Quorum sensing (QS) is a cell-cell communication mechanism that connects members in various microbial systems. Conventionally, a small number of QS entries are collected for specific microbes, which is far from being able to fully depict communication-based complex microbial interactions in human gut microbiota. In this study, we propose a systematic workflow including three modules and the use of machine learning-based classifiers to collect, expand, and mine the QS-related entries. Furthermore, we develop the Quorum Sensing of Human Gut Microbes (QSHGM) database (http://www.qshgm.lbci.net/) including 28,567 redundancy removal entries, to bridge the gap between QS repositories and human gut microbiota. With the help of QSHGM, various communication-based microbial interactions can be searched and a QS communication network (QSCN) is further constructed and analysed for 818 human gut microbes. This work contributes to the establishment of the QSCN which may form one of the key knowledge maps of the human gut microbiota, supporting future applications such as new manipulations to synthetic microbiota and potential therapies to gut diseases.

Terpene synthases (TPS) catalyze the cyclization of the acyclic prenyl diphosphate precursor and are responsible for the abundance of the natural product terpene in nature. The biosynthesis of terpenoid by nonseed plant TPS is not well understood due to the highly dynamic feature of the cyclization reaction and the unavailability of the enzyme structure. Here we discovered a class I TPS JeST4 from Jungermannia exsertifolia and elucidated its catalytically active structure in complex with the substrate and the key carbocation intermediates during catalysis. We found that D106 is critical for the enzyme's activity by mediating the gate formed with R294 or R225. Further, we identified two hotspot regions from the coevolution study and computational simulations, and the G91S and R242K mutations improved the conversion rate by 39-and 11-fold, respectively. Remarkably, in both variants, R294 is able to stabilize the substrate pyrophosphate group, analogous to the dominating interaction network observed in the distantly related bacterial TPSs. Further, NMR and molecular dynamics simulations indicated the formation of an unusual C10(S)-bicyclogermacrene. Our research demonstrates the capacity of computing-informed engineering of nonseed plant TPS. The discovery of the new TPS enzyme from nonseed land plant and computing-guided engineering would potentiate the exploitation of the ultracheap enzyme as a potential biocatalyst for the production of the valuable terpenoid products.