key: cord-0758880-w9ogp7h8 authors: Saxena, Shalini; Meher, Kranti; Rotella, Madhuri; Vangala, Subhramanyam; Chandran, Satish; Malhotra, Nikhil; Palakodeti, Ratnakar; Voleti, Sreedhara R; Saxena, Uday title: In silico and in vitro Demonstration of Homoharrintonine’s Antagonism of RBD-ACE2 Binding and its Anti-inflammatory and anti-thrombogenic Properties in a 3D human vascular lung model date: 2021-05-03 journal: bioRxiv DOI: 10.1101/2021.05.02.442384 sha: d9eddcfffdba1f8d8fad75fa52ee94e84eab93b2 doc_id: 758880 cord_uid: w9ogp7h8 Since 2019 the world has seen severe onslaught of SARS-CoV-2 viral pandemic. There is an urgent need for drugs that can be used to either prevent or treat the potentially fatal disease COVD-19. To this end, we screened FDA approved antiviral drugs which could be repurposed for COVID-19 through molecular docking approach in the various active sites of receptor binding domain (RBD). The RBD domain of SARS-CoV-2 spike protein is a promising drug target due to its pivotal role in viral-host attachment. Specifically, we focussed on identifying antiviral drugs which could a) block the entry of virus into host cells, b) demonstrate anti-inflammatory and/or anti-thrombogenic properties. Drugs which poses both properties could be useful for prevention and treatment of the disease. While we prioritized a few antiviral drugs based on molecular docking, corroboration with in vitro studies including a new 3D human vascular lung model strongly supported the potential of Homoharringtonine, a drug approved for chronic myeloid leukaemia to be repurposed for COVID-19. This natural product drug not only antagonized the biding of SARS-CoV-2 spike protein RBD binding to human angiotensin receptor 2 (ACE-2) protein but also demonstrated for the first time anti-thrombogenic and anti-leukocyte adhesive properties in a human cell model system. Overall, this work provides an important lead for development of rapid treatment of COVID-19 and also establishes a screening paradigm using molecular modelling and 3D human vascular lung model of disease to identify drugs with multiple desirable properties for prevention and treatment of COVID-19. SARS-CoV-2 virus has devastated the world since late 2019. The virus when left untreated leads to COVID-19 disease and is often fatal. The disease begins with entry of the virus into the human host cells. Coronaviruses (CoVs), including SARS-CoV, MERS-CoV, and SARS-CoV-2, are cytoplasmically replicating, positive-sense, single-stranded RNA viruses with four structural proteins i.e. Spike protein (S), envelope (E) protein, membrane (M) protein, and nucleocapsid (N) protein [1] . This entry of SARCS-CoV-2 into host cells is mediated by a specific receptor binding domain (RBD) on the surface of spike protein. The entry of the virus into the host cell requires the spike or S-protein to be cleaved in two steps. The first step is the binding of the virus to the ACE-2 protein and this is accomplished by the cellular proteases acting at the region between S1 and S2, namely, the transmembrane serine protease TMPRSS2. S1 and S2 subunits are responsible for viral-receptor binding and virus-host cell membrane fusion, respectively. The S1 subunit bears the receptor-binding domain (RBD) on its C-terminal domain. The RBD itself contains the receptor-binding motif (RBM), which actually comes into contact with the carboxypeptidase domain of the ACE-2 molecule. The RBD engages with a counter receptor on human cells called angiotensin receptor 2 (ACE-2) and gains entry. Once inside the human cells, it uses host machinery to replicate itself. Thus, antagonism of binding of RBD to ACE-2 could be a preventive strategy for this infection. Generally, the S protein plays a crucial role in eliciting the immune response during disease progression also. Thus S protein RBD domain, if inhibited, should prevent viral attachment, fusion and entry, and thus make it an attractive target over other targets [2] . c) The activation of endothelial cells by virus leading to recruitment of inflammatory cells such as monocytes into the lung. d) Uncontrolled production of inflammatory cytokines "cytokine storm" further exaggerating lung dysfunction. e) A prothrombogenic environment which promotes blood clotting in lung capillaries. All these events translate into a hostile and highly inflammatory environment in the lungs. There is a rapid decline in lung function due to lung cell death, fibrosis