key: cord-270622-aofva2ab
authors: Li, Qizhang; Wang, Zhiying; Zheng, Qiang; Liu, Sen
title: Potential clinical drugs as covalent inhibitors of the priming proteases of the spike protein of SARS-CoV-2
date: 2020-08-26
journal: Comput Struct Biotechnol J
DOI: 10.1016/j.csbj.2020.08.016
sha: 
doc_id: 270622
cord_uid: aofva2ab

In less than eight months, the COVID-19 (coronavirus disease 2019) caused by the SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) virus has resulted in over 20,000,000 confirmed cases and over 700,000 deaths around the world. With the increasing worldwide spreading of this disease, the lack of effective drugs against SARS-CoV-2 infection makes the situation even more dangerous and unpredictable. Although many forces are speeding up to develop prevention and treatment therapeutics, it is unlikely that any de novo drugs will be available in months. Drug repurposing holds the promise to significantly save the time for drug development, since it could use existing clinic drugs to treat new diseases. Based on the “steric-clashes alleviating receptor (SCAR)” strategy developed in our lab recently, we screened the library of clinic and investigational drugs, and identified nine drugs that might be repurposed as covalent inhibitors of the priming proteases (cathepsin B, cathepsin L, and TMPRSS2) of the spike protein of SARS-CoV-2. Among these hits, five are known covalent inhibitors, and one is an anti-virus drug. Therefore, we hope our work would provide rational and timely help for developing anti-SARS-CoV-2 drugs.

27 Since its outbreak in December 2019, the COVID-19 (coronavirus disease 2019) disease from the 28 infection of the SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) virus has caused 29 over 20,000,000 confirmed cases and over 700,000 deaths in over 180 countries/regions as of August 30 12, 2020 (https://coronavirus.jhu.edu). Although most COVID-19 patients can recover from the 31 disease, some of the severe patients might suffer from long-term health issues including irreversible 32 lung damages [1] and fertility compromise [2] . In addition to the devastating health crisis, the 33 ongoing COVID-19 pandemic is also inflicting heavy losses on the global economy due to city 34 lockdowns and glitches in supply chains (https://www.imf.org/en/Topics/imf-and-covid19). Facing 35 the escalating risk from COVID-19, the whole world is intensively working on the discovery of 36 prevention and treatment options for SARS-CoV-2 infection [3] . According to the statistics of the 

According to a recent study [9], both the 57 endosomal cysteine proteases cathepsin B/L (CatB/L) and the transmembrane protease serine type 2 58 (TMPRSS2) can prime the S protein of SARS-CoV-2. Meanwhile

Therefore, similar with SARS-CoV [11] and MERS-CoV

Historically, the drug discovery practice mainly focuses on non-covalent drugs due to potential off-64 target effects and toxicity issues of irreversible covalent drugs

However, recent years have 65 witnessed the resurgence of covalent drugs because many people have realized that compared to non-66 covalent drugs, covalent drugs might have extra advantages including: (i) better biochemical 67 efficiency since they are more competitive than non-covalent endogenous substrates and co-factors 68 [14]; (ii) lower patient burden and less drug resistance due to lower and less frequent dosing

To 70 help the discovery of covalent drugs, we previously established a "steric-clashes alleviating receptor 71 (SCAR)" strategy [17] for the in silico docking and screening of covalent drugs enlightened by in 72 silico protein design

org) containing approved and in-trial drugs with known warhead groups targeting 77 cysteine (CatB/CatL) or serine (TMPRSS2). Then, SCARdock was used to computationally screen 78 potential covalent inhibitors of human CatB, CatL, and TMPRSS2. After careful filtering and 79 evaluation, we identified five (trapoxin B, neratinib, HKI-357, domatinostat and (Z)-dacomitinib) 80 potential covalent inhibitors for CatB, three (neratinib, HKI-357 and (Z)-dacomitinib) for CatL

The 3D structures of the indicated proteins were downloaded from the RCSB database 96

The small molecule inhibitors were extracted from the complex structures mentioned above. The 103 structures were visually checked, and incorrect bonds/atoms were manually corrected in IQmol 104 (version 2.14.0). MGLTools (version 1.5.6) was used to generate the corresponding PDBQT files for

