CDK4/6 inhibitors and the pRB-E2F1 axis suppress PVR and PD-L1 expression in triple-negative breast cancer

Abstract

Immune-checkpoint (IC) modulators like the poliovirus receptor (PVR) and programmed death ligand 1 (PD-L1) attenuate innate and adaptive immune responses and are potential therapeutic targets for diverse malignancies, including triple-negative breast cancer (TNBC). The retinoblastoma tumor suppressor, pRB, controls cell growth through E2F1-3 transcription factors, and its inactivation drives metastatic cancer, yet its effect on IC modulators is contentious. Here, we show that RB-loss and high E2F1/E2F2 signatures correlate with expression of PVR, CD274 (PD-L1 gene) and other IC modulators and that pRB represses whereas RB depletion and E2F1 induce PVR and CD274 in TNBC cells. Accordingly, the CDK4/6 inhibitor, palbociclib, suppresses both PVR and PD-L1 expression. Palbociclib also counteracts the effect of CDK4 on SPOP, leading to its depletion, but the overall effect of palbociclib is a net reduction in PD-L1 level. Hydrochloric acid, commonly used to solubilize palbociclib, counteracts its effect and induces PD-L1 expression. Remarkably, lactic acid, a by-product of glycolysis, also induces PD-L1 as well as PVR. Our results suggest a model in which CDK4/6 regulates PD-L1 turnover by promoting its transcription via pRB-E2F1 and degradation via SPOP and that the CDK4/6-pRB-E2F pathway couples cell proliferation with the induction of multiple innate and adaptive immunomodulators, with direct implications for cancer progression, anti-CDK4/6- and IC-therapies.

Fig. 1. RB-loss correlates with reduced expression of immune hallmarks and increased RNA expression of CD274 and PVR in mouse and human breast cancer.

A GSEA comparing immune-related pathways in Rb^Δp53^Δ (red) with p53^Δ (blue) TNBC-like mammary tumors, visualized by Cytoscape. NES normalized enrichment score. B Correlation analysis of immune markers (right labels) with molecular signatures (bottom labels) in patient samples of 1302 mixed breast cancer and 205 TNBC. C Correlation level of CD274 (PD-L1) and PVR with indicated RB-related pathways. D Summary of ENCODE analysis for the PVR and CD274 promoter region (see Supplementary Fig. S3A). *P < 0.05, **P < 0.001, ***P < 0.0001, ^&P = 0.0659.

Fig. 2. Direct Regulation of PVR and PD-L1 by pRB and E2F1.

A Immunoblot validation of E2F1 expression following adenovirus-mediated transduction of E2F1 (Ad.E2F1) in indicated TNBC lines. B Immunoblot analysis of glycosylated and non-glycosylated PVR in multiple TNBC lines transduced with Ad.GFP or Ad.E2F1, or C depleted for RB1 via RNAi. D Immunoblot analysis of PVR response to stable depletion of RB1 by shRNA. E Left, immunoblot analysis of PD-L1 in RB1-deficient TNBC lines transduced with Ad.GFP or Ad.E2F1. Right, cumulative PD-L1 quantification normalized to ponceau (black) or to GAPDH (green). F Immunoblot analysis of PD-L1 in three different RB(+) TNBC lines following acute knockdown of RB1 via RNAi, 3 days (D3) post transfection. G Immunoblot analysis of PD-L1 in TNBC lines with stable RB1 knockdown. H Immunoblot analysis of PD-L1 in RB1-deficient (MDA-MB-468) or RB1-depleted (HCC38-shRNA) TNBC cell lines transduced with Ad.RB1. Numbers above immunoblots denote band intensity normalized to loading control.

Fig. 3. The CDK4/6 inhibitor palbociclib suppresses PVR and PD-L1 expression, and counteracts the effect of SPOP depletion on PD-L1.

A Immunoblot analysis showing the effect of 2-day palbociclib treatment on ngPVR and gPVR expression. B, C Immunoblot analysis of PD-L1 in TNBC lines treated with increasing (B) or high (C) concentrations of palbociclib for 2 days. D Immunoblot analysis of PD-L1 in TNBC lines with stable knockdown of RB1, treated with/without palbociclib for 2 days. E Immunoblot analysis of PD-L1 in livers of mice on mixed background treated with palbociclib (140 mg/kg; gavage; 5 consecutive days/week) for 28 days and probed with AF1019 antibody (left) or in livers of C57BL/6 mice treated with palbociclib (140 mg/kg; gavage) for 7 consecutive days, probed with MAB90781 antibody (right). Bottom, PD-L1 quantification. Pure palbociclib (FA65120) was suspended in sodium lactate (50 nM, PH 4.0), which was used in the vehicle control mice. F Heatmap analysis of RNA-seq data (GSE177054) of MDA231 treated with ribociclib or ribociclib/D4476 for 2 days. White boxes denote out-of-range intensity. G Immunoblot analysis of PD-L1 in indicated tumor cells following knockdown of SPOP by RNAi over 2 (D2) or 3 (D3) days. H Immunoblot analysis of PD-L1 in indicated TNBC lines with/without SPOP knockdown after 2 days (D2) and treated with/without palbociclib for 24 h. I Immunoblot analysis of PD-L1 and SPOP in indicated TNBC lines after palbociclib treatment for 48 h. Numbers above immunoblots denote band intensity normalized to loading control.

Fig. 4. Effect of confluency, palbociclib treatment duration, hydrochloric and lactic acids on PVR and PD-L1 expression in TNBC cells.

A Immunoblot analysis of PD-L1 in TNBC lines seeded at various densities in 6-cm plates and treated with 1 µM palbociclib for ~2 days. Bottom, PD-L1 quantification of (A) and Supplementary Fig. 7A. Statistics calculated by pairwise analysis of seeding densities and treatment durations. B Immunoblot analysis of TNBC lines treated with various concentrations of hydrochloric acid (HCl) for 2 days. Right, PD-L1 quantification and analysis by Welch’s t-test. C Immunoblot analysis of MDA231 treated with/without palbociclib and/or with/without indicated solvents. D Immunoblot analysis of PD-L1 and PVR in different TNBC lines treated with physiological concentrations of lactic acid for 2 days. Numbers above immunoblots denote band intensity normalized to loading control. E A model suggesting that in response to mitogenic signals, CDK4/6 controls PD-L1 turnover by inducing its degradation via Culin3^SPOP and its transcription via pRB-E2F1. PD-L1 in red and green depicts old or newly synthesized glycosylated protein, respectively. F A model suggesting that the CDK4/6-RB-E2F axis modulates immune surveillance by transcriptionally regulating PD-L1, PVR and potentially other IC modulator genes. Solid and segmented arrows indicate validated (herein) and unvalidated targets, respectively; single arrows denote pRB and E2F1 binding sites as well as strong E2F1 recruitment by ChIP-seq analysis, whereas double arrows denote no or weak E2F1 recruitment.