Advertisement

High throughput screening for drugs that inhibit 3C-like protease in SARS-CoV-2

Open AccessPublished:January 13, 2023DOI:https://doi.org/10.1016/j.slasd.2023.01.001

      Abstract

      The SARS coronavirus 2 (SARS-CoV-2) pandemic remains a major problem in many parts of the world and infection rates remain at extremely high levels. This high prevalence drives the continued emergence of new variants, and possibly ones that are more vaccine-resistant and that can drive infections even in highly vaccinated populations. The high rate of variant evolution makes clear the need for new therapeutics that can be clinically applied to minimize or eliminate the effects of COVID-19. With a hurdle of 10 years, on average, for first in class small molecule therapeutics to achieve FDA approval, the fastest way to identify therapeutics is by drug repurposing. To this end, we developed a high throughput cell-based screen that incorporates the essential viral 3C-like protease and its peptide cleavage site into a luciferase complementation assay to evaluate the efficacy of known drugs encompassing approximately 15,000 clinical-stage or FDA-approved small molecules. Confirmed inhibitors were also tested to determine their cytotoxic properties. Medicinal chemistry efforts to optimize the hits identified Tranilast as a potential lead. Here, we report the rapid screening and identification of potentially relevant drugs that exhibit selective inhibition of the SARS-CoV-2 viral 3C-like protease.

      Keywords

      Introduction

      Nearly three years into the COVID-19 pandemic, SARS-CoV-2 continues to spread worldwide, and highly contagious variants continue to emerge throughout the world, leading to death rates at times as high as 400 per day in the USA [
      • Lopez Bernal J.
      • Andrews N.
      • Gower C.;
      • et al.
      Effectiveness of Covid-19 vaccines against the B.1.617.2 (Delta) variant.
      ]. Many deaths are driven by respiratory and cardiac issues, especially in elderly populations or those with pre-existing conditions. Multiple safe and effective vaccines were developed and administered at an unprecedented pace; however, global vaccine-uptake has lagged, and widespread immunity seems impossible to achieve. Taken together, emerging variants and continued spread of the virus drive the need for therapeutics which can mitigate the effects of infection. Identifying antiviral therapeutics using approved drugs is one way to rapidly defend ourselves from the severity of these viral infections while vaccine distribution continues worldwide.
      Two of the most well-characterized drug targets in coronaviruses are the proteases 3CLpro (also called Mpro) and PLpro, the papain-like protease [
      • Baez-Santos Y.M.
      • St John S.E.
      • Mesecar A.D.
      The SARS-coronavirus papain-like protease: structure, function and inhibition by designed antiviral compounds.
      ,
      • Zhang L.
      • Lin D.
      • Sun X.
      • et al.
      Crystal structure of SARS-CoV-2 main protease provides a basis for design of improved alpha-ketoamide inhibitors.
      ]. Both enzymes are essential for processing the polyproteins that are translated from the viral RNA. We previously published on the development of a cell-based high throughput screen to identify inhibitors of the PLpro enzyme [
      • Smith E.
      • Davis-Gardner M.E.
      • Garcia-Ordonez R.D.
      • et al.
      High-throughput screening for drugs that inhibit papain-like protease in SARS-CoV-2.
      ]. 3CLpro is a 33.8kDa enzyme that functions by digesting the polyprotein at 11 conserved sites [
      • Jin Z.
      • Du X.
      • Xu Y.
      • et al.
      Structure of M(pro) from SARS-CoV-2 and discovery of its inhibitors.
      ]. Inhibitors of this protease would limit viral replication, and therefore limit infection in patients. At the beginning of the pandemic, and at the time of this HTS campaign in early 2020, there were no drugs on the market which targeted either 3CLpro or PLpro. Lopinavir and ritonavir were the first drugs used in clinical trials to treat COVID-19 targeting 3CLpro [
      • Cui W.
      • Yang K.
      • Yang H.
      Recent progress in the drug development targeting SARS-CoV-2 main protease as treatment for COVID-19.
      ]. These drugs, which are used in combination therapy for patients of HIV were tested for the treatment of COVID-19. Unfortunately, little to no effect was seen in the clinical trials. Fortunately, 3CLpro inhibitors such as PF-07321332 have recently been identified that have clinical efficacy and appear to be effective to treat the COVID-19 infections [
      • Owen D.R.
      • Allerton C.M.N.
      • Anderson A.S.
      • et al.
      An oral SARS-CoV-2 M(pro) inhibitor clinical candidate for the treatment of COVID-19.
      ]. Now, this and other inhibitors have been found using alternative strategies and screening efforts, all being done since our experiments were conducted [
      • Moghadasi S.A.
      • Esler M.A.
      • Otsuka Y.
      • et al.
      Gain-of-signal assays for probing inhibition of SARS-CoV-2 Mpro/3CLproin living cells.
      ,
      • Dey-Rao R.
      • Smith G.R.
      • Timilsina U.
      • et al.
      A fluorescence-based, gain-of-signal, live cell system to evaluate SARS-CoV-2 main protease inhibition.
      ,
      • Rawson J.M.O.
      • Duchon A.
      • Nikolaitchik O.A.
      • et al.
      Development of a cell-based luciferase complementation assay for identification of SARS-CoV-2 3CLpro inhibitors.
      ,
      • Rothan H.A.
      • Teoh T.C.
      Cell-based high-throughput screening protocol for discovering antiviral inhibitors against SARS-COV-2 main protease (3CLpro).
      ,
      • Wen W.
      • Chen C.
      • Tang J.
      • et al.
      Efficacy and safety of three new oral antiviral treatment (molnupiravir, fluvoxamine and Paxlovid) for COVID-19:a meta-analysis.
      ]. However, continued evaluation for broad utility and escape mutants is ongoing and hence, multiple thereapeutics are needed, setting the stage for our discovery efforts targeting 3CLpro.
      In this study, we aim to identify drugs that specifically inhibit the 3CLpro enzyme. This was done using transient expression system expressing 3CLpro coupled to a luciferase-based reporter complementation assay.
      We have implemented this assay principle to run a 1536-well high throughput assay, testing approximately 15,000 compounds inclusive of 3 libraries of compounds that have either some prior clinical testing or were molecularly docked versus the 3CLpro enzyme. At the completion of the campaign, we found 18 compounds that had selectivity for inhibiting the 3CLpro enzyme, without causing cytotoxicity. These compounds were subjected to various secondary assays to show mode of action.