The small 110 molecules were docked into the corresponding pockets of the proteins with AutoDock Vina (version 111 1.1.2) [27]. The docking process did not consider the flexibility of the protein. The space coordinates 112 of the S atom (for Cys) or the O atom (for Ser) in the wild-type protein were used for calculating the 113 atom distances of the bonding atoms in the warhead groups. The distance cutoff between the bonding 114 atoms and the S/O atom in the protein was set to 1.8 Å, indicating that the conformation with a 115 distance above 1.8 Å is not accepted. Since the results (Table 1) had distances between 1.2-1.8 Å, no 116 score punishment for steric conflict was applied for the cases in this study. For each ligand, top 10 117 poses were used for evaluation

Following the SCARdock 125 protocol [17], these reactive residues were computationally mutated to glycine to generate the SCAR 126 proteins for docking

From the RCSB database, we obtained four X-ray structures of human CatB and seventeen X-ray 131 structures of human CatL (Supplementary Table 1). To identify the most suitable structures for

Supplementary Table 1), the protein structure from 1CSB (PDB ID) was chosen for CatB, and the 135 protein structure from 5MAE (PDB ID) was chosen for CatL. The human TMPRSS2 structure was 136 obtained from the SWISS-MODEL repository since there were no X-ray structures available. These 137 structures were then used for SCARdock screening, and after distance and score filtering, we 138 identified five potential covalent inhibitors for CatB

Overall, three cysteine covalent warheads and two serine covalent warheads were observed in the 140 identified drugs (Figure 1F)

Trapoxin B contains an epoxide 144 warhead. For this warhead, the nucleophilic attack might occur on the ring carbon next to the 145 carbonyl carbon, and then a covalent bond can form between the sulfur atom of Cys29 and the 146 bonding carbon of the oxirane moiety, accompanied by the ring opening and the formation of a 147 hydroxyl group (Figure 1F & Figure 3A). Neratinib (HKI-272) and HKI-357 contain a nitrile 148 warhead. For this warhead, a covalent thioimidate bond might form at the electrophilic nitrile carbon 149 after the attack of the cysteine sulfur atom (Figure 3B & 3C). Domatinostat (4SC-202) and (Z)-150 dacomitinib are amide-based ligands. A covalent bond might form between the β-carbon and the 151 cysteine sulfur atom (Figure 3D & 3E)

Three drugs, i.e. neratinib (HKI-272), HKI-357 and (Z)-dacomitinib, were identified as potential

This result is not unexpected though, since CatB and CatL are homologous and 158 have very similar 3D structures (Figure 1B & 1C). Thus, neratinib (HKI-272), HKI-357 and (Z)-159 dacomitinib are potential covalent inhibitors for both CatB and CatL. Although the docked poses 160 were slightly different on these two proteins, the warheads of these drugs were also at the positions 161 suitable for covalent bonding (Figure 3F-H)

)-boceprevir and (R)-boceprevir, were 165 identified as potential TMPRSS2 covalent inhibitors (Figure 5A). Lodoxamide, aceneuramic acid and 166 aleplasinin possess an α-ketoacid group, whereas (S)-and (R)-boceprevir have an α-ketoamide group

Both of these groups could be used as covalent warheads targeting serine (Figure 1F)

Figure 5B, the warheads in these inhibitors are positioned well for the covalent bonding between the 169 bonding atoms and the hydroxyl group of TMPRSS-Ser441

The ongoing 174 COVID-19 pandemic caused by the SARS-CoV-2 virus is undoubtedly reminding us his warning 175 was not a hoax. The high infection rate and mortality ratio of COVID-19 are unexpected [31], and the 176 SARS-CoV-2 virus has made 2020 a difficult year for a lot of people in the world. Although this 177 virus would unlikely kill over 10 million people, it is still posing an unprecedent threat to both the 178 health and the economy of the whole world due to the shortage of effective prevention and treatment 179 therapeutics so far

During the infection of coronaviruses, the spike (S) protein mediates host recognition and binding

However, the S protein needs to be cleaved and primed by the host cell before the virus can enter and 184 hijack the host cell

CatL, and TMPRSS2 of the human cell can prime the S protein of 185 SARS-CoV-2

targeting these priming proteases 186 might be an effective choice to disrupt the infection of this virus. Based on our recent work 187 [17,19,20], we adopted the SCARdock protocol to repurpose clinic drugs as potential inhibitors of 188 these priming proteases in this study. We identified several clinic drugs that might be useful as the 189 covalent inhibitors of CatB

