The potential of modular design for brand new proteins that do not yet exist in the natural world are the latest in a recent series of developments toward custom-designing proteins. Scientists with the University of Washington used the Berkeley Lab Advanced Light Source (ALS) for some of their research. They collaborated with Susan Tsutakawa, Greg Hura, and Kathryn Burnett, of the Molecular Biophysics & Integrated Bioimaging Division, who work at the SIBYLS beamline where some crystallography studies of repeat protein molecule structures took place.
Computational methods for protein structure have come a long way. Accuracy is extremely high (1.2 Å RMSD for carbon backbone) when a homologous structure can be found. However, for proteins with novel folds and designer proteins, if there is no high resolution structure by NMR or crystallography, it has been extremely difficult to know whether the prediction is correct and how accurate it is. In this landmark study to create new families of a helical folds, the Baker laboratory used a modular assembly process varying helix length and linker length to create over a hundred designs of novel proteins and tested these proteins for soluble expression. Eighty-three were expressed and showed helical content; 15 structures could be crystallized and matched the protein design, indicating only 20% accuracy of the models.
However, Tsutakawa, Hura, and Burnett, of the Tainer laboratory and the SIBYLS beamline at the ALS analyzed these structures by small angle X-ray scattering (SAXS). SAXS measures all the electron pair distances, providing a vector based structural analysis. A predicted scattering curve from the models could then be compared to the experimental data, using the robust VR algorithm created by Greg Hura. Based on this criteria, which uses the central region of the scattering curve, 43 models matched the experimental data, bringing the validation of the design system to over 50%.
In a few cases where the SAXS data did not match the design or the crystal data, they showed that the proteins were indeed forming oligomers in solution and in two cases for the crystallographic data, they were able to identify oligomers from the crystal lattice that matched the experimental solution data. Thus, this study showed the success of a helical protein de novo design, that the a helical fold protein universe is larger than what is currently known, and that SAXS is a powerful means of validating protein models and revealing their structures in solution. Read the full story on the University of Washington’s online NewsBeat.