Non-covalent SARS-CoV-2 Mpro inhibitors developed from in silico screen hits

Abstract

M^pro, the main protease of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is essential for the viral life cycle. Accordingly, several groups have performed in silico screens to identify M^pro inhibitors that might be used to treat SARS-CoV-2 infections. We selected more than five hundred compounds from the top-ranking hits of two very large in silico screens for on-demand synthesis. We then examined whether these compounds could bind to M^pro and inhibit its protease activity. Two interesting chemotypes were identified, which were further evaluated by characterizing an additional five hundred synthesis on-demand analogues. The compounds of the first chemotype denatured M^pro and were considered not useful for further development. The compounds of the second chemotype bound to and enhanced the melting temperature of M^pro. The most active compound from this chemotype inhibited M^pro in vitro with an IC₅₀ value of 1 μM and suppressed replication of the SARS-CoV-2 virus in tissue culture cells. Its mode of binding to M^pro was determined by X-ray crystallography, revealing that it is a non-covalent inhibitor. We propose that the inhibitors described here could form the basis for medicinal chemistry efforts that could lead to the development of clinically relevant inhibitors.

Figure 1

Development of M^pro inhibitors from in silico screening hits. (A) Graph showing the inhibitory activity of the five validated in silico screening hits tested at a final concentration of 40 μM in an in vitro protease-activity assay. After addition of the FRET substrate, fluorescence was acquired at 10 min intervals over 60 min. For each compound, the increase in fluorescence intensity was normalized to the DMSO control. GC376, a previously described M^pro inhibitor, served as a positive control. Blank, reaction omitting M^pro; RFU, relative fluorescence units. (B) Chemical structure of the five validated compounds. (C) Chemical structure of Z222979552, the most active dihydro-quinolinone compound obtained after two rounds of chemical structure similarity searches of the REAL space library of molecules. (D) Dose–response curves for compound Z222979552 examined at 0.4, 1, 3, 5, 10 and 20 μM final compound concentrations, in absence (solid lines) or presence (dashed lines) of 0.1 μg protein lysate. DMSO, GC376 and blank controls are as described above. (E) Thermal shift assay performed in the presence of DMSO or 20 μM of compounds Z222979552 or GC376. The graphs show the derivatives of the melting curves used to calculate the melting temperature of M^pro.

Figure 2

Crystal structure of M^pro in complex with the dihydro-quinolinone compound Z222979552. (A) Surface representation of SARS-CoV-2 M^pro with secondary structure elements colored green and compound Z222979552 shown as a stick model. The omit electron density map corresponding to the ligand compound is shown (σ = 1.0). (B) Stick model of compound Z222979552 and of select residues at the active site of M^pro showing the key interactions between the protease and its inhibitor.

Figure 3

Compound Z222979552 reduces SARS-CoV-2 replication in Vero E6 cells in a dose-dependent manner. (A) SARS-CoV-2 titers upon treatment with compound Z222979552 or DMSO control. Viral titers are displayed as fifty-percent tissue culture infective dose (TCID₅₀)/ml 24 h post-infection. TCID₅₀ values correspond to the viral titers (log₁₀ scale) required to kill fifty-percent of infected host cells. The experiment was performed in triplicate and the results are shown as means and standard deviations. (B) Immunofluorescence staining of infected Vero E6 cells for double-stranded RNA (dsRNA). The cells were treated with compound Z222979552 or DMSO and were stained 24 h post-infection. Green, dsRNA; blue, DAPI. One representative image out of three biological replicates is shown. (C) Z222979552-mediated cytotoxicity, determined using Vero E6 cells treated with the compound or DMSO for 24 h. The experiment was performed in triplicate and the results are shown as means and standard deviations.