RK-33, a small molecule inhibitor of host RNA helicase DDX3, suppresses multiple variants of SARS-CoV-2

SARS-CoV-2, the virus behind the deadly COVID-19 pandemic, continues to spread globally even as vaccine strategies are proving effective in preventing hospitalizations and deaths. However, evolving variants of the virus appear to be more transmissive and vaccine efficacy toward them is waning. As a result, SARS-CoV-2 will continue to have a deadly impact on public health into the foreseeable future. One strategy to bypass the continuing problem of newer variants is to target host proteins required for viral replication. We have used this host-targeted antiviral (HTA) strategy that targets DDX3X (DDX3), a host DEAD-box RNA helicase that is usurped by SARS-CoV-2 for virus production. We demonstrated that targeting DDX3 with RK-33, a small molecule inhibitor, reduced the viral load in four isolates of SARS-CoV-2 (Lineage A, and Lineage B Alpha, Beta, and Delta variants) by one to three log orders in Calu-3 cells. Furthermore, proteomics and RNA-seq analyses indicated that most SARS-CoV-2 genes were downregulated by RK-33 treatment. Also, we show that the use of RK-33 decreases TMPRSS2 expression, which may be due to DDX3s ability to unwind G-quadraplex structures present in the TMPRSS2 promoter. The data presented support the use of RK-33 as an HTA strategy to control SARS-CoV-2 infection, irrespective of its mutational status, in humans.


