key: cord-0935565-rpxpj54q
authors: Zeng, Jingkun; Weissmann, Florian; Bertolin, Agustina P.; Posse, Viktor; Canal, Berta; Ulferts, Rachel; Wu, Mary; Harvey, Ruth; Hussain, Saira; Milligan, Jennifer C.; Roustan, Chloe; Borg, Annabel; McCoy, Laura; Drury, Lucy S.; Kjaer, Svend; McCauley, John; Howell, Michael; Beale, Rupert; Diffley, John F.X
title: Identifying SARS-CoV-2 Antiviral Compounds by Screening for Small Molecule Inhibitors of Nsp13 Helicase
date: 2021-04-08
journal: bioRxiv
DOI: 10.1101/2021.04.07.438808
sha: af20d9f24802088289502a90ff82c3a59318d626
doc_id: 935565
cord_uid: rpxpj54q

The coronavirus disease 2019 (COVID-19) pandemic, which is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a global public health challenge. While the efficacy of vaccines against emerging and future virus variants remains unclear, there is a need for therapeutics. Repurposing existing drugs represents a promising and potentially rapid opportunity to find novel antivirals against SARS-CoV-2. The virus encodes at least nine enzymatic activities that are potential drug targets. Here we have expressed, purified and developed enzymatic assays for SARS-CoV-2 nsp13 helicase, a viral replication protein that is essential for the coronavirus life cycle. We screened a custom chemical library of over 5000 previously characterised pharmaceuticals for nsp13 inhibitors using a FRET-based high-throughput screening (HTS) approach. From this, we have identified FPA-124 and several suramin-related compounds as novel inhibitors of nsp13 helicase activity in vitro. We describe the efficacy of these drugs using assays we developed to monitor SARS-CoV-2 growth in Vero E6 cells.

More than 100 million people have been infected and more than 2.5 million people have died worldwide due to the COVID-19 pandemic as of late February 2021 (1). The number of infections continues to rise, with more than 150,000 daily infections globally, highlighting the importance of developing effective vaccines and therapeutics for the prevention and treatment of COVID-19.

Vaccines are under mass roll-out nevertheless concerns over potentially vaccine-resistant variants have been growing. It has been shown in cell culture experiments that the virus has the potential to evolve mutant strains that can evade neutralizing antibodies produced by the body to combat infections (2) , and reports have emerged that certain variants could cause re-infections (3) . Many of the vaccines being used, e. g. the COVID-19 vaccines produced by Pfizer, Moderna and AstraZeneca-Oxford, target the spike protein on the surface of SARS-CoV-2 (4) . Multiple mutations of the spike protein have been found around the globe (5) . Although there is no clear evidence of vaccine-evading variants yet, vaccines against rapidly evolving structural proteins might not protect against all newly emerging strains and are unlikely to have pan-coronavirus efficacy. With these uncertainties regarding SARS-CoV-2 vaccines, multiple layers of protection and treatment against COVID-19 are needed including the identification of therapeutic drugs that can interfere with viral entry or viral propagation is of utmost importance. Nonetheless, therapeutic options for COVID-19 are currently limited. De novo development of antiviral therapies generally requires between 10 to 17 years (6) . The repurposing of drugs originally developed for other uses could provide a practical approach for the fast identification, characterisation and deployment of antiviral treatments (6, 7) . The repurposing of approved or investigational drugs exploits existing detailed information on drug chemistry together with human pharmacology and toxicology, allowing rapid clinical trials and regulatory review (6) . This strategy has proven useful so far, with a few repurposed medicines having been authorised by different regulatory agencies to treat COVID-19, such as remdesivir, an antiviral developed to treat Ebola (8, 9) . Remdesivir is a pro-drug inhibitor of the RNA-dependent RNA polymerase (RdRp) that shows inhibitory activity against all three strains, SARS-CoV-1, MERS-CoV and SARS-CoV-2, of the coronavirus outbreaks in this century (10) (11) (12) . Drug resistance to monotherapies may develop rapidly, particularly in RNA viruses where mutations occur frequently, thus it would be useful to have multiple antiviral drugs (13, 14) .

