2.1 Introduction of LLPS-based compartments into E. coli
Since the establishment of LLPS-based compartment in prokaryotes has not been recorded in detail in previous studies4, we introduced several phase modules and recorded condensate nucleation, growth, and maintenance in E. coli . The major phase module we tested included FUSLCD (fused in sarcoma low complexity domain) fused with truncated GCN4. FUSLCD was known for its ability to phase separate via pi-pi interaction[28, 29], while GCN4 could further facilitate phase separation via oligomerization[30] (Figures 2A and 2B). This fusion protein formed membraneless compartments in the following experiments. During the dynamic formation process, some condensates were observed to nucleate around cell poles and move around them (Figure 2C, Video S1), while others emerged elsewhere in the cytosol, then moved around along the cellular membrane, and fused with each other, especially with the ones near cell poles (Figure 2D, Video S2). These phenomena are much like the behavior of liquid droplets[31]. Interestingly, nearly all condensates finally took up the intracellular space in a regular pattern, much like polar bodies[32], and the pattern was homogeneous across different cells and different expression levels of this phase module (Figure 2E, Supplementary Figure 1). Other phase separating proteins, such as cryptochrome 2 (CRY2), known for its homo-oligomerization after a period of blue light induction, formed compartments at only one pole of the cell (Supplementary Figure 2)[33].
To further investigate the mechanism underlying the pole-localized pattern of LLPS-based compartment, it was attempted to dissect genome exclusion and pole attraction of protein aggregates by expressing phase module in elongated E. coliftsZ ), which had duplicated genomes but no septum between them. It turned out that the condensates formed regularly among genomes rather than only targeting the poles (Supplementary Figure 3). This phenomenon was similar to the results of Winkler et. al , indicating that the curvature of a cell wall did not play a role in the pole-localization[34, 35]. Additionally, it was found that the compartments formed only along one side of the cellular major axis, opposite to the side nucleoids occupied (Supplementary Figure 3). Combining those results together, it was reasoned that the unfavorable contact between nucleoid and phase module and the stochastic movements of the droplet-like condensates could determine its polar localization.
Then, the fluidity of the compartment formed by the phase module was investigated, which showed similar recovering capability as that in eukaryotes (Figure 2F)[36]. Remarkably, the compartments formed in our experiments were smaller compared to those formed inside eukaryotic cells or in vitro due to the limited space inside a bacterial cell[19, 21, 37], which could be the reason why they showed whole condensate bleaching. The recovery of fluorescence could be the result of protein exchange from the other compartments in the cells mediated by diffusing scaffold proteins in cytosol, instead of the periphery of the bleached site (Figure 2F). This property indicated that protein (and other molecules) were allowed to diffuse across the condensate boundary, providing the basis for POI recruitment[19].