Computational De Novo Protein and Peptide Design

Although this symposium was scheduled for Wednesday afternoon, we had sizeable interest and attendance due to the significance of the topic and excellence of the speakers. The emphasis on the development of biologics as drugs has increased dramatically over the last few years. In fact it is well accepted that biologic drugs will drive pharmaceutical future market growth. For example, biologics represent over 30 percent of Abbott’s development pipeline, while Roche, obtained two-thirds of its pharma sales in 2010 from biologics (a report by MedCity News, November 20, 2012).

Biologics are specifically being developed for use in cancer, diabetes, rheumatoid arthritis, and orthopedics. Examples include Abbott’s Humira, Amgen’s Enbrel, and Biogen Idec’s Rituxin as representative best-selling biologics.

It is with this in mind that Francisco Hernandez-Guzman (Accelrys, United States) began our session with “Novel in silico prediction algorithms for the design of stable and more effective proteins.” Dr Hernandez-Guzman discussed new computational tools available from Accelrys to calculate the energetic effects of mutations on protein stability and protein-protein binding affinity including pH dependence, and use of these methods for successful protein design. He presented a validation example for the method examining binding affinity of OMTKY3 inhibition of proteinase B residues with 19 P1 residue mutations, and an in silico design of an IgG antibody with enhanced pH selective binding to improve its serum half life.

Woody Sherman (Schrödinger, United States) followed with a discussion of “Advanced structural modeling of biologics with BioLuminate.” Bioluminate is Schrödinger's new product that provides for computational protein-protein docking (Piper), antibody modeling, long loop predictions and protein mutational studies. Dr. Sherman presented examples illustrating prediction of thermal stability and mutational effects on protein-protein binding energy with the barnase/barstar complex; thermal stability of SH3 domain mutants; and CDR antibody modeling predictions and de novo approaches to H3 loop modeling.

Enrico Purisima (National Research Council of Canada) rounded out the session with a presentation on using computational methods of directed evolution to enhance affinity maturation. Dr. Purisima’s lab has used a combination of three in silico methods (SIE, FoldX and Rosetta) to redesign antibodies to enhance their interactions and affinities to VEGF-A and HER2. The designed antibodies were then tested with experimental screening methods using surface plasmon resonance, which resulted in redesign of the binding interface.

The symposium demonstrated the usefulness and potential contribution of computational methodologies to modeling, predicting, and redesigning protein-protein interactions with applications to the design of biologics (antibodies).

Rachelle Bienstock, Symposium Organizer