Detecting native-like properties in combinatorial libraries of de novo proteins

Abstract

Combinatorial methods based on binary patterning of polar and nonpolar residues have been used to generate large libraries of de novo alpha-helical proteins. Within such libraries, the ability to find structures that resemble natural proteins requires a rapid method to sort through large collections of proteins and detect those possessing 'native-like' features. The current paper presents such a method and applies it to an initial collection of de novo proteins. We present a method to identify proteins with native-like properties from libraries of de novo sequences expressed in vivo. A novel 'rapid prep' freeze/thaw procedure was used to prepare samples; chromatographic purification was not required. The semi-crude samples were analyzed for native-like features by one-dimensional 1H NMR spectroscopy. Using this method, we demonstrate that native-like features can readily be observed for several proteins among a collection of sequences designed by binary patterning of polar and nonpolar amino acids. Native-like properties can be detected using a method that requires neither isotopic enrichment nor chromatographic purification. The method is inexpensive, rapid, and suitable for parallel processing. It can therefore be employed to screen for native-like properties among large collections of de novo sequences. Using this method, we demonstrate that although the binary code strategy does not explicitly design tertiary packing, it can nonetheless generate proteins that possess native-like properties. The use of combinatorial methods to produce large collections of proteins coupled with the availability of a rapid assay for detecting native-like properties will facilitate the design and isolation of novel proteins with desirable properties.