Targeted therapies elicit seemingly paradoxical and poorly understood effects on tumor immunity. Here, we show that the MEK inhibitor trametinib abrogates cytokine-driven expansion of monocytic myeloid-derived suppressor cells (mMDSC) from human or mouse myeloid progenitors. MEK inhibition also reduced the production of the mMDSC chemotactic factor osteopontin by tumor cells. Together, these effects reduced mMDSC accumulation in tumor-bearing hosts, limiting the outgrowth of KRas–driven breast tumors, even though trametinib largely failed to directly inhibit tumor cell proliferation. Accordingly, trametinib impeded tumor progression in vivo through a mechanism requiring CD8+ T cells, which was paradoxical given the drug's reported ability to inhibit effector lymphocytes. Confirming our observations, adoptive transfer of tumor-derived mMDSC reversed the ability of trametinib to control tumor growth. Overall, our work showed how the effects of trametinib on immune cells could partly explain its effectiveness, distinct from its activity on tumor cells themselves. More broadly, by providing a more incisive view into how MEK inhibitors may act against tumors, our findings expand their potential uses to generally block mMDSC expansion, which occurs widely in cancers to drive their growth and progression. Cancer Res; 76(21); 6253–65. ©2016 AACR.
The main purpose of targeted kinase inhibitor development for cancer therapy has been to precisely block oncogenic signaling in tumor cells. However, many other cells in the tumor microenvironment (TME), including immune cells, rely upon the same signaling pathways for normal activity. For instance, T cells, critical for controlling the growth of immunogenic tumors (1–4), require the Ras–MAPK signaling cascade following antigen activation for proliferation and effector function (5). The attractiveness of targeting these pathways in tumor cells has led to concerns that an on-target side effect would disrupt beneficial antitumor immunity (6, 7).
Indeed, proper T-cell activation and proliferation is impaired by pharmacologic inhibition of MEK signaling, both with the FDA-approved drug trametinib (6, 7) and other compounds (8). In fact, only mutation-specific BRAF inhibitors elicit some immunostimulatory effects, which can be enhanced by macrophage depletion (9). However, although trametinib impairs T-cell function in vitro, it does not limit the effectiveness of either adoptive cell therapy (10) or checkpoint blockade (11) in mouse models and can actually synergize with these immunotherapies. Although signaling from common gamma chain cytokines reduces sensitivity of T cells to trametinib in vivo (12), another potential explanation is that trametinib acts on tumor and stromal cells in the TME that, overall, ameliorate its immunomodulatory effects.
The occurrence of neoplasia results in a chronic inflammatory response that promotes the pathologic expansion and recruitment of myeloid-derived suppressor cells (MDSC; refs. 13–16). Immune suppression by MDSCs is a critical factor in the ability of tumor cells to avoid adaptive immune responses (13, 17–20). Although the impact of trametinib on MDSC mobilization has not been studied, MEK signaling is known to promote the lineage commitment of myeloid cells from hematopoietic stem cells and multipotent progenitor cells (21). Similar findings have also implicated MEK's target proteins, ERK1/2, in the development of myeloid cells (22). However, the requirement for MEK–ERK signaling in tumor-driven MDSCs expansion and the susceptibility of this pathologic axis to small-molecule MEK inhibition have yet to be studied.
A possible outcome of therapeutic MEK-signaling inhibition is alterations in the secretion of cytokines by intrinsically inflammatory, Ras-mutated tumor cells. Among these, osteopontin has been implicated in the recruitment of macrophages into tumors (23), and its expression is positively correlated with CD204+ M2-like macrophages (24). Osteopontin secreted by tumor cells has also been reported to drive the expansion of MDSCs in the spleens of tumor-bearing mice through activation of the ERK1/2–MAPK pathway in myeloid progenitors (25).
To examine the effects of trametinib on the tumor immunoenvironment, we dissected the role of trametinib in restricting the growth of a KRas-driven breast tumor cell line in immunocompetent mice. We find that the treatment of tumor-bearing mice with trametinib results in a reduction in monocytic MDSCs (mMDSC) that allows CD8+ T cells to control tumor growth. The impairment of MDSC mobilization is attributable to a direct blockade of MDSC expansion from bone marrow precursors and a reduced secretion of chemotactic molecules by tumor cells. Our study enhances understanding of trametinib's antitumor efficacy by demonstrating a mechanism of enhanced immunity through a reduction in pathologic MDSC mobilization.
