Search

Background OX40 has been widely studied as a target for immunotherapy with agonist antibodies taken forward into clinical trials for cancer where they are yet to show substantial efficacy. Here, we investigated potential mechanisms of action of anti-mouse (m) OX40 and anti-human (h) OX40 antibodies, including a clinically relevant monoclonal antibody (mAb) (GSK3174998) and evaluated how isotype can alter those mechanisms with the aim to develop improved antibodies for use in rational combination treatments for cancer.Methods Anti-mOX40 and anti-hOX40 mAbs were evaluated in a number of in vivo models, including an OT-I adoptive transfer immunization model in hOX40 knock-in (KI) mice and syngeneic tumor models. The impact of FcγR engagement was evaluated in hOX40 KI mice deficient for Fc gamma receptors (FcγR). Additionally, combination studies using anti-mouse programmed cell death protein-1 (mPD-1) were assessed. In vitro experiments using peripheral blood mononuclear cells (PBMCs) examining possible anti-hOX40 mAb mechanisms of action were also performed.Results Isotype variants of the clinically relevant mAb GSK3174998 showed immunomodulatory effects that differed in mechanism; mIgG1 mediated direct T-cell agonism while mIgG2a acted indirectly, likely through depletion of regulatory T cells (Tregs) via activating FcγRs. In both the OT-I and EG.7-OVA models, hIgG1 was the most effective human isotype, capable of acting both directly and through Treg depletion. The anti-hOX40 hIgG1 synergized with anti-mPD-1 to improve therapeutic outcomes in the EG.7-OVA model. Finally, in vitro assays with human peripheral blood mononuclear cells (hPBMCs), anti-hOX40 hIgG1 also showed the potential for T-cell stimulation and Treg depletion.Conclusions These findings underline the importance of understanding the role of isotype in the mechanism of action of therapeutic mAbs. As an hIgG1, the anti-hOX40 mAb can elicit multiple mechanisms of action that could aid or hinder therapeutic outcomes, dependent on the microenvironment. This should be considered when designing potential combinatorial partners and their FcγR requirements to achieve maximal benefit and improvement of patient outcomes.

Mouse syngeneic tumor models are widely used tools to demonstrate activity of novel anti-cancer immunotherapies. Despite their widespread use, a comprehensive view of their tumor-immune compositions and their relevance to human tumors has only begun to emerge. We propose each model possesses a unique tumor-immune infiltrate profile that can be probed with immunotherapies to inform on anti-tumor mechanisms and treatment strategies in human tumors with similar profiles. In support of this endeavor, we characterized the tumor microenvironment of four commonly used models and demonstrate they encompass a range of immunogenicities, from highly immune infiltrated RENCA tumors to poorly infiltrated B16F10 tumors. Tumor cell lines for each model exhibit different intrinsic factors in vitro that likely influence immune infiltration upon subcutaneous implantation. Similarly, solid tumors in vivo for each model are unique, each enriched in distinct features ranging from pathogen response elements to antigen presentation machinery. As RENCA tumors progress in size, all major T cell populations diminish while myeloid-derived suppressor cells become more enriched, possibly driving immune suppression and tumor progression. In CT26 tumors, CD8 T cells paradoxically increase in density yet are restrained as tumor volume increases. Finally, immunotherapy treatment across these different tumor-immune landscapes segregate into responders and non-responders based on features partially dependent on pre-existing immune infiltrates. Overall, these studies provide an important resource to enhance our translation of syngeneic models to human tumors. Future mechanistic studies paired with this resource will help identify responsive patient populations and improve strategies where immunotherapies are predicted to be ineffective.