More interestingly, all these drugs 194 are covalent pan-HER (human epidermal growth factor receptor) kinase inhibitors targeting the 195 nucleophilic cysteine in the ATP binding site of EGFR and/or HER2 [34]. This fact indicated that 196 these three molecules are electrophilic, and it is highly possible that they can form covalent bonds 197 with the indicated cysteine residues of CatB and CatL if the non-covalent binding affinity is high 198 enough. Nontheless, we want to note that in our docking results, the nitrile groups

However, there were docked poses of these 201 two drugs with suitably positioned acrylamide groups in less optimal poses (Supplementary Figure 202 1). In addition, domatinostat (4SC-202) and trapoxin B were identified as potential CatB covalent 203 inhibitors. Interestingly, both domatinostat (4SC-202) and trapoxins B are the inhibitors of histone 204 deacetylases (HDACs), and they have been used for the treatment of

HKI-272), HKI-357 and dacomitinib might be the most

As of the identified hits of TMPRSS2, the most attractive drug might be boceprevir, which is a first-208 generation inhibitor of hepatitis C virus non-structural protease 3 (HCV NS3) [37]. Based on the 209 docking score and the SCAR enriching score (Table 1), (S)-boceprevir might be better than (R)-210 boceprevir. Since boceprevir has anti-virus activity

212 aceneuramic acid is used for the treatment of hereditary inclusion body myopathy [39], and 213 aleplasinin was developed to treat Alzheimer disease [40]. A note is that the screening of TMPRSS2 214 inhibitors was based on a homology model of human TMPRSS2, which might make the result not as 215 reliable as the CatB/L cases. However

A theoretical method to evaluate if a ligand could bind to two different proteins is to compare the 218 similarity of the binding pockets of the target proteins

Presumably, the protein-ligand 219 interactions between a same ligand and different proteins should be similar, although the extent to 220 which should vary on a case-by-case basis

Therefore, to validate the binding potential of our 221 computational hits with the target proteins, the ligand-protein interactions were generated

This analysis also 225 showed that (S)-boceprevir formed more interactions with TMPRSS2 than (R)-boceprevir did, which, 226 in agreement with their docking scores and SCAR enriching scores

Taken together, using our SCARdock protocol, we identified nine drugs that might be repurposed as 229 the covalent inhibitors of the priming proteases of the S protein of SARS-CoV-2. Among these nine 230 drugs, neratinib (HKI-272), HKI-357, dacomitinib, and boceprevir might be highly potential with 231 moderate side effects

did the 237 computational work

This work was supported by the grants from National Natural Science Foundation of China 241 (31670768, 31971150)

COVID-19): Lessons From Severe Acute Respiratory Syndrome 250 and Middle East Respiratory Syndrome

Rising Concern on Damaged Testis of COVID-19 Patients

Evaluation and 255 Treatment Coronavirus

Deconstructing the drug development process: the new face of innovation

A bibliometric review of drug repurposing

Drug repurposing: 261 progress, challenges and recommendations

More than 80 clinical trials launch to test coronavirus treatments

Remdesivir 266 for the Treatment of Covid-19 -Preliminary Report

CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven 270

TMPRSS2 and 272 ADAM17 cleave ACE2 differentially and only proteolysis by TMPRSS2 augments entry 273 driven by the severe acute respiratory syndrome coronavirus spike protein

Simultaneous treatment of 276 human bronchial epithelial cells with serine and cysteine protease inhibitors prevents severe 277 acute respiratory syndrome coronavirus entry

Middle East Respiratory Syndrome Coronavirus 280 Infection Mediated by the Transmembrane Serine Protease TMPRSS2

The resurgence of covalent drugs

Drug discovery considerations in the development of 285 covalent inhibitors

Ligand Conformational Bias Drives 288

Enantioselective Modification of a Surface-Exposed Lysine on Hsp90

Identification of the 291 Clinical Development Candidate MRTX849, a Covalent KRASG12C Inhibitor for the 292 Treatment of Cancer

Discovery of Covalent Ligands via Noncovalent Docking 294 by Dissecting Covalent Docking Based on a "Steric-Clashes Alleviating Receptor (SCAR

Discovery of novel inhibitors of human S-297 adenosylmethionine decarboxylase based on in silico high-throughput screening and a non-298 radioactive enzymatic assay

Potential covalent drugs targeting the main protease of the SARS-300

CoV-2 coronavirus

Repurposing Clinical Drugs as AdoMetDC Inhibitors 302 Using the SCAR Strategy

Trapoxin, an antitumor cyclic 304 tetrapeptide, is an irreversible inhibitor of mammalian histone deacetylase

