Identification and characterization of aurintricarboxylic acid as a potential inhibitor of SARS-CoV-2 PLpro

Abstract

As the global health crisis due to evolution of mutations in SARS-CoV-2 continues, it is important to develop several effective antivirals to control the disease. Targeting papain-like protease (PLpro) of SARS-CoV-2 for drug development is a promising strategy due to its dual role in promoting viral replication and dysregulating host immune responses. Here, we screened a library of compounds to find potential inhibitors of PLpro. We find aurintricarboxylic acid (ATA) inhibits PLpro with K_i and IC₅₀ values of 16 μM and 30 μM, respectively. The binding of ATA to PLpro was further characterized using isothermal titration calorimetry, differential scanning fluorimetry, dynamic light scattering and circular dichroism spectrometry. In vitro assays showed the antiviral potential of ATA with IC₅₀ of 50 μM. In vivo efficacy was studied in Syrian hamsters and the results are being discussed.

1. Introduction

Development of effective antivirals against SARS-CoV-2 is crucial to contain the current COVID-19 pandemic and future coronavirus outbreaks. Some of the important strategies for drug development include targeting essential steps in the viral life cycle, like blocking the interaction of viral spike protein with host ACE2 receptor, inhibiting viral genome replication by viral RNA dependent RNA polymerase (RdRp) and associated replication machinery, and inhibiting two viral proteases (Main protease and Papain like protease) [1], [2], [3]. Papain-like protease (PLpro) of SARS-CoV-2 is a part of a large multi-domain viral protein (Nsp3) and performs the essential role of cleaving viral polyproteins (pp1a/1ab) at three sites to yield mature non-structural proteins, Nsp1, Nsp2, and Nsp3, which are required for virus replication. Besides, PLpro is also implicated in the evasion of host antiviral immune responses through deubiquitination and deISG15ylation of several host/viral proteins hampering the signaling cascades of the innate immune system [4], [5], [6], [7]. Therefore, targeting PLpro for therapeutic intervention will not only suppress the viral replication but also help in preventing the dysregulation of host innate immune responses.

Several potential inhibitors of PLpro have been reported earlier using various assays. The generic drug, 6-thioguanine (6-TG), was reported to inhibit SARS-CoV PLpro with K _i value of 10–20 μM [8]. Recently, 6-TG has also been shown to inhibit SARS-CoV-2 replication in Vero-E6 cells with an EC ₅₀ of ~2 μM [9]. Another FDA approved drug, disulfiram, was shown to inhibit PLpro from SARS-CoV and MERS [10], possibly through different mechanisms. However, clinical efficacy of these two approved drugs is limited. GRL0617 is another well-characterized noncovalent inhibitor, first identified against PLpro from SARS-CoV [11]. It showed IC ₅₀ value of 2.4 μM against SARS-CoV-2 PLpro, however, lacks sufficient stability and potency for further antiviral development [12], [13]. Several modifications of GRL0617 have since been reported and the best analogue (ZN-3-80) had IC ₅₀ value of 0.59 μM [13]. Their in vivo efficacy, however, has not yet been reported. A few structurally diverse molecules, with different scaffold than GRL0617, also show inhibition in low micromolar range [14], [15]. Recently, Armstrong et al., engineered nanobodies that bind to PLpro with nanomolar affinity [16]. Since many potent inhibitors fail due to their in vivo toxicity and efficacy issues, there is need to find novel PLpro inhibitors.

Here, we performed high-throughput biochemical screening to identify novel inhibitors of SARS-CoV-2 PLpro using our optimized assay conditions reported earlier [17]. We found one of the compounds, aurintricarboxylic acid (ATA), inhibits the viral enzyme in the low micromolar range. ATA has been shown to inhibit replication of several viruses through different mechanisms but its binding and inhibition of any viral proteases have not yet been reported. Here, we extensively characterized the interaction of ATA with PLpro using various biophysical methods and its inhibition potential was further explored in vitro and in vivo.

