Science 357, 168–175 (2017)

Credit: AAAS

Most proteins have evolved for function rather than stability, which makes it difficult to ascertain the connections between sequence and folding and, in turn, confounds de novo protein design. Combining the expression of synthetic DNA libraries with proteolysis resistance assays, Rocklin et al. have now systematically determined the stability of over 15,000 designed miniproteins in a high-throughput fashion. Following the screen, each sequence was assigned a 'stability score', and a subset was validated by structural analyses including circular dichroism and NMR. In-depth analysis of the features of successfully designed stable miniproteins revealed how multiple determinants (such as buried nonpolar surface area) are balanced, leading to refinements in the design and modeling procedures. With iterative rounds of design, synthesis, and stability testing, the number of stable designs multiplied from 200 in the first round to over 1,800 in the fourth round, spanning four topologies, only one of which is found in natural miniproteins. Comparison to naturally occurring miniproteins further revealed that 774 of the designed miniproteins identified in this study outperformed the most stable natural miniprotein in the Protein Data Bank. This collection of thousands of new stable miniproteins with diverse topologies may themselves have applications in synthetic biology, and the insights gleaned from the way their global and sequence determinants contribute to folding will be useful for improving computational design strategies.