SARS-CoV-2-mediated dysregulation of metabolism and autophagy uncovers host-targeting antivirals

Abstract

Viruses manipulate cellular metabolism and macromolecule recycling processes like autophagy. Dysregulated metabolism might lead to excessive inflammatory and autoimmune responses as observed in severe and long COVID-19 patients. Here we show that SARS-CoV-2 modulates cellular metabolism and reduces autophagy. Accordingly, compound-driven induction of autophagy limits SARS-CoV-2 propagation. In detail, SARS-CoV-2-infected cells show accumulation of key metabolites, activation of autophagy inhibitors (AKT1, SKP2) and reduction of proteins responsible for autophagy initiation (AMPK, TSC2, ULK1), membrane nucleation, and phagophore formation (BECN1, VPS34, ATG14), as well as autophagosome-lysosome fusion (BECN1, ATG14 oligomers). Consequently, phagophore-incorporated autophagy markers LC3B-II and P62 accumulate, which we confirm in a hamster model and lung samples of COVID-19 patients. Single-nucleus and single-cell sequencing of patient-derived lung and mucosal samples show differential transcriptional regulation of autophagy and immune genes depending on cell type, disease duration, and SARS-CoV-2 replication levels. Targeting of autophagic pathways by exogenous administration of the polyamines spermidine and spermine, the selective AKT1 inhibitor MK-2206, and the BECN1-stabilizing anthelmintic drug niclosamide inhibit SARS-CoV-2 propagation in vitro with IC₅₀ values of 136.7, 7.67, 0.11, and 0.13 μM, respectively. Autophagy-inducing compounds reduce SARS-CoV-2 propagation in primary human lung cells and intestinal organoids emphasizing their potential as treatment options against COVID-19.

Fig. 1. SARS-CoV-2 causes accumulation of key metabolites in infected cells.

a, b Analysis and regulation of significantly altered pathways of mock- and SARS-CoV-2-infected (24 h p.i., MOI = 0.1) VeroFM cells (a) or Calu-3 cells (b). The y-axis shows the (median) log2 fold change (FC) of all significantly altered metabolites of the indicated pathway while the –log10 corrected p-value (false discovery rate (FDR)) is shown on the x-axis. The size of the circles illustrates the number of significantly changed metabolites in relation to all metabolites of a specific pathway. c Analysis of the autophagic pathway and the involved metabolites: ‘amino acids’ and ‘GSH metabolism’ (orange), ‘nucleotides’ (blue), ‘glycolysis’ (green), ‘polyamine metabolism’ (red) and ‘O-GalNAcylation’ (purple) in mock- and SARS-CoV-2-infected (24 h p.i.) VeroFM and Calu-3 cells. For a–c Error bars represent SEM. n = 4 biological samples per group of one experiment. All p-values were determined by a two-way ANOVA and Tukey´s post hoc test. FDRs were adjusted using the Benjamini-Hochberg method. Abbreviations: 1,3-BPG, 1,3-bisphosphoglyceric acid; 3-PGA, 3-phosphoglyceric acid; ADP, adenosine diphosphate; AKT1, RAC-alpha serine/threonine-protein kinase; ala, alanine; AMP, adenosine monophosphate; AMPK, AMP-activated protein kinase; arg, arginine; asn, asparagine; asp, asparagine; ATP, adenosine triphosphate; aut, autophagosome; BECN1, beclin-1; bio., biosynthesis; CTP, cytidine triphosphate; CoA, coenzyme A; cys, cysteine; cysgly, cysteinylglycine; cyst, cystathionine; dCTP, deoxycytidine triphosphate; dicarbox., dicarboxylate; elF5AH, eukaryotic translation initiation factor 5A hypusinated; EP300, histone acetyltransferase p300; F1P, fructose 1-phosphate; F6P, fructose 6-phosphate; F-1,6-BP, fructose 1,6-bisphosphate; FKBP51, 51 kDa FK506-binding protein; G6-P, glucose 6-phosphate; gln, glutamine; glu, glutamic acid; gly, glycine; glyox, glyoxylate; GSH, glutathione (reduced); GTP, guanosine triphosphate; hcys, homocysteine; his, histidine; ile, isoleucine; lac, lactic acid; leu, leucine; lys, lysine; mal, malic acid; met, methionine; met., metabolism; modific., modification; mTORC1, mechanistic target of rapamycin complex 1; NAcput, N-acetylputrescine; NAcspd, N-acetylspermidine; orn, ornithine; PEP, phosphoenolpyruvic acid; phe, phenylalanine; PHLPP, PH domain leucine-rich repeat-containing protein phosphatase; PPP, pentose phosphate pathway; pro, proline; prot., protein; put, putrescine; pyr, pyruvic acid; resid., residue; ROS, reactive oxygen species; S7P, sedoheptulose-7-phosphate; SAH, S-adenosylhomocysteine; SAM, S-adenosylmethionine; SAT1, diamine acetyltransferase 1; SKP2, S-phase kinase-associated protein 2; ser, serine; spd, spermidine; spm, spermine; tau, taurine; TFEB, transcription factor EB; thr, threonine; trp, tryptophan; tyr, tyrosine; UDP-GalNAc, UDP-N-acetylgalactosamine; UDP, uridine diphosphate; UTP, uridine triphosphate; val, valine.

