The advance of therapeutic proteins represents a revolution in clinical practice, but use of protein drugs requires consideration of their immunogenicity and potential to elicit an anti-biotherapeutic immune response (aBIR) in human patients. Such aBIRs can manifest a range of complications ranging from loss of efficacy to life-threatening anaphylactic shock, and mitigating immunogenicity is a key aspect of biotherapeutic development. While numerous factors influence protein immunogenicity, one critical feature is the source, i.e. proteins of non-human origin are disproportionately immunogenic. Given the immense therapeutic potential of foreign proteins, a variety of deimmunization strategies have been considered. Some methods, such as antibody humanization, are highly effective but limited to a narrow subset of protein classes. Others, such as conjugation to polyethylene glycol, are widely applicable but typically lead to a substantial loss of protein function. Modern protein engineering has enabled genetic approaches wherein immunogenic epitopes are deleted by site-directed mutagenesis, but current methods are costly, time and labor intensive, and have shown limited success. In contrast, computational tools for protein analysis are fast, efficient, and increasingly accurate. In this proposal, it is hypothesized that novel optimization algorithms can be leveraged to design protein variants that simultaneously reduce protein immunogenicity while maintaining high level functionality.
Aim 1 will test the hypothesis that for a given therapeutic protein, there exists a predictable and optimizable spectrum of trade-offs between the competing goals of reduced immunogenicity and high-level functionality. Ten engineered variants of a 2-lactamase therapeutic candidate (P992L) will be designed with a range of weights on the two objective functions: deimmunization vs. functionality. The P992L variants will be produced and assayed for activity, thermostability and immunogenicity.
Aim 2 will test the hypothesis that combinatorial protein libraries can be computationally optimized for functional therapeutic candidates. The optimization algorithms will be extended to enable design of deimmunized combinatorial protein libraries. Five libraries, having a range of relative weights for deimmunization vs. functionality, will be constructed and screened with a high throughput functional assay. The proportion of library members exhibiting high level functionality will be quantitatively determined, and the results will be benchmarked against the original library design parameters.
Aim 3 seeks to evaluate and refine the models of immunogenicity and functionality that underlie the objective functions of the design algorithm. An enhanced model of protein structure will be integrated into the design algorithm, and a new panel of P992L proteins will be constructed and experimentally evaluated. The results of these analyses will subsequently be used to update the optimization objectives and algorithm, and produce new variants, thereby closing the loop between computation and experiment. Successfully achieving these aims will yield broadly applicable algorithms for engineering powerful and immune-tolerant therapeutic proteins.
The advance of therapeutic proteins represents a revolution in clinical practice, but use of protein drugs requires consideration of their immunogenicity and potential to elicit an anti-biotherapeutic immune response in human patients. In this proposal, it is hypothesized that novel optimization algorithms can be leveraged to design protein variants that simultaneously reduce protein immunogenicity while maintaining high level functionality. Ultimately, such a technology could provide the medical community with access to a host of new immunotolerant protein drugs, redefining standards of care for cancer, drug-resistant viral and bacterial infections, strokes, heart attacks, auto-immune disorders, and other devastating diseases.
|Choi, Yoonjoo; Furlon, Jacob M; Amos, Ryan B et al. (2018) DisruPPI: structure-based computational redesign algorithm for protein binding disruption. Bioinformatics 34:i245-i253|
|Salvat, Regina S; Verma, Deeptak; Parker, Andrew S et al. (2017) Computationally optimized deimmunization libraries yield highly mutated enzymes with low immunogenicity and enhanced activity. Proc Natl Acad Sci U S A 114:E5085-E5093|
|Hua, Casey K; Gacerez, Albert T; Sentman, Charles L et al. (2017) Computationally-driven identification of antibody epitopes. Elife 6:|
|Choi, Yoonjoo; Verma, Deeptak; Griswold, Karl E et al. (2017) EpiSweep: Computationally Driven Reengineering of Therapeutic Proteins to Reduce Immunogenicity While Maintaining Function. Methods Mol Biol 1529:375-398|
|Griswold, Karl E; Bailey-Kellogg, Chris (2016) Design and engineering of deimmunized biotherapeutics. Curr Opin Struct Biol 39:79-88|
|Choi, Yoonjoo; Ndong, Christian; Griswold, Karl E et al. (2016) Computationally driven antibody engineering enables simultaneous humanization and thermostabilization. Protein Eng Des Sel 29:419-426|
|Blazanovic, Kristina; Zhao, Hongliang; Choi, Yoonjoo et al. (2015) Structure-based redesign of lysostaphin yields potent antistaphylococcal enzymes that evade immune cell surveillance. Mol Ther Methods Clin Dev 2:15021|
|Zhao, Hongliang; Verma, Deeptak; Li, Wen et al. (2015) Depletion of T cell epitopes in lysostaphin mitigates anti-drug antibody response and enhances antibacterial efficacy in vivo. Chem Biol 22:629-39|
|Verma, Deeptak; Grigoryan, Gevorg; Bailey-Kellogg, Chris (2015) Structure-based design of combinatorial mutagenesis libraries. Protein Sci 24:895-908|
|Salvat, Regina S; Parker, Andrew S; Choi, Yoonjoo et al. (2015) Mapping the Pareto optimal design space for a functionally deimmunized biotherapeutic candidate. PLoS Comput Biol 11:e1003988|
Showing the most recent 10 out of 19 publications