SARS-CoV-2 hijacks folate and one-carbon metabolism for viral replication

Abstract

The recently identified Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) is the cause of the COVID-19 pandemic. How this novel beta-coronavirus virus, and coronaviruses more generally, alter cellular metabolism to support massive production of ~30 kB viral genomes and subgenomic viral RNAs remains largely unknown. To gain insights, transcriptional and metabolomic analyses are performed 8 hours after SARS-CoV-2 infection, an early timepoint where the viral lifecycle is completed but prior to overt effects on host cell growth or survival. Here, we show that SARS-CoV-2 remodels host folate and one-carbon metabolism at the post-transcriptional level to support de novo purine synthesis, bypassing viral shutoff of host translation. Intracellular glucose and folate are depleted in SARS-CoV-2-infected cells, and viral replication is exquisitely sensitive to inhibitors of folate and one-carbon metabolism, notably methotrexate. Host metabolism targeted therapy could add to the armamentarium against future coronavirus outbreaks.

Fig. 1. Metabolic changes induced by early SARS-CoV-2 infection.

a Schematic of the metabolic profiling approach. Vero E6 TMPRSS2 + cells were infected with concentrated SARS-CoV-2 at a MOI = 2 or mock-infected with virus-depleted flow-through, infected for 8 h and profiled by LC/MS and RNAseq in parallel. b Immunofluorescence of nucleoprotein (IF-Np) and fluorescence in situ hybridization (FISH) for + strand viral genomic RNA (FISH-gRNA) and merge with Hoechst stained nuclei in infected vs mock-infected cells. See also Supplementary Fig. 1b. The experiment was reproduced in at least six independent experiments. c Volcano plot visualization of 8 h SARS-CoV-2 versus mock vs infected Vero E6 RNAseq from n = 3 datasets. Selected inflammatory (red) and metabolism (blue) pathway genes are shown. P-value generated with DESeq under default setting. d Box plot visualization of RNAseq reads in SARS-CoV-2 versus mock-infected cells. n = 3 biologically independent samples were examined over one independent experiment. Data are presented as mean values ± SD. One-way ANOVA with multiple comparison using the Sidak method. e PCA of 106 intracellular metabolites, as determined by LC-MS in SARS-CoV-2-infected (red) or mock-infected (gray) cells, from n = 6 biologically independent replicates. f Volcano plot visualization of log2 fold change (x-axis) and -log10(P value; y-axis) of intracellular metabolites measured by LC-MS. Significantly increased or decreased metabolites related to glycolysis, de novo purine synthesis, 1C metabolism/ transsulfuration pathway, amino acids, histidine catabolism, urea cycle/polyamine metabolism, and de novo pyrimidine synthesis are labeled. n = 6 biologically independent samples were examined over one independent experiment, P-values were generated with two-tailed P value from Student’s t test. g Fold change of intracellular glucose, lactate, de novo purine, and one-carbon metabolite levels detected by LC-MS in SARS-CoV-2 and mock-infected cells. Mock-infected levels were set to 1. All barplots show mean ± standard deviation (SD). *P < 0.05, **P < 0.01, or ***P < 0.001 from Student’s two-tailed t test. Druggable targets are labeled in red. See also Supplementary Fig. 4. Source data are provided as a Source Data file.

Fig. 2. SARS-CoV-2 induced glycolysis and one-carbon metabolism supports viral RNA and protein expression, replication, and cytopathic effect.