Optimization of 6,7-Disubstituted-4-(arylamino)quinoline-3-carbonitriles as Orally Active

Irreversible Inhibitors of Human Epidermal Growth Factor Receptor-2 Kinase Activity

Irreversible protein kinase inhibitors

A novel molecular mechanism to explain mutations of the HCV 314 protease associated with resistance against covalently bound inhibitors

ZINC 15 -Ligand Discovery for Everyone

ROSETTA3: an 319 object-oriented software suite for the simulation and design of macromolecules

AutoDock Vina: Improving the speed and accuracy of docking with a 322 new scoring function, efficient optimization, and multithreading

Specialized roles for cysteine cathepsins in health and 325 disease

A Review of Small Molecule Inhibitors and Functional Probes of 327 Human Cathepsin L

Hepatocyte growth factor is 329 a preferred in vitro substrate for human hepsin, a membrane-anchored serine protease 330 implicated in prostate and ovarian cancers

Real estimates of mortality 333 following COVID-19 infection

Rapid repurposing of drugs for COVID-19

The P2/P2′ sites affect the substrate cleavage of TNF-α 338 converting enzyme (TACE)

Small-molecule EGFR tyrosine kinase 340 inhibitors for the treatment of cancer

RETRACTED: Design 343 and synthesis of CHAP31, trapoxin B and HC-toxin based bicyclic tetrapeptides disulfide as 9 344 potent histone deacetylase inhibitors

Phase I 347 study of domatinostat (4SC-202), a class I histone deacetylase inhibitor in patients with 348 advanced hematological malignancies

The effect of the first-351 generation HCV-protease inhibitors boceprevir and telaprevir and the relation to baseline 352 NS3 resistance mutations in genotype 1: experience from a small Swedish cohort

Efficacy of 355 lodoxamide eye drops on mast cells and eosinophils after allergen challenge in allergic 356 conjunctivitis

Clinical trial data available for UX001, aceneuramic acid extended-release

The Apparent uPA/PAI-1 Paradox in Cancer: More than 360 Meets the Eye

362 Identify drug repurposing candidates by mining the protein data bank. Briefings in 363

Proteins are shown in gray 368 ribbons. The nucleophilic cysteines are shown in sticks and colored in green. (F) Reaction 369 mechanisms of the amide

Figure 2. Structures of the SCARdock hits for CatB. The bonding atoms in the warheads are pointed 372 out by arrows

Docked poses of the identified hits of CatB (A-E) and CatL (F-H). The proteins are shown 374 in surface or ribbons, and the drugs are shown in sticks. The cysteine residues for covalent binding 375 are colored in yellow. The displayed items for CatB are trapoxin B (A)

The displayed items for CatL are 377 neratinib (HKI-272) (F), HKI-357 (G) and (Z)-dacomitinib (H). The putative covalent bonding atoms 378 in the drugs are

The binding details between the SCARdock hits and CatB/CatL. The docked poses are as 380 same as in Figure 3. For CatB: trapoxin B (A), neratinib (HKI-272) (B), HKI-357 (C), domatinostat 381 (4SC-202) (D) and (Z)-dacomitinib (E); for CatL: neratinib (HKI-272) (F)

The proteins are shown in ribbons with key residues shown in sticks, and the drugs 383 are shown in sticks

385 boceprevir and (R)-boceprevir. (B) The docked poses of the hits in TMPRSS2. The proteins are 386 shown in surface or ribbons, and the drugs are shown in sticks. The serine residues for covalent 387 binding are colored in red. The putative covalent bonding atoms in the drugs are indicated by arrows. 388 (C) The binding details between the SCARdock hits and TMPRSS2. The proteins are shown in 389 ribbons with key residues shown in sticks, and the drugs are shown in sticks

The plots were 392 prepared with LigPlot+. Ligands and protein side chains are shown in ball-and-stick representation, 393 with the ligand bonds colored in purple. Hydrogen bonds are shown as green dotted lines, while the 394 spoked arcs represent protein residues making nonbonded contacts with the ligands. The red circles 395 and ellipses indicate protein residues that are in equivalent 3D positions when the structural models 396 are superposed. The PDB codes of the experimental complex structures were shown in the 397 parentheses

Competing financial interests: The authors declare no competing financial interests. 13