Materials and Methods
WT C57BL/6 female 6- to 8-week-old mice were procured from the NCI (Bethesda, MD) or Charles River Laboratories. OT1 C57BL/6-Tg (TcraTcrb)1100Mjb/J transgenic mice were obtained from The Jackson Laboratory. All mice were randomized into treatment groups.
Cell lines and media
The Brpkp110 primary mammary tumor cell line was generated by culturing a mechanically dissociated B6 L-Stop-L-KRasG12Dp53flx/flxL-Stop-L-Myristoylated p110α−GFP/+ primary breast tumor mass (26). Tumor cells were passaged a total of 10 times before monoclonally deriving the Brpkp110 cell line. Cells used in this study underwent <4 subsequent passages. Lewis lung carcinoma cells (LLC1) were obtained from ATCC and cultured for <4 passages. All cell lines and lymphocytes were cultured in R10 (RPMI1640 (CellGro, with l-glutamine), 10% FBS, penicillin (100 IU/mL), streptomycin (100 μg/mL), l-glutamine (2 mmol/L), sodium pyruvate (0.5 mmol/L), and β-mercaptoethanol (50 μmol/L).
Cell proliferation assays
Cells were plated in 96-well plates and the next morning, trametinib (dissolved in DMSO) was diluted into the wells 1:1,000, so that the final concentration of DMSO was 0.1%. Cell proliferation was measured 48 hours later with the CellTiter 96 MTS assay (Promega) according to the manufacturer's instructions.
Tumor inoculation and treatments
Brpkp110 tumors were initiated by injecting 5 × 105 cells into the subcutaneous axillary region. LLC tumors were initiated by an injection of 2 × 105 cells intraperitoneally. Mice were treated by oral gavage once daily with a dose of 1.0 mg/kg trametinib (GSK-1120212, LC Laboratories) suspended in a vehicle solution of 10% PEG-300 (Sigma-Aldrich) and 10% Cremophor EL (EMD Millipore) in sterile dH2O. For MDSC adoptive transfer, mice were injected intravenously with 3–5 × 106 MDSCs differentiated for 4 days from bone marrow with 50% Brpkp110-conditioned media. For CD8 and natural killer (NK) cell depletion, mice were injected with anti-CD8α (clone YTS 169.4) or anti-NK1.1 (clone PK136) on day 3 (500 μg/mouse) and day 10 (250 μg/mouse) after tumor inoculation. All antibodies, including isotype controls (anti-CD8a: clone LTF-2, anti-NK1.1: clone C1.18.4), were purchased from Bio X Cell and injected intraperitoneally in sterile PBS. Tumor volume was calculated as 0.5 × (L × W2), where L is the larger dimension.
For flow cytometry analysis, tumors were dissected and mechanically dissociated through a 70-μm nylon cell strainer. For cell sorting, tumors were dissected, minced with a scalpel, digested in RPMI1640 containing 1 mg/mL collagenase (Type V, Sigma-Aldrich) for 1 hour at 37°C. Cells were dissociated through a 70-μm nylon cell strainer, followed by passage through a 23 G needle.
In vitro MDSC differentiation, suppression, and chemotaxis
Bone marrow from naïve mice was cultured for 4 days with IL6 (40 ng/mL, PeproTech) and GM-CSF (40 ng/mL, PeproTech) or media containing 50% tumor-conditioned media, prepared by filtering supernatant from a confluent flask of tumor cells through a 0.45-μm membrane. For suppression assays, MDSCs were either added to 2 × 105 CellTrace-labeled WT splenocytes simultaneously activated with anti-CD3 (500 ng/mL, clone 2C11, Tonbo) and anti-CD28 (1 μg/mL, clone 37.51, Tonbo) or 2 × 105 CellTrace-labeled OT-I splenocytes simultaneously activated with OVA257–264 peptide (1 μmol/L, GenScript) in 96-well plates. Proliferation of T cells was measured 3 days later. For chemotaxis assays, MDSCs were separated into Ly6G+ and Ly6G− fractions with anti-Ly6G MicroBeads (Miltenyi Biotec) according to the manufacturer's protocol. Chemotaxis was measured toward recombinant carrier-free osteopontin (R&D Systems) after 1 hour on 3-μm filter plates (Ly6G+ cells) or 4 hours on 5-μm filter plates (Ly6G− cells; Neuro Probe).
