We put forward two questions: Why can phase separation in cells produce membrane-less organelles? And how can we design our system to implement its intended functions?
Like oil in water, the contents of cells can separate into droplets. According to physical principles, the process where materials self-assemble into organelles is described as ‘phase separation’, which is the conversion of a single-phase system into a multiphase system. In general, materials flow to regions with low chemical potential instead of low concentration. Finally, the components are no longer distributed uniformly but form granules locally, which can act as organelles in the cell.
Thus, the main work to synthesize an organelle is to implement phase separation in a cell. Then, how can we do it? Composition can switch rapidly through multivalence. Our design was inspired by recent studies showing that multivalence drives protein phase separation and formation of synthetic organelles. What’s more, we take our inspiration from existing living systems and previous work. For example, intrinsic disordered regions are an indicator of large-scale phase separation in the cell. They interact with each other through van der Waals forces, hydrophobic effects and electrostatic attraction. There are many interactions like this in nature, such as FKBP and Frb, SUMO and SIM, SH3 and PRM, phyB and PIF6. Thus, we can make good use of them to induce the self-assembly of our designed organelles and regulate them in various ways.
In conclusion, multivalence drives the self-assembly of proteins and interaction binds the parts together. Therefore, interaction can induce phase separation and multivalence can make larger assemblies, which are two essential elements in our design that ensure the formation of synthetic organelles.
To design multivalent modules, it is not ideal to use multiple repeated domains, which will not only make the protein extremely large and cause difficulty in molecular cloning, but also may be problematic for making transgenic organisms. Thus, instead of using multiple repeats, we turned to de novo-designed homo-oligomeric coiled coils, which we named HOTags (Homo-Oligomeric Tags). They are short peptides, ~30 amino acids[1], and are therefore ideal tags to introduce multivalence. There are seven coiled coils previously characterized in protein de novo design studies. It has been proved in previous work by Shu Xiaokun’s lab that HOTag3 and HOTag6 are the most robust in driving aggregate formation over a wide range of protein concentrations, so we chose them.
To design interaction modules, we tested many components and fused them to the N-termini of HOTag3 or HOTag6. Some of them spontaneously self-assemble and some are inducible. Moreover, we can regulate them using various different inducers and promoters with different intensities.
Post-translational modifications by the small ubiquitin-like modifier (SUMO) are crucial events in cellular response to radiation and a wide range of DNA-damaging agents. Previous studies have shown that SUMO mediates protein-protein interactions by binding to a SUMO-interacting motif (SIM) on receptor proteins. Furthermore, recent studies have shown that a protein with ten repeats of human SUMO3 (polySUMO) and a protein with ten repeats of SIM (polySIM) can phase separate in vitro[3]. Therefore, we chose SUMO3 and SIM as a pair of interaction modules that can drive the formation of synthetic organelles spontaneously. For plasmid construction, in order to make synthetic organelles visible, we chose mCherry (red fluorescent protein) and yEGFP (yeast-enhanced green fluorescent protein) as reporters. Then, we fused mCherry between the C-terminus of SIM and the N-terminus of HOTag6. Similarly, we fused yEGFP between the C-terminus of SUMO and the N-terminus of HOTag3. We used the resulting constructs to transform yeast and proved that they can be stably co-expressed. If the system works, we will find red granules co-localized with green granules in cells under a fluorescence microscope.
The interaction between FKBP and Frb can be robustly induced by rapamycin. Rapamycin is a 31-membered macrolide antifungal antibiotic. It binds with high affinity (Kd=0.2 nM) to the 12-kDa FK506 binding protein (FKBP), as well as to a 100-amino acid domain of the mammalian target of rapamycin (mTOR), known as FKBP-rapamycin binding domain (Frb)[5]. Thus, we chose these two proteins as a pair of interaction modules and assembled them onto a yeast plasmid the same as in the construction of SUMO and SIM and introduced them into yeast. If we add rapamycin to the yeast, we will see red granules co-localized with green granules in cells under a fluorescence microscope. Synthetic organelles become real!
