INTRODUCTION. We previously showed, using the KPC (Kras^{LSL G12D/+}; p53 ^r172H/+; Pdx1-Cre) mouse model of pancreatic cancer (PDAC), that sequential treatment with the DNA hypomethylating agent (HMA) decitabine (DAC) followed by aPD-1 initially enhanced anti-tumor effects, yet tumors developed resistance linked to a unique M2-polarized Chil3+ myeloid subtype. Therefore, we aimed to characterize the mechanisms of polarization and functional significance of these myeloid cells in order to develop strategies to target them in combination with DAC and aPD1. We focused on cytokines that may be responsible for an unfavorable myeloid polarization, and HDAC inhibitors that may show immunologic and cytotoxic synergy with DAC therapy.
METHODS. Bone marrow (BMDM) and peritoneum (PM) derived myeloid cells were used to test the effect of conditioned media (CM) from primary KPC cell lines after treatment with DAC (KPC-SN-DAC). CM treated myeloid cells were used to test their effect on T cells compared with direct and indirect effect of DAC. We used expression profiling of myeloid cells and cytokine arrays to identify and test immune targets. The WST1 assay and the Bliss/Loewe consensus were used to assess cytotoxic synergy. Candidate drugs were combined with DAC (0.2 mg/kg s.c. daily) and aPD1 (twice weekly 200 ug) in the orthtopic KPC model.
RESULTS. BMDM and PM treated with KPC-SN-DAC resulted in an HMA-specific M2-like gene signature. Notably DAC alone had minimal myeloid polarization effects. CD8⁺ T cells co-cultured with KPC-SN-DAC polarized macrophages showed significant differences in activation and exhaustion markers, thus confirming the myeloid cells’ immunosuppressive properties. To investigate which factors may trigger this polarization, we probed the KPC-SN-DAC using cytokine arrays and identified a number of candidate cytokines induced or secreted from KPC cells specifically following HMA treatment. None of the individual cytokine agonist/antagonist treatments were able to revert the myeloid polarization or T cell effect. We then evaluated four different HDAC inhibitors with reported immunomodulatory effects and non-overlapping HDAC isoform specificities: Romidepsin (RM), Domatinostat (DM), Pracinostat (PR) and Entinostat (EN). RM, DM, PM showed cytotoxic synergy at ultra-low doses and DM and PR reverted DAC-induced cytokine secretion. In the orthotopic model, DAC+PR+aPD1 displayed the most significant tumor volume reduction, prolonged survival more than any agent or dual agent combination and lead to a significant reduction in Chil3 expression compared to DAC+aPD1 treatment.
CONCLUSIONS. Our findings identify a functionally immune suppressive HMA specific myeloid effect. The addition of a second epigenetic agent enhances the therapeutic efficacy resulting in prolonged survival and reverses the suppressive myeloid cell related effects of DAC + aPD1.

Stephen Muzyka

Appearances