Human bone marrow
All patients with stage I–II lung cancer, who were scheduled for surgical resection, consented for tissue collection of a portion of their tumor and/or blood for research purposes at the Hospital of the University of Pennsylvania (Philadelphia, PA) and The Philadelphia Veterans Affairs Medical Center (Philadelphia, PA) after obtaining consents that had been approved by their respective Institutional Review Boards. All patients selected for entry into the study met the following criteria: (i) histologically confirmed pulmonary squamous cell carcinoma or adenocarcinoma; (ii) no prior chemotherapy or radiotherapy within 2 years; and (iii) no other active malignancy. Bone marrow cell suspension was obtained from the rib fragments that were removed from patients as part of their lung cancer surgery. The single-cell suspension was obtained by vigorous pipetting of cells flushed from bone marrow and passing the disaggregated cells through a 70-μmol/L nylon cell strainer. Total bone marrow cells were cultured in 6-well plates (2 × 106 leukocytes/well) in 3 mL of complete IMDM [IMDM (CellGro, with l-glutamine and 25 mmol/L HEPES), 15% FBS, penicillin (100 IU/mL), streptomycin (100 μg/mL), l-glutamine (2 mmol/L), and β-mercaptoethanol (50 μmol/L)] supplemented with 40 ng/mL human IL6 (PeproTech) and 40 ng/mL human GM-CSF (PeproTech) for 4 days. Cells were stained for surface marker expression, fixed in 1% paraformaldehyde, and analyzed by flow cytometry.
RNA from FACS-purified cell populations was isolated using TRIzol (Invitrogen) according to the manufacturer's instructions. cDNA was synthesized using the High-Capacity cDNA RT Kit (Applied Biosystems) and analyzed by qPCR with SYBR Green (Applied Biosystems) on a 7500 Fast machine (Applied Biosystems). Primers used were Arg1-fwd, 5′-GGAATCTGCATGGGCAACCTGTGT-3′; Arg1-rev, 5′-AGGGTCTACGTCTCGCAAGCCA-3′; Nos2-fwd, 5′- GTTCTCAGCCCAACAATACAAGA-3′; Nos2-rev, 5′- GTGGACGGGTCGATGTCAC-3′; TBP-fwd, 5′-CACCCCCTTGTACCCTTCAC-3′; TBP-rev, 5′-CAGTTGTCCGTGGCTCTCTT-3′.
Cells were lysed in RIPA buffer (Thermo Scientific) with protease inhibitors (Complete Protease Inhibitor Cocktail Tablets, Roche) and phosphatase inhibitors (Halt Phosphatase Inhibitor, Thermo Scientific, and Na3VO4, 1 mmol/L) and cleared by centrifugation. Proteins were quantified by BCA assay (Thermo Scientific), diluted in reducing Laemmli buffer, denatured by incubation at 95°C, run on mini Protean TGX Ready Gels (Bio-Rad Laboratories), transferred to a PVDF membrane, blocked, and incubated with primary antibodies for p-ERK1/2 (Cell Signaling Technology, clone D13.14.4E) and β-actin (Sigma, clone AC-15). Immunoreactive bands were developed using horseradish peroxidase–conjugated secondary antibodies (Bio-Rad Laboratories) and ECL substrate (GE Healthcare).
Tissues were embedded in Tissue-Tek OCT and frozen. Endogenous peroxidases were quenched from acetone-fixed sections (8 μm) by incubating in 0.3% H2O2 for 10 minutes at room temperature. Following quenching, sections were blocked using 3% goat serum, followed by staining with antibodies against Ki67 (clone D3B5, Cell Signaling Technology). IHC using the ABC Kit (Vector Laboratories) was performed according to the manufacturer's instructions, and sections were counterstained with hematoxylin. Slides were then imaged at 10× objective magnifications on a Nikon E600 upright microscope with a Nikon DS-Ri1 digital camera. Nikon NIS-Elements software was used for image acquisition and image stitching of the entire tumor. Image-Pro Plus 7 analysis software was used to measure the percentage of Ki67-stained nuclei within each sample. Percentage of stained nuclei was calculated as the total area of brown stained nuclei divided by the total area of the sample.
