TL;DR
This study compares two common mouse models for HPV-positive oropharyngeal cancer (OPSCC) – subcutaneous (SC) and orthotopic (base of tongue, BOT) – using single-cell RNA sequencing. They found that SC models better mimic the early, immune-active phase of human OPSCC, while BOT models better reflect the later, immune-suppressed state, offering crucial guidance for selecting appropriate models for immunotherapy research.
What Problem Does This Paper Address?
Evaluating new cancer therapies, especially immunotherapies, relies heavily on preclinical models. For HPV-positive oropharyngeal squamous cell carcinoma (OPSCC), a cancer increasingly linked to HPV infection, understanding how these models mimic the human tumor immune microenvironment (TIME) is critical. The TIME is a complex ecosystem of immune cells that significantly influences cancer progression and treatment response. While both subcutaneous (SC) and orthotopic (base of tongue, BOT) mouse models are used, their immunological similarities and differences compared to human OPSCC, particularly across different stages of tumor progression, remain largely unclear. This lack of clarity hinders the accurate prediction of how potential therapies will perform in humans.
Methods
The researchers employed single-cell RNA sequencing (scRNA-seq) to analyze the immune cell populations within both SC and BOT murine models at different tumor stages (early and late). This allowed them to identify distinct immune cell types and their functional states. They also performed multiplex immunohistochemistry (mIHC) to validate their scRNA-seq findings spatially within the tumor tissues. To assess drug responsiveness, they analyzed the expression of inhibitory receptor genes on T cells and conducted in vivo therapeutic trials using immune checkpoint inhibitors (ICIs) like anti-PDCD1 and anti-CTLA4 in both mouse models.
Datasets
The study utilized two types of murine models:
- Subcutaneous (SC) murine models: Established by implanting HPV-positive OPSCC cells in the flank of mice.
- Orthotopic (BOT) murine models: Established by implanting HPV-positive OPSCC cells in the base of the tongue.
Additionally, they analyzed scRNA-seq data from human HPV-positive OPSCC samples (early and late stages) for comparative analysis.
Key Insights and Ideas
- SC models capture early immune activity: The SC models showed an increase in effector T cells as tumors grew, mirroring the initial immune response seen in early-stage human OPSCC. This suggests they are good for studying early anti-tumor immunity.
- BOT models reflect immune exclusion: In contrast, BOT models exhibited a decline in cytotoxic T cells and an increase in immunosuppressive myeloid cells, which is characteristic of advanced, "immune-excluded" human tumors. This makes them valuable for studying late-stage immune evasion.
- Distinct T-cell dynamics: While SC models showed increasing T-cell infiltration, they also accumulated immunosuppressive cells like Tregs and IL17A+ γδ T cells. BOT models, however, showed a progressive loss of cytotoxic T cells.
- Myeloid cells differ significantly: BOT models were dominated by M2-polarized macrophages, known for their immunosuppressive roles, whereas SC models had a more dynamic myeloid landscape.
- Differential ICI response: Crucially, the models predicted different responses to ICIs. Early-stage BOT models mirrored human responses to PDCD1 blockade, while late-stage SC models better reflected human responses to CTLA4 blockade. This highlights how model choice impacts drug sensitivity predictions.
Key Results
- scRNA-seq identified distinct immune cell heterogeneity in SC and BOT models, with SC models showing a stage-dependent increase in T-cell infiltration, while BOT models showed a decline.
- Human OPSCC samples displayed a lymphoid-dominant TIME with increasing T-cell infiltration during progression, more closely resembling the SC model's early stages.
- BOT models showed a significant accumulation of M2 macrophages and angiogenic neutrophils, contributing to an immunosuppressive TIME.
- In vivo drug studies revealed that BOT models responded better to PDCD1 blockade, while SC models responded better to CTLA4 blockade, aligning with observations in human OPSCC.
Limitations and Future Work
The authors acknowledge several limitations:
- Cell line specificity: The use of a single cell line (mEERL) might not fully capture the molecular heterogeneity of clinical OPSCC, potentially limiting the generalizability of findings.
- Low-abundance cell populations: scRNA-seq analysis of human samples captured limited numbers of certain immune cells (macrophages, DCs, neutrophils), which could affect the resolution of their analysis.
- Species-specific differences: While efforts were made to compare murine models to human data, inherent species-specific immunobiology and methodological differences in tissue acquisition can lead to quantitative discrepancies.
Future work could involve:
- Humanized models: Utilizing humanized mouse models or spatial transcriptomics to better recapitulate the complex lymphoid-tumor crosstalk seen in humans.
- Investigating resistance mechanisms: Further exploring why SC models, despite showing initial responses to PDCD1 blockade, might develop resistance due to the presence of suppressive T-cell populations.
- Broader ICI evaluation: Investigating responses to other ICIs like LAG3 and TIGIT, where the models showed different predictions compared to human responses.
Summary updated