Discovery of SARS-CoV-2 Papain-like Protease Inhibitors through a Combination of High-Throughput Screening and a FlipGFP-Based Reporter Assay.

The papain-like protease (PLpro) of SARS-CoV-2 is a validated antiviral drug target. Through a fluorescence resonance energy transfer-based high-throughput screening and subsequent lead optimization, we identified several PLpro inhibitors including Jun9-72-2 and Jun9-75-4 with improved enzymatic inhibition and antiviral activity compared to GRL0617, which was reported as a SARS-CoV PLpro inhibitor. Significantly, we developed a cell-based FlipGFP assay that can be applied to predict the cellular antiviral activity of PLpro inhibitors in the BSL-2 setting. X-ray crystal structure of PLpro in complex with GRL0617 showed that binding of GRL0617 to SARS-CoV-2 induced a conformational change in the BL2 loop to a more closed conformation. Molecular dynamics simulations showed that Jun9-72-2 and Jun9-75-4 engaged in more extensive interactions than GRL0617. Overall, the PLpro inhibitors identified in this study represent promising candidates for further development as SARS-CoV-2 antivirals, and the FlipGFP-PLpro assay is a suitable surrogate for screening PLpro inhibitors in the BSL-2 setting.

Introduction

The COVID-19 pandemic has led to 170,812,850 confirmed cases and 3,557,586 deaths as of June 2, 2021, rendering it the worst pandemic since the 1918 Spanish flu. The etiological agent of COVID-19 is SARS-CoV-2, a single-stranded positive-sense RNA virus that belongs to the beta coronavirus genus.[1] Two additional coronaviruses within the same genus, SARS-CoV, and MERS-CoV, have caused epidemics in humans with mortality rates of 9.6% and 34.3%, respectively. Although SARS-CoV-2 has a lower mortality rate of 2.1% compared to SARS-CoV and MERS-CoV, it has led to a far greater death toll due to its higher transmission.[2] SARS-CoV-2 differs from SARS-CoV and MERS-CoV in that it has a long incubation time after the initial infection (1–2 weeks), and a large percentage of infected patients continue to shed the virus while being asymptomatic, presenting a daunting task for surveillance and containment.[3]

Two mRNA vaccines developed by Pfizer/BioNtech and Moderna and one adenovirus-based vaccine by Johnson and Johnson have been approved by FDA in the United States.[4] For small molecule antivirals, remdesivir received FDA approval on October 22, 2020.[5] Although the polymerase of SARS-CoV-2 has a proofreading function, it continues to mutate at a rate about 10–6 per site per cycle.[3] Several variants have already emerged and have widely circulated among humans since the beginning of the pandemic.[6] Therefore, there is a dire need for additional antivirals with a novel mechanism of action. Antivirals are not substitutes for vaccines but rather important complements that can be used for the treatment of infection from both wild-type (WT) and variant viruses. Among the viral proteins that have been actively pursued as SARS-CoV-2 antiviral drug targets, the main protease (Mpro) and papain-like protease (PLpro) are the most promising ones.[7,8] Mpro and PLpro are involved in the proteolytic digestion of the viral polyproteins pp1a and pp1ab, yielding individual functional viral proteins for the replication complex formation. PLpro cleaves at three sites with the recognition sequence “LXGG↓XX”.[9] PLpro has been shown to play additional roles in dysregulating the host immune response and impairing the host type I interferon antiviral effect through its deubiquitinating and deISG15ylating (interferon-induced gene 15) activities, respectively.[10−12] SARS-CoV-2 PLpro cleaves ISG15 and polyubiquitin modifications from cellular proteins, and inhibition of PLpro led to the accumulation of ISG15-conjugates and polyubiquitin conjugates.[13] While SARS-CoV PLpro prefers ubiquitinated substrates, SARS-CoV-2 PLpro prefers the ISGlyated proteins as substrates.[10−12] PLpro is part of a membrane -anchored multidomain protein named nonstructural protein 3 (nsp-3), an essential component of the replicase–transcriptase complex. The pleiotropic roles of SARS-CoV-2 PLpro make it a promising antiviral drug target. Substantial morbidity and mortality associated with COVID-19 infection is caused by cytokine storm,[14] and suppressing host immune response using dexamethasone and baricitinib has been shown to provide therapeutic benefits in the treatment of severe infections.[15,16]

Significant progress has been made in developing SARS-CoV-2 Mpro inhibitors,[7,8,17−20] and the Pfizer compounds PF-07304814 and PF-07321332 are currently in phase 1 clinical trials.[21] In comparison, PLpro represents a more challenging drug target, and GRL0617 remains one of the most potent PLpro inhibitors reported to date despite several high-throughput screening and medicinal chemistry optimization campaigns.[9,10,12,22,23]GRL0617 was originally developed as a deubiquitinase inhibitor and was later identified as a SARS-CoV PLpro inhibitor through a high-throughput screening.[23] As SARS-CoV-2 and SARS-CoV PLpro share a sequence identity of 83% and similarity of 90%, GRL0617 was also repurposed for SARS-CoV-2 PLpro, and it was reported to inhibit SARS-CoV-2 PLpro with IC50 values of around 2 μM and SARS-CoV-2 viral replication with EC50 values around 20 μM from multiple studies.[12,13,22,24]

In this study, we report our progress in developing novel SARS-CoV-2 PLpro inhibitors. Using the fluorescence resonance energy transfer (FRET)-based enzymatic assay, we conducted a high-throughput screening against the Enamine 50K diversity compound library and identified two hits Jun9-13-7 and Jun9-13-9 with single-digit micromolar IC50 values. Subsequent lead optimization led to the discovery of several hits with sub-micromolar potency in the enzymatic assay. Notably, we developed the FlipGFP assay for quantifying the intracellular PLpro inhibition, which can be conducted in the biosafety level 2 (BSL-2) setting. We found a positive correlation between the results from the FlipGFP-PLpro assay and the antiviral assay, suggesting that the FlipGFP-PLpro can be applied to faithfully predict the cellular antiviral activity of PLpro inhibitors. The X-ray crystal structure showed that binding of GRL0617 to the wild-type (WT) SARS-CoV-2 PLpro induced a conformational change in the BL2 loop to the more closed conformation. Molecular dynamics (MD) simulations revealed that the replacement of the carboxamide group in GRL0617 to the trialkyl ammonium in Jun9-72-2 and Jun9-75-4 results in stronger ionic hydrogen bonding interaction between the N–H+ group with the side chain of Asp164 and in a more optimal fitting of the receptor binding area. Overall, the SARS-CoV-2 PLpro inhibitors reported herein represent promising hits for further development as SARS-CoV-2 antivirals, and the FlipGFP-PLpro assay is useful in testing the cellular activity of PLpro inhibitors in the BSL-2 setting.