Brpkp110 cells were cultured in serum-free RPMI with DMSO or trametinib for 40 hours. Supernatants were collected, centrifuged, passed through a 0.22-μm filter, and concentrated by centrifugation in Amicon 3000 MWCO tubes (EMD Millipore). Concentrated supernatants were run 0.5 cm on a NuPage 12% Gel with MES buffer, extracted, digested with trypsin, and subjected to LC/MS-MS analysis by the Wistar Proteomics Facility.
Zombie Yellow (BioLegend) was used for all live/dead staining. For mouse experiments, antibodies from the following companies were used: Tonbo Biosciences: CD45.2-PerCP/Cy5.5 (104), CD11b-APC (M1/70), F480-PerCP/Cy5.5 (BM8.1), CD45-PE/Cy7 (30-F11), and CD3-FITC (145-2C11); and BioLegend: F480-PE/Cy7 (BM8), Ly6G-FITC (1A8), Ly6C-APC/Cy7 (HK1.4), I-A/I-E-PacBlue (M5/114.15.2), CD4-APC (RM4-5), CD25-APC/Cy7 (PC61), Foxp3-PacBlue (MF-14), CD8b-PerCP/Cy5.5 (YTS156.7.7), and H-2Kb-PE (AF6-88.5). For human experiments, antibodies from the following companies were used: Tonbo Biosciences: CD45-PE/Cy7 (HI30); BioLegend: CD14-APC (HCD14), CD15-FITC (HI98), HLA-DR-APC/Cy7 (L243), CD33-PerCP/Cy5.5 (WM53), and CD11b-PacBlue (ICRF44); and Sino Biological: VNN2-PE (04). Samples were run on a BD LSRII cytometer and analyzed by FlowJo.
Osteopontin concentrations were measured using ELISA Kits (RayBiotech) according to the manufacturer's instructions. Plasma was isolated from peripheral blood of mice by centrifugation in lithium heparin tubes (Becton Dickinson) and stored at −80°C. Intratumoral fluid was isolated from advanced Brpkp110 tumors after careful excision and blotting on gauze tissue (to remove excess fluid) by squeezing the tumor through a 10-mL syringe (BD, Luer-Lok Tip) into a microcentrifuge tube, followed by two centrifugation steps to obtain debris-free liquid, which was stored at −80°C.
Human gene expression analysis from trametinib/dabrafenib-treated melanomas
Normalized gene expression data for GSE61992 (27) were downloaded from the series matrix file in the Gene Expression Omnibus public database at the National Center for Biotechnology Information (Bethesda, MD). In this dataset, a pretreatment biopsy was taken from 9 patients with BRAFV600-mutant metastatic melanoma that were treated with dabrafenib plus trametinib, followed by biopsies taken from at least one progressing melanoma metastasis resected within 6 weeks of RECIST-determined disease progression. Human gene names were matched to Illumina HumanHT-12 V4.0 probe IDs and gene expression values were averaged for genes with multiple probes. Expression data were averaged for sample replicates and analyzed in a paired t test for pretreatment versus progressing lesions (the earliest biopsy taken from progressing lesions).
Unless indicated otherwise, all data shown represent means with SEM. All hypothesis testing was two-sided, and a significance threshold of P = 0.05 was used. Unpaired t tests were performed unless indicated otherwise. Analyses were carried out using GraphPad Prism software. Experiments were repeated at least twice unless otherwise indicated.
All animals were maintained in specific pathogen-free barrier facilities and used in accordance with the Institutional Animal Care and Use Committee of the Wistar Institute, Philadelphia, PA).
Trametinib reduces mMDSC accumulation in tumor-bearing mice
We have previously demonstrated that MEK inhibitors, including trametinib, inhibit T-cell responses in vitro and in vivo, which can be rescued with T-cell cytokines (12). To dissect the effects of trametinib treatment on other compartments of the immunoenvironment of solid tumors, we analyzed the effect of trametinib against the progression of transplantable Brpkp110 breast tumors, derived from syngeneic autochthonous p53/KRas/PI3K-driven primary breast adenocarcinomas (12, 17, 26). Trametinib alone dramatically delayed breast cancer growth in multiple independent experiments (Fig. 1A). This was unexpected because, in addition to the direct T cell–suppressive effects of trametinib, Brpkp110 cells were only weakly sensitive to trametinib (70% growth compared with vehicle) in vitro (Fig. 1B). Accordingly, trametinib did not significantly reduce Ki67 staining in Brpkp110 tumors (Fig. 1C and D), indicating that in vivo proliferation of tumor cells was largely unaffected by direct MEK inhibition. We therefore reasoned that trametinib-induced changes in the microenvironment might be responsible for the decrease in Brpkp110 tumor growth.