SARS-CoV-2 is a positive-sense single stranded RNA virus that encodes at least nine enzymatic activities in two overlapping large polyproteins pp1a and pp1ab (15, 16) . Once expressed in the host cell, pp1a and pp1ab are processed by virus-encoded proteases into 16 non-structural proteins (nsps) (15, 16) . Coronavirus nsp13, one of the non-structural proteins, is a superfamily 1B (SF1B) helicase that can unwind DNA or RNA in an NTP-dependent manner with a 5' to 3' polarity (17) (18) (19) (20) . Moreover, nsp13 harbours RNA 5'-triphosphatase activity that could play a role in viral 5' RNA capping (18, 20, 21) . Nsp13 is highly conserved among SARS-like coronaviruses with 99.8% sequence identity (600 out of 601 amino acids) between SARS-CoV-1 and SARS-CoV-2 (22). Importantly, nsp13 is a key component of replication-transcription complexes (RTC) and is indispensable for the coronavirus life cycle, making it a promising target for pan-coronavirus antivirals (23-26).

As part of a larger project to identify small molecule inhibitors of all SARS-CoV-2 enzymes, we report the development of a high throughput fluorescence resonance energy transfer (FRET)-based assay for nsp13 helicase activity in vitro. We used this assay to screen a custom compound library of over 5000 previously characterised pharmaceuticals. We identify FPA-124 and suramin-like compounds as novel nsp13 inhibitors that also show antiviral activity in a cell-based viral proliferation assay.

For SARS-CoV-1 nsp13, it has previously been shown that GST-tagged nsp13 expressed in insect cells is more active than MBP-or 6xHis-tagged nsp13 expressed in bacteria (27, 28). Therefore, we used a baculovirus-insect cell expression system to express and purify two differently tagged SARS-CoV-2 nsp13 versions, GST-nsp13 and 3xFlag-His6-nsp13 (FH-nsp13). Both nsp13 variants were purified using affinity chromatography based on GST or the 3xFlag tag followed by gel filtration Figure S3A) . Nsp13 was incubated with compounds for 10 min, then the substrates were dispensed to start the reaction ( Figure 3A) . The screen was performed at two compound concentrations, 1.25 and 6.25 µM. We observed that, on the day of the screen, the reaction was complete by 10 minutes (Supplementary Figure S3B) . The time required to dispense substrates to all the wells within a plate to start the reaction was ~10 seconds, while the time required to read all wells was ~75 seconds (Supplementary Figure S3C-D) . We applied a positional correction to the calculation of the initial velocity of reactions for each plate to compensate for this time delay (Figure 3B-C) . A total of 339 compounds showed >20% reduction of corrected initial velocity or showed >10% reduction of endpoint signals at either 1.25 µM or 6.25 µM compound concentration. After manual inspection of their kinetic curves, 142 of the 339 compounds showed a clear effect in HTS reactions and were selected as primary hits (example in Figure 3D ). To find specific nsp13 inhibitors, we initially selected 35 of the primary hits that displayed >30% reduction of corrected initial velocity for subsequent validation experiments. (Figure 3E ).

