Background Effective CD3 T-cell engagers (TCEs) must balance potency with potential toxicities, such as cytokine release syndrome. The CD3- and tumor-binding arms should function together to maximize tumor-cell killing while fine-tuning cytokine release. However, most TCEs in clinical development are derived from a small number of CD3-binding antibodies, such as SP34–2, limiting the pairs of parental antibodies that allow for optimal immunological synapse formation.

To address these barriers, AbCellera developed a complete TCE platform: a discovery engine to identify diverse antibodies with high specificity for tumor targets, a portfolio of fully-human, developable, CD3-binding antibodies that are distinct from SP34–2, a clinically-validated multispecific engineering platform, and a high-throughput strategy to rapidly identify TCE candidates.

Previously, we demonstrated that identical CD3-binding antibodies result in variable functional profiles when paired with different tumor targets, supporting the value of beginning with diverse parental antibodies. But the full extent to which CD3-binding parameters impact TCE function has not been well-understood due to the limited number of anti-CD3 antibodies that have been used to generate TCEs. In this study, we designed and engineered our largest panel of bispecific TCEs to date and used high-throughput functional assessments to gain novel insights into how CD3-binding properties impact function. We present here an overview of those insights, along with a demonstration of how we leverage these lessons to rationally select CD3-binders for different tumor targets.

Methods To deepen our understanding of CD3-binding parameters that impact function, we paired hundreds of CD3-binders with a single tumor-binding paratope. We assessed bispecifics using our high-throughput primary T cell functional assays, including tumor-cell killing and release of IFNγ, TNFα, IL-2, and IL-6. We performed a clustering analysis to identify distinct functional profiles, and aggregated these data with extensive analyses of our CD3-binders to gain insights into how CD3-binding properties impact function.

Results We compared TCE functional readouts across a range of monovalent CD3-binding parameters, including affinity, on- and off-rates, subunit specificity, and epitopes, revealing the impact of several parameters on TCE function. We then leveraged these insights, combined with our functional clustering analysis, to select diverse CD3-binding antibodies for multiple bispecific TCE programs.

Conclusions We integrated large parental and bispecific antibody datasets to gain insights into parameters that impact TCE function. By combining these insights with our previous data demonstrating the impact of tumor target on TCE function, we have developed a strategy to rationally select tumor- and CD3-binding antibody pairs for each tumor target.