To elucidate possible changes in the immunoenvironment, we treated mice with established tumors for 3 days (to minimize differences in tumor burden) and analyzed populations of tumor-infiltrating leukocytes. Despite the sensitivity of T cells to MEK inhibition (12), we did not observe a decrease in CD8+ or CD4+ T cells at the tumor beds in trametinib-treated tumor-bearing mice (Fig. 2A). There also was no difference in the proportion of PD-1+ intratumoral CD8+ or CD4+ T cells in trametinib-treated mice (Supplementary Fig. S1A). The proportion of Foxp3+ Tregs and differentiated (MHC-II+) macrophages also did not significantly change after trametinib treatment (Fig. 2B). In contrast, trametinib induced a dramatic reduction in the accumulation of specifically CD11b+MHC-II−Ly6ChiLy6G− mMDSCs, while CD11b+MHC-II−Ly6ClowLy6G+ polymorphonuclear MDSCs (PMN-MDSC) were not reduced, whether analyzed as a proportion of total leukocytes (Fig. 2C) or as a proportion of CD11b+MHC-II− myeloid cells (Fig. 2D and E). The apparent increase in PMN-MDSCs as a percentage of CD11b+MHC-II− cells was entirely due to a reduction in mMDSCs because overall percentages of PMN-MDSCs among leukocytes remained constant while the respective percentages of M-MDSCs were reduced. Confirming their identity as immunosuppressive cells, these mMDSCs expressed high levels of arginase and Nos2 (Fig. 2F) and were capable of suppressing the proliferation of activated T cells (Fig. 2G). The selective reduction in mMDSCs was also observed in the spleens of trametinib-treated tumor-bearing mice (Fig. 2H and I), suggesting a systemic effect on mMDSC mobilization.
Preferential reduction in mMDSCs mobilization was not restricted to breast cancer, because mice gavaged with trametinib also induced a reduction in the accumulation of mMDSCs in the solid tumors (Fig. 3A) and spleen (Fig. 3B and C) of mice with intraperitoneal LLC tumors. Accordingly, LLC disease progression was also delayed (Fig. 3D). Of note, LLC proliferation in vitro, unlike Brpkp110 cells, is acutely sensitive to trametinib (Figs. 1B and 3E). However, trametinib only has a modest survival benefit, suggesting that the direct effect of trametinib on tumor cells is not translated in vivo, even though the drug effectively inhibited pERK at tumor beds (Fig. 3F). Therefore, trametinib's ability to decrease MDSC accumulation appears to be the predominant mechanism underlying delayed LLC progression in vivo. Together, these data demonstrate that trametinib decreases the accumulation of immunosuppressive mMDSCs at tumor beds in multiple tumor models, resulting in significant delays in tumor growth independent of in vitro sensitivity of tumor cells to the drug.
Trametinib abrogates mMDSC expansion by inhibiting the Ras–MAPK pathway in myeloid precursors
To dissect the mechanism whereby trametinib treatment inhibits the mobilization of mMDSCs, we stimulated myelopoiesis in bone marrow cells with a combination of GM-CSF and IL6 (28). Recapitulating our in vivo observations, trametinib significantly impaired the expansion of mMDSCs in a dose-dependent manner (Fig. 4A and B). A reduction in the expansion of PMN-MDSCs was also observed with trametinib treatment, although to a much smaller degree than for M-MDSCs, further supporting the preferential inhibition of trametinib on the mobilization of myeloid cells of the monocytic lineage. Similar inhibitory effects were observed when MDSCs were differentiated from bone marrow using Brpkp110-conditioned media (Fig. 4C and D), a system that effectively promotes expansion of MDSCs capable of suppressing T-cell proliferation (Fig. 4E). A separate MEK inhibitor (AS-703026) and an ERK inhibitor (SCH772984) also demonstrated dose-dependent inhibition of mMDSC differentiation from bone marrow cultured by Brpkp110-conditioned media (Supplementary Fig. S2), confirming this effect is not restricted to trametinib. We verified that trametinib abrogated MEK signaling at both early and late stages in bone marrow cultures incubated with Brpkp110-conditioned media (Fig. 4F). Importantly, we investigated whether trametinib inhibited myeloid progenitor expansion through cytotoxic effects on MDSCs. Although there was a slight increase in the percentage of dead cells collected from cultures after MDSC expansion in the presence of trametinib (Fig. 4G), it was not observed at earlier time points, and it was not sufficient to account for the significant reduction in live cells at the end of MDSC expansion with trametinib (Fig. 4H). There was also no increase in the apoptosis of in vitro–derived MDSCs cultured in trametinib for 4 hours (Fig. 4I and J). These results indicate that blocking MEK signaling likely acts to prevent differentiation of precursor cells instead of increasing the apoptosis of MDSCs their precursors.