In a first validation experiment, we tested the compounds using the same assay conditions as in the HTS and determined the compound concentration at which half maximal nsp13 inhibition (IC50) was observed. We found that 27 of the 33 tested compounds showed nsp13 inhibition with apparent IC50 values < 30 µM (Supplementary Figure S4 and Table 1 , second column). We then characterised the hits under more physiological conditions where the DNA substrate was replaced by its RNA counterpart, and the ATP concentration was increased from 100 µM to 2 mM. Nonspecific inhibition of enzymes due to colloidal aggregation of compounds is the most common source of false-positives in high-throughput inhibitor screens (31-34). The presence of non-ionic detergents can reduce colloid formation, which can lead to a right-shift of dose-response curves for many aggregation-prone compounds (35). In order to uncover potential aggregators, we added 0.02% Tween-20 to the reaction. Surprisingly, under these conditions only 5 out of 27 compounds still inhibited nsp13 and showed similar IC50 values as in the first validation experiment (Supplementary Figure S5 and Table   1 , third column). The remaining 22 compounds lost the strong inhibition shown in the first experiment, as indicated by the increase in their IC50 values ( To test directly if some of our compounds could inhibit nsp13 activity via a detergent-sensitive aggregation mechanism (32, 35), we tested 15 selected compounds in the presence or absence of Tween-20 with otherwise identical assay conditions (Supplementary Figure S6) . The presence of Tween-20 reduced the inhibition of 13 of the 15 tested compounds, causing an IC50 shift greater than 4-fold (Supplementary Figure S6 and Table 1 , fourth column), suggesting the nsp13 inhibition shown by these drugs in HTS and in the first validation experiment could be due to an aggregation effect. Colloidal aggregators absorb and inhibit enzymes without specificity (31, 33). Potential aggregators can be further confirmed by counter-screening against other enzymes (33). Hence, we tested whether these compounds could inhibit the SARS-CoV-2 RNA-dependent RNA polymerase (RdRp) using a FRET-based assay developed in an accompanying manuscript (Bertolin et al.).

Compounds with mild detergent sensitivity that still showed moderate to weak nsp13 inhibition in the presence of detergent, also inhibited RdRp in a similar fashion (e.g., Avasimibe, Supplementary Figure S7 and Supplementary Table S1 ). On the other hand, compounds that no longer inhibited nsp13 in the presence of detergents did not inhibit RdRp (e.g., A-385358, Supplementary Figure S7 and Supplementary Table S1 ). All together, these results suggest that many of the compounds identified in the nsp13 HTS were aggregation-based inhibitors. These findings are consistent with Aggregator Advisor, an open access tool that advises on the likelihood of colloidal aggregation based on criteria including lipophilicity (LogP>3.5) and on a structural similarity index (>85% similarity compared to reported aggregators) (36). We analysed the selected hits using this method and, not surprisingly, 21 out of 22 detergent-sensitive compounds showed LogP>3.5 (Supplementary Table   S2 ).

After this first round of analysis, we re-examined our primary hit list, and selected another 12 hit compounds with a LogP<3.6 and with no significant similarity to known aggregators (Supplementary Table S3 and Figure 3E ) (36). We also tested two published nsp13 inhibitors, SSYA10-001 (37, 38) and myricetin (39). We determined the IC50 values of these compounds in the presence or absence of Tween-20 ( Table 2 After all the above analyses, we found 7 compounds that inhibited nsp13 with no detergent-sensitivity (PDK1/Akt/Flt Dual Pathway Inhibitor, suramin and suramin-related compounds, and FPA-124, Supplementary Figure S10). PDK1/Akt/Flt Dual Pathway Inhibitor showed promiscuous inhibition in several other screens, including the RdRp screen, nsp3 papain-like protease screen, nsp5 main protease screen and nsp15 endonuclease screen (Biochem J, this issue). Therefore, it was excluded from further analysis. Suramin and suramin-related compounds (NF 023, PPNDS, Evans Blue and Diphenyl Blue) are heavily negatively charged molecules and may inhibit nucleic acid-binding proteins by binding to positively charged protein regions (40-44). They were also identified as hits in the RNAdependent RNA polymerase (RdRp) screen (Bertolin et al. Biochem J, this issue) but not in other parallel screens, consistent with the ability of these polyanionic compounds to strongly inhibit certain nucleic acid-binding proteins (41, (45) (46) (47) . Notably, a kinase inhibitor, FPA-124, inhibited nsp13 with a micromolar IC50 of ~9 µM, and its IC50 value was not affected by the addition of the non-ionic detergents Tween-20 or Triton X-100 (Supplementary Figures S8-S9) , suggesting its nsp13 inhibition was unlikely to be due to aggregation. Moreover, FPA-124 did not score as a hit in the RdRp screen nor parallel screens against other SARS-CoV-2 enzymes (Biochem J, this issue), indicating its inhibitory activity is likely target specific. Examples of the observed IC50 curves are shown in Figure 4 .

