hide
Free keywords:
-
Abstract:
Despite the groundbreaking successes of computational design in recent years, simultaneous improvements of design throughput and accuracy are continually needed to achieve better experimental success rates, and tackle more difficult design problems. We describe Damietta 1; a novel protein design framework that maximises computational efficiency by tensorising energy calculations, and improves accuracy by relying on a self-consistent scoring function. This scoring function is not trained or contaminated by any learnt parameters, but relies purely on physics-based force field. We deploy these design concepts to tackle three design problems with different levels of difficulty, yielding agonists and antagonists of growth factor signaling pathways with therapeutic potential. First, we use Damietta to design epidermal growth factor (EGF) inhibitors based on an EGF receptor template structure. Testing only two designs, they were capable of binding EGF and inhibiting its signaling in cells. Second, we also use Damietta to create stabilised variants of the granulocyte-colony stimulating factor (G- CSF), which could proliferate G-CSF-dependent cell line.Third, we design a bispecific, single-domain cytokine, capable of engaging two different cytokine receptors (here, we start by a IL3-Rα/G-CSFR combination). Such a “novokine” possesses a novel fold, and can serve as a non-natural cytokine with novel function. These applications exemplify the design of proteins with therapeutic potential, and demonstrate Damietta to be applicable for a range of protein design and engineering problems.