To support the clinical relevance of MDSC inhibition by trametinib, we next obtained human bone marrow isolates from 7 different lung cancer patients undergoing partial rib resection and induced MDSC expansion by culturing in GM-CSF and IL6 (28). Although a more diverse population of myeloid cells was observed in these cultures in comparison with murine bone marrow, a CD14+CD15intHLA-DR−/lowCD11b+CD33+ population of immature myeloid cells (corresponding to mMDSCs; ref. 29) did reproducibly expand (Fig. 5A and B). As in our mouse bone marrow and in vivo experiments, trametinib induced in a dose-dependent reduction preferentially in CD14+CD15int cells, consistently observed across all patient samples (Fig. 5A and C).
To extend our observations to the clinical use of trametinib, we analyzed published gene expression data of BRAFV600-mutant melanoma lesions biopsied from patients before and after treatment with trametinib plus dabrafenib (30). We found significant posttreatment decreases in two genes associated with mMDSCs, IL4R and VNN2 (Supplementary Fig. S3). High IL4Rα expression correlates with immunosuppressive abilities of mMDSCs from cancer patients (31), whereas VNN2 (also known as GPI-80 or vanin-2) expression has been reported on immunosuppressive CD14+HLA-DR−/low MDSCs (32). Indeed, we confirmed that the CD14+CD15intHLA-DR−/low cells expanded in our cultures expressed VNN2, whereas CD14−CD15highHLA-DR−/low cells lacked surface VNN2 (Fig. 5B). Because the gene expression data we analyzed were taken after tumor lesions progressed on trametinib/dabrafenib treatment, once tumor cells acquired drug resistance, decreases in MDSC-associated genes likely indicate tumor cell–independent activity of BRAF/MEK inhibition and suggest that trametinib may reduce mMDSCs in human patients.
Taken together, these results indicate that trametinib, by inhibiting the Ras–MAPK pathway in myeloid precursors, disproportionately decreases the mobilization of immature mMDSCs in mice and humans. This potentially explains the decreased accumulations of mMDSCs in the periphery and in the tumor beds in vivo, and it suggests a novel mechanism that possibly contributes to the antitumor activity of trametinib in human patients.
CD8+ T cells are required for optimal antitumor activity of trametinib
Our results mentioned above suggest that trametinib may be effective in the Brpkp110 tumor model by decreasing the expansion of immunosuppressive MDSCs, thus allowing antitumor T-cell responses to control tumor growth. This mechanism, however, requires a functional effector response by tumor-infiltrating T cells in the presence of trametinib treatment. We found that short-term treatment of tumor-bearing mice with trametinib did not decrease the frequencies of CD8+ or CD4+ tumor-infiltrating T cells capable of producing IFNγ when restimulated ex vivo (Fig. 6A). In fact, after 10 days of trametinib gavage, the frequencies of IFNγ-producing CD8+ and CD4+ intratumoral T cells was increased compared with control treatment (Fig. 6B). The general decrease in functional effector T cells at this later time point most likely reflects an increasing burden of immunosuppression in the TME, yet the finding that trametinib is able to forestall this progression to effector T cell dysfunction suggests that antitumor immunity may be important during trametinib treatment. To examine whether T-cell immunity contributes to the efficacy of trametinib, we depleted CD8α+ T cells during trametinib treatment of Brpkp110-bearing mice. Although trametinib was able to prevent tumor growth in mice treated with an irrelevant IgG, trametinib was markedly less effective at limiting tumor growth in mice lacking CD8α+ T cells (Fig. 6C). These data demonstrate that tumor-infiltrating effector T cells are not restricted by trametinib in vivo and are actually necessary for the full efficacy of trametinib. In contrast, tumors did not grow significantly faster when trametinib-treated mice were depleted of NK cells (Fig. 6D).
