Fragments vs HNF4: a chemical probe

Driven by the spectacular success of diabetes and obesity drugs, metabolism is a hot therapeutic area. Much of the focus has been on GPCRs such as GLP-1R, but there are plenty more potential targets. For example, the two isoforms of hepatocyte nuclear factor 4 (HNF4α and HNF4γ) are nuclear receptors that control transcription of genes associated with metabolic homeostasis; mutations in HNF4α lead to an inheritable form of diabetes. But the biology is complicated and not fully understood. In a new J. Med. Chem. paper, Daniel Merk and collaborators at Goethe University Frankfurt and elsewhere describe a chemical probe.

 

The story actually starts with an (open-access) 2020 paper in Int. J. Mol. Sci., in which the researchers screened 480 fragments at 50 µM in a reporter gene assay. The fragments were derived from FDA-approved drugs and roughly rule-of-three compliant, though with some more lipophilic members.

 

HNF4 is activated by binding to naturally occurring fatty acids, and the cell-based assay could detect ligands that altered gene activity, either by increasing it (agonists) or decreasing it (inverse agonists). After follow-up dose response assays and binding validation via isothermal titration calorimetry (ITC), the researchers found one low micromolar agonist and two low micromolar inverse agonists. The most potent of these became the subject of the new paper.

 

Structure-activity studies on compound 4 revealed that both the carboxylic acid and hydroxyl moieties are important for binding, and that affinity could be improved by growing the fragment. Interestingly, while compound 20 acts as an inverse agonist, compound 23 is an agonist. A crystal structure of compound 23 bound to the ligand binding domain of HNF4α shows that it binds in the same site and in a similar manner as the natural ligand myristic acid. Further medicinal chemistry ultimately led to compound 46, an agonist with low nanomolar activity in cells.

 

 

Compound 46 was selective against nine other nuclear receptors and not toxic up to 10 µM. The compound was also active in a separate reporter assay using the human insulin promoter. Purified recombinant HNF4α normally retains a bound fatty acid, and the researchers measured binding affinity of compound 46 with and without this natural ligand. The binding affinity of compound 46 measured in the absence of the fatty acid was stronger (low nanomolar by ITC), demonstrating competition, as expected.

 

This is a nice fragment to lead success story in academia. It is also an uncommon example where the primary assay was cell-based, and structural biology did not play a significant role in compound optimization. No ADME data are provided, so in vivo studies may be premature, but with its low molecular weight and high ligand efficiency compound 46 is well positioned for further optimization.