Cervical cancer is one of the most common cancers affecting women, and the mortality-to incidence ratio (MIR) of cervical cancer varies between developed and developing regions, North America had the lowest MIR (0.36), followed by MIR in Europe (0.40), while Africa had the highest MIR (0.68). Infection with human high-risk papillomaviruses (HR-HPV) is the most important driver of cervical cancer, but after persistent HR-HPV infection, infected cervix cells still need many years to transform into cancer. Since 90% of failures in current treatments are related to drug resistance, the development of more effective therapies is urgently required. However, many new therapies have to be terminated because of the unsatisfactory drug efficacy at clinical phases, which reflects the current flaws of preclinical drug models. In this context, establishing precise in vitro models to screen out the therapeutic targets and to assess the new drug efficacy against cervical cancer is vital. In this study, based on patient-derived cervical organoids, multi-omic analysis including RNA sequencing and HLA-restricted peptidome sequencing was used to investigate dysregulated genes, biology processes, and HLA-presented epitopes in HPV-transformed and cancerous cervical organoids. Totally 6,515 HLA class I and 719 HLA class II restricted peptides were yielded, thereafter the immunogenicity of the epitopes was predicted with public machine learning-based tools. Interestingly, a large number of upregulated genes and highly immunogenic epitopes in HPV+ and cancer samples were associated with DNA repair. Thus, the role DNA repair-related molecules appear to play in cervical cancer therapies and prognosis was further explored. I first constructed a DNA repair-related gene score (DRGscore) with the public cervical cancer dataset of the Cancer Genome Atlas Program (TCGA-CESE). In the following, I observed a negative correlation between DRGscore and the prognosis of cervical cancer patients. The tumor microenvironment of patients with lower DRGscore contained more immune effectors and less immune suppressors, which suggested the negative correlation between DRGscore and sensitivity to immunotherapy. This finding could provide clues for patient medication guidance: high DRGscore patients might benefit from DNA repair inhibitors to reduce chemoradiotherapy resistance and gain a better prognosis. In contrast, low DRGscore patients rather might be sensitive to different immunotherapy including immune checkpoint inhibitors, and adoptive T cell therapy. Due to HPV infection and viral genome integration, HPV+ and cervical cancer cells are generally under stress conditions, which induces the intracellular accumulation of phosphoantigens (pAgs) that can be recognized by Vγ9Vδ2 T cells in an HLA-independent manner. The DNA repair activities in host cells are exploited by HPV for replication, which may provide some potential ligands recognized by Vγ9Vδ2 T cell receptors. To figure out our hypotheses, I investigated the Vγ9Vδ2 T cell cytotoxic effect against HPV-transformed and cervical cancer cells based on our cervical organoid model. Importantly, the cytotoxic effect was significantly enhanced against HPV+ and cancer organoids compared to healthy organoids in the presence or the absence of bromohydrin pyrophosphate (BrHPP), a synthetic phosphoantigen which activates Vγ9Vδ2 T. Moreover, the BrHPP dependent cytotoxic effect was found strikingly to be reduced to the level in non-activated Vγ9Vδ2 T group with the blockade of butyrophilin-subfamily members (BTN3A, BTN2A1). Besides, CD107a over expression was also associated with BrHPP stimulation. Both pieces of evidence indicated the BTN-subfamily members were mainly responsible for the cytotoxic effect with BrHPP activation and the self-active Vγ9Vδ2 T cells induced indiscriminate killing against HPV+, cancer cells but also healthy cells via degranulation including granzyme B (GzmB), platelet rich fibrin (Prf), interferon gamma (IFNγ) secretion. While the mechanism of the specific cytotoxicity against HPV+ and cancer organoids in the BrHPP absence condition still remained vague. I concluded based on the bioinformatic analysis of multi-omic data that the differential recognition of non-healthy cells by Vγ9Vδ2 T cells depends on multiple-ligand interactions, which should be further explored in the future. Overall, within my thesis project, stable patient-derived cervical organoid lines were established, which acted as a promising in vitro model for screening cervical cancer associated epitopes and for exploring T cell-induced cytotoxic effects. Thus, the organoid-T cell coculture model that I established could prove useful for testing additional modulators (e.g. small molecules, antibodies, tumor sensitizers, etc.). Consequently, my investigations on the Vγ9Vδ2 T-mediated killing effect against HPV-transformed cervical cancer cells have shed light on the potential future utilization of Vγ9Vδ2 T in clinical settings.
