Research ArticleCancer

Improved antitumor activity of immunotherapy with BRAF and MEK inhibitors in BRAFV600E melanoma

See allHide authors and affiliations

Science Translational Medicine  18 Mar 2015:
Vol. 7, Issue 279, pp. 279ra41
DOI: 10.1126/scitranslmed.aaa4691

Melanoma’s triple threat

Combination therapy is the favored approach to fight drug-resistant cancer. For BRAF-mutated melanoma, combining a BRAF inhibitor and checkpoint inhibitors was hoped to improve the antitumor response; however, an early clinical trial was stopped because of liver toxicity. Hu-Lieskovan et al. test the addition of MEK [MAPK (mitogen-activated protein kinase) kinase] inhibitors to this combination therapy in an effort to potentiate the MAPK inhibition of BRAF inhibitors while concurrently decreasing the toxicity. They show in a mouse model of BRAFV600E-driven melanoma that triple therapy with BRAF and MEK inhibitors together with adoptive cell transfer (ACT) immunotherapy induced complete tumor regression in a manner consistent with immune activation. In addition, replacing ACT with anti-PD1 in the triple therapy had similar results, supporting the testing of MEK and BRAF inhibitions with various immunotherapies in patients with BRAF-mutated melanoma.

Abstract

Combining immunotherapy and BRAF targeted therapy may result in improved antitumor activity with the high response rates of targeted therapy and the durability of responses with immunotherapy. However, the first clinical trial testing the combination of the BRAF inhibitor vemurafenib and the CTLA4 antibody ipilimumab was terminated early because of substantial liver toxicities. MEK [MAPK (mitogen-activated protein kinase) kinase] inhibitors can potentiate the MAPK inhibition in BRAF mutant cells while potentially alleviating the unwanted paradoxical MAPK activation in BRAF wild-type cells that lead to side effects when using BRAF inhibitors alone. However, there is the concern of MEK inhibitors being detrimental to T cell functionality. Using a mouse model of syngeneic BRAFV600E-driven melanoma, SM1, we tested whether addition of the MEK inhibitor trametinib would enhance the antitumor activity of combined immunotherapy with the BRAF inhibitor dabrafenib. Combination of dabrafenib and trametinib with pmel-1 adoptive cell transfer (ACT) showed complete tumor regression, increased T cell infiltration into tumors, and improved in vivo cytotoxicity. Single-agent dabrafenib increased tumor-associated macrophages and T regulatory cells (Tregs) in tumors, which decreased with the addition of trametinib. The triple combination therapy resulted in increased melanosomal antigen and major histocompatibility complex (MHC) expression and global immune-related gene up-regulation. Given the up-regulation of PD-L1 seen with dabrafenib and/or trametinib combined with antigen-specific ACT, we tested the combination of dabrafenib, trametinib, and anti-PD1 therapy in SM1 tumors, and observed superior antitumor effect. Our findings support the testing of triple combination therapy of BRAF and MEK inhibitors with immunotherapy in patients with BRAFV600E mutant metastatic melanoma.

INTRODUCTION

The recent breakthroughs brought by the clinical use of immune checkpoint inhibition in cancer provide an exciting promise of long-term responses in clinically significant numbers of patients (15). Strategies to extend this low-frequency event to most patients have become the focus of cancer immunotherapy research. In BRAF mutant melanoma, the combination of BRAF inhibitors and immunotherapy has been tested in both preclinical models and clinical trials (69). This is based on the targeting of the BRAFV600E driver mutation, present in about 50% of metastatic melanomas, and the immunosensitization effects of BRAF inhibitors through increased antigen presentation (1012), antigen-specific T cell recognition (10, 13), homing of immune effector cell to the tumors (12, 14, 15), and improved T cell effector functions (6, 16). However, the benefit of this combination in preclinical models has been modest (69), whereas substantial liver toxicity was observed in the first clinical trial combining the BRAF inhibitor vemurafenib and the CTLA4 blocking antibody ipilimumab (17). Both the improved effector function and the toxicities were attributed to the paradoxical activation of the MAPK pathway by vemurafenib in BRAF wild-type cells (18).

MEK [MAPK (mitogen-activated protein kinase) kinase] inhibitors, on the other hand, can potentiate the antitumor effects in the melanoma cells (19) and reduce toxicity associated with BRAF inhibitors (18), given their ability to inhibit MAPK signaling in cells with and without a BRAF mutation (20). In addition, MEK inhibitors have demonstrated potential of immunosensitization by up-regulation of tumor antigen expression and presentation (10, 21), serving as a rational addition to the BRAF inhibitor and immunotherapy combination. However, there is theoretical concern that a MEK inhibitor could dampen immune effector functions, given that in vitro studies have shown impaired T cell proliferation and functions with MEK inhibition (10, 22). Alternatively, when combining with BRAF inhibitors, MEK inhibitors might balance the potential overreacting effector cells to avoid exhaustion and improve the tumor microenvironment by influencing the cytokine production and immunosuppressive cell populations in the tumor microenvironment (20). Using a syngeneic BRAFV600E mutant melanoma mouse model (6), we tested the hypothesis that the addition of a MEK inhibitor would enhance the immunosensitization effects of BRAF inhibition, with increased antitumor activity and decreased toxicity.

RESULTS

Enhanced in vivo antitumor activity with pmel-1 adoptive cell transfer (ACT), dabrafenib, and/or trametinib

We derived a BRAFV600E mutant murine melanoma SM1, syngeneic to fully immunocompetent C57BL/6 mice, from a spontaneously arising melanoma in BRAFV600E transgenic mice (6). Besides the presence of the BRAFV600E transversion, SM1 also has CDKN2A gene deletion and BRAF and MITF gene amplification and is only moderately sensitive to vemurafenib (6). Here, we first confirmed the downstream MAPK pathway inhibition of SM1 after treatment with dabrafenib, trametinib, or the combination in vitro by down-regulated phosphorylated extracellular signal–regulated kinase (pERK) (Fig. 1A). To further explore the drug effects on effector T cells, we treated gp10025-33–activated pmel-1 mouse splenocytes with serial dilutions of dabrafenib, trametinib, or dabrafenib plus trametinib. Western blot analysis at 24 hours of treatment showed paradoxical activation of the MAPK pathway with dabrafenib alone at medium and high concentrations, evidenced by increased pERK (fig. S1A). Trametinib alone or with dabrafenib blocked the MAPK pathway even at low doses. However, in vitro cell viability (MTS) assay with concentration up to 40 μM dabrafenib and 2 μM trametinib did not show any decreased cell viability at 72 hours (fig. S1B).

