I

Screening or selection for compacity and solubility

Fig. 9.6 (A) Schematic representation of binary patterning used for amphiphilic a-helix and b-strand. Polar and nonpolar residues are shown in dark grey and light grey; (B) Principle used for the combinatorial design of novel proteins from secondary structure modules.

Folded proteins

Fig. 9.6 (A) Schematic representation of binary patterning used for amphiphilic a-helix and b-strand. Polar and nonpolar residues are shown in dark grey and light grey; (B) Principle used for the combinatorial design of novel proteins from secondary structure modules.

tures. Thus, a simple binary code of polar and nonpolar residues arranged in the appropriate order can drive polypeptide chains to collapse into globular a-helical folds. Dozens of proteins from this initial library have been purified and characterized. They showed circular dichroism spectra characteristic of a-helical proteins, and some of them had native like properties, such as NMR chemical shift dispersion [52], cooperative chemical and thermal unfolding [53], and protection

of hydrogen exchange [54]. However most of the proteins from this initial library formed fluctuating structures and were probably molten globules. Using a typical molten globule-like protein from the original 74-residue library, a second generation library was constructed to increase length of the four helices (102-residue library) [55]. Biophysical characterization of five proteins by circular dichroism and NMR-measurements showed that stability and native-like properties were improved compared to the initial protein. This was confirmed by a structural study by NMR of a protein isolated from this improved library [56]. The experimentally determined structure was indeed a four-helix bundle as specified by the design (Fig. 9.7). It demonstrated also that the designed protein is not a molten globule and forms a unique structure.

In another study the Pliickthun group also used the approach of binary patterning to design novel proteins [57]. The length and amino acid composition of the modules were determined according to rules deduced from secondary structure elements observed in natural proteins. For example, a serine followed by a proline or a glutamic acid were used to design the N-terminal caps of a-helices while the helices themselves were encoded semi-randomly with a binary patterning of polar and nonpolar residues having a propensity to form a-helices. The building blocks were thus generated at the DNA level and randomly assembled until the average DNA fragment length, corresponding to proteins of about 100 amino acids, was reached. Several libraries were built to generate proteins with a mixture of a-helices and b-turns, a combination of b-strands and b-turns, or a combination of a-helices, b-strands and b-turns.

Arbitrarily selected clones from these libraries were tested for expression in E. coli. The proportion of clones with detectable expression was found to be between about 8% and 84% depending on the library. Quite a high fraction of proteins in these synthetic libraries were resistant to cytoplasmic degradation. However, the solubility, which is a characteristic of most of natural globular proteins, was moderate for expressible clones, with 10% to 60% of the proteins found in the soluble fraction. Further characterization by circular dichroism, size exclusion chromatography and sedimentation equilibrium experiments showed that some members from the all a-helices library were indeed helical, possessed a defined oligomerization state and showed cooperative chemical unfolding behavior. However, these proteins also showed properties consistent with a molten-globule state. By contrast, the all b-strand library led mainly proteins prone to aggregation. These results showed that an unexpected proportion of proteins built-up from structural secondary elements without b-strands possessed several of the favorable properties of natural proteins.

In order to avoid fastidious screening, it is an advantage to have a method to select folded proteins from large libraries of random proteins (see Section 9.3.1.1). Such an approach was developed to perform selections by ribosome display [41], based on two observations. Firstly, upon folding, a globular protein hides most of its hydrophobic residues. Secondly, once folded natural proteins often show a certain resistance to proteases due to their compactness. Thus, with a model system, the authors used a combination of hydrophobic interaction chromatography and proteolysis under limiting conditions and were able to select for a natural folded protein (a fragment of protein D from phage lambda) from a mixture including three other proteins previously obtained from the library of secondary structure modules. The potential of this selection approach is evident, and it will probably be applied to the search of folded proteins from combinatorial libraries.

Was this article helpful?

0 0

Post a comment