      Materials and methods

      Plasmids

      Plasmids encoding a constitutively active Renilla luciferase (RLuc), and cleavage-activated Firefly luciferase (FLuc) reporters were previously described [
      • Smith E.
      • Davis-Gardner M.E.
      • Garcia-Ordonez R.D.
      • et al.
      High-throughput screening for drugs that inhibit papain-like protease in SARS-CoV-2.
      ]. To construct a SARS-CoV-2 3CLpro specific reporter, a codon-optimized gene fragment encoding the DnaE intein, C-terminal FLuc fragment, cleavage peptide, and N-terminal FLuc fragment was cloned into the CMVR expression vector between NotI and BamHI restriction sites. The cleavage peptide sequence (amino acids AVLQ↓SGFR) was derived from the native sequence of the nsp4-nsp5 junction from the Wuhan-IVDC-HB-01-2019 EPI_ISL_402119 isolate. This cleavage peptide is identical to that of SARS-CoV-1; therefore, the same reporter could be used for both viruses. The coding sequence for 3CLpro with a C-terminal hemagglutinin (HA) tag was synthesized and cloned into the CMVR expression vector from the same reference sequence. Analogous FLuc reporter constructs were synthesized for Polio 3Cpro (accession KX162707.1, peptide: AKVQ↓GPGFD), and the HIV protease (matrix/capsid peptide: VSQNY↓PIVQ) as controls. Corresponding HA-tagged SARS-CoV-1 3CLpro (Urbani strain, accession AY278741_1), Poliovirus 3Cpro, and HIV protease constructs were also cloned in the CMVR backbone from the same reference sequences. Plasmids were amplified and purified from DH5α Escherichia coli using Invitrogen Maxiprep kits (Invitrogen, Waltham, MA) and fully sequenced to confirm the correct sequence.

      MaxCyte transient transfection

      The MaxCyte transfection system was chosen over lipid based methods due to its superior scalability and affordability [
      • Steger K.
      • Brady J.
      • Wang W.
      • et al.
      CHO-S antibody titers >1 gram/liter using flow electroporation-mediated transient gene expression followed by rapid migration to high-yield stable cell lines.
      ]. Briefly, 293T cells were grown in DMEM supplemented with 10% HI FBS and 1% Anti-anti (all media reagents from Life Technologies). At 70–90% confluent, the 293T cells are harvested and resuspended in MaxCyte Electroporation Buffer at 1e8cells/mL. DNA is added to the cells at a total concentration of 200µg/ml in the following ratios: 27% 3CLpro or Empty Vector for high control cells, 67% FLUC-Reporter-3CLpro, 7% Renilla Plasmid (may be used for built-in cytotoxicity analysis, but we did not). The cells are electroporated using MaxCyte cassettes and the MaxCyte device per manufacturer's instructions. The cells are incubated for 20 min prior to seeding in flasks for a 4-h incubation. The cells were harvested and stored in liquid nitrogen to be used during HTS.