Results and Discussion

Expression and Characterization of SARS-CoV-2 PLpro

Two constructs of SARS-CoV-2 PLpro were expressed in Escherichiacoli, one with a Hig-tag (PLpro-His) and another without the tag (PLpro). To profile the proteolytic activity of PLpro in cleaving the viral polyprotein, we developed a FRET-based enzymatic assay with the peptide substrate 4-((4-(dimethylamino)phenyl)azo)benzoic acid (Dabcyl)-FTLRGG/APTKV-5-[(2-aminoethyl)amino]naphthalene-1-sulfonic acid (Edans), which corresponds to the nsp2 nsp3 junction from the SARS-CoV-2 polyprotein. The enzymatic activity kcat/Km of PLpro-His and PLpro was 340 M–1 s–1 and 255 M–1 s–1 (Table S1), respectively, which is consistent with previous reports.[12,24] SARS-CoV-2 PLpro was also reported to have deubiquitinating and deISGylating activities.[10−13,22] Accordingly, we characterized the deubiquitinating and deISGylating activities of SARS-CoV-2 PLpro-His using the Ub-AMC and ISG-AMC substrates, respectively, in the enzymatic assay. It was found that SARS-CoV-2 PLpro-His is more efficient in cleaving the ubiquitin (Ub) and ISG15 (ISG) modifications than the viral polyprotein, with kcat/Km values of 1070 and 1.67 × 105 M–1 s–1 (Table S2), respectively. This substrate preference is in agreement with results reported previously,[24] and SARS-CoV PLpro was also reported to have a similar substrate preference.[25] Significantly, the deISGylating activity is 156-fold higher than the deubiquitinating activity, which is consistent with previous reports that SARS-CoV-2 PLpro prefers ISG15 over ubiquitin.[9−13]

High-Throughput Screening of the Enamine 50K Diversity Library against the SARS-CoV-2 PLpro and Hit Validation

The HTS assay was optimized in 384-well plates using the FRET substrate, which gave a Z′ factor of 0.668 with a signal-to-noise ratio (S/B) of 11.2, indicating that this was a robust assay (Figure ). We then performed the HTS against the enamine library, which consists of 50 240 structurally diverse compounds. GRL0617 was included as a positive control.

Figure 1

The 384-well high-throughput screening assay for SARS-CoV-2 PLpro. The signal to base ratio (S/B) is 11.2, and the calculated Z′ factor is 0.688.

Hits showing more than 50% inhibition (Figure S1) were repurchased from Enamine and titrated in the FRET-based enzymatic assay to determine the IC50 values (Figure A and Table S3). In parallel, a differential scanning fluorimetry (DSF) assay was performed as a secondary assay to characterize the binding of the hits with SARS-CoV-2 PLpro (Figure B and Table S3). The most potent two hits, Jun9-13-7 and Jun9-13-9 (Figure C), had IC50 values of 7.29 ± 1.03 and 6.67 ± 0.05 μM, respectively. Jun9-13-7 and Jun9-13-9 also increased the thermal stability of SARS-CoV-2 PLpro by 2.98 ± 0.09 and 2.18 ± 0.29 °C (Table S3), which is consistent with their enzymatic inhibition. In comparison, GRL0617 had an IC50 value of 2.05 ± 0.12 μM and increased the protein stability by 3.52 ± 0.27 °C in the DSF assay (Table S3). The potency of GRL0617 in inhibiting SARS-CoV-2 PLpro from our study is consistent with recent reports.[10−13] The rest of the hits had weak enzymatic inhibition (IC50 > 10 μM) and showed marginal binding to PLpro (Table S3); therefore, they were not further pursued. Both Jun9-13-7 and Jun9-13-9 also inhibit the deubiquitinating and deISGylating activities with IC50 values ranging from 4.93 to 12.51 μM (Figure D and Table S4). In contrast, neither of these two compounds inhibited SARS-CoV-2 Mpro up to 200 μM (Figure S2), suggesting the inhibition of SARS-CoV-2 PLpro is specific. The binding of Jun9-13-7 and Jun9-13-9 to SARS-CoV-2 PLpro was further characterized using the native mass spectrometry (Figure E). It was shown that both Jun9-13-7 and Jun9-13-9 showed dose-dependent binding to PLpro with binding stoichiometry of one drug per PLpro, similar to the positive control GRL0617. Enzymatic kinetic studies showed that compounds Jun9-13-7 and Jun9-13-9 are noncovalent inhibitors with Ki values of 3.96 and 2.10 μM, respectively (Figure S3). The Lineweaver–Burk plots yielded an intercept at the Y-axis, suggesting that both compounds are competitive inhibitors similar to GRL0617 (Figure S3).

Figure 2

HTS and hit validation of SARS-CoV-2 PLpro inhibitors. (A) IC50 values of the screening hits in the FRET-based enzymatic assay, the red line indicates the IC50 = 10 μM. (B) Differential scanning fluorimetry assay of the screening hits in stabilizing the SARS-CoV-2 PLpro. (C) Chemical structures of GRL0617, Jun9-13-7, and Jun9-13-9. (D) Inhibitory activity of Jun9-13-7 and Jun9-13-9 against SARS-CoV-2 PLpro using Ub-AMC and ISG-AMC substrates. (E) Native MS binding assay of Jun9-13-9 and Jun9-13-7 to SARS-CoV-2 PLpro.

Lead Optimization of SARS-CoV-2 PLpro Inhibitors

To further optimize the enzymatic inhibition of Jun9-13-7 and Jun9-13-9, 13 structural analogues were purchased from Enamine (Figure A), and 34 compounds were synthesized (Figure B) to elucidate the structure–activity relationships (SAR). It was found that a hydroxyl substitution on the left phenyl ring is critical for the activity, as methylation led to significant loss of enzymatic inhibition (Jun9-13-9 vs Jun9-25-4). The methyl substitution on the methylene linker is also important for the enzymatic inhibition (Jun9-13-9 vs Jun9-26-2). Similarly, the ortho-methyl or chloride substation on the right phenyl ring is critical for the activity (Jun9-13-7 vs Jun9-29-5; Jun9-13-7 vs Jun9-13-4). Next, guided by this initial SAR results, 34 analogues were designed and synthesized (Figure B). Nine compounds had IC50 values less than 1 μM including Jun9-75-4 (IC50 = 0.62 μM), Jun9-85-1 (IC50 = 0.66 μM), Jun9-84-3 (IC50 = 0.67 μM), Jun9-87-1 (IC50 = 0.87 μM), Jun9-72-2 (IC50 = 0.67 μM), Jun9-87-2 (IC50 = 0.90 μM), Jun9-87-3 (IC50 = 0.80 μM), Jun9-75-5 (IC50 = 0.56 μM), and Jun9-53-2 (IC50 = 0.89 μM). Among them, Jun9-75-4 was the most potent PLpro inhibitor with an IC50 of 0.62 μM, a 10-fold increase compared to Jun9-13-9 (IC50 = 6.67 μM). Jun9-75-4 is also 3-fold more potent than GRL0617 (IC50 = 2.05 ± 0.12 μM), representing one of the most potent PLpro inhibitors reported to date.

