147f De Novo Protein Design with Flexible Templates and Its Application to the Redesign of Complement 3a

Ho Ki Fung1, Christodoulos A. Floudas1, Dimitrios Morikis2, John D. Lambris3, and Li Zhang4. (1) Department of Chemical Engineering, Princeton University, Engineering Quadrangle, Olden Street, Princeton, NJ 08544, (2) Department of Chemical and Environmental Engineering, University of California at Riverside, A319 Bourns Hall, Riverside, CA 92521, (3) Department of Pathology and Laboratory Medicine, University of Pennsylvania, 401 Stellar Chance, Philadelphia, PA 19104, (4) Department of Chemistry, University of California at Riverside, University of California at Riverside, Riverside, CA 92521

De novo peptide or protein design starts with a flexible 3-dimensional protein backbone and involves the search for all amino acid sequences that will fold into such a template ([1]-[3]) and exhibit higher stability and/or higher activity. It is of paramount importance to designing improved inhibitors, engineering proteins with higher stability, conferring novel properties to catalytic sites of enzymes, and speeding up the drug discovery process [5].

Our current de novo protein design framework consists of two stages: (i) in silico sequence selection, and (ii) fold validation. We proved that our O(n2) in silico sequence selection model was superior in computational performance than several other O(n) formulations in literature [4]. Recently, we have devised two new sequence selection models that handle explicitly true protein backbone flexibility [6], as exhibited by multiple X-ray or NMR structures. One applied the single structure algorithm on a weighted average of the multiple structures, and the other employed binary distance bin variables to account for each of the structures, but imposed constraints on Cα position pairs (i,k) and (k,p) to eliminate solutions which would otherwise suggest physically impossible structures [6]. Both models were developed based on our discovery of an efficient way for linearizing the sequence selection model for single protein structure [6]. The models treat true backbone flexibility explicitly because they consider protein conformations of all possible combinations of continuous dihedral angle and Cα-Cα distance values between bounds determined by the X-ray or solution structures. As far as fold validation is concerned, we are applying a highly computationally efficient method, CYANA, on a full-atomistic scale driven by the AMBER forcefield, together with a local energy minimization package TINKER, for calculating fold specificities of the new sequences from the stage one models.

Confirmed by experimental data, our framework already proved to be highly successful in redesigning Compstatin, a 13-residue therapeutic peptide that inhibits Complement 3 and is used for treating unregulated complement activation ([1]-[2]). In this work, the framework will be extended to the design of a larger protein system, C3a. C3a is a 77-residue cleavage component of Complement 3 that is generated during the complement activation cascade. De novo design of C3a was based upon both the X-ray crystal structure elucidated by Huber et al. [7] and our own structures of the protein generated using molecular dynamics simulation [8]. The design was aimed at obtaining an antagonist that binds to the C3aR receptor better than C3a. In view of the multiple structures corresponding to the design template for C3a, the novel formulations using both a weighted average structure and binary distance bin variables were employed in the design work. Sequences for the antagonist were predicted and are being synthesized for experimental validation by our research collaborators.

[1]- J.L. Klepeis and C.A. Floudas and D. Morikis and C.G. Tsokos and E. Argyropoulos and L. Spruce and J.D. Lambris. "Integrated Computational and Experimental Approach for Lead Optimization and Design of Compstatin Variants with Improved Activity." J. Am. Chem. Soc. 125 (2003): 8422-8423.

[2]- J.L. Klepeis and C.A. Floudas and D. Morikis and C.G. Tsokos and J.D. Lambris. "Design of Peptide Analogs with Improved Activity Using a Novel de Novo Protein Design Approach."Ind. Eng. Chem. Res.43 (2004): 3817-3826.

[3]- C.A. Floudas and H.K. Fung and S.R. McAllister and M. Mönnigmann and R. Rajgaria. "Advances in Protein Structure Prediction and De Novo Protein Design: A Review." Chem. Eng. Sci. 61 (2006): 966-988 .

[4]- H.K. Fung and S. Rao and C.A. Floudas and O. Prokopyev and P.M. Pardalos and F. Rendl. "Computational Comparison Studies of Quadratic Assignment Like Formulations for the In Silico Sequence Selection Problem in De Novo Protein Design."J. Comb. Optim. 10 (2005): 41-60 .

[5]- C.A. Floudas. "Research Challenges, Opportunities and Synergism in Systems Engineering and Computational Biology." AIChE J.51 (2005): 1872-1884.

[6]- H.K. Fung and M.S. Taylor and C.A. Floudas. "Novel Formulations for the Sequence Selection Problem in De Novo Protein Design with Flexible Templates." J. Comb. Optim. (2006): accepted for publication.

[7]- R. Huber and H. Scholze and E.P. Paques and J. Deisenhofer. "Crystal Structure Analysis and Molecular Model of Human C3a Anaphylatoxin."Hoppe-Seylers Z Physiol. Chemie. 361 (1980): 1389-1399.

[8]- H.K. Fung and M.S. Taylor and C.A. Floudas and D. Morikis and J.D. Lambris. "Redesigning Complement 3a into Flexible Templates from Both X-ray Crystallography and Molecular Dynamics Simulation." (2006): in preparation.