      3CLpro 1536 well luciferase assay

      The 3CLpro and Empty vector cells were thawed and counted. Compounds were pre-spotted onto fresh assay plates with either 5 nL (for 10 mM stocks of ReFRAME), or at 20 nL (for 1 mM or 2.5 mM stocks of Pathogen Box or Target Mol). The cells were seeded at 2500 cells/well or 5e5cells/mL in 293T growth media using a BioRaptr FRD at 5 µL /well. The plates were briefly spun at 1000 RPM and incubated for 48 h at 37C, 5% CO2, 95% RH. After a 48-hincubation, the plates were removed from the incubator and allowed to equilibrate to room temperature for 15 min. ONE-Glo (Promega, Madison, WI) luciferase reagent was added at 5 µL /well with the BioRaptr FRD and the plates were again briefly spun. After a 10-min incubation at room temperature, the luminescence was measured using a ViewLux (PerkinElmer, Waltham, MA) for 30 s. The high control was Empty Vector + FLUC wells, and the low control and data wells have 3CLpro+FLUC + compound or vehicle (DMSO). The protocol can be found in Table 1.
      Table 1D. A table representing the EC50s of the 4 compounds selected from the HTS campaign.
      OrderStepConditionComments
      1Maxcyte transfect and freeze down cells3e7/mlCells are transfected with 200 µg/ml total DNA Low control and sample field = 3CL:PRO+FLUC High control cells = EV+FLUC DNA Ratio 27% 3CLPRO (or EV), 67% FLUC, 7% RLUC
      2Pre-spotted compoundsEither 5 or 20 nlCompounds acoustically dispensed in Corning Solid white TC plates
      3Thaw and seed cells5 µl2.5K/well in growth media (DMEM + 10% HI FBS + 1% Anti-Anti)
      4Incubation48 h37C + 5% CO2
      5Incubation15 minRoom Temperature
      6Add One Glo5 µl/wellLuciferase reagent, spin
      7Incubation10 minRoom Temperature in the dark
      8ReadLuminescenceViewLux

      3CLpro 1536 well cell titer Glo cytotoxicity luciferase assay

      The same cells from the primary assay were plated and incubated the same as the primary 3CLpro assay. Instead of the ONE-Glo reagent, Cell Titer Glo (Promega, Madison, WI) luciferase reagent was added at 5 µL /well with the BioRaptr FRD and the plates were again briefly spun. After a 10-min incubation at room temperature, the luminescence was measured using a ViewLux (PerkinElmer, Waltham, MA) for 30 s. The high control wells contained no cells, and the low control and data wells have 3CLpro+FLUC + compound or vehicle (DMSO).

      Post HTS confirmation assay

      Following the completion of screening all three libraries the most active and selective drugs were subjected to testing under the following conditions. HEK293T cells were transiently transfected in 6-well plates using jetPRIME transfection reagent (Polyplus, Illkirch-Graffenstaden, France) according to manufacturer's instructions at the same ratios used in the MaxCyte transfection procedure. After 4 hours, transfection complexes were removed, and cells were reseeded into 96-well plates containing compounds at a density of 20,000 cells per well. Plates were then incubated at 37 °C for 48 h. Firefly and renilla luciferase luminescence were detected using Promega Dual Glo kit according to manufacturer's instructions. This procedure was followed using the SARS-CoV-2 3CLpro, SARS-CoV-2 3CLpro, Polio 3Cpro, and HIVpro reporter systems using plasmids with their analogous peptides based on the details referenced in the plasmid methods.

      3CLpro western blot analysis

      HEK293T cells were transfected with jetPRIME (Polyplus, Illkirch-Graffenstaden, France) and treated as described above. 48 hours post-transfection, cells were lysed with RIPA buffer and blotted for FLuc (Abcam, Boston, MA), RLuc (Thermofisher), HA (Millipore Sigma), and actin (Abcam, Boston, MA) MG132 (Biomol International, Plymouth Meeting, PA).

      3CLpro biochemical inhibition assay

      His-SUMO SARS-CoV-2-PLpro (1564-1877) Expression and purification

      SARS-CoV-2 [Severe acute respiratory syndrome coronavirus 2]-PLpro (1564−1877, MN908947.3) amino acid sequence was codon optimized for E. coli expression, subcloned, and sequence verified (GenScript) into pE-SUMO-pro AMP vector (LifeSensors). Vector was transformed into One Shot BL21(DE3) competent cells (Thermo Scientific) and plated onto LB-AMP plates (InvivoGen). Transformants (7 colonies) were inoculated in 100 mL Terrific Broth (TB) media supplemented with 50 µg/mL Carbenicillin and incubated overnight at 37 °C with shaking to saturation (OD600 ≥2). Overnight culture (∼1:50 dilution) was used to inoculate fresh TB media supplemented with 50 µg/mL Carbenicillin. 3 L culture was incubated at 37 °C with shaking to OD600 ∼0.6) and induced by adding IPTG to a final concentration of 0.5 mM and cultured for an additional 24 h at 20 °C with shaking. Cells were harvested by centrifugation and cell pellet was stored at −80 °C. Cell pellet was thawed on ice for 15 min and resuspended in Lysis Buffer (50 mM HEPES pH=8.0, 500 mM NaCl, 10 mM Imidazole, 10% Glycerol, DNase I (5µg/mL) and 1X SigmaFast protease inhibitor (Sigma) at 5 mL per gram of wet weight. Cells were lysed using a French Press (Avestin) following manufacturer instructions. Cell lysate was centrifuged at 51,428 x g for 60 min at 4 °C. Using an ÄKTApurifier system maintained at 4 °C, clear supernatant was loaded into a Ni-NTA Superflow column (Qiagen) previously equilibrated with 10 column volumes of Ni-Buffer A (50 mM HEPES pH=8.0, 500 mM NaCl, 10 mM Imidazole, 10% Glycerol). The column was washed with same buffer until the A280 returned to baseline value. Bound protein was eluted with a linear gradient of Ni-Buffer B (50 mM HEPES pH=8.0, 500 mM NaCl, 250 mM Imidazole, 10% Glycerol) and 1 mL fractions were collected. Samples of eluted protein were analyzed by SDS-Page gel (BioRad) and Odyssey Western Blot system (probed with anti-SUMO-tag (Abcam Rabbit Anti-Smt3 antibody). Confirmed fractions were pooled, concentrated and buffer exchanged with Storage Buffer (50 mM HEPES pH=8.0, 500 mM NaCl, 10% Glycerol) to near 190 µM using Amicon Ultra-15 10 kDa units (Thermo Scientific). Final concentration was ∼9.0 mg/mL and purity estimated >80%