Figure 3

SAR of SARS-CoV-2 PLpro inhibitors. (A) Analogues of GRL0617 purchased from Enamine. (B) Synthetic compounds designed based on the SAR results. Potent compounds with IC50 values less than 1 μM are highlighted in blue.

Development of FlipGFP Assay for Testing the Cellular Activity of SARS-CoV-2 PLpro Inhibitors

One of the challenges in SARS-CoV-2 antiviral drug discovery is that SARS-CoV-2 is a biosafety level 3 (BSL-3) pathogen, which limits the number of drug candidates that can be screened. To help prioritize lead compounds for the antiviral assay with infectious SARS-CoV-2, which requires BSL-3 facility, we developed a cell-based FlipGFP assay for SARS-CoV-2 PLpro that is suitable for testing the intracellular activity of PLpro inhibitors in the BSL-2 setting. The two major advantages of a cell-based PLpro assay over the FRET-based enzymatic assay are that (1) it can eliminate compounds that are either cytotoxic or membrane impermeable, and (2) substrate cleavage in the cell cytoplasm recapitulates the physiological process of viral polyprotein cleavage by PLpro in a virus-infected cell. It is known that cysteine proteases are susceptible to redox active compounds as well as nonspecific alkylating chemicals such as ebselen.[26,27] The FlipGFP-PLpro assay is expected to rule out such promiscuous compounds since the substrate is cleaved under the reducing intracellular environment.

In the assay design, the 10th and 11th β-strands from the GFP protein were separated from the rest of the GFP β-barrel (β-strands 1–9) (Figure A).[28−30] The 10th and 11th β-strands were linked through the PLpro cleavage site and a heterodimerized coiled coils E5/K5. In the absence of the PLpro, the 10th and 11th β-strands are restrained and unable to associate with the GFP β-barrel 1–9. When the cleavage site is digested by the PLpro, the 11th β-strand then flips its orientation and associates with GFP β-barrel 1–9 together with the 10th β-strand, leading to restoration of the green fluorescence signal (Figure A). A red fluorescent protein mCherry was included within the construct via a “self-cleaving” 2A peptide to act as the transfection control (Figure B), and the normalized ratio of green fluorescence signal over red fluorescence signal is proportional to the enzymatic activity of PLpro. Cells transfected with FlipGFP-PLpro but without the PLpro showed no green fluorescence signal (Figure C, sixth row), suggesting host proteases are unable to cleave the PLpro substrate sequence, thereby eliminating the background signal interference. Specifically, little or no GFP signal was observed when the cells were transfected with SARS-CoV-2 PLpro and a construct containing either the TEV cleavage site (FlipGFP-TEV) (Figure C, fourth row) or the Mpro cleavage site (FlipGFP-Mpro) (Figure C, third row). Similarly, little or no GFP signal was observed when the cells were transfected with SARS-CoV-2 Mpro and a construct containing the PLpro cleavage site (FlipGFP-PLpro) (Figure C, fifth row). In contrast, strong green fluorescence signals were observed when the cells were transfected with PLpro and FlipGFP-PLpro (Figure C, seventh row) or Mpro and FlipGFP-Mpro (Figure C, second row).

Figure 4

Development of cell-based FlipGFP assay for the quantification of the cellular activity of SARS-CoV-2 PLPro inhibitors. (A) Design principle for the cell-based FlipGFP assay. (B) Sequence of the flipped GFP β10–11 and cleavage site. (C) FlipGFP-PLPro assay development. 293T cells that were not transfected (Ø), or transfected with FlipGFP-Mpro and SARS-CoV-2 Mpro plasmids, or FlipGFP-Mpro and SARS-CoV-2 PLpro plasmids, or FlipGFP-TEV and SARS-CoV-2 PLpro plasmids, or FlipGFP-PLpro and SARS-CoV-2 Mpro plasmids, or FlipGFP-PLpro plasmid alone, or FlipGFP-PLpro and SARS-CoV-2 PLpro plasmids (details are described in Materials and Methods). Representative images of FlipGFP-PLpro assay with the positive control GRL0617 (D) and the negative control GC376 (E). (F) Dose–response inhibition PLpro in the FlipGFP-PLpro assay by nine compounds selected from the FRET-based enzymatic assay.

With the established assay condition, we then screened nine most potent PLpro inhibitors with IC50 values less than 1 μM from the FRET-based enzymatic assay (Figure ). GRL0617 and GC376 were included as positive and negative controls, respectively. Compounds were added 3 h post transfection, and GFP and mCherry fluorescence signals were measured at 48 h post transfection. A dose-dependent decrease of the GFP signal was observed with increasing concentrations of GRL0617 (Figure D), and quantification of the normalized GFP/mCherry ratio gave an EC50 value of 9.29 ± 3.45 μM. As expected, GC376 had no effect on the intensity of green fluorescence signal (EC50 > 60 μM) (Figure E), suggesting the FlipGFP assay is suitable for the screening of PLpro inhibitors. Among the nine compounds tested in the cell-based FlipGFP assay, compounds Jun9-53-2, Jun9-72-2, Jun9-75-4, Jun9-85-1, and Jun9-87-1 had EC50 values less than 10 μM, while compounds Jun9-84-3 and Jun9-87-3 were less active with EC50 values of 17.07 μM and 10.16 μM, respectively. Compounds Jun9-75-5 and Jun9-87-2 were not active (EC50 > 50 μM), despite their potent activity in the FRET-based enzymatic assay (Figure Jun9-75-5, IC50 = 0.56 μM; Jun9-87-2, IC50 = 0.90 μM).