Introduction
The causative agent of the worldwide COVID-19 pandemic is the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). Vaccines are generally considered as the first line of defense against such a large-scale viral spread. Basically, vaccines prime the host's immune system, which then coordinates a direct attack against specific virions. Indeed, rapid deployment of several safe COVID-19 vaccines has proven very effective in attenuating infections and limiting deaths. Nevertheless, as with all viruses, SARS-CoV-2 has been evolving/mutating and variants that are more transmissible have emerged (Joshi et al., 2021;Singh et al., 2021). Moreover, it is possible that the continuous evolution of SARS-CoV-2 will result in a vaccine resistant variant(s) (Singh et al., 2021) even as the efficacy of vaccines is waning (Chemaitelly et al., 2021;Fowlkes et al., 2021;Goldberg et al., 2021). Accordingly, SARS-CoV-2 will continue to have a major impact on the future of public health if development of alternative therapeutic strategies are not pursued.
Direct-acting antiviral agents (DAAs) and host-targeted antivirals (HTAs) have been developed into effective alternatives to vaccine resistance or for use in combination with vaccinations (Kumar et al., 2020). Here, we have used a HTA directed against the human host DEAD (Asp-Glu-Ala-Asp)-box RNA helicase, DDX3X (DDX3) and demonstrated a potential of providing an anti-SARS-CoV-2 therapy. We developed a small molecule inhibitor of DDX3, RK-33, that is designed to be a competitive inhibitor of the DDX3 ATP-binding site and has been demonstrated to abrogate its RNA helicase function (Bol et al., 2015;Xie et al., 2016;Heerma van Voss et al., 2018) which is required for the translation of complex structured 5′capped mRNAs. In addition, using highly accurate physical and biochemical methodologies such as, analytical ultracentrifugation, binding kinetics, and isothermal titration calorimetry, it has been demonstrated that RK-33 binds DDX3 within the Walker I motif that contributes to ATP-binding (Yang et al., 2020). Our rationale was based on evolutionary studies of SARS-CoV-2, which have revealed that it is closely related to SARS-CoV with a similar sized genome that has nearly identical organization and hence, replication (van den Born et al., 2005;Stammler et al., 2011;Yang and Leibowitz, 2015; Coronaviridae Study Group of the International Committee on Taxonomy of Viruses, Petrosillo et al., 2020;Yoshimoto, 2020). SARS-CoV-2 replication initially proceeds through a 5′cap-dependent translation using the host's translation complexes (Cencic et al., 2011). Consequently, SARS-CoV-2 also has a 5′capped mRNA genome that requires cap-dependent translation (Gordon et al., 2020b). Given the strong evidence that DDX3 participates in 5′capped translation where it forms functional complexes with several eukaryotic translation initiation co-factors including eIF4E, eIF4F, eIF4G and eIF3 (Lee et al., 2008;Geissler et al., 2012;Shih et al., 2012;Soto-Rifo and Ohlmann, 2013) and is a required functioning unit in promoting translational as a part of the 80S translation initiation complex (Geissler et al., 2012), it follows that DDX3 would be utilized in the translation of the SARS-CoV-2 genome. Consistent with this line of reasoning, recent proteomics analyses found DDX3 colocalized with SARS-CoV-2 viral particles and the latter required DDX3 for virion production (Ciccosanti et al., 2021), DDX3 interacted with SARS-CoV-2 RNA (Kamel et al., 2021), and that DDX3 was part of the SARS-CoV-2 interactome (Schmidt et al., 2021).
Similarly, DDX3 also participates in the resolution of both RNA and DNA G-quadraplex structures and such structures have been predicted to form in the SARS-CoV-2 genome (Herdy et al., 2018;Ji et al., 2021;Russon et al., 2021), in the promoter region of the host TMPRSS2 gene (Shen et al., 2020), as well as in the 5′UTR of the TMPRSS2 transcript. TMPRSS2 expression is also required for SARS-CoV-2 entry into cells (Baughn et al., 2020;Hoffmann et al., 2020;Qiao et al., 2020). In addition, three related lines of evidence indicate that DDX3 can be considered as a prime host protein to target (Winnard et al., 2021): (a) several viruses (18 species from 12 genera and 11 families) have been demonstrated as requiring DDX3 for efficient virion production; (b) investigators have actively pursued the targeting of DDX3 with small molecule inhibitors as an antiviral strategy (Schroder, 2011;Valiente-Echeverria et al., 2015;Brai et al., 2016Brai et al., , 2019aBrai et al., ,b, 2020aKukhanova et al., 2020;Quaranta et al., 2020;Yang et al., 2020); and (c) RK-33 has effectively been used to target DDX3 and abrogate virion production in Dengue, West Nile, Zika, Respiratory Syncytial, and human Parainfluenza Type-3 viral infections (Yang et al., 2020).
Our results indicate that regardless of the SARS-CoV-2 isolate (Lineage A, and Lineage B Alpha, Beta, and Delta variants), RK-33 treatment of Calu-3 cells infected with these isolates display significantly reduced viral load by one to three log orders. Consistent with this finding, proteomics and RNA-seq analyses of the infected/RK-33 treated cells demonstrated the downregulation of most of the SARS-CoV-2 proteins and genes and host TMPRSS2 expression was decreased as well. Collectively, the data obtained indicates that the use of RK-33 to abrogate host DDX3 functions is a viable option for the treatment of SARS-CoV-2 infection as well as subsequent variants that may arise.

Virus inhibitor
RK-33 was dissolved in 100% DMSO to make a 10mM stock. For a cell-based assay the stock was diluted in complete cell media to required concentrations.

CC 50 and EC 50
For CC 50 assays cells were maintained as described above in 96-well opaque white wall plates for 24 h prior to RK-33 drug treatment. Cells (2.8 × 10 4 ) were incubated with RK-33 in serially diluted concentrations from 50-0.39 μM for 24 h. Cell viability was estimated by CellTiter-Glo assay (Promega) in accordance with manufacturer's recommendation. For EC 50 assays Calu-3 cells were maintained as described above in 12-well plates for 24 h prior to EC 50 assays. Cells were treated with RK-33 in decreasing concentrations from 5-0.3 μM for 1 h followed by 1 h of SARS-CoV-2 Lineage A variant infection at multiplicity of infection (MOI) of 0.1. After incubation, viral infectant was removed, cells were washed once with PBS and fresh complete medium supplemented with RK-33 in equivalent concentrations was added. Virus containing cell media was collected 48 hours post infection (hpi) and virus titers were estimated by standard plaque assay as described elsewhere (Juarez et al., 2013;Mendoza et al., 2020).

