key: cord-0708706-04vmyj4d authors: Li, Lingyu; Ma, Liyan; Hu, Yue; Li, Xiaoxue; Yu, Meng; Shang, Hai; Zou, Zhongmei title: Natural biflavones are potent inhibitors against SARS-CoV-2 papain-like protease date: 2021-10-12 journal: Phytochemistry DOI: 10.1016/j.phytochem.2021.112984 sha: 2e9a3443e80e59756b821d80001817e01c04f1a4 doc_id: 708706 cord_uid: 04vmyj4d Papain-like protease (PL(pro)) is a key enzyme encoded by SARS-CoV-2 that is essential for viral replication and immune evasion. Significant suppression of viral spread and promotion of antiviral immunity can be achieved by inhibition of PL(pro), revealing an inspiring strategy for COVID-19 treatment. This study aimed to discover PL(pro) inhibitors by investigating the national compound library of traditional Chinese medicines (NCLTCMs), a phytochemical library comprising over 9000 TCM-derived compounds. Through virtual screening and enzymatic evaluations, nine natural biflavones were confirmed to be effective PL(pro) inhibitors with IC(50) values ranging from 9.5 to 43.2 μM. Pro-ISG15 cleavage assays further demonstrated that several biflavones exhibited potent inhibitory effects against PL(pro)-mediated deISGylation, a key process involved in viral immune evasion. Herein, we report the discovery, antiviral evaluation, structure-activity relationship elucidation and molecular docking investigation of biflavones as potent inhibitors of SARS-CoV-2 PL(pro). Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent of COVID-19, continues to ravage mankind throughout the world. As of 25 July 2021, there have been over 192 million confirmed cases of COVID-19 including 4.1 million deaths reported to the World Health Organization. Nonetheless, there is still a lack of effective treatment for COVID-19, and therapeutic drugs are desperately needed. PL pro is a key enzyme encoded by SARS-CoV-2 that recognizes and cleaves the LXGG consensus sequence (Leu-X-Gly-Gly, X refers to unspecific amino acids) of both viral and host proteins (Klemm et al., 2020) . Cleavage of the viral polyprotein at the LXGG site generates non-structural proteins (nsp1-3) to participate in the assembly of the viral replicase complex, which initiates replication and transcription of the viral genome (Hu et al., 2021a) (Fig. 1A) . Beyond the ability to process viral proteins, ample evidence indicates that PL pro can also manipulate host proteins to evade antiviral responses. For example, recent studies demonstrated that SARS-CoV-2 PL pro could bind to, interact with and finally cleave the interferon stimulated gene product-15 (ISG15) modifier from melanoma differentiation-associated protein 5 (MDA5) to escape immune surveillance. MDA5 is the crucial viral RNA sensor to detect cytosolic SARS-CoV-2 RNA and thereby triggers innate immune response in an ISG15-dependent manner (Liu et al., 2021a; Yin et al., 2021) . However, SARS-CoV-2 PL pro is capable of removing the ISG15 modifier from MDA5 by cleaving the LRGG motif at the Cterminus of ISG15 (Fig. 1B) , thus terminating MDA5-mediated antiviral signalling. This event is termed deISGylation, a viral evolutionary strategy to evade host innate immunity. Likewise, studies by Shin et al. (2020) . showed that PL pro could facilitate immune evasion by deISGylating interferon regulation factor 3 (IRF3), which in turn disrupts type I interferon-mediated antiviral signalling. In addition to interfering with ISG15 modification, SARS-CoV-2 PL pro could also cleave Lys48 ubiquitin chains from host proteins to promote immune evasion but with much less preference and lower selectivity than ISG15 substrates (Freitas et al., 2020; Klemm et al., 2020) . searching for novel PL pro inhibitors, especially those with inhibitory potential against the deISGylation activity of PL pro . Traditional Chinese medicine (TCM) is an extraordinary reservoir of antiviral agents, from which a wide variety of phytochemicals have been proven to be effective PL pro inhibitors. For instance, tanshinones derived from Salvia miltiorrhiza were found to be specific and selective inhibitors of PL pro (Park et al., 2012b) . Furthermore, flavonoids isolated from Angelica keiskei (Park et al., 2016) and Broussonetia papyrifera (Park et al., 2017) displayed significant inhibition of PL pro . In addition, polyphenols from Paulownia tomentosa demonstrated dose-dependent inhibitory effects on both the proteolytic and deubiquitination activities of PL pro (Cho et al., 2013) . In addition to the above phytochemicals, previous studies also identified diarylheptanoids from Alnus japonica (Park et al., 2012a) and cinnamic amides from Tribulus terrestris (Song et al., 2014) as potent inhibitors of PL pro . Enlightened by this, we aimed to discover novel