Rapid discovery and classification of inhibitors of coronavirus infection by pseudovirus screen and amplified luminescence proximity homogeneous assay

Abstract

To identify potent antiviral compounds, we introduced a high-throughput screen platform that can rapidly classify hit compounds according to their target. In our platform, we performed a compound screen using a lentivirus-based pseudovirus presenting a spike protein of coronavirus, and we evaluated the hit compounds using an amplified luminescence proximity homogeneous assay (alpha) test with purified host receptor protein and the receptor binding domain of the viral spike. With our screen platform, we were able to identify both spike-specific compounds (class I) and broad-spectrum antiviral compounds (class II). Among the hit compounds, thiosemicarbazide was identified to be selective to the interaction between the viral spike and its host cell receptor, and we further optimized the binding potency of thiosemicarbazide through modification of the pyridine group. Among the class II compounds, we found raloxifene and amiodarone to be highly potent against human coronaviruses including Middle East respiratory syndrome coronavirus (MERS-CoV), severe acute respiratory syndrome coronavirus (SARS-CoV), and SARS-CoV-2. In particular, using analogs of the benzothiophene moiety, which is also present in raloxifene, we have identified benzothiophene as a novel structural scaffold for broad-spectrum antivirals. This work highlights the strong utility of our screen platform using a pseudovirus assay and an alpha test for rapid identification of potential antiviral compounds and their mechanism of action, which can lead to the accelerated development of therapeutics against newly emerging viral infections.

Graphical abstract

Fig. 1

Combination of a pseudovirus assay and protein-based assays enabled identification of novel inhibitors targeting the early stage of coronavirus infection. a) Schematic representation of a compound screen using a pseudovirus presenting the spike of MERS-CoV. To generate the pseudovirus, a lentivirus backbone, luciferase gene, and the spike were transfected onto 293T cells. In the compound screen, Huh-7 cells were treated with compounds and infected with the pseudovirus, after which the accumulated luciferase signal was measured. b) Schematic representation of the alpha test. In the alpha test, the receptor binding domain (RBD) of the spike of MERS-CoV was immobilized on the donor bead while its cellular receptor, hDPP4 was attached to an acceptor bead. c) Summary of classification of hits from the pseudovirus screen and subsequent alpha test: ∼16% entry-specific (spike-specific) compounds among 1126 hits. d) Structure of three identified hits: thiosemicarbazide, amiodarone, and raloxifene.

Fig. 2

Thiosemicarbazide (class I) shows target specificity, and its analogs show regions for their antiviral potency. a) Structure of newly-synthesized thiosemicarbazide derivatives. The modified moieties compared to compound 1 or 1f are highlighted in green (2-pyridine), purple (thiosemicarbazide), orange (pyridine), and blue (p-trifuloromethylphenyl). b) Potency (IC₅₀ or EC₅₀ (μM)), cytotoxicity (CC₅₀ (μM)), and selective index (SI = CC₅₀/EC₅₀ (MERS-CoV infection assay)) of thiosemicarbazide derivatives as evaluated in alpha test, pseudovirus assay, and immunofluorescence-based MERS-CoV infection assay. CC₅₀ values that are >50 μM were considered as 50 μM when calculating SI. n.d; not determined. c) Schematic presentation of the ELISA using hDPP4 and the RBD of the S1 of MERS-CoV spike. d-f) Three compounds, 1 (d), 1g-8 (e), and 1j (f) tested for their inhibiting activity on the binding between hDDP4 and RBD. Each data point represents the mean of triplicate assays with ±SEM, and their IC₅₀ values were calculated through curve fitting analysis using Prism 6.0.

Fig. 3

Thiosemicarbazide inhibits MERS-CoV infection by blocking the interaction between RBD and hDPP4. a) Schematic presentation of immunofluorescence-based MERS-CoV infection assay. b) Antiviral potency of compound 1 and 1g-8 against MERS-CoV infection. The cells were treated with individual compounds at the time of MERS-CoV infection at an MOI of 0.0625. The cells were further incubated for 24 h followed by immunofluorescence imaging. MERS-CoV infection was determined using a rabbit anti-MERS-CoV spike antibody (green), and cell viability was measured with Hoechst 33342 (red). Concentrations of 3.125 μM and 50 μM were respectively selected as the low and high doses for each compound. Chloroquine (CHQ) at a concentration of 10 μM was included as a control. c-left) Docking models for compounds 1 and 1j bound to the MERS-CoV receptor binding domain (RBD) (PDB: 4L3N). The molecular surface of MERS-CoV RBD is presented in gray while compound 1 is marked in magenta. c-right) Superimposition of the RBD docked with 1 (the model in left) on the X-ray structure of RBD complexed with hDPP4 (PDB: 4L72). The gray molecular surface is MERS-CoV RBD (PDB id: 4L3N, 4L72) and the green ribbon is hDPP4. d-e) Comparison of docked configurations of 1 (d) and 1j (e) to the binding site in the RBD. Key interactions between ligand and RBD are presented by dashed lines: orange is a hydrogen bond, cyan is a pi-pi interaction, and green is a pi-cation interaction. Ligands are colored by atom types, with carbon as magenta in 1, cyan in 1j, and silver in RBD; nitrogen is blue; oxygen is red; and fluorine is green.

