Fritzemeier, Kai;Renner, Steffen;Drepper, Friedel;Haehnel, Wolfgang - De novo protein design: combination of computational and chemical screening methods

ECCB 2002 Poster sorted by: Author | Number

Next | Previous poster (in order of the view you have selected)

Title: De novo protein design: combination of computational and chemical screening methods	P40
Fritzemeier, Kai; Renner, Steffen; Drepper, Friedel; Haehnel, Wolfgang haehnel@uni-freiburg.de Institut für Biologie II/Biochemie, Albert-Ludwigs-Universität Freiburg - Germany

The aim of protein design is to create novel proteins with tailored structural and functional properties. Extensive research efforts have provided a wide variability in biomimetic molecules. However, the lack in understanding the folding of proteins has limited the number of de novo protein variants. Recently developed methods for computational protein design can support a chemical screening for de novo proteins as pre-synthesis screen of a large virtual peptide library.

Computational design
The solution of the protein folding problem i.e. the correct prediction of the folded structure from the amino acid sequence has not yet been solved. De novo design of proteins is an opposite approach to the inverse folding problem. A partial solution has been found by design algorithms searching systematically for the best assembly of sidechain structures in a given fixed backbone target. The assumption of a fixed backbone dramatically reduces the complexity of the problem.
In our approach the sequence space is converted into a geometric space. To model the geometric specificity of sidechain placement the torsional flexibility of the amino acid side chain is represented as a set of discrete favourable conformations, the rotamers [1]. This assumption makes the search problem sizeable. An efficient way to handle the resulting large conformational space is the dead-end-elimination (DEE) [2,3] algorithm which discards all rotamers that are not part of the global minimum energy conformation (GMEC). In our implementation of different DEE-algorithms we use the nonbond energy terms of the CHARMM forcefield represented by the electrostatic potential and the Lennard-Jones potential as physico-chemical scoring function for possible structures. The energies are precomputed as interaction energies between sidechain and backbone and between pairs of sidechains. The values of interaction energy are stored into matrices on which the different optimisation procedures operate. This procedure leads to an significant decrease of calculation time.
If DEE converges, which means that only one rotamer remains per residue position, the resulting structure corresponds to the GMEC for the given backbone structure. In the other case we use the Branch-and Terminate algorithm [4] to search for the GMEC in the remaining residues. In our implementation both, DEE and Branch-and-Terminate are able to handle different forms of sequence symmetry which makes the synthesis of the received molecule significant easier. The calculation of the best sequences for a variety of slightly different backbone structures provides us a good starting point for a further exploration of the conformational space using our approach of a permutative screen through chemically synthesised peptide libraries.

Chemical synthesis of designed proteins
A modular strategy of template assembled synthetic proteins (TASP)[5,6] is used for chemical synthesis of the computer designed proteins. Up to four different secondary structure forming peptides are stepwise assembled on a cyclic decapeptide, the template, using thioether bonds. The use of template assembled proteins has three important advantages: 1. It is possible to synthesise pure proteins with more than 120 amino acids in chemical synthesis. 2. The template determines the position and the orientation of the secondary structure elements and therefore avoids problems resulting from an alternative global folding. 3. The template decreases the entropy Delta S of the system and increases the stability of the molecule.
A further advantage of template proteins is the possibility to bind the template reversibly to cellulose membranes [7]. With parallel spot synthesis we are able to synthesise several hundred different proteins by permutative combination of secondary structure forming peptides. This approach allows to explore the sequence space around a starting structure which is obtained by a sequence design algorithm. With a diode array spectrometer we have screened directly on the solid support a set of 400 heme binding proteins for differences in the redox potential [7] and another library for enzymatic activity. It is also possible to screen a protein library for different functions after exchange of non covalently bound cofactors. The screening for an optimal function of the target molecule leads to proteins optimised for that function. They are not yet optimised for a local tightly packed structure as in natural proteins. We hope that the combination of the computationally design and the combinatorial synthesis will help to find proteins with new function.

[1] Ponder, J. and Richards, F. (1987) Tertiary templates for proteins - use of packing criteria in the enumeration of allowed sequences for different structural classes. J. Mol. Biol. 193, 775-791.
[2] Dahiyat, B. and Mayo, S. (1997) De novo protein design: fully automated sequence selection. Science 278, 82-87.
[3] Desmet, J., De Maeyer, M., Hazes, B. and Lasters, I. (1992) The dead end elimination theorem and its use in protein side-chain positioning. Nature 356, 539-542.
[4] Gordon, B. and Mayo, S. (1999) Branch-and-Terminarte: a combinatorial optimisation algorithm for protein design. Structure 7, 1089-1098.
[5] Mutter, M., Altman, E., Altman, H.K., Hersperger, R., Koziej, P., Nebel, K., Tuchscherer, G. and Vuilleumier, S. (1988) The Construction of New Proteins. PartIII. Artificial Folding Units by Assembly of Amphiphilic Secondary Structures on a Template. Helv. Chim. Acta 71, 835-847.
[6] Rau, H.K. and Haehnel, W. (1996) De-Novo Design of Redox Proteins. Ber. Bunsenges. Phys. Chem. 100, 2052-2056.
[7] Rau, H., DeJonge, N. and Haehnel, W. (2000) Combinatorial Synthesis of four-Helix bundle Hemoproteins for tuning the cofactor properties. Angew. Chem. Int. Ed. 39, 250-253.