RK-33 Is a Broad-Spectrum Antiviral Agent That Targets DEAD-Box RNA Helicase DDX3X

Viral disease is one of the greatest burdens for human health worldwide, with an urgent need for efficacious antiviral strategies. While antiviral drugs are available, in many cases, they are prone to the development of drug resistance. A way to overcome drug resistance associated with common antiviral therapies is to develop antivirals targeting host cellular co-factors critical to viral replication, such as DEAD-box helicase 3 X-linked (DDX3X), which plays key roles in RNA metabolism and the antiviral response. Here, we use biochemical/biophysical approaches and infectious assays to show for the first time that the small molecule RK-33 has broad-spectrum antiviral action by inhibiting the enzymatic activities of DDX3X. Importantly, we show that RK-33 is efficacious at low micromolar concentrations in limiting infection by human parainfluenza virus type 3 (hPIV-3), respiratory syncytial virus (RSV), dengue virus (DENV), Zika virus (ZIKV) or West Nile virus (WNV)—for all of which, no Food and Drug Administration (FDA)-approved therapeutic is widely available. These findings establish for the first time that RK-33 is a broad-spectrum antiviral agent that blocks DDX3X’s catalytic activities in vitro and limits viral replication in cells.


Introduction
Viral infections pose a risk to human health worldwide, with an urgent need for efficacious antiviral strategies. The relatively few antivirals in use typically function by targeting a specific viral protein critical to viral replication, and hence are only effective against a single pathogen. Although potentially effective in the short-term, their use can lead to selection for drug resistant strains that threaten the long-term effectiveness of the treatment [1]. To mitigate the risk of developing drug-resistant strains, one strategy is to instead target host cellular cofactors critical to viral replication.
DEAD-box helicase 3 X-linked (DDX3X) is a cellular adenosine 5 -triphosphate (ATP)-dependent RNA helicase that hydrolyzes ATP to unwind duplex RNA, thereby remodeling RNA/RNA-protein complexes; these catalytic functions are attributable to highly conserved sequence motifs distributed over two recombinase A (RecA)-like helicase domains [2,3]. As an ATP-dependent RNA helicase, DDX3X has roles in all aspects of RNA metabolism including transcriptional regulation, pre-mRNA

Analytical Ultracentrifugation
Sedimentation velocity experiments were conducted using an Optima Analytical Ultracentrifuge (Beckman Coulter, Brea, CA, USA) at a temperature of 20 °C. Protein was diluted in 20 mM Tris, pH 8.0, 150 mM NaCl, 10% (v/v) glycerol, 1 mM TCEP and where relevant, 100 μM RK-33 or an equivalent % DMSO was added prior to centrifugation. In total, 380 μL of sample and 400 μL of reference solution were loaded into a conventional double sector quartz cell and mounted in a K230E. c(s) distributions from fits to data collected at 290 and 410 nm were plotted as a function of s 20,w for DDX3X in the presence of 100 µM RK-33 at 290 nm (blue) and 410 nm (pink), 100 µM RK-33 alone (410 nm, gray) and DDX3X with an equivalent % dimethyl sulfoxide (DMSO) (DDX3X alone, 290 nm, black). The residual plots are shown in insets. Data in (C,D) are each representative of two independent experiments. (E,F) Calorimetry studies of RK-33 binding to DDX3X. RK-33 at (E) 50 µM and (F) 200 µM was injected into DDX3X (17 µM) in the sample cell. Data were fit to a one-site binding model. The stoichiometry deduced from data fitting to the two experiments was 1.05 ± 0.2, with a K d of 33 ± 2 µM and ∆H of −19 ± 0.3 kcal/mol. Reported values represent the average of the two experiments, with average errors of the fits to the experimental data reported by Origin 7.0.

Analytical Ultracentrifugation
Sedimentation velocity experiments were conducted using an Optima Analytical Ultracentrifuge (Beckman Coulter, Brea, CA, USA) at a temperature of 20 • C. Protein was diluted in 20 mM Tris, pH 8.0, 150 mM NaCl, 10% (v/v) glycerol, 1 mM TCEP and where relevant, 100 µM RK-33 or an equivalent % DMSO was added prior to centrifugation. In total, 380 µL of sample and 400 µL of reference solution were loaded into a conventional double sector quartz cell and mounted in a Beckman 8-hole An-50 Ti rotor (Beckman Coulter, Brea, CA, USA). Samples were centrifuged at a rotor speed of 40,000 rpm and the data were collected continuously at multiple wavelengths (290 and 410 nm). Solvent density (1.0331 g/mL at 20 • C) and viscosity (1.385 cP at 20 • C), as well as estimates of the partial specific volume (0.7247 mL/g for DDX3X at 20 • C), were calculated using the program SEDNTERP (Durham, NH, USA) [24]. Sedimentation velocity data were fitted to a continuous size (c(s)) distribution model using the program SEDFIT (Bethesda, MD, USA) [25].

