DNA-encoded libraries come of age as companies move closer to using the tags to identify novel compounds for drug discovery
At the end of 2015, in a Eureka post, I gave a brief introduction to DNA-Encoded Libraries (DELs) and suggested that they would be a hot topic for medicinal chemistry in 2016. So, I was naturally interested in a recent article published by Roche summarizing some of the company’s experiences in this area. My attention was particularly drawn to the two compounds highlighted by the author which have proceeded to clinical trials and are clearly derived from hits obtained from screening DELs.
Interestingly, both of these examples emanate from GSK (which acquired Praecis Pharmaceuticals in 2007 including its DEL technology) and are documented in peer-reviewed journal articles.
The discovery of the RIP1 kinase inhibitor GSK2982772 is recounted in a paper from the GSK group at Collegeville, PA, working in collaboration with a UK-based academic group. The paper cites an earlier article describing the synthesis of an 800-million-member DNA-encoded library by the Praecis group which suggests this was the source of the initial hit (designated GSK′481). GSK′481 demonstrated remarkable properties for a screening hit: as well as excellent potency in binding and functional assays for RIP1 kinase, the compound also showed complete specificity for the target when screened at 10mM in two large kinase panels. Even better, the benzoxazepinone scaffold constituted a novel kinase inhibitor template.
However, the compound was too lipophilic and insoluble which gave rise to poor pharmacokinetic (PK) properties. (Poor solubility bedevils many modern drug candidates but it is vital that drugs that are to be administered orally dissolve well in the gastrointestinal tract so that they can be absorbed into the bloodstream). Thus, the GSK team undertook a lead optimization program to improve the properties of GSK′481 guided in part by a computational model of the structure of the target kinase.
After a thorough exploration of the various parts of the scaffold, the final compound was obtained by replacement of the isoxazole by a triazole giving GSK2982772 (a relatively small change overall). This compound is 100-fold less lipophilic than the initial hit and thus more soluble. Consequently, its PK profile was much improved, allowing its progression, after further pre-clinical testing, to clinical candidate status. Post-nomination, the compound was successfully co-crystallised with RIP1 kinase (see RCSB entry 5TX5) revealing an allosteric (type III) binding mode.
In January 2018, GSK2982772 completed a Phase II clinical trial for psoriasis. However, most recently, dosing in a further safety evaluation study has been suspended pending evaluation of cases of asymptomatic arrhythmia adverse events, detected during planned telemetry monitoring. The road to market is never a smooth one…
The second compound, the soluble epoxide hydrolase (sEH) inhibitor GSK2256294, was discovered and reported by the GSK groups in Waltham, MA and King of Prussia, PA. In this case, DELs based on a triazine scaffold were used for hit-finding leading to the identification of the compound denoted “ELT hit”. This was a potent sEH inhibitor (pIC50 = 8.1), although its molecular properties, including molecular weight (517 Da.), polar surface area (141 Å2) and lipophilicity (ClogP = 4.7, calculated using ACD/Labs GALAS model), were rather high for orally delivered drugs and so required optimization.
Lead optimization chemistry led to an improved compound denoted “ELT delivered asset” which was deemed suitable for preclinical development. This compound had a similar molecular weight to the original hit (506 Da.) but reduced lipophilicity (ClogP = 3.7) and polar surface area (98 Å2) and improved potency (pIC50 = 8.5). Further optimization focused on improving so-called “developability” parameters, including aqueous solubility and oral bioavailability, and led finally to the identification of the clinical candidate GSK2256294. Notably, this has a significantly reduced molecular weight (422 Da.) and a further reduction in PSA (92 Å2) albeit at the cost of slightly increased lipophilicity (ClogP = 4.1). GSK2256294 is currently undergoing Phase II testing in healthy volunteers to assess the compound’s effect on tissue sEH activity and insulin sensitivity.
Examples such as these suggest that DEL technology is capable of delivering readily optimizable hit compounds to medicinal chemistry teams and are perhaps part of the reason why the number of deals between DEL specialists (such as X-Chem and HitGen) and pharma/biotech companies have mushroomed in the last year or so. While further developments in the underlying technologies may be anticipated, it seems that DNA-Encoded Libraries have come of age and are taking their place alongside other hit-finding technologies in the continuing search for new and improved therapeutics.
DNA-encoded chemistry: enabling the deeper sampling of chemical space. Nature Reviews Drug Discovery 2017, 16, 131–147. doi:10.1038/nrd.2016.213
DNA-Encoded Library Technology: A Brief Guide to Its Evolution and Impact on Drug Discovery. Annual Reports in Medicinal Chemistry 2017, 50, 1-15. doi: 10.1016/bs.armc.2017.09.002
How DNA-encoded libraries are revolutionizing drug discovery. Chemical and Engineering News 2017, 95 (25), 28-33. https://cen.acs.org/articles/95/i25/DNA-encoded-libraries-revolutionizing-drug.html
So What Do You Get From DNA-Encoded Libraries? http://blogs.sciencemag.org/pipeline/archives/2016/07/11/so-what-do-you-get-from-dna-encoded-libraries