Streamline - machine learning guided development of novel GLP-1 receptor agonists with improved drug properties

Randi Mikkelsen; Nina Buch-Manson; Claudia Hjorringgaard; Nielsen Jens Christian; Mads Nygaard; Lisbeth Elster; Silas Rasmussen; Kristoffer Rigbolt; Niels Vrang; Louise Dalboge

doi:10.1530/endoabs.110.EP858

EP858

Prev Next

Section Contents Cite

Endocrine Abstracts (2025) 110 EP858 | DOI: 10.1530/endoabs.110.EP858

ECEESPE2025 ePoster Presentations Metabolism, Nutrition and Obesity (164 abstracts)

Streamline - machine learning guided development of novel GLP-1 receptor agonists with improved drug properties

Randi Mikkelsen ¹ , Nina Buch-Månson ¹ , Claudia Hjørringgaard ¹ , Jens Christian Nielsen ¹ , Mads Nygaard ¹ , Lisbeth Elster ¹ , Silas Rasmussen ¹ , Kristoffer Rigbolt ¹ , Niels Vrang ¹ & Louise Dalbøge ¹

Author affiliations

¹Gubra, Hørsholm, Denmark

JOINT2771

We have developed streaMLine, an innovative platform for peptide drug discovery that greatly shortens the time from initial hit to clinical drug candidate. The platform allows for high throughput synthesis and screening. Thousands of peptides are systematically screened in in vitro assays and on chemical and physical parameters, whereby the streaMLine platform enables complete sequence exploration and simultaneous optimization of key parameters. Using the streaMLine platform, we developed novel GLP-1R agonists from a secretin peptide backbone, to demonstrate how high throughput screening of peptide libraries and machine learning guided drug design can be applied to accelerate drug discovery. We systematically synthesized and screened a total of 2,688 peptides in a parallelized optimization workflow. Using this approach, we were able to generate potent, selective, and long-acting GLP-1R agonists with improved physicochemical properties. To validate the developed QSAR pipeline, we conducted an in-depth profiling of a developed GLP-1R agonist, GUB021794. GUB021794 was tested for cross-reactivity towards other receptors in the glucagon superfamily of receptors and showed no activation of the GIPR, GLP-2R or GCGR when tested at concentrations up to 3000 nM. With a GLP-1R potency of 0.018 nM and a SCTR potency of 190 nM, GUB021794 thus showed high receptor selectivity for GLP-1R. In addition, GUB021794 (S.C., 10 nmol/kg QD) showed potent body weight loss in diet-induced obese (DIO) mice and a half-life in rats compatible with once-weekly dosing in humans. Overall, the preclinical data demonstrate that GUB021794 is a potent, long-acting GLP-1R agonist promoting robust body weight loss in DIO mice comparable to clinically approved GLP-1R agonists such as semaglutide.