and hypoxia. Most other organs also fail in due time due to direct viral intervention and or hypoxia due to lack of oxygen supply by blood. But such events also provide choke points for therapeutic intervention. For example, a drug that combines inhibiting the binding of virus to human cells and also possess anti-inflammatory and anti-thrombogenic properties may be more desirable than a drug with just antiinflammatory property or entry antagonism alone. Table-1 lists our proposed wish list of properties that may be desirable for a drug to be repurposed for prevention and treatment. can also mimic monocyte adhesion by adding monocytes to the 3D bioprinted system. So, using one model we are able to screen for several features seen in human disease. We chose a few selected antiviral drugs with demonstrated efficacy against other viruses such MERS and SARS. The idea behind this was that since they have antiviral activity to begin with, we are likely to guarantee ourselves of one potential therapeutic activity. Secondly, since they are all drugs approved for human use, they can be rapidly deployed for COVD-19 treatment. Since the entry of virus initiates the pathological cascade of evets, we first rank ordered the drugs based on their ability to block the binding of RBD to ACE-2 using molecular docking studies. After that we systematically tested them for a) inhibition of binding of RBD to ACE- Cresset Flare software was used for molecular docking studies against the spike protein SARS-CoV-2 (http://www.cresset-group.com/flare/) [4] . The 2D structures of all the antiviral drugs were downloaded from Drugbank database [5] and prepared using Flare software. Atom force field parameterization was assigned and hydrogen atoms were added in the structure. Further, energy minimized was done for all the drugs, nonpolar hydrogen atoms were merged, and rotatable bonds were defined. Later, ligand minimization has been carried out in Flare by Minimize tool by using Normal calculation methods. The RBD of spike glycoprotein SARS-CoV-2 is used for present study. It is clear that spike proteins represent potential targets for anti-SARS-CoV-2 drugs hampering the interaction between human ACE-2 and the viral RBD will block the entry of the virus into the human cells. The 3D structure of RBD binds to ACE-2 receptor (PDB ID: 6M0J) were downloaded from Protein Data Bank (PDB) (https://www.rcsb.org) [6] . The protein has two chain A and E, the A chain has ACE-2 receptor and E chain has RBD domain. The RBD domain has been save in to PBD format for further studies. The target protein preparation was carried out in Flare software with default settings. Missing residues, hydrogen's and 3D protonation were carried out on the target protein. Protein minimization has been carried out in Flare by Minimize tool by using Normal calculation methods. Binding site was generated Accelrys Discovery Studio visualizer 3.5 (Copyright© 2005-12, Accelrys Software Inc.) to explore potential binding sites of the RBD protein using receptor cavities tools. Based on a grid search and "eraser" algorithm, the program defines where a binding site is. The binding sites were displayed as a set of points (point count) and the volume of each of cavity was calculated as the product of the number of site points and the cube of the grid spacing. Volume of each site were calculated and further saved and exported in to Flare for advance analysis. All the 23 antivirals were docked in the active site of RBD by using Flare docking module of Cresset software. All the compounds were subjected to docking using Lead Finder (LF) and the predicted binding poses were analysed [7] docking analysis and visualization of hits led to the identification of 7 drugs based on rank score (RS), and binding energy (ΔG) that bind selectively to the RBD site-1 and site-2. The above assay was performed using SARS-CoV-2 sVNT ready to use kit by Genscript, which is a competition ELISA, mirroring the viral neutralization process. In the first step, a mixture of horse radish peroxidase-RBD (HRP-RBD) and controls/drugs (drugs at the concentrations of 50µM, 100µM and 200µM) were incubated at 37 o C for an hour to allow the binding of drugs to HRP-RBD. Following the incubation, these mixtures were added to a capture plate, which was a 96 well microplate coated with human ACE-2 receptor to permit the binding of any unbound HRP-RBD and the ones bound to drugs to ACE-2 