2. Materials and methods

2.1. Cloning, expression, purification and mutagenesis of PLpro

The recombinant SARS-CoV-2 PLpro was cloned, expressed, purified and extensively characterized, and detailed procedures were published earlier [17]. Briefly, the codon optimized nucleotide sequence of PLpro, encoding residues 746-1059 of Nsp3, was synthesized, cloned into pET28a vector, and the protein was expressed in E. coli BL21(DE3) cells. The protein was purified from soluble fraction of the cell lysate using Ni-NTA affinity chromatography, followed by size exclusion chromatography. Purity of the protein was checked on SDS-PAGE (Supplementary Fig. S1a). A single major peak eluting from Superdex 200 increase 10/300 column (GE Healthcare), corresponding to the PLpro monomer (Supplementary Fig. S1b), suggested the protein is pure enough for enzyme inhibition studies. The purified protein, concentrated to 10 mg/ml, was stored at 4 °C in the storage buffer (50 mM HEPES pH 7.5, 100 mM NaCl and 5 mM DTT) for further studies.

For site-directed mutagenesis, oligonucleotide primers were designed using Quick Change Primer Design server (https://www.agilent.com/store/primerDesignProgram.jsp). Two mutagenic primers for each mutation site were annealed to the template plasmid carrying the insert for native PLpro and the entire plasmid was amplified in PCR reaction using Phusion DNA polymerase. The PCR products were treated with the restriction endonuclease (DpnI) at 37 °C for 1 h to digest the template plasmid. The reaction mixture was then transformed into competent cells of E. coli DH5α strain. Plasmids carrying the putative mutations in the insert were isolated and the presence of the mutations was confirmed through DNA sequencing carried out by Eurofins Genomics, India. All the mutant proteins were expressed in BL21(DE3) cells using the procedures similar to the one used for native PLpro.

2.2. High throughput screening

The experiments were performed in 96-well black non-treated plates (Thermo Scientific, Denmark) using excitation and emission filters of 355 nm and 430 nm, respectively in CLARIOstar Plus multi-mode microplate reader (BMG Labtech, Germany). The assay buffer contained 50 mM MES pH 6.5, 100 mM NaCl, 0.5 mM EDTA, 5 mM DTT and 0.1 mg/ml BSA. The stock solutions of all the screened compounds were prepared at 100 mM concentration in DMSO. The initial screening assays were performed in a reaction volume of 50 μl containing 200 nM enzyme, 300 μM test compound and 20 μM fluorogenic tetrapeptide substrate, Z-Leu-Arg-Gly-Gly-AMC. To identify compounds causing interference in the fluorescence signal in the given wavelength range, control wells were set up for each compound containing all the components of the reaction mixture except the enzyme. IC₅₀ value was calculated from the three/four parameters non-linear regression fitting of velocity data for different concentrations of the compounds using GraphPad Prism software (www.graphpad.com). All the plotted data in Fig. 1 A, B are mean values from three independent experiments. The same procedures were followed for determination of IC₅₀ values of ATA against the mutant proteins.

Fig. 1.

Biochemical characterization of PLpro inhibition by ATA (A) Concentration-dependent inhibition of PLpro enzymatic activity by ATA; [PLpro] = 200 nM, [substrate] = 70 μM; Inset: Chemical structure of ATA containing three phenyl rings with -COOH and -OH groups in each ring (B) Enzymatic activity of PLpro at different concentrations of substrate (Z-LRGG-AMC: 50, 250, 500, 750 μM) and ATA (0–500 μM) to estimate K_i. FU: Fluorescence units.

2.3. Isothermal titration calorimetry

The compounds showing inhibition of PLpro enzyme activity were subjected to isothermal titration calorimetry for the binding parameter analysis using Malvern's MicroCal iTC200. The protein concentration was optimized at 25 μM and the compounds were taken at 0.7 mM concentration in 30 mM Tris pH 7.5, 100 mM NaCl and 5 mM DTT. The total no. of injections was set at 19, sample cell temperature at 25 °C, reference power at 5 μcal/s, initial delay of 120 s was given with a stirring speed of 800 rpm. All titrations were performed with an initial injection of 0.4 μl followed by 18 identical injections of 2 μl with a duration of 4 s per injection and a spacing of 120 s between successive injections. The heat of dilution obtained from the titration of compound in buffer were averaged and subtracted from each injection. The interaction parameters were calculated using MicroCal iTC200 analysis software.