      Fluorescence based SARS-CoV-2 PLpro enzymatic assay

      In order to determine the IC50 values for the lead inhibitors in the SARS2 target based assay a modified inhibition assay (21 µL), from previously published method, was performed in triplicate in a 384-well plate format [
      • Baez-Santos Y.M.
      • Barraza S.J.
      • Wilson M.W.
      • et al.
      X-ray structural and biological evaluation of a series of potent and highly selective inhibitors of hµMan coronavirus papain-like proteases.
      ]. After protein, peptide and buffer optimizations, the final His-SUMO-SARS-CoV-2 PLpro (1564−1877) enzyme concentration was 0.15 µM. The assay was performed at 25 °C using HEPES pH=7.4 as the assay buffer and the enzyme activity was monitored by measuring the PLpro mediated release of 7-amino-4-methylcoµmarin (AMC) from the 25 µM ZRLRGG-AMC peptide substrate (Bachem), using the EnVision multimode plate reader (PerkinElmer) with umbelliferon filters (Excitation wavelength of 360 nm (with 40 nm bandwidth) and emission wavelength of 460 nm (with 50 nm bandwidth). Briefly, using a HiBase black PS 30 µL 384 wells plate (Greiner Bio-One), 7 µL of 3X concentration His-SUMO SARS-CoV-2 PLpro (1564−1877) enzyme and 7 µL of 3X concentration inhibitor (Final 12 points 1:3 dose response starting at 30 µM) were mixed and incubated for 30 min at 25 °C before adding/mixing 7 µL of 25 µM ZRLRGG-AMC peptide. All reagents were dissolved in assay buffer. Plates were incubated for 30 min at 25 °C before fluorescence was measured.
      Following reader acquisition, we normalized all data using the average raw data per test drug conc minus the average raw of the DMSO control divided by the average raw data for the highest concentration of Disulfiram minus the average raw of DMSO)*100. IC50s were derived using an unconstrained 4 parameter fit with GraphPad Prism.
      In this assay we validated experimental performance using recently identified inhibitors of PLpro enzyme activity, Disulfiram and GRL-0617, with an anticipated IC50 of 2 µM and 0.7 µM respectively.

      Screening libraries

      ReFRAME library

      CALIBR formerly partnered with the Bill and Melinda Gates Foundation to form an integrated platform of drug candidates. As part of this the ReFRAME collection was born, which contains greater than 13,000 purchased or resynthesized FDA-approved/registered drugs (∼40%), as well as investigational new drugs currently or previously in any phase of clinical development (∼60%).

      Pathogen box library

      The Pathogen Box library is a 400 diverse, drug-like molecules active against neglected diseases of interest provided by the Medicines for Malaria Venture.

      TargetMol library

      The TargetMol company performed CADD in silico docking using the Swiss-Model Homology Modelling process to generate reliable protein models or 3D protein structures of Receptor Binding Domain (RBD) of spike protein, ACE2, viral papain like protease (Plpro) and main protease (3CLpro, also named 3-chymotrypsin-like protease). Based on these protein structures, TargetMol selected 474 top-ranked based on docked molecules and virtually screening against 15,376 compound structures.

      Tranilast and its analogs

      Initial screening indicated that tranilast, (E)-2-(3-(3,4-dimethoxyphenyl)acrylamido)benzoic acid, is a modestly potent and nontoxic 3CLpro inhibitor, with EC50 values ranging from 38-88 µM in different runs. Many analogs of tranilast are commercially available and thus 21 analogs were purchased or synthesized by amide coupling (see supplementary Fig. 2, panels B & C). 9 compounds were found to be modestly more potent that tranilast, though none consistently displayed a sub-micromolar EC50 value. Trends noted are that 8 of the 9 most potent compounds bear tranilast's aniline o-CO2H group. Saturation of the double bond reduces activity, suggesting covalent engagement of 3CLpro. Adding a CN group (making the acrylamide covalent but reversible) does not dramatically affect activity. Replacing the aniline with saturated rings reduces activity. Some analogs that displayed EC50 values lower than tranilast had Emax values less than 50% (most coumarin-like analogs, for example) or had concerns with low solubility in the assay, as noted. Perhaps the most promising tranilast analog seen was SR-32229, in which the methoxy groups of tranilast are replaced with chlorine atoms. This compound has low-µM affinity (Supplementary Fig. 2, panel D, red line), intermediate between the positive control disulfiram (green line) and tranilast (blue line), and with Emax ∼100% of tranilast.

