Despite successful vaccine development, drugs against SARS-CoV-2 are still needed to manage active infections. Building on existing knowledge regarding the involvement of GSK3β, we aimed at identification of compounds active against this key human kinase as potential coronaviridae inhibitors. Screening a targeted library provided a high rate of active compounds with greater than 50% inhibition of viral infection, with nearly half the compounds conferring some inhibitory effect against SARS-CoV-2 and HCoV-229E. Compared to the ‘hit’ rate of non-targeted host-directed therapy screens of ~3%   or ~0.1% in non-specific screens , our results demonstrate a high benefit-to-investment ratio. More importantly, this screen demonstrates that more general host-kinase screens, which typically have 1-4 compounds representing each host kinase type, might be insufficient to rule-out specific kinases as potential targets. This is made evident through our study, as many GSK3β inhibitors in this library did not pass the 50% activity threshold. As such, a larger, focused screen might be a more comprehensive approach to determine the host-pathogen involvement of specific host kinases compared to general representative screens. Although beyond the scope of this study, targeted libraries also offer fascinating opportunities for further structure-function relationship studies.
Comparative analysis of SARS-CoV-2 and HCoV-229E screens suggests that viral infectivity reduction caused by GSK3β inhibition is not limited to SARS-CoV  and SARS-CoV-2, and thus may serve as a general target for development of drugs against various coronaviridae. Considering the timeline for new drug development and that approval is lengthy and continued coronavirus spread is likely, the ongoing COVID-19 pandemic necessitates exploratory development of such broad-spectrum antivirals. This is further supported by conserved arginine-serine rich domains, target regions for GSK3 phosphorylation, in N protein sequences across coronaviruses . Despite limited 20-30% overall N protein sequence similarity, the domains maintain repeat motifs conducive for repeat phosphorylation .
Using SARS-CoV-2 nucleocapsid as a measurement of viral infection resulted in a robust Z’ score for the screen (Z’ =0.6) when compared alternatively to the dsRNA marker (Z’=0.3, supplementary Table 1), likely due to ability of the antiviral dsRNA antibody to cross react with the host cell dsRNA. One viral marker may be insufficient to draw conclusions regarding viral infection, as in the case of GSK3β, due to potential target-marker interactions. Although treated cells do accumulate nucleocapsid, we found a strong agreement between nucleocapsid and dsRNA marker levels in their response to the inhibitors tested, suggesting that in this high-content screening assay, nucleocapsid can be used as a marker for viral infection. Screen inhibition readouts were similar (mean difference of 10%) to those in nucleocapsid expression. Media-released virus measured using plaque assay quantification confirmed that the inhibitory effect of GSK3β-inhibition extends to SARS-CoV-2 assembly and maturation.
T-1686568 was selected due to its high selectivity index. Similar to the validation of viral infection measurements using different viral markers, utilising different cell lines is important for any robust immortalized cell line screen, and particularly for compound activity against SARS-CoV-2, given the many reporting mismatches of screen results between different cell lines . To guard against this, we internally validated our screen through the use of different cell lines and found a consistent T-1686568 inhibitory effect. Of particular note, host-targeting inhibitors designed against viral entry mechanisms, such as ACE2 and TMPRSS2, present higher variability in the expression of these host factors between cell lines . Our observations suggest GSK3β is a conserved pathway critical to SARS-CoV-2 infection in many tissues, and strongly recommend it be considered in the selection of targets for drug development. This is further supported through the observation of SARS-CoV-2 inhibition in patient-derived colon organoids, which contain multiple different types of infectible cells .
Mass spectrometry studies have demonstrated that the RS domain of the N protein that encompasses residues Ser-176 to Ser-276 undergoes extensive phosphorylation (Supplemental Figure 3). Using synthetic peptides modeled after these phosphosites, we have shown that they can be directly phosphorylated by recombinant GSK3β, provided that these peptides are prephosphorylated at priming sites. Recombinant PKC-alpha was able to phosphorylate several of the priming sites for the GSK3β in vitro, although other kinases may be responsible for this in vivo. Our findings complement the recent in vivo studies of Liu et al. , who showed that GSK3 acted upstream of the N protein to mediate its phosphorylation, and by site-directed mutagenesis determined that the suspected Ser-188 and Ser-206 priming phosphosites for GSK3 recognition were required for phosphorylation-dependent mobility shifts of the N protein on SDS-PAGE gels by Western blotting. Thr-205 may also act a priming site for phosphorylation of the Ser-201 and Ser-197 sites that are known to be phosphorylated in cells and predicted for GSK3 targeting as well.
The SARS-CoV-2 N protein is an abundant RNA-binding protein critical for viral genome packaging, yet the molecular processes and characteristics of this function are not fully understood. The N protein is a phosphoprotein that associates with the viral RNA genome to form the ribonucleoprotein core . The N protein contains three dynamic disordered regions that house putative transiently-helical binding motifs. The two folded domains interact minimally such that full-length N protein is a flexible and multivalent RNA-binding protein.
The phosphorylation of the RS domain in SARS-CoV and SARS-CoV-2 has been implicated in the regulation of N protein binding to RNA, multimerization and subcellular location [28–32] Studies with SARS-CoV-2 indicate that changes in the phosphorylation status of the RS domain induces profound alterations in the association of multiple nucleocapsid proteins with a single viral RNA in a structured oligomer with RNA-protein and protein-protein interactions to switch to one that permits more viral genome processing . Following N protein phosphorylation in the RS domain, the RNA-protein complex is able to recruit the stress granule protein G3BP1 and suppress the G3BP-dependent host immune response . In our study, the blockage of N protein phosphorylation by GSK3 appeared to result in partial accumulation of the N protein, and completely prevented spike protein production to allow formation of the virus particles.
Several recent studies [33, 35–40] have shown that the N protein also undergoes liquid-liquid phase separation when mixed with RNA, and polymer theory predicts that the same multivalent interactions that drive phase separation also engender RNA compaction. Our future studies are geared toward validating our hypothesis that phosphorylation of the SARS-CoV-2 N protein by GSK3β is needed for forming a physiological ‘phase separation’ state required for the N protein / RNA assembly and budding.
Interestingly, evidence in the literature suggest that the spike protein is also a target for phosphorylation  that may affect its expression level and cellular trafficking. Coincidently, we noticed that the reduced levels of the Spike protein S1 subunit (Figure 3) are consistent with T-1686568 induced decrease of the viral titer. Together, our findings strongly support inhibition of GSK3 as an effective anti-viral strategy for treatment of COVID-19 and other coronaviruses at their earliest stages.