2.4. Differential scanning fluorimetry

Differential scanning fluorimetry (DSF) experiments were performed using Bio-Rad CFX96 Real-time PCR instrument. 25 μl of the sample mix in Bio-rad hard-shell 96-well white plate contained 2.5 μM protein, 0.39–200 μM ATA, and 1× SYPRO Orange dye in 50 mM HEPES, pH 7.5. After an initial incubation at 20 °C for 5 min, the temperature was increased to 95 °C at a rate of 1 °C/min. The fluorescence readings were taken at 0.5 °C increments. Fluorescence values for samples containing PLpro and PLpro with various concentrations of ATA were corrected by subtracting the fluorescence values of control wells containing all the corresponding components (buffer, 1× SYPRO Orange dye and various concentrations of ATA) except the protein.

2.5. Dynamic light scattering

The Dynamic light scattering experiments were performed in a low volume disposable sizing cuvette (ZEN0112 of Malvern, UK) on Zetasizer Nano-ZS instrument (Malvern Panalytical Ltd., UK) for PLpro (30 μM) alone and PLpro (30 μM) with ATA (100 μM) in a buffer containing 30 mM Tris pH 7.5, 100 mM NaCl and 5 mM DTT. The samples were equilibrated at 20 °C for 30 s followed by three measurements, with every measurement having 15 runs of 10 s each. The data were analyzed using the software provided with the instrument.

2.6. Circular dichroism spectroscopy

CD spectra were recorded on J-815 CD spectrometer (Jasco) equipped with a Peltier-type thermostatic cell holder at 20 °C using 1 mm optical path length quartz cuvette. The measurements were made at 1 nm data interval, 50 nm/min scan speed in the far UV region (200–260 nm) with three accumulations using 10 μM PLpro and varying ATA concentrations (0–200 μM). Time-course measurements were made using 10 μM protein and 100 μM ATA for 4 h. Both, ATA and PLpro, were prepared in 1× PBS. The data were analyzed using spectra analysis software supplied with the instrument and composition of secondary structure elements were estimated using CDNN 2.1 software [18].

2.7. In silico docking

Crystal structure of PLpro for molecular docking was extracted from Protein Data Bank (PDB id:6wrh) and blind docking was performed using the SeeSAR software suite (from BiosolveIT) taking entire PLpro structure as potential binding site. ATA coordinates were downloaded from PubChem (from National Center for Biotechnology Information). For each binding site, five hundred different poses were generated and docked using FlexX. These poses were scored using Hyde and sorted based on the scores for different binding parameters, such as binding affinity, ligand-lipophilicity efficiency (LLE), ligand efficiency (LE), torsion quality, inter and intramolecular clashes, etc. Based on these scores, the best binding mode of the ligand for each site was selected.

2.8. In vitro toxicity and antiviral assay of ATA

To perform in vitro antiviral assay of ATA against SARS-CoV-2, cytotoxicity of the compound was first assessed in Vero E6 cells. The cells, seeded in a 96-well plate at 80 % confluency, were treated with various concentrations of ATA (up to 1 mM) and the MTT assay was performed after 48 h to evaluate the viability of cells. For the antiviral assay, Vero E6 cells seeded in a 96-well plate at 80 % confluency were infected with SARS-CoV-2 at 0.1 m.o.i. (multiplicity of infection) for 2 h. The inoculum was subsequently aspirated and different concentrations of ATA (25, 75, 100, 200, 300, and 400 μM) in fresh media were added to the cells, followed by the collection of respective supernatants at 24 h post-infection. RNA was isolated from the supernatant and RT-qPCR was performed using primers specific for the viral spike, nucleocapsid, and ORF1a. The mean Ct values obtained from the RT-qPCR analysis was used for estimating the viral RNA copy numbers. IC₅₀ for ATA was determined in GraphPad Prism using four parameters variable slope non-linear regression fitting.

2.9. Evaluation of antiviral potential of ATA in Syrian hamster

The in vivo antiviral potential of ATA was investigated in 24 Syrian hamsters, divided into four groups (one control and three treatment groups) of six animals each. On day zero, all the animals were infected intra-nasally with SARS-CoV-2 (TCID₅₀ = 10⁵). While no treatment was given to the control group, different doses of ATA were given to three treatment groups (A, B, and C) at 15, 30, and 45 mg/kg body weight, respectively. All the animals were monitored daily, and their body weights were recorded. The treatment was initiated on day zero, 4 h post-infection, through oral route (in 100 μl solution). The doses were repeated to the treatment groups every day for four days (day 1–4) and the animals were euthanized on day 5. Throat swabs were taken and lungs were harvested from all the animals for further processing. Viral loads in throat swabs and lung tissues were estimated using RT-qPCR. The left lung was used for viral load estimation and the right lung was used for histopathological evaluations. The significance of differences among groups was calculated using Welch's t-test (one-tailed).