      Screening data acquisition, normalization, representation, and analysis

      All data files were uploaded into the Scripps institutional HTS database (Symyx Technologies, Santa Clara, CA) for plate QC and hit identification. Activity for each well was normalized on a per-plate basis using the following equation:
      %activity=100×(TestwellMedianLowControlMedianHighControlMedianLowControl)
      (1)


      Where “High Control” represent wells containing cells transfected with empty vector + Fluc; while “Low Control” represents wells containing DMSO and cells transfected with the 3CLpro + Fluc and finally the “Data Wells” contain the same with test compounds. The Z′ and S:B were calculated using the High Control and Low Control wells. In each case, a Z′ value greater than 0.5 was required for a plate to be consider acceptable [
      • Zhang J.H.
      • Chung T.D.
      • Oldenburg K.R.
      A simple statistical parameter for use in evaluation and validation of high throughput screening assays.
      ].
      Thus, in this assay, by design, inhibitors of 3CLpro will decrease the luminescence as compared to untreated cells. The high throughput assay design can be found in Fig. 1.
      Fig 1:
      Fig. 1Schematic of Assay Design. A. Schematic of the HA-tagged 3CLpro protein encoded by SARS-CoV-2 nsp5 . B. Schematic of the FLuc reporter construct. C- and N-terminal portions of Firefly luciferase gene are separated by a target peptide containing the cleavage sequence from the junction of nsp4/5. N- and C-terminal DnaE inteins assist in dimerization once cleaved (Q is the cleavage site). C. Schematic of the assay. In the presence of 3CLpro, the target peptide is cleaved and FLuc domains dimerize and are catalytically active.

      1536-well format assay optimization

      Due to the excellent “out of the box” performance with Z′s consistently greater than 0.5, only limited optimization was needed. Any changes that were made were limited to the transfection method, creating a freezer ready assay and cell titer upon seeding for the assay. Assay optimization was done to scale the production of transiently transfected cells at a level compatible with dozens of 1536 well plates, which were frozen in bulk for later use in the HTS assay. A head-to-head comparison of the JetPRIME system to MaxCyte was done and results were similar if not better using the MaxCyte. Fresh vs frozen transient cultured cells were also compared again with no deleterious effect on Z′s. Based on Z′ analysis using the controls and criteria as defined above, we chose the best parameter to proceed with including 2500 cells per well.

      Results

      Primary HTS

      The first step of the HTS campaign was screening the 3CLpro inhibitor assay against the ReFRAME library [
      • Janes J.
      • Young M.E.
      • Chen E.
      • et al.
      The ReFRAME library as a comprehensive drug repurposing library and its application to the treatment of cryptosporidiosis.
      ]. In this assay, 13,135 compounds were tested at a single concentration in singlicate at a final nominal concentration 10 µM. Raw assay data was imported into Scripps’ corporate database and subsequently analyzed using Symyx software. Activity of each compound was calculated on a per-plate basis using the equation shown in the methods.
      Assay performance was excellent with an average Z′ of 0.80±0.05 and an average signal-to-background ratio (S:B) of 45.93±4.14 (n=12 plates). A summary of the results of the primary screening assay are shown in Fig. 2 and Table 2. A mathematical algorithm was used to determine active compounds. Here we applied what we call a “DMSO Cutoff”. In this case the average activity of all DMSO treated wells from multiple interleaved DMSO only plates, plus 3 times the standard deviation value for the same set of data was used as a cutoff parameter, i.e., any compound that exhibited greater percent activation than the cutoff parameter was declared active. Using this “DMSO Cutoff” criteria of 20.0% the primary assay yielded 1028 active compounds (“hits”). We compared these hits to the PLpro confirmed actives, 202 in total, from our previous HTS efforts and which reduce the number of compounds of interest to 862. The ReFrame collection is in limit supply, and as such, limited numbers of molecule are allowed per regulation for retest. Therefore, we assessed the average +5SD of all compounds tested, selected hits above that cut-off and compared them to the 862 actives described above. The final list numbered 418 compounds which was accepted for resupply by CALIBR.
      Fig 2:
      Fig. 23CLpro ReFRAME Primary HTS assay results. Graphed is a single point scatterplot of all 13,135 compounds tested. Each dot graphed represents the activity result of a well containing test compound (black dots) or controls (red, violet, and green dots).
      Table 2HTS campaign.
      LibraryStageConc.# Samples# Replicates# PlatesZ'S:BHit Cutoff# Hits
      ReFramePrimary10 µM13,1351120.80±0.0545.93±4.1420.0%1028
      Confirmation10 µM418310.72±0.0433.32±2.4831.1%75
      Counterscreen10 µM418330.90±0.0227.81±0.1420.0%143
      Titration20 µM233 1:3 Dilution30.71±0.0826.38±2.40IC50<10 µM20
      Titration CS20 µM233 1:3 Dilution30.83±0.0330.11±1.43IC50<10 µM2
      Pathogen BoxConfirmation4 µM1495390.76±0.0854.10±7.2028.2%50
      +Target MolTitration10 µM503 1:3 Dilution30.72±0.0533.63±3.73IC50<4 µM27
      Titration CS10 µM503 1:3 Dilution30.92±0.0224.79±0.82IC50<4 µM5

