Fig 1.
UCK inhibition limits cellular response to molnupiravir and its cell-active metabolite NHC.
A. Schematic representation of cellular roles of UCK in pyrimidine salvage pathway. UCKs govern the first- and rate-limiting step of pyrimidine salvage pathway. It catalyzes the mono-phosphorylation of uridine and cytidine into uridine monophosphate (UMP) and cytidine monophosphate (CMP), respectively, which upon subsequent phosphorylation by CMPK and NDPK are converted to the tri-phosphorylated nucleotides. Cytidine triphosphate (CTP) could be further converted from uridine triphosphate (UTP) by CTPS. In the cell viability-based screening, UCK inhibitor CPU, UCK/CTPS dual inhibitor DAU, and finally CTPS inhibitor CPC were employed, as highlighted in the schematic representation. UCK, uridine cytidine kinase 1/2; CTPS, CTP synthase 1/2; CMPK, cytidine monophosphate kinase; CDA, cytidine deaminase; NDPK, nucleoside-diphosphate kinase; CPU, cyclopentenyl uracil; DAU, 3-Deazauridine; CPC, cyclopentenyl cytosine. B. Focused nucleoside analogue drug library screening identified UCK as a limiting factor for the cellular response of molnupiravir. AML HL-60 and THP1 cells were treated with nucleoside analogues drugs in the presence or absence of 100 µM CPU, 12 nM CPC, and 1 µM DAU for 96 hours before cell viabilities were determined using a resazurin reduction assay. Relative viability percentages were determined by normalizing background-subtracted fluorescence signals at 584 nm to DMSO controls. In both HL-60 and THP1 cells, inhibition of UCK by CPU and DAU, but not CTPS inhibition by CPC, retarded the cytotoxicity of molnupiravir and azacitidine. C-D. Inhibition of UCK consistently antagonized NHC, metabolite of molnupiravir, in multiple cell lines. In C, relative cell viability curves in multiple leukemia cell lines treated with NHC, in the presence or absence of CPU, CPC, or DAU; in D, the relative cell viability data as presented in C were further used to determine synergy scores (δ), calculated using SynergyFinder. A δ > 10 reflects synergy, and a δ < -10 reflects antagonism. E-F. Cellular UCK levels positively correlated with CPU-induced NHC antagonism. In E, Western blot analysis of UCK expression levels in multiple cell lines, where lysates containing equal amount of protein were separated by SDS-PAGE, probed with UCK1 and UCK2 specific antibodies, and then subject to densitometry analysis. UCK1 and UCK2 expressions in the cell lines were normalized to the levels in HL-60 cells, and were presented as relative levels. In F, cell lines in E were subjected to resazurin reduction assay upon 96-hour treatment with NHC or DMSO control, in the presence or absence of CPU. Synergy scores were subsequently determined based on the cell viability data, as described in D, and further demonstrated positive correlations with both UCK1 and UCK2 expression levels. r and p values of Spearman correlation analysis are shown.
Fig 2.
UCK1 and UCK2 preferentially interact with molnupiravir metabolite NHC in their ATP-bound state, similar to other known UCK substrates.