Cell infection
Calu-3 cells (8 × 10 5 ) were plated in 12-well plates 24 h prior to drug treatment. Cells were then incubated with RK-33 at 5 μM concentration for 24 h followed by infection with SARS-CoV-2 variant as indicated, at MOI of 0.1. The infectant was removed after 1h and cells were washed once with PBS and fresh complete medium supplemented with RK-33 at 5 μM concentration was added. Cells infected with SARS-CoV-2 as well as virus containing supernatants were collected 48 h post infection.

RNA-Seq
Total RNA was extracted from virus containing cell lysates using Trizol LS (Invitrogen, Carlsbad, CA) and purified using a column purification method (Qiagen, Germany) in accordance with manufacturer's recommendations that were modified to isolate both small and large RNA species. RNA concentration and preliminary quality control was estimated on a NanoDrop spectrophotometer (ThermoFisher Scientific) while RNA-Seq was performed commercially (BGI Americas, San Jose, CA) as detailed in Supplementary Methods. Four biological replicates were analyzed from each of the treatments (Uninfected cells treated with DMSO, Uninfected cells treated with RK-33, Infected cells treated with DMSO, and Infected cells treated with RK-33).
RNA-Seq data is in the process of being uploaded to the data server.

Proteomics
Proteomics was performed at the Johns Hopkins University School of Medicine Mass Spectrometry and Proteomics Core using 50 μg of each sample as detailed in Supplementary Methods. Biological replicates of four were analyzed as in RNA-Seq.

RT-qPCR and RNA isolation
Viral RNA was extracted from virus containing cell media and from cell lysates using Trizol LS (from Invitrogen, cat# 10296010) and purified by Direct-Zol RNA purification kit (from Zymo research, cat#R2052) in accordance with manufacturer's recommendations. SARS-CoV-2 RNA (RdRp and E genes) was amplified by the one-step Logix Smart-2 COVID-19 real-time PCR reagents kit from Co-Diagnostics, Inc. (Salt Lake City, UT, United States). Normalization was performed using RNaseP gene amplified from host RNA. Copy numbers were estimated based on the included quantified positive control reagent.

Mutational analysis
The genome sequence of the Lineage A variant was retrieved from GenBank accession MW811435, a direct submission of NR-52281 sequencing at ATCC. The Alpha variant genome sequence was retrieved from GenBank accession MZ376737. The Beta variant genome sequence was retrieved from GenBank accession MZ345001. The Delta variant genome sequence was reconstructed from GenBank accession MN908947 with variations as reported on the NR-55611 Certificate of Analysis. Of particular note is a deletion within ORF7a (AA 44-100) with a frameshift that results in extension of the protein sequence with a new stop codon downstream of ORF7b.
Protein sequences for 25 SARS-CoV-2 proteins were extracted from the four genome sequences. For each protein, a four-isolate multiple sequence alignment was performed in CLC Genomics Workbench with a progressive alignment algorithm using default parameters. Differences with respect to the Lineage A variant were noted. Evolutionary distances of proteins were estimated using the WAG substitution model (Whelan and Goldman, 2001).
Three-dimensional structural models of the reference A lineage spike protein homotrimer in open and closed conformations were obtained from the Zhang Lab. 1 This model was generated using the D-I-TASSER/ C-I-TASSER pipeline (Zheng et al., 2021). The protein sequence of the structural model is identical to that of the Lineage A variant used in this study. Residues with variations found in the Alpha, Beta and delta variants were visualized on this structure with the Visual Molecular Dynamics program (Humphrey et al., 1996).

