Optibrium Inc.’s President, Dr. Tamsin Mansley, breaks down all you need to know about the challenges of drug metabolism, and the in silico approaches to overcome these problems.
Determining metabolic fate is crucial in drug discovery. Horror stories are often shared around expensive, late-stage failures due to unexpected metabolism. Many challenges crop up, including poor metabolic stability, low bioavailability, unforeseen drug-to-drug interactions, issues from genetic polymorphisms, and the formation of reactive or toxic metabolites. Early in silico modelling can help to prevent any problems down the line.
Identifying the Enzyme Culprits
The first step to optimise metabolism is understanding which enzymes and isoforms are primarily responsible for your compound’s metabolism. Then, you can identify the sites on your molecule that these enzymes are metabolising, and how to design your compound to block these.
There are a range of different enzyme families which may be involved in metabolism. For example, cytochrome P450s, aldehyde oxidases and flavin-containing monooxygenases can cause oxidation of your compound’s functional groups. Sulfotransferases and uridine diphosphate glucuronosyl-transferases cause conjugation of compounds to polar groups. Additionally, within each enzyme family, there are numerous different isoforms which are functionally similar enzymes that differ slightly in amino acid sequence.
Using classical categorisation models, it is possible to quickly determine which enzyme families and isoforms are most likely to metabolise a specific atomic site. This can indicate compounds which can be metabolised by multiple enzymes, with multiple routes of clearance. There are two main reasons you might want compounds with multiple routes of clearance.
Firstly, genetic polymorphisms between individual patients may mean different isoforms of enzymes are present or absent and in varying concentrations. Therefore, in situations where only one isoform is responsible for drug metabolism, issues related to toxic drug build-up may arise in certain populations.
Similarly, single clearance routes increase risks from drug-to-drug interactions. Co-administered drugs may inhibit or induce action by certain drug metabolising enzymes, causing variability in a patient’s exposure to the relevant drug. By ensuring multiple routes of drug clearance, these effects can then be mitigated.
Mapping Metabolic Liabilities
Knowing which enzymes cause your compounds’ metabolism is only half the battle. To optimise metabolic stability, you also need to identify where your compounds may be metabolised. To model this regioselectivity we can take a dual approach, considering both the reactivity and the accessibility of each atomic site to metabolism.
The reactivity of a specific site on a compound to a particular metabolic reaction can be modelled with quantum mechanical simulations. These physics-based methods take a holistic view of a molecule and the electronic distributions within it and hence the electron flow within a reaction pathway. The reactivity of each site on a molecule will be specific to the enzyme family, but will not vary between isoforms of the same enzyme.
The accessibility component of a regioselectivity model is influenced by the substrate’s molecular shape and functional groups, along with the particular enzyme’s active site structure. This means accessibility will be specific to each isoform and enzyme family. The particular steric and/or polar features within both the enzyme binding pocket and the substrate will determine the substrate’s orientation and whether a particular area can access the active site; thus, some sites will be less vulnerable to metabolism than others. Accessibility effects can be modelled using descriptors rooted on each site of metabolism on the ligand.
Reactivity and accessibility effects for each enzyme and isoform can be combined using robust machine learning models, trained on high-quality data and tested on an independent test set. By applying these regioselectivity models, a good comprehension can be gained into which labile sites need to be blocked for increased metabolic stability.
