SARS-CoV–2 rapidly replicates in Caco–2 cells
To investigate potential antiviral compounds against SARS-CoV–2, we established a highly permissive SARS-CoV–2 cell culture model, characterized by fast progression of viral infection with visible cytopathogenic effect (CPE) already after 24 hours (Fig. 1a). To determine if productive viral infection takes place in this model, we measured viral RNA copies in supernatant during a 24 hour time period (Fig. 1b). The number of SARS-CoV–2 RNA molecules increased continuously after infection (Fig. 1c), indicating that the virus undergoes full replicatory cycles in our cell culture model. Taken together, we established a functional SARS-CoV–2 cell culture model that allows investigation of the different steps of the SARS-CoV–2 life cycle in cells.
Translation inhibitors prevent SARS-CoV–2 replication
To determine the temporal profile of SARS-CoV–2 infection, we infected Caco–2 cells with SARS-CoV–2 at a multiplicity of infection of one, cultured them for a range of 2–24 hours and quantified translatome and proteome changes by mePROD proteomics compared to mock- infected samples (Fig. 2a, Supplementary Table 1). In each of three replicates, we quantified relative translation rates and relative protein levels for approximately 4,200 proteins and over 7,000 proteins, respectively. Dimension reduction by principal component analysis (PCA) showed that replicates clustered closely and revealed a first separation of infected samples from mock controls after 6 hours of infection (Fig. 2b). The translational landscape consecutively reshaped with longer infection times. Many RNA viruses decrease cellular protein synthesis, as has been suggested for SARS-CoV–114,15. However, when monitoring global translation rates, only minor changes were observed with a maximum of a 23% decrease in translation after 10 hours of infection with SARS-CoV–2 (Fig. 2c and Extended Data Fig. 1). To examine the kinetics of viral protein expression, we quantified the translation of viral proteins across time points (Fig. 2d). As expected, viral protein synthesis increased over time, starting 6 hours after transfection. Viral protein translation continuously accelerated, suggesting an increase in either translation efficiency or mRNA levels of viral genes. To identify pathways potentially important for virus amplification in cells, we determine host proteins that correlated with translation kinetics of viral proteins. Standardized (Z score), averaged profiles of all quantified viral proteins were used as reference profiles, distance to this profile calculated for all quantified host proteins, and a network analysis carried out for the top 10% quantile of nearest profiles (Extended Data Fig. 1b). Pathway analyses of the network revealed extensive remodelling of translational patterns of the host translation machinery itself, potentially explaining how a significant change in global protein synthesis is avoided (Fig. 2c, e, Extended data Fig. 1c). In addition, we detected significant enrichment of components of several other pathways, such as antigen presentation or vesicular trafficking.
Host translation has been targeted previously11,16 to inhibit replication of diverse coronaviruses such as SARS-CoV–1 or MERS-CoV with FDA approved small molecules (Extended Data Fig. 1d). In contrast to other viruses, for which host protein synthesis has been reported to be repressed to allow increased synthesis of viral proteins14,15, our data suggested that SARS-CoV–2 only caused minor changes in host translation capacity (Fig. 2c). Thus, we speculated that SARS-CoV–2 replication might be more sensitive to translation inhibition, since viral proteins compete with host proteins for efficient translation. We tested two translation inhibitors—cycloheximide (inhibitor of translation elongation) and emetine (inhibits 40S ribosomal protein S14)—for their ability to reduce SARS-CoV–2 replication.
Antiviral activity of these translation inhibitors against different coronaviruses had been observed previously (Extended Data Fig. 1d)11,16. All compounds caused significant inhibition of SARS-CoV–2 replication at non-toxic concentrations (Fig. 2f, g and Extended Data Fig. 1e, f). Taken together, analysing the translatome of cells infected with SARS-CoV–2 revealed the temporal profile of viral and host protein responses leading to the discovery of translation inhibitors as potent inhibitors of SARS-CoV–2 replication in cells.