      Secondary assays

      Confirmation screen

      After completion of cherry-picking and pre-spotting, the confirmation 3CLpro assay used the same reagents and detection system as the primary screening assay, testing each of the 418 compounds at a single concentration (nominally 10 µM) in triplicate. Assay performance of the 3CLpro HTS confirmation assay was consistent with previous experiments with an average Z′ of 0.72 and an average signal-to-background ratio (S:B) of 33.32. Again, we ascertained a “DMSO Cutoff,” and in this case 31.1% activity was used to obtain 75 hits or a 17.94% hit rate. A summary of the results of the confirmation is in Table 2.

      Cytotoxicity counterscreen

      The counterscreen assay was run as described above on the same set of 418 pre-spotted in triplicate at a single concentration (nominally 10 µM). Assay performance of the cytotoxicity counterscreen was excellent with an average Z′ of 0.90 and an average S:B of 27.81. With a “DMSO cutoff” implemented, cytotoxic compounds were found with activity above 20.0% and of the 418 compounds 143 were found to be cytotoxic. The results can also be found in Table 2. After comparison of the actives found in the confirmation and counterscreen, it was determined that 23 compounds selectively inhibit the 3CLpro without considerably affecting the cytotoxicity.

      Concentration response assay

      After completion of cherry-picking and pre-spotting, the concentration response assay used the same reagents and detection system as the primary screening assay but tested each of the 23 compounds in triplicate with a 10 point 1:3 dilution series. Assay performance was good with an average Z′ of 0.71±0.08 and an average signal-to-background ratio (S:B) of 26.38±2.40 (n=3 plates). For each test compound, percent activation was plotted against compound concentration. A four-parameter equation describing a sigmoidal dose-response curve was then fitted with adjustable baseline using Assay Explorer software (Symyx Technologies Inc.). The reported IC50 values were generated from fitted curves by solving for the X-intercept value at the 50% activation level of the Y-intercept value. The following rule was used to declare a compound as “active” or “inactive”: Compounds with an IC50 greater than 10 µM were considered inactive. Compounds with an IC50 equal to or less than 10 micromolar were considered active. Of those, 20 compounds were active in the 3CLPRO assay.

      Cytotoxicity concentration response counterscreen

      The counterscreen assay performance of the 3CLPRO INH cytotoxicity assay was good with an average Z′ of 0.83±0.03 and an average S:B of 30.11±1.43 (n=3 plates). 2 compounds showed IC50<10 µM. The data from the 3CLpro and the cytotoxicity counterscreen can be seen in Table 2. The results show that 18 compounds showed selective inhibition of the 3CLpro without causing cytotoxicity. All IC50 data has been deposited in the ReFRAME database under the following link (https://reframedb.org/assays/A00486).

      Additional libraries tested

      To expand on the limited number of potentially selective hits identified above we sought to test other libraries that would allow us to diversify our panel of test inhibitors using drugs or drug like molecules that had been molecularly docked to the 3CLpro or PLpro target or were potentially useful as known pathogen inhibitors. Thus, we tested both the Target Mol and Pathogen Box libraries which were also screened in the same formats as the ReFRAME library. The data from these libraries can be found in Table 2. The Target Mol library had 22 compounds that showed at least partial selectivity.

      Downstream characterization

      The outcomes of the HTS campaign directed us to look at one compound from the ReFrame library, Tranilast, as well three compounds from the Target Mol library, AST-487, Epoxomicin, and Carfilzomib. The structures and dose curves from the HTS titration and counterscreen assays can be found in Fig. 3A. These compounds, along with analogs of Tranilast were tested in the 96 well cell-based confirmation assay format, showing inhibition levels against a number of viral proteases, including SARS-CoV-2 3CLpro, SARS-CoV-1 3CLpro, Polio 3Cpro, and the unrelated HIV protease (Fig. 3B). Tranilast exhibited SARS 3CLpro specific inhibition. None of the Tranilast analogs showed any inhibition in either the HTS assay format or the confirmation cell-based assay (data not shown). AST-487 showed some selective SARS 3CLpro inhibition but only at the highest concentration tested. Epoxomicin and Carfilzomib confirmed the inhibition seen in the HTS campaign but were less selective and inhibited all the proteases tested. As Epoxomicin and Carfilzomib are described as protease inhibitors, we hypothesized that the observed non-selective inhibition might be due to a change in protein levels. To investigate this, we looked at protein expression by western blot. Surprisingly, we found a dose-dependent decrease in protein expression of SARS-CoV-2 3CLpro-HA suggesting that protease degradation may explain our assay results (data not shown). However additional studies showed a decrease in all exogenously expressed constructs, including all RLuc, FLuc, and all viral proteases, but no appreciable change in actin levels in cells treated with epoxomicin and carfilzomib. This indicates that their robust inhibition in the dual reporter assay is due to an unidentified off-target effect. All plasmids utilized in these studies contain a CMV promoter, so there may be a non-proteolytic mechanism for this observation (Supplemental Fig. 1).
      Fig 3:
      Fig. 3Downstream characterization of 4 HTS active inhibitors. The 4 compounds selected from the HTS campaign with their dose response results and structures are shown in A. The blue curves represent the 3CLpro response while the red is the cytotoxicity response. These 4 compounds were confirmed in the 96-well confirmation assays against 6 different viral constructs (B). Tranilast and analogs were tested in the enzymatic inhibition assay and a representative graph with the Disulfiram control, Tranilast, and the most active inhibitor 32229-1 dose responses is shown in C. The corresponding table of results from the different assays is displayed in D.
      After confirming activity of Tranilast in the cell-based assays, we also wanted to test the enzymatic function of the 3CLpro in the presence of Tranilast or any of its analogs. Disulfiram was used as a positive control and Tranilast gave an IC50 of 37 µM, like the cell-based confirmation assay results. The analogs showed a range of EC50s, with SR-32236-1 showing the best EC50 at 20 µM. The concentration response curves from the control Disulfiram, Tranilast, and the most active analog SR-32229-1 can be found in Fig. 3C. A table representing the EC50s of the 4 compounds selected from the HTS campaign is described in Fig. 3D. The full table of EC50s from the Tranilast analogs, along with their structures can be found in Supplemental Fig. 2.