A-B. Melting profiles of recombinant UCK1 in the presence of potential ligands, determined using differential scanning fluorometry (DSF) assays. Recombinant UCK1 (5 µM) was mixed with 5 mM molnupiravir (MOV), its phosphorylatable metabolite NHC, and known substrates azacitidine (AzaC) or cytidine (C), in the presence or absence of 5 mM ATP. Subsequently, protein thermal stability was examined using DSF. Mean fluorescence signals (solid line) ± SEM (dashed line) of a representative experiment performed in quadruplicate are shown in (A, left panel); melting temperatures (Tm) were then determined as the minima of negative derivative of the melting curve (A, right panel). In B, mean change of Tm (ΔTm) compared to protein only group of n = 4 independent experiments are shown. Similar to the known substrates of UCK1 cytidine and azacitidine, the phosphorylatable metabolite of molnupiravir NHC, but not molnupiravir, effectively engaged recombinant UCK1 upon the addition of ATP, evidenced by significant Tm changes compared to UCK1 bound to ATP only. Ordinary one-way ANOVA tests (Bonferroni’s multiple comparisons tests) were performed across treatment groups [ΔTm (protein only+ATP) vs. ΔTm (protein only+C), p < 0.0001, t = 12.73, DF = 26; ΔTm (protein only+ATP) vs. ΔTm (protein only+MOV), p = 0.3244, t = 1.815, DF = 26; ΔTm (protein only+ATP) vs. ΔTm (protein only+NHC), p < 0.0001, t = 7.641, DF = 26; ΔTm (protein only+ATP) vs. ΔTm (protein only+AzaC), p < 0.0001, t = 8.469, DF = 26], where asterisks signify statistical significance (****p ≤ 0.0001). C-D. Melting profiles of recombinant UCK2 in the presence of potential ligands, determined using DSF assays as described in A-B. Recombinant UCK2 (3.5 µM) was incubated with 5mM compounds alone or in the presence of 5 mM ATP, before protein thermal stabilities were determined using DSF. Similar to UCK1, NHC effectively engaged recombinant UCK2, but only in the presence of ATP. Mean fluorescence signals (solid line) ± SEM (dashed line) of a representative experiment performed in quadruplicate are shown (C, left panel); melting temperatures (Tm) were then determined as the minima of negative derivative of the melting curve (C, right panel). In D, mean change of Tm (ΔTm) compared to protein only group of n = 2 independent experiments performed in triplicate are shown. Ordinary one-way ANOVA tests (Bonferroni’s multiple comparisons tests) were performed across treatment groups [ΔTm (protein only+ATP) vs. ΔTm (protein only+C), p = 0.048, t = 3.00, DF = 6; ΔTm (protein only+ATP) vs. ΔTm (protein only+NHC), p = 0.0007, t = 7.343, DF = 6], where asterisks signify statistical significance (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001). E-F. Change of Tm of ATP-bound UCK1 (E) and UCK2 (F), in the presence of increasing concentrations of NHC or C, determined using DSF. Recombinant UCK1 and UCK2 were incubated with 5 mM ATP, alone or together with increasing concentrations of NHC or C, before protein Tm were determined using DSF as described in the previous sections. Change of Tm (ΔTm) was calculated by normalizing to protein + ATP only group. Mean ΔTm ± SEM of n = 4 independent experiments performed in quadruplicate are shown.
Fig 3.
UCK2 is the preferred kinase in phosphorylating molnupiravir metabolite NHC.
A-B. Saturation curves of UCK1 (A) or UCK2 (B) -mediated phosphorylation of cytidine, using ADP-Glo-coupled activity assay. C-D. Saturation curves of UCK1 (C) or UCK2 (D) -mediated phosphorylation of NHC, using ADP-Glo-coupled activity assay. In A-D, mean initial rates ± SEM of n = 2 independent experiments performed in triplicates are shown. E. Kinetic parameters of UCK1 and UCK2-mediated phosphorylation of NHC, in comparison to cytidine, a canonical substrate of UCK1 and UCK2. Kinetic parameters ± SEM were determined via Michaelis-Menten equation (GraphPad Prism) based on the saturation curve shown in A-D.
Fig 4.
X-ray crystal structure of human UCK1-NHC-AMPPNP.
A. Surface representation of the UCK1-NHC-AMPPNP tetramer where individual monomers are colored light pink, light cyan, light green and wheat. NHC and AMPPNP are depicted as spheres colored yellow and magenta, respectively. B. Cartoon representation of the UCK1-NHC-AMPPNP monomer. Alpha-helices are coloured green, beta-strands are coloured blue, 310-helices are colored orange and loop regions are coloured light grey. The catalytic magnesium ion is shown as a gray sphere. NHC and AMPPNP (labeled as ATP) are shown as sticks colored yellow and magenta, respectively. C. Electron density for the NHC binding site. D. Electron density for the ATP binding site. The 2Fo-Fc maps are contoured at 1.0 σ (blue) and the Fo-Fc maps are contoured at +3.5 σ (green) and -3.5 σ (red). Water molecules are shown as red spheres. Hydrogen bonds are depicted as dashed lines. Figure produced with PyMOL (version 3.0.4, Schrödinger).
Fig 5.
UCK2 expression dictates the anti-SARS-CoV-2 efficacy of NHC.
