Bernard Busetta

An analysis of the prediction of secondary structures

Abstract An analysis of the prediction of the secondary structures was performed for 38 proteins of known structures, using the methods of Chou and Fasman (Biochemistry 13, 211–245) and Garnier et al. ( J. Mol. Biol. 120, 97–120). To improve the efficiency of the prediction we introduce a possible use of the distribution of the hydrophobic residues. The prediction level is clearly better for proteins with a single type of secondary structure (all-α or ail-β) than for proteins with a mixed type (α+β or α/β). A previous knowledge of the protein type greatly improves the prediction level. However, some information about this type may be inferred from the protein sequence. In those proteins for which two domains of different type can be defined the accuracy of the prediction appears better for the carboxyl terminal part than for the amino one.

Examination of folding patterns for predicting protein topologies

From the prediction of protein secondary structures, formation energies may be estimated for each incipient nucleus of the folding process. The averaging which may be performed on large families of distantly related proteins improves the accuracy of measurement of these energies and the efficiency of the prediction of the different steps involved in the protein folding (secondary structures, domain boundaries, topologies, etc.) and allows the description of new folding patterns.

The prediction of protein topologies

The estimates of the different contributions to the free energy of folded proteins are derived from analysis of protein X-ray structures, and introduced in the conformational analysis of the protein tertiary structures at a very macroscopic level. Different predictions of protein topologies are reported in the case of all-alpha and all-beta proteins.

Synthesis and CD studies of an 88-residue peptide containing the main receptor binding site of HTLV-I SU-glycoprotein.

Essential HTLV‐I biological functions, like host‐cell receptor recognition, depend on the structural motives on the surface glycoprotein gp46. We defined a peptide of 88 amino acids [Arg147‐Leu234] corresponding to the central part of the protein sequence, where major neutralizing epitopes are localized. After evaluating the feasibility of its chemical synthesis, the chosen sequence was realized using the stepwise solid‐phase methodology. Multiple chromatographic purification steps were required to obtain a sample suitable for structural analysis. Correct folding was supported by strong binding of monooclonal antibodies, recognizing known exposed immunodominant regions. Circular dichroism studies confirmed a non‐random conformation of at least 70–80% of the synthetic peptide. Investigation of the 3D‐structure of the synthetic peptide will provide useful information for future vaccine and drug‐design strategies.

Conformational analysis of melittin using the residual representation

The conformational analysis of polypeptide chains, for which the closing of the S-S bridges requires some special arrangement of the molecule, usually provides a small number of possible conformations. In [I], we described the use of the residual representation to search for the most probable conformations of apamin, a short polypeptide neurotoxin from bee venom, with two intramolecular S-S bridges. When such constraints do not exist, the number of starting conformations must be larger and the number of final possible conformations too. Here we intend to use the same residual representation to search for the most probable conformations of melittin, another peptide from bee venom which displays haemolytic activity. Melittin is a polypeptide chain of 26 amino acids, without any S-S bridge, the N-terminal part of which is a cluster of mainly hydrophobic residues (I-20) followed by a strongly basic C-terminal portion (21-26) [2]: be a continuous process; it is impossible to reach a final correct conformation by the simple energy minimisation of a single set of initial torsion angles (for instance all the residues in extended conformations a! = 220’). Thus, from a completely extended chain it is impossible to recover a real helical form through energy minimisation [ 31. Generally the mathematical minimisation of the complete energy function leads to different final conformations depending on the initial values given to the torsion angles. In the residual representation, each residue displays two most probable conformations: the extended form a! = 220’ and the helical form a! = 40’. To pass from one conformation to the other it is necessary to jump over an energy barrier. To study all the possible conformations of a polypeptide chain it would be necessary to assign each residue both possible conformations. For a protein of n residues the number of starting sets would be 2”. It is possible to reduce drastically this huge number, if in a first step we consider only the stable secondary structures.

The use of folding patterns in a multisolution approach of the all-ß protein three-dimensional structure. A ß-roll structure is predicted in the retroviral glycoprotein

An automatic macromolecular modelling package of unknown protein structures was developed using the intimate correlation which appears between the observed X-ray structures and their associated predicted folding patterns. The method can be considered as a generalization of both the combinatorial [1] and the template identification [2,3] approaches which were proposed some years ago, and provide a fast way of selecting structural motifs to build new proteins. As an illustration, the tertiary fold of the all-beta-domain of the retroviral outermembrane glycoprotein is proposed.

Improving the residual representation of proteins

Abstract A new, two-sphere, residual representation of protein is described, with a simulation of the different components of the correspondent energy, van der Waals interactions, hydrogen bonds at the level of the main chain, solvent interactions, and polar group interactions. Different figures of merit which may be used in conformational analysis of proteins are introduced. A study of bovine trypsin pancreatic inhibitor and homologous proteins illustrates the possibilities of the new representation in a multisolution process and the use of the different figures of merit in improving refined conformations.

Constrained Refinement Based on NOE and Chemical Shift Information: The Monomer Form of Arginine–Vasopressin‐like Insect Factor

Via the refinement process of the monomer form of an arginine–vasopressin‐like insect factor, the paper analyses the most relevant NMR information to define the solution structure of a flexible peptide. The relative importance of the different NOE constraints is discussed.