RNA-seq demonstrates RK-33 treatment downregulates multiple viral transcripts and alters the host transcriptional landscape
To study the effect of RK-33 and SARS-CoV-2 on the human transcriptome, we performed RNA-seq on isolated RNA from uninfected Calu-3 cells treated with DMSO (DMSO control), cells treated with RK-33 (RK-33 treated DMSO control), cells infected with SARS-CoV-2 and treated with DMSO (Virus), and cells infected with SARS-CoV-2 and treated with RK-33 (RK-33 treated virus). A non-toxic dose of RK-33 (5 μM) was used for these studies and cells were infected at a MOI of 0.1. All samples were collected at 48 hpi. RNA-Seq was mapped separately to human genome and SARS-CoV-2 genome. Principal Components Analysis (PCA) of host RNA showed a good separation between the infected samples and the control treated samples, while the RK-33 treated samples were clustered together. As expected, RK-33 and SARS-CoV-2 infection were the drivers for sample segregation ( Figure 1A; Supplementary Figure S1). Hierarchical clustering separated all RNA samples according to treatment with the viral infected cells being the most different compared to the other three groups ( Figure 1B). RNA of most SARS-CoV-2 genes (ORF1ab, S, ORF3a, E, M, ORF7a, ORF7b, ORF8, N, ORF10) exhibited a reduction when treated with RK-33 compared to virus infected control cells ( Figure 1C). Volcano plots were produced to compare host RNA expression between RK-33 treated, viral infected cells and untreated, viral infected cells ( Figure 1D). We also compared effects of viral infection on untreated Calu-3 cells ( Figure 1E). Many of the genes that were upregulated by the virus were significantly downregulated by RK-33 treatment. We studied genes that were significantly dysregulated (>1.1 FC, p < 0.05) by RK-33 treatment of virally infected Calu-3 cells (Infected RK-33 vs. Infected DMSO), by virus infection of the cells (Infected DMSO vs. Uninfected DMSO), and with the control (Uninfected RK-33 vs. Uninfected DMSO). The gene sets were plotted on a Venn diagram ( Figure 1F), and show the largest overlap exists between the RK-33 treated virally infected cells and the virally infected cells, indicating the widespread effects of SARS-CoV-2 on the genetic landscape. Reactome pathway analysis (Fabregat et al., 2017;Jassal et al., 2020) revealed that the top five pathways enriched in viral infected DMSO treated control cells were Metabolism of proteins, Gene expression (transcription), Metabolism, RNA Pol II transcription, and Translation ( Figure 1G). The top five Reactome pathways enriched in viral infected RK-33 treated cells were Metabolism of proteins, Metabolism, Post-translational protein modification, Gene expression (transcription), and RNA Pol II transcription ( Figure 1H). It is important to note that host translation is less significantly impacted after cells are treated with RK-33. Gene set enrichment analysis of the uninfected cells and RK-33 treated virus infected cells are shown in Supplementary Figures S2, S3. Analysis of genes dysregulated by SARS-CoV-2 infection and RK-33 treatment expectedly showed a larger effect (Supplementary Figure S3). We also compared our gene set with published data (Blanco-Melo et al., 2020) which showed a large commonality both in Calu-3 gene sets and in the deceased COVID-19 lung tissue set (Supplementary Figure S4). In addition, RNA-seq QC results indicated that all samples had around 45 million clean reads, but alignment QC indicated SARS-CoV-2 infected Calu-3 cells had around 33% unique matches in the human genome compared to the rest of the samples at 88% (Supplementary Figure S5). This is an important result, as it is known that the Nsp1 degrades host RNA so that it can hijack the host cell (Huang et al., 2011). This reversion of viral to host RNA in the RK-33 treated SARS-CoV-2 infected cells demonstrates the role of DDX3 in SARS-CoV-2 replication and the importance of targeting DDX3 by RK-33.

