Dissertation Title:  Tandem Display for Library-vs-Libary Screening

Abstract: 

Protein-protein interactions underpin a broad array of biological processes and represent essential targets for therapeutic development. Despite their central importance, established screening methodologies face substantial limitations regarding scope, throughput, and scalability. Established in vitro display methods, such as ribosome and mRNA display, can interrogate exceptionally large libraries (theoretically up to 10e15 variants) against single immobilized targets. However, these methods are inherently limited when attempting to simultaneously screen complex libraries against equally diverse target sets due to significant logistical and technical constraints.

As an example, consider the neuroscientist’s goal of a molecularly annotated connectome, which would require developing antibodies against every protein in the proteome of the model species. To individually express and purify tens of thousands of target proteins as is done in standard display approaches would be prohibitively costly and labor intensive. Additionally, in order to disambiguate which library members were bound to which targets, it would be necessary to conduct the screen in an arrayed fashion, significantly increasing the experimental complexity and sequencing costs. It would thus require well-funded and significant, coordinated effort to achieve such a goal just once, reducing its utility as a workday research tool.

To overcome these constraints, we propose a novel, ultra-high-throughput screening platform leveraging an enhanced form of cDNA display. This platform would enable concurrent interrogation of extensive protein interaction libraries irrespective of length, approaching an interaction complexity ceiling of 10e15 that is constrained predominantly by current DNA synthesis and sequencing technologies, rather than intrinsic technological barriers.

As an initial demonstration of the platform’s utility, we screen designed pairs of covalent peptide binders in the same manner as the SpyCatcher/SpyTag system. The characterization of an expanded palette of covalent peptide binders would materially contribute to diagnostics development, brain mapping, and combinatorial assembly of genetic circuits within synthetic biology, of direct utility for academics and industrial scientists.

Establishing such a broadly applicable, scalable protein-protein interaction screening technology would prove impactful across various scientific and therapeutic fields, including therapeutic nanobody discovery, pandemic prevention and planning, and detailed characterization of complex multi-protein or small molecule-dependent interactions. Further, the data generated by surveying the interactomic landscape of collections of proteins will address a significant gap in the training sets of protein structure prediction algorithms, which are largely trained on successful interactions of often-unknown affinity rather than unsuccessful ones. This technology thus has the potential to substantially advance capabilities within synthetic biology, protein engineering, therapeutic discovery and derisking, and chemical biology.

Committee members: 

Edward Boyden
Y. Eva Tan Professor in Neurotechnology
MIT
 
George Church
Robert Winthrop Professor of Genetics
Harvard Medical School
 
Alex K. Shalek
J.W. Kieckhefer Professor in IMES and Chemistry
MIT