The nuclear enzyme poly(ADP-ribose) polymerase (PARP), in cooperation with homologous recombination repair (HRR), plays a crucial role in repairing damaged DNA repair and maintaining genome integrity (Gibson and Kraus, 2012). Homologous recombination deficiency (HRD) renders tumors more vulnerable to the pharmacological inhibition of PARP, as it impairs the repair of DNA damage, leading to synthetic lethality (Bryant et al., 2005, Farmer et al., 2005). This pivotal concept underpins the clinical application of PARP inhibitors (PARPis) in cancer therapy (Curtin and Szabo, 2020, Singh et al., 2024), particularly ovarian cancer and breast cancer (Litton et al., 2018, Mirza et al., 2016, Moore et al., 2018). Furthermore, a growing body of preclinical research aims to expand the use of PARPis to other malignancies (Geng et al., 2023, Ma et al., 2023, Passiglia et al., 2023, Xiong et al., 2022). However, most clinical cases treated with PARPis inevitably develop resistance (Chiappa et al., 2021, Dias et al., 2021). As a result, numerous studies have focused on developing new combination regimens to overcome the clinical limitations of PARPis (Li et al., 2022, Madorsky Rowdo et al., 2024, Teng et al., 2024).

Beyond synthetic lethality, PARPis also harness immune responses to eliminate tumors. This is demonstrated by enhanced tumor-targeting immune responses following PARPis treatment (Xia et al., 2024) and reduced PARPis effectiveness after T-cell depletion (Pantelidou et al., 2019). Given the increased T cell infiltration and enhanced PD-L1 expression induced by PARPis (Shen et al., 2019), it is likely that immune checkpoint blockade (ICB) targeting PD-1/PD-L1 to reinvigorate dysfunctional tumor-infiltrated T cells could significantly enhance the anti-tumor capacity of PARPis. Unfortunately, the objective response rate (ORR) of PARPis in combination with pembrolizumab in either advanced triple-negative breast cancer (TNBC) (Vinayak et al., 2019) or recurrent ovarian cancer (Konstantinopoulos et al., 2019) cohort is under 20 %. Further research is needed to clarify the mechanisms behind resistance to PARPis, whether used alone or in combination with pembrolizumab, and to develop more effective combinatorial strategies.

In addition to T cells, natural killer (NK) cells are another major cytotoxic immune cell with a rapid and powerful non-specific killing capacity, playing a pivotal role in antitumor immune responses (Shimasaki et al., 2020). The function of NK cells can be modulated by ligands expressed on tumor cells, such as PD-L1 and major histocompatibility complex (MHC) molecules, including classic MHC and nonclassic HLA-E/G (Chiossone et al., 2018, Shimasaki et al., 2020). Specifically, PD-L1 binding to PD-1 blunts the tumor-killing effect of NK cells (Concha-Benavente et al., 2018, Hsu et al., 2018). MHC class I molecules facilitate antigen presentation to CD8+ T cells but suppress signal transduction via the immunoreceptor tyrosine-based inhibitory motif (ITIM) after binding to killer immunoglobulin-like receptors (KIRs) on NK cells (Bruijnesteijn et al., 2018, Raulet and Vance, 2006). HLA-E engages with CD94/NKG2A heterodimer to deliver inhibitory signals into NK cells (Carretero et al., 1997, Gornalusse et al., 2017, Kaiser et al., 2008), and blocking NKG2A has been shown to effectively delay tumor progression (Andre et al., 2018, Salome et al., 2022). HLA-G, originally identified as a key inhibitory factor in the context of pregnancy to protect the fetus from maternal immune attack, also plays a role in immune suppression in cancer (Ferreira et al., 2017). As the immunosuppressive functions of HLA-G in tumor immunology are increasingly recognized, HLA-G and its classic receptor ILT2 are emerging as potential novel immune checkpoints (Lin and Yan, 2021). Unlike PD-L1, which is broadly expressed in normal cells, the expression pattern of HLA-G is highly restricted, making HLA-G-based therapeutics more precise and less toxic. Antibodies targeting HLA-G or ILTs, either alone or in combination with existing immune checkpoint inhibitors, have become the focus of several clinical trials aimed at maximizing immunotherapeutic benefits.

In this study, we demonstrated that PARPis promote NK cell accumulation without enhancing their cytotoxic potential, as shown through analyses of clinical specimens. This effect was closely linked to a conspicuous elevation of HLA-G expression within tumor cells via an EGFR-AKT/mTOR-TFEB axis. The potential of a combined therapeutic approach involving PARPis and HLA-G blockade was further validated using humanized patient-derived xenograft (PDX) models. Collectively, our findings highlight a hitherto unexplored mechanism by which PARPis promote immune evasion and propose a promising combination strategy to improve tumor outcomes.