Ion channels regulate electrical activity in excitable cells and are key pharmaceutical targets for conditions such as cardiac arrhythmias and neuropathic pain. Despite their significance, achieving subtype selectivity remains a challenge in ion channel drug development, limiting the therapeutic applicability of existing modulators. Therefore, new molecular tools with strong potency and subtype selectivity are essential to advance ion channel therapeutics. Recent advances in deep learning now enable the de novo design of protein binders targeting clinically relevant proteins based solely on their structure. Such de novo protein binders hold promise as a new class of biologics with remarkable selectivity, stability, and versatility for therapeutic engineering. Concurrently, a surge in ion channel structural data provides a strong foundation for design efforts for these challenging targets. Here, we leveraged deep learning to design de novo protein modulators for ion channels, including KV1.3 for autoimmune diseases and SK2 for atrial fibrillation. Our strategies generated pore blockers and gating modulators, with final candidates expressed, purified, and tested via electrophysiology. Electrophysiological data demonstrate that our first-generation KV1.3 and SK2 pore-blocking binders exhibit nanomolar inhibitory potency. Subsequent optimization of lead KV1.3 candidates yielded a binder with low-nanomolar potency and over 10-fold selectivity against the homologous off-target channels. Notably, we reached these results by screening just 15 KV1.3 and 5 SK2 binders. We are currently performing experimental structural validation of these designs. Our results highlight the potential of de novo protein design in advancing ion channel-targeted therapeutics with potential enhanced efficacy and reduced side effects.
Mateos et al. (Sun,) studied this question.