Since the report of the three-dimensional (3D) structure of whale myoglobin solved by John Kendrew, protein structure has been a key piece of information for biological sciences. The knowledge of the structure allows us to know how a chain with a defined sequence of residues is folded in space. Usually, protein folding results in clusters with a high number of residues giving rise to protein domains. The domains are very stable structures which contain some highly conserved residues allowing the protein to exert its activity. Thus, these macromolecules are able to work as catalists of chemical reactions, as carriers for small molecules, as the generators of mechanical movement and to perform even more complex functions such as the communication between cells. Protein sequence databases are much bigger than protein databases of known 3D structures. Therefore, a central problem for protein science is to increase the number of protein structures.

Molecular modeling is a procedure that allows the calculation of the structure of a protein only if another similar protein with known 3D structure is already available. The latter is called a reference structure. When sequence comparison between them yields at least 25% identity, it is possible to use the reference structure as a mould, containing the main fold, to be transfered to the protein for which only the sequence is known. When more than one reference structure is found, they can also be compared with each other. However, this comparison uses the fold in order to find the closest related regions which are called structural conserved regions (SCR). Computers with fast calculation and graphic display capabilities are presently required for this kind of work, eventhough more tools are becoming available on personal computers.

In our work we used molecular modeling to obtain the structure of manganese peroxidase (MnP13-1) from Ceriporiopsis subvermispora, a white-rot basidiomycete able to degrade lignin. The aminoacid sequence of this enzyme was deduced by Dr. Rafael Vicuña et al. from the sequence of the corresponding gene. The MnP13-1 sequence was compared with those of all other peroxidase sequences currently available. The comparison yielded over 60 % identity in particular with the manganese peroxidase (MnP) from Phanerochaete chrysosporium, another white-rot fungus previously studied. When 3D information is available, structural homology yields more information than traditional sequence homology. So far, lignin peroxidase (LiP) and manganese peroxidase from P. chrysosporium, as well as Arthromyces ramosus peroxidase (ARP) have known 3D structures. After structural comparison was conducted, LiP and MnP resulted to be closely related to each other (0.85 A/res deviation) than ARP with LiP or ARP with MnP (1.0 A/res deviation each pair). The same comparison gave us the SCR´s. The pairwise sequence comparison of the above proteins with MnP13-1 allowed us to calculate scores for each of the SCR´s. This comparison revealed that most MnP13-1 boxes were closely related to LiP and MnP, specially to regions between I125-R147 on MnP13-1 and V112-R134 on LiP. On the contrary, E148-G183 (box 4) and R266-327 (box 7) are regions more similar to ARP than to LiP. These segments of lower scores lead to think in the evolution of these proteins, considering that the former includes helix E on ARP and the latter the helix 7A on LiP.

The modeling of MnP13-1 was done transfering the atom positions for each residue on the conserved boxes, mainly from MnP to MnP13-1. Differences such as the insertion of residues TGGN and TDSP on MnP13-1 were handled in a different way. The first insertion was modeled as a loop between S231 and D232. The TDSP insertion located at the C-end part of the polypeptide chain was built residue by residue in extended geometry. The model was completed by the optimization of the potential energy starting with the positions of atoms provided by the transfering step. The optimization moved the coordinates shortly in order to reduce the potential energy. The final model differed 1.7 A/res in average from the starting model and it contained very low energy. The optimized model presented TDSP insertion self stabilized by local contacts. Moreover, this chain segment pointed to the binding site of Mn2+, which is an essential metal ion for enzyme function. Therefore some interactions should be expected between both parts of the proteins due to their opposite charges.

Another characteristic provided by the model is that the geometry of Ca2+ binding is conserved in same way than the Ca 2+ binding of LiP and MnP. The geometry of Mn2+ binding in MnP13-1 is conserved too as it is in MnP. The putative binding site for veratryl alcohol was also modeled on MnP13-1 and the similarity of the geometry of this site suggested the binding of aromatic substrates on this protein too.

The modeling of MnP13-1 from reference structures with a high sequence homology revealed structure datails that are unique to MnP13-1 that could differentiate this protein from other members of MnP protein family. The new characteristics observed on the MnP13-1 model should be tested experimentally.