Correction: Multi-targeting of functional cysteines in multiple conserved SARS-CoV-2 domains by clinically safe Zn-ejectors

Correction for ‘Multi-targeting of functional cysteines in multiple conserved SARS-CoV-2 domains by clinically safe Zn-ejectors’ by Karen Sargsyan et al., Chem. Sci., 2020, 11, 9904–9909, DOI: 10.1039/D0SC02646H.

New infectious viruses causing epidemics/pandemics such as the SARS-CoV-2 require near-term effective and practical strategies to treat virus-infected patients. This is because developing specic antiviral drugs/vaccines takes time and in the interim, lives are lost/disrupted. One short-term strategy is to leverage the non-specicity of some FDA-approved drugs to target critical viral proteins. 1 Here, we present a multi-targeting strategy combining evolutionary (conserved protein domains) and physical (factors controlling Zn 2+ -bound Cys reactivity) principles to identify new drug targets in conserved viral domains and applied it to the SARS-CoV-2. We show that clinically safe Znejector drugs, disulram and ebselen, can target highly conserved Zn 2+ -binding and/or catalytic cysteines ( Fig. 1) in multiple conserved viral domains essential for SARS-CoV-2 replication.
Conserved cysteines and Zn 2+ play crucial roles in viral infections. 2 Cysteines can serve catalytic and/or structural roles (by binding metal ions and forming disulde bridges) in viral enzymes/proteins. Zn 2+ is an essential cofactor of many viral proteins, as it induces protein folding and stabilizes the local structure. 3,4 Viral Zn-sites containing reactive cysteines bound to structural Zn 2+ cations (termed labile Zn-sites) may serve as drug targets, as the Zn 2+ -bound thiolates can react with Zn-ejectors, leading to loss of Zn 2+ and viral protein structure/function. [5][6][7][8] Our previous studies 7,9 had revealed the key factors determining which Zn 2+ -bound cysteines in proteins are reactive: labile Znsites are likely to be Zn-Cys 4 or Zn-Cys 3 His sites lacking hydrogen bonds to any Zn-bound thiolate, which would reduce the Zn-bound thiolate's negative charge and reactivity (ESI Fig. S1 †). We then used these guidelines to identify drug-target proteins containing labile structural Zn-sites. 8,10 To circumvent toxicity due to undesirable targeting of essential human proteins, we had proposed using clinically safe (FDA-approved or in phase II/III clinical trials) Zn-ejecting agents that do not affect crucial host proteins (ESI Table S1 †) to target putative labile Zn-sites in viral proteins. 8 We showed that disulram, an anti-alcoholism drug, can eject Zn 2+ from the predicted labile Zn-Cys 4 site in the hepatitis C virus NS5A protein, inhibiting viral replication, and that inhibition was enhanced when disulram was combined with interferon-a. 8 Following our work, disulram was found to eject Zn 2+ and inhibit replication in other viruses, notably SARS-and MERS-CoV papain-like proteases (PL pro ), 11 which along with the main protease (M pro ) cleave the pp1a and pp1ab replicase polyproteins into 16 nonstructural proteins (nsps). 12 Since SARS-CoV PL pro remained inactivated aer removing unbound disulram, but was reactivated by reductant, disulram may also form a covalent adduct with the catalytic cysteine. 11 The above ndings suggest that a possible strategy to combat infectious coronaviruses with no approved drugs/vaccines is to employ clinically safe Zn-ejector drugs to target multiple essential Zn 2+ -bound and/or catalytic cysteines in conserved viral domains. Coronaviruses belonging to the order Nidovirales are amenable to our proposed strategy since they employ cysteine proteases (M pro , PL pro ) and share conserved core replicase domains. 12 Thus, we analyzed the SARS-CoV-2 genome (GenBank: MN908947.3) comprising 5 0 -methylated cap, genes encoding nonstructural and structural proteins, and 3 0 -polyadenylated tail. We focused on the large replicase polyprotein pp1ab because its cleavage products (nsp7-nsp16) are involved in viral replication. 12 By searching the pp1ab sequence for conserved domains using the Conserved Domain Database, 13 we found 18 such domains (ESI Table S2 †). For each conserved domain found, we searched the Protein Data Bank (PDB) 14 for <3Å structures from other coronaviruses sharing similar function and high sequence identity using BLASTp 15 (ESI Scheme S1 †). We then checked each structure for Zn-(Cys 4 /Cys 3 His) sites, lacking hydrogen bonds to the Zn-bound thiolates. Although nsp12 (6nur, 6nus) and nsp14 (5c8s, 5c8t, 5c8u) structures have Zn-(Cys 4 /Cys 3 His) sites, their poor ($3.1Å) resolution prohibited reliable hydrogen-bond analysis of these Zn-sites.