RNA-Dependent ATP Hydrolysis Assay
DDX3X-mediated ATP hydrolysis was measured using a modification of the protocol of Epling et al. [26], except that activity was measured in the presence of poly(I:C), a synthetic analogue of dsRNA (InvivoGen, San Diego, CA, USA), to stimulate ATPase activity. Reactions were performed in pentaplicate in a final volume of 50 µL in 96-microwell assay plates (Corning, Corning, NY, USA) and reagents diluted using diethylpyrocarbonate (DEPC)-treated water. Reactions consisted of ATPase reaction buffer (20 mM Tris-Cl pH 7.5, 1.5 mM DTT, 1.5 mM MgCl 2 ), 200 nM DDX3X (residues 1-580) or the DDX3X K230E (residues 1-580) variant as a negative control, 0.02 mg/mL poly(I:C), with RK-33 (0-250 µM) or equivalent (2% (v/v)) DMSO, and 0.3 mM ATP (Sigma-Aldrich). Following incubation for 60 min at room temperature, 100 µL Biomol ® Green (Enzo Life Sciences, Farmingdale, NY, USA), a colorimetric reagent that binds to free inorganic phosphate (Pi), was added to all wells. After 15 min at room temperature, absorbance at 620 nm, which correlates with the detection of Pi, was measured using a FLUOstar Omega plate reader (BMG Labtech, Ortenberg, Germany). Phosphate standards (Enzo Life Sciences, Farmingdale, NY, USA) were serially diluted (0-40 µM) to generate a standard curve to enable quantitation of the Pi generated by DDX3X catalytic activity in the assay.

RNA Unwinding Assay
RNA binding assays were performed based on the protocol of Kim and Seo [27], but with the modifications described below.

Plaque Assay
BHK-21 (used for DENV-2) or Vero (used for RSV, hPIV-3, ZIKV and WNV) cells were seeded into 24-well plates at a density of 2 × 10 5 cells/well and grown overnight in culture medium at 37 • C with 5% CO 2 prior to infection. The virus inoculum was removed and replaced with semisolid overlays of 0.8% aquacide II (Calbiochem, San Diego, CA, USA) in DMEM containing 2% FBS, and the mixture was incubated at 37 • C with 5% CO 2 . After 3-4 days, the cells were fixed with neutral buffered formalin (Sigma-Aldrich, St. Louis, MO, USA) for 2 h at room temperature, rinsed with tap water, and stained with 1% crystal violet for 10 min. The stain was removed by rinsing the cells with tap water, and the viral plaques were counted visually. Dose-response curves were plotted from the plaque number (plaque forming units (pfu)/mL) versus the logarithmic value of the concentrations of RK-33. Statistical analysis was performed using GraphPad Prism 8 software. Outlier points were excluded.

Results
RK-33 is a REN with anticancer properties [11,[13][14][15][16] speculated to target the ATP binding site of DDX3X [11]. Since DDX3X plays a role in the replication of various viruses [5][6][7][8], we set out here for the first time to confirm RK-33 s ability to bind directly to DDX3X, inhibit its enzymatic activities, and test its ability to inhibit a range of viral infections.

