AI-powered Design Of Liver X Receptor Modulators

Artificial intelligence (AI) is a valuable asset in accelerating the drug development process and reducing costs. The chemical space contains an estimated 10^60 potentially pharmacologically active molecules, a figure that exceeds the total number of molecules on Earth.[1] AI methods help to narrow down the number of relevant molecules that could address a given target protein and thus save costs and time in the experimental testing. Instead of large compound libraries with millions of compounds, the number of substances to be tested can be reduced substantially.

In this study, we selected the nuclear liver X receptors (LXRs, NRH2/3) as a target, which function as transcription factors. These receptors activate genes involved in reverse cholesterol transport, fatty acid synthesis, and inflammatory responses.[2,3] Therapeutically, LXRs hold potential for treating atherosclerosis, metabolic disorders, and certain cancers due to their role in lipid homeostasis and inflammation.[2,3] However, the development of LXR-targeted drugs is hindered by adverse effects such as hepatic steatosis and hypertriglyceridemia, necessitating the design of selective modulators to mitigate these limitations.[2] Since LXRα is the dominant subtype in the liver, selective LXRβ agonists have the potential to prevent hepatic steatosis while maintaining the beneficial effects on cholesterol efflux. Conversely, inverse LXR agonists could be employed to develop antilipogenic drugs, in particular for metabolic disorders such as metabolic (dysfunction) associated fatty liver disease (MAFLD) and metabolic dysfunction-associated steatohepatitis (MASH).[4] Hence, there is a strong need for novel LXR modulators that can be further developed into clinical candidates.

Here, we utilize a chemical language model (CLM) trained on SMILES strings from literature LXR agonists to design novel LXR modulators.[5] This approach enabled the identification of an LXR agonist with a markedly enhanced off-target profile in comparison to literature LXR agonist T0901317, as well as an inverse LXR agonist with antilipogenic activity.