PIP4K2C inhibition reverses autophagic flux impairment induced by SARS-CoV-2

Abstract

In search for broad-spectrum antivirals, we discovered a small molecule inhibitor, RMC-113, that potently suppresses the replication of multiple RNA viruses including SARS-CoV-2 in human lung organoids. We demonstrated selective dual inhibition of the lipid kinases PIP4K2C and PIKfyve by RMC-113 and target engagement by its clickable analog. Advanced lipidomics revealed alteration of SARS-CoV-2-induced phosphoinositide signature by RMC-113 and linked its antiviral effect with functional PIP4K2C and PIKfyve inhibition. We discovered PIP4K2C's roles in SARS-CoV-2 entry, RNA replication, and assembly/egress, validating it as a druggable antiviral target. Integrating proteomics, single-cell transcriptomics, and functional assays revealed that PIP4K2C binds SARS-CoV-2 nonstructural protein 6 and regulates virus-induced impairment of autophagic flux. Reversing this autophagic flux impairment is a mechanism of antiviral action of RMC-113. These findings reveal virus-induced autophagy regulation via PIP4K2C, an understudied kinase, and propose dual inhibition of PIP4K2C and PIKfyve as a candidate strategy to combat emerging viruses.

Fig. 1:. RMC-113 inhibits SARS-CoV-2 infection in vitro and in human ALOs with a high genetic barrier to resistance.

a, Chemical structure of RMC-113. b, Rescue assay for virus-induced cell lethality. RMC-113 (10 μM) was incubated with Vero E6-eGFP cells for 20 hours followed by SARS-CoV-2 infection. eGFP signal measured at 96 hpi indicates cell survival. c and d, Fluorescence images (c) and corresponding graph (d) of Vero-eGFP cells rescued from SARS-CoV-2-induced lethality by RMC-113 (Belgium-GHB-03021 strain, MOI = 0.05). e and j, Dose response to RMC-113 of SARS-CoV-2 infection [black, USA-WA1/2020 strain, MOI = 0.05 (e), 1 (j)] and cell viability (blue) in Calu-3 cells (e) or ALO-derived monolayer supernatants (j) via plaque and alamarBlue assays at 24 (e) or 48 (j) hpi, respectively. f and i, Schematics of the experiments shown in g and j, k, l, respectively. g, Vero E6-TMPRSS2 cells were infected with rSARS-CoV-2-nLuc virus (MOI = 0.05) and passaged daily under RMC-113 (0.1–0.3 μM) or DMSO over nine passages. Viral titers were measured by plaque assays. h, Dose response to RMC-113 of rSARS-CoV-2-nLuc virus harvested after nine passages under RMC-113 or DMSO via luciferase assays. k, Dose response to RMC-113 of SARS-CoV-2 (MOI = 1) nucleocapsid copy number in ALO lysates measured by RT-qPCR assays at 48 hpi. l, Confocal IF microscopy images of F-actin (violet), nucleocapsid (green), and DAPI (blue) in naive and SARS-CoV-2–infected ALOs, pretreated with DMSO or RMC-113 (5 μM) at 24 hpi. Representative merged images at x40 magnification are shown. Scale bars: 50 μm. Data is representative (c, d, g, j, l), or a combination (e, h, k) of 2 independent experiments, each with 2–3 biological replicates. Data in e, h, j, and k are relative to DMSO. Means ± SD are shown.

Fig. 2:. RMC-113 is a selective kinase inhibitor that stably binds PIKfyve and PIP4K2C.

a, Kinase abundance ratio between RMC-113 (0.1, 1.0, and 10 μM)- and DMSO-treated SUM159 cell lysates measured following affinity purification via multiplexed inhibitor beads kinome profiling and analyzed by mass spectrometry (MIB/MS). Shown are the Log2 fold change values of a subset of the screen panel (see Supplementary Fig. 2a). Data means ± SD of 2 replicates are shown. b and c, In vitro dose response to RMC-113 of PIKfyve activity (b) and PIP4K2C binding (c). d, Biochemical parameters for RMC-113 and SRN2–002. ND=Not determined e, Chemical structure of SRN2–002. f, Dose response to SRN2–002 of SARS-CoV-2 (MOI = 0.05) infection (black) and cell viability (blue) in Calu-3 cells via plaque and alamarBlue assays at 24 hpi, respectively. g, PIKfyve and PIP4K2C expression in lysates of SARS-CoV-2-infected A549-ACE2 cells following incubation with SRN2–002 individually or in combination with RMC-113, UV irradiation, and pull down by streptavidin beads measured via Western blotting at 24 hpi. Shown are representative membranes of 2 experiments. Lanes: 1: SRN2–002, no UV; 2: SRN2–002, no pull-down enrichment; 3: SRN2–002 (1 μM); 4: SRN2–002 (5 μM); 5: competitive inhibition of SRN2–002 (1 μM) with RMC-113 (10 μM); 6: competitive inhibition of SRN2–002 (5 μM) with RMC-113 (50 μM). h, Putative binding mode of RMC-113 into the ATP-binding pockets of PIKfyve and PIP4K2C based on microsecond timescale MD simulations. A representative snapshot with key interactions are shown. Binding pocket residues with >10% interaction frequencies to RMC-113 are shown. Residues with positive charge are highlighted with blue molecular surface; negative charge, red; hydrophobic, green; polar, cyan. RMC-113 is shown in yellow (carbon) stick model. H-bonds are illustrated with purple dashed lines, π–π interactions with green. See Supplemental Fig. 2f for more detailed simulation interactions. i and k, Summary of the observed main protein-ligand interactions in the MD simulations of PIKfyve (i) and PIP4K2C (k). Interactions with >10% frequency are displayed. j and l, Aggregate of protein-ligand interactions (residues with >10%) in the simulations of PIKfyve (j) and PIP4K2C (l). Data in (i and j) consist of 72 μs, and (k and l) consist of 20 μs, both analyzed each 1 ns.