RK-33 Binds Directly to the Active Site of DDX3X
To confirm direct binding of RK-33 to recombinant human DDX3X ( Figure 1A), we exploited the spectroscopic properties of RK-33, which absorbs strongly between 330 and 430 nm ( Figure 1B), in multi-wavelength analytical ultracentrifugation experiments. We collected sedimentation velocity data at 290 and 410 nm; the latter enabled the specific detection of RK-33 either alone or in complex with DDX3X, or DDX3X K230E that harbors a substitution within motif I (Walker A) and is deficient in ATP hydrolysis and RNA-unwinding [6]. At 290 nm and in the absence of RK-33, both DDX3X ( Figure 1C, black) and DDX3X K230E ( Figure 1D, black) sedimented as single species with sedimentation coefficients (s 20,w ) of 3.6S. DDX3X data collected at both 290 and 410 nm and in the presence of 100 µM RK-33 also revealed a 3.6S species, indicating RK-33 bound directly to DDX3X ( Figure 1C, pink). However, when DDX3X K230E was incubated in the presence of 100 µM RK-33 this overlapping species was absent ( Figure 1D, pink), indicating a lack of complex formation. These findings imply that RK-33 binds residue K230 within the Walker A motif, which forms part of the DDX3X active site.
The thermodynamics of RK-33 binding to DDX3X was measured by ITC ( Figure 1E,F). RK-33 bound weakly ( Figure 1E), with only low c value hyperbolic curves [33] obtained when using higher concentrations of RK-33 ( Figure 1F). Fitting to a one-site binding model enabled estimation of the dissociation constant (K d ) of 33 µM.

RK-33 Inhibits DDX3X Catalytic Activities
To assess the effect of RK-33 binding on DDX3X, increasing concentrations of the agent were added to a colorimetric assay detecting DDX3X's ability to hydrolyze ATP and generate free Pi in the presence of poly(I:C). Inhibition was clearly evident, with a half maximal inhibitory concentration (IC 50 ) of~40 µM RK-33 (Figure 2A, black); the extent of maximal inhibition of DDX3X corresponded to the low levels of activity of the catalytically inactive DDX3X K230E mutant control [6] in either the absence or presence of RK-33 (Figure 2A, red). Next, we tested whether RK-33 could inhibit DDX3X from unwinding a double-stranded RNA (dsRNA) using a helicase unwinding assay, where the fluorescent RNA products are analyzed by Next, we tested whether RK-33 could inhibit DDX3X from unwinding a double-stranded RNA (dsRNA) using a helicase unwinding assay, where the fluorescent RNA products are analyzed by quantitative polyacrylamide gel electrophoresis (PAGE). Purified recombinant DDX3X ( Figure 1A) was incubated with an RNA duplex, in which one strand bears a 5 Cy5 fluorescent label, in the absence or presence of increasing concentrations of RK-33. In the absence of RK-33, DDX3X ( Figure 2B) but not the DDX3X K230E control ( Figure 2C), was capable of unwinding the RNA duplex, as indicated by the appearance of ssRNA. As expected, DDX3X K230E remained incapable of unwinding RNA in the presence of RK-33 ( Figure 2D, quantified in Figure 2F). In contrast, DDX3X unwinding activity was progressively impaired by increasing concentrations of RK-33, with an IC 50 of 35 µM ( Figure 2E, quantified in Figure 2F) aligning well with the results from the ATPase assay (Figure 2A). Together, these findings establish that RK-33 binding can inhibit DDX3X's ability to both hydrolyze ATP and unwind RNA, consistent with RK-33 binding to the Walker A motif.
unwinding RNA in the presence of RK-33 ( Figure 2D, quantified in Figure 2F). In contrast, DDX3X unwinding activity was progressively impaired by increasing concentrations of RK-33, with an IC50 of 35 μM ( Figure 2E, quantified in Figure 2F) aligning well with the results from the ATPase assay ( Figure 2A). Together, these findings establish that RK-33 binding can inhibit DDX3X's ability to both hydrolyze ATP and unwind RNA, consistent with RK-33 binding to the Walker A motif.

RK-33 Is a Potent Broad-Spectrum Antiviral Agent
Since RK-33 has only been tested for its anticancer properties, we tested its ability to limit infection by key members of the Paramyxoviridae (RSV and hPIV-3) and Flaviviridae (WNV, DENV-2, ZIKV) families. Vero cells were infected at an MOI of 1 with RSV [34], hPIV-3, DENV-2 [30,35], ZIKV [32,35] or WNV [35], followed 2 h later by the addition of increasing concentrations of RK-33. Then, 22 h later, virus production was quantified by plaque assays and qRT-PCR analysis of the cell supernatant (DENV-2, ZIKV, WNV, hPIV-3) and cell lysates (RSV). Strikingly, RK-33 potently inhibited not only DENV-2, ZIKV and WNV ( Figure 3A-C) but also RSV and hPIV-3 replication ( Figure 3D,E), with EC50s of ≤10 μM (see Table 1 for pooled data). These results were consistent with qRT-PCR analysis indicating that RK-33 inhibited all viruses with comparable EC50 values ( Figure 4, Table 1). Table 1. Pooled half maximal effective concentration (EC50) data from replicate plaque assays and RT-qPCRs.   Table 1 for pooled data from the two assays.   Results represent the mean +/− standard deviation (SD) for duplicate wells from a single assay, representative of two (n = 2) independent experiments. See Table 1 for pooled data from the two assays.  Table 1 for pooled data from the two assays.
Importantly, the viral inhibitory effects of RK-33 were not due to cytotoxic effects, as indicated by XTT cell viability assays, which revealed that RK-33 concentrations below 30 μM had no impact on Vero cell numbers/viability (Supplementary Figure S1).