Fig. 1. Enhanced in vivo antitumor activity with pmel-1 ACT plus dabrafenib (D) and/or trametinib (T).

(A) Western blot analysis of MAPK pathway. SM1 cells were treated with serial dilutions of D, T, or D + T for 1 and 24 hours. L: low dose [D (0.1 μM)/T (0.005 μM)]. M, medium dose [D (5 μM)/T (0.25 μM)]; H, high dose [D (20 μM)/T (1 μM)]. (B) In vivo tumor growth curves with four mice in each group (mean ± SD). SM1-bearing C57BL/6 mice were treated with D (30 mg/kg), T (0.15 mg/kg), or the combination via daily oral gavage, starting when tumors were 3 to 5 mm. (C) Schema of pmel-1 ACT model. C57BL/6 mice had myeloid-depleting total body irradiation (TBI) followed by bone marrow transplant (BMT) and SM1 tumor injections. When tumors reached 3 mm, 3 million gp10025-33 peptide–activated pmel-1 splenocytes (pmel-1 transgenic mice carrying TCR specific for melanoma antigen gp100) were injected. Wild-type C57BL/6 mouse splenocytes activated by CD3 and CD28 were mock ACT controls. Both ACTs were followed by high-dose IL-2 for 3 days. Daily oral gavage of vehicle control (V), D (30 mg/kg), T (0.6 mg/kg), or the combination was started on the day of ACT. i.v., intravenously; s.c., subcutaneously; i.p., intraperitoneally. (D) In vivo SM1 tumor growth curves with three to four mice in each group (mean ± SD), after D, T, and ACT treatments. P < 0.0001 by unpaired t test on day 30, pmel-1 ACT + D + T versus pmel-1 ACT + T, or versus mock ACT + D + T.

We then tested the antitumor effects of dabrafenib, trametinib, and the combination in vivo against established SM1 tumors in syngeneic mice. SM1 tumors established subcutaneously in C57BL/6 mice responded to combination therapy of dabrafenib and trametinib, with a statistically significant difference in growth inhibition when compared with tumors treated with dabrafenib or trametinib alone or vehicle control (Fig. 1B, P = 0.002, unpaired t test, n = 4, mean ± SD). We then tested the combinatorial effect of dabrafenib, trametinib, and immunotherapy using the pmel-1 ACT model, which is based on T cells transgenic for a T cell receptor (TCR) recognizing the murine melanosomal antigen gp100 (23), endogenously expressed by SM1 (Fig. 1C). Myeloid-depleted C57BL/6 mice with bone marrow transplant and established subcutaneous SM1 tumors received ACT of gp10025-33 peptide–activated splenocytes obtained from pmel-1 mice. Wild-type C57BL/6 mouse splenocytes activated nonspecifically by CD3 and CD28 were administered as mock ACT controls. Both ACTs were followed by 3 days of high-dose interleukin-2 (IL-2) injections. In three replicate experiments, the triple combination therapy of dabrafenib and trametinib with pmel-1 ACT provided superior antitumor activity against established SM1 tumors with complete tumor regression, which was not observed with pmel-1 ACT plus either BRAF or MEK inhibitor therapy alone or with mock ACT with both dabrafenib and trametinib (Fig. 1D).

Increased effector T cell homing to the tumors associated with both dabrafenib and trametinib

To analyze the mechanism of improved antitumor activity with the triple combination therapy, we first evaluated the expansion and change in distribution of adoptively transferred cells in vivo. Tumors and spleens were harvested on day 5 after ACT and stained for CD3, CD8, and Thy1.1 (expressed by the pmel-1 mice but not the wild-type C57BL/6 mice). There was a statistically significant increase of CD3+Thy1.1+ (adoptively transferred pmel-1 effector) and CD3+CD8+ (both endogenous and adoptively transferred effector) cells in the tumors treated with dabrafenib, trametinib, or the combination of dabrafenib plus trametinib when compared to vehicle-treated mice (unpaired t test, n = 3, mean ± SD, Fig. 2, A and B). On the other hand, effector cells harvested from the spleen did not show a statistically significant difference in distribution between the treatment groups (Fig. 2A). To analyze the effects on the whole animal, we genetically labeled the adoptively transferred cells with the firefly luciferase transgene to track these cells in vivo using bioluminescence imaging (Fig. 2C, imaged 5 days after ACT). The quantitative analysis of luciferase activity over time in the living mice showed peaked tumor-infiltrating effector T cell 5 days after ACT. SM1 tumors treated with triple combination therapy, pmel-1 ACT plus dabrafenib, or pmel-1 plus trametinib showed a significantly higher accumulation of adoptively transferred effector cells than those tumors treated by pmel-1 ACT alone (Fig. 2D, unpaired t test, n = 4, mean ± SD).

Fig. 2. Increased tumor-infiltrating lymphocytes (TILs) with pmel-1 ACT plus dabrafenib and/or trametinib in SM1 tumors.

