regulate the functional activities of NK cells.
Vemurafenib treatment altered expression of immune ligands
To determine how vemurafenib treatment of melanoma cells interferes with their immunogenicity to NK cells, we compared the expression of NK receptor ligands by melanoma targets treated with DMSO or vemurafenib. We focused here on the ligands of NCR3/NKp30, NKG2D receptors, classical HLA class I, and nonclassical HLA-E molecules involved in melanoma cell lysis by NK cells (representative histograms on Fig. 5A). Melanoma cells expressed high levels of B7H6, a stress-induced molecule from the B7 family that is a ligand of NKp30. B7H6 molecule was highly expressed (50% to 95% of positive cells; Fig. 5B) by all the cell lines except A375 (25%). In four (SK28, M14) or three (M88, RPMI, M119, MelS) independent experiments, expression of B7H6 was higher in vemurafenib-treated cells (Fig. 5B).
Modulation of NK receptor ligands expression by melanoma cell lines in response to vemurafenib. A, FACS histograms from one representative experiment with SK28 cells are depicted. Graphs summarizing the results of 4 to 9 independent experiments showing the expression percentages of B7-H6 (B), MICA and ULBP2 (C), and the expression level (median of fluorescence intensity values) of HLA-A, -B, -C, and -E molecules (D).
Each cell line expressed a unique pattern of NKG2D ligands (NKG2D-L) that was altered by vemurafenib treatment (Fig. 5C). The SK28, M88, M199, and MelS cell lines expressed less MICA (Fig. 5C, left) following treatment by vemurafenib. In addition, ULBP2 was expressed on >40% of cells in SK28, M14, A375, and M199 cell lines; this expression was decreased by vemurafenib on all the cell lines except on M14 cells (Fig. 5C, right).
Following vemurafenib treatment, the median of fluorescence intensity values of HLA-A, -B, and -C molecules were higher in all cell lines but M199 (Fig. 5D, left). In addition, vemurafenib induced increased expression of HLA-E molecules on all cell lines except M88 (Fig. 5D, right). Thus, all seven BRAF-mutated cell lines, which differ in their mutational status, responded similarly to treatment by vemurafenib.
Vemurafenib treatment altered the expression of activating NK receptor ligands (decreased NKG2D-L, increased NKp30 ligand) and increased expression of HLA class I and HLA-E molecules, ligands of inhibitory receptors (KIRs, NKG2A) that decreased NK cell anti-tumor function. The shift in the balance between activating and inhibitory signals caused by vemurafenib likely underlies the changes in NK-cell degranulation and the reduced lysis of vemurafenib-treated cell lines by activated NK cells. In the BRAF-nonmutated cell lines, B7-H6, MICA, or ULBP2 expressions were not affected (Supplementary Fig. S5C), whereas HLA class I and HLA-E MFI values were decreased by 10%–40% (Supplementary Fig. S5D).
Vemurafenib treatment did not affect shedding of NKG2D ligands
To determine whether changes in expression of NK receptor ligands on the surface of vemurafenib-treated cells resulted from increased shedding of ectodomains of cell surface proteins, we measured the soluble NKG2D-L (MICA and ULBP2) in the supernatants of vemurafenib-treated cells. We found less soluble ULBP2 (sULBP2) produced by all cell lines except M88 and MelS (Fig. 6A). As a control, we treated the melanoma cells with Batimastat, an inhibitor of MMP. Batimastat reduced amounts of sULBP2 (Fig. 6A) while increasing amounts of cell surface ULBP2, confirming that ULBP2 expression could be regulated by MMP in melanoma cell lines (Fig. 6B). These results indicate that vemurafenib did not alter membrane expression of NKG2D-L through increased shedding.
Decreased expression of NKG2D ligands does not involve shedding. Secretion of soluble ULBP2 (sULBP2) measured by ELISA in the supernatants of melanoma cell cultures treated for 24 hours with vemurafenib (10 μmol/L) batimastat (MMP inhibitor) or DMSO as control (A, n = 3). Expression of cell surface ULBP2 by melanoma cells treated with batimastat for 48 hours was assessed by flow cytometry (B, one representative of four independent experiments).
