Targeting Isoprenoid Biosynthesis to Treat Human MalariaAlexandre S. Lawrenson1, Neil G. Berry1, Raman Sharma1, Paul M. O’Neill1, Steve A. Ward2, Audrey Odom3, Ovadia Lazari4, David Cronk4
|1Robert Robinson Laboratories, Department of Chemistry, University of Liverpool, Crown Street, Liverpool, L69 7ZD, UK|
2Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L3 5QA, U.K.
3Washington University School of Medicine, Campus Box 8208, St. Louis, MO 63110, USA
4Biofocus DPI, Chesterford Research Park, CB10 1XL, UK.
|Malaria remains a major threat to global health, with over 250 million cases per year and one million deaths annually, primarily in children under the age of five. The parasite that causes malaria is largely resistant to older therapies such as chloroquine, and is increasingly resistant to the frontline agents that have been developed (e.g. semi-synthetic artemisinins). Therefore, the continued discovery of new antimalarial agents is essential. The non-mevalonate (or MEP) pathway represents an essential biosynthetic route used by plants, algae and eubacteria to generate isoprenoid precursors. The MEP pathway has also been validated in pathogenic organisms such as P. falciparum. As this pathway is absent in mammalian systems, small molecules that inhibit the enzymes of this pathway are therefore expected to be well-tolerated when used therapeutically, and represent attractive targets for the development of novel antimalarial chemotherapeutics. Chemical validation of the pathway came from studies of the small molecule fosmidomycin, which inhibits the pathway via inhibition of 1-deoxy-d-xylulose-5-phosphate reductoisomerase (IspC), and kills the malaria parasite with minimal side effects. |
The work described here is concerned with targeting an alternative enzyme of the MEP pathway, 4-diphosphocytidyl-2C-methyl-d-erythritol cytidylyltransferase (IspD). Previous work showed that small molecules could indeed inhibit the activity of IspD and that it poses a promising new target. Chemoinformatics was employed to screen the BioFocus library of compounds using a number of seed compounds. These seed structures were identified as the natural substrates of IspC/IspD, as it hoped that mimetics may be possible inhibitors. Additionally, a literature survey yielded a limited number of weak inhibitors of both enzymes, and potential metal binding groups were also considered as IspD is a metalloprotein where an Mg2+ atom is found to be crucial for activity, therefore metal group binders may inhibit turnover. Fingerprint similarity searching using several different molecular fingerprint methods was performed to screen the BioFocus library for the nearest neighbours of the seed structures according to their Tanimoto coefficient values. Further to this, in order to encompass some diversity into the chemotypes identified, a carefully chosen Tanimoto cut-off (supported by the literature) was selected, with these compounds also included in the results for their scaffold hopping potential. The combined results yielded several thousand hits, which were each in turn docked and ranked using crystal structures of both IspC/IspD to assess their binding strengths, according to a previously validated docking protocol. Compounds were ultimately filtered to include only those with favourable solubility, and resulted in a diverse subset of 5,000 compounds. HTS was performed on this subset of compounds to assess their in vitro activity against IspD, resulting in 76 hits ranging in activity from a few hundred nM to several µM. LCMS was used to check the purity of the compounds resulting in a final list of 54 active structures. Several chemotypes from within this set are now undergoing in depth chemical optimisation.
The HTS data also formed the basis of an additional round of chemoinformatics. With an increased set of compounds available as seeds, it was hoped that this would further enrich the virtual screening. The structures were used to perform a host of virtual screening methods of varying complexity, including fingerprint similarity searching, Bayesian classification, principal component analysis, artificial neural networks, random forests, support vector machine learning and k nearest neighbour models. These methods were again applied to the BioFocus library and a consensus approach was used, with only compounds that had much support being selected. Compound filtering and diversity analysis resulted in a rationally selected set of compounds which will be tested for their in vitro IspD activity through HTS.