hide
Free keywords:
-
Abstract:
Growth factors are signaling molecules coordinating the complex functionality of multicellular organisms during development and homeostasis. Since aberrant expression of growth factors can cause diverse disorders such as cancer, autoimmune and cardiovascular diseases, growth factors and their receptors are central targets for therapeutic modulation. One of the options to manipulate signaling interactions is to use protein-based binders that are highly specific and able to target various molecular surfaces. Here, we present two different strategies of computational protein design to obtain inhibitors against growth factors which are key modulators of tumor progression. The first approach requires the structure of a native growth factor:growth factor receptor complex and aims to re-engineer a natural binding domain to make it more soluble, more stable, or more affine. In contrast, the second approach relies only on the structure of a target epitope and takes advantage of new software for massive-scale docking of a target site against a protein structure database to select the high shape complementary scaffolds. Adopting the first approach, we designed inhibitors of epidermal growth factor (EGF) using a single domain of EGF receptor as a template. Experimental evaluation of only two designed candidates revealed that both of them are solubly expressed, stable, and bind EGF with nanomolar affinities (i.e. 5-fold stronger than a native domain). Furthemore, we showed that one design inhibits EGF-induced proliferation of epidermoid carcinoma cells with IC50 of 0.5 nM. Using the second strategy, we designed inhibitors of vascular endothelial growth factor (VEGF) based on two different scaffolds. The binding affinities of the designs (16 candidates) to VEGF range from nano- to micromolar levels. X-ray structure determination of one of the candidates showed atomic-level agreement with the design model. Moreover, the best designs showed the ability to inhibit proliferation of VEGF-dependent cells. Thus, our results demonstrate the feasibility of the rational and generalizable approaches to design high-affinity protein binders against predefined conformational motifs.
