 |
MOLECULAR MODELLING OF ENZYMES: FROM 3D STRUCTURE TO INDUSTRIAL AND PHARMACEUTICAL APPLICATIONS
Lucia Gardossi and Paolo Braiuca, Dipartimento di Scienze Farmaceutiche, Università di Trieste
1. Selectivity
The Penicillin G Acylase (PGA) from E. Coli is a heterodimer of more than 700 residues. Its catalytic site can be divided into 2 subsite: an acylic cleft, very specific towards phenylacetic residues and an aminic subsite, much larger and less specific, capable of accepting a number of different substrates, especially aminoacids.
The selectivity of PGA has been studied by a double approach: firstly GRID was used to study the chemical features of the active site. The second step was the "in-site" substrate construction on the basis of GRID molecular field interactions, and the simulation of tetrahedral intermediates. Conformational analysis was performed by means of dynamics simulations and the AMBER force field, following the "classical tetrahedral intermediate approach". The results confirm what predicted by GRID, with aromatic carbon perfectly superimposed to the molecular interaction field (MIF) generated by aromatic carbon probe, the possibility of hydroxyl substitution on the phenyl ring and a bad interaction with nitro substituted substrates. This simple model can predict with a satisfactory agreement our experimental results.
2. Enantioselectivity
The combined use of GRID and tetrahedral intermediate construction has also provided explanation of the observed enantioselectivity of PGA towards L-enantiomers of aromatic aminoacids.
The MIFs generated by hydrophobic probe and carbonyl oxygen probe can easily describe the aminic subsite as divided into 2 different zones: a hydrophobic and a polar one (Fig. 1).
Figure 1. The aminic subsite can be divided into a hydrophobic zone (green) and a polar one (yellow)
The docking simulation and the conformational search of the tetrahedral intermediates show that L- aminoacids and their esters position their carbonyl group in the polar zone mentioned before, taking good interactions with a number of polar residues, in particular the Arg B:263, while their side chains find a room in the hydrophobic zone.
The steric feature of the D-enantiomers keep the carbonyl group from positioning in the polar zone and from interacting with the Arg B:263 and this causes a destabilization of about 6-7 kcal/mol.
On the other side the conformation of D- and L- enantiomers of aliphatic aminoacids is very similar, having in both cases the carbonyl located in the polar area, so that there is no enantiopreference as also experimentally confirmed.
3. GRID/PCA
In the perspective of improving enzyme performances and meet a specific chemical requirement we have considered also structures of PGA either mutants or coming from different microbiological sources.
The structure of a PGA from a mutant strain of Providencia Rettgeri was recently resolved. The enzyme seems to have an altered specificity, also if there are not clear experimental evidences. The combined use of GRID and Chemiometric analysis can fastly and accurately compare the selectivity of two or more enzymes (or receptors) and it can also point out any significant difference among the enzymes considered. The method, known as GRID/PCA, was developed by Prof. Cruciani and co-workers, at the University of Perugia, and it uses GRID output which is analysed by means of Principal Component Analysis (PCA).
The PCA score plot and loading plot allow to identifies the variable responsible for the different selectivity and to refold them into three-dimensional data (Figure 2).
The analysis pointed out a different selectivity towards hydrophobic and halogen probes due to a mutation of a Met into Leu.
Figure 2. One of the identified zone of different selectivity
The model can be recalculated using different GRID directives, taking into account, for example, the mobility of side chains.
4. Current Research
Our current research concerns many different topics. We are working on improving the simulation of different environments, since this is of major importance for our research because we perform biotranformations mostly in non-aqueous media. We have started to apply and evaluate the suitability (efficiency) of the methods available for adequate solvent simulation with the aim of identify a suitable method (for our needs) with a good precision/computational time ratio.
We are working also on the design of a homology model of the PGA from Alcaligenes Faecalis. Its 3D structure is not known yet, but we would like to verify the potential application of this enzyme in our experimental research. For this purpose our goal is to verify the differences in selectivity between the PGA from the two different sources.
For a more rational approach of the investigation of PGA catalised processes our research includes also the investigation of the reaction mechanism of PGA on quantum mechanical basis.
We intend to apply all the methods mentioned so far to a different hydrolase: the Glutaryl Acilase, which plays a fundamental role in the industrial production of cephalosporins. The enzyme belongs to the super-family of PGA but it is completely different as far as thesubstrate selectivity is concerned. On this aspects we have started a GRID/PCA comparison with PGA.
Finally, we are collaborating with prof. Spalluto and Dr. Moro (Padova) for the construction of a 3D-QSAR model of a series of A3 inhibitors, by CoMFA and GRID/Golpe methods.
|