Proteomics analysis indicated RK-33 treatment suppresses multiple viral proteins and alters the host response to infection
To study changes in the human and viral proteome after infection with SARS-CoV-2 and RK-33 treatment, we performed proteomics analysis on SARS-CoV-2 infected Calu-3 cells treated with DMSO solvent control (Virus) and on the RK-33 treated and virus infected Calu-3 cells (RK-33 treated virus) as described above. Other controls used were uninfected Calu-3 cells treated with RK-33 (RK-33 treated DMSO control) and Calu-3 cells treated with DMSO (DMSO control). PCA showed clear separation between virally infected and uninfected cells as well as virally infected cells treated with DMSO, and virally infected cells treated with RK-33 ( To study how RK-33 treatment affected Calu-3 infected cells, we plotted the data using Volcano plots ( Figures 2C, D). As expected, most of the SARS-CoV-2 proteins (P0DTC2 spike glycoprotein, P0DTC3 ORF3a, P0DTC5 membrane protein, P0DTC8 ORF8, P0DTC9 nucleoprotein, P0DTD1 replicase polyprotein 1ab, P0DTD2 ORF9b, P59595 nucleoprotein) were significantly upregulated (>7.5 FC) after infection of Calu-3 cells. Subsequently, we performed pathway analysis using Reactome to compare RK-33 treated infected cells with control infected cells. The five most significant pathways that were enriched following SARS-CoV-2 infection were interferon alpha/beta signaling, ISG15 antiviral mechanism, viral mRNA translation, SRP-dependent cotranslational protein targeting to membrane, and peptide chain elongation ( Figure 2H). In contrast, RK-33 treatment of SARS-CoV-2 infected cells results in a shift of pathways being impacted, with the five most significant pathways enriched being Mitochondrial translational initiation, Mitochondrial translational elongation, Mitochondrial translational termination, Antigen Presentation: Folding, assembly and peptide loading of class I MHC, and SRP-dependent cotranslational protein targeting to membrane ( Figure 2I). We also queried the STRING database (Szklarczyk et al., 2019) and found pathways in RK-33 treated cells that were positively enriched (Endosomal/vacuolar pathway and Regulation of Complement cascade) and negatively enriched (Mitochondrial translation and Mitochondrial translation initiation). Pathways in virus infected DMSO control cells that were positively enriched included Regulation of IFNA signaling and Interferon alpha/beta signaling while Translation of Replicase and replication transcription complex and Endosomal sorting complex required for transport were negatively enriched (Supplementary Figure S8). These data indicate that RK-33 treatment suppresses viral protein production corresponding to changes in the host proteomic landscape especially as it relates to innate immune response pathways.

RK-33 inhibits HCoV-OC43 and SARS-CoV-2 production
Given that DDX3 has been shown to play a regulatory role in the viral biogenesis, we initially tested the impact of RK-33 on coronavirus replication using human coronavirus OC43 (betacoronavirus) as a BSL-2 model of SARS-CoV-2. The CC50 value of RK-33 on Rhabdomyosarcoma (RD) muscle cells (highly susceptible to HCoV-OC43 infection), was determined to be 3.22 μM ( Figure 3A). Next the antiviral impact of RK-33 was tested on RD cells using non-toxic concentrations of RK-33 (0.1 and 1 μM). Treatment with 1 μM of RK-33 decreased HCoV-OC43 infectious titers by over 100-fold (> ~ 2 logs; Figure 3B). As can be seen in Figure 3C, HCoV-OC43 infection of RD cells results in extensive cytopathic effect (CPE) by 4 dpi (left panel) that was partially relieved by 1 μM of RK-33 treatment (right panel). These data indicate that RK-33 is effective against a betacoronavirus, which encouraged us to test its impact on SARS-CoV-2.
Thus, since in our study we used Calu-3 cells (lung cancer cell line) as a model cell line for SARS-CoV-2 infection, our initial step in antiviral testing was to establish the CC50 of RK-33 in Calu-3 cells. For this purpose, cells were incubated with RK-33 in concentrations ranging from 0.39-50 μM for 24 h. Cell viability was measured by CellTiter-Glo assay which measures ATP production. The established CC50 for Calu-3 cell lines was 13.48 μM ( Figure 3D). Next, the EC50 of RK-33 against SARS-CoV-2 was determined by treatment of Calu-3 cells with 0.31-5 μM RK-33 or 0.05% DMSO for 24 h prior to infection. RK-33 and DMSO treated cells were infected with SARS-CoV-2 Washington isolate (Lineage A) at an MOI of 0.1 and 48 hpi the amount of infectious virus in the supernatants was determined by plaque assay. Overall, RK-33 displayed potent anti-SARS-CoV-2 effect on Calu-3 cells with an EC 50 < 1 μM ( Figure 3E), resulting in a selectivity index (CC50/ EC50) of 14.5. To confirm these results, a similar experiment was performed with cells treated with DMSO or 5 μM RK-33. In agreement with the results in Figure 3E, RK-33 treatment had a significant three log reduction in SARS-CoV-2 viral titers ( Figure 3F) RK-33 treatment also resulted in a reduction of ~2.5-3 log10 of intracellular viral RNA production as measured by RT-qPCR analysis using primers recognizing RdRp and E genes/ regions ( Figure 3G). Finally, a one log reduction in extracellular viral RNA (from cell-free supernatants) was also observed ( Figure 3H) in RK-33 treated samples.