      Discussion

      COVID-19 continues to stress our hospitals as variants have led to mass outbreaks in the unvaccinated and highly susceptible individuals. Because of this, it is imperative to find additional therapies to treat these patients. By developing a cell-based complementation assay, we are able to quickly screen compounds and or drugs to determine their effectiveness at inhibiting the 3CLpro-mediated cleavage. At the time of the assay development and compound screening, no small molecule controls had ever been identified. GC376 was identified after completion of this effort [
      • Fu L.
      • Ye F.
      • Feng Y.
      • et al.
      Both Boceprevir and GC376 efficaciously inhibit SARS-CoV-2 by targeting its main protease.
      ]. The assay sensitivity and specificity may be further validated by using this as a control going forward.
      The HTS campaign led to a few possible 3CLpro inhibitor leads, that were advanced to enzymatic and other secondary studies. All said, 18 compounds were found to have moderate selectivity in respect to the cytotoxicity counterscreen. We performed both SARS-CoV-1 and SARS-CoV-2 3CLpro assays, along with HIVpro and Polio 3Cpro cell-based follow-up studies and enzyme inhibition assays on 4 compounds that were interesting based on their moderate level of preferential activity in the primary SARS-CoV-2 assay compared to their cytotoxicity profile. These are Tranilast, AST-487, Epoxomicin, and Carfilzomib.  Tranilast confirmed activity in the cell-based assay as well as the enzymatic inhibition assay. Although the analogs of Tranilast did not show any effect in the cell-based confirmation assay or the initial HTS cell-based assay, several showed inhibition in the enzymatic assay, with one compound showing a slight improvement compared to Tranilast. This difference in activity between assays may be explained by low cellular availability of these drugs.
      While we were successful at rapidly screening libraries of drugs that may be repurposed, we were not able to identify drugs that demonstrated more than moderate selectivity over cytotoxicity while demonstrating inhibition of enzymatic activity in the SARS-CoV-2 biochemical assay. The robustness of the cell and biochemical based approaches and the considerable reproducibility of the hits from the vast diversity of drugs found in each library allowed us to narrow our search to a few discreet compounds. We were limited in terms of pharmacologic controls to implement during the March of 2020 time frame that would allow us to gauge the sensitivity of the cell-based and biochemical assays. In fact, at the beginning of this effort no known SARS-CoV-2 3CLpro inhibitors existed. Subsequently, multiple publications have come forth that indicate possible control inhibitors [
      • Ma C.
      • Hu Y.
      • Townsend J.A.
      • et al.
      Ebselen, disulfiram, carmofur, PX-12, tideglusib, and shikonin are nonspecific promiscuous SARS-CoV-2 main protease inhibitors.
      ]. As bona fide inhibitors are now available, we will implement them for future validation of our HTS assay to help hone the sensitivity. The biochemical approach developed as part of the downstream validation may also prove to be the better path forward to identify hits and then triage with the arsenal of well poised cell-based assays we have in place.
      To date, we have successfully completed an HTS campaign, identifying potential inhibitors of the 3CLpro enzyme. We have shown that we can identify compounds which were further characterized by inhibiting the enzymatic activity of the 3CLpro enzyme. To end, the limited number of compounds identified as hits along with their lack of potency, opens the door for much larger and diverse library screening against this target, which may afford first in class inhibitors.
      Supplemental Fig. 1: 293T cells were transfected as in 3B with RLuc, FLuc-reporter, and SARS-CoV-2 3CLpro-HA as indicated, and treated with increasing concentrations of Tranilast, AST, Epoxomicin, Carfilzomib, or MG132 or left untreated (lanes 2, left blot, lanes 2-3 right blot). After 48 h, cell lysates were blotted for RLuc, FLuc, HA, and actin.

