Search

Iron is vital for all living creatures but becomes toxic when it exceeds the levels catered for by their natural cellular buffering mechanisms, causing free radical formation via the Fenton reaction and ultimately, therefore, leading to oxidative stress. In such situations, for example in repeatedly transfused patients suffering from β-thalassaemia or sickle cell anaemia, iron chelation therapy is required. Desferioxamine (DFO), the most widely used therapeutic chelator, is a hexadentate ligand possessing a very high affinity for Fe3+, but it is not orally active. 3-hydroxypyridin-4-ones (HPOs), the more recently introduced synthetic alternatives, have also been shown to be useful therapeutic chelators, demonstrating the essential affinity and selectivity for Fe3+ along with good oral activity. Deferiprone, a typical HPO, has emerged as a prominent therapeutic, able to remove accumulated excess iron from the heart and mitochondria. However, because Deferiprone has some drawbacks of relatively high metabolic instability and a side effect of lowering white blood cell count for a small number of patients, the search continues for other such synthetic chelators with improved properties. In the work reported here, various computational studies have been performed to aid in the rational design of Fe3+ chelators, with their physicochemical properties (pKa, Fe3+ affinity, hydration and membrane permeability) predicted by means of quantitative structure property relationship (QSPR) methods, quantum mechanical (QM) calculations, and molecular dynamic (MD) simulations. The pKa and Fe3+ affinity were also studied experimentally with a novel approach devised for dealing with ligands possessing substituents with hydrogen bond donor and/or acceptor groups. The pKa values predicted using QSPR and QM static calculations (Gaussian 09) were found to differ very significantly from the experimentally determined values. When data from the QM static calculations were combined with a regression model, however, the pKa predictions were significantly improved, with the predicted values then within ± 0.2 log units of the experimental values, and computing times of the order of 1 day per molecule. These calculations also allowed the determination of possible deprotonation sequences for the predicted compounds. Further pKa predictions were made by means of QM MD calculations, using Car-Parrinello molecular dynamics (CP-MD). These simulations were found to yield pKa values within ±0.3 log units of the experimental values but involved much longer computing times (of the order of 20 days per molecule). In addition, however, the CP-MD simulations also provided valuable insights into the atomistic details of the proton transfer mechanism and the solvation structure and dynamics at all stages of the reaction. For the three HPOs studied, it was observed that proton transfer takes place along a chain of three H20 molecules, although direct hydrogen bonds were observed to form transiently. The (Fe3+) log /C, predictions for the HPOs were made using an entirely novel QM-based methodology (and without knowledge of the chelator pKa values), yielding log K-\ values within ± 0.32 log units of the experimental values. For the preparation of membrane permeability study, novel Chemistry at HARvard Molecular Mechanics (CHARMM) force fields specifically for use in HPO simulations were developed. These new force fields were validated using Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) MD simulations of the chelators’ behaviour in aqueous solution.