Mutation status of SARS-CoV-2 proteins
Along with its worldwide spread, SARS-CoV-2 virus has mutated resulting in genome-wide amino acid variations across the four variants of varying lineages as displayed in Figure 4A. Protein N in the Alpha and Delta variants shows greater divergence from Lineage A than the spike proteins of these same variants. Proteins E, nsp7, and ORF3a have diverged more than the spike in the Beta variant, and the Delta variant displays some level of divergence from Lineage A in the M, nsp2, nsp4, nsp6, and NendU proteins. Due to large deletions and a frameshift, the ORF8 protein in the Delta variant has diverged from Lineage A to a greater extent than spike protein 2 (Supplementary Table 1). Furthermore, Figure 4B displays amino acid substitutions and deletions with respect to the Lineage A variant in sixteen of the proteins. This includes large truncations in ORF7a (Beta and Delta variants) and ORF8 (Alpha variant) and a high number of variations in the hACE2 receptorinterfacing spike proteins. Spike protein amino acid substitutions and deletions with respect to Lineage A variant are further visualized along the protein sequence at higher resolution in Figure 4C. These variations occur throughout the spike protein molecular structure, including within the S1-receptor-bindingdomain as seen in the 3D models in Figure 4D. Details of mutations can be found in Supplementary Table 1.

RK-33 inhibits Alpha, Beta, and Delta variants of SARS-CoV-2
Due to the diversity between different viral variants, we next tested the effect of RK-33 on replication and virion production of different of SARS-CoV-2 variants. RK-33 treatment of cells resulted in a significant reduction (2-4 log fold) of infectious Frontiers in Microbiology 08 frontiersin.org titers of the Alpha, Beta, and Delta variants ( Figure 5A). Analysis of the intracellular and extracellular RNA (RdRp and E genes) also showed significant difference between RK-33 treated and DMSO control samples in the amount of the viral RNA production (Figures 5B-G). In conclusion, RK-33 displayed potent antagonistic SARS-CoV-2 effects against all tested variants.

RK-33 effects TMPRSS2 and reduces spike-pseudotyped lentivirus constructs in vitro
The transmembrane serine protease TMPRSS2 has been identified as the protease that cleaves spike protein of SARS-CoV-2 thus facilitating entry (Hoffmann et al., 2020). To determine if Comparison of mutation status of SARS-CoV-2 variants. (A) Radar plot displaying estimated protein sequence evolutionary distance to the Lineage A variant for the Alpha (purple), Beta (green), and Delta (blue) variants. Only proteins that contain substitutions in at least one of the variants are shown. Proteins are ordered to group similar distances together for each isolate. *Delta ORF7a protein sequence has an estimated distance of 0.67 from Lineage A due to a deletion resulting in a frameshift. **Delta ORF8 protein sequence has an estimated distance of 0.02 from the Lineage A variant. These distances are plotted at the limits of the plot for clarity. Frontiers in Microbiology 10 frontiersin.org abrogating DDX3 activity by RK-33 would affect TMPRSS2 expression, we treated Calu-3 cells with increasing concentrations of RK-33 for 48 h and probed for the levels of TMPRSS2 and DDX3 by immunoblotting. As shown in Figure 6A, RK-33 (5 μM) was able to reduce TMPRSS2 expression by 50% in 48 h (Fiji / ImageJ). To determine if the decrease in TMPRSS2 level by RK-33 had any biological significance on viral entry, we treated Calu-3 cells with RK-33 for 24 h, followed by infection with a SARS-CoV-2 Spikepseudotyped GFP lentivirus construct. Two days following infection, we counted the GFP cells using a random unbiased approach. As shown in Figure 6B, use of RK-33 at CC5 (3 μM), reduced the infected cell numbers by 50%. This supports our hypothesis that dysregulating DDX3 by RK-33 can affect TMPRSS2 expression resulting in decrease in SARS-CoV-2 lentivirus infection of Calu-3 cells.

