Moitessier Research Group

Although medicinal research has made tremendous advancements in the past decades, the expense and time associated with today’s drug design process is still overwhelming and the time it takes for a conceived lead compound to emerge as a marketed drug is well above a decade. Drug companies has invested large sums of money and time in high throughput screening (HTS), which still responds with low hit rates. Avoiding unproductive syntheses is a worthy goal not only in terms of cost and time efficiency but also environmental responsibility - as the amount of waste products from organic syntheses are greatly reduced. We believe a computational program that could provide us with highly accurate predictions of biological activity will be one solution to this critical problem.

In the present project, computational, organic and biological chemists team up in a feedback loop involving software development, computer-assisted drug design, modern chemical synthesis, and biochemical testing. Within this system, group members follow the lead compound from its moment of conception to its final performance in biochemical testing. The outcomes of this compound’s development are then reported to those members developing the software, and modifications or improvements in the software occur before the whole cycle begins again. The vested interest that the group shares for each prospective candidate helps to expedite the feedback process. We have decided to focus on three particular enzymes – Dipeptidyl Peptidase IV (DPP-IV), Cyclophilin A (Cyp A), and Prolyl Oligopeptidase (POP) – implicated primarily in diabetes, AIDS, and Alzheimer’s disease, respectively. These enzymes have been also implicated in cancer (DPP-IV) and rheumatoid arthritis (Cyp A) research.

At the heart of this medicinal chemistry project is our innovative software, FITTED (Flexiblity Induced Through Targeted Evolution) which seeks to improve current drug design by accurately predicting the biological activity of proposed target compounds. We are developing this software in the hopes that in the future it can be used universally and applied to a large scope of physiological macromolecules, greatly improving efficiency in the world of drug design. The software we are inventing recognizes some of the most challenging elements of macromolecule structure and environment; molecular flexibility and solvation effects are two realistic aspects of chemical systems that are repeatedly ignored or greatly simplified in the use and programming of many other currently available modeling software.

Our immediate goals for the software include the improvement of our genetic algorithm for docking to flexible proteins, the development of a coarse-grained / knowledge-based force field, and the inclusion of explicit bridging water molecules. Much progress has been made so far in all these areas, and all the while the software is being evaluated for its actual success in real systems outside the computer. Our first generation of synthesized compounds has coincided with the preliminary and non automated version of our software; the next generation of compounds will be in response to the revised versions, and so on. The very first synthesized compounds are presently evaluated for their inhibitory potency. Throughout the project, there will be constant dialogue between the programmers and the medicinal chemists which will help to improve both the candidate compounds and the actual software being produced.

top

Catalyst discovery and computers. The catalyst discovery process as practiced now is time consuming , expensive, and often serendipitous . Computational prediction would enable the chemists to reduce the cost and accelerate the lead generation and optimization steps . Until recently, conventionally available computer power has been insufficient to tackle the combinatorial problem of molecular design through virtual screening. This computational power is now available and the time is ripe for the development of more predictive methods. Indeed, the use of computer-aided molecular design methods in the field of drug development is well established but focused almost exclusively on ground state structures. Meanwhile, few groups have focused on the development of rapid and predictive tools for de novo design of asymmetric organic reactions which require the more challenging and complex study of flexible transition state (TS) structures. Apart from the highly computer time-intensive density functional theory (DFT), ab initio and semi-empirical methods, which are restricted to small systems, little attention has been directed to TS structure analysis by fast and accurate in silico screening methods. In particular, few predictive molecular mechanics-based models have been designed to address TS structures. To this end, w e have recently reported a simple yet accurate strategy for describing transition states using molecular mechanics (MM) methods and a genetic algorithm to account for flexibility of the TS structures. Development of a fully automated computational method for rapid virtual screening of potential asymmetric catalysts is underway and recently entered the validation stage.

When developing a virtual screening method, the computational chemists have to balance accuracy and speed. In this prospect, molecular mechanics appears as a method of choice. As transition state force fields (TSFF) and SEAM approaches, our developed computational method predicts the transition states (TSs) structure and energies using molecular mechanics principles but in a less time-consuming manner. Preliminary validation of this method in combination with a genetic algorithm as a search engine revealed the potential of such an easy-to-use approach. Further improvement and validations are being carried out.

Predictions will next be compared with experimental data. This experimental vs. observed quantitative structure-efficacy relationship will be necessary to identify and fix the weaknesses of the method (e.g., ? stacking poorly reproduced due to the choice of a specific force field, poor estimation of missing parameters which requires the improvement of the parameter estimator). A new cycle of optimization will then be started.

Carbohydrate-based Catalysts. With this method in hand, we designed, screened and selected bifunctional catalysts for synthesis. Monosaccharide units are attractive synthons. They are also cheaper and more abundant than amino acids and are characterized by a high density of stereocentres and functionalities. Therefore, a judicious choice of the carbohydrate core would properly position the appendages, which are anchored to or interacting with the hydroxyl groups. The potential stereodiversity of carbohydrate-based templates (made possible by the availability of a wide range of monosaccharides) allows sufficient flexibility for finely tuning the geometric and electronic properties of the catalyst. To date, this high complexity has been believed to be an obstacle to their use in asymmetric synthesis. In the context of virtual screening, this is viewed as an asset. Application to asymmetric reduction of ketones and asymmetric addition to carbonyl are being investigated. The designed catalysts take advantage of the numerous oxygen atoms of the carbohydrate synthons to bind metals and/or Lewis acids.

The long term goal of this research is the delivery of a suite of programs or protocols that will have been experimentally validated for catalyst discovery and the delivery of efficient catalysts in a time and cost efficient manner.

top

Protecting/Directing groups for Regioselective Carbohydrate manipulations. In order to screen libraries of oligosaccharides, which are potential ligands of proteins, it is crucial to develop suitable strategies for combinatorial ou parallel preparation of such polymeric molecules. Due to the polyol nature of the monosaccharides, several protection/deprotection steps are required. In an attempt to decrease the number of protection and deprotection transformations, we decided to study the relative reactivity of the free secondary hydroxyl groups of monoprotected D-glucose derivatives. For this purpose, we have applied the first version of the software to the design of an efficient protecting/directing group for carbohydrates used for their regioselective functionalization. In this study, virtual screening of protecting groups on position 6 was carried out and a pyridyl-containing moiety was identified as having a significant effect on the hydrogen bond network (Fig. 4). Experimental data confirmed the predicted intramolecular hydrogen bonding network and the enhanced regioselectivity in acetylation and glycosylation reactions.

Open glycosylation has been seen as a promising but challenging strategy for effective preparation of oligosaccharides. Use of these particularly activated species as acceptors in glycosylation reactions would provide a practical route to oligosaccharides.

top