Suramin, NF 023 and FPA-124 are compounds whose enzymatic inhibition is largely insensitive to detergent addition (Figure 4A-C) . On the other hand, navitoclax and linoleic acid are examples of compounds with high LogP values and strong sensitivity to detergents suggesting that their inhibitory effect could be attributed to colloidal aggregation occurring at certain micromolar concentrations ( Figure 4D-E) .

We also tested these compounds in a gel-based nsp13 helicase assay ( Figure 4G) . We used an RNA substrate consisting of a Cy3 strand and an unlabelled complementary strand with a 5' overhang. A competitor strand was included to capture the unlabelled strand. Reaction products were then separated by native PAGE and detected by Cy3 fluorescence. We tested the identified detergentinsensitive compounds -FPA-124, suramin, NF 023, PPNDS, Evans Blue, Diphenyl Blue together with the 2 published SARS-CoV-1 nsp13 inhibitors, myricetin and SSYA10-001, at a compound concentration of 15 µM in the presence of 0.01% Triton X-100. All 6 identified compounds clearly reduced the generation of single-stranded Cy3 strand, confirming that the compounds inhibited helicase activity ( Figure 4G ). As expected, based on their IC50 values in the presence of detergent ( Table 2) , myricetin and SSYA10-001 inhibited nsp13 less efficiently under these conditions.

We next evaluated potential antiviral activity of the compounds against SARS-CoV-2 in Vero E6 cells.

Our experiments thus far identified 6 compounds that inhibited nsp13 in biochemical assays. Of these, suramin, PPNDS, NF 023, Evans Blue and Diphenyl Blue are structurally related compounds containing at least one polysulfonated naphthyl group, which is believed to be the critical pharmacophore (40, 41). Suramin is the prototypical naphthalene polysulfonated compound and most studied drug of this group, so we tested suramin in our cell-based experiments. The other validated inhibitor was FPA-124, which is a cell permeable selective AKT inhibitor (48) . We also included the two published SARS-CoV-1 nsp13 inhibitors SSYA10-001 and myricetin (37-39). To test an effect on viral replication, the compounds were added to Vero E6 cells, and then cells were infected with SARS-CoV-2 at a multiplicity of infection of 0.5. After 22 hours cells were fixed and analysed by immunofluorescence ( Figure 5A ). Immunofluorescent detection of SARS-CoV-2 N protein was used as a read out for viral replication in cell culture (see Material and Methods). The compounds were tested over a range of concentrations, and the half maximal effective concentration (EC50) for each compound was calculated (Figure 5B-C) . Suramin and FPA-124 showed viral inhibition with EC50 values of 9.9 µM and 14 µM, respectively, while the published nsp13 inhibitors, myricetin and SSYA10-001, presented a weaker viral inhibition potency (EC50=32 µM and 81 µM, respectively). Of these four tested compounds, only FPA-124 showed considerable inhibition of cellular growth at high concentrations (100-300 µM), perhaps due to inhibition of AKT kinase (Supplementary Figure   S11A ) (48) .

Combining two drugs with antiviral activity and different viral or cellular targets can result in improved outcomes compared to antiviral monotherapy (49) . Therefore, we decided to investigate potential synergy between nsp13 inhibitors and the RNA-dependent RNA polymerase inhibitor remdesivir (10) (11) (12) . Following the same infection protocol described in Figure 5A , the compounds were again tested over a range of concentrations in the presence of 1 µM remdesivir. At this concentration, remdesivir alone inhibited viral infection by less than 20%. None of the evaluated nsp13 inhibitors showed significant synergy with remdesivir under these assay conditions (Supplementary Figure S11B ).