Discussion
Mutations affecting the MAPK pathway provide opportunities for inhibitors to target oncogenic pathways in cancer cells. Currently, two inhibitors, vemurafenib and dabrafenib (GSK2118436), are approved for the treatment of malignant melanoma with activating BRAF mutations. Despite initially favorable clinical responses, most patients with BRAFV600E-mutated metastatic melanoma relapse after several months of treatment. Melanoma cells express different stress-induced molecules (6, 10, 14) and present altered expression of HLA class I (15) and HLA-E molecules (16, 17, 18) that are recognized by NK cells, potent cytotoxic antitumor effectors. In this study, we focused on the effect of vemurafenib on the immunogenicity to NK cells of melanoma cell lines with different BRAF mutational statuses. Our results show that targeted therapy alters visibility of the tumor cells to NK cells and interferes with NK-cell activities.
Treatment by vemurafenib affected ligand expression of two NK receptors involved in the lysis of melanoma cells. Vemurafenib treatment downregulated expression of MICA and ULBP2, the NKG2D-L expressed by the cell lines, and increased expression of B7-H6, a ligand of NKp30. In addition, the drug increased expression of HLA class I and HLA-E molecules that inhibit the lysis by NK cells through KIRs and NKG2A receptors. The reduced functional capacities of NK cells and the reduction in NK-mediated cell lysis were probably due to the effect of vemurafenib on activating NK receptor ligands. Vemurafenib did not alter proliferation of cell lines with an intact BRAF gene, nor affect expression of B7-H6, MICA, or ULBP2. However, HLA class I and HLA-E expression decreased. The effects of vemurafenib are thus dependent on mutations in the BRAF gene.
We showed that vemurafenib increased expression of B7-H6, a ligand of the activating NKp30 receptor (19, 20) that is commonly expressed by tumor cells (21). In inflammatory conditions, interaction of NKp30/B7-H6 induces NK-cell activation against tumor cells (22). B7-H6 shRNA treatment dampened sensitization of tumor cells to NK-mediated lysis.
Treatment of tumor cells with various therapeutics, including chemotherapy (cisplatin, 5-fluorouracil), radiotherapy, nonlethal heat shock, and cytokine therapy (TNF), upregulates expression of B7-H6 in tumor cells and enhances tumor sensitivity to NK-cell cytolysis (23). Vemurafenib also triggers ER stress (24) and may upregulate B7H6 expression, in agreement with the present results.
The variable inhibition of surface expression of NKG2D-L in response to vemurafenib is likely due to the mutations unique to each cell line. SK28 cells bear an EGFR-activating mutation that affects regulation of NKG2D-L in response to stress through relocalization of AUF1 proteins that destabilize NKG2D-L (25). In this cell line, BRAF and EGFR mutations likely contribute to the NKG2D-L alteration observed in vemurafenib-treated cells. Alternatively, valproic acid, an HDAC inhibitor, upregulated NKG2D-L through the phosphorylation of ERK (26). Cell regulatory pathways such as ERK and AKT may influence NKG2D-L expression on stressed cells (27, 28); inhibition of ERK phosphorylation decreased these ligands.
Although shedding caused by the proteolytic activity of a disintegrin and metalloproteinase domain–containing protein (ADAM-10 and 17) regulated expression of ULBP2 (29), reduced expression of NKG2D-L by vemurafenib likely involves another mechanism. We showed that the reduction of ULBP2 on the cell surface of vemurafenib-treated melanoma cells was not due to shedding. In contrast, the MMP inhibitor, Batimastat, increases ULBP2 expression by reducing its shedding. We suggest that vemurafenib, which inhibits ERK phosphorylation, may induce MICA and ULBP2 downregulation without affecting their shedding. Thus, vemurafenib would not induce NKG2D internalization.
ADAM-17 controlled expression and shedding of the NKp30 ligand, B7-H6. Reduced shedding of B7-H6 led to higher membrane expression of the molecule on targets (30). We suggest that vemurafenib, through inhibition of ERK phosphorylation, alters ADAM-17 shedding and trafficking, thus regulating expression of NKp30 ligand (31, 32).
BRAFV600E can influence a decreased basal MHC-I expression, and a BRAF inhibitor can potentiate the induction of MHC molecules by IFNγ and IFNα2b (33). Our data showing an upregulation of HLA-A, -B,- C, and -E molecules in response to vemurafenib concur with these previous observations and indicate involvement of the inhibitory pathway in NK/vemurafenib-treated target interactions. Resistance to BRAF inhibitor was associated with HLA class I downregulation (34). These observations suggest that anti-NKG2A (35) or anti-KIRs (36) may be interesting options for releasing NK-cell antitumor capacities. We have shown that vemurafenib treatment modulates two ligands of activating NK receptors, indicating the immunoregulatory effect of BRAF inhibitors. The bystander effect of vemurafenib, by which it inhibited BRAF signaling through pERK inhibition, may be responsible for alteration of the phenotypic profile of melanoma cells and may control their visibility to NK cells.