As mentioned above, interactions can be formed not only by inducers such as rapamycin, but also spontaneously, just as SUMO and SIM. So can we combine these two ways of interactions? To solve this problem, we did further studies about phytohormone and found ABA. Abscisic acid (ABA) is an important phytohormone that regulates plant stress responses. Proteins from the PYR-PYL-PCAR family were identified as ABA receptors[6]. Upon binding to ABA, a PYL protein associates with type 2C protein phosphatases (PP2Cs) such as ABI1 and ABI2, inhibiting their activity[7]. Previous structural and biochemical observations have provided insight into PYL-mediated ABA signaling and given rise to a working model. In the absence of ABA signaling, PP2Cs are fully active and PYLs exist as inactive homodimers in cells, unable to bind or inhibit PP2Cs, mainly due to the incompatible conformation of CL2loop[7]. In response to ABA binding, the CL2 loop undergoes a conformational rearrangement to close onto the ABA-bound pocket, then, the interaction between PYLs and PP2Cs can be formed. Here we chose PYL1 and ABI1 as a pair of interaction modules. Then, we assembled them onto yeast plasmid as the same as the construction of FKBP and Frb and transformed them into yeast. Based on the interaction of PYL1 and ABI1, we can get a wonderful scene: In the absence of ABA, the synthetic organelles composed only of PYL1 appear, because of the homodimers of PYL1. And after we add ABA into yeast, ABI1 can enter the organelles with the interaction of ABI1 and PYL1, and we can see red droplets colocalize with green droplets in cells through fluorescence microscope. In this way, new components can enter the original organelles and the time of occurrence can be regulated as it is inducer-mediated regulation. So it give our designs and functions more possibilities.
Now, we have artificially designed phase separation in cells and synthesized membraneless organelles. But how can we fulfill intended functions with synthetic organelles? Here, we propose two ideas. We reserve two sites to implement functions, which means function modules, such as enzymes in metabolism, proteins in signaling pathway, transcription factors in transcription and so on, have two sites in our designs.
Just as we characterized synthetic organelles with fluorescent proteins, we can fuse function modules between the C-termini of the interaction modules and the N-termini of HOTags. Then, the function modules can be fused into S to implement intended functions.
We introduced a specific protein, an anti-GFP nanobody, which is very small (only 13 kDa, 1.5nm-2.5nm) and has a high affinity (0.59 nM). It is a camelid antibody against GFP[8]. We can fuse GFP between the C-termini of the interaction modules and the N-termini of HOTags, and fuse function modules to the C-terminus of the anti-GFP nanobody. Then, with the help of the interaction between the anti-GFP nanobody and GFP, SPOT can load function modules, and targeted functions can be realized. You may ask: How does an anti-GFP nanobody improve the design? Firstly, it will not make the protein very large and will reduce its effect on the structure of the function modules, which can ensure the quality of the designed functions. Secondly, it can bring components not belonging to the original structure to synthetic organelles, which can extend their function. Thirdly, it is easy to regulate the expression of target proteins. Thus, nanobodies may have a surprise in store!
We implemented phase separation in vivo and synthesized artificial membrane-less organelles. The main challenge in synthesizing an organelle is to implement phase separation in a cell, so we stress the importance of interactions and multivalence. For these two aspects, we offered our ideas and their feasibility was analyzed. Finally, we proposed two approaches to implement that target functions. We believe that in the near future, making “millions of dollars” by harnessing phase separation in cells will no longer be a dream!
[1] Zhang, Q., Huang, H., Zhang, L., Wu, R., Chung, C. I., Zhang, S. Q., ... & Shu, X. (2018). Visualizing Dynamics of Cell Signaling In Vivo with a Phase Separation-Based Kinase Reporter. Molecular cell, 69(2), 334-346.
[2] Woolfson, D. N., Bartlett, G. J., Burton, A. J., Heal, J. W., Niitsu, A., Thomson, A. R., & Wood, C. W. (2015). De novo protein design: how do we expand into the universe of possible protein structures?. Current opinion in structural biology, 33, 16-26.
[3] Banani, S. F., Rice, A. M., Peeples, W. B., Lin, Y., Jain, S., Parker, R., & Rosen, M. K. (2016). Compositional control of phase-separated cellular bodies. Cell, 166(3), 651-663.
[4] Husnjak, K., Keiten-Schmitz, J., & Müller, S. (2016). Identification and characterization of SUMO-SIM interactions. In SUMO (pp. 79-98). Humana Press, New York, NY.
[5] Banaszynski, L. A., Liu, C. W., & Wandless, T. J. (2005). Characterization of the FKBP Rapamycin FrB Ternary Complex. Journal of the American Chemical Society, 127(13), 4715-4721.
[6] Park, S. Y., Fung, P., Nishimura, N., Jensen, D. R., Fujii, H., Zhao, Y., ... & Alfred, S. E. (2009). Abscisic acid inhibits type 2C protein phosphatases via the PYR/PYL family of START proteins. science, 324(5930), 1068-1071.
[7] Yin, P., Fan, H., Hao, Q., Yuan, X., Wu, D., Pang, Y., ... & Yan, N. (2009). Structural insights into the mechanism of abscisic acid signaling by PYL proteins. Nature structural & molecular biology, 16(12), 1230.
[8] Ries, J., Kaplan, C., Platonova, E., Eghlidi, H., & Ewers, H. (2012). A simple, versatile method for GFP-based super-resolution microscopy via nanobodies. Nature methods, 9(6), 582.