The SARS-CoV-2 protein nsp13 possesses helicase activity and is essential for viral replication and proliferation (24-26, 50). Moreover, nsp13 is the most conserved non-structural protein within the coronavirus family, making it a very promising target for the development of pan-coronavirus antivirals (22) . In this study, we developed a robust FRET-based helicase assay and used it for highthroughput inhibitor screening against the SARS-CoV-2 helicase nsp13 testing over 5000 previously characterised pharmaceuticals. We report the identification of FPA-124 and suramin-like compounds as novel inhibitors of nsp13. We performed comparative analyses between the two most promising inhibitors reported in this work (FPA-124 and suramin) and two SARS-CoV-1 nsp13 inhibitors described in the literature (myricetin (39) and SSYA10-001 (37, 38)). We showed that FPA-124 and suramin have lower IC50 against nsp13 in vitro and lower anti-viral EC50 in cell-based assays than the two published compounds.

Colloid formation by aggregation-prone compounds is one of the main sources of false positives in drug discovery (33, 34). Adding non-ionic detergents can reduce colloid formation and thus advise on the specificity of HTS hits (31, 35). All subsequent screens in this series of papers used detergents in the primary screens. Nsp13 inhibition by FPA-124 or suramin showed no or little detergent sensitivity.

We performed the SARS-CoV-2 nsp13 HTS in the absence of detergents and obtained 142 primary hits. We then tested a selection of 47 hits with different non-ionic detergents in validation assays. Indeed, myricetin has been reported to have an aggregator-like behaviour (51) (52) (53) and to form colloidal aggregates detectable by dynamic light scattering (54) , raising concerns about the specificity of its reported SARS-CoV-1 nsp13 inhibition (39). We tested a selection of the detergent sensitive hits against a different enzyme, the SARS-CoV-2 RNA-dependent RNA polymerase (RdRp). We found they had similar inhibitory effect on both nsp13 and RdRp further suggesting their mode of action could involve non-specific protein capture by colloid formation. Nonetheless, detergent sensitivity does not completely rule out a bona fide inhibitory effect (36). Sometimes this promiscuous activity occurs solely at different concentrations than specific binding-based inhibition. Therefore, some of the compounds that presented a detergent-dependent decrease in inhibition reported in this work may still have bona fide inhibitory activity against nsp13 at certain concentrations.

FPA-124 is a cell-permeable Akt inhibitor that induces apoptosis in multiple cancer cell lines (48) . We have shown that this compound has an IC50 of ~9 µM and is not detergent-sensitive in in vitro assays.

FPA-124 was not a hit in any of the other screens in this series (Biochem J, this issue) providing further evidence for specificity. Due to the nature of this compound's original cellular target, cytotoxicity is expected. Indeed, FPA-124 showed cytotoxicity at 100 µM (Supplementary Figure   S11A) . Suramin, originally synthesised by Bayer in 1916, is a clinically approved drug mainly used to treat river blindness and sleeping sickness (41, 55). We show that suramin and several of its structurally similar compounds are novel inhibitors of SARS-CoV-2 nsp13. Suramin inhibited nsp13 in vitro with an IC50 of ~1 µM and inhibited viral growth in cell-based assays with an EC50 of ~10 µM. A wide range of antiviral effects have been reported for suramin, as it inhibits Zika virus, dengue virus, chikungunya virus, HIV, hepatitis C virus, herpes simplex type-1 virus and recently SARS-CoV-2 (41, 56) . Suramin seems to inhibit multiple steps in viral infection and replication: it interferes with virusreceptor interaction and hence viral host cell binding and uptake (57, 58) , it interferes with viral helicase activities ((45) and this work) and it interferes with viral RNA polymerase activity (46, 59) .

