Intratumoral Hypoxia and ECM Remodeling Drives Tumor Immune Evasion in Patient-derived Tumor-immune Tunable Tissue Engineered Models of High-grade Serous Carcinoma

Introduction: : Ovarian cancer is considered the silent killer due to late diagnosis and recurrence and resistance to chemotherapy and immunotherapy, with an overall survival rate lower than 20% in the advanced stage. High-Grade Serous Ovarian Carcinoma (HGSC) is the most frequent and lethal subtype. More than 80% of patients with HGSC present immunologically cold tumors defined by exclusion of immune cells and overexpression of immunosuppressive molecules, even though the presence of tumor-infiltrated lymphocytes (TILs), such are CD4 and CD8+ T cells, in the ovarian tumor microenvironment (TME) correlated with better outcomes. The heterogenous and complex HGSC-TME is characterized by hypoxia (low oxygen levels) which alters cell behavior, causes resistance to therapies, and induces extracellular matrix (ECM) remodeling through activation of cancer-associated fibroblasts (CAFs). CAFs participate in the dysregulation of collagen secretion through TGF-β signaling leading to stiff ECM that acts as a physical barrier for the infiltration of TILs and immunotherapy diffusion into the TME. Current studies compare hypoxic TME (1.5% O2) with hyperoxic culture models (21% O2) even though physiologically relevant oxygen levels in the healthy ovary are around 7% O2. Therefore, there is an increasing need to develop biomimetic preclinical models that could recapitulate the physiologically relevant oxygen levels in normal ovary and hypoxic HGSC-TME. Our objective was to develop and validate physiologically relevant tissue immuno-engineered patient-derived 3D models to study immunologically cold HGSC-TME and identify possible therapeutic targets. In the absence of these models, the development of effective therapeutics strategies targeting hypoxic immunologically “cold” tumors would remain difficult.

Materials and

Methods: : A patient-derived 3D model was bioengineered using human plasma from HGSC patients based on crosslinking fibrinogen into fibrin and engineered to recreate physoxia and normal ECM in healthy ovary (nECMPhys) and hypoxia and aberrant tumor ECM remodeling in the HGSC-TME (tECMHyp) (Fig. A). Patient-matched HGSC dissociated tumor biopsies were cultured in the presence of CAFs in the 3D model forming a precision-based model. Autologous peripheral blood mononuclear cells (PBMCs) were challenged to infiltrate into the 3D scaffold or multi-cultured with HGSC and CAFs cells inside the scaffold to study tumor-stroma-immune interactions. The oxygen levels in the scaffolds were validated using optical sensor. The models were characterized by second harmonic generation (SHG) microscopy to detect collagen fibers, immunohistochemistry (IHC) for collagen I and α-SMA expression, western blot to study TGF-β signaling, and stiffness measurement by atomic force microscopy. An unbiased dimensionality reduction analysis by flow cytometry was used to study the activation and cytotoxicity of TILs when multi-cultured with HGSC and CAFs and to determine the role of intratumoral hypoxia and ECM remodeling in immune infiltration. Galunisertib, a TGF-β1 receptor inhibitor, was used to target the hypoxia-induced ECM remodeling. Human ovaries were decellularized by exposure to low-dose detergents. The decellularized ECM (dECM) was characterized by IHC and recellularized with HGSC and CAFs. Immune infiltration studies in dECM were evaluated by confocal microscopy.

Results, Conclusions, and Discussions:: The biomechanical properties of the model were optimized to recapitulate the hypoxic HGSC TME (0.7% O2, tECMHyp) when incubated in a hypoxia incubator (1.5% O2) or the physoxic ovary (7% O2, nECMPhys) when incubated in a standard incubator (21% O2). (Fig B). Collagen fibers were detected by SHG only in the presence of CAFs confirming their key role in ECM remodeling (Fig. C). Intratumoral hypoxia induced enhanced collagen I secretion and activation of CAFs (α-SMA) in their coculture with HGSC cells (Fig. D), and also a higher expression of TGF-β in tECMHyp vs nECMPhys (Fig. E). The hypoxia-induced ECM remodeling by CAFs correlated with significantly higher stiffness in tECMHyp when compared to nECMPhys (Fig. F). Moreover, intratumoral hypoxia enhanced the effector functions of the CD8+ TILs, by increasing activation markers (IL-2) and cytotoxicity (granzyme B, GrB), but also increased immunosuppressive T regs (CD4-FOXP3-IL-2+) (Fig. H). Immunophenotyping studies revealed significantly impaired immune infiltration of CD8+, CD4+, B cells, macrophages and myeloid-derived suppressive cells (MDSC) in tECMHyp when compared to nECMPhys (Fig. H). Pre-treatment with TGF-β receptor inhibitor rescued the impaired infiltration in tECMHyp (Fig.I), suggesting an implication of intratumoral hypoxia and CAFs in ECM remodeling and immune infiltration through TGF-β signaling. Our results were validated in dECM from human ovary that conserves native-like ECM composition, where intratumoral hypoxia impaired PBMCs infiltration and was also rescued by galunisertib (Fig.J). Critically, these results suggest that intratumoral hypoxia acts as a friend and foe in the HGSC-TME by enhancing CD8 activation and cytotoxicity to kill the tumor cells but also increases an immunosuppressive TME with aberrant ECM remodeling impairing immune infiltration. In conclusion, we have bioengineered and functionally characterized a biomimetic tunable tissue engineered 3D model to study the role of hypoxia and ECM remodeling in tumor-immune interactions. Our results are expected to have positive impact by providing a paradigm shifting understanding of intratumoral hypoxia when compared to physiologically relevant physoxic ovarian oxygen levels and its influence in immune evasion and cellular interactions, and turning immunologically “cold” tumors into “hot” ones by targeting the hypoxia-induced dysregulated ECM remodeling.

Acknowledgements and/or References (Optional):: Project supported by NIH/NCI R21CA259158.