BACKGROUND

Ferroptosis has been implicated in the regulation of the tumor immune environment; however, its precise effect on immune checkpoint inhibitors remains contradictory.

OBJECTIVE

To elucidate the "double-edged sword effect" of a key ferroptosis-related factor in regulating the immune microenvironment.

METHODS

This study utilized single-cell RNA sequencing (scRNA-seq) analysis to characterize the tumor microenvironment in ovarian cancer samples from immunotherapy cohorts. Following quality control and variable gene screening, data from The Cancer Genome Atlas (TCGA), Genotype-Tissue Expression (GTEx), GENE EXPRESSION OMNIBUS (GEO), bulk, and spatial transcriptome databases were analyzed. The AddModuleScore_UCell function was employed for gene set scoring by evaluating the expression patterns of specific gene features in single-cell datasets, which were found to correlate with interactions between tumor cells and stromal cells, recognized as key contributors in the immunosuppressive milieu. Immunohistochemistry, western blot, and multiplex immunohistochemistry (mIHC) analyses were employed to explore the HMOX1/TGF-β1/PI3K/AKT/NF-κB(p65) signaling pathways. In vitro findings were further validated in a mouse model. The correlation between risk factors and progression-free survival (PFS) was analyzed using Cox regression and Kaplan-Meier methods.

RESULTS

We demonstrated decreased expression of the ferroptosis-activating gene HMOX1 in ovarian cancer epithelial cells, while being upregulated in macrophages. Ovarian cancer (OV) epithelial cells with HMOX1 inhibition could secrete TGF-β1 to activate three macrophage subtypes: SPP1+, FOLR2+ and C1QC+ via the PI3K/AKT/NF-κB (p65) pathway. The up-regulation of HMOX1 in macrophages also activated these three macrophage subtypes via the NF-κB pathway. Both pathways simultaneously inhibited Cytotoxic T Lymphocyte (CTL) activation and contributed to the immunosuppressive microenvironment of ovarian cancer, as demonstrated in both in vitro and in vivo models. Targeting HMOX1 alone, whether through activation or inhibition, was only effective in modulating a single pathway while simultaneously inducing negative feedback on the opposing pathway, demonstrating the "double-edged sword effect" of HMOX1 in regulating the immune microenvironment.

CONCLUSION

Overall, we proposed and validated two strategies targeting HMOX1 to improve the efficacy of PD-1 inhibitors, and confirmed that HMOX1, TGF-β1, SPP1, FOLR2, and C1QC could be used to construct models predicting the efficacy of immune checkpoint inhibitors.