Fig. 2. SARS-CoV-2 limits autophagy signaling and blocks autophagic flux.

a Protein levels and phosphorylation status of selected autophagy-relevant proteins in SARS-CoV-2-infected VeroFM cells at 8 h, 24 h, or 48 h post infection (h p.i.) were analyzed by Western blotting. For analysis of ATG14 oligomers (virtual blot, bottom panel, left) proteins were cross-linked 2 h prior to cell harvest and analyzed by Wes (ProteinSimple) capillary electrophoresis 8-48 h p.i. P-values were determined by one-way ANOVA, Bonferroni post hoc test. b Fluorescence microscopy of transfected and SARS-CoV-2-infected VeroFM cells expressing pH-sensitive tandem fluorescent-tagged LC3B-mRFP/EGFP showed that low pH autophagolysosomes (AL, red) were reduced compared to autophagosomes (AP, green + red = yellow) in virus-infected cells. Microscopic read-out was done by a scientist blind to the experimental conditions. For mock (n = 44 cells) and SARS-CoV-2-infected (n = 46 cells) VeroFM cells were analyzed. P-values were determined by two-way ANOVA, Tukey´s post hoc test, mean with SEM. Scale bar = 10 µM. c Accumulation of autophagy marker P62 in SARS-CoV-2-infected VeroFM cells. d Elevated LC3B-II levels in SARS-CoV-2-infected and bafilomycin A1 (BafA1)-pretreated VeroFM cells indicate virus-induced autophagic flux inhibition at 8 and 24 h p.i. In all panels, error bars denote SEM derived from n = 3 biologically independent samples derived from one experiment. P-values were determined by two-way ANOVA, Sidak post hoc test. p ≤ 0.05 (*), p ≤ 0.01 (**), p ≤ 0.001 (***), p ≤ 0.0001 (****), p > 0.05 (not significant, ns). Abbreviations: AKT1, RAC-alpha serine/threonine-protein kinase; AMPK, AMP-activated protein kinase; ATG14, autophagy-related 14; ATG16L1, autophagy-related 16 like 1; BafA1, bafilomycin A1; BECN1, Beclin-1 protein; CDK2, cyclin-dependent kinase 2; EGFP, enhanced green fluorescent protein; LC3B, microtubule-associated protein 1 A/1B light chain 3B; LXRXX(pS/pT), phospho-AMPK substrate motif; mRFP, monomeric red fluorescent protein; mTORC1/2, mechanistic target of rapamycin complex 1/2; PRAS40, proline-rich AKT1 substrate 1; Rheb, Ras homolog enriched in brain; SKP2, S-phase kinase-associated protein 2; TSC1/2, tuberous sclerosis 1/2; ULK1, Unc-51-like kinase 1; VPS34, phosphatidylinositol 3-kinase catalytic subunit type 3.