a +strand gRNA FISH, Np IF, and merge with Hoechst stained nuclei of Vero-E6 TMPRSS2 + cells cultured in media with 25 mM glucose versus galactose as the sugar source and infected with SARS-CoV-2. b Mean ± SD fold change of live Vero E6 TMPRSS2 + cell number and median tissue culture infectious dose (TCID50) presented as fluorescent-focus units (FFU) per ml of culture supernatant at 48 h post infection of cells cultured in glucose versus galactose from n = 3 biologically independent replicates. c FISH analysis of +strand gRNA, IF for Np, and merge with Hoechst stained nuclei in SARS-CoV-2-infected Vero E6 TMPRSS2 + cells treated with DMSO or 100 nM piericidin A (PierA). d Mean ± SD fold change live cell number from n = 3 biologically independent replicates, as in c. e Mean ±  TCID50 from n = 3 biologically independent replicates in Vero E6 TMPRSS2 + , as in b. f Phase microscopic images of SARS-CoV-2 versus mock-infected Vero E6 TMPRSS + cells cultured for 48 h with DMSO, 1 μM of methotrexate (MTX), 30 μM hypoxanthine (hypo), 100 μM thymidine, or 1 mM formate, as indicated. Yellow scale bar indicates 100 μm. g Mean ± fold change live cell # and TCID50/ml from samples collected as in f from three biologically independent replicates. h FISH microscopic analysis of viral gRNA, IF of Np, and merge with Hoechst stained nuclei in SARS-CoV-2-infected Vero E6 TMPRSS2 + cells treated for 48 h with the indicated conditions. Yellow arrows indicate representative cells with high gRNA (red) but low Np (green) signal. i Ratios of +strand gRNA FISH versus Np IF signals from 500 Vero E6 TMPRSS2 + cells from 20 random fields for each condition in h are shown. j FISH microscopic analysis of viral gRNA, IF of Np, and merge with Hoecshst stained nuclei in SARS-CoV-2-infected A549 ACE2 + cells treated with the indicated conditions. Yellow arrows indicate representative cells with high gRNA (red) but low Np (green) signal. k Fold change mean ± SD live cell # and TCID50/ml from A549 ACE2 + samples collected as in j from n = 3 biologically independent replicates. l Flow-FISH analysis of Np subgenomic RNA in SARS-CoV-2-infected A549 ACE2 + cells treated with the indicated conditions. Of note, the leftmost peak in each row indicates uninfected cells. m Mean ± SD of Np subgenomic RNA mean fluorescence intensity (MFI) values from n = 3 biologically independent replicates, as in l. In all panels, cells were infected at MOI = 0.1 for 48 h. Microscopy images are representative of at least n = 3 biologically independent values. P-values in this figure were calculated by one-way ANOVA with multiple comparisons using Sidak method. Source data are provided as a Source Data file.

Fig. 3. SARS-CoV-2 induced serine one-carbon metabolism supports viral RNA and protein expression, replication, and cytopathic effect.

a Phase microscopic images of SARS-CoV-2 versus mock infected Vero E6 TMPRSS2 + cells cultured with DMSO, 10 μM of the dual SHMT1/2 inhibitor SHIN1 or 10 μM SHIN1 + 1 μM formate, as indicated. White scale bar indicates 100 μm. The experiment was reproduced in at least six independent experiments. b Mean ± SD fold change TCID50 (left) and live cell (right) from n = 3 biologically independent replicates, as in a. c IF of Np, FISH for +strand gRNA, and merge with Hoechst stained nuclei in mock-infected or SARS-CoV-2-infected Vero E6 TMPRSS2 + cells treated with DMSO, SHIN1, or SHIN1 + formate. d Mean ± SD fold change TCID50 (top) and live cell (bottom) values in SARS-CoV-2-infected A549 ACE2 + cells, treated with the indicated conditions, from n = 3 biologically independent replicates. e IF of Np, FISH for +strand gRNA, and merge with Hoechst stained nuclei in mock-infected or SARS-CoV-2-infected A549 ACE2 + cells treated with DMSO, SHIN1, or SHIN1 + formate. f Flow-FISH analysis of Np subgenomic RNA in SARS-CoV-2-infected A549 ACE2 + cells treated with the indicated conditions. g Mean ± SD values from n = 3 biologically independent replicates of viral subgenomic RNA Flow-FISH MFI values, as in f. h Immunoblot analysis of whole cell lysates from Cas9 + TMPRSS2 + Vero E6 expressing control, SHMT1 or SHMT2 sgRNAs. i Mean ± SD fold change TCID50 (left) and live cell (right) values from Vero E6 TMPRSS2 + with control, SHMT1 or SHMT2 targeting sgRNAs infected by SARS-CoV-2 from n = 3 biologically independent replicates are shown. j FISH of subgenomic Np RNA, IF of Np, FISH for +strand gRNA, and merge with Hoechst stained nuclei in cells with control, SHMT1 or SHMT2 targeting sgRNAs infected by SARS-CoV-2. In all panels, cells were infected at MOI = 0.1 for 48 h. Microscopy images are representative of at least n = 3 biologically independent values. P-values in this figure were calculated by one-way ANOVA with multiple comparisons using Sidak method. Source data are provided as a Source Data file.

Fig. 4. Schematic of SARS-CoV-2 induced one-carbon metabolism in support of viral replication.

SARS-CoV-2 induces glycolysis and one-carbon metabolism at the post-transcriptional level in newly infected cells. Serine metabolism, particularly by cytosolic SHMT1 produces carbon units for de novo purine synthesis in support of massive viral subgenomic RNA synthesis, non-structural protein expression, and viral replication.