receptor. After incubating the microplate at 37 o C for an hour, plate was washed four times using wash buffer in order to remove any unbound circulating HRP-RBD_ drug complexes. Washing step was followed by addition of a colour substrate; tetramethyleneblue (TMB), turning the colour to blue. The reaction was allowed to run for 15 minutes followed by the quenching using stop solution turning the colour from blue to yellow. This final solution was read at 450 nm. The absorbance of the sample is inversely proportional to the inhibition of RBD's binding to the human ACE-2 by the drug. Human IL-1ẞ ELISA was performed to detect the presence of IL-1ẞ which is a key mediator of inflammatory response. The ELISA was performed using Human IL-1ẞ high sensitivity kit sold by Invitrogen. The sample for our ELISA consisted of cell culture media (referred to as sample hereafter) collected after treating the cells with drugs at 100uM. Samples were added to the microplate precoated with human IL-1ẞ antibody which captures the IL-1ẞ present in the samples. A secondary anti-human IL-1ẞ antibody conjugated to biotin was added to the plate. Following an overnight incubation, microplate was washed six times using wash buffer in order to remove any unbound biotin conjugated anti-human IL-1ẞ antibody. Streptavidin-HRP was then added which binds to biotin conjugated antibody and the plate was incubated at room temperature on a shaker for an hour. After the incubation, plate was washed again following the same process as the previous wash step and an amplification reagent I was added to the wells. Following the incubation of 15 minutes and a wash, amplification reagent II was added. After incubation of half an hour in dark and a wash step later, a substrate solution was added turning the colour to blue. The reaction was terminated using a stop solution (turning the colour from blue to yellow) after 15-20 minutes. This final solution was read at 450nm. The OD of the sample is directly proportional to the amount of human IL-1ẞ present in it. Human thrombodulin ELISA was performed using a ready to use kit by R and D systems. Samples (media collected from our 3D vascular lung system) were added to a microplate precoated with monoclonal antibody specific for human thrombodulin. This was followed by incubation period of two hours (allowing any thrombodulin present in the sample to bind to the monoclonal antibody) the plate was washed with wash buffer four times. After washing away any unbound thrombodulin, an enzyme-linked monoclonal antibody specific for human thrombodulin was added to the plate. Another wash step was performed in order to remove any unbound antibody-enzyme reagent after completion of two hours of incubation. A colour substrate was then added turning the colour to blue. Reaction was quenched after about 15-20 minutes using a stop solution which turned the colour from blue to yellow. Absorbance was read at 450 with wavelength correction set to 540 nm or 570 nm. For the 3D vascular lung model three types of cell were grown: In the 3D-vascular lung model four layers were bioprinted, first layer was collagen layer, (30µl Currently, there are no specific effective antiviral treatments for COVID-19, although most of the COVID-19 patients have mild or moderate courses, up to 5%-10% can have severe, potentially life threatening course. Thus, there is an urgent need for effective drugs [8] . Our in silico strategy helps us to find new repurposed antiviral drugs that can attach to residues at the site of binding of the RBD to the ACE-2. In the current study we had discovered several potential binding sites for molecules that can occupy such druggable pockets so as to inhibit virus-ACE-2 binding in vitro. The X-ray model of the RBD was used to identify 3 possibly druggable pockets where drugs might bind. The active site volume and binding surface area of three pocket is representation in Docking analysis and visualization of selected antivirals led to the identification of 7 drugs that bind selectively to the both the RBD sites. Total 7 drugs were identified based on rank score (RS), and binding energy (ΔG). These include Homoharrigtonine, Triparanol, Lopinavir, Ritonavir, Astemizole, Amodiaquine, and Fluspirilene. Our goal was to find those molecules which have great binding affinity towards both the active sites of RBD domain. Detail binding analysis of selected drugs towards active site of spike protein