In silico ADMET: applications and new paths

The early and simultaneous consideration of potency, selectivity and ADMET (absorption, distribution, metabolism, excretion and toxicity) has become a characteristic feature of modern drug discovery. Today, prediction methods are an integral part of technology platforms for lead finding and optimization. Besides the efficient application of proven prediction methods, the field needs new developments yielding robust, precise models. Protein structure-based approaches in ADMET prediction might be one of those new paths.

In early drug discovery, lead selection and lead optimization can be regarded as multi-dimensional optimization problems. Besides pharmacological activity on a primary target and selectivity towards other drug targets, ADMET properties have a prominent role as important optimization parameters. In order to reduce the complexity of the underlying optimization problems it is important to understand the relationship between chemical structure and some of the aforementioned parameters. In silico ADMET is at the heart of this business. It tries to relate structural features or physicochemical properties of small molecules with pharmacodynamic or toxicologic endpoints. Thus, in silico ADMET helps to focus lead optimization efforts on compounds with a higher probability of success.

In this issue of Drug Discovery Today Editor’s Choice, I would like to refer to three recent reviews that highlight the importance of in silico ADMET from different angles. I will review the various methods of communicating and visualizing physicochemical properties as well as new orthogonal approaches to predict ADMET properties.
The consideration of drug-like physicochemical properties during the hit-to-lead and lead optimization processes is now well accepted and optimal ranges for key molecule descriptors, such as size, lipophilicity, polarity, rotatable bond and aromatic ring counts have been suggested. Timothy Ritchie, Peter Ertl and Richard Lewis have reviewed a number of approaches used to represent molecule properties graphically in the context of oral ‘drug-likeness’, with the goal of improving the decision making of medicinal chemists during the drug discovery process. They conclude that both simple and more sophisticated graphing and visualization techniques have successfully been applied to assist medicinal chemists in designing and selecting molecules with better properties.
In our review “Utility of protein structures in overcoming ADMET-related issues of drug-like compounds” my colleagues Friederike Stoll, Andreas Göller and I focus on protein–structure based methods for overcoming ADMET problems during lead optimization. The number of solved X-ray structures of proteins relevant for ADMET processes of drug molecules has increased remarkably over recent years. A list of the proteins that are key players for ADMET processes is provided, including their PDB-IDs. While it is still not really possible to use the structural information available on the P-glycoprotein, the hERG channel and cytochrome P450 enzymes to “design out” unwanted ADMET properties, convincing examples are listed in which the X-ray structures of human serum albumin and pregnane X receptor were used to rationally overcome a low fraction unbound and cytochrome P450 3A4 induction.
A very similar review by Gautier Moroy, Virginie Martiny, Philippe Vayer, Bruno Villoutreix and Maria Miteva with the title “Toward in silico structure-based ADMET prediction in drug discovery” extends the list of structures for ADMET proteins for phase II metabolizing enzymes (UDP-glucuronosyl-transferase, sulfotransferases) and the plasma protein α1-acid glycoprotein. They conclude that ADMET proteins seem to be particularly challenging topics for structure-based design because they are often promiscuous, with flexible and sometimes multiple binding sites. However, they are confident that protein structure-based approaches constitute a major progression in the field and will open new avenues toward ADMET predictions at the atomic level.
I hope that you enjoy reading the articles provided with this newsletter. I think they illustrate how different approaches in the prediction of ADMET properties support the early drug discovery process and help to increase its efficiency.
Alexander Hillisch is a Director of Medicinal Chemistry and Head of Computational Chemistry at Bayer Pharma AG, Wuppertal, Germany. Since 2003 his team supports drug discovery efforts in cardiology and oncology indication areas with computational chemistry, chemoinformatics, in silico ADMET and structural bioinformatics techniques.
From 1998 to 2003 he headed the research group "Structural Bioinformatics and Drug Design" at EnTec GmbH, Jena, Germany, a subsidiary of Schering AG, Berlin. There he was project manager in preclinical research and involved in the computer aided design and pharmacological characterization of drugs against gynecological diseases and cancer.
He conducted his Ph.D. thesis at the Institute of Molecular Biotechnology (IMB), Jena in the area of biophysics (NMR, FRET) and molecular modeling. Alexander Hillisch received his Ph.D. in Biochemistry with Prof. Peter Schuster in 1998 and his diploma in Pharmacy in 1995 from the University of Vienna, Austria.

He is author of 35 research papers, 35 patent applications and two books. Alexander teaches “Molecular pharmacology and Drug Design” at the University of Cologne, from which he received a honorary professorship in 2010.

Share this article

More services


This article is featured in:
The View From Here


Comment on this article

You must be registered and logged in to leave a comment about this article.