3. Results

3.1. High throughput screening

Enzymatic assays of PLpro were performed using fluorogenic tetrapeptide substrate, Z-Leu-Arg-Gly-Gly-AMC. An in-house library of 350 drug-like compounds, including a set of natural product compounds (listed in Supplementary Table S1), was screened against PLpro using optimized assay conditions reported earlier [17]. To consider only specific inhibitors, 5 mM DTT was included in the assay buffer. To identify potential hits, 6-thioguanine – a known inhibitor of PLpro [8], [9], was taken as a reference and compounds showing a decrease in fluorescence better than 6-thioguanine at 300 μM concentration were selected for further screening. The second round of screening with selected compounds was performed in a concentration-dependent manner where the concentrations of the compounds varied from 0 to 300 μM at a fixed concentration of the substrate (20 μM) and the enzyme (200 nM). Among all the compounds tested, significant inhibition was achieved only by ATA. Concentration-dependent inhibition of PLpro enzymatic activity showed IC ₅₀ for ATA as 30 μM (Fig. 1A). To determine the type of inhibition and the inhibitory constant (Ki), the enzymatic assays were performed at four concentrations of the substrate (50, 250, 500, 750 μM) with varying concentrations of ATA (0 μM–500 μM) (Fig. 1B). The kinetics data were fitted into four models of enzyme inhibition (competitive, non-competitive, uncompetitive, and mixed) using SigmaPlot (Systat Software, San Jose, CA) software. The Akaike Information Criterion corrected (AICc) values were used to find out the plausible mode of inhibition. The kinetic data fitted best in the non-competitive model of enzyme inhibition with Ki value of 16 μM.

3.2. Isothermal titration calorimetry

The energetics of binding of ATA to PLpro was studied using isothermal titration calorimetry (ITC) (Fig. 2A). Thermodynamic data, fitted into one-site binding model, yielded dissociation constant (K _D) of 11.3 μM with enthalpy (ΔH) and entropy (ΔS) changes of −2.47 kcal/mol and 14.3 cal/mol/deg, respectively (Table 1 ). Thermodynamic data show favorable enthalpy and entropy contributions to the overall binding energy (ΔG) suggesting binding of ATA to PLpro comprises hydrophobic and hydrogen bonding interactions. To further confirm the non-competitive mode of binding of ATA to PLpro, ITC experiments were also performed in presence of GRL0617, a known competitive inhibitor of PLpro [11]. First, the enzyme was preincubated with GRL0617 followed by titration using ATA (Fig. 2B), and the thermogram and the derived results were compared with those from titration of ATA without preincubation with GRL0617 (Fig. 2A). And second, the enzyme was preincubated with ATA before titrating with GRL0617 (Fig. 2D) and compared with those without preincubation with ATA (Fig. 2C). In both the cases, there were no significant differences in thermodynamic parameters due to preincubation with the other inhibitor suggesting ATA binds to PLpro non-competitively, at a site different from the GRL0617-binding site (Table 1).

Fig. 2.

Thermodynamic and kinetic parameters of ATA and GRL0617 binding to PLpro. Top panel: Raw data of calorimetric titration showing exothermic heat changes with successive injections; bottom panel: Integrated binding isotherm plotted against molar ratio (ATA: PLpro); ITC experiment showing heat changes upon (A) ATA binding to PLpro and (B) ATA binding to PLpro preincubated with GRL0617 for 2 h. (C) GRL0617 binding to PLpro, (D) GRL0617 binding to PLpro preincubated with ATA for 2 h.

Table 1.

Thermodynamic parameters of interaction of ATA and GRL0617 to PLpro.