SARS-CoV-2 was studied in detail. Interaction analysis of drugs with spike protein SARS-CoV-2 (RBD) were carried out to identify the compound having highest binding affinity with target protein. Active site-1 composed of Arg454, Phe456, Arg457, Lys458, Glu465, Arg466, Asp467, Ile468, Ser469, Glu471, Thr473, Gln474 and Pro491 amino acid residues, while the site-2 composed of Leu335, Cys336, Pro337, Phe338, Gly339, Trp436, Phe342, Asn343, Val362, Ala363, Asp364, Val367, Leu368, Ser371, Ser373 and Phe374 amino acid residues, as shown in Figure- 1. Active site-1 is more hydrophobic in nature as compare to active site-2. Active sites of RBD domain of S1 protein. The LF rank score is an indicator of the binding affinity of protein-ligand complex. The LF rank for selected drugs in the pockets site-1 & site-2 is described in Table-3 Out of 23 antivirals, we selected 7 drugs for in vitro biological evaluation based on rank score and ΔG along with their interactions with the protein. We also selected two low scoring drugs (Toremifene and Chloroquine) for in vitro biological evaluation to confirm ours in silico studies. As shown in Figure-2 , the graph represents the correlation between rank score and binding energies of all the selected drugs in RBD site-1 and site-2. All the drugs have more or less similar range of score and binding energies. Also, the selected drugs well occupied the active site cavity (S1-1 and S1-2). Docking analysis of Ritonavir in the site-1 revealed that it has making three strong hydrogen bonding interactions with Arg457 and another with Asp467. The drug is also involved in one hydrogen bond with Asn460 and cation-pi interactions with Lys458. In the binding site-2, the drug was involved in hydrogen bonding interaction with Asn343 and Trp436. The compound was found placed proximal to various hydrophobic amino acids such as Leu335, Gly339, Phe342, Ala344, Thr438, Asn437, Asn440, Ile441, Ser371, Leu368 &Phe374 and hence exhibited hydrophobic interactions. Apart from these interactions the compound was further stabilized by pi-pi interaction with Phe373 and Phe338. Amodiaquine is an aminoquinoline analog used for the therapy of malaria. Amodiaquine has been linked to severe cases of acute hepatitis which can be fatal, for which reason it is recommended for use only as treatment and not for prophylaxis against malaria. It shows good LF Rank score and ΔG in both the active sites. In site 1, the drug is making two hydrogen bonding interactions with Lys458, and Arg466. Interaction analysis in active site-2 revealed that it has hydrogen bonding interactions with Phe342, Val367 amino acid residues (Figure- In this experiment, we explored the ability of the short listed drugs on blocking binding of RBD to human ACE-2 using a commercially available competition ELISA assay. As shown below in Figure-8 , several of the drugs showed inhibitory activity. But the stand out drug was Homoharringtonine with near complete inhibition at higher concentrations. Ritonavir and chloroquine also showed activity but the maximum activity obtained (<50%) was much less than that of Homoharringtonine. These data support further evaluation of Homoharringtonine for anti-inflammatory and anti-thrombogenic properties. In this experiment, we explored the inherent potential of the drugs on blocking secretion of IL-1β, a cytokine known to be involved in COVID-19. We used LPS (lipopolysaccharide) as a stimulus to induce the secretion since this is a well-established inflammatory stimulant and its use does not need BLA3 safety lab unlike what is needed for using live virus. As expected LPS induced secretion of IL-1β. But none of the drugs tested showed any inhibitory activity, as shown in Figure-9 . In this experiment, we explored the ability of drugs on blocking secretion of thrombomodulin, an endothelial surface protein which binds thrombin and modulates blood clotting. Inflammatory stimulus such as LPS and cytokines activate the secretion of this protein thus rendering is less functional. Secreted thrombomodulin is less effective at clotting function so drugs which reduce its secretion (therefore retaining it on endothelial surface) would be expected to have a beneficial effect. As shown below in Figure- In this experiment, we explored the ability of the drugs to reduce monocyte adhesion to the endothelium induced by inflammatory stimulus such as LPS. The adhesion of monocytes to the endothelium is the first step