Discussion
SARS-CoV-2 has been successfully targeted in a few ways. Prophylactic vaccines are available that reduce/eliminate hospitalization and associated medical complications due to COVID-19. Presently, other treatments are limited to the antiviral nucleoside analog Remdesivir from Gilead Biosciences and a monoclonal antibody cocktail from Regeneron. However, Remdesivir is only prescribed in severe cases, and both require intravenous infusion making it difficult to obtain without hospitalization. Moreover, Remdesivir has shown mixed results in reducing the time to recovery in hospitalized patients (Beigel et al., 2020;Malin et al., 2020;Spinner et al., 2020;Wang et al., 2020;Garibaldi et al., 2021). Another nucleoside analog drug Molnupiravir from Merck reduces the risk of hospital admission or death by approximately 50% in patients at risk for poor outcomes (Mahase, 2021).
On the other hand, identifying virus-host interacting factors could reveal new targets to mitigate medical complications associated with SARS-CoV-2 infection. Over the years, several roles of a host protein, DDX3, have been identified as crucial for replication and production of many viruses (Squeglia et al., 2020;Ciccosanti et al., 2021). Interestingly, DDX3 has also been identified, using proteomics, to be involved in SARS-CoV-2 replication (Squeglia et al., 2020;Ciccosanti et al., 2021). Moreover, computational and system biology approaches have indicated different proteins of SARS-CoV-2 to be associated with cellular organelle functions that are regulated by DDX3 activity (Gordon et al., 2020a). In this study, we used a small molecule inhibitor of DDX3, RK-33, to study its impact of SARS-CoV-2. Using the Lineage A variant, we showed that RK-33 reduced viral titers by over 10,000-fold by plaque assay. These data were further validated by RT-qPCR (RdRp and envelope) of the intra-and extracellular SARS-CoV-2 RNA of the RK-33 treated Lineage A infected Calu-3 cells. This work is in agreement with a study showing that inhibition of DDX3 by two different DDX3 inhibitors, RK-33 and C-4B, impacted SARS-CoV-2 RNA production, but this study did not determine the impact on viral titers by plaque assay (Ciccosanti et al., 2021).
Based on this encouraging finding, we performed RNA-seq and proteomics analysis on SARS-CoV-2 infected Calu-3 cells, before and after RK-33 treatment. PCA of the RNA showed the major effect of RK-33 on the cells. It was interesting to note that in the virus infected Calu-3 cells, more than 80% of the RNA species was that of SARS-CoV-2, while in the RK-33 group, the host mRNA species was 80% represented. This is indicative that perturbing DDX3 by RK-33 restored many of the host cellular functions while suppressing SARS-CoV-2 replication. This could be clearly seen in both the heat map and the PCA plot with RK-33 treatment serving to "normalize" expression of the virus infected cells toward the uninfected cell type gene expression. Volcano plots also demonstrated how the high gene expression patterns caused by the virus infection were modulated by RK-33 treatment of the cells. The sensitivity of RK-33 to reduce Lineage A titers is indicative of the potential of targeting host proteins for SARS-CoV-2 treatment.
To further characterize the effect of RK-33 on the host cellular machineries, we performed proteomics on SARS-CoV-2 infected Calu-3 cells. Proteomic-based PCA analysis showed that RK-33 treated virus infected samples segregated from the untreated infected samples. The results indicated that RK-33 was the major determinant of segregation in the PCA plot. Subsequent pathway analysis showed that many of the downregulated pathways were associated with virus attachment, entry, assemble and release. Proteomics analysis clearly demonstrated that most of the proteins involved in SARS-CoV-2 genome replication and transcription (ORF1 ab, ORF3A, Spike, Membrane, ORF7A and Nucleoprotein) were significantly downregulated by RK-33 in SARS-CoV-2 infected Calu-3 cells. The ability of RK-33 to reduce viral titers of SARS-CoV-2 Lineage A could reflect on the dependence of SARS-CoV-2 on DDX3 for its replication. This is probably because at the molecular level SARS-CoV-2 usurps DDX3 in the host cell to its advantage for replication and for pathogenesis. This is in parallel to what is already demonstrated by many other viruses that utilize DDX3 for viral entry, processing, and egress from the host cells (Winnard et al., 2021). Furthermore, as DDX3 has been shown to interact with nucleocapsid protein of SARS-CoV-2 (Ciccosanti et al., 2021), it is plausible that the use of RK-33 is directly interfering with SARS-CoV-2 replication machinery as well.
