Scrutiny of chimeric antigen receptor activation by the extracellular domain: experience with single domain antibodies targeting multiple myeloma cells highlights the need for case-by-case optimization

Introduction: Multiple myeloma (MM) remains incurable, despite the advent of chimeric antigen receptor (CAR)-T cell therapy. This unfulfilled potential can be attributed to two untackled issues: the lack of suitable CAR targets and formats. In relation to the former, the target should be highly expressed and reluctant to shedding; two characteristics that are attributed to the CS1-antigen. Furthermore, conventional CARs rely on scFvs for antigen recognition, yet this withholds disadvantages, mainly caused by the intrinsic instability of this format. VHHs have been proposed as valid scFv alternatives. We therefore intended to develop VHH-based CAR-T cells, targeting CS1, and to identify VHHs that induce optimal CAR-T cell activation together with the VHH parameters required to achieve this. Methods: CS1-specific VHHs were generated, identified and fully characterized, in vitro and in vivo. Next, they were incorporated into second-generation CARs that only differ in their antigen-binding moiety. Reporter T-cell lines were lentivirally transduced with the different VHH-CARs and CAR-T cell activation kinetics were evaluated side-by-side. Affinity, cell-binding capacity, epitope location, in vivo behavior, binding distance, and orientation of the CAR-T:MM cell interaction pair were investigated as predictive parameters for CAR-T cell activation. Results: Our data show that the VHHs affinity for its target antigen is relatively predictive for its in vivo tumor-tracing capacity, as tumor uptake generally decreased with decreasing affinity in an in vivo model of MM. This does not hold true for their CAR-T cell activation potential, as some intermediate affinity-binding VHHs proved surprisingly potent, while some higher affinity VHHs failed to induce equal levels of T-cell activation. This could not be attributed to cell-binding capacity, in vivo VHH behavior, epitope location, cell-to-cell distance or binding orientation. Hence, none of the investigated parameters proved to have significant predictive value for the extent of CAR-T cell activation. Conclusions: We gained insight into the predictive parameters of VHHs in the CAR-context using a VHH library against CS1, a highly relevant MM antigen. As none of the studied VHH parameters had predictive value, defining VHHs for optimal CAR-T cell activation remains bound to serendipity. These findings highlight the importance of screening multiple candidates.