Fig. 3:. PIKfyve and PIP4K2C are essential for SARS-CoV-2 infection and mediate the antiviral effect of RMC-113.

a, f, j Schematics of the experiments in b, c (a); g, h (f); and k, l, m (j). b, Confirmation of siRNA-mediated knockdown by RT-qPCR in Calu-3 cells. Shown are gene expression levels normalized to GAPDH relative to respective gene levels in siNT control at 48 hours post-transfection. c, Viral titers (PFU/ml) and cell viability (blue) in Calu-3 cells transfected with the indicated siRNAs at 24 hpi with SARS-CoV-2 (MOI = 0.05) via plaque and alamarBlue assays, respectively. d, Structures of PIKfyve and/or PIP4K2C inhibitors. e, Dose response of rSARS-CoV-2-nLuc (black, USA-WA1/2020 strain, MOI = 0.05) infection and cell viability in Calu-3 cells via luciferase and alamarBlue assays at 24 hpi, respectively. g and h, Rescue of rVSV-SARSCoV-2-S infection under RMC-113 treatment upon ectopic expression of WT PIKfyve (g) and PIP4K2C (h), their kinase-dead mutants or empty control plasmids measured by luciferase assays at 24 hpi in Vero cells. i, Phosphoinositides and associated kinases. k and l, Fold change (FC) of peak area ratio of the indicated phosphoinositides in SARS-CoV-2-infected (MOI = 0.5) vs. uninfected (k) and RMC-113- vs. DMSO-treated infected Calu-3 cells (l), as measured via PRMC-MS. m, Product-to-substrate ratios in RMC-113- vs. DMSO-treated cells infected with SARS-CoV-2 (k, l). Data is relative to siNT (b, c) or DMSO (e, g, h). Data are a combination (b, c, e, g, and h) or representative of 2 independent experiments with 2–4 replicates each. Means ± SD are shown (b, c, g, h). k, l, m represent one of two independent experiments. See an associated experiment in Supplemental Fig. 5e-g and Table S2. *P ≤ 0.05, **P < 0.01, ***P <0.001, ****P <0.0001 by 1-way ANOVA followed by Dunnett’s (b, c) or Tukey’s multiple-comparison test (g, h).

Fig. 4:. PIP4K2C is required for temporally distinct stages of the SARS-CoV-2 life cycle, whereas PIKfyve is required for viral entry only.

a, Schematic of the time-of-addition experiments shown in b, c. b and c, Calu-3 cells were infected with WT SARS-CoV-2 (MOI = 1). At the indicated times, RMC-113 (10 μM) (b), apilimod (10 μM) (c), or DMSO were added. Supernatants were collected at 10 hpi, and viral titers measured by plaque assays. d and f, Schematics of the experiments shown in e, h (d) and g, i (f). e, Dose response to RMC-113 of WT SARS-CoV-2 entry (MOI = 1) in Calu-3 cell lysates measured by RT-qPCR assays at 2 hpi. g, Dose response to RMC-113 and ensitrelvir of viral RNA replication measured by luciferase assay in Vero E6 cells 24 hours post-transfection of in vitro transcribed nano-luciferase reporter-based SARS-CoV-2 subgenomic non-infectious replicon. h, WT SARS-CoV-2 (MOI = 1) entry measured in Calu-3 cells depleted of the indicated kinases using corresponding siRNAs by RT-qPCR at 2 hpi. i, Viral RNA replication and cell viability (blue) measured by luciferase and alamarBlue assays, respectively, in Vero E6 cells depleted of the indicated kinases, 24 hours post-transfection of in vitro transcribed nano-luciferase reporter-based SARS-CoV-2 subgenomic non-infectious replicon. Data are a combination (b, c, e, g, h) or representative (i) of two independent experiments with 2–4 replicates each. Means ± SD are shown. Data is relative to DMSO (b, c, e, g) or siNT (h, i). **P < 0.01, ****P <0.0001 by 1-way ANOVA followed by Turkey’s (b,c) or Dunnett’s (h, i) multiple-comparison test. ns= non-significant.

