Subcellular localization of PD-L1 and cell-cycle-dependent expression of nuclear PD-L1 variants: implications for head and neck cancer cell functions and therapeutic efficacy

The programmed cell death 1 ligand 1 (PD-L1)/programmed cell death protein 1 (PD-1) axis is primarily associated with immunosuppression in cytotoxic T lymphocytes (CTLs). However, mounting evidence is supporting the thesis that PD-L1 not only functions as a ligand but mediates additional cellular functions in tumor cells. Moreover, it has been demonstrated that PD-L1 is not exclusively localized at the cellular membrane. Subcellular fractionation revealed the presence of PD-L1 in various cellular compartments of six well-characterized head and neck cancer (HNC) cell lines, including the nucleus. Via Western blotting, we detected PD-L1 in its well-known glycosylated/deglycosylated state at 40-55 kDa. In addition, we detected previously unknown PD-L1 variants with a molecular weight at approximately 70 and > 150 kDa exclusively in nuclear protein fractions. These in vitro findings were confirmed with primary tumor samples from head and neck squamous cell carcinoma (HNSCC) patients. Furthermore, we demonstrated that nuclear PD-L1 variant expression is cell-cycle-dependent. Immunofluorescence staining of PD-L1 in different cell cycle phases of synchronized HNC cells supported these observations. Mechanisms of nuclear PD-L1 trafficking remain less understood; however, proximity ligation assays showed a cell-cycle-dependent interaction of the cytoskeletal protein vimentin with PD-L1, whereas vimentin could serve as a potential shuttle for nuclear PD-L1 transportation. Mass spectrometry after PD-L1 co-immunoprecipitation, followed by gene ontology analysis, indicated interaction of nuclear PD-L1 with proteins involved in DNA remodeling and messenger RNA (mRNA) splicing. Our results in HNC cells suggest a highly complex regulation of PD-L1 and multiple tumor cell-intrinsic functions, independent of immune regulation. These observations bear significant implications for the therapeutic efficacy of immune checkpoint inhibition.