A. Schematic representation of the high content imaging-based infectivity assay setup. Lung alveolar basal epithelial A549/ACE2 cells, engineered to overexpress SARS-CoV receptor ACE2, were infected with SARS-CoV-2 at a MOI of 0.07 overnight, before viral infection was assayed via immunofluorescence staining against the viral nucleocapsid protein (NC) and cell nucleus (DAPI). Infectivity was determined as percentage of DAPI/NC double-positive cells out of DAPI single-positive cells, which was further normalized to infectivity level in virus-infected wildtype group and presented as relative infectivity %. B. Western blot analysis of A549/ACE2 cell lysates 72 hours post-transfection with UCK2-targeting (siUCK2) or control (siNT) siRNAs. Left panel: Representative Western blot image. Right panel: densitometry analysis of n = 4 independent experiments, where protein signals were normalized to β-tubulin and relative to that of siNT-treated cells. C-F. siRNA-mediated downregulation of UCK2 in A549/ACE2 cells antagonized the anti-SARS-CoV-2 efficacy of NHC (C-D), but not another antiviral nucleoside analogue remdesivir (Rem, E-F). At 72 hours post-transfection with UCK2-targeting (siUCK2) or control (siNT) siRNAs, A549/ACE2 cells were re-seeded and infected with SARS-CoV-2 overnight in the presence of antivirals or the drug diluent control DMSO, before viral infectivity was determined using the high content imaging-based infectivity assay as depicted in A. In C and E, representative immunofluorescence staining images of infected A549/ACE2 cells, scale bars represent 100 μm; in D and F, mean relative infectivity % ± SEM of n = 3 independent experiments performed in duplicate are shown. G. UCK2 knockdown antagonized NHC through limiting the intracellular levels of its tri-phosphorylated active metabolite NHC-TP. A549/ACE2 cells transfected with UCK2-targeting (siUCK2) or control (siNT) siRNAs were treated with 100 µM NHC for 6 hours before cell lysates were harvested for the measurement of intracellular NHC-TP levels. The NHC-TP levels were normalized to total NTP (NHC-TP/total NTP) or total purine-TP (NHC-TP/total NTP G + ATP) levels. Mean NHC-TP/total NTP and NHC-TP/total NTP G + ATP, relative to siNT group, ± SEM of n = 2 independent experiments are shown, together with individual experiment values. Student’s t-tests were performed across treatment groups – for relative NHC-TP/total NTP [siNT vs. siUCK2, p = 0.03, t = 21.5, df = 1]; for relative NHC-TP/total G + ATP [siNT vs. siUCK2, p = 0.03, t = 21.6, df = 1], where asterisk signifies statistical significance (*p ≤ 0.05).
Fig 6.
UCK inhibition antagonizes the anti-SARS-CoV-2 efficacy of NHC.
A-B. CPU-mediated UCK inhibition in A549/ACE2 cells dose-dependently antagonized the anti-SARS-CoV-2 efficacy of NHC. A549/ACE2 cells were infected with SARS-CoV-2 overnight in the presence of dose-response matrix composed of the antivirals/DMSO and CPU/DMSO, before viral infectivity was determined using the high content imaging-based infectivity assay as depicted in Fig 5 A. In A, left panel, mean relative infectivity % ± SEM of n = 4 independent experiments performed in duplicate are shown; right panel, mean antiviral EC50 of NHC with increasing concentration of CPU are further shown together with color-coded individual experiment values. Paired t-tests were performed across treatment groups – 0 µM CPU vs. 10 µM CPU, p = 0.10; 0 µM CPU vs. 22.5 µM CPU, p = 0.08; 0 µM CPU vs. 50.6 µM CPU, p = 0.05; 0 µM CPU vs. 114 µM CPU, p = 0.03; 0 µM CPU vs. 256 µM CPU, p = 0.003; where asterisk signifies statistical significance (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001). In B, representative immunofluorescence staining images of infected A549/ACE2 cells, scale bars represent 100 μm. C. UCK2 inhibition by CPU significantly hampered the accumulation of intracellular NHC-TP. A549/ACE2 cells were treated with DMSO or 256 µM CPU for 1 hour before cells were further treated with 100 µM NHC for another 5 hours. Cell lysates were subsequently harvested, and the intracellular nucleotide levels were determined. The NHC-TP levels were normalized to total NTP levels and then to DMSO control samples. Mean of the resulting relative NHC-TP levels ± SEM of n = 3 independent experiments are shown. Paired t-test was performed across treatment groups [Relative NHC-TP (0 µM CPU) vs. Relative NHC-TP (256 µM CPU), p = 0.0006, t = 40.51, df = 2], where asterisks signify statistical significance (***p ≤ 0.001). D-E. CPU-mediated UCK inhibition in A549/ACE2 cells did not antagonize remdesivir. Experiment was carried out as described in A-B. In D, mean relative infectivity % ± SEM of n = 3 independent experiments performed in duplicate are shown. In E, representative immunofluorescence staining images of infected A549/ACE2 cells, scale bars represent 100 μm.