inhibitors of SARS-CoV-2 PL pro by investigating NCLTCMs, a phytochemical library constructed by our group currently possessing over 9000 entities of TCM-derived compounds. For the rapid search and identification of PL pro inhibitors, structure based virtual screening was initially performed to filter potential candidates. The hit compounds were then subjected to enzymatic evaluations to confirm their inhibitory activities. Finally, a panel of natural biflavones (Fig. 2, 1-9 ) ranked as the most potent PL pro inhibitors from NCLTCMs, with anti-proteolytic IC50 values ranging from 9.5 to 43.2 μM. Gratifyingly, pro-ISG cleavage assays further demonstrated that several biflavones exhibit significant inhibition of the deISGylation activity of PL pro , offering good prospects in attenuating PL pro -mediated immune evasion. It is also worth mentioning that all biflavones in this study were derived from TCM. Among them, amentoflavone (1), ginkgetin (3), CoV-2 (Hu et al., 2021b; Zhang et al., 2020) . Our findings suggested that these biflavones might act as antiviral ingredients of the above TCMs by inhibiting PL pro . This article also reports a detailed kinetic investigation, structure-activity relationship elucidation and molecular docking analysis of biflavones. Structure based virtual screening is an effective strategy for lead compound discovery (Ghosh et al., 2006) . The present study established a molecular docking model based on the existing crystal structure of SARS-CoV-2 PL pro (PDB: 7JRN) and performed virtual screening of all 9032 entries of NCLTCMs. The CDOCKER score (indicated as -CDOCKER interaction energy) was set as the criterion to filter candidates, and molecules that scored over 30 were selected as hit compounds. The virtual screening process yielded 152 hit compounds, which were subjected to fluorogenic enzymatic assays to corroborate their inhibitory potencies. The assay utilized Z-RLRGG-AMC, a pentapeptide resembling the consensus cleavage sequence of PL pro labelled with 7-amino-4-methylcoumarin (AMC) at the C-terminus. The AMC motif was non-fluorescent when covalently conjugated, but became dramatically fluorescent upon cleavage by PL pro , thus enabling efficient determination of antiproteolytic activity (Ratia et al., 2008) . To eliminate promiscuous inhibitors, the assays were performed in the presence of 10 mM dithiothreitol (DTT) and 100 mM sodium chloride, for which DTT was employed as a reducing agent and electrophile trap, while sodium chloride was added to preclude unspecific electrostatic interactions. Psoralidin, a previously reported natural inhibitor of SARS-CoV PL pro (Kim et al., 2014) , was selected as the positive control. By this method, we determined the anti-proteolytic activities for all hit compounds, 46 of which achieved 50% inhibition at a 50 μM concentration. Among them, a panel of natural biflavones (1-9) were found to be most prominent, with IC50 values ranging from 9.5 to 43.2 μM. Detailed results for virtual screening and experimental evaluation are listed in Table 1 , including CDOCKER scores, binding energies, binding interactions and IC50 values. To gain kinetic insights, inhibition constants (Ki) for all biflavones were determined by Dixon plots. As shown in Table 1 . The intersections also gave clues for inhibition mode, by which upper x-axis intersections indicated competitive or mixed type inhibition, as in the cases of all biflavones except morelloflavone (6). The intersection for 6 fell on the x-axis (Fig. 3F) , suggesting that 6 alone manifested non-competitive inhibition. Explanations for this phenomenon will be discussed in depth below. In summary, a panel of natural biflavones were identified as the most potent PL pro inhibitors from NCLTCMs via virtual screening and experimental corroboration. Because a variety of flavones were proven to be potent PL pro inhibitors in previous reports (Cho et al., 2013; Park et al., 2016; Park et al., 2017) , the biflavones from NCLTCMs attracted our particular interest and inspired further investigation. To correlate the chemical structure of the biflavones with their inhibitory potency against PL pro , we investigated the structure-activity relationship (SAR) for the biflavones. Since biflavones are structurally composed of two flavone monomers connected by C-C bonds (1-6) or C-O-C bonds (7-9), as indicated by the thick red lines in Fig. 2 , the influence of connection type was first examined. The C-O-C connections revealed a more pronounced effect on anti-proteolytic activity than the C-C connections. As presented in Table 1 , 4′-O-6′′-connected biflavone 7 displayed the strongest activity with an IC50 value of 9.5 μM. Another biflavone with the same connection type (8) showed a milder yet tolerated IC50 value of 26.3 μM. Meanwhile, 4′-O-3′′′-connected biflavone 9 ranked as the third-best inhibitor with an IC50 value of 22.8 μM. In contrast, C-C connections were less tolerated, and only biflavone 1 demonstrated a satisfactory IC50 value of 13.0 μM, whereas the IC50 values for all other C-C-type biflavones (2-6) ranged from 29.8 to 43.2 μM, which were all lower than those of C-O-C-type biflavones. To further explore the influence of the connection bond, the monomeric counterparts of biflavones, apigenin and acacetin, were also assessed for their antiproteolytic activities. However, dramatic loss of activity was observed. The IC50 values for the monomers decreased to 75.7 to 91.2 μM, which was 8-to 10-fold weaker than that of dimer 7, suggesting that the dimeric skeleton was indispensable for PL pro inhibition. We then surveyed the substitution effect of the biflavones, and found that the presence of hydroxy groups was more beneficial than methoxy groups. For instance, of all 8-3′′′-connected biflavones (1-5), non-methylated compound 1 conferred the most potent activity with an IC50 value of 13.0 μM, whereas replacement of hydroxy groups J o u r n a l P r e -p r o o f with methoxy groups led to a 2.3-to 3.3-fold reduction in anti-proteolytic activity, as evidenced by the IC50 values of 29.8 to 43.2 μM for 2-5. Similarly, 7 and 8 both characterized a 4′-O-6′′-connected skeleton, nonetheless, non-methylated biflavone 7 showed 2.8-fold more potent activity than its di-methylated counterpart (8), with IC50 values of 9.5 μM and 26.3 μM for 7 and 8, respectively. We next analysed the binding mode of the biflavones with the aid of molecular docking. The interactions between SARS-CoV-2 PL pro and a representative C-C-type biflavone (1) were first explored. As illustrated in Fig. 4A , biflavone 1 could dock well into the P3/P4 region of the substrate-binding pocket of PL pro with a "two-winged" pattern, by which the two flavone fragments concurrently occupy the P3 and P4 pockets to block substrate access. Multiple hydrogen bonds dominated the binding interactions. As depicted in Fig. 4D The "two-winged" binding pattern of biflavone revealed a plausible explanation for the dimeric-dependent inhibition, in which the inability to co-occupy the P3 and P4 pockets by flavone monomers accounted for 8-to 10-fold activity loss compared to dimer 7. In the case of morelloflavone (6), the "two-winged" binding pattern was impaired by a 3-8′′′-connection. With the bulky flavone motif substituted at the 3-position, the resulting steric hindrance significantly impeded the binding of 6 with the substratebinding pocket (Fig. 4C/4F) . As a consequence, 6 only formed one hydrogen bond with PL pro , resulting in a poor docking score of 48.2 and a weak binding energy of -91.2 kcal/mol for 6. In agreement with these findings, 6 exhibited mild anti-proteolytic activity with an IC50 value of 36.4 μM, which was the second least active of all nine biflavones. Moreover, the partial incompetency of 6 for the PL pro substrate-binding pocket was further supported by its non-competitive inhibition profile, as determined by Dixon plot (Fig. 3F) . Given the pivotal role of deISGylation in PL pro -mediated immune evasion, we next evaluated whether biflavones could inhibit the deISGylation activity of PL pro by pro-ISG15 cleavage assays. Pro-ISG15 is the precursor of ISG15 harbouring the LRGG↓TEPGGRS sequence at the C-terminus, which could be cleaved by PL pro after the LXGG recognition motif. The cleavage led to a reduction in molecular weight by approximately 0.8 kDa, and the corresponding protein band shift could be efficiently detected and quantified by SDS-PAGE analysis (Klemm et al., 2020; Shin et al., 2020) . Accordingly, inhibition of deISGylation could be determined by the proteolysis ratio of pro-ISG15. Using the above approach, we assessed the inhibitory activities of all biflavones at an initial concentration of 20 μM. In line with recent studies (Klemm et al., 2020; Shin et al., 2020) , SARS-CoV-2 PL pro demonstrated robust deISGylation activity, by which the proteolysis ratio of pro-ISG15 exceeded 95% in a 10 min period (Fig. 5A, second lane). However, the deISGylation process was significantly inhibited in the presence of biflavones, as shown in Fig. 5E . Notably, four biflavones (3, 4, 7 and 9 Biflavones with inhibition rates greater than 50% at 20 μM (2-4 and 7-9) were further investigated for their inhibitory potencies at decreased concentrations of 10 μM, 5 μM and 2.5 μM (Fig. 5B-5D ). All of the tested compounds displayed dose-dependent inhibition, and the inhibition rate is plotted in Fig. 5F-5G . Analogous to anti-proteolytic activity, inhibitory