Volsurf computational method applied to the prediction of stability of thermostable enzymes.

A computational model for the quantitative prediction of protein thermostability has been developed by means of the Volsurf method. A data set of 22 enzymes of reported thermostability in water systems, for the most part coming from thermophilic and hyperthermophilic organisms, has been built up. Molecular descriptors of the protein surface have been calculated and their role in the stabilization of the macromolecule has been analyzed by a multivariate statistical approach. The resulting regression model has shown a good predictivity and it has been able to quantitatively identify some structural requirements correlated with protein stability. The method can be the basis for a new computational support tool in rational protein design, which is complementary to the existing methods based on the sequence analysis.

Computational methods to rationalize experimental strategies in biocatalysis.

Computational methods are more and more widely applied in biocatalysis to gain rational guidelines, to orient experimental planning and, ultimately, to avoid expensive and time-consuming experiments. In this respect, molecular modelling, multivariate statistical analysis and chemometrics in general are useful computational tools, although they follow completely different investigative approaches.

Optimized polymer-enzyme electrostatic interactions significantly improve penicillin G amidase efficiency in charged PEGA polymers.

Hydrolytic yields as high as 80% were obtained by using penicillin G amidase (PGA) on substrates anchored on optimized positively charged PEGA polymers. By increasing the amount of permanent charges inside the polymer, electrostatic interactions between the positively charged PEGA(+) and the negatively charged PGA (pI = 5.2-5.4) were strengthened, thus favouring the accessibility of the bulky enzyme (MW = 88 kDa) inside the pores. The effect of different amounts of charges on polymer swelling and protein retention inside the polymer was investigated and correlated to the enzyme efficiency demonstrating that electrostatic interactions predominate over swelling properties in determining enzyme accessibility.

An innovative application of the "flexible" GRID/PCA computational method: study of differences in selectivity between PGAs from Escherichia coli and a Providentia rettgeri mutant.

The original GRID/PCA technique was adapted for the development of a tool potentially useful for the plan of a research strategy in rational enzyme design. The use of the MOVE directive of GRID made it possible to partially take into account protein flexibility, and the multivariate analysis was used as an instrument for focusing only on relevant information related to the differences in enzyme substrate selectivities. The comparison of two different penicillin G acylases, from Escherichia coli and from Providentia rettgeri, was used as a case study; these enzymes are very similar and their reported selectivities differ only for a couple of mutations around the active site. The "flexible" GRID/PCA method was able to correctly predict the observed selectivity differences caused not only by mutations of residues of the active site but also by long range effects on substrate selectivity due to sequence mutations on residues not directly involved in substrate recognition.

Nonswelling macroporous synbeads for improved efficiency of solid-phase biotransformations.

An application of novel, highly porous nonswelling resins (Synbeads) for enzymatic catalysis on solid supports is reported. These new resins combine easy handling of the beads, chemical stability, improved accessibility of proteins and higher productivity relative to swelling polymers. The present study demonstrates that the resin porosity greatly affects the efficiency in solid-phase biotransformations and that Synbead resins are valuable alternatives to swelling polymers for solid-phase chemistry and biocatalysis. The present study investigates the influence of key parameters, such as porosity and reactive functional-group density, on the reaction efficiency.

A homology model of penicillin acylase from Alcaligenes faecalis and in silico evaluation of its selectivity.

A three-dimensional model of the relatively unknown penicillin acylase from Alcaligenes faecalis (PA-AF) was built up by means of homology modeling based on three different crystal structures of penicillin acylase from various sources. An in silico selectivity study was performed to compare this homology model to the structure of the Escherichia coli enzyme (PA-EC) in order to find any selectivity differences between the two enzymes. The program GRID was applied in combination with the principal component analysis technique to identify the regions of the active sites where the PAs potentially engage different interactions with ligands. These differences were further analyzed and confirmed by molecular docking simulations. The PA-AF homology model provided the structural basis for the explanation of the different enantioselectivities of the enzymes previously demonstrated experimentally and reported in the literature. Different substrate selectivities were also predicted for PA-AF compared to PA-EC. Since no crystallographic data are available for PA-AF to date, the three-dimensional homology model represents a useful and efficient tool for fully exploiting this attractive and efficient biocatalyst, particularly in enantioselective acylations of amines.

GRID/tetrahedral intermediate computational approach to the study of selectivity of penicillin G acylase in amide bond synthesis.

Molecular modelling was used to investigate the catalytic site of penicillin G acylase (PGA) by building up a simple enzyme-ligand model able to describe and predict the enzyme selectivity. The investigation was based on a double computational approach: first, the GRID computational procedure was applied to gain a qualitative description of the chemical features of the PGA active site; second, a classical "transition state approach" was used to simulate the tetrahedral intermediates and to evaluate their energies. GRID calculations employed different probes which gave a complete description of the chemical interactions occurring upon binding of different ligands, thus indicating those structures having good affinity with the active site of the enzyme. Tetrahedral intermediates were constructed on the basis of GRID results and provided both geometrical features and energies of enzyme-substrate interaction. Such energies were compared to experimental kinetic data obtained in the enzymatic acylation of L-phenylglycine methyl ester using various methyl phenylacetate derivatives. The good agreement of computational results with experimental evidence demonstrates the validity of the model as a rapid and flexible tool to describe and predict the enzyme selectivity.