Fig. 3. SARS-CoV-2 replication-dependent and cell-specific regulation of autophagy in vivo.

a Enhanced relative LC3B-II/HSC70 and P62 protein levels in lung samples of SARS-CoV-2-infected compared to mock-infected hamsters (dotted line). Syrian hamsters (Mesocricetus auratus, n = 3 per group, one experiment) were infected or mock-infected (heat-inactivated virus) with 1x10e5 plaque-forming units SARS-CoV-2 by intranasal instillation. Lung samples were taken at days 2, 3, 5, 14 post infection, and protein lysates were analyzed by Western blotting. P-values were determined by two-way ANOVA, Tukey´s post hoc test, mean with SD. b Immunohistochemistry (IHC). Increased amount of P62 and LC3B-positive cells in formalin-fixed and antibody-stained lung samples from deceased COVID-19 (n = 6), pneumonia (n = 3), and control patients (n = 3). For each sample, 4 randomly chosen microscopic fields (40x) were independently rated by two persons. P-values were determined by one-way ANOVA, Dunnett´s, mean with SEM. Scale bar = 100 µM. c Dot plots depicting scaled average expression of autophagy and immune genes for selected cell types from postmortem lung patient samples separated according to disease duration representing patients deceased within 14 days (early, n = 3) and after 14 days (late, n = 4). d Expression profile of distinct cell types from the olfactory mucosa of COVID-19 patients (n = 8) separated according to viral loads and compared to healthy controls (n = 5). Cut-off for the categories “low” (n = 5) and “high” (n = 3) was 10e5 GE per ml swab-derived liquid. Scaled expression levels are color-coded and the percentage of cells expressing the gene is size-coded. Differential expression analyses were calculated with MAST, and p-values were adjusted with the Benjamini-Hochberg method. For comparisons to the uninfected control, significance is indicated by a black circle. For comparisons between conditions (early vs. late or low viral load vs. high viral load) significance is depicted by an asterisk. p ≤ 0.05 (*), p ≤ 0.01 (**), p ≤ 0.001 (***), p ≤ 0.0001 (****), p > 0.05 (not significant, ns). Ave. Exp. = average expression, Pct. Exp. = percent of cells expressing the gene.

Fig. 4. Autophagy induction limits SARS-CoV-2 growth in vitro.

Schematic representation of autophagy signaling indicating site of action of small-molecule compounds (orange columns) or siRNA knockdown (green columns) used for pathway modulation. VeroFM cells were infected with SARS-CoV-2 (MOI = 0.0005) and treated with the indicated compounds: polyamines spermidine (spd, 100 µM) and spermine (spm, 100 µM), AMPK modulator AICAR (25 µM), mTORC1 inhibitor rapamycin (rap, 300 nM), VPS34 and ULK1 inhibitors SAR405 (1 µM) and MRT68921 (5 µM), AKT1 inhibitor MK-2206 (1 µM), BECN1-stabilizers SMIP004 (S004, 10 µM), valinomycin (val, 5 µM), niclosamide (nic, 10 µM) or DMSO (vehicle, dashed lines). For siRNA knockdown, VeroFM cells were transfected with 60 nM siRNA targeting ATG5, ATG7, FIP200, BECN1, and infected with SARS-CoV-2 48 h later. SARS-CoV-2 infectious virus particle units (PFU) and genome equivalents (GE) per ml were determined by plaque assay and real-time RT-PCR at 24 h p.i., respectively. Data are presented as fold differences (see Supplementary Fig. 11 for unprocessed data). In all panels, error bars =SEM derived from n = 3 (n = 2 for spd, spm, AICAR) biologically independent samples from one experiment. Statistics were done for experiments with n = 3 using a two-way ANOVA, Dunnett’s, and one-way ANOVA for siRNA knockdown experiments (green). Compounds with significant SARS-CoV-2 inhibition (spm, spd, MK-2206, S004, val, nic) in VeroFM cells were confirmed in Calu-3 cells (see Supplementary Fig. 11f–g). p ≤ 0.05 (*), p ≤ 0.01 (**), p ≤ 0.001 (***), p ≤ 0.0001 (****), p > 0.05 (not significant, ns). Abbreviations: AICAR, 5-Aminoimidazole-4-carboxamide ribonucleotide; AKT1, RAC-alpha serine/threonine-protein kinase; AMPK, AMP-activated protein kinase; ATG5, autophagy-related 5; ATG7, ubiquitin-like modifier-activating enzyme ATG7; ATG12, ubiquitin-like protein ATG12; aut, autophagosome; BECN1, Beclin-1 protein; CLEAR, coordinated lysosomal expression and regulation; cmpd, compound; EP300, histone acetyltransferase p300; elF5AH, eukaryotic translation initiation factor 5A, hypusinated; FIP200, FAK family kinase-interacting protein of 200 kDa; FKBP51, 51 kDa FK506-binding protein; GE, genome equivalents; lys, lysosome; mTORC1, mechanistic target of rapamycin complex 1; nic, niclosamide; PFU, plaque-forming units; phag, phagophore; PHLPP, PH domain leucine-rich repeat-containing protein phosphatase; rap, rapamycin; S004, SMIP-004; SKP2, S-phase kinase-associated protein 2; spd, spermidine; spm, spermine; TFEB, transcription factor EB; ULK1, unc-51-like kinase 1; val, valinomycin; VPS34, phosphatidylinositol 3-kinase catalytic subunit type 3.

