PARP Inhibitor Upregulates PD-L1 Expression and Enhances Cancer-Associated Immunosuppression

Abstract

Purpose: To explore whether a cross-talk exists between PARP inhibition and PD-L1/PD-1 immune checkpoint axis, and determine whether blockade of PD-L1/PD-1 potentiates PARP inhibitor (PARPi) in tumor suppression.Experimental Design: Breast cancer cell lines, xenograft tumors, and syngeneic tumors treated with PARPi were assessed for PD-L1 expression by immunoblotting, IHC, and FACS analyses. The phospho-kinase antibody array screen was used to explore the underlying mechanism of PARPi-induced PD-L1 upregulation. The therapeutic efficacy of PARPi alone, PD-L1 blockade alone, or their combination was tested in a syngeneic tumor model. The tumor-infiltrating lymphocytes and tumor cells isolated from syngeneic tumors were analyzed by CyTOF and FACS to evaluate the activity of antitumor immunity in the tumor microenvironment.Results: PARPi upregulated PD-L1 expression in breast cancer cell lines and animal models. Mechanistically, PARPi inactivated GSK3β, which in turn enhanced PARPi-mediated PD-L1 upregulation. PARPi attenuated anticancer immunity via upregulation of PD-L1, and blockade of PD-L1 resensitized PARPi-treated cancer cells to T-cell killing. The combination of PARPi and anti-PD-L1 therapy compared with each agent alone significantly increased the therapeutic efficacy in vivoConclusions: Our study demonstrates a cross-talk between PARPi and tumor-associated immunosuppression and provides evidence to support the combination of PARPi and PD-L1 or PD-1 immune checkpoint blockade as a potential therapeutic approach to treat breast cancer. Clin Cancer Res; 23(14); 3711-20. ©2017 AACR.

Conflict of interest statement

Conflicts of Interest

The authors have declared that no conflict of interest exists.

©2017 American Association for Cancer Research.

Figures

Figure 1. PARPi upregulates PD-L1 protein expression in breast cancer cells

(A) MDA-MDA-MB-231 and BT549 cells were treated with 10 μM olaparib or 10 nM talazoparib for 24 hours, and subjected to immunoblotting with the indicated antibodies. PD-L1 knockout (K/O) cells were included as a negative control. (B) PD-L1 expression in PARP1 knockdown (K/D), PARP1 knockout (K/O), and MDA-MB-231 parental cells by immunoblotting. (C and D) The indicated MDA-MB-231 cells were subjected to FACS analysis for cell surface PD-L1 expression. (E) SUM149 cells were treated with the indicated concentrations of olaparib for 10 days, and cell surface PD-L1 expression was determined by FACS.

Figure 2. PARPi upregulates PD-L1 expression in xenograft tumors

(A) BT549, (B) SUM149 and (C) MDA-MB-231 cells were inoculated into mammary fad pad of nude mice, and the mice with established tumors were treated with olaparib or rucaparib. Tumors were then isolated to evaluate PD-L1 expression by immunoblotting (A and B) or IHC staining (C). Black arrowheads indicate the detected PD-L1 signals. Scale bar, 50 μm.

Figure 3. GSK3β inactivation is required for PARPi-mediated PD-L1 upregulation

(A) SUM149 cells were treated with olaparib, and subjected to human phospho-Kinase Array. The top two responding kinases were Chk2 and GSK3α/β. (B) SUM149 cells were treated with the indicated concentrations of olaparib for 10 days, and subjected to immunoblotting with the indicated antibodies. (C) BT549 parental or GSK3β knockout cells were treated with 10 μM olaparib for 24 hours. PD-L1 expression was evaluated by immunoblotting.

Figure 4. Olaparib attenuates PD-1 binding and T-cell-mediated cell death in TNBC cells

(A) FACS analysis of cell surface PD-1 binding of MDA-MB-231 cells treated with 10 μM olaparib or 10 nM talazoparib for 24 hours. (B) FACS analysis of PARP1 knockdown (K/D), PARP1 knockout (K/O), and parental MDA-MB-231 cells. (C) SUM149 cells were treated with the indicated concentrations of olaparib for 10 days. (D) MDA-MB-231 cells expressing nuclear RFP protein were first treated with or without olaparib (10 μM) for 3 hours and then co-cultured with or without activated peripheral blood mononuclear cells (PBMCs). Left, quantitation showing the number of live cells per well, counting the number of red fluorescent objects, normalized to that at the zero time point. Right, the percent of T cell-meditated tumor cell killing observed at 72 hours in activated PBMC co-culture with control or olaparib-treated cells (normalized to co-culture without PBMCs). (E) Left, representative merged images showing red fluorescent (nuclear restricted RFP), and green fluorescent (Caspase 3/7 substrate) objects in MDA-MB-231 cells co-cultured with activated PBMCs at 0 and 72 hours. Images were taken using the IncuCyte Zoom microscope. Right, quantitation showing the number of live cells following treatment with olaparib (10 μM), PD-L1 antibody (PD-L1 Ab; 10 μg/ml), or the combination co-cultured with activated PBMCs for 72 hours. The number of live cells (red fluorescent objects) were counted and normalized to that at the zero time point. * P

Figure 5. PARPi-induced PD-L1 upregulation suppresses anticancer immunity and blockade of PD-L1 potentiates PARPi

(A) EMT6 cells were treated with 10 μM olaparib for 24 h. Cell surface PD-L1 were analyzed by FACS. (B) Representative images of tumors after olaparib and/or anti-PD-L1 antibody treatment at the indicated time points in the EMT6 syngeneic mouse model. (C) Effects of olaparib and/or anti-PD-L1 antibody treatment on tumor growth in EMT6 syngeneic mouse model treated (n = 8). Tumors were measured at the indicated time points and dissected for tumor cell PD-L1 expression analysis, TIL analysis, and pathological analysis at endpoint. (D) Hematoxylin and eosin (H&E) and Ki-67 staining of EMT6 tumors. Scale bar, 50 μm. (E) Cell surface PD-L1 expression in EMT6 cells derived from EMT6 mouse tumors. (F) Cytotoxic CD8+ T cells population (IFNγ+ CD8+ CD3+ CD45+) in TILs isolated from EMT6 tumors by CyTOF analysis. * P < 0.05

Figure 6. Inverse correlation between PAR and PD-L1 in surgical specimens of breast cancer

Top, representative images of IHC staining of PAR and PD-L1 in human breast cancer tissues (n = 116). Scale bar, 50 μm. Bottom, correlation analysis between PAR and PDL-1 was analyzed using the Pearson Chi-Square test (P = 0.02). A P value of less than 0.05 was considered to be statistically significant.