S. no.	ITC experiments	N (stoichiometry)	KD (dissociation constant) (M)	ΔH (enthalpy change) (cal/mol)	ΔS (entropy change) (cal/mol/deg)
1.	ATA binding to PLpro	1.31	11.3 × 10−6	−2.469 × 103	14.3
2.	ATA binding to PLpro pre-incubated with GRL0617 for 2 h	1.30	11.3 × 10−6	−2.403 × 103	14.6
3.	GRL0617 binding to PLpro	0.912	2.8 × 10−6	−11.240 × 103	−12.3
4.	GRL0617 binding to PLpro pre-incubated with ATA for 2 h	1.07	2.8 × 10−6	−9.401 × 103	−6.09

3.3. Biophysical characterization of PLpro-ATA interaction

The effect of binding of ATA on the PLpro structure was investigated by various biophysical methods. Differential scanning fluorimetry (DSF) experiments were performed to assess whether attachment of ATA to PLpro would lead to partial unfolding of the protein. The protein unfolding was monitored by recording fluorescence from 1× Sypro orange dye in the sample while heating them from 20 °C to 90 °C. The fluorescence from samples containing PLpro (2.5 μM) alone and PLpro (2.5 μM) with various concentrations of ATA (0.78 μM to 200 μM) were recorded as shown in Fig. 3A. It was found that with increasing concentrations of ATA, there was a consistent decrease in dye fluorescence suggesting stabilizing effects of ATA on PLpro structure. DSF data, therefore, indicates that ATA does not induce any partial unfolding of PLpro, instead it stabilizes the protein structure.

Fig. 3.

Biophysical characterization of PLpro-ATA interaction (A) DSF data for PLpro with different concentrations of ATA; [PLpro] = 2.5 μM. (B) Monitoring particle size distribution during PLpro-ATA interaction using DLS; [PLpro] = 30 μM, [ATA] = 100 μM. CD spectra showing secondary structural changes in PLpro, (C) upon ATA binding; [PLpro] = 10 μM, and (D) during 4 h duration; [PLpro] = 10 μM, [ATA] = 100 μM.

Further, whether ATA binding leads to protein aggregation was explored by dynamic light scattering (DLS), Fig. 3B. It was found that the protein solution was monodisperse with less than 10 nm in size and there was no significant change in the size of particles on addition of ATA even after 90 min. DLS experiments, therefore, suggest that ATA doesn't induce aggregation of PLpro.

Further, effects of ATA binding on the secondary structure of PLpro were investigated using circular dichroism (CD) spectrometer. CD spectra were recorded for PLpro (10 μM) and PLpro (10 μM) with different concentrations of ATA (15–200 μM), and for up to 4 h duration (Fig. 3C, D). The comparative analysis of secondary structural elements in PLpro and PLpro in presence of ATA shows that ATA is not able to completely denature the protein even when ATA concentration is 20 times higher than that of the protein (Supplementary Table S3).

Therefore, DSF, DLS, and CD experiments together suggest that ATA doesn't induce any protein unfolding, aggregation or significant changes in protein secondary structure. Hence, loss in enzymatic activity of PLpro by ATA is not due to protein denaturation or aggregation but likely due to specific interaction of ATA with PLpro. The thermodynamic parameters obtained from ITC experiments (Table 1) show that these interactions involve hydrophobic and hydrogen bonding interactions. Further, enzyme inhibition studies and ITC experiments in the presence of GRL0617 show the mode of interaction of ATA with PLpro is non-competitive.

3.4. Molecular docking and mutational studies

Our attempts to grow co-crystals of SARS-CoV-2-PLpro with ATA didn't succeed. Therefore, probable mode of binding of ATA to PLpro was explored by molecular docking and site-directed mutagenesis. Blind docking showed three probable binding sites of ATA (Sites 1, 2 & 3) with similar binding affinities (Fig. 4 ) and these sites were away from the catalytic site and different from the site (S3/S4) [11] where several known competitive inhibitors bind. The first binding site (Site 1) corresponds to the SUb2 site [19] where N-terminal domain of ISG15 or the second molecule of diubiquitin substrate binds to the enzyme. Interestingly, in one of the reported crystal structures of PLpro with 3, 4-Dihydroxybenzoic acid (PDB id: 7ofu), methyl ester was found to bind at the same site. Further, ATA, being negatively charged, may have affinity for this site as phosphate and chloride ions were found to bind at the site in several crystal structures (PDB id: 6wrh, 7qch,7qci, 7qcj, 7qck, 7qcm, 7ofs,7oft, 7ofu). Other two sites where ATA was found to bind in molecular docking studies were in the grooves between the palm and the thumb domains (Site 2), and the thumb and ubiquitin-like (Ubl1) domains (Site 3). Though, no ligands have been reported to bind PLpro at Site 2, several crystal structures showing ligands bound at Site 3 are there in the PDB (7ofs, 7oft, 7qcl 7qch, 7qcj, 7qck).