Cellular Antiviral Activity of PLpro Inhibitors against SARS-CoV-2

To determine whether there is a correlation between the FlipGFP-PLpro assay results and the cellular antiviral activity of PLpro inhibitors, we first tested the nine PLpro inhibitors selected from the FRET assay with IC50 values less than 1 μM against SARS-CoV-2 (USA-WA1/2020) in Vero E6 cells. GRL0617 inhibited SARS-CoV-2 with an EC50 of 23.64 μM. Compounds Jun9-72-2, Jun9-75-4, Jun9-84-3, Jun9-85-1, and Jun9-87-1 had more potent antiviral activity than GRL0617 with EC50 values of 6.62 μM, 7.88 μM, 8.31 μM, 7.81 μM, and 10.14 μM, respectively (Figure A). Compounds Jun9-53-2 and Jun9-87-3 had similar antiviral activity as GRL0617 with EC50 values of 25.19 μM and 22.34 μM, respectively. In contrast, Jun9-75-5 and Jun9-87-2 were not active (EC50 > 60 μM) (Figure S4).

Figure 5

Antiviral activity of PLpro inhibitors against SARS-CoV-2 in Vero E6 and Caco-2 hACE2 cells. (A) Antiviral activity of PLpro inhibitors against SARS-CoV-2 in Vero E6 cells. (B) Antiviral activity of PLpro inhibitors against SARS-CoV-2 in Caco2-hACE2 cells. Antiviral assay results for compounds Jun9-75-5 and Jun9-87-2 are shown in Figure S4.

To further confirm the antiviral activity, we tested the same set of compounds against SARS-CoV-2 in Caco2-hACE2 cells. Caco2-ACE2 expresses TMPRSS2 and is a physiologically relevant cell line for SARS-CoV-2 replication.[31−33]GRL0617 inhibited SARS-CoV-2 replication in Caco2-hACE2 cells with an EC50 of 19.96 μM, and seven PLpro inhibitors Jun9-53-2, Jun9-72-2, Jun9-75-4, Jun9-84-3, Jun9-85-1, Jun9-87-1, and Jun9-87-3 showed improved antiviral activity with EC50 values ranging from 7.90 to 16.22 μM (Figure B). Jun9-75-5 and Jun9-87-2 were not active (EC50 > 60 μM) (Figure S4), which is consistent with the results from the Vero E6 cells.

Overall, three PLpro inhibitors Jun9-72-2, Jun9-85-1, and Jun9-87-1 were identified as potent SARS-CoV-2 antivirals with EC50 values at or less than 10 μM when tested in both the Vero E6 and Caco2-hACE2 cell lines.

Correlation between the Results from the FlipGPF PLpro Assay and the Antiviral Assay

Plotting the FlipGFP-PLpro assay results with the antiviral assays results showed that there is a positive correlation in both the Vero E6 and Caco2-hACE2 cell lines with R2 values of 0.86 and 0.89, respectively (Figure A,B). Specifically, compounds Jun9-75-5 and Jun9-87-2 with weak activity in the FlipGFP-PLpro assay (EC50 > 60 μM and 55.07 μM, Figure E) also had no antiviral activity against SARS-CoV-2 (Vero E6 and Caco2-hACE2 cells EC50 > 60 μM, Figure S4). The remaining seven compounds which had potent activity in the FlipGFP-PLpro assay also showed potent antiviral activity in both the Vero E6 and Caco2-hACE2 cells (Figure A,B). These results suggest that the FlipGFP-PLpro assay can be used to faithfully predict the cellular antiviral activity of PLpro inhibitors against infectious SARS-CoV-2. Although the FRET-based enzymatic assay is typically used to select compounds for the antiviral assay, we found there is a poor correlation between the FRET assay results and the cellular antiviral assay results (Figure C,D). Taken together, the correlation plots highlighted the advantage of the FlipGFP-PLpro assay in prioritizing lead compounds for the antiviral assay with infectious SARS-CoV-2.

Figure 6

Correlation of the FlipGFP-PLpro, FRET assay results with the antiviral assay results. (A) Correlation of the results between FlipGFP-PLpro assay and the antiviral assay in Vero E6 cells. (B) Correlation of the results between FlipGFP-PLpro assay and the antiviral assay in Caco2-hACE2 cells. (C) Correlation of the results between FRET assay and the antiviral assay in Vero E6 cells. (D) Correlation of the results between FRET assay and the antiviral assay in Caco2-hACE2 cells.

X-ray Crystal Structure of SARS-CoV-2 PLpro in Complex with GRL0617

The complex structure of SARS-CoV-2 PLpro with GRL0617 was determined at 2.50 Å resolution, providing insight into its mechanism of inhibition. There are two monomers per asymmetric unit in the P21 space group. Unambiguous electron density reveals that GRL0617 binds to the S3–S4 subpockets of PLpro (Figure A). The naphthalene ring is positioned in the S4 site, where it forms hydrophobic interactions with Pro247 and Pro248. Upon ligand binding, Tyr268 flips inward (Figure B) to π-stack with the naphthalene and benzene rings. Connecting the naphthalene to the amide is a methylene linker that is substituted with a methyl group. Here we show that this methyl inserts directly into the core of the S4 subpocket. This section is nearly superimposable with the γ and δ1 carbons of the P4 leucine for the ISG15 substrate in the previously determined complex structure (Figure C), demonstrating the nonpolar features of the S4 site as well as the complementarity of the methyl moiety with the core of this subpocket. The amide nitrogen of GRL0617 serves as a hydrogen bond acceptor for the side chain of Asp164, while the amide oxygen accepts a hydrogen bond from the mainchain amide of Gln269. The disubstituted benzene spans the central substrate channel, partially occupying the P5–P3 substrate mainchain binding site, where it forms π–π interactions with the side chains of Tyr268/Gln269, and the backbone amides of Gly163/Asp164. The ortho-methyl group projects toward the catalytic core forming hydrophobic interactions with the S2 site, forcing Leu162 slightly outward compared with the apo structure, while the meta-nitrogen orients toward the S5 site, causing Gln269 to swing inward to accept a hydrogen bond.

Figure 7

Complex structure of SARS-CoV-2 PLpro with GRL0617. The protein and ligand of the SARS-CoV-2 complex are colored in light green and dark green, respectively. C111* indicates the catalytic cysteine. (A) Binding mode of GRL0617 with an unbiased Fo – Fc map, shown in gray, contoured at 2σ. Hydrogen bonds are shown as red dashed lines. (B) Superimposition with apo SARS-CoV-2 PLpro (salmon, PDB ID 6WZU). Significant rearrangement is observed in the loop comprising residues Asn267-Cys270 upon GRL0617 binding. These movements are indicated with arrows. (C) Superimposition with the terminal five residues of the ubiquitin-like protein ISG15 substrate (orange) from the complex with SARS-COV-2 PLpro (PDB ID 6XA9, showing only the ligand). The atoms of the Leu side chain at the P4 position are labeled. (D) Superimposition with the catalytic mutant C/S111 in complex with GRL0617 (PDB ID 7JIR, light blue and dark blue). OAc indicates an acetate molecule in the C/S111 mutant active site that may have caused the structural differences between the WT and mutant complexes.