Regulation of cellular pathways by SARS-CoV–2 infection
Next, we characterized changes in cellular protein networks upon SARS-CoV–2 infection at the level of total protein abundance. To obtain a general understanding of host proteome changes after infection, we analysed system-wide changes in protein levels over time (Fig. 3a, Supplementary Table 2). While only minor host proteome changes were observed at early infection time points, the proteome underwent extensive modulation 24 hours post infection. Hierarchical clustering identified two main clusters of proteins regulated: The first cluster consisted of proteins reduced during infection and mainly included proteins belonging to cholesterol metabolism (Extended Data Fig. 2 and Supplementary Table 3). The second cluster was composed of proteins increased by infection and revealed strong increases in RNA modifiers, such as spliceosome components, and carbon metabolism (Fig. 3b, c, Extended data Fig. 3a and Supplementary Table 4). Remarkably, for most spliceosome components increased by SARS-CoV–2, direct binding to viral proteins of SARS-CoV–1 or other coronaviruses has been shown (Extended data Fig. 3b)9,17–21. Thus, we tested whether inhibition of splicing or glycolysis may be able to prevent SARS-CoV–2 replication. Addition of pladeinolide B, a spliceosome inhibitor targeting splicing factor SF3B122, prevented viral replication at conditions non-toxic to the host cells (Fig. 3d, Extended Data Fig. 3c), revealing splicing as an essential pathway for SARS-CoV–2 replication and potential therapeutic target.
Next, we assessed the propensity to inhibit the other main cellular cluster increased upon SARS-CoV–2 infection—carbon metabolism. Indeed, inhibition of glycolysis by 2-deoxy-D- glucose (2-DG), an inhibitor of hexokinase (i.e. glycolysis), had previously been shown to be effective against other viruses in cell culture and suppressed rhinovirus infection in mice23.
Blocking glycolysis with non-toxic concentrations of 2-DG prevented SARS-CoV–2 replication in Caco–2 cells (Fig. 3e and Extended Data Fig. 3d). Together, our quantitative analyses of proteome changes upon SARS-CoV–2 infection revealed host pathways changes upon infection and revealed spliceosome and glycolysis inhibitors as potential therapeutic agents for COVID–19.
Kinetic proteome profiling to identify potential antiviral targets
In order to better understand host proteins co-increased with viral proteins and to provide with potential additional inhibitors of SARS-CoV–2 replication, we next analysed for proteins showing a similar abundance trajectory over time as viral proteins (Fig. 4a). We computed distance and false discovery rate for each protein compared to the averaged profile of all viral proteins and filtered the data for all proteins with a FDR below 0.01 and performed functional interaction network analysis. Furthermore, in line with our analyses of translatome changes, we found RNA modifying components and metabolic pathways enriched (Extended Data Fig. 3b, c). Next, we carried out gene ontology (GO) analysis of all proteins showing an abundance trajectory significantly similar to the viral protein profile (Supplementary Table 5). We identified a major cluster of metabolic pathways enriched in our data, which consisted of diverse nucleic acid metabolism sub-pathways (Fig. 4b). Coronavirus replication depends on availability of cellular nucleotide pools24,25 Thus, we tested the effect of inhibitors of nucleotide synthesis on SARS-CoV–2 replication in cells. We did not observe any antiviral effects of brequinar, an inhibitor of dihydroorotate dehydrogenase, required for de novo pyrimidine biosynthesis, tested at a maximum concentration of 10 µM. In contrast, ribavirine, which inhibits inosine monophosphate dehydrogenase (IMPDH), the rate-limiting enzyme in de novo synthesis of guanosine nucleotides, inhibited SARS-CoV–2 replication at low micromolar concentrations (Fig. 4c and Extended Data Fig. 4a). This is consistent with previous studies showing that targeting IMPDH inhibits replication of coronaviruses HCoV- 43, CoV-NL63 and MERS-CoV16, but strikingly not of SARS-CoV–126. This suggests ribavirine, which is an approved antiviral medication, may be regarded as a possible candidate for further testing.
Additionally, components of the protein folding machinery showed a behaviour highly comparable to the viral proteins (Extended Data Fig. 3e), consistent with a perturbation of host cell proteostasis due to the higher folding load resulting from high translation rates of viral proteins in the cytosol and ER. Therefore, we tested effects of NMS–873, a small molecule inhibitor of the AAA ATPase p97, on SARS-CoV–2 replication. p97 is a key component of proteostasis affecting protein degradation, membrane fusion, vesicular trafficking and disassembly of stress granules27. NMS–873 has been shown to inhibit influenza A and B replication with decreased expression and mislocalization of viral proteins as suggested mechanism28. NMS–873 completely inhibited SARS-CoV–2 at low nanomolar concentrations (Fig. 4d and Extended Data Fig. 4b). In summary, analyses of the effects of SARS-CoV–2 infection on the host cell proteome revealed major readjustments in cellular function, particularly of splicing, proteostasis and nucleotide biosynthesis. Compounds modulating these pathways prevented SARS-CoV–2 replication in cells.