Head-on and co-directional RNA polymerase collisions orchestrate bidirectional transcription termination.

Genomic DNA is a crowded track where motor proteins frequently collide. It remains underexplored whether these collisions carry physiological function. In this work, we develop a single-molecule assay to visualize the trafficking of individual E. coli RNA polymerases (RNAPs) on DNA. Based on transcriptomic data, we hypothesize that RNAP collisions drive bidirectional transcription termination of convergent gene pairs. Single-molecule results show that the head-on collision between two converging RNAPs is necessary to prevent transcriptional readthrough but insufficient to release the RNAPs from the DNA. Remarkably, co-directional collision of a trailing RNAP into the head-on collided complex dramatically increases the termination efficiency. Furthermore, stem-loop structures formed in the nascent RNA are required for collisions to occur at well-defined positions between convergent genes. These findings suggest that physical collisions between RNAPs furnish a mechanism for transcription termination and that programmed genomic conflicts can be exploited to co-regulate the expression of multiple genes.

Programming cell-surface signaling by phase-separation-controlled compartmentalization.

The landscape of cell-surface signaling is formidably complex. Robust tools capable of manipulating the spatiotemporal distribution of cell-surface proteins (CSPs) for dissecting signaling are in high demand. Some CSPs are regulated via multivalency-driven liquid-liquid phase separation (LLPS). Employing the robustness and versatility of LLPS, we decided to engineer LLPS-based tools for precisely manipulating CSPs. We generated membrane-tethering LLPS systems by fusing multivalent modular phase-separation scaffold pairs with CSP binders. Phase separation of the scaffold pairs, concomitant compartmentalization of CSPs on membranes, and cluster-dependent signaling outputs of CSPs require membrane recruitment of one or both scaffolds. We also engineered orthogonal phase-separation systems to segregate CSPs into mutually exclusive compartments. The engineered phase-separation systems can robustly cluster individual CSPs, co-cluster two or more CSPs, or segregate different CSPs into distinct compartments on cell surfaces. These tools will enable the dissection of complicated cell-signaling landscapes with high precision.

PHF1 compartmentalizes PRC2 via phase separation.

Polycomb repressive complex 2 (PRC2) is central to polycomb repression as it trimethylates lysine 27 on histone H3 (H3K27me3). How PRC2 is recruited to its targets to deposit H3K27me3 remains an open question. Polycomb-like (PCL) proteins, a group of conserved PRC2 accessory proteins, can direct PRC2 to its targets. In this report, we demonstrate that a PCL protein named PHF1 forms phase-separated condensates at H3K27me3 loci that recruit PRC2. Combining cellular observation and biochemical reconstitution, we show that the N-terminal domains of PHF1 cooperatively mediate target recognition, the chromo-like domain recruits PRC2, and the intrinsically disordered region (IDR) drives phase separation. Moreover, we reveal that the condensates compartmentalize PRC2, DNA, and nucleosome arrays by phase separation. Luciferase reporter assays confirm that PHF1 phase separation promotes transcription repression, further supporting a role of the condensates in polycomb repression. Based on our findings, we propose that these condensates create favorable microenvironments at the target loci for PRC2 to function.

Rapid regulations of metabolic reactions in Escherichia coli via light-responsive enzyme redistribution.

Protein-based condensates have been proposed to accelerate biochemical reactions by enriching reactants and enzymes simultaneously. Here, we engineered those condensates into a photo-activated switch in Escherichia coli (PhASE) to regulate enzymatic reactions via tuning the spatial correlation of enzymes and substrates. In this system, scaffold proteins undergo liquid-liquid phase separation (LLPS) to form light-responsive compartments. Tethered with a light-responsive protein, enzymes of interest (EOIs) can be recruited by those compartments from cytosol within only a few seconds after a pulse of light induction and fully released in 15 min. Furthermore, we managed to enrich small molecular substrates simultaneously with enzymes in the compartments and achieved the acceleration of luciferin and catechol oxidation by 2.3- and 1.6-folds, respectively. We also developed a quantitative model to guide the further optimization of this demixed regulatory system. Our tool can thus be used to study the rapid redistribution of proteins, and reversibly regulate enzymatic reactions in E. coli.