To determine that a reduction in MDSCs contributes to the activity of trametinib, we adoptively transferred ex vivo–derived MDSCs into Brpkp110-bearing mice treated with trametinib and found that tumor growth was accelerated (Fig. 6E). The complementary observations that the efficacy of trametinib can be reduced by either depleting CD8α+ T cells or exogenously restoring MDSC levels suggest a mechanism in which trametinib reduces myeloid-driven immune suppression to facilitate the capacity of cytotoxic T cells to control tumor growth.
Trametinib abrogates the production of osteopontin by KRas-mutated tumor cells
We also reasoned that trametinib could decrease immunosuppression in the TME by directly modulating the expression of coinhibitory ligands or cytokines by tumor cells. Treatment of Brpkp110 cells in vitro with trametinib did not decrease PD-L1 expression, and in fact, trametinib slightly increased PD-L1 when Brpkp110 cells were exposed to IFNγ (Supplementary Fig. S1B). Instead, LC/MS-MS analysis of culture supernatants from Brpkp110 cells revealed that several cytokines were dramatically altered in response to trametinib treatment. A number of factors known to be involved in recruitment of myeloid cells, such as CSF1, CCL2, and CX3CL1 (Fig. 7A), were decreased after trametinib treatment. Most importantly, osteopontin (SPP1) was decreased >18-fold upon trametinib treatment. We confirmed the trametinib-driven decrease in osteopontin secretion in separate ELISA experiments (Fig. 7B). We focused on osteopontin because it has been reported to induce MDSC expansion (25) and tumor recruitment of macrophages (24). Supporting the relevance of our proteomic analysis, the tumor-driven increase of osteopontin in plasma was abrogated by short-term (3 days) treatment with trametinib in Brpkp110-bearing mice (Fig. 7C).
Importantly, when in vitro–derived Ly6G+ and Ly6G− MDSCs were isolated with Ly6G-MACS separation (Fig. 7D), they were able to migrate toward a gradient of recombinant osteopontin (Fig. 7E and F). This trafficking was clearly chemotactic because migration was greatly diminished when osteopontin was supplied on the same side of the transwell chamber as the cells. Although the osteopontin concentration required for chemotaxis (100 μg/mL) is much higher than the concentration observed in the plasma (<1 μg/mL), we found that the concentration of osteopontin in the intratumoral fluid obtained from Brpkp110 tumor samples exceeded 200 μg/mL (Fig. 7G). Therefore, the physiologic range of osteopontin concentrations in tumor-bearing mice should induce migration of MDSCs from peripheral blood into tumor tissue. Together, these results indicate that trametinib also reduces MDSCs in the tumor by reducing the production of chemotactic cytokines by tumor cells through a direct anti-inflammatory effect that is independent of changes in tumor cell proliferation.
Here, we show that trametinib controls malignant progression by abrogating the mobilization of MDSCs into tumors by directly impairing differentiation from bone marrow precursors and indirectly by reducing tumor-secreted chemotactic molecules. Ultimately, immune suppression in the tumor is reduced, which explains why the therapeutic effectiveness of trametinib depends on the activity of antitumor T cells.
Our data offer mechanistic insight into the apparent inconsistency between in vitro and in vivo effects of trametinib on antitumor adaptive immunity. Thus, we and others have previously shown that trametinib has a direct inhibitory effect on proliferation and effector function in both naïve and memory T cells in vitro (6, 7, 12). In contrast, recent reports indicate that trametinib does not limit the effectiveness of adoptive T-cell therapy (10) or checkpoint blockade with antibodies against PD-1, PD-L1, and CTLA-4 (11) in other mouse models. We previously showed that the ability of common γ chain cytokines, which were administered exogenously or were likely upregulated in response to immunotherapy, to rescue the immunosuppressive activity of trametinib could explain some of these discrepancies (12). Our current work offers a more comprehensive picture of how trametinib impacts multiple cell types in tumor-bearing hosts. We demonstrate that trametinib reduces the accumulation of a major immunosuppressive cell compartment in our KRas-driven tumor model, namely MDSCs. This occurs at two levels: On one hand, inflammation-induced MDSC expansion clearly depends on the MAPK pathway, which is abrogated upon MEK inhibition, in both human and mouse bone marrow precursors. On the other hand, trametinib reduces tumor-derived osteopontin secretion, which correlates with a decrease in mMDSCs. In addition, our study identifies for the first time the chemoattractant activity of osteopontin on MDSCs at physiologically relevant concentrations. Together, these combined effects, in addition to the rescuing activity of endogenous cytokines, compensate for the direct inhibitory effects of trametinib on T cells. This ultimately results in CD8 T cells, but not NK cells, being paradoxically required for the full efficacy of trametinib.
