The inability to accurately decipher the relationship between protein sequence and structural stability presents a major difficulty in predicting the effects of mutation on protein folding. We have analyzed this complicated relationship using rigorous high-throughput methods to explicitly test hypotheses that have been generated by previous de novo design studies and gain a more complete understanding of protein folding and structure. Our studies focus on the sequence-stability-function relationship of the four-helix bundle protein Rop through the means of combinatorial repacking of the hydrophobic core. Using a novel in vivo screen that utilizes GFP as a reporting phenotype, we are able to screen large libraries for functional variants, representing Rop mutants that are able to achieve a native-like fold. These functional variants from a modestly repacked library have been sequenced via high-throughput colony sequencing technology to accumulate a data set of over 200 unique Rop variants differing only in packing within a small part of the hydrophobic core. To gain insight into the thermodynamic consequences of core packing on stability, we have developed a high-throughput thermal scanning (HTTS) assay to assess the relative stabilities of the sequenced active variants (as well as a set of inactive variants for comparative analysis). This robust data set of sequence and stability information suggests that the packing of the hydrophobic core of Rop is quite lenient in regards to function, but more stringent in regards to stability and native-like structure. Interestingly, a large portion of the functional variants are molten globular in structure as a result of poor core packing, and large differences in stability are evident even with very small differences in primary sequence. These results suggest that packing of the hydrophobic core in stable, native-like proteins is akin to fitting together of a jig-saw puzzle, but that elements of oil drop-like behavior do exist to some extent. This has obvious implications for in silico effective energy functions (EEFs) aimed at structure prediction. To this end, we are currently using our empirical data to compare and contrast with different predictive EEFs. In addition, we are using biophysical methods such as circular dichroism, 1H-15N HSQC, and X-ray crystallography to provide detailed information of the structure of many of these interesting core variants.
Finally, we have also used the four-helix bundle model to examine other design-related phenomena, including examination of the effect of core packing on conformational specificity. This collaborative effort produced very interesting results that show that very small changes in core packing can produce dramatic consequences in protein folding. Rop has also provided an excellent proof-of-principle model system for developing an elegant method to examine protein-protein interactions in a robust, HT approach called TAG scanning. This technology, outlined in chapter 5, is extremely well-suited for cases in which little is known about the structure or binding partners of the target, providing a widely applicable method for the growing field of proteomics.