Machine Learning-Assisted Optimized Production of Quorum Quenching Anthraquinones in Rhubarb.

Abstract

Background: This first attempt study focused on assessing the quorum quenching (QQ) capabilities of four keystone anthraquinones—chrysophanol, emodin, aloe emodin, and rhein— and their analogs found in the medicinal rhubarb herb. Virtual screening was utilized to determine the affinity of these compounds against quorum-sensing receptors. The anthraquinone with the highest binding score will be prioritized for optimized production via machine learning. Methods: 45 anthraquinone compounds and two quorum sensing receptors, SdiA(AHL type) and LsrB(AI-2 type) were obtained for virtual screening. An optimized random forest with bootstrap resampling was used for feature importance to see how parameters such as concentrations and k constants influence chrysophanol production. Significant Findings: Virtual screening methods reveal that chrysophanol compounds could not only demonstrate bioenergy-simulating and electron-shuttling capabilities from prior studies but also have the potential to inhibit bacterial communication, through quorum quenching. It was observed that chrysophanol 8-(6-galloylglucoside) and chrysophanol 8-(6-O-galloyl-beta-D-glucopyranoside) have the highest QQ potential with an affinity of -12.508 and -10.133 respectively. Random forest revealed the importance of high consumption of reactant and the influence of rhein and aloe emodin in the production of chrysophanol. Inhibition of hydroxylase while upregulation of oxidase could potentially lead to higher chrysophanol yield.