In their seminal study, Hu-Lieskovan and colleagues (10) also looked at the proportions of MDSCs in tumors and spleens of mice treated with trametinib. However, they did not observe a decrease in mMDSCs and instead found a decrease in PMN-MDSCs in tumors of mice treated with trametinib and dabrafenib (10), whereas our study shows a preferential suppressive effect on the mobilization of mMDSCs. This difference might be due to different treatment schemes. In their experiments, trametinib was combined with either dabrafenib or adoptive cell therapy, and it was not tested as a single agent as it was in our study. In addition, their study used a BRAF-driven melanoma tumor, which may respond differently than our KRas-driven breast tumor in the secretion of inflammatory mediators under MEK inhibition. Nevertheless, our in vitro studies clearly demonstrate that MEK inhibition abrogates cytokine-induced MDSC expansion, with a preferential direct effect on mMDSCs. In addition, in a murine study of allograft transplantation, rapamycin treatment induced the expansion of mMDSCs (defined as CD11b+GR1int), but this effect was prevented by trametinib administration (33). These combined findings show that in some systems, mMDSC expansion can be blunted by MEK inhibition.
Our results also highlight the importance of tumor microenvironmental and systemic responses to trametinib. We observed that LLC cells are more sensitive to trametinib in vitro than KRas-mutated Brpkp110 tumor cells. However, in vivo trametinib is not more effective against LLC than Brpkp110 tumors. It is therefore likely that trametinib's effectiveness depends on the reliance of at least some tumors on the expansion and the recruitment of immunosuppressive MDSCs, rather than the direct cytotoxic or cytostatic effects of trametinib on cancer cells.
Overall, our study offers novel mechanistic understanding to reconcile inconsistent effects of trametinib in vitro and in vivo and explains how, by influencing multiple cells types in tumor-bearing hosts, MEK inhibition could have overall permissive effects on protective antitumor immunity. Subsequent analyses of trametinib's effectiveness as a function of MDSC burden and activity in cancer patients will further substantiate whether its therapeutic activity is immune dependent.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: M.J. Allegrezza, T.L. Stephen, J.R. Conejo-Garcia
Development of methodology: M.J. Allegrezza, M.R. Rutkowski, N. Svoronos, A. Perales-Puchalt, J.R. Conejo-Garcia
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.J. Allegrezza, A. Perales-Puchalt, K.K. Payne, S. Singhal, E.B. Eruslanov, J. Tchou
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.J. Allegrezza, A. Perales-Puchalt, K.K. Payne, J. Tchou
Writing, review, and/or revision of the manuscript: M.J. Allegrezza, M.R. Rutkowski, T.L. Stephen, N. Svoronos, A. Perales-Puchalt, K.K. Payne, S. Singhal, E.B. Eruslanov, J.R. Conejo-Garcia
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.J. Allegrezza, J.M. Nguyen
Study supervision: J.R. Conejo-Garcia
We thank the Wistar Flow Cytometry, Proteomics, and Imaging facilities at Wistar, as well as the Gabrilovich laboratory, for technical assistance.
This study was supported by the grants R01CA157664, R01CA124515, R01CA178687 (J.R. Conejo-Garcia), The Jayne Koskinas & Ted Giovanis Breast Cancer Research Consortium at Wistar, Ovarian Cancer Research Fund (OCRF) Program Project Development awards, RO1CA187392 (E.B. Eruslanov), T32CA009171 (M.J. Allegrezza and N. Svoronos), T32CA009140 (K.K. Payne), Ann Schreiber Award (OCRF; A. Perales-Puchalt), and Cancer Center Support Grant (CCSG) CA010815 (D. Altieri) to The Wistar Institute.
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.