17-β-estradiol reduces surface PD-L1 expression in estrogen receptor-positive breast cancer but not type 1 endometrial cancer cells.

Dear Editor, We report here that 17-β-estradiol (E2), through ERα (estrogen receptor α), inhibits IFN-γ-induced surface PD-L1 (programmed death-ligand 1, CD274) level in advanced ER+/HER2− breast cancer (BC) but not in triple negative BC (TNBC) or endometrial cancer (EC). This study implies that mTOR and MAPK pathways can reduce the surface PD-L1 level and diminish immune evasion via ERα. It also has important implications on endocrine resistance and the limitations of mTOR or MAPK-targeted therapies for treating advanced ER+/HER2– BC. Tumour cell surface PD-L1 binds to the PD-1 (programmed death 1, CD279) receptor on the surface of activated tumour-infiltrating lymphocytes (TILs), leading to tumour immune evasion. However, high-level surface PD-L1 is observed in ER− BC and EC tissues and the TILs1, 2 but rarely in ER+ BC tumors.2 Additionally, it remains unclear why ER+ patients are less responsive than those with ER− cancers to anti-PD-1/PD-L1 therapy. E2 functions primarily through the two receptors of ERα and ERβ in both cancers. Approximately 60%–65% of BC cases are ER+/HER2– luminal A subtype,3 and 80% of EC are of type 1 ER+ cancers.4 E2 reduces the IRF-1 protein and mRNA levels in murine splenocytes5 and MCF-7 cell line,6 respectively, implying that E2 may reduce the surface PD-L1 in both diseases. However, how E2 regulates tumour cell surface PD-L1 expression in ER+ BC and type1 EC is elusive. By qRT-PCR, BC cell lines MCF-7 (ERα+/PD-L1trace) and MDA-MB-231 (ERα−/PD-L1high), and EC cell lines Ishikawa (ERα+/PD-L1trace, type 1) and TEN (ERα−/PD-L1trace, type 2) were selected; flow cytometry was employed for this study. We examined the PD-L1 transcripts in a cohort of 25 EC specimens and detected a significantly higher PD-L1 transcriptional level in the type 2 EC group than in non-EC, but not for type 1 EC (Tables S1–S4 and Figures S1 and S2). Since IFN-γ is a major driver of surface PD-L1 expression through the IFN-γ-JAK-STAT-IRF1 pathway in multiple cancers,7 we exposed these cells to different concentrations of IFN-γ for 24 h and demonstrated that IFN-γ strongly increased the surface PD-L1 level in MCF-7, MDA-MB-231 and TEN. This did not happen to Ishikawa (Figure 1A–D), even after 3 days of exposure to 50 ng/ml IFN-γ (Figure 1G,H). Comparatively, MCF-7 showed a strong response, with the surface PD-L1 level peaking at 48 h and declining non-significantly at 72 h (Figure 1E,F). The same trend occurred after the cells were exposed to 10 ng/ml IFN-γ in a phenol-red free medium containing 5% CSFBS (charcoal-stripped fetal bovine serum) for 3 days (Figure 1I,J). We exposed the four cell lines to 10 ng/ml IFN-γ, 10 nM E2, and a combination of both in a phenol-red free medium containing 5% CSFBS during an eight-day culture. We showed that E2 alone did not alter the PD-L1 level on the surface of all four cell lines but greatly downregulated the IFN-γ-induced PD-L1 at the surface and transcriptional levels in MCF-7 co-exposed to E2 and IFN-γ, compared with IFN-γ exposure alone (Figure 2A–F), suggesting that it is not endogenous but the IFN-γ-induced surface PD-L1 that was downregulated by E2 in the advanced ER+/HER2− BC in vitro. As E2 affected ERα+/ERβ+ MCF-7 instead of the ERα−/ERβ+ MDA-MB-231, we used two ERα antagonists tamoxifen (Figure 3A) and ICI 182,780 (Fulvestrant) (Figure 3B), and the ERβ antagonist PHTPP (Figure 3C) to block both receptors, respectively, in MCF-7. We demonstrated that ERα contributed to the E2's downregulation at the surface and transcriptional (Figure 3D) levels. Since the PD-L1 level in the ERα−/ERβ+ A549 lung cancer cell line is strongly upregulated by IFN-γ via the JAK/STAT/IRF-1 pathway,8 our outcome was further validated in this line that ERβ was not involved (Figure 3E). Our finding accords with a newly published work by Hühn et al.,9 who showed that total E2 deprivation or fulvestrant treatment increased the surface PD-L1 expression in MCF-7 through ERα. Their report designated changes in MFI, however, we observed changes in the proportion (%) of the PD-L1 expressing cells. This suggests that E2 may protect ER+/HER2– BC by interacting with ERα and reducing the quantity of IFN-γ-induced surface PD-L1 expressing cells. Thus, such a protective role by E2 may slow the progression of ER+ BC's deterioration. This may explain why ER+/HER2– BC expresses less PD-L1, less malignance than the basal TNBC subtype, and less sensitivity to anti-PD-1/PD-L1 therapy than ER− BC. We confirmed that the JAK/STAT/IRF1 pathway regulates the IFN-γ-induced surface PD-L1 level (Figure 4,1 and 2A). Co-exposure to BEZ235 or U0126 combined with E2 and IFN-γ significantly restored the proportion of the E2-downregulated IFN-γ-induced surface PD-L1 expressing cells in MCF-7 (Figure 4,2C), but LY294002 or other kinase inhibitors (Table S5) were not capable (Figure 4,2B,C). Additionally, BEZ235 performed at a much greater recovery rate than U0126, significantly higher than those exposed to IFN-γ in MCF-7 (Figure 4,2D). This outcome implies that both the mTOR and the MAPK pathways cooperate in inhibiting immune evasion through ERα by reducing the IFN-γ-induced surface PD-L1 expression in ER+ BC. We found that tamoxifen or ICI 182,780 potentiates IFN-γ to upregulate surface PD-L1 expression, even when E2 is present in MCF-7 (Figure 3A,B). The higher the antiestrogen dose, the greater the PD-L1 surface expression (Figure 3A,B) as ICI 182, 780 inhibited ER in a dose-dependent manner.10 This indicates that prolonged administration of antiestrogens may induce tumour cell surface PD-L1 expression by disrupting a dynamic protective effect built by the E2-ERα-IFN-γ-mTOR-MAPK axis in the tumour milieu of the advanced ER+/HER2– BC patients, triggering tumour immune evasion. Together, these findings may provide causal reasons for antiestrogen resistance and the limited efficacy of mTOR or MAPK-targeted therapies for treating advanced ER+/HER2– BC patients. Additionally, the increased tumour cell surface PD-L1 induced by the inhibitors of ERα, mTOR or MAPK may potentially augment the sensitivity of ER+ BC cells to immune checkpoint inhibitors. E2 reducing the IFN-γ-induced cancer cell surface PD-L1 may be the main reason why high-level PD-L1 protein in ER+ BC tumours is rare. Antiestrogens or targeted therapies combined with anti-PD-1/PD-L1 regiments could be more beneficial in treating advanced ER+/HER2– BC patients. We thank Associate Professor Pamela Pollock (Queensland University of Technology, Australia) for the TEN cell line; Dr. Zhe Yang for the A549 cell line; Dr. Lili Wang and Dr. Jason Lynch for invaluable suggestions and help with FACS techniques; Dr. Lili Huang for helpful advice in statistical analyses; Dr. Bin Zhang (CSIRO, Australia), Dr. Shanli Zhu (Wenzhou Medical University, P. R. China), Mr. Yang Chen, Dr. Rasha Mosa and Dr. Rui Li for technical support. We thank the School of Biomedical Sciences Facility Center at The University of Queensland for supporting this study. This work was also supported by funding from NHMRC and scholarships from APA and UQGSS. The authors declare no conflict of interest. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.