SARS-CoV-2 nsp8 forms dimer and tetramer in solution
We expressed and purified SARS-CoV-2 nsp8 with a C terminal 8xHis-tag. During the purification process, we found that nsp8 forms multi-oligomer and nucleic acid contamination when the purification buffer contains 300 mM NaCl. To avoiding the contamination by nucleic acids, the concentration of NaCl in purification buffer was increased to 1 M. It is interesting that nsp8 was eluted from gel filtration column as two peaks (Fig. 1A). The oligomerization state of nsp8 from two elution peaks was analyzed using dynamic light scattering (DLS) and small angle neutron scattering (SANS). The calculated molecular weight of nsp8 from the first and second elution peak was 90-93 kDa and 40-49 kDa, respectively (Fig. 1B). As the theoretical molecular weight of nsp8-His is 22.98 kDa, data indicate that in addition to the purified dimer of nsp8 as reported previously 14, a tetramer form of nsp8 was identified. Then, we characterized the shape of nsp8 dimer and tetramer based on the DLS and SANS measurements. Hydrodynamic radius (Rh) as measured by DLS reflects the solvated protein size, while radius of gyration (Rg) as measured by SANS reflects mostly the compositional distribution of hydrogenated protein molecules. The ratio of Rg to Rh (Rg/Rh) offer the shape information of protein molecules15. The larger Rg/Rh values of nsp8 dimers than those of nsp8 tetramers imply the variation of the shape of molecules from non-spherical to globular (Fig. 1C and D).
Destabilization of both nsp8 dimer and tetramer as salt concentration decreasing
Interestingly, DLS analysis demonstrated that the Rh of nsp8 dimer and tetramer increased with decreasing NaCl concentration (Fig. 1C). This corroborated well with the SANS measurements that revealed the increase of the Rg at lower NaCl concentrations for both nsp8 dimer and tetramer (Fig. 1D). Both Rh and Rg of dimers and tetramers increase noticeably when the concentration of NaCl is decreased, indicating a looser structure in the buffer containing low concentration of NaCl. These data may suggest that lowering salt concentration destabilizes the structure of both nsp8 dimers and tetramers. To test the hypothesis, we assessed the thermo-stability of nsp8 dimer and tetramer, and revealed that both the melting temperature (Tm) and the onset temperature of aggregation (Tagg) indeed decreased at lower NaCl concentration for both nsp8 dimer and tetramer (Fig. 1E and F). Furthermore, quantitative comparison of Tm and Tagg values of the two forms of nsp8 indicates that the dimer form is more stable than the tetramer one.
Different phase separation behavior of nsp8 dimer and tetramer
The expanded size and decreased stability with decreasing NaCl concentration may imply that nsp8 dimer and tetramer can undergo phase separation at low NaCl concentration. To test the hypothesis, we directly diluted nsp8 with a low NaCl concentration buffer. Liquid-like droplets formed in 1 mg/mL nsp8 solution containing 50 mM NaCl (Fig. 2A). When the concentration of NaCl was increased to 100 mM, nsp8 dimer failed to undergo phase separation at 1 mg/mL (Fig. 2A), whereas at 2 mg/mL, nsp8 dimer also formed liquid-like droplet in the buffer containing 100 mM NaCl (Fig. 2B). For fluorescent microscopy observation, nsp8 was labeled with His-tag labeling dye RED-tris-NTA. Fluorescent microscopy confirmed that nsp8 dimers form liquid-like droplets at 1 mg/mL nsp8 in the 50 mM NaCl buffer, and nsp8 condensates can be dissolved upon increasing the NaCl concentration to 275 mM by mixing nsp8 dimer (1 mg/mL) dissolved in 500 mM NaCl with volume ratio 1: 1 (Fig. 2C). The liquid-like nature of the droplet was further confirmed by fluorescence recovery after photobleaching (FRAP) experiment as rapid recovery of fluorescence was observed after photobleaching (Fig. 2D). These indicated that nsp8 dimer undergoes liquid-liquid phase separation in solution in a protein and NaCl concentration dependent manner.
The thermal stability analysis clearly showed that the nsp8 tetramer is significantly less stable than the dimer (Fig. 1E and F). Then, we tested whether nsp8 tetramers exhibit different phase separation behaviors compared to the dimer. Consistent with our hypothesis, nsp8 tetramer forms solid-like sediments instead of liquid-like droplet at concentration of 1 mg/mL in the buffer containing 50 mM or 100 mM NaCl (Fig. S2). Solid-like condensates can still be observed even as the concentration of nsp8 tetramer decreased to 0.25 mg/mL with low NaCl concentration, which can be reversed by simply increasing NaCl concentration to 275 mM (Fig. 2E).
To investigate whether the N terminal IDR is essential for phase transformation, we carried out a truncation of nsp8 by deleting the N terminal 76 residues (nsp8ΔN76). Similar to the wild type nsp8, nsp8ΔN76 also forms homodimer and tetramer in solution (Fig. S3). However, the deletion of N terminal 76 residues unequivocally abolished droplet formation of nsp8 dimer (Fig. 2F), suggesting the importance of the N terminal IDR for the LLPS of the nsp8 dimer. Interestingly, with the deletion of N terminal 76 residues, nsp8 tetramer mutants were less aggregated at 50 mM NaCl compared to the wild type nsp8 tetramer (Fig. 2G). The observation suggested that N terminal IDR play a key role in over-aggregation of nsp8 tetramers at decreasing NaCl concentration.
RNA modulates LLPS of nsp8 tetramer
N-terminus of nsp8 is a defined RNA binding motif, which can be stabilized by binding with RNA 8, 12. We speculated that RNA binding may induce nsp8 tetramer transition from solid-like condensate to liquid-liquid phase separation at low NaCl concentration. To confirm this, we designed a 12-nt ssRNA (R12) with sequence derived from SARS-CoV-2 genome adjacent to poly (A). When mixing RED-tris-NTA labeled nsp8 tetramer (0.25 mg/ml) with R12, liquid-like droplet can be observed with RNA concentration up to 75 µM (Fig. 3A). To verify whether RNA modulating LLPS of nsp8 tetramer depends on the sequence and the length, LLPS experiments were performed in the presence of 12-nt poly U (U12) or 22-nt poly U (U22) (Fig. 3A). We found that, when mixed with 75 µM U12, nsp8 tetramer still forms solid-like structure instead of liquid-like droplet. However, nsp8 tetramer turned to liquid-like droplet when mixing with 22-nt poly U (U22) at concentration as low as 7.5 µM. Thus, data suggested that LLPS of nsp8 tetramer depend on both RNA sequence and length. To confirm that the phase separated droplets were formed by nsp8 binding with RNA, we synthesized 6-FAM labeled R12. Fluorescent microscopy demonstrated that nsp8 and RNA were co-localized in the phase separated droplet (Fig. 3B).
nsp8 forms condensates in living cells
Then, we asked whether nsp8 also undergoes liquid-liquid phase separation in live human cell line. To address this notion, nsp8 fusion with monomeric red fluorescent protein mCherry at C-terminal (nsp8-mCherry) was over-expressed in Hela cells. By imaging this cell line using fluorescent microscopy, we observed that over-expressed nsp8 could form condensates in cytoplasm, while over-expressed mCherry was diffuse in the entire cell (Fig. 4). Taken together, results suggested that SARS-CoV-2 nsp8 undergo phase separation in solution, as well as in live cells.