(A) Quantification of TILs. Splenocytes and TILs harvested on day 5 after ACT were counted and analyzed by flow cytometry for Thy1.1/CD3/CD8 staining (three mice in each group, mean ± SD). Percentage of effectors (CD3+CD8+ or CD3+Thy1.1+) was shown to be statistically significantly changed by unpaired t test in several subgroups (CD3+CD8+ TILs: P = 0.049, pmel D versus pmel V; P = 0.02, pmel T versus pmel V; P = 0.004, pmel D + T versus pmel V; P = 0.035, pmel D + T versus pmel T; CD3+Thy1.1+: P = 0.03, pmel D versus pmel V; P = 0.02, pmel T versus pmel V; P = 0.006, pmel D + T versus pmel V; P = 0.047, pmel D + T versus pmel T). (B) Representative flow data of percentage of CD3+Thy1.1+ TILs are shown. (C) In vivo bioluminescence imaging (BLI) of adoptively transferred lymphocytes. Pmel-1 transgenic T cells were transduced with a retrovirus–firefly luciferase and used for ACT. Representative figure on day 5 depicts four replicate mice per group. (D) Quantification of BLI of serial images with region of interest (ROI) analysis at the site of tumors (counts per pixel) obtained through day 18 after ACT of luciferase-expressing pmel-1 T cells (four mice per group, mean ± SD). On day 5, P = 0.0009, pmel D versus pmel V; P < 0.0001, pmel T versus pmel V; P < 0.0001, pmel D + T versus pmel V; P = 0.01, pmel T or pmel D + T versus pmel D (unpaired t test, n = 4).

No impaired effector function in vivo associated with trametinib

Given the concern that MEK inhibitors might impair effector T cell function (10), we evaluated the effect of dabrafenib and trametinib on T cell effector function both in vitro and in vivo. When gp10025-33–activated pmel-1 mouse splenocytes were exposed in vitro to increasing concentrations of dabrafenib and trametinib for 72 hours, there was a significant decrease of interferon-γ (IFN-γ)–producing effector cells associated with medium and high concentrations of both dabrafenib and trametinib (Fig. 3A, unpaired t test, n = 3, mean ± SD). To test the effects of the combination therapy in vivo, we harvested tumors and spleens 5 days after ACT and analyzed the activation state of T cells. Adoptively transferred effectors T cells (CD3+Thy1.1+) collected from mice treated with pmel-1 ACT plus trametinib or triple combination therapy did not show statistically significant differences compared to pmel-1 plus vehicle control in the ability to respond to short-term ex vivo restimulation with the gp10025-33 antigen, assessed by IFN-γ secretion (Fig. 3, B and C, unpaired t test, n = 3, mean ± SD). We then tested the direct effect of dabrafenib and trametinib on lymphocyte cytotoxicity in vivo independent of their effects on SM1 tumor cells using an in vivo cytotoxicity assay, where the targets are syngeneic splenocytes devoid of the BRAFV600E mutation and pulsed with gp10025-33 or control peptides (Fig. 3D). Pmel-1 ACT induced potent cytotoxic effects against splenocytes pulsed with the gp10025-33 peptide, but not against the control OVA (ovalbumin) peptide (34% lower gp10025-33 target intensity, P = 0.01 pmel-1 V versus mock V, unpaired t test, n = 3, mean ± SD). The cytotoxicity increased with systemic treatment of pmel-1 ACT plus dabrafenib, but this was not statistically significant (Fig. 3, E and F, and fig. S1D), and there was no difference in cytotoxic activity with pmel-1 plus trametinib and triple combination therapy when compared to pmel-1 ACT plus vehicle control. Therefore, the addition of trametinib to dabrafenib did not change the functionality of adoptively transferred pmel-1 cells in terms of their ability to release immune-stimulating cytokines and intrinsic antigen-specific lytic activity in vivo.

Fig. 3. Dabrafenib, trametinib, or combination treatment impairs effector T cell function in vitro but not in vivo.

(A) In vitro study of cytokine-producing function of effector cells. Gp10025-33–activated pmel-1 mouse splenocytes were treated at serial dilutions of D, T, or D + T for 72 hours. L, low dose [D (0.1 μM)/T (0.005 μM)]; M, medium dose [D (5 μM)/T (0.25 μM)]; H, high dose [D (20 μM)/T (1 μM)]. Cells were analyzed by fluorescence-activated cell sorting (FACS) for CD3/CD8/IFN-γ staining. Bar graphs of percentage of IFN-γ expressing CD3+CD8+ cells are shown (mean ± SD). P = 0.002, D M or D H versus D L; P = 0.045, D H versus D M; P = 0.003, T M or T H versus T L; P = 0.0002, D + T M or D + T H versus D + T L (unpaired t test, n = 3). (B) In vivo effect on cytokine production upon antigen restimulation. SM1 tumor–bearing C57BL/6 mice received pmel-1 ACT with or without D and T. On day 5 after ACT, spleens and TILs were isolated for intracellular IFN-γ staining analyzed by FACS after 5-hour ex vivo exposure to the gp10025-33 peptide. Percentage of IFN-γ expressing CD3+Thy1.1 cells in the spleen and tumor was normalized to Pmel + V (mean ± SD). (C) Gating strategy and representative flow data are shown. (D) Schema of the in vivo cytotoxic T cell assay. C57BL/6 mice received ACT of 5 × 104 pmel-1 splenocytes and daily D, T, D + T, or vehicle via oral gavage. On day 5, mice received an intravenous challenge with CFSE (carboxyfluorescein diacetate succinimidyl ester)–labeled target cells (splenocytes pulsed with gp10025-33 peptide or control OVA peptide). Gp10025-33–pulsed targets were pulsed with 10 times more concentration of CFSE- than OVA-pulsed cells. Ten hours later, splenocytes were harvested and analyzed by FACS. (E) Bar graph representation of the in vivo cytotoxicity study result (mean ± SD). P = 0.01, pmel V versus mock V (34% down, unpaired t test, n = 3). (F). Representative flow data are shown.

