Adipose PD-L1 Modulates PD-1/PD-L1 Checkpoint Blockade Immunotherapy Efficacy in Breast Cancer

Abstract

Programmed death-ligand 1 (PD-L1) and its receptor programmed cell death protein 1 (PD-1) modulate antitumor immunity and are major targets of checkpoint blockade immunotherapy. However, clinical trials of anti-PD-L1 and anti-PD-1 antibodies in breast cancer demonstrate only modest efficacy. Furthermore, specific PD-L1 contributions in various tissue and cell compartments to antitumor immunity remain incompletely elucidated. Here we show that PD-L1 expression is markedly elevated in mature adipocytes versus preadipocytes. Adipocyte PD-L1 prevents anti-PD-L1 antibody from activating important antitumor functions of CD8⁺ T cells in vitro. Adipocyte PD-L1 ablation obliterates, whereas forced preadipocyte PD-L1 expression confers, these inhibitory effects. Pharmacologic inhibition of adipogenesis selectively reduces PD-L1 expression in mouse adipose tissue and enhances the antitumor efficacy of anti-PD-L1 or anti-PD-1 antibodies in syngeneic mammary tumor models. Our findings provide a previously unappreciated approach to bolster anticancer immunotherapy efficacy and suggest a mechanism for the role of adipose tissue in breast cancer progression.

Keywords: Breast cancer; PD-1; PD-L1; PPARgamma antagonist; adipocyte; combination therapy; immune checkpoint; immunotherapy; inflammation and cancer; new targets.

Figure 1.

Adipocytes express high levels of PD-L1. (a) PD-L1 mRNA by PCR in 3T3-L1 and 3T3-F442A pre- and post-adipogenesis. (b) PD-L1 protein in cells by WB before and after adipogenesis. (c) Diagram for in vitro adipogenesis (left) and PD-L1 protein expression at different stages of adipogenesis in 10T1/2 (right). aP2 is an adipogenic marker and β-actin is the loading control. (d) Comparison of PD-L1 protein by WB in 10T1/2 pre/post adipogensis and PD-L1 WT/KO B16 melanoma cells. (e) Representative immunofluorescence images of PD-L1 (red), plasma membrane marker wheat germ agglutinin (WGA, green) and nuclear marker DAPI (blue) in pre- and post-adipogenic 10T1/2 cells. (f) Immunostaining of PD-L1 and CD36 using WAT from C57BL/6 mice. (g) WB of PD-L1 and aP2 proteins in adipose stromal cells (ASC) and adipocytes from human breast tissue. One representative result from three donor samples are shown here. β-Actin is used as the loading control. Values represent mean ± SD.

Figure 2.

3’ UTR-mediated control of PD-L1 expression during adipogenesis. (a) Diagram of the pGL3-Cd274/Pd-l1 promoter reporter construct. (b) IFNγ stimulation of Cd274 promoter-driven luciferase in B16 and 3T3-L1 cells. (c) Cd274 promoter-driven luciferase reporter activity before and after adipogenesis. (d) Diagram of the pmirGLO-CD274 3’UTR reporter construct. (e) Cd274 3’UTR luciferase reporter activity before and after adipogenesis.

Figure 3.

Adipose PD-L1 attenuates anti-PD-L1 antibody effect in vitro. (a) Flow chart of the coculture assay. (b) Percentage of IFNγ⁺ T cells in CD8⁺ cells following co-culturing of total splenocytes, anti-PD-L1 antibody, and adipocytes. (c) Immunofluorescent staining of PD-L1 (red), WGA (green) and DAPI (Blue) in doxycycline (Dox, 1μg/ml) treated PD-L1-inducible 3T3-L1 cells. (d) Percentage of IFNγ⁺ T cells in CD8⁺ cells following co-culturing of total splenocytes, anti-PD-L1 antibody, and preadipocytes. Dox was included in all co-culture wells throughout the experiment. Induced ectopic PD-L1 in 3T3-L1 derivative cells is shown in the inlet. Data expressed as mean ± SD.

Figure 4.

PPARγ antagonist GW9662 represses adipose PD-L1 expression. (a) Oil Red O staining and Western blotting of aP2, PD-L1, and β-actin in 3T3-L1 during adipogenesis, with or without GW9662 treatment. (b) Representative PD-L1 immunofluorescent staining of subcutaneous WAT from DMSO- and GW9662-treated mice (left). Fold change (normalized to vehicle-treated group) of fluorescence intensity of PD-L1 is shown on the right. (c-f) Fold changes of mean fluorescence intensity (MFI) from specific cells types isolated from either sham (vehicle) or GW9662-treated mice are shown for splenocytes (c), lung cells (d), thymus cells (e), fat tissue macrophages (CD45+ CD3-CD11b+ F4/80+) (f). Values represent mean ± SD. N = 5 per group.

Figure 5.

GW9662 boosts anti-PD-L1/PD-1 antibody efficacy in vivo. (a-d) Tumor study with treatment of GW9662 and anti-PD-L1. E0771 mammary tumor growth curve (a), representative tumor image (b) and tumor weight (c) upon harvest. Survival curve is shown in (d). Treatments are as follow. Sham: vehicle + anti-IgG (n = 13), vehicle + anti-PD-L1 (n = 13), GW + anti-IgG (n = 13) and anti-PD-L1 + GW (n = 13). (e-f) Tumor study with treatment of GW9662 and anti-PD-1. E0771 mammary tumor growth (e) and survival (f). Sham: vehicle + anti-IgG (n = 4), vehicle + anti-PD-1 (n = 4), GW + anti-IgG (n = 5) and anti-PD-1 + GW (n = 4). Data are mean± SEM. P values for the tumor growth curve in (a) indicate the statistics for the last measurement time point. P values for the survival analysis in (d) test the null hypothesis that the indicated two survival curves are identical.

Figure 6.

GW9662 boosts anti-tumor immunity in E0771 TILs. Immunophenotyping of various tumor infiltrating lymphocytes (TILs) after treatments: (a) Total CD8⁺ cells, and in the CD8+ gate: (b) IFNγ⁺, (c) CD44⁺, (d) PD-1⁺, (e) CD4⁺ in CD3⁺, and in CD4+ gate: (f) CD25⁺FOXP3⁺, (g) IFNγ⁺, (h) CD44⁺, Sham: vehicle + anti-IgG (n = 5); α–PD-L1: vehicle + anti-PD-L1 (n = 4); GW: GW9662 + anti-IgG (n = 4); α–PD-L1 + GW: anti-PD-L1 + GW9662 (n = 5). Data are mean ± SD.

Figure 7.

Model for effects of adipogenesis on antitumor immunity.