Fig. 4.

Probable binding sites of ATA to PLpro by molecular docking studies.

To further explore the sites for ATA binding, residues making hydrogen bonding interactions with ATA in the docked structures were mutated to Alanine (Ala) and inhibition of mutant PLpros by ATA was studied by enzymatic assays using the fluorogenic tetra-peptide substrate (Table 2 , Fig. S2). It was found that when two residues at Site 1 (N128 and H175) are mutated to Ala, the effect on IC₅₀ values for ATA are different. While N128A mutation increased the IC₅₀ value from 30 μM (with native protein) to 69 μM suggesting its potential role in ATA binding, N175A mutation did not alter the IC₅₀ value (31 μM) significantly. It is likely that at Site 1, ATA binds in an orientation different from what was predicted by molecular docking, and doesn't involve interaction with residue H175. For Site 2, side chain of only one residue (R82) was found to make hydrogen bond with ATA, and mutating the residue to Ala increased the IC₅₀ value to 111 μM suggesting R82 is also involved in ATA binding. For Site 3, when either of the two residues (D37 or Y56) are mutated to Ala, the IC₅₀ values for ATA increased to 53 μM suggesting both residues contribute to the binding of ATA to PLpro. Therefore, the mutational studies indicate that ATA binds to PLpro at multiple sites.

Table 2.

IC₅₀ values of PLpro variants.

S. no	Sites	PLpro variants	IC50 (μM)
1.		Native	30
2.	Site 1	N128A	69
3.		H175A	31
4.	Site 2	R82A	111
5.	Site 3	D37A	53
6.		Y56A	53

3.5. In vitro toxicity and antiviral assay of ATA

To explore the in vitro antiviral potential of ATA, cytotoxicity of ATA was first assessed in Vero E6 cells. The cells were treated with 1.0 mM ATA and the MTT assay was performed after 48 h. The results showed 95 % of the cells were viable suggesting ATA is nontoxic to the cells at this concentration. Therefore, antiviral potential of ATA was evaluated at various nontoxic concentrations below 1.0 mM. Vero E6 cells were infected with SARS-CoV-2 and treated with different concentrations of ATA (0, 25, 75, 100, 200, 300, and 400 μM). The viral load after ATA treatment was estimated in terms of viral RNA copy numbers by performing RT-qPCR. There was a sharp decrease in viral RNA copy numbers as compared to the control (no ATA) when concentration of ATA increased from 25 μM to 75 μM (Table 3 ) suggesting the IC₅₀ value of ATA lies between these two concentrations. Four parameters variable slope non-linear regression fitting of the dose-response curve showed IC₅₀ as 50 μM (Fig. 5 ).

Table 3.

Reduction in the viral RNA copy numbers with increasing concentrations of ATA.

Concentrations of ATA (μM)	Viral RNA copy no./ml	% reduction
0	2482	0
25	2404	3.1
75	393	84.2
100	207	91.7
200	193	92.2
300	71	97.1
400	25	99.0

Fig. 5.

Dose-response curve for in vitro IC₅₀ value determination of ATA.

3.6. Evaluation of in vivo antiviral potential of ATA in Syrian hamster

In vivo antiviral potential of ATA was evaluated in Syrian hamsters following a procedure described earlier [20]. ATA was administered orally to the SARS-CoV-2-infected animals in three treatment groups, A, B, and C at 15, 30, and 45 mg/kg body weight, respectively. The schedule of ATA treatment is shown in Fig. 6A.

Fig. 6.

In vivo antiviral assay of ATA. (A) Treatment design in Syrian hamsters. (B) Body weights of individual animals from Day 0–5. Viral RNA levels in (C) throat swabs and (D) lung tissues of SARS-CoV-2-infected animals from the untreated control (six animals) and three treatment groups (six animals each). Horizontal bars represent mean values.

