Improving computational protein design for real-world applications

From the Stoddard Lab, Basic Sciences Division, and the Bradley Lab, Public Health Sciences Division

Drugs – for much of human history we have dedicated ourselves to the task of finding new and creative ways to package and deliver them in order to treat our myriad maladies. There are, of course, the old classics - we’ve ingested them, smoked them, injected them, rubbed them on our skin. But with the explosion of modern biotechnology has come a profusion of ever more creative strategies. We’ve packaged them in viruses and then intentionally infected ourselves. We’ve loaded them onto tiny robots to deliver them with surgical precision. Perhaps the most famous recent case is the mRNA technology underlying the development of the COVID-19 vaccines. Drs. Barry Stoddard and Phil Bradley, Professors in Fred Hutch’s Basic Sciences and Public Health Sciences Divisions, take a different approach – designing synthetic proteins that can act as scaffolds for drug display and delivery.

Though they are initially synthesized as linear strings of amino acids, the proteins in our cells quickly fold themselves into complex three-dimensional shapes that are crucial for their functions. Deciphering how the sequence of a protein informs the shape it will adopt has been a perduring challenge in biochemistry. However, recent progress in computational power and predictive algorithms based on artificial intelligence have led to major advances in this area. And better understanding the relationship between sequence and structure has led to the possibility of computationally designing new proteins with interesting and useful features. “The use of computational protein design to create new biomolecules leads us to question what additional types of protein folds, including those that have never been seen before in nature, might be designed in silico and then adopted for use in real life,” said Dr. Stoddard. In an exceptional example of collaborative synergy, Dr. Stoddard, an expert in structural biology, has teamed up with Dr. Bradley, an expert in computational protein design, to bring this vision to life. Their labs are particularly focused on a class of synthetic proteins called circular tandem repeat proteins (cTRPs) – repeated protein sequences that self-assemble into large circular structures. “[cTRPs] can display significant stability and solubility, a wide range of sizes, and are useful as protein display particles for biotechnology applications. However, cTRPs also demonstrate inefficient self-assembly from smaller subunits,” they wrote in a recent article in Communications Biology that reports a new class of cTRPs with improved self-assembly capabilities.

cTRPs form from the linking together of several smaller protein subunits. The relatively poor self-assembly of previous generations of cTRPs, the authors noted, resulted from “relatively limited contacts and small surface areas that are involved in packing between…subunits.” They therefore sought to design a new, bulkier class of cTRPs, which they termed thick cTRPs (tcTRPs), that would provide more contact area between subunits. This work began with computational models – the authors used a molecular modeling software called Rosetta to design subunits that were predicted to have improved assembly. The group then selected two promising designs of different sizes – one with 9 subunits (tcTRP9) and one with 24 subunits (tcTRP24) - for which they synthesized the proteins and directly assessed their assembly characteristics. Analysis of the tcTRP9 proteins’ mass, thermal stability, and crystal structure validated that it did indeed assemble as the authors predicted. The tcTRP24 protein required some tweaking of its original sequence to improve the accuracy of assembly, but the authors ultimately generated a protein that they determined by Cryo-Electron Microscopy to assemble correctly with >95% accuracy.

Having succeeded in developing tcTRPs with improved assembly, the authors finally sought to test whether these proteins could be functionalized – a process in which the subunits are fused to effector protein domains to allow them to influence some biological function. They first fused SH2 protein domains to their tcTRPs and found that these domains maintained their sequence-specific phospho-tyrosine peptide binding function. As a second test, they fused a modified antibody, called a nanobody, against SARS-CoV-2 and found that this conferred the particles with a potent ability to neutralize the virus.

These findings, Dr. Stoddard explained, illustrate how “new designed protein fold…can be used for the arrangement and display of functional protein domains for use in biotech and/or biomedical applications.” Looking forward, he is excited to explore the potential of this approach to computational protein design. “Are there limits to this technology, and if so, what are they? Are certain types of protein folds that look nice on a computer screen nonetheless impossible for a peptide chain to form? We continue to push the capabilities of computational protein design, as well as our ability to arm our designed molecules with functional proteins (binding domains, enzymes, signaling molecules) to see what is possible and useful.”

tcTRP structure
structure of the circular tcTRP9 protein fused to three anti-SARS-CoV-2 nanobodies. Image provided by Dr. Barry Stoddard.

This work was supported by the National Institutes of Health and Fred Hutchinson Cancer Research Center.

Fred Hutch/UW Cancer Consortium members Barry Stoddard and Phil Bradley contributed to this work

Hallinan JP, Doyle LA, Shen BW, Gewe MM, Takushi B, Kennedy MA, Friend D, Roberts JM, Bradley P, Stoddard BL. Design of functionalised circular tandem repeat proteins with longer repeat topologies and enhanced subunit contact surfaces. Commun Biol. 2021 Oct 29;4(1):1240. doi: 10.1038/s42003-021-02766-y. PMID: 34716407; PMCID: PMC8556268.