In parallel to our study, the X-ray crystal structures of SARS-CoV-2 PLpro in complex with GRL0617 and its analogues were also released by others with PDB IDs of 7CMD,[24]7CJM (C111S),[13]7JIR (C111S, Snyder457),[22]7JIT (C111S, Snyder495), 7JIV (C111S, Snyder530), and 7JIW (C111S, Snyder530). Notably, we are among the first ones to crystallize GRL0617 with the WT SARS-CoV-2 PLpro.

One of the unique aspects of GRL0617 is that it does not interact with the catalytic core but instead binds to a distal portion of the active site. Other research groups have determined complex structures of PLpro with GRL0617 with its catalytic cysteine, Cys 111, mutated to a serine, presumably to increase its propensity to crystallize (PDB ID, 7JIR (2.1 Å) and 7CJM (3.2 Å)).[13,22] When the three structures are compared, the GRL0617 adopts a nearly identical pose. Minor differences in the side chain conformations of Glu 167 and Gln 269 are observed. However, there is a significant difference in the pose of Leu 162 between the WT and the C111S mutants (Figure D). In our WT structure, Leu 162 inserts into the core of the protein, where it maintains an interatomic distance of 3.4 Å with the catalytic cysteine. In contrast, Leu 162 of both C111S structures flips outward, toward the solvent. In the higher-resolution structure (PDB ID 7JIR), an acetate from the crystallization condition is modeled in the active site. When superimposed with our WT structure, this acetate clashes with Cys 111 (closest distance 2.5 Å) and Leu 162 (3.0 Å). In the lower-resolution C111S mutant complexed with GRL0617 (PDB ID 7CJM), no acetate is modeled, but Leu 162 adopts the same conformation as the higher-resolution C111S structure (PDB ID 7JIR). Further inspection of the 2Fo – Fc map of 7JCM reveals that there is unmodeled density corresponding to the acetate from PDB ID 7JIR. Interestingly, this experiment did not use acetate in their crystallization condition. Therefore, the density in the catalytic core of both C111S structures likely corresponds to a species of unknown identity that preferentially interacts with a serine residue.

Molecular Dynamics Simulations of SARS-CoV-2 PLpro with GRL0617, Jun9-53-2, Jun9-72-2, and Jun9-75-4

The binding interactions between the GRL0617 and the PLpro protein in the X-ray structure with PDB ID 7JRN and the stability of the X-ray structure were further explored using 100 ns MD simulations. The MD simulations show that the complex formed is stable (Figure A–C) and did not deviate significantly from the starting crystallographic structure in the protease S4/S3 area, having RMSD values smaller than ca. 2.4 Å for the protein and ca. 2 Å for the ligand (Figure C). The MD simulations further verified the stability of the binding interactions inside the broad binding cavity of SARS-CoV-2 PLpro observed in the X-ray structure, as inspected from the MD simulation trajectory and shown in frequency interaction and RMSD plots (Figure B). The naphthalene ring of the ligand is positioned in the hydrophobic S4 site, according to the specific binding features of a ISG15 peptidic substrate, e.g., with LXGG sequence (PDB ID 4MOW),[22] where it forms T-shaped π–π stacking with Tyr268 and has dispersion interactions with Pro248, Tyr264, and occasionally with Pro247, while the phenyl ring of the ligand can interact with L162 (Figure A,B). Hydrogen bonding interactions stabilize the ligand, e.g., between GRL0617 amide CO and the main chain NH of Gln269; the GRL0617 amide NH and the Asp164 side chain carboxylic acid, and occasionally between the anilino amino group of GRL0617 and Tyr 268 side chain hydroxy group (Figure A,B).

Figure 8

Molecular dynamics simulations of SARS-CoV-2 PLpro with GRL0617, Jun9-53-2, Jun9-72-2, and Jun9-75-4. In (A), (D), (G), and (J) are shown representative frames from 100 ns MD simulations of the complexes between the GRL0617, Jun9-53-2, Jun9-72-2, and Jun9-75-4 inside SARS-CoV-2 PLPro (protein = light blue ribbon and sticks; the ligand’s carbons are shown in green; nitrogen and oxygen are shown in blue and red, respectively; hydrogen bonding interactions are shown with red dashes). In (B), (E), (H), and (K) are shown stabilizing interactions inside the binding area of PLpro with inhibitors; hydrogen bonding interactions are depicted in blue, π–π stacking in yellow, hydrophobic interactions in brown, and water bridges in green. The binding interactions are considered important when the frequency bar is ≥0.2. In (C), (F), (I), and (L) are shown the RMSD plots of Cα carbons of the protein (blue diagram) and of the ligand (red diagram).

Using the structure with PDB ID 7JRN as a template, we docked the potent analogues Jun9-53-2, Jun9-72-2, and Jun9-75-4 in the SARS-CoV-2 PLpro drug-binding site. The stability of the docking poses was explored inside the hydrophobic S4 area using 100 ns MD simulations. The MD simulations show that the complexes formed are stable when the two methyl groups of the CH(CH3)-N(CH3) moiety are in the same side of the space, in a gauche position, in agreement with the observed RMSD values of ligands and protein which are both smaller than ca. 2 Å (Figure F,I,L).

Compared to GRL0617, in Jun9-53-2, Jun9-72-2, and Jun9-75-4 the carboxamide group has been replaced with a methylamino group increasing the ligand–receptor hydrogen bonding interaction strength due to the presence of a donor N–H+ group. Thus, in all complexes with Jun-compounds, the N–H+ group is engaged in strong ionic bonding interactions with the side chain carboxylic acid of Asp164 (Figure D,G,J,E,H,K) throughout the simulation, shifting the ligands from Q269 toward M208 in the S4 area (Figure D,G,J). In this binding orientation, the R166 side chain is moved to stabilize the side chain carboxylic acid anion of D164 with anionic hydrogen bonding (Figure D,G,J). All the ligands are stabilized inside the binding area by forming T-shaped π–π stacking between the ligand naphthalene ring and Tyr264 and hydrophobic interactions between naphthalene ring and Pro248, Pro247 (Figure D,G,J,E,H,K). In the ligand Jun9-75-4 the indole ring NH forms hydrogen bonding interactions with Gln269, while in ligand Jun9-72-2 the donor phenol hydroxyl group forms a hydrogen bond with the main chain NH group of L162, which is occasionally bridged with a water molecule (Figure D,G,J,E,H,K).