Improved tumor microenvironment when trametinib was combined with dabrafenib and ACT

To evaluate the effect of dabrafenib, trametinib, and the combination on other cellular components of the tumor microenvironment, we harvested spleens and tumors 5 days after ACT and studied the cell populations by multiplex FACS. We first analyzed the myeloid-derived suppressor cells (MDSCs), a heterogeneous group of cells that induce tumor-associated immune suppression. MDSCs (Gr+CD11b+) were present at a low level in SM1 tumors (<5%), and their level increased with antigen-specific ACT (fig. S1C). Pmel-1 ACT plus dabrafenib significantly increased MDSCs in the tumors when compared to the pmel-1 ACT plus vehicle control group (P = 0.02, unpaired t test, n = 3, mean ± SD), with a nonsignificant trend of decreased MDSCs in the spleen. Trametinib alone or dabrafenib plus trametinib with pmel-1 ACT did not change MDSCs in the tumors or spleens when compared to pmel-1 ACT plus vehicle. MDSCs consist of two major subsets: cells with granulocytic phenotype that express Ly6G marker (PMN-MDSC, Ly6CLoLy6GHiCD11b+), and cells with monocytic phenotype expressing Ly6C marker (MO-MDSC, Ly6CHiLy6GLoCD11b+) (fig. S1E). These two subsets of MDSCs have distinct functions in infection, autoimmune diseases, and cancer (24). We examined these two subsets of MDSCs, presented as percentage of CD11b+ cells. In Fig. 4 (A and B), there was a significant shift of MDSC subsets in the tumors toward increased MO-MDSCs and decreased PMN-MDSCs, associated with dabrafenib, trametinib, and combination treatments. Whereas in the spleen, other than a decreased percentage of MO-MDSC with pmel-1 ACT plus dabrafenib–treated mice versus pmel-1 ACT plus vehicle control (P = 0.06, unpaired t test, n = 3), there was no significant change among the different treatment groups. We then analyzed mature tumor-associated macrophages (TAMs; F4/80+CD11b+). Both pmel-1 ACT plus dabrafenib and triple combination therapy significantly increased macrophages in the tumors (Fig. 4, C and E), but to a lesser extent with triple combination, whereas no change was seen with pmel-1 ACT plus trametinib. There was no significant change in macrophages in the spleen among the different treatment groups. Analysis of another immunosuppressive cell population, the T regulatory cells (Tregs; CD4+CD25+FOXp3+), showed a significantly increased percentage in tumors with pmel-1 ACT plus dabrafenib treatment (fig. S1F, P = 0.002, unpaired t test, n = 3) but no other significant change with the other combination therapies in both tumor and spleen (Fig. 4, D and E). These results indicate that dabrafenib, when combined with pmel-1 ACT and IL-2, increases macrophage and Tregs infiltration in the tumor microenvironment, thus providing a potential mechanism for the suboptimal antitumor effect with this combination. This effect on macrophage and Tregs can be overcome by the addition of trametinib. Moreover, both dabrafenib and trametinib can shift the ratio of MDSC subsets from PMN-MDSC to predominantly MO-MDSC.

Fig. 4. Dabrafenib and trametinib changed the cellular components of the tumor microenvironment.

On day 5 after ACT, spleens and tumors were isolated and stained with fluorescent-labeled antibodies and then analyzed by FACS, with three mice in each group (mean ± SD). (A) MO-MDSC (CD11b+Ly6CHiLy6GLo) presented as percentage of CD11b+ cells. *P = 0.06, pmel D versus pmel V in spleen; P = 0.009, pmel V versus mock V in tumor (unpaired t test, n = 3). (B) PMN-MDSC (CD11b+Ly6CLowLy6GHi) presented as percentage of CD11b+ cells. *P = 0.002, mock D + T versus mock V in tumor (unpaired t test, n = 3). (C) Analysis of macrophages (F4/80+CD11b+). *P = 0.04, pmel D versus pmel V; P = 0.002, pmel D + T versus pmel V, both in tumors (unpaired t test, n = 3). (D) Analysis of Tregs (CD4+CD25+FOXp3+). *P = 0.002, pmel D versus pmel V in tumors (unpaired t test, n = 3). (E) Gating strategy and representative FACS plots in tumors.

Increased immune-related gene expression by both dabrafenib and trametinib treatments

To better understand the impact of dabrafenib and trametinib on the tumor microenvironment, we compared the gene expression profiling of SM1 tumors by microarray analysis after a 5-day treatment with dabrafenib, trametinib, or the combination with pmel-1 or mock ACT. Two to three replicates were prepared per treatment group after the samples had passed quality control by gel electrophoresis (fig. S2A). The expression level of each gene was averaged across samples and used for further analysis. As shown in Fig. 5A by principal components analysis (PCA) and in fig. S2B by hierarchical clustering of global gene expression, the biological replicates cluster together closely and separate from each other according to the different treatments. Clustering of immune-related genes obtained from the gene ontology (GO) consortium under the GO term of immune system process (www.geneontology.org; GO:0002376) showed three different patterns of gene expression changes (Fig. 5B). Cluster A genes are up-regulated after treatment with dabrafenib, trametinib, or the combination in mice receiving either mock ACT or pmel-1 ACT, and included melanoma antigens and many MAPK pathway genes. Also among them are CD274 (PD-L1) and CSF-1R. Cluster B genes are up-regulated by dabrafenib, trametinib, or the combination when combined with antigen-specific pmel-1 ACT only, but not with the mock ACT. Many of these genes are major histocompatibility complex (MHC) molecules, cytokines, and chemokines or their receptors (table S2). CD8, granzyme B, and IFN-γ are within this group. CD83, a dendritic cell maturation marker (25), CD86, a ligand expressed on antigen-presenting cells that binds to CD28 and CTLA4 (26), and CSF-1, are all in this gene cluster. Finally, cluster C genes are down-regulated by dabrafenib, trametinib, or the combination in mice that received either mock ACT or pmel-1 ACT. Again, this cluster included many MAPK pathway genes, as well as VEGF (vascular endothelial growth factor), a previously reported target of BRAF and MEK inhibition (27, 28). Another interesting gene in this cluster is CD276 (B7-H3), which was reported recently to be associated with advanced stage of melanoma progression (29). Further clustering of samples according to chemokines and their receptors showed a general trend of increased expression of these genes in the dabrafenib-, trametinib-, or combination-treated tumors, especially with triple combination therapy (Fig. 5C). Expression of some genes previously reported to be regulated by BRAF inhibitors or mentioned above is shown in fig. S2C.