During the observation period of five days, no significant weight loss (Fig. 6B), gross complications, or mortality were observed in any groups (i.e., treatment or untreated control groups). RT-qPCR showed reduction of virus RNA copy numbers in the throat swab samples for the ATA-treated groups as compared to the untreated control (Fig. 6C). The significance of the reduction in viral load in the treatment groups as compared to the untreated control was calculated using one-tailed Welch's t-test. Treatment groups A & B showed 0.4-log reduction (p-value: 0.0592 and 0.0580, respectively) while group C showed 1-log reduction (p-value: 0.0439) in the virus RNA levels as compared to the untreated control.

In lung tissues, however, there was no significant reduction of the virus RNA levels in the treatment groups as compared to the untreated control (Fig. 6D). Also, no noticeable difference in the lung pathology was observed among the untreated and ATA treated groups (Fig. 7A, B).

Fig. 7.

Lung pathology. (A) Representative images of lungs and (B) haematoxylin and eosin (H&E) stained lung tissues suggesting no significant differences in lung pathology among the untreated control and ATA treated animals.

4. Discussion

ATA has been shown to possess antiviral activity against a diverse group of viruses, including HIV [21], [22], [23], Enterovirus 71 [24], influenza A & B [25], [26], Vaccinia [27], Hepatitis C [28], [29], [30], Zika [31], and SARS-CoV [32]. In the case of SARS-CoV, it was initially speculated that ATA would inhibit RdRp [33], but no experimental evidences have since been reported. Here we find that ATA inhibits enzymatic activity of viral PLpro. Our further exploration showed that aurin, the parent compound of ATA without carboxylate groups, didn't inhibit PLpro suggesting a crucial role of carboxylates of ATA in promoting binding to PLpro. Molecular docking studies show three probable binding sites of ATA in PLpro. Though ATA is known to gradually form oligomers in solution, we find no significant difference in PLpro inhibition activities by the freshly prepared ATA and the oligomeric ATA (ATA solution left at 277 K for a week). The CD experiments further show no denaturation of the enzyme by ATA suggesting specific inhibition of PLpro.

During the course of our work, two reports showing antiviral activity of ATA against other targets of SARS-CoV-2 have also been published [34], [35]. David et al. show binding of ATA to the receptor binding domain (RBD) of viral spike protein leading to the inhibition of RBD-ACE2 interactions [34], while Canal et al. report the inhibition of exoribonuclease activity of Nsp10/14 by ATA [35]. The fact that ATA inhibits multiple SARS-CoV-2 targets makes it a unique molecule having potential to hamper different steps of viral life cycle, and therefore, a promising candidate for therapeutic development.

In vivo evaluation of the antiviral potential of ATA in SARS-CoV-2-infected Syrian hamsters show a significant reduction in viral load from the oral swabs but no reduction in the viral load of lung tissues. The reasons for different outcomes from lung tissues and oral swabs could be due to the fact that ATA is a highly polar molecule with limited cell permeability. Therefore, when ATA is administered orally, it might not reach in effective concentration in the lungs to kill the virus but reduction in viral load in the throat indicates its effectiveness in killing the virus when in direct contact. Poor cell permeability, however, would make ATA significantly less toxic and it may be developed as an oral gargle or a nasal cleaning solution to reduce the viral load, and thereby, decreasing person-to-person transmission/infection. Further, administration of ATA in Syrian hamsters through a different route, such as nebulization or nasal inhalation may increase its availability in the lung tissues and can reduce viral shedding and spread of infection there. Though the cell culture and animal model studies show limited toxicity of ATA [31], [36], [37], further studies need to be performed to assess its safety and effectiveness.

5. Conclusions

Here, we screened a library of compounds using high-throughput biochemical assay developed earlier [17] to find potential inhibitors of SARS-CoV-2 PLpro. We identified ATA as a novel inhibitor of PLpro and evaluated its binding characteristics using various biophysical techniques. We also assessed its in vitro antiviral potential against SARS-CoV-2. Our in vivo studies show that ATA reduces viral load in the throat when administered orally to Syrian hamsters, however, no improvement in lung pathology was observed. Its therapeutic efficacy through other routes of administration needs to be explored.

CRediT authorship contribution statement

Rimanshee Arya: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Visualization, Writing – original draft, Writing – review & editing. Vishal Prashar: Conceptualization, Methodology, Validation, Formal analysis, Writing – original draft, Writing – review & editing, Supervision. Mukesh Kumar: Conceptualization, Methodology, Validation, Formal analysis, Writing – original draft, Writing – review & editing, Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary Information

Data availability

Data will be made available on request.