Based on these promising results, we evaluated the effect of RK-33 on viral titers in the Alpha, Beta, and Delta variant-infected cells. RK-33 was able to reduce the viral loads by 2-3 log folds. This is not unexpected as targeting host factors essential for viral replication, such as DDX3, would have similar outcomes for all the variants. As is evident from the spike protein modeling, there is a higher concentration of spike protein mutation within the ACE-2 binding region, but this does not affect the ability of RK-33 to reduce viral titers across all the variants. Also, the amino acid substitutions lead to increased evolutionary distances in different proteins among the four isolates used in this study. Thus, the four variants used in this study could serve as diverse models adapted to provide a growth advantage or independent evolution of SARS-CoV-2 leading to increased virulence and pathogenesis. Furthermore, we showed that RK-33 was able to reduce TMPRSS2 expression in Calu-3 cells. A key aspect of the TMPRSS2 gene is the presence of guanine-rich tracts of DNA in its gene promoter that can form G-quadruplex structures. These structures can be targeted by benzoselenoxanthene derivatives that stabilize TMPRSS2 G-quadruplexes in vitro, thus downregulating TMPRSS2 gene expression (Shen et al., 2020). In this context, DDX3 was found to facilitate both translation of general complex secondary structures including G-quadruplexes (Lai et al., 2008(Lai et al., , 2010, as well as mRNAs with secondary structure in immediate vicinity to their m 7 GTP cap (Soto-Rifo et al., 2012). We have presented data that demonstrates that use of RK-33 even at CC5, reduced the SARS-CoV-2 infection rate by 50%. In support of our findings, a recent paper by Liu et al. (2022) demonstrated that the use of G-quadruplex specific stabilizers could reduce SAR-CoV-2 infection in a pseudovirus cell system and in mouse models.
These data indicate that targeting DDX3 by RK-33 for SARS-CoV-2 treatment could potentially reduce SARS-CoV-2 virulence and pathogenicity by two different mechanisms. First, it can decrease the infection rate by downregulating TMPRSS2 protein expression, and second, reduce the pathogenicity, by suppressing SARS-CoV-2 replication. This two-prong mechanism of action would be extremely effective at suppressing viral replication of any newly emerging variants of concern, which is in line with an antiviral that is targeting a host factor. We fully appreciate that the impact of RK-33 on TMRPSS2 expression is only one mechanism by which RK-33 may be impacting SARS-CoV-2 production. Indeed, our proteomic and transcriptomic analysis points to multiple cellular events being altered in infected cells by RK-33 treatment, such as innate immune responses, translation, and membrane trafficking, each of which are important for viral production.
In conclusion, we report on the importance of DDX3 and its targeting by the small molecule inhibitor RK-33 which was developed earlier by rational drug design that resulted in a unique structural scaffold with little or no toxicity in animals (Kondaskar et al., 2010;Bol et al., 2015;Xie et al., 2016;Heerma van Voss et al., 2018). Use of RK-33 on Calu-3 cells resulted in suppressing virus titers from all four isolates and has the potential to treat SARS-CoV-2 infection. We are the first to present a promising HTA strategy to tackle the different variants associated with SARS-CoV-2 as well as subsequent variants that may arise.

Data availability statement
The data presented in the study are deposited in the Gene Expression Omnibus (GEO) repository, accession number GSE209800.