Fig. 5. Microarray analysis of tumors treated by dabrafenib, trametinib, or the combination of dabrafenib and trametinib with pmel-1 ACT or mock ACT.

On day 5 after ACT, tumors were isolated and snap-frozen immediately (two to three mice in each group). RNA isolation was done after all samples were collected. (A) PCA of gene expression profile of the tested samples. (B) Clustering of immune-related genes with analysis of variance (ANOVA) filter, P < 0.05. Gene names in individual clusters are listed in tables S1 to S3. (C) Clustering of chemokines and their receptors. (D) Clustering of MDAs and MHC class I and II molecules.

Increased tumor antigen and MHC expression in tumors treated by triple combination

One mechanism wherein tumors can evade the immune attack is through decreased tumor antigen or MHC expression. Analysis of the microarray data showed that both melanoma antigen and MHC expression are low in mock ACT plus vehicle and pmel-1 ACT plus vehicle (Fig. 5D). Dabrafenib and trametinib significantly increased melanoma antigen expression in both pmel-1 ACT– and mock ACT–treated tumors, with the highest expression seen after triple combination treatment. However, the up-regulation of MHC molecules was restricted to the antigen-specific pmel-1 ACT–treated tumors with dabrafenib or/and trametinib, but not with mock ACT plus dabrafenib and trametinib, indicating that tumor-specific effector cells are important in the mechanism of MHC up-regulation.

Up-regulation of PD-L1 and enhanced in vivo antitumor activity with dabrafenib and/or trametinib with PD1 blockade

Activated T cells express PD1, which can bind to PD-L1 ligand up-regulated on tumor cells, and inhibit T cell effector functions. Recently, PD1 and PD-L1 were found to be increased in melanoma tumor samples from patients treated with BRAF inhibitors (12), and addition of a MEK inhibitor suppressed the production of PD-L1 by in vitro study of melanoma cell lines (30). The up-regulation of IFN-γ, granzyme B expression, and PD1 in cluster B indicated increased effector T cell activation and function with dabrafenib and trametinib (Fig. 6A). However, the up-regulation of PD-L1 suggested an adaptive immune resistance mechanism induced by the presence of effector T cells (Fig. 6A). Flow cytometry analysis of PD-L1 expression of SM1 tumors after 5 days of treatment was consistent with the microarray gene expression data (Fig. 6B), whereas no significant change was observed in the spleen samples. To test whether the increased IFN-γ in the tumor milieu is sufficient for the up-regulation of PD-L1, we treated SM1 cells with increasing concentrations of IFN-γ and harvested cells for flow cytometry after 18 hours. The result showed that IFN-γ (1 μg/ml) up-regulated PD-L1 on the surface of SM1 cells by more than 10-fold, similar to the B16-positive control cells (Fig. 6C). This up-regulation of PD-L1 provided a rationale for the combination of PD1 blockade therapy with dabrafenib and trametinib. Immunocompetent C57BL/6 mice with established subcutaneous SM1 tumors received PD1 antibody or isotope control (200 μg) via intraperitoneal injection every 5 days, starting when the tumors reached 4 to 6 mm2. Consistent with a previous report (8), SM1 is innately resistant to PD1 antibody therapy alone. In three replicate experiments, the combined therapy of dabrafenib, trametinib, and anti-PD1 provided superior antitumor activity against established SM1 tumors compared with anti-PD1 plus either therapy alone or isotope control with both dabrafenib and trametinib (Fig. 6D).

Fig. 6. Up-regulation of PD-L1 and triple combination of dabrafenib, trametinib, and PD1 blockade is superior in antitumor effect against SM1.

(A) Heat map representation of CD8, granzyme B, IFN-γ, PD1, and PD-L1 gene expression from microarray data (PD-L1: P = 0.01, mock D + T versus mock V; P = 0.004, pmel T versus pmel V; P = 0.004, pmel D + T versus pmel V; P = 0.03, pmel T versus pmel D + T; unpaired t test, n = 3). (B) Percentage of PD-L1–expressing cells in the spleen and tumors 5 days after ACT and drug treatments started; three mice in each group (mean ± SD). P = 0.006, mock D + T versus mock V; P = 0.04, pmel D versus pmel V; P = 0.007, pmel T versus pmel V; P = 0.001, pmel D + T versus pmel V. (C) Expression of PD-L1 on SM1 after 18 hours of stimulation with IFN-γ at different concentrations. B16 cells served as a positive control. (D) In vivo SM1 tumor growth curves after D, T, and anti-PD1 treatments; four mice in each group (mean ± SD). SM1 tumor–bearing C57BL/6 mice received anti-PD1 (Merck DX400; 200 μg) via intraperitoneal injection every 4 days, starting when tumors reached 3 to 5 mm. Daily oral gavage of vehicle control (V), D (30 mg/kg), T (0.6 mg/kg), or the combination was started on the same day as anti-PD1.

DISCUSSION

It has been previously reported that MEK inhibitors might be detrimental to T cell responses to cancer mainly on the basis of in vitro studies (10, 22). However, by using an immunocompetent mouse model of BRAFV600E mutant melanoma, we demonstrate that the addition of the MEK inhibitor trametinib significantly improves the antitumor effect of the BRAF inhibitor dabrafenib and two modes of immunotherapy, ACT and PD1 blockade, via improved effector T cell homing to the tumors, preserved effector function, increased tumor antigen and MHC expression, cytokine release, and attenuated immunosuppressive cells up-regulated by the BRAF inhibitor in the tumor microenvironment.