Fig. 5:. RMC-113 reverses SARS-CoV-2-induced impairment of autophagic flux.

a, Schematic of the experiment shown in c, d. b, Autophagy flux: a shift from autophagosomes (yellow, RFP+/GFP+) to autolysosomes (red, RFP+/GFP-). c, Representative confocal microscopic images of A549-ACE2 cells transfected with GFP-RFP-LC3 tandem plasmid and infected with SARS-CoV-2 (MOI = 0.5), treated with DMSO, RMC-113 (5 μM), apilimod (5 μM) or CQ (1 μM) for 24 hours and stained for nucleocapsid (violet). Representative merged images at x40 magnification are shown Scale bars: 10 μm. Zoomed-in images show autophagosomes (yellow) and autolysosomes (red). d, Autolysosome-to-autophagosome ratio (autophagy flux) in 70 single cells (c). e, Expression levels of autophagy markers in uninfected and SARS-CoV-2-infected (MOI= 1) A549-ACE2 cell lysates treated with RMC-113 (5μM) and/or CQ (1μM) or DMSO at 24 hpi via Western blotting. Numbers represent expression signals relative to DMSO averaged from four membranes. f, Schematic of the viscRNA-seq analysis. g, Dot plot depicting marker genes used to annotate the indicated cell populations. Color indicates expression level in cpm; dot size indicates the fraction of cells expressing the marker. h and i, UMAP embedding of the scRNA-seq dataset indicating distinct cell types (h) or SARS-CoV-2 transcripts (i). j and k, Pathway enrichment in AT2 cells infected vs. uninfected (j) and RMC-113 vs. DMSO treated (k) ALOs. l, Heatmap showing the log2 fold change in the expression of autophagy-related genes between infected vs. uninfected (DMSO), RMC-113 vs. DMSO (uninfected); and RMC-113 vs. DMSO (infected) in AT2-like cells at 24 hpi. See, Supplementary Fig. 8b. Black rectangles highlight transcripts with significant changes measured by Wilcoxon test. m-o, Box plots showing the expression level of the indicated genes in individual AT2 cells at 24 hpi. Horizontal lines indicate the first, second (median) and third quartiles; whiskers extend to ±1.5× the interquartile range. P values by two-sided Wilcoxon test with Benjamini-Hochberg correction are shown. Data is a combination (d, j-o,) or representative (c, e) of 3 independent experiments. Means ± SD are shown (d). ***P < 0.001, ****P <0.0001 by 1-way ANOVA followed by Dunnett’s multiple comparison test (d).

Fig. 6:. PIP4K2C binds SARS-CoV-2 NSP6 and mediates virus-induced impairment of autophagic flux.

a, Schematic of the experiment shown in b. b, Representative confocal microscopic images of A549-ACE2 cells transfected with the indicated siRNAs and GFP-RFP-LC3 tandem plasmid, and infected with SARS-CoV-2 (MOI = 0.5) for 24 hours and stained for nucleocapsid (violet). Representative merged images at x40 magnification are shown. Scale bars: 10 μm. Zoomed-in images show autophagosomes (yellow) and autolysosomes (red). c, Autolysosome-to-autophagosomes ratio (autophagy flux) in 27 single cells (b). d, Expression levels of p62, LC3I and LC3II following transfection of the indicated siRNAs in uninfected A549-ACE2 cell lysates (left panel) and SARS-CoV-2-infected cells at 24 hpi (MOI = 0.5) (right panel). Numbers represent the expression signal relative to DMSO averaged from 3 membranes. e, PIP4K2C interactions with 15 SARS-CoV-2 nonstructural proteins and empty plasmid as measured via protein-fragment complementation assay (PCAs) in HEK 293T cells. Dots depict the mean normalized luminescence ratio (NLR) values generated from 2 independent experiments with 3 replicates each. The dotted line depicts the cutoff (NLR>10) used to define PIP4K2C-interacting proteins (depicted in green), representing greater than two SDs above the mean NLR of a non-interacting reference set. NLR, normalized luminescence ratio. f, Proposed model for the roles of PIP4K2C and PIKfyve in SARS-CoV-2 infection and the mechanism of antiviral action of RMC-113. Data is a representative (b, d, e) of 2 independent experiments. Data are relative to siNT (c, d). Means ± SD are shown (c). **P < 0.001, ****P <0.0001 by 1-way ANOVA followed by Dunnett’s multiple comparison test (c).