Highly Selective\nNovel Heme Oxygenase‑1 Hits\nFound by DNA-Encoded Library Machine Learning beyond the DEL Chemical\nSpace

DNA-encoded library (DEL) technology, especially when\ncombined\nwith machine learning (ML), is a powerful method to discover novel\ninhibitors. DEL-ML can expand a larger chemical space and boost cost-effectiveness\nduring hit finding. Heme oxygenase-1 (HO-1), a heme-degrading enzyme,\nis linked to diseases such as cancer and neurodegenerative disorders.\nThe discovery of five series of new scaffold HO-1 hits is reported\nhere, using a DEL-ML workflow, which emphasizes the model’s\nuncertainty quantification and domain of applicability. This model\nexhibits a strong extrapolation ability, identifying new structures\nbeyond the DEL chemical space. About 37% of predicted molecules showed\na binding affinity of <i>K</i><sub>D</sub> < 20 μM,\nwith the strongest being 141 nM, amd 14 of those molecules displayed\n>100-fold selectivity for HO-1 over heme oxygenase-2 (HO-2). These\nmolecules also showed structural novelty compared to existing HO-1\ninhibitors. Docking simulations provided insights into possible selectivity\nrationale.