Increased numbers of TILs have been reported in biopsies of patients treated with BRAF inhibitors (12, 14, 15), with an increase in clonality after BRAF inhibition and a better response in those patients who had a high proportion of preexisting dominant TCR clones (31), suggesting that the T cell infiltration may be an antigen-driven recruitment into regressing tumors. In a xenograft model where a BRAF mutant human melanoma cell line was transduced with gp100 and H-2D to allow recognition by gp100-specific pmel-1 mouse T cells, treatment with the BRAF inhibitor vemurafenib significantly increased the tumor infiltration and enhanced the antitumor activity of adoptively transferred T cells in vivo (27). This increased TIL infiltration was thought to be primarily mediated by decreased VEGF production by tumors via C-myc. Analysis of human melanoma biopsies before and during BRAF inhibitor treatment confirmed the down-regulation of VEGF. Another syngeneic BRAF mutant melanoma model with PTEN−/− background tested the combination of vemurafenib and PD1 and/or CTLA4 blockade (9), and showed significantly increased T cell infiltration with vemurafenib alone. On the contrary, in an inducible BRAF mutant melanoma model, there was decreased TIL infiltration after BRAF inhibition (7). In our SM1 model, a previous study combining vemurafenib and pmel-1 ACT did not show increased TIL infiltration but did show improved effector function, likely through paradoxical activation of the MAPK pathway in T cells by vemurafenib (6, 16). With dabrafenib, titrated to higher concentrations to optimize antitumor activity, we observed both increased tumor T cell infiltration and improved functions. This is possibly due the different potency and concentration of the two BRAF inhibitors (we had used a lower dose with vemurafenib) (6). We also observed decreased VEGF expression by microarray in the tumors treated with dabrafenib. Trametinib increased T cell homing to tumors, evidenced by in vivo imaging, ex vivo single tumor cell phenotyping, and microarray analysis. Evaluation of in vivo effector cell function by both cytokine-releasing capacity and cytotoxic activity in vivo showed preserved effector functions after treatment with single-agent trametinib or dabrafenib plus trametinib.

One of the suggested mechanisms by which BRAF inhibitors may sensitize tumor to the immune system is through up-regulation of melanocyte differentiation antigens (MDAs) and MHC expression in BRAF mutant melanoma (10-12). MEK inhibitors, on the other hand, can increase MDA expression in both BRAF mutant and wild-type melanoma cells (10, 21). This should result in improved antigen-specific T cell recognition (10, 13). It has been shown that MDA expression was significantly decreased in patients at the time of progression on BRAF inhibitors and restored when subsequent combined BRAF and MEK inhibitors were given (12). Consistent with the previous reports, our data suggest increased MDA and MHC expression in tumors treated by the combination therapy. The up-regulation of MDAs is a drug effect, also seen in dabrafenib and trametinib combination with mock ACT. On the other hand, up-regulation of MHC is specific to the combination of dabrafenib, trametinib, or dabrafenib plus trametinib with pmel-1, but not with mock ACT, indicating that antigen-specific effector activation is crucial for this regulation. Analysis of immune-related genes in the microarray data indicated other genes that are regulated in similar fashion: up- or down-regulated by drugs in both pmel-1 and mock ACT, or up-regulated by drugs in antigen-specific ACT only. Clustering of chemokines and their receptors showed overall increased expression of many of these genes in tumors treated with the triple combination.

Despite the theoretical promise of increased tumor-infiltrating T cells with improved function and increased tumor MDA and MHC expression, the combination of dabrafenib and pmel-1 ACT had only modest antitumor effects, with a small but statistically significant difference from mock ACT and pmel-1 ACT with vehicle control. This raised the possibility that the immunosuppressive cells (MDSCs, TAMs, and Tregs) in the tumor microenvironment might inhibit effector T cell function. One of the main factors that negatively regulate the immune system is MDSCs, a heterogeneous population of immature myeloid cells (CD11b+Gr1+ in mice) that are significantly expanded in patients with cancer and have been shown to correlate negatively with prognosis and overall survival (32). MDSCs have been shown to not only suppress immune responses but also promote tumor growth and expansion in different tumor types (3337). We observed an interesting shift of MDSC subtypes in the SM1 tumors, from the PMN-MDSC subset to the MO-MDSC subset, associated with dabrafenib, trametinib, or combination treatment. This was correlated with significantly up-regulated TAMs (F4/80+CD11b+) and Tregs (CD4+CD25+FoxP3+) after treatment with dabrafenib combined with pmel-1 ACT, suggesting a potential immune evasion pathway. Microarray data showed concurrent up-regulation of CSF-1, CSF-1R, and the dendritic cell maturation marker CD83. Trametinib, however, did not increase TAMs or Tregs when combined with pmel-1 ACT, and further attenuated the effect by dabrafenib when combined with both dabrafenib and pmel-1 ACT. This might have accounted for the significantly different antitumor effects of triple combination therapy compared to dabrafenib plus pmel-1 ACT.

Activated T cells express PD1, which in turn binds to its ligand PD-L1 and negatively regulates TCR signaling (3840). Increased T cell exhaustion markers, including TIM3, PD1, and PD-L1, were noted in tumor samples from patients treated with BRAF inhibitors, suggesting a potential resistance mechanism (12). One in vitro study of melanoma cell lines showed increased PD-L1 expression by BRAF inhibitor–resistant melanoma cells, mediated by c-JUN and STAT3 signaling, and addition of a MEK inhibitor suppressed the expression of PD-L1 (30). Our study showed improved effector activity manifested by increased IFN-γ and granzyme B expression, which coincides with the up-regulation of PD1 and PD-L1, unique to the tumor milieu. We also showed that IFN-γ is sufficient for the up-regulation of PD-L1 on SM1, consistent with published reports (39), which also explained why MEK inhibitor–treated tumors also had elevated PD-L1 expression, and again stressed the importance of the in vivo environment to study immune responses. Given that CD8 T cells are the effectors of PD1 blockade, and the up-regulation of PD-L1 seen with both dabrafenib and trametinib treatments when combined with antigen-specific ACT, we tested the triple combination of dabrafenib, trametinib, and anti-PD1 therapy, and observed significant synergy of this combination to inhibit SM1 tumor growth. This antitumor activity was most significant with the triple therapy, but was also notable for the combination of either dabrafenib or trametinib plus anti-PD1 therapies.