Fig. 5. Polyamine supplementation and autophagy induction modulate cellular metabolism and limit SARS-CoV-2 growth in primary human lung cells and intestinal organoids.

a Analysis and regulation of significantly altered pathways of SARS-CoV-2-infected VeroFM cells (24 h p.i.) upon treatment with spermine (spm, 100 µM, thin outlined circles) or niclosamide (nic, 10 µM, solid outlined circles). Analysis was done as described in Fig. 1 (n = 4 biological independent replicates per group, one experiment). All p-values were determined by a two-way ANOVA and Tukey´s post hoc test. FDRs were adjusted using the Benjamini-Hochberg method. b Primary human airway epithelial cells (n = 6, from two independent experiments with n = 3) were infected with SARS-CoV-2 (MOI = 0.1) and treated with spm, nic, or vehicle as mentioned in a for 24 h and 48 h, respectively. P-values were determined by two-way ANOVA, Dunnett’s, mean with SEM. c Human intestinal organoids (n = 8, from two independent experiments with n = 4) were infected with SARS-CoV-2 (MOI = 0.05) and treated with spm, nic, or vehicle as above for 48 and 72 h, respectively. Data are presented as fold change PFU/ml or GE/ml. Two-way ANOVA, Dunnett’s, mean with SEM was used for p-value calculation. Technical outliers were removed resulting in n = 7 for spm, 72 h p.i. (PFU/ml), and nic, 48 h p.i. (GE/ml) as well as n = 6 for veh, 72 h p.i. (GE/ml). d VeroFM cells were infected with SARS-CoV-2 (MOI = 0.0005) and treated with increasing concentrations of spermidine (spd), spm, MK-2206, and nic (n = 3 biological replicates per concentration from one experiment, mean with SEM) to calculate the 50% inhibitory concentration (IC₅₀). Technical outliers were removed resulting in n = 2 for nic, 0.046 µM, 48 h p.i. (PFU/ml). Limit of detection (LOD) of the plaque assay was 50. Data are presented as percentages of the negative control (see Supplementary Fig. 14 (raw data) and Fig. 10 (cytotoxicity assays)). p ≤ 0.05 (*), p ≤ 0.01 (**), p ≤ 0.001 (***), p ≤ 0.0001 (****), p > 0.05 (not significant, ns). Abbreviations: GE, genome equivalents; LOD, limit of detection; met., metabolism; nic, niclosamide; PFU, plaque-forming units; spd, spermidine; spm, spermine; veh, vehicle.