Fig. 4

Amiodarone (class II) inhibits the early stage of coronavirus infection. a) Antiviral activity of amiodarone demonstrated by an immunofluorescence-based MERS-CoV infection assay. In total, ten different concentrations (1–50 μM) of amiodarone were examined, and immunofluorescence images obtained from low (3.12 μM) and high (50 μM) compound concentrations were selected. Green signals represent cells infected with MERS-CoV and red signals indicate cell survival. b) Viral mRNA from cells infected with MERS-CoV quantified by qRT-PCR. Each data point represents the mean ± SEM of triplicate assays. c) Time-of-addition experiment performed by addition of 10 μM amiodarone at different time points during MERS-CoV infection. Drugs were added at 1 h before infection (−1), at the time of infection (0), or at various hours post-infection (+1 ∼ +6), and the extent of inhibition of the MERS-CoV infection was quantified by the immunofluorescence signals. Chloroquine and lopinavir were included at a concentration of 10 μM for comparison. Each data point represents the mean ± SEM of triplicate assays. d) Antiviral activity of amiodarone examined against the HCoV-229E and HCoV-OC43 viruses. Viruses were infected into Vero cells in the presence of amiodarone, and after 24 h of infection, viral mRNA was quantified by qRT-PCR. Each data point represents the mean ± SEM of triplicate assays. e) Graphical presentation of potential mechanisms of the antiviral activity of amiodarone. Amiodarone, as a weak base, could lead to an alkalization of the lysosome and therefore act as a potential inhibitor of lysosomal acid sphingomyelinase. ASM, acid-sphingomyelinase; AM, acid-sphingomyelin.

Fig. 5

Raloxifene inhibits the early stage of coronavirus infection. a) Antiviral activity of raloxifene demonstrated by an immunofluorescence-based MERS-CoV infection assay as previously noted in Fig. 4a. Green signals represent cells infected with MERS-CoV and red signals represent cell survival. b) In total, ten different concentrations of raloxifene were tested in the immunofluorescence assay, and EC₅₀ and cytotoxicity were calculated through curve fitting analysis using Prism 6. c) Viral mRNA quantified by qRT-PCR from the Vero cells infected with MERS-CoV. d) Potency of raloxifene against HCoV-229E or HCoV-OC43 measured by qRT-PCR as detailed in Fig. 4d. e) Time-of-addition experiment using 10 μM of raloxifene. Experiments were performed as described in Fig. 4c. f) Cellular cholesterol levels in A549 cells treated with 10 μM raloxifene for 12 h. g) Expression levels of selected genes involved in cholesterol metabolism quantified by qRT-PCR after 12 h of treatment with 10 μM raloxifene. CH25H; cholesterol 25-Hydroxylase, FASN; fatty acid synthase, MSR1; macrophage scavenger receptor 1, SQLE; squalene epoxidase. Each data point represents the mean ± SEM of triplicate assays.

Fig. 6

Benzothiophene analogs (class II) show regions crucial for antiviral potency. a) Structure activity relationship (SAR) analysis performed using newly synthesized benzothiophne derivatives. The modified moieties of the phenyl group of compound 2 are highlighted in green or orange color, and the thiophene moiety is highlighted in purple. b) Potency (EC₅₀ (μM)) of tested benzothiophene derivatives evaluated by MERS-CoV pseudovirus assay and immunofluorescence-based MERS-CoV infection assay, cytotoxicity (CC₅₀ (μM)), and selective index (SI = CC₅₀/EC₅₀ (MERS-CoV infection assay)). CC₅₀ values that are >50 μM were considered as 50 μM when calculating SI. n.d; not determined. c) Antiviral activity of 2a-1 and 2a-4 examined by immunofluorescence-based MERS-CoV infection assay as noted in Fig. 4a. Green signals present the cells infected with MERS-CoV and red signals present all cells. d-e) Antiviral potency of 2a-1 (d) and 2a-4 (e) against HCoV-229E or HCoV-OC43 measured by qRT-PCR as noted in Fig. 4d. Each data point represents the mean of triplicate assays with ±SEM.

Fig. 7

Compounds inhibiting the early stage of coronavirus infection show potency against both SARS-CoV and SARS-CoV-2 infection. a) Antiviral activity of the selected broad-spectrum compounds (amiodarone, raloxifene, 2a-1, and 2a-4) demonstrated by immunofluorescence-based SARS-CoV infection assay. Among the ten different concentrations (1–50 μM) of each compound examined, images at low (3.12 μM) and high (50 μM) concentrations were selected. Green signals represent cells infected with SARS-CoV and red signals represent cell survival. b-c) Potency of the selected drugs against SARS-CoV2 infection demonstrated with a cell-based infection model. Cells were treated with 10 μM of amiodarone, 2a-1, and 2a-4 (b), raloxifene (c) at the time of infection with SARS-CoV-2, and viral mRNA was quantified by qRT-PCR after 24 h of infection. For comparison, 10 μM of remdesivir was included. d-g) In vivo potency of raloxifene against SARS-CoV-2 demonstrated using a hamster infection model. d) Schematic presentation of SARS-CoV2 infection and sampling. SARS-CoV-2 was infected through nasal inoculation and animals were treated with raloxifene during a two-day infection period at 12 mg/kg or 36 mg/kg doses. On day 2, viral titer in the hamsters was measured by qRT-PCR. e-f) Viral titers of SARS-CoV-2 in lungs quantified on day 2 by qRT-PCR (e) or TCID₅₀ (f). g) Viral titer of nasal turbinate quantified on day 2 by TCID₅₀. h) Body weight of the hamsters was measured daily for the 2-day infection period and they remained normal. Each data point represents the mean ± SEM of triplicate assays.