Fig 7.
UCK is critical for an optimal selectivity of NHC.
A-B. Depletion of UCK2 significantly reduced the selectivity (EC50 (24h)/CC50 (96h)) of NHC by 50%. A549/ACE2 cells transfected with UCK2-targeting (siUCK2) or control (siNT) siRNAs were treated with NHC for 4 days before cell viability was determined using resazurin reduction assay. Alternatively, cells were infected with SARS-CoV-2 overnight in the presence of NHC before viral infectivity was determined using high content imaging-based infectivity assay. Mean viability/infectivity relative to DMSO group ± SEM of n = 2 independent experiments are shown in A, and were further curve fitted using a nonlinear curve fitting model with variable slope (GraphPad Prism) to generate the antiviral EC50 and growth inhibitory CC50 values of NHC. EC50 (24h)/CC50 (96h) was then used to indicate drug selectivity. In B, mean EC50 (24h)/CC50 (96h) ± SEM, and mean fold change of EC50 (24h)/CC50 (96h) ± SEM are shown. Student’s t-tests were performed across treatment groups [EC50 (24h)/CC50 (96h) (siNT) vs. EC50 (24h)/CC50 (96h) (siUCK2), p = 0.041; fold change (siNT) vs. fold change (siUCK2), p = 0.041)], where asterisks signify statistical significance (*p ≤ 0.05). C-D. Inhibition of UCK activity by CPU dose-dependently reduced the selectivity (EC50 (24h)/CC50 (96h)) of NHC, by up to over 50%. A549/ACE2 cells were treated with a dose-response matrix of CPU and NHC for 4 days before cell viability was determined using resazurin reduction assay, or alternatively, cells were infected with SARS-CoV-2 overnight before viral infectivity was determined. Mean viability/infectivity relative to DMSO group ± SEM of n = 2-4 independent experiments are shown in C, and were further curve fitted using a nonlinear curve fitting model with variable slope (GraphPad Prism) to generate the antiviral EC50 and growth inhibitory CC50 values of NHC. EC50 (24h)/CC50 (96h) was then used to indicate drug selectivity. In D, EC50 (24h)/CC50 (96h) ± SEM, and mean fold change ± SEM are shown. Student’s t-tests were performed across treatment groups [EC50 (24h)/CC50 (96h) (0 µM CPU) vs. EC50 (24h)/CC50 (96h) (10 µM CPU), p = 0.368; EC50 (24h)/CC50 (96h) (0 µM CPU) vs. EC50 (24h)/CC50 (96h) (22.5 µM CPU), p = 0.259; EC50 (24h)/CC50 (96h) (0 µM CPU) vs. EC50 (24h)/CC50 (96h) (50.6 µM CPU), p = 0.195; EC50 (24h)/CC50 (96h) (0 µM CPU) vs. EC50 (24h)/CC50 (96h) (256 µM CPU), p = 0.042; fold change (0 µM CPU) vs. fold change (10 µM CPU), p = 1.04E-7; fold change (0 µM CPU) vs. fold change (22.5 µM CPU), p = 0.636; fold change (0 µM CPU) vs. fold change (50.6 µM CPU), p = 0.107; fold change (0 µM CPU) vs. fold change (256 µM CPU), p = 8.14E-5], where asterisks signify statistical significance (*p ≤ 0.05, ****p ≤ 0.0001).