Conclusion

Given the tremendous impact of the COVID-19 pandemic, the SARS-CoV outbreak in 2003 was a dire warning that was gravely overlooked in retrospect. Looking forward, it is imperative that therapeutics are developed that are not only effective against SARS-CoV-2 but against future strains of similar coronaviruses. PLpro is a high-profile drug target, partially because it is highly conserved between SARS-CoV and SARS-CoV-2, sharing 83% sequence similarity. Inhibitors like GRL0617 are equally effective against both viruses, with a Ki of 0.49 μM and 0.57 μM, against SARS-CoV PLpro and SARS-CoV-2 PLpro.[23] Likewise, all critical active site residues that interact with GRL0617 are conserved. Consequently, the binding poses are nearly identical (Figure S5). These similarities would indicate that PLpro inhibitors might retain their activity against beta coronaviruses that might emerge in the future.

Previous attempts to discover SARS-CoV-2 PLpro inhibitors through HTS have failed to identify hits with improved enzymatic inhibition and cellular antiviral activity.[12,13] Structural analogues of GRL0617 were also designed and synthesized; however, none showed improved enzymatic inhibition.[22] Part of the reason for the difficulty in targeting SARS-CoV-2 PLpro is the lack of S1 and S2 pockets, which leaves only S3 and S4 pockets for inhibitor binding. The majority of the cysteine protease inhibitors are covalent inhibitors targeting the catalytic cysteine,[34] and it remains a challenge to develop noncovalent cysteine protease inhibitors with a similar potency as the covalent inhibitors. Among the reported SARS-CoV or SARS-CoV-2 PLpro inhibitors, GRL0617 is one of the most potent compounds. However, it had weak antiviral activity (Vero E6: EC50 = 23.64 μM; Caco2-hACE2: EC50 = 19.96 μM). In this study, we aim to identify more potent SARS-CoV-2 PLpro inhibitors through a HTS. On the basis of two promising hits Jun9-13-7 and Jun9-13-9, a library of analogues was designed and synthesized, among which several compounds had sub-micromolar IC50 values in the FRET-based enzymatic assay. To alleviate the burden of relying on BSL-3 facility to test the antiviral activity of PLpro inhibitors, we developed the cell-based FlipGFP-PLpro assay, which can be used to quantify the intracellular enzymatic inhibition of PLpro in a BSL-2 lab. The FlipGFP-PLpro assay is a close mimetic of the virus-infected cell in which PLpro cleaves its substrate in the native intracellular reducing environment. The advantage of the FlipGFP-PLpro assay over the standard FRET-based enzymatic assay is that it can rule out compounds that are either cytotoxic or membrane impermeable or nonspecifically modifying the catalytic cysteine through oxidation or alkylation. Our results showed there is a positive correlation between the results of FlipGFP-PLpro assay and the antiviral assay in both the Vero E6 and Caco2-hACE2 cells. In contrast, the correlation between the FRET assay results and the antiviral assay results is poor. The FlipGFP-PLpro assay can be performed in the BSL-2 setting, which alleviates the resources and financial burdens associated with screening a large number of compounds in the BSL-3 facility. This is expected to speed up the drug discovery process. In total, three PLpro inhibitors Jun9-72-2, Jun9-85-1, and Jun9-87-1 were identified as potent SARS-CoV-2 antivirals with EC50 values at or less than 10 μM when tested in both the Vero E6 and Caco2-hACE2 cell lines.

We also solved the X-ray crystal structure of the wild-type SARS-CoV-2 PLpro in complex with GRL0617. Binding of GRL0617 to SARS-CoV-2 induced a conformational change in the BL2 loop to the more closed conformation. In contrast, a larger inhibitor VIR251 stabilizes the BL2 loop in the open conformation.[9] The intrinsic flexibility of the BL2 loop implies that structurally diverse inhibitors might be able to fit in the S3–S4 pockets.

As shown by the MD simulations, the replacement of the carboxamide group in GRL0617 to the trialkyl ammonium in Jun9-53-2, Jun9-72-2, and Jun9-75-4 affects the binding interactions inside the receptor-binding region. In comparison to GRL0617, the N–H+ group in Jun9-53-2, Jun9-72-2, and Jun9-75-4 is engaged in strong ionic hydrogen bonding interactions with a side chain of Asp164, participating in another stabilizing ionic hydrogen bonding interactions with Arg166, which pulls the ligands inside the receptor-binding region from the hydrogen-bonded Gln269 to a new T-shaped π–π stacking with the Ty264 instead of Tyr268 in GRL0617. All the four ligands form hydrophobic interactions between naphthalene ring and Pro248. Overall, these features might explain the higher potency of Jun9-53-2, Jun9-72-2, and Jun9-75-4 compared to GRL0617.

In conclusion, the SARS-CoV-2 PLpro inhibitors discovered in this study represent promising hits for further development as SARS-CoV-2 antivirals, the FlipGFP-PLpro assay is a suitable surrogate for testing the cellular activity of PLpro inhibitors in the BSL-2 setting, and the results can be used to help prioritize leads for the antiviral assay.

Materials and Methods

Cell Lines and Viruses

VERO E6 cells (ATCC, CRL-1586) were cultured in Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 5% heat-inactivated FBS in a 37 °C incubator with 5% CO2. Caco2 cells expressing human ACE2 (Caco2-hACE2) were established by transducing Caco2 cells (ATCC HTB-37) with lentiviral particles derived with pWPI-IRES-Puro-Ak-ACE2 (a gift from Sonja Best; Addgene plasmid #154985).

SARS-CoV-2, isolate USA-WA1/2020 (NR-52281), was obtained through BEI Resources and propagated once on VERO E6 cells before it was used for this study. Studies involving the SARS-CoV-2 were performed at the UTHSCSA biosafety level-3 laboratory by personnel wearing powered air-purifying respirators.

Protein Expression and Purification

Detailed expression and purification of C-terminal His tagged SARS-CoV-2 PLPro (PLpro-His) were described in our previous publication.[8] Briefly, the SARS-CoV-2 papain-like protease (PLpro) gene (ORF 1ab 1564–1876) from strain BetaCoV/Wuhan/WIV04/2019 with E. coli codon optimization in the pET28b(+) vector was ordered from GenScript. The pET28b(+) plasmid was transformed into BL21(DE3) cells, and protein expression was induced with 0.5 mM IPTG when the OD600 was around 0.8 for 24 h at 18 °C. Then cells were harvested and lysed, the PLpro-His protein was purified with a single Ni-NTA resin column, and eluted PLpro-His was dialyzed against a 100-fold volume dialysis buffer (50 mM Tris [pH 7.5], 150 mM NaCl, 2 mM DTT) in a 10 000 kDa molecular weight cutoff dialysis tubing.

