Decoding binary reaction model products (novel quercetin-glucose adducts): From identification to bitterness masking...

This study aims to search for quercetin derivatives that may form during coffee roasting and clarify their flavor characteristics. A binary reaction model comprising only quercetin and glucose was innovatively employed to simulate roasting and to investigate whether substitution reactions would occur between the two compounds at high temperatures. Through meticulous analysis and purification of the reaction products, we isolated four novel quercetin-glucose adducts that had not been previously reported. Furthermore, we successfully detected one of these adducts (HF-3) in roasted coffee beans with different pretreatments, with a concentration range of 0.135-0.423 μg/mL. Sensory evaluation demonstrated that HF-3 exhibited a high threshold for both bitterness (355 μmol/L) and astringency (328 μmol/L), while effectively masking the bitterness of some bitter molecules, particularly quinine (45.7 %) and caffeine (42.1 %). Computer simulation technologies confirmed that the competition between HF-3 and caffeine for accessing the active pocket was responsible for the observed bitterness masking effect.