Discussion
Even though viral infections pose a significant threat to human health, antivirals to combat them are few and often lead to the development of drug resistance. A strategy to avoid drug resistance is Dose-response curves were plotted from the vRNA copies versus the logarithmic value of the concentrations of RK-33. Results represent the mean +/− SD for duplicate wells from a single assay, representative of two (n = 2) independent experiments. See Table 1 for pooled data from the two assays. Importantly, the viral inhibitory effects of RK-33 were not due to cytotoxic effects, as indicated by XTT cell viability assays, which revealed that RK-33 concentrations below 30 µM had no impact on Vero cell numbers/viability (Supplementary Figure S1).

Discussion
Even though viral infections pose a significant threat to human health, antivirals to combat them are few and often lead to the development of drug resistance. A strategy to avoid drug resistance is to develop antivirals targeting host cell co-factors critical to viral replication. DDX3X is a prime host target for multiple viruses of medical significance, being required for replication of HIV-1 [5,6] and flaviviruses such as Japanese encephalitis virus (JEV) [36] and HCV [7,8].
RK-33 is a REN, containing the imidazo[4,5-e] [1,3]diazepine ring system that is known for its anticancer properties [9,11,15,16]. RK-33 has been suggested to target DDX3X based on coimmunoprecipitation studies and work examining DDX3X's yeast orthologue Ded1p (~53% sequence identity) [11]. Here, using AUC and ITC, we establish for the first time that RK-33 binds directly to DDX3X at low µM concentrations, likely within the Walker A motif that contributes to ATP-binding. Consistently, we also confirm using an ATPase assay and fluorescence-based RNA unwinding assay that RK-33 inhibits DDX3X's ATPase and RNA unwinding activities at µM concentrations. Bol and colleagues [11] reported inhibition of helicase activity of yeast Ded1p at lower (nM) RK-33 concentrations. The basis of this difference in sensitivity is likely due to differences in the proteins themselves. The DDX3X used in this study lacked the C-terminal 82 residues that play a role in oligomerization, which is important for RNA duplex binding and DDX3X's catalytic activity [28,37]. Further to this full-length DDX3X and Ded1p are only 53% identical, have different substrate preferences, and differ quantitatively in their RNA unwinding catalytic efficiency [38], meaning that the nature of the RNA substrate differs for the two enzymes. This is also consistent with other studies showing marked differences in the efficacy of compounds against the same enzyme from a range of bacterial species [38].
REN analogues that modulate the helicase activity of target viral or host helicases have been shown to have antiviral activity against HIV-1 [21], HCV [22], JEV [22], WNV [22,23] and hepatitis B [19,20]. Consistent with the antiviral properties of REN derivatives and of agents targeting DDX3X, our virus infection studies reveal here for the first time that the DDX3X-targeting REN derivative RK-33 also has antiviral properties. Notably, RK-33 exhibited broad-spectrum antiviral activity, inhibiting the replication of RSV from the Paramyxoviridae family, and the closely related flaviviruses DENV and WNV. Further to this, we are the first to show that RK-33 can also inhibit the replication of additional members of the Paramyxoviridae and Flaviviridae families, namely hPIV-3 and ZIKV, respectively. Our findings suggest a hitherto unrecognized role for DDX3X in hPIV-3 and ZIKV replication and may further expand the list of viruses for which anti-DDX3X agents may be effective and reinforce the antiviral properties of REN derivatives.
Collectively, our findings support the idea that DDX3X-targeting agents have potential as broad-spectrum antiviral agents, and that RK-33 is one such agent, although it should be stressed that demonstration of RK-33 selectivity for DDX3X in a cellular context remains the focus of further study in this laboratory. Our data further suggest that RK-33 s ability to block DDX3X's ATPase and RNA unwinding activities is the basis of its broad-spectrum antiviral properties.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.