      Declaration of Competing Interest

      The authors declared no potential conflict of interest with respect to the research, authorship and or publication of this article.

      Acknowledgments

      We thank Lina DeLuca (Lead Identification, Scripps Florida) for compound management.

      Appendix. Supplementary materials

      • Supplemental Fig. 2: A. Tranilast and disulfiram were control compounds. B. 15 direct analogs were purchased. C. 6 related compounds were also purchased or synthesized. D. SR-32229 is a low-µM inhibitor.

      References

        • Lopez Bernal J.
        • Andrews N.
        • Gower C.;
        • et al.
        Effectiveness of Covid-19 vaccines against the B.1.617.2 (Delta) variant.
        N Engl J Med. 2021; 385: 585-594
        • Baez-Santos Y.M.
        • St John S.E.
        • Mesecar A.D.
        The SARS-coronavirus papain-like protease: structure, function and inhibition by designed antiviral compounds.
        Antiviral Res. 2015; 115: 21-38
        • Zhang L.
        • Lin D.
        • Sun X.
        • et al.
        Crystal structure of SARS-CoV-2 main protease provides a basis for design of improved alpha-ketoamide inhibitors.
        Science. 2020; 368: 409-412
        • Smith E.
        • Davis-Gardner M.E.
        • Garcia-Ordonez R.D.
        • et al.
        High-throughput screening for drugs that inhibit papain-like protease in SARS-CoV-2.
        SLAS Discov. 2020; 25: 1152-1161
        • Jin Z.
        • Du X.
        • Xu Y.
        • et al.
        Structure of M(pro) from SARS-CoV-2 and discovery of its inhibitors.
        Nature. 2020; 582: 289-293
        • Cui W.
        • Yang K.
        • Yang H.
        Recent progress in the drug development targeting SARS-CoV-2 main protease as treatment for COVID-19.
        Front Mol Biosci. 2020; 7616341
        • Owen D.R.
        • Allerton C.M.N.
        • Anderson A.S.
        • et al.
        An oral SARS-CoV-2 M(pro) inhibitor clinical candidate for the treatment of COVID-19.
        Science. 2021; 374: 1586-1593
        • Moghadasi S.A.
        • Esler M.A.
        • Otsuka Y.
        • et al.
        Gain-of-signal assays for probing inhibition of SARS-CoV-2 Mpro/3CLproin living cells.
        mBio. 2022; 13 (-22): e00784
        • Dey-Rao R.
        • Smith G.R.
        • Timilsina U.
        • et al.
        A fluorescence-based, gain-of-signal, live cell system to evaluate SARS-CoV-2 main protease inhibition.
        Antiviral Res. 2021; 195105183
        • Rawson J.M.O.
        • Duchon A.
        • Nikolaitchik O.A.
        • et al.
        Development of a cell-based luciferase complementation assay for identification of SARS-CoV-2 3CLpro inhibitors.
        Viruses. 2021; 13: 173
        • Rothan H.A.
        • Teoh T.C.
        Cell-based high-throughput screening protocol for discovering antiviral inhibitors against SARS-COV-2 main protease (3CLpro).
        Mol Biotechnol. 2021; 63: 240-248
        • Wen W.
        • Chen C.
        • Tang J.
        • et al.
        Efficacy and safety of three new oral antiviral treatment (molnupiravir, fluvoxamine and Paxlovid) for COVID-19:a meta-analysis.
        Ann Med. 2022; 54: 516-523
        • Steger K.
        • Brady J.
        • Wang W.
        • et al.
        CHO-S antibody titers >1 gram/liter using flow electroporation-mediated transient gene expression followed by rapid migration to high-yield stable cell lines.
        J Biomol Screen. 2015; 20: 545-551
        • Baez-Santos Y.M.
        • Barraza S.J.
        • Wilson M.W.
        • et al.
        X-ray structural and biological evaluation of a series of potent and highly selective inhibitors of hµMan coronavirus papain-like proteases.
        J Med Chem. 2014; 57: 2393-2412
        • Zhang J.H.
        • Chung T.D.
        • Oldenburg K.R.
        A simple statistical parameter for use in evaluation and validation of high throughput screening assays.
        J Biomol Screen. 1999; 4: 67-73
        • Janes J.
        • Young M.E.
        • Chen E.
        • et al.
        The ReFRAME library as a comprehensive drug repurposing library and its application to the treatment of cryptosporidiosis.
        Proc Natl Acad Sci USA. 2018; 115: 10750-10755
        • Fu L.
        • Ye F.
        • Feng Y.
        • et al.
        Both Boceprevir and GC376 efficaciously inhibit SARS-CoV-2 by targeting its main protease.
        Nat Commun. 2020; 11: 4417
        • Ma C.
        • Hu Y.
        • Townsend J.A.
        • et al.
        Ebselen, disulfiram, carmofur, PX-12, tideglusib, and shikonin are nonspecific promiscuous SARS-CoV-2 main protease inhibitors.
        ACS Pharmacol Transl Sci. 2020; 3: 1265-1277