Our results are limited to preclinical models and need to be validated in clinical trials. Several phase 1 clinical trials are ongoing, combining BRAF and MEK inhibitors with immunotherapies such as anti-CTLA4, anti-PD1, anti–PD-L1, and ACT (20), and our data suggest that the presence of a MEK inhibitor will improve the effects of the combined therapy as opposed to the concern of limiting T cell responses to cancer.

MATERIALS AND METHODS

Study design

The primary research objective was to evaluate combinatorial strategies of BRAF inhibitor, MEK inhibitor, and ACT or PD1 blockade. The overall study design was a series of controlled laboratory experiments in mice, as described in the sections below. In all experiments, animals were assigned to various experimental groups in random. The experiments were replicated two to three times as noted. For the experiments reporting isolation of TILs, three mice per group were used for each experiment, with two to three replicates. All outliers were included in the data analysis.

Mice, cell lines, and reagents

C57BL/6 mice (Thy1.2, The Jackson Laboratory) and pmel-1 (Thy1.1) transgenic mice were bred and kept under defined-flora, pathogen-free conditions at the Association for the Assessment and Accreditation of Laboratory Animal Care–approved animal facility of the Division of Experimental Radiation Oncology, University of California, Los Angeles (UCLA), and used under the UCLA Animal Research Committee protocol #2004-159. The SM1 murine melanoma was generated from a spontaneously arising tumor in BRAFV600E mutant transgenic mice as previously described (6). The tumor was minced and implanted into C57BL/6 mice for in vivo experiments. Part of the minced tumor was plated under tissue culture conditions as SM1 cell line, maintained in RPMI (Mediatech) with 10% fetal calf serum (Omega Scientific), 2 mM l-glutamine (Invitrogen), and 1% (v/v) penicillin, streptomycin, and amphotericin (Omega Scientific). Dabrafenib and trametinib were obtained under a material transfer agreement (MTA) with GlaxoSmithKline (GSK). Dabrafenib and trametinib were dissolved in dimethyl sulfoxide (DMSO; Fisher Scientific) and used for in vitro studies. For in vivo studies, dabrafenib and trametinib were suspended in an aqueous mixture of 0.5% hydroxypropyl methyl cellulose (HPMC) and 0.2% Tween 80 (Sigma-Aldrich). One hundred microliters of the suspended drug was administered by daily oral gavage into mice at 30 mg/kg of dabrafenib and/or 0.6 mg/kg of trametinib when tumors reached 5 mm in diameter. Mouse PD1 antibody (DX400) was obtained under an MTA with Merck.

Cell viability assays

SM1 cells, naïve C57BL/6 splenocytes, or activated pmel-1 splenocytes were seeded in 96-well flat-bottom plates (5000 cells per well) with 100 ml of 10% fetal calf serum medium and incubated for 24 hours. Serial dilutions of dabrafenib, trametinib, or DMSO vehicle control, in culture medium, were added to each well in triplicate and analyzed following the MTS assay (Promega).

Western blotting

Cells were washed with ice-cold phosphate-buffered saline and then lysed using a lysis buffer containing 10 mM tris (pH 7.4), 100 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM NaF, 20 mM Na4P2O7, 2 mM Na3VO4, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100, 10% glycerol, leupeptin (10 μg/ml), aprotinin (60 μg/ml), and 1 mM phenylmethanesulfonyl fluoride. Equal amounts of protein extracts were separated by using 8 or 10% SDS–polyacrylamide gel electrophoresis and then transferred to a polyvinylidene difluoride membrane (Bio-Rad Laboratories Inc.). After blocking for 1 hour in a tris-buffered saline containing 0.1% Tween 20 and 5% bovine serum albumin, the membrane was probed with various primary antibodies, followed by secondary antibodies conjugated to horseradish peroxidase. The immunoreactivity was revealed by use of a Pierce ECL kit (Thermo Scientific), and the densities of the protein bands were quantified by ImageJ software. Primary antibodies included pERK Thr204/205, ERK, pAKT Ser473, AKT, and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (Cell Signaling Technology).

Pmel-1 ACT in vivo model

C57BL/6 mice were treated with lymphoid-depleting (500 cGy) or myeloid-depleting (900 cGy) total body irradiation followed by bone marrow transplant and subcutaneous SM1 tumor injection, and then received 5 × 106 gp10025-33 peptide–activated pmel-1 splenocytes intravenously when tumors reached 3 to 5 mm in diameter as previously described (29, 30) and daily intraperitoneal administration of 50,000 IU of IL-2 for 3 days. Activated splenocytes from wild-type C57BL/6 mice were controls. BRAF inhibitor dabrafenib, MEK inhibitor trametinib, vehicle, or the combination of dabrafenib and trametinib was given daily by oral gavage from the day of ACT. Tumors were followed by caliper measurements three times per week.

Flow cytometry analysis

SM1 tumors harvested from mice were digested with collagenase (Sigma-Aldrich). Splenocytes and cells obtained from digested SM1 tumors were stained with antibodies to CD3 BV605 (clone 17A2), Ly6C FITC (fluorescein isothiocyanate) (clone AL-21), PD-L1/CD274 PE (phycoerythrin) (clone MIH5) (Becton Dickinson Biosciences), CD8a BV421 (clone 53-6.7) (BioLegend), Ly6G (Gr1) PerCP 5.5 (clone RB6-8C5), CD11b APC (allophycocyanin) (clone M1/70), F4/80 Pacific Blue/eFluor 450 (clone BM8), CD25 APC (PC61.5), and CD4 FITC (RM4-5) (eBioscience), and analyzed with LSR II or FACSCalibur flow cytometers (Becton Dickinson Biosciences), followed by analysis using FlowJo software (FlowJo LLC) as previously described (30). Intracellular staining of IFN-γ was done as previously described (30). Intracellular staining of Foxp3 PE (FJK-16s) (eBioscience) was done according to the manufacturer’s recommendations.

