Rearranging and concatenating a native RTX domain to understand sequence modularity

Abstract

The use of repetitive peptide sequences forming predictable secondary structures has been a key paradigm in recent efforts to engineer biomolecular recognition. The modularity and predictability of these scaffolds enables precise identification and mutation of the active interface, providing a level of control which non-repetitive scaffolds often lack. However, the majority of these scaffolds are well-folded stable structures. If the structures had a stimulus-responsive character, this would enable the allosteric regulation of their function. The calcium-responsive beta roll-forming repeats in toxin (RTX) domain potentially offer both of these properties. To further develop this scaffold, we synthesized a set of RTX peptides ranging in size from 5 to 17 repeats, with and without C-terminal capping. We found that while the number of repeats can be altered to tune the size of the RTX face, repeat ordering and C-terminal capping are critical for successful folding. Comparing all of the constructs, we also observed that native configuration with nine repeats exhibited the highest affinity for calcium. In addition, we performed a comparison on a set of known RTX-containing proteins and find that C-terminal repeats often possess deviations from the consensus RTX sequence which may be essential for proper folding. We further find that there seems to be a narrow size range in which RTX domains exist. These results demonstrate that the deviations from the consensus RTX sequence that are observed in natural proteins are important for high-affinity calcium binding and folding. Therefore, the RTX scaffolds will be less modular as compared with other, non-responsive scaffolds, and the sequence-dependent interactions between different repeats will need to be retained in these scaffolds as they are developed in future protein-engineering efforts.