Introduction: Emerging evidence indicates a critical role for both the phosphoinositide-3-kinase (PI3K)/mammalian target of rapamycin (mTOR) and mitogen associated protein kinase (MAPK) pathways in the pathogenesis and progression of head and neck squamous cell carcinoma (HNSCC). Combining PI3K/mTOR and MAPK pathway targeting therapies results in synergistic tumor growth inhibition and prevents therapeutic resistance observed with inhibition of either pathway alone. Understanding how these targeted therapies alter programmed death ligand (PD-L) expression in the tumor microenvironment is critical for the rational combination of targeted and immune checkpoint antibody therapies. Here, we characterize baseline PD-L expression on different cellular subsets and how this expression is altered following mTOR inhibition with rapamycin and MAPK inhibition with the MEK1/2 inhibitor PD325901 (PD901) in a syngeneic murine model of oral cavity cancer.

Methods: Ras-mutant mouse oral cancer (MOC) cells demonstrating a co-activated PI3K/mTOR pathway were transplanted into immunocompetent C57BL/6 mice. Mice were treated with systemic rapamycin (1.5 mg/kg QOD), PD901 (1.5 mg/kg QD), combination or control for 21 days following primary tumor growth to 0.1 cm3. At the end of the treatment period, control and treated mice were euthanized, tumor harvested and processed, and PD-L1/2 expression was characterized on CD45-CD31- MOC tumor cells, CD45-CD31+ endothelial cells, and tumor-infiltrating CD45+ hematopoietic cells including Gr1+CD11b+ myeloid derived suppressor cells (MDSCs), CD11b+Ly6ClowF4/80+ macrophages (TAMs) and CD4+Foxp3+ regulatory T-cells (Tregs) using flow cytometric analysis. Results analyzed are from 7 individual tumors for each treatment condition in two biological replicate experiments.

Results: PD-L1 was expressed on 62.2% MOC tumor cells with an average mean fluorescence intensity (MFI) of 450 at baseline. Tumor cell PD-L1 expression could be stratified by CD44 expression (p<0.001) suggesting co-activation. Further characterization revealed PD-L1 expression on 24.4% of endothelial cells (avg MFI 208), 43.5% of MDSCs (avg MFI 614), 90.7% of TAMs (avg MFI 4167) and 58.6% of Tregs (avg MFI 330). PD-L2 was expressed on less than 5% of all cell subsets analyzed. Following targeted therapy in-vivo, MEK inhibition alone resulted in a significant decrease in PD-L1 expression on MOC tumor cells (p<0.001), MDSCs (p<0.05), TAMs (p<0.01) and Tregs (p<0.001) but a significant increase in tumor endothelial cell PD-L1 expression (p<0.05). Rapamycin alone appears to induce little change in PD-L1, but produces additive effects when combined with MEK inhibition to enhance or suppress PD-L1 expression in different cell subsets (p<0.001 in MOC tumor cells, TAMs, Tregs and endothelial cells, p<0.01 in MDSCs).

Conclusions: PD-L1 but not PD-L2 was expressed at baseline in varying degrees on all tumor cell subsets analyzed. MEK and mTOR targeted inhibition variably alters PD-L1 expression in-vivo. Interestingly, while targeted therapy reduced PD-L1 expression on most cell subsets analyzed, PD-L1 expression on endothelial cells was enhanced. Knowledge of how different targeted and standard cytotoxic anti-cancer treatments alter checkpoint inhibitor expression and validation of this murine data in human clinical trial samples will be critical for the rational design of combination targeted and immune checkpoint antibody-based immunotherapies.