Putative labile Zn-sites were found in the SARS-CoV structures of PL pro subdomain of nsp3, nsp10 Zn-nger protein, and nsp13 helicase (Table 1). To obtain the corresponding SARS-CoV-2 sequences, we aligned the SARS-CoV PL pro /nsp10/nsp13 and the SARS-CoV-2 pp1ab sequences using BLASTp 15 and obtained excellent alignment (ESI Table S3 †). The SARS-CoV-2 PL pro structure was homology-modeled from the SARS-CoV 4m0w_A structure using MODELLER, 16 whereas the SARS-CoV-2 nsp10 and nsp13 structures were derived from the respective SARS-CoV structures (2xyq_B and 6jyt_B) by point mutations using SCRWL4 (ref. 17) since their sequences differ by only 1-2 residues. These modeled structures conrm the absence of hydrogen bonds to the Zn-bound thiolates (ESI Fig. S2 †). Furthermore, model structures of the SARS-CoV-2 PL pro /M pro with the catalytic cysteine covalently modied by disulram/ ebselen obtained using QM/MM energy minimization show that the active site can accommodate a covalent adduct, thus inhibiting enzyme activity (ESI Fig. S3 †).
To conrm that disulram and ebselen are covalently bound to the cysteines in PL pro and nsp10, the molecular weights (MWs) of PL pro and nsp10 before and aer adding these two zinc-ejectors were measured by mass spectrometry. The PL pro MALTI-TOF mass spectrum (Fig. 2c, top panel) revealed a major peak with a measured MW of 37 120 Da close to PL pro 's calculated MW (37 125 Da). Disulram-treated PL pro (Fig. 2c, middle panel) had an additional peak at 37 269 Da, suggesting half of disulram was bound in PL pro . Ebselen-treated PL pro (Fig. 2c, bottom panel) had additional peaks at 37 394 and 37 669 Da, indicating one and two ebselen molecules were bound, respectively. For nsp10 with a calculated MW of 14 066 Da, the MALTI-TOF mass spectra (Fig. 2d) exhibited additional peaks at 14 202 Da (corresponding to 0.5 disulram-bound nsp10) and 14 339 Da (corresponding to one ebselen-bound nsp10), suggesting that both drugs were covalently bound to cysteines in nsp10. To further verify that the drug was attached to cysteines involved in Zn 2+ -binding in PL pro , disulram-and ebselentreated PL pro were digested by trypsin, and analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS). The MWs of the PL pro peptides, 183  Since disulram and ebselen are in clinical use, we further determined if they could inhibit PL pro by measuring protease activity in cleaving a uorogenic substrate (Dabcy-FTLKGGYAPTKVTE-Edans) in the absence and presence of varying concentrations of each Zn 2+ -ejector drug. We found that disulram and ebselen inhibited PL pro activity with IC 50 of 7.52 AE 2.13 mM and 2.36 AE 0.16 mM, respectively ( Fig. 3a and b). Interestingly, compared to disulram, ebselen displayed slightly stronger inhibition of PL pro activity despite its less potent Zn-ejecting ability (Fig. 2a). In SARS-CoV-2 M pro that lacks a Zn 2+ -site, ebselen (IC 50 ¼ 0.67 AE 0.09 mM) also showed stronger inhibition than disulram (IC 50 ¼ 9.35 AE 0.18 mM). 18 This suggests that ebselen may be more effective in targeting the catalytic cysteine than disulram.