The expression and purification of untagged SARS-CoV-2 PLpro (PLpro) were carried out as follows: the SARS-CoV-2 PLpro gene (ORF 1ab 1564–1876) was subcloned from the pET28b(+) to pE-SUMO vector according to the manufacturer’s protocol (LifeSensors Inc., Malvern, PA). The forward primer with the Bsa I site is GCGGTCTCAAGGTGAAGTTCGCACCATCAAAGTTTTTACC; the reverse primer with a Xba I site is GCGGTCTCTCTAGATTACTTGATGGTGGTGGTGTAGCTGTTCTC. SUMO-tagged protein was expressed and purified as PLpro-His protein. The SUMO tag was removed by incubation with SUMO protease 1 at 4 °C overnight, and the free SUMO tag was removed by application of another round of Ni-NTA resin. The purity of the protein was confirmed with a SDS-PAGE gel.

The expression and purification of SARS-CoV-2 Mpro with unmodified N- and C-termini were reported in previous studies.[8]

Peptide Synthesis

The SARS-CoV-2 PLpro FRET substrate Dabcyl-FTLRGG/APTKV(Edans) and the SARS-CoV-2 Mpro FRET substrate Dabcyl-KTSAVLQ/SGFRKME(Edans) were synthesized by solid-phase synthesis through iterative cycles of coupling and deprotection using the previously optimized procedure.[35]

Ub-AMC and ISG15-AMC were purchased from Boston Biochem (catalog no. U-550-050 and UL-553-050, respectively).

Compound Synthesis and Characterization

Details for the synthesis procedure and characterization for compounds can be found in the Supporting Information.

Enzymatic Assays

The high-throughput screening was carried out in 384-well format. One microliter of 2 mM library compound was added to 50 μL of 200 nM PLpro-His protein in a PLpro reaction buffer (50 mM HEPES pH 7.5, 5 mM DTT and 0.01% Triton X-100) and was incubated at 30 °C for 1 h. The reaction was initiated by adding 1 μL of 1 mM PLpro FRET substrate. The end-point fluorescence signal was measured after 3 h incubation at 30 °C with a Cytation 5 image reader with filters for excitation at 360/40 nm and emission at 460/40 nm. The final testing compound concentration is ∼40 μM, and the FRET substrate concentration is ∼20 μM; a control plate as in Figure was included in every batch of screening.

The diversity compound library consisting of 50,240 compounds was purchased from Enamine (catalog no. 781270).

For the measurements of Km/Vmax: with Peptide-Edans as a substrate, the final PLpro protein concentration is 200 nM, and the substrate concentration ranges from 0 to 200 μM; with Ub-AMC as a substrate, the final PLpro protein concentration is 50 nM, and the Ub-AMC concentration ranges from 0 to 40 μM; with ISG15-AMC as a substrate, the final PLpro protein concentration is 2 nM, and the ISG15-AMC concentration ranges from 0 to 15 μM. The reaction was monitored in a Cytation 5 image reader with filters for excitation at 360/40 nm and emission at 460/40 nm at 30 °C for 1 h. The initial velocity of the enzymatic reaction was calculated from the initial 10 min enzymatic reaction and was plotted against the substrate concentrations in Prism 8 with a Michaelis–Menten function.

For the IC50 measurement with FRET peptide-Edans substrate: the reaction was carried out in 96-well format with 200 nM PLpro protein as described previously.[7,8] For the IC50 measurements with Ub-AMC or ISG15-AMC substrate, the reaction was carried out in 384-well format. The final PLpro protein concentration is 50 nM, and substrate concentration is 2.5 μM when Ub-AMC is applied; the final PLpro protein concentration is 2 nM, and substrate concentration is 0.5 μM when ISG15-AMC is applied.

For the Lineweaver–Burk plots of GRL0617, Jun9-13-7, and Jun9-13-9, the assay was carried as follows: 50 μL of 400 nM PLpro protein was added to 50 μL of reaction buffer containing testing compound and various concentrations of FRET peptide-Edans substrate to initiate the enzyme reaction. The initial velocity of the enzymatic reaction with and without testing compounds was calculated by linear regression for the first 10 min of the kinetic progress curve, and then plotted against substrate concentrations in Prism 8 with the Michaelis–Menten equation and linear regression of double reciprocal plot.

The main protease (Mpro) enzymatic assays were carried out in Mpro reaction buffer containing 20 mM HEPES pH 6.5, 120 mM NaCl, 0.4 mM EDTA, 20% glycerol, and 4 mM DTT as described previously.[7,8,26,27]

Cell-Based FlipGFP-PLpro Assay

Plasmid pcDNA3-TEV-flipGFP-T2A-mCherry was ordered from Addgene (catalog no. 124429). SARS-CoV-2 PLpro cleavage site LRGGAPTK or SARS-CoV-2 Mpro cleavage site AVLQSGFR was introduced into pcDNA3-FlipGFP-T2A-mCherry via overlapping PCRs to generate a fragment with SacI and HindIII sites at the ends. SARS-CoV-2 Mpro and PLpro expression plasmids pcDNA3.1 SARS2Mpro and pcDNA3.1 SARS2 PLpro were ordered from Genscript (Piscataway NJ) with codon optimization.

For transfection, 96-well Greiner plate (catalog no. 655090) was seeded with 293T cells to overnight 70–90% confluency. A total of 50 ng of pcDNA3-flipGFP-T2A-mCherry plasmid and 50 ng of protease expression plasmid pcDNA3.1 were used each well in the presence of transfection reagent TransIT-293 (Mirus). Three hours after transfection, 1 μL of testing compound was added to each well at 100-fold dilution. Images were acquired 2 days after transfection with a Cytation 5 imaging reader (Biotek) GFP and mCherry channels and were analyzed with Gen5 3.10 software (Biotek). SARS-CoV-2 PLpro protease activity was calculated by the ratio of GFP signal sum intensity over the mCherry signal sum intensity. The FlipGFP-PLP assay IC50 value was calculated by plotting the GFP/mCherry signal over the applied compound concentration with a four-parameter dose–response function in Prism 8. The mCherry signal alone was utilized to determine the compound cytotoxicity.

Differential Scanning Fluorimetry (DSF)

The thermal shift binding assay (TSA) was carried out using a Thermo Fisher QuantStudio 5 Real-Time PCR system as described previously.[7,8,26,27] Briefly, 4 μM SARS-CoV-2 PLpro protein (PLpro) in PLpro reaction buffer was incubated with 40 μM of compounds at 30 °C for 30 min. 1× SYPRO orange dye was added, and the fluorescence of the well was monitored under a temperature gradient range from 20 to 90 °C with 0.05 °C/s incremental step. Measured Tm was plotted against the compound concentration with one-site binding function in Prism 8.

