Background: Natural Killer (NK) cells have intrinsic anticancer activity that can be redirected toward acute myeloid leukemia (AML) with chimeric antigen receptor (CAR) engineering. Here, we study the functional consequences of CAR binding affinity and targeted epitope on CAR-NK cell activation, cytolytic synapse formation, and antitumor activity. Methods: We characterized NK-92 and primary NK cell populations expressing variant affinity AML-specific CARs containing single-chain variable fragments (scFvs, 26292 or 7G3) targeting two epitopes on CD123. 26292 affinity variants were discovered through directed evolution of an error-prone mutagenic library, while 7G3 affinity variants were previously reported. The resulting CAR-NK cell panel was studied with in vitro binding, activation, and cytotoxicity studies and in mouse xenograft models. Results: 26292 and 7G3 CARs of variable CD123 binding affinities were highly expressed in NK cells and conferred antigen-specific activation in vitro. High-resolution imaging demonstrated greater clustering of high-affinity 7G3 CAR-NK cells and consequent AML target cell death in a short-term time lapse. Low-affinity 7G3 CAR-NK cells exhibited enhanced antigen density discrimination with greater membrane-proximal signaling, cytokine production, and cytotoxicity. In longer-term assays, low-affinity 7G3 CAR-NK cells demonstrated more sustained killing of AML cells. In vivo testing highlighted greater expansion of low-affinity 7G3 CAR-NK cells in two xenograft models. Conclusions: Expression of 26292 and 7G3 CARs with a range of CD123 binding affinities in NK cells leads to antigen-specific activation and cytotoxicity against AML. Affinity-based differences in functional activation and antitumor activity are dependent on time course and are scFv/epitope specific.

Traumatic peripheral nerve injuries (PNI), present with symptoms ranging from pain to loss of motor and sensory function. Difficulties in intraoperative visual assessment of nerve functional status necessitate intraoperative nerve conduction studies (INCSs) by neurosurgeons and neurologists to determine the presence of functioning axons in the zone of a PNI. This process, also referred to as nerve “inching”, uses a set of stimulating and recording electrode hooks to lift the injured nerve from the surrounding surgical field and to determine whether an electrical stimulus can travel through the zone of injury. However, confounding electrical signal artifacts can arise from the current workflow and electrode design, particularly from the mandatory lifting of the nerve, complicating the definitive assessment of nerve function and neurosurgical treatment decision-making. The objective of this study is to describe the design process and verification testing of our group’s newly designed stimulating and recording electrodes that do not require the lifting or displacement of the injured nerve during INCSs. Ergonomic in vivo analysis of the device within a porcine model demonstrated successful intraoperative manipulation of the device, while quantitative nerve action potential (NAP) signal analysis with an ex vivo simulated “inching” procedure on healthy non-human primate nerve tissue demonstrated excellent reproducible recorded NAP fidelity and the absence of NAP signal artifacts at all points of recording. Lastly, electrode pullout force testing determined maximum forces of 0.43 N, 1.57 N, and 3.61 N required to remove the device from 2 mm, 5 mm, and 1 cm nerve models, respectively, which are well within established thresholds for nerve safety. These results suggest that these new electrodes can safely and successfully perform accurate PNI assessment without the presence of artifacts, with the potential to improve the INCS standard of care while remaining compatible with currently used neurosurgical technology, infrastructure, and clinical workflows.