Maria João Ramos

Computational Biochemistry, University of Porto

Maria João Ramos



Research Areas

Enzyme Catalysis:

We have been performing the study of catalytic and inhibition mechanisms of several enzymes, such as Class I Ribonucleotide Reductase, Pyruvate Formate Lyase, Farnesyltransferase, Fumarate Reductase, Cox-2, Uroporphyrinogen III Decarboxylase, Glutathione Transferase, Superoxide Dismutase, Thioredoxin, HIV-1 Protease, Reverse Transcriptase and Integrase, PLP-dependent enzymes, HMG-CoA-Reductase, Asparaginase, Renin, ACE-II and several glucosidades, among others.
We have been using large QM/MM enzyme models and high theoretical levels, with QM regions of over 100 atoms described at the DFT level with large basis sets, embedded in complete enzyme models described at the MM level. We have been treating the interaction betweem the quantum and classical regions with the ONIOM methodology.

Molecular Dynamics of Proteins:

We have been addressing many aspects of enzyme catalysis via molecular dynamics simulations. These include conformational changes, detection of water channels and water hydration sites, evaluation of protein flexibility, or refinement of enzyme:substrate complexes obtained from docking/modelling studies. Cu, Zn Superoxide Dismutase, Farnesyltransferase, Gluthatione Transferase or HIV-1 Reverse Transcriptase are some of the studied systems. Parameter Development is also an area in which we have been working on, e.g. we have developed molecular mechanics parameters for  metal enzymes in several ligand environments. The parameters have been obtained by fitting to DFT potential energy surfaces and are committed to the Amber force field. Parameters for biological membranes have also been developed.

Computational Mutagenesis:

Protein complexation regulates a large number of cellular events, and to interfere with protein:protein complexes is of the utmost therapeutic importance. Alanine scanning mutagenesis of protein-protein interfacial residues is currently performed to detect the hot spots for protein complexation. These are the regions that must be drug-targeted.
We have developed a computational protocol, based on MM-PBSA calculations, that predicts differences in binding free energies between the wild-type and alanine mutated complexes with an average unsigned error of 0.80 kcal/mol, and a maximum error of 2.5 kcal/mol. It was benchmarked with a set of 46 mutations, and permits a systematic scanning mutagenesis of protein-protein interfaces. We have recently shown that the method is as accurate as TI but at a fraction of the computational cost.

Software Development:

We have  developed a multi-stage docking algorithm, MADAMM,  which allows for full ligand flexibility as well as receptor flexibility. A user friendly graphic interface has been  implemented too. The algorithm has produced very good results, in terms of its capacity to obtain the binding pose of a protein:ligand complex in systems subjected to significant binding-pocket reorganization upon ligand binding.

We have also developed other softwares in the last years.  VSLab, CompASM, VolArea, and ChemBioTracker are those that have been already made available to the scientific community. VSLab offers a simple and convenient way of conducting virtual screening campaigns with AutoDock (Scripps), simplifying all computational procedures and offering additional analysis tools. CompASM is a tool that performs and analyses Alanine Scanning Mutagenesis of proteins interfaces (using the Amber software) in a simple, convenient and semi-automated way. VolArea measures surface areas and volumes of molecules with a fast algorithm, adequate for treating large numbers of structures from molecular dynamics trajectories. ChemBioTracker identifies networks of interatomic interactions in complex biological systems.
All these algorithms are directed to facilitate drug discovery, which we carry out with the collaboration of industrial partners.

Drug Discovery:

Drug discovery is an area in which we have been working on. Basically, we have been exploring ways of optimizing lead compounds. Usually we start from a structure of a receptor and try to find hit compounds that bind the receptor with virtual screening techniques. After experimental validation, we try to perform subtle modifications in the drug which improve its affinity (calculated with FEP/TI techniques) without compromising its ADME properties.