Suramin is a large symmetrical molecule carrying two polysulfonated naphthyl urea groups containing six negative charges at physiological pH and therefore, the basis for its many targets is likely to be its ability to strongly bind positively charged regions in proteins such as polymerases (42, 44). Indeed, we show that suramin and related compounds also inhibited the SARS-CoV-2 RdRp with micromolar IC50s without detergent sensitivity. The polyanionic nature of suramin also confers low cell membrane permeability. However, suramin can be taken up by endocytosis and the uptake rate can be enhanced by liposomal delivery (60).

We validated suramin, suramin-like compounds and FPA-124 as nsp13 inhibitors that could be subjected to careful structural optimization to generate clinically more useful compounds in the hope of increasing antiviral potency and reducing cytotoxicity. Co-structures of nsp13 and these smallmolecule inhibitors could highly benefit these optimisation processes. The viral RNA polymerase inhibitor, remdesivir, is currently the only FDA-approved small molecule antiviral for the treatment of COVID-19 patients (10, 61) . However, it is far from being a silver bullet as it has been shown to be only modestly effective in treating very sick patients (10, 61) . Thus, combining remdesivir with a mechanistically distinct drug (e. g. drugs that target other SARS-CoV-2 enzymatic activities) may improve antiviral efficacy and reduce the likelihood of emerging drug resistance (62) . Nonetheless, our experiments in Vero cells showed there was no synergistic inhibition on viral replication when combining remdesivir and our nsp13 inhibitors. Recent studies on host-virus interactions suggest that SARS-CoV-2 nsp13 may also be implicated in immune suppression as it targets several host proteins involved in innate immune signalling pathways such as the interferon pathway and NF-κB pathway (63) . Vero E6 cell experiments are useful tools for studying viral replication, but it would be interesting to test our nsp13 inhibitors in other cell lines that have intact innate immune responses in the future (64) .

In conclusion, we have identified novel small-molecule inhibitors of SARS-CoV-2 nsp13 helicase. We provide evidence that suramin and FPA-124 can be considered as leads that deserve further evaluation, as both compounds were found to inhibit nsp13 in vitro and exhibit antiviral activity against SARS-CoV-2 in a relevant cell culture model.

FH-nsp13 and GST-nsp13 were expressed in baculovirus-infected insect cells. The coding sequence of SARS-CoV-2 nsp13 (NCBI reference sequence NC_045512.2) was codon-optimised for S. frugiperda and synthesized (GeneArt, Thermo Fisher Scientific). Nsp13 was subcloned into the biGBac vector pLIB (65) 

A FRET-based fluorescence-quenching approach was designed to monitor nucleic acid strand separation catalysed by nsp13. The assay uses a forked duplex DNA or RNA substrate (15-bp duplex with a 20-nt oligo-dT 5' overhang). One strand contains a Cy3 fluorophore at the 5' end (Cy3 strand, DNA or RNA) and the other strand a Black Hole Quencher-2 (BHQ-2, DNA) or an Iowa black RQ quencher (AbRQ, RNA) at the 3' end (quencher strand). A DNA competitor strand that is complementary to the Cy3 strand prevents substrate reannealing. HPLC-purified DNA and RNA oligonucleotides were purchased from Eurofins genomics and IDT respectively with the following sequences: 

The oligonucleotides used in the radiolabelled assays are the following: nM RNA substrate. The unwinding reactions were quenched at various times with a quench solution containing 50 mM EDTA. The aliquots then were mixed with 5x loading buffer (TBE buffer, 15% Ficoll, 1 µg/ml bromophenol blue, 1 µg/ml xylene cyanol FF). The products of the unwinding reactions were separated on a 20% native-PAGE (0.5X TBE). The gels were dried and then exposed to a Phospho-Imager screen overnight, imaged using Phospho-Imager software.