In vivo cytotoxicity assay

The assay was conducted as previously described (30). In brief, splenocytes from naïve wild-type C57BL/6 mice were pulsed with gp10025-33 peptide (50 mg/ml) or the same amount of control OVA257-264 peptide. After 1 hour of incubation, gp10025-33-pulsed wild-type splenocytes were labeled with 6 nM CFSE for 10 min at 37°C, whereas control OVA257-264–pulsed splenocytes were differentially labeled with a 10-fold dilution of CFSE (0.6 nM). Cells were injected intravenously into experimental mice at 5 days after pmel-1 ACT. After 10 hours, five mice per group were sacrificed, and their spleens were examined for the presence of CFSE-labeled cells. Percent cytotoxic activity was calculated as number of live gp10025-33–pulsed splenocytes divided by the number of live OVA257-264–pulsed splenocytes, which were distinguished on the basis of the 10-fold difference in CFSE fluorescence by flow cytometry.

Bioluminescence imaging

Pmel-1 splenocytes were retrovirally transduced to express firefly luciferase as previously described (29) and used for ACT. Bioluminescence imaging was carried out with a Xenogen IVIS 200 Imaging System (Xenogen/Caliper Life Sciences) as previously described (22, 23).

Microarray data generation and analysis

Total RNAs were extracted using the RNeasy Micro Kit (Qiagen) from SM1 tumors. Complementary DNAs were generated, fragmented, biotinylated, and hybridized to the GeneChip Mouse 430 V2 Arrays (Affymetrix). The arrays were washed and stained on a GeneChip Fluidics Station 450 (Affymetrix); scanning was carried out with the GeneChip Scanner 3000 7G (Affymetrix); and image analysis was done with the GeneChip Command Console Scan Control (Affymetrix). Microarray analyses were performed in the R statistical programming environment, and Bioconductor suite of packages was used (41). Expression data were normalized, background-corrected, and summarized using the robust multi-array average (RMA) algorithm implemented in the R “affy” package (42). Hierarchical clustering was performed using the Euclidean distance as the similarity metric with average linkage clustering. Clustering results were visualized by heat maps generated using the R “NMF” package (43).

Statistical analysis

Descriptive statistics such as number of observations, mean values, and SD were reported and presented graphically for quantitative measurements. Normality assumption was checked for outcomes before statistical testing. For measurements such as tumor volume, percentage of TILs, quantified imaging data, and cytokine expression levels, pairwise comparisons between treatment groups were performed by unpaired t tests. All hypothesis testing was two-sided, and a significance threshold of 0.05 for P value was used. Analyses were carried out using GraphPad Prism (version 6) software (GraphPad Software).

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/7/279/279ra41/DC1

Fig. S1. In vitro study of effects of dabrafenib and trametinib on effector T cell and gating strategies.

Fig. S2. Microarray analysis quality control, all gene clustering, and expression of interested genes.

Fig. S3. Source data of Western blots in Fig. 1A.

Table S1. Immune A genes.

Table S2. Immune B genes.

Table S3. Immune C genes.

Table S4. Source data of tumor growth curves in Figs. 1B, 1D, and 6D.

REFERENCES AND NOTES

Acknowledgments: We are grateful to X. Wang from UCLA Statistics Core for her assistance in statistical analysis. We want to thank L. Liu, T. Gilmer, L. Nakamura, and J. Offord from GSK for their assistance regarding dabrafenib and trametinib. We also want to thank N. Restifo, Surgery Branch, National Cancer Institute, Bethesda, MD, for providing pmel-1 (Thy1.1) transgenic mice, and R. Prins from UCLA for maintaining the pmel-1 mice. Funding: Funded by NIH grants P01 CA168585 and P50 CA086306, the Dr. Robert Vigen Memorial Fund, the Ressler Family Foundation, the Wesley Coyle Memorial Fund, and the Garcia-Corsini Family Fund (to A.R.). S.H.-L. was supported by NIH grant T32 CA09297, ASCO (American Society of Clinical Oncology) Young Investigator Award, and Tower Foundation Research Grant. S.M. was supported by UCLA Final Year Dissertation Fellowship (2014–2015). B.H.M. was supported by the Rio Ortega Scholarship from the Hospital 12 de Octubre, Madrid, Spain. L.R. was supported by the V Foundation-Gil Nickel Family Endowed Fellowship in Melanoma Research and a grant from the Spanish Society of Medical Oncology (SEOM) for Translational Research in Reference Centers. T.G.G. is supported by the National Cancer Institute/NIH (P01 CA168585 and R21 CA169993), an American Cancer Society Research Scholar Award (RSG-12-257-01-TBE), a Melanoma Research Alliance Established Investigator Award (20120279), the National Center for Advancing Translational Sciences UCLA CTSI (Clinical and Translational Science Institute) Grant UL1TR000124, and a CONCERN Foundation CONquer CanCER Now Award. J.T. is supported by the NIH Ruth L. Kirschstein Institutional National Research Service Award #T32-CA009120. Flow cytometry was performed in the UCLA Jonsson Comprehensive Cancer Center (JCCC) and Center for AIDS Research Flow Cytometry Core Facility that is supported by NIH awards CA-16042 and AI-28697, and by the JCCC, the UCLA AIDS Institute, and the David Geffen School of Medicine at UCLA. Author contributions: S.H.-L., S.M., B.H.M., L.R., L.G., E.M.P., R.C.K., and B.C.-A. performed the experiments. J.T. and T.G.G. analyzed the gene expression profiling. S.H.-L. and A.R. designed the experiments and wrote the manuscript. Competing interests: E.M.P. is an employee of Merck. The other authors declare that they have no competing interests. Data and materials availability: Dabrafenib and trametinib were obtained through an MTA with GSK. DX400 murine PD1 antibody was obtained through an MTA with Merck. The gene expression profiling is available at www.ncbi.nlm.nih.gov/geo/ (accession number: GSE64102).
View Abstract

Navigate This Article