Native Mass Spectrometry

Before MS analysis, the protein was buffer exchanged using two Micro Bio-Spin columns (Bio-Rad) and diluted into 0.2 M ammonium acetate to a concentration of 10 μM. Each drug was diluted with 100% ethanol to concentrations of 100, 200, and 400 μM. Imidazole, a charge reducing reagent, was added to each sample to stabilize the drug-bound state at a final concentration of 40 mM. For each sample, 0.5 μL of ligand was added and dried down in each tube prior to the addition of 0.5 μL of 40 mM DTT, 0.5 μL of 400 mM imidazole, and 4 μL of protein for a final concentration of 4 mM DTT and 8 μM protein. Final ligand concentrations were either 10, 20, or 40 μM.

Native mass spectrometry (MS) was performed using a Q-Exactive HF quadrupole-Orbitrap mass spectrometer with the Ultra-High Mass Range research modifications (Thermo Fisher Scientific) as described in our previous publications.[7,8] Data were deconvolved and analyzed with UniDec.[21]

Immunofluorescence Assay

An antiviral immunofluorescence assay was carried out as previously described.[8,36] Briefly, Vero E6 cells or Caca2-hACE2 cells in 96-well plates (Corning) were infected with SARS-CoV-2 (USA-WA1/2020 isolate) at a MOI of 0.1 in DMEM supplemented with 1% FBS. Immediately before the viral inoculation, the tested compounds in a 3-fold dilution concentration series were also added to the wells in triplicate. The infection proceeded for 24 h without the removal of the viruses or the compounds. The staining and quantification procedures are described in our previous publications.[8] Briefly, the cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton-100, blocked with DMEM containing 10% FBS, and stained with a rabbit monoclonal antibody against SARS-CoV-2 NP (GeneTex, GTX635679) and an Alexa Fluor 488-conjugated goat antimouse secondary antibody (ThermoFisher Scientific). Hoechst 33342 was added in the final step to counterstain the nuclei. Fluorescence images of approximately 10 000 cells were acquired per well with a 10× objective in a Cytation 5 (BioTek). The total number of cells, as indicated by the nuclei staining, and the fraction of the infected cells, as indicated by the NP staining, were quantified with the cellular analysis module of the Gen5 software (BioTek).

Crystallization and Structure Determination

SARS-CoV-2 PLpro-His (PLpro-His) protein was concentrated and loaded to a HiLoad 16/60 Superdex 75 size exclusion column (GE Healthcare) pre-equilibriated with 20 mM Tris pH 8.0 and 5 mM NaCl. Peak fractions were pooled and incubated with GRL0617 in a 1:1 molar ratio for 1 h at room temperature and then concentrated to 8 mg/mL. PLpro crystals were grown in a hanging-drop, vapor-diffusion apparatus by mixing 0.75 μL of 8 mg/mL PLpro-GRL0617 with 0.75 μl of well solution (30% PEG 4000, 0.2 M Li2SO4, and 0.1 M Tris pH 8.5). Crystals were transferred to a cryoprotectant solution containing 30% PEG 4,000, 0.2 M Li2SO4, 0.1 M Tris pH 8.5, and 15% glycerol, before being flash frozen in liquid nitrogen.

X-ray diffraction data for SARS-CoV-2 PLpro + GRL0617 was collected on the SBC 19-BM beamline at the Advanced Photon Source (APS) in Argonne, IL, and processed with the HKL2000 software suite.[37] The CCP4 versions of MOLREP were used for molecular replacement using a previously solved apo SARS-CoV-2 PLpro structure, PDB ID: 6WZU as a reference model.[38] Rigid and restrained refinements were performed using REFMAC, and model building was performed with COOT.[39,40] Protein structure figures were made using PyMOL (Schrödinger, LLC).

MD Simulations

MD simulations were carried out to the bound GRL0617, Jun9-53-2, Jun9-72-2, and Jun9-75-4 with PLpro prepared as described previously from the experimental structure of SARS-CoV-2 PLpro with GRL0617 (PDB ID 2JRN). Each complex was solvated using the TIP3P[41] water model. Using the “System Builder” utility of Schrodinger Desmond v.11.1, each complex was embedded in an orthorhombic water box extending beyond the solute 10 Å in the x,y,z direction leading to 14 500 waters. Na+ and Cl– ions were placed in the water phase to neutralize the systems and to reach the experimental salt concentration of 0.150 M NaCl. The total number of atoms was ca. 48 000.

The OPLS-2005 force field[42,43] was used to model all protein and ligand interactions and lipids. The particle mesh Ewald method (PME)[44,45] was employed to calculate long-range electrostatic interactions with a grid spacing of 0.8 Å. van der Waals and short-range electrostatic interactions were smoothly truncated at 9.0 Å. The Langevin thermostat[46] was utilized to maintain a constant temperature in all simulations, and the Berendsen barostat[47] was used to control the pressure. Periodic boundary conditions were applied (73 × 102 × 65) Å3. The equations of motion were integrated using the multistep RESPA integrator[48] with an inner time step of 2 fs for bonded interactions and nonbonded interactions within a cutoff of 9 Å. An outer time step of 6.0 fs was used for nonbonded interactions beyond the cutoff. Each system was equilibrated in MD simulations with a default protocol for water-soluble proteins provided in Desmond, which consists of a series of restrained MD simulations designed to relax the system while not deviating substantially from the initial coordinates.

The first simulation was a Brownian dynamics run for 100 ps at a temperature of 10 K in the NVT (constant number of particles, volume, and temperature) ensemble with solute heavy atoms restrained with a force constant of 50 kcal mol Å–2. The Langevin thermostat[46] was applied in the NVT ensemble and a MD simulation for 12 ps with solute heavy atoms restrained with a force constant of 50 kcal mol Å–2. The velocities were randomized, and MD simulation for 12 ps was performed in the NPT (constant number of particles, pressure, and temperature) ensemble and a Berendsen barostat[47] with solute heavy atoms equally restrained at 10 K and another one at 300 K. The velocities were again randomized, and unrestrained MD simulation for 24 ps was performed in the NPT ensemble. The above-mentioned equilibration was followed by 100 ns simulation without restraints. Two simulations were performed in a workstation with GTX 970. The visualization of the produced trajectories and structures was performed using Maestro or programs Chimera[49] and VMD.

Safety Statement

No unexpected or unusually high safety hazards were encountered.