De novo proteins are proteins designed from first principles, as opposed to natural proteins, or mutants of natural proteins. De novo proteins offer chemists and biochemists unique opportunities to understand and predict the structure of newly discovered proteins, and to perform functions that may be inaccessible with natural proteins or enzymes. The problem in designing de novo proteins is that there are far too many conformations available to polypeptides within a small energy range to predict structure with any confidence. Our approach toward de novo protein design largely circumvents the multi-minima problem by drastically reducing the number of available conformations. We do this by taking advantage of our ability to synthesize molecules, and in particular, we exploit the high rigidity and well defined structure of organic hosts such as cavitands, which can act as scaffolds. Such hosts can be used both for their hydrophobic binding pockets and as scaffolds to preorganize poly-peptidic units into a tertiary protein structure. Peptides are designed that will fold into alpha-helices that contain a hydrophilic and a hydrophobic face. Once attached to the scaffold the peptides form well-defined helical bundles.
The problem with de novo design is control. How can you be sure the structure will be helical? How can you be sure four helices will be involved, not three or five? How can you control the arrangement of the helices, parallel versus anti-parallel? Use of a rigid organic scaffold and organic synthesis, we solve these problems. The only remaining challenge is to create monomeric native-like structures. This means there should be no aggregation of the proteins, and their parts should be conformationally specific; there should be one predominant conformation.
We have designed peptide sequences that, when folded into an alpha-helix, will present hydrophobic and hydrophilic faces. For example, EELLKKLEELLKKG (E = glutamic acid, K = lysine, L = leucine, G = glycine) positions the five leucines (i.e., five hydrophobic sec-butyl groups) on one face. The other face contains the salt-bridged pairings of the positive and negative charges of the glu and lys.
We can link the peptides to the organic scaffold via a variety of linkages. Typically, we chloro-acetylate the N-terminus of the peptides and react this electrophilic group selectively with thiols from our cavitands or other templates/scaffolds. Alternatively, cysteines incorporated into the peptide allow us to attach the peptides at any position in the sequence, for example via disulfides formed with the template.
The peptides are readily made on our automated peptide synthesizer, cleaved from the resin, purified, and then coupled to the template. The ensuing protein is then purified and characterized. Most of our initial information comes from circular dichroism (CD), which crudely, is like a UV spectrum of chiral moieties. CD gives a measure of overall helicity. CD is a secondary structural determinant, and tells nothing of the tertiary structure or packing of the protein. We measure stability by monitoring the CD spectra in the presence of denaturants such as guanidine hydrochloride. We have found the bulk of our de novo proteins to be helical and highly stable.
The final question then is native-like structure. Do the proteins exist in single conformations? This is not only a challenge to design in, it is also difficult to diagnose. One method to explore native-like structure is NMR. Conformational specificity is marked by sharp signals, and a high dispersion of signals. This we have indeed seen in several of our proteins. Another method to probe for conformational specificity is exchange of N-H groups with D2O. This can be done easily by NMR simply by dissolving the protein in D2O and watching the NH signals disappear over time. We have had some NHs linger for weeks, which is on par with natural proteins. All in all, some of our best designs appear to be highly stable and native-like.
Current efforts are toward further understanding what makes a protein native-like, making less stable proteins, which are more like natural proteins, making anti-parallel bundles, and making catalysts. As for catalysts, our systems have great potential because we can design in virtually any functional group wherever we want, and our proteins are highly stable at high temperatures, where reaction rates will be enhance further still.
Adam R. Mezo, and John C. Sherman, J. Am. Chem. Soc., 1999, 121, 8983-8994.
Ashley S. Causton and John C. Sherman Bioorg. Med. Chem. 1999, 7, 23-28.
Bruce C. Gibb, Adam R. Mezo and John C. Sherman Tetrahedron Lett. 1995, 36, 7587-7590.