The screen was performed in 384-well Greiner black flat-bottom plates (Greiner 781076). A custom compound library containing over 5000 compounds assembled from commercial sources (Sigma, Selleck, Enzo, Tocris, Calbiochem, and Symansis) were distributed in 24 plates using an Echo 550 (Labcyte) across column 3 to column 22. 2.5, or 12.5 nL of a 10 mM stock of the compounds dissolved in DMSO were arrayed and pre-dispensed into the assay plates using an Echo 550 (Labcyte), before being sealed and stored at -80 °C until screening day. All 384 wells on the plates contain 1 µL DMSO. First, 10 µL 2x Nsp13 mix (6 nM FH-nsp13) in HTS assay buffer (20 mM HEPES pH 7.6, 20 mM NaCl, 5 mM MgCl2, 1 mM DTT and 0.1 mg/ml BSA) was dispensed into columns 2 -23 of the plates to incubate with the compounds for 10 min at room temperature. Then 10 µL 2x substrate mix (200 µM ATP, 360 nM DNA substrate, 1800 nM competitor) in HTS assay buffer was dispensed to start the enzymatic reaction. The plates were then spun briefly and transferred to a Tecan microplate reader to monitor changes in fluorescence signal in a kinetic mode. The fluorescence signal was first read at 2 min of reaction and then read every 1.5 min for 10 cycles. The screening for nsp13 inhibitors was done twice, one with a final compound concentration of 1.25 µM and one with 6.25 µM.

The first fluorescence signal at 2 min after reaction start in the screen was used as the initial velocity (V0) to compare nsp13 activity in different wells. MATLAB was used to process data. The initial velocity for each compound is first normalised against the DMSO controls in column 23 of each plate as following:

Because there was a time delay in reading different wells by the microplate reader, reactions in wells that were read later progressed further than wells that were read earlier in each cycle, resulting in higher signals towards the end of a microplate. This positional variation is corrected as following:

For each plate, a linear regression of Normalised_V0 is fitted against well numbers of wells in between column 2 and column 23. A slope value (S) and an intercept value (I) from this linear regression were obtained and used to do the positional correction for the Nth well of a 384-well plate.

For endpoint analysis, fluorescent signals from 6.5 min, 8 min and 9.5 min were taken average when the enzymatic reaction had finished in the DMSO control wells. Compounds that gave more than 10% inhibition in endpoint analysis were considered hits and were collated with hits from initial velocity analysis. Plate 14 from the 1.25 µM screen, plate 1 and 5 from 6.25 µM screen were excluded from data analysis because of faulty handling of the plates. In total, there were 339 hits after initial selection.

For preliminary evaluation of the 339 hits, a kinetic plot in which the y-axis shows fluorescent signals, and the x-axis shows timepoints was drawn for each hit. In addition, signals from 10 wells that were before and after the hit well in a 384-well plate were plotted in the same graph with the hit. Because the wells were close to each other, their reaction time was similar and thus one would expect their curves overlap if there was no inhibition of the reaction. And a real hit would produce lower signal than the neighbouring wells. We manually inspected the curves for all the 339 hits and confirmed that 142 of them gave obvious lower signal than their neighbouring wells.

To be used as a reference for the hit selection method described above, the Z-score was calculated for each compound (Supplementary Figure S3G) as following:

where mean(Corrected_V0) is the average of corrected initial velocity values of all compounds and SD(Corrected_V0) is the standard deviation of corrected initial velocity values of all compounds.

To assess the quality of the screen, the Z-factor was calculated for each plate based on corrected initial velocity values as following:

The average Z-factor for the screen after positional correction is 0.53. This value suggests the screen is good, as described by Zhang et al. (67) . Table 1 , were included for comparison purposes. 