To our knowledge, we are the rst to combine knowledge of conserved coronavirus domains and the key factors controlling Zn-bound cysteine reactivity 9 to identify previously unknown druggable Zn-sites in multiple SARS-CoV-2 domains. The labile Zn-sites discovered in SARS-CoV-2 are attractive drug targets, as they are highly conserved among coronaviruses and play vital structural/catalytic roles in key proteins in the viral life cycle: the Zn-binding ability of PL pro is crucial for structural integrity and protease activity. 19 PL pro not only cleaves the viral polyproteins, but also helps SARS/MERS-CoV to evade the host innate immune response through deubiquitinating and deI-SGylating enzymatic activities. 19,20 The labile Zn-site in nsp10, a crucial cofactor for multiple replicative enzymes such as nsp14 and nsp16, 21 plays an important structural role. 22 The Znbinding domain of nsp13 helicase, which catalyzes dsRNA/ dsDNA unwinding, is vital for helicase activity. 23 Whereas most studies focus on designing inhibitors or repurposing drugs to target a specic viral enzyme/protein such as M pro , 1 our study shows that the same Zn-ejecting drug can attack highly conserved cysteines in multiple viral targets. Disul-ram, used since 1951 with a recommended daily dose of #500 mg, and ebselen are considered to be clinically safe. 24,25 Both may serve as multi-targeting drugs acting at various stages of the virus life cycle: they can target Zn-bound cysteines in PL pro , nsp10, and possibly nsp13, and/or catalytic cysteines in PL pro and M pro enzyme. Crippling both PL pro and M pro enzymes needed to cleave the replicase polyprotein 1a and pp1ab would likely inhibit SARS-CoV-2 replication. Indeed, both disulram and ebselen were found to decrease the number of viral RNA copies at 10 mM concentration in SARS-CoV-2-infected Vero E6 cells and ebselen was further shown to inhibit SARS-CoV-2 with a EC 50 of 4.67 AE 0.80 mM in plaque-reduction assay. 18 Disulram/ebselen may serve as a broad-spectrum anti-viral since the domains containing the labile Zn-sites are highly conserved across several types of coronaviruses: disulram can inhibit SARS-CoV, MERS-CoV, and SARS-CoV-2 PL pro in vitro, whereas most SARS-CoV PL pro inhibitors are inactive against MERS-CoV PL pro . 26 A possible advantage of targeting multiple conserved domains is that the virus has to undergo simultaneous appropriate mutations of the different targeted domains to develop drug resistance. 8 We propose combining disulram/ebselen with other FDA-approved drugs, which have immune-modulatory/anti-inammatory properties and/or anti-viral effect, to potentially inhibit SARS-CoV-2 replication synergistically. To test this possibility, we chose the zinc ionophore, hydroxychloroquine, because it can downregulate pro-inammatory cytokines 27 and can increase the Zn 2+ level inside a cell. 28 Increasing intracellular Zn 2+ concentration has been shown to inhibit SARS-CoV nsp12 RNAdependent RNA polymerase, the core enzyme of a multiprotein replication and transcription complex. 29 We evaluated antiviral synergy between disulram/ebselen and hydroxychloroquine by pretreating Vero E6 cells with the two drugs at various concentrations for 1 h at 37 C, followed by incubation with SARS-CoV-2 for 1 day at 37 C (see Methods). We then determined the anti-SARS-CoV-2 activity of each drug or drug combination using immunouorescence assay to detect SARS-CoV-2 N protein expression (green in Fig. 4a). Consistent with previous results, 18 disulram and ebselen exhibit an estimated IC 50 of 17.5 mM and 23.3 mM, respectively, based on the quantication of SARS-CoV-2 N protein expression (Fig. 4b). The SARS-CoV-2 infection rate in Vero E6 cells treated with a given concentration of hydroxychloroquine and/or disulram/ebselen is shown as the mean and corresponding standard deviation of three replicates in Fig. 4c. Disulram/ebselen combined with hydroxychloroquine exhibited enhanced antiviral effect compared to each drug alone with p values < 0.05. For example, 12.5 mM disulram combined with 5 mM hydroxychloroquine did not affect cell viability, but reduced viral infection compared to disulram alone (p value ¼ 0.007) or hydroxychloroquine alone (p value ¼ 0.014). This provides proofof-concept for combining disulram/ebselen with other safe drugs to synergistically inhibit SARS-CoV-2 by targeting multiple conserved viral regions/pathways.
In summary, this study offers a possible strategy to tackle outbreaks of coronaviruses by leveraging the non-specicity of clinically safe Zn-ejector drugs combined with broad-spectrum antivirals to target multiple conserved domains essential for viral replication. Our general strategy based on evolutionary and physical principles can be used to identify druggable Zn-sites in other non-coronaviruses employing essential cysteines and Zn 2+ in conserved viral domains.

Conflicts of interest
The authors declare no competing nancial interests. The antiviral activities of disulfiram and ebselen against SARS-CoV-2 were determined on Vero E6 cells using immunofluorescence assay to detect SARS-CoV-2 N protein expression (green). (b) Viral infection was quantified by a high-content image analysis system and the average infection rate of no drug treatment was set as 100% for calculation of the 50% inhibitory concentration (IC 50 ). For the 50% cytotoxic concentration (CC 50 ), Vero E6 cells treated with the indicated compound were assayed by Cell Counting Kit-8. IC 50 and CC 50 were calculated by Prism software. (c) The SARS-CoV-2 infection rates in Vero E6 cells treated with hydroxychloroquine (HCQ) plus disulfiram or ebselen were determined as described above, and shown as means and standard deviations (n ¼ 3). The average infection rate of six sets of experiments with no drug treatment was set as 100%. The p values were calculated by student's t test.