We found that vemurafenib did not affect the degranulation potential of NK cells in response to cytokines, results agreeing with evidence that vemurafenib did not alter specific antigen recognition, and that cytotoxicity was preserved in T cells (37). However, circulating NK-cell numbers were increased in patients treated with vemurafenib (38). A BRAF inhibitor (PLX4720, a research analogue of vemurafenib) increased pERK1/2, expression of CD69, and IFNγ secretion by IL2-activated NK cells (39).
Currently, treatment of patients bearing a BRAF-mutated melanoma involves inhibitors that target both mutated BRAF and MEK (40). MEK inhibition alters expression of the main NK receptors and the function of cytokine-activated NK cells, but the combined BRAF and MEK inhibitors do not (41).
Therapeutics intended to limit cancer cell growth by acting on MAPK inhibitors should also be considered in terms of their impact on immunosurveillance (42). Indeed, BRAF inhibition augments melanoma antigen expression and maintains T-cell function (43). PLX4720 shows antimetastatic properties in a murine model of BRAF-mutated melanoma that requires host NK cells and perforin (44). Treatment downregulates tumor CCL2 and correlates with reduced tumor growth and increased NK and T-cell infiltration of the tumors (45). MEK inhibitors increase antigen-specific T cells within the tumor, sparing their cytotoxicity. Combinations with anti–PD-L1 treatment show a synergic effect of tumor growth inhibition (46). However, inhibition of BRAF in a murine model of human melanoma is associated with decreased tumor-resident lymphocytes and resistance to CTLA-4 mAb (47).
A better understanding of the off-target efficacy of MAPK inhibition affecting tumor–host interactions would be an important source of strategies aimed at facilitating antitumor immune responses. Our findings indicate that synergy between targeted therapies and NK-cell–based immunotherapy may open new opportunities for the design of clinical trials. Targeting inhibitory pathways in NK-cell–tumor interactions may be complementary to small-molecule inhibitors for the treatment of advanced melanoma. The prospect of combining NK-cell–based immunotherapy with approaches to target the immunosuppressive tumor microenvironment or immune checkpoints, such as KIR blockade, is especially relevant to the treatment of solid tumors (48, 49) and tumors refractory to targeted therapies.
Disclosure of Potential Conflicts of Interest
L. Zitvogel is a consultant/advisory board member for Lytix Biopharma and Transgene. E. Vivier is a consultant/advisory board member for Innate-Pharma. B. Dreno has honoraria from the speakers bureau of BMS, LEO, Novartis, and Roche. M.F. Avril is a consultant/advisory board member for steering committee for trial MO25743: Vemurafenib in Brain Metastasis of Melanoma Patients. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: A. Frazao, M. Colombo, F. Bouquet, A. Toubert, A. Caignard
Development of methodology: A. Frazao, E. Fourmentraux-Neves, M. Messaoudene, M.F. Avril
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Frazao, M. Colombo, S. Rusakiewicz, L. Zitvogel, F. Vely, F. Faure, B. Dreno, E. Pasmant, M.F. Avril
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Frazao, M. Colombo, E. Fourmentraux-Neves, F. Bouquet, A. Savina, E. Pasmant, M.F. Avril
Writing, review, and/or revision of the manuscript: A. Frazao, M. Colombo, M. Messaoudene, B. Dreno, F. Bouquet, A. Savina, A. Toubert, M.F. Avril, A. Caignard
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Benlalam
Study supervision: A. Caignard
Other (provided an essential reagent): E. Vivier
Grant Support
This work was supported by Roche Pharmaceuticals, Fondation ARC, Institut National du Cancer (2011-PLBIO-06, for M. Colombo and E. Fourmentraux-Neves, and PAIR Melanoma 2013-066), Ligue Nationale Contre le Cancer (for M. Messaoudene), and Cancéropôle Ile de France (for A. Frazao).
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.
Acknowledgments
Data acquisition was performed at the cytometry and immuno-biology facility (CYBIO) and the proteomics facility (3P5) of the Cochin Institut. We thank Dalila Darmoul for helpful discussion and for providing batimastat.
Footnotes
Note: Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/).
- Received December 21, 2016.
- Revision received April 10, 2017.
- Accepted May 25, 2017.
- ©2017 American Association for Cancer Research.
References
Volume 5, Issue 7