activity against deISGylation also favoured C-O-C connections. As shown in Fig. 5G, at a concentration of 10 μM, C-O-C-type biflavones 7, 8, and 9 achieved 65.6% to 98.0% inhibition, whereas the inhibition rate for C-C-type biflavones 2, 3, and 4 was only 53.2% to 71.1% (Fig. 5F ). When we further reduced the concentration to 5 μM, none of the biflavones except a C-O-C-type biflavone (9) reached 50% inhibition. Surprisingly, 9 even achieved 60.7% inhibition at concentrations down to 2.5 μM, demonstrating promising potential to attenuate PL promediated deISGylation. As mentioned above, 4′-O-methylochnaflavone (9) is a naturally occurring biflavone in Lonicera japonica Thunb. (common name: honeysuckle). L. japonica is not only the most prescribed antiviral TCM, but also a food-medicine herb widely used in daily tea drinks, soft drinks and cosmetics (Shang et al., 2011) . Notably, abundant studies indicated that L. japonica is effective for coronaviruses. For instance, cytopathic morphology-based assays by Wu et al. (2004) . demonstrated that L. japonica extracts had significant antiviral effects against SARS-CoV. Furthermore, the antiviral efficacy of L. japonica was confirmed by accumulating clinical evidence. During the battle against SARS-CoV-2, greater than 85% of COVID-19 patients in China received TCM treatment and clinical results revealed that TCM could significantly alleviate the symptoms of mild patient (Wang et al., 2021; Yang et al., 2020) . Among these TCMs, two L. japonica formulas, namely "Lianhua Qingwen The X-ray structure of wild-type PL pro of SARS-CoV-2 complexed with a noncovalent small molecule inhibitor was obtained from the PDB database (PDB ID: 7JRN). The binding site was defined as an 8 Å radius sphere around the centroid of the co-crystallized ligand. The SMILES strings of 9032 NCLTCMs compounds were imported into BIOVIA Discovery Studio 2019 (Dassault Systèmes, San Diego, USA) to build a ligand database, and the ligands were further minimized using the prepare ligands protocol as implemented in Discovery Studio software. The virtual screening process was performed using the CDOCKER algorithm, followed by CHARMM27 force field minimization and binding energy calculations with an explicit solvent model. The hydrogen bond distance was measured using the measure wizard of PyMOL 2.4 (The PyMOL Molecular Graphics System, Schrödinger, LLC). The 3D illustrations of the ligand-protein interactions were rendered in PyMOL. All of the fluorogenic measurements were performed in triplicate, and the blotting experiments were performed in duplicate. The IC50, Ki and inhibition rate of pro-ISG15 cleavage are expressed as the mean ± standard error (SD) if not otherwise specified. Lingyu Li performed all experiments and drafted the manuscript; Liyan Ma performed part of the activity determination work; Yue Hu and Xiaoxue Li participated in the collection and preparation of samples; Meng Yu and Hai Shang participated in the virtual screening process; Zhongmei Zou designed the study and revised the manuscript. The authors declare no conflict of interest. Diarylheptanoids from Alnus japonica inhibit papain-Like protease of severe acute respiratory syndrome coronavirus Tanshinones as selective and slowbinding inhibitors for SARS-CoV cysteine proteases Chalcones isolated from Angelica keiskei inhibit cysteine proteases of SARS-CoV Evaluation of polyphenols from Broussonetia papyrifera as coronavirus protease inhibitors A noncovalent class of papain-like protease/deubiquitinase inhibitors blocks SARS virus replication Potent, Novel SARS-CoV-2 PLpro Inhibitors Block Viral Replication in Monkey and Human Cell Cultures Papain-like protease regulates SARS-CoV-2 viral spread and innate immunity Flavonoids from the aerial parts of Lonicera japonica Papain-Like Protease (PLpro) Inhibitory Effects of Cinnamic Amides from Tribulus terrestris Fruits Efficacy and Safety of Traditional Chinese Medicine in Coronavirus Disease COVID-19): A Systematic Review and Meta-Analysis Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia (Trial Version 7) Small molecules targeting severe acute respiratory syndrome human coronavirus Traditional Chinese Medicine in the Treatment of Patients Infected with 2019-New Coronavirus (SARS-CoV-2): A Review and Perspective Naturally Occurring Cytotoxic [3′→8″]-Biflavonoids from Podocarpus nakaii MDA5 governs the innate immune response to SARS-CoV-2 in lung epithelial cells The efficacy and safety of Jinhua Qinggan granule (JHQG) in the treatment of coronavirus disease 2019 (COVID-19): A protocol for systematic review and meta analysis The authors are grateful to the financial support of the National Mega-Project for Innovative Drugs (2019ZX09735002).