Escape from neutralizing antibodies by SARS-CoV-2 spike protein variants

Genomic evidence for reinfection with SARS-CoV-2: a case study

Viral targets for vaccines against COVID-19

The Impact of Mutations in SARS-CoV-2 Spike on Viral Infectivity and Antigenicity

Drug repositioning: identifying and developing new uses for existing drugs

Rapid repurposing of drugs for COVID-19

Controlled Trial of Ebola Virus Disease Therapeutics

Therapeutic efficacy of the small molecule GS-5734 against Ebola virus in rhesus monkeys

Remdesivir for the Treatment of Covid-19 -Final Report

Mechanism of SARS-CoV-2 polymerase stalling by remdesivir

Remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (2019-nCoV) in vitro

Why are RNA virus mutation rates so damn high?

Emerging SARS-CoV-2 mutation hot spots include a novel RNA-dependent-RNA polymerase variant

Coronavirus biology and replication: implications for SARS-CoV-2

The Architecture of SARS-CoV-2

SF1 and SF2 helicases: family matters

Multiple enzymatic activities associated with severe acute respiratory syndrome coronavirus helicase

The human coronavirus 229E superfamily 1

helicase has RNA and DNA duplex-unwinding activities with 5'-to-3' polarity

The severe acute respiratory syndrome (SARS) coronavirus NTPase/helicase belongs to a distinct class of 5' to 3' viral helicases

Human coronavirus 229E nonstructural protein 13: characterization of duplex-unwinding, nucleoside triphosphatase, and RNA 5'-triphosphatase activities

Structural bases of norovirus RNA dependent RNA polymerase inhibition by novel suramin-related compounds

In silico screening for human norovirus antivirals reveals a novel non-nucleoside inhibitor of the viral polymerase

Synthesis, molecular characterization, and biological activity of novel synthetic derivatives of chromen-4-one in human cancer cells

Combinatorial drug therapy for cancer in the postgenomic era

Architecture of a SARS-CoV-2 mini replication and transcription complex

Identification and prediction of promiscuous aggregating inhibitors among known drugs

Comprehensive mechanistic analysis of hits from high-throughput and docking screens against beta-lactamase

Complementarity between a docking and a high-throughput screen in discovering new cruzain inhibitors

Aggregating behavior of phenolic compounds--a source of false bioassay results?

Human African trypanosomiasis

Suramin Inhibits SARS-CoV-2 Infection in Cell Culture by Interfering with Early Steps of the Replication Cycle

The approved pediatric drug suramin identified as a clinical candidate for the treatment of EV71 infection-suramin inhibits EV71 infection in vitro and in vivo

The Suramin Derivative NF449 Interacts with the 5-fold Vertex of the Enterovirus A71 Capsid to Prevent Virus Attachment to PSGL-1 and Heparan Sulfate

Structurebased inhibition of Norovirus RNA-dependent RNA polymerases

Delivery of suramin as an antiviral agent through liposomal systems

Compassionate Use of Remdesivir for Patients with Severe Covid-19

Drug repurposing screens and synergistic drug-combinations for infectious diseases

A SARS-CoV-2 protein interaction map reveals targets for drug repurposing

Defectiveness of interferon production and of rubella virus interference in a line of African green monkey kidney cells (Vero)

biGBac enables rapid gene assembly for the expression of large multisubunit protein complexes

New baculovirus expression tools for recombinant protein complex production

A Simple Statistical Parameter for Use in Evaluation and Validation of High Throughput Screening Assays

Molecular and biological characterization of human monoclonal antibodies binding to the spike and nucleocapsid proteins of severe acute respiratory syndrome coronavirus

We thank Anne Early for assistance. This work was supported by the Francis Crick Institute, which receives its core funding from Cancer Research UK (FC001066, FC001030), the UK Medical Research Council (, FC001030), and the Wellcome Trust (FC001066, FC001030). This work was also funded by a Wellcome Trust Senior Investigator Award (106252/Z/14/Z) to J.F.X.D. FW and BC have received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement Nos 844211 and 895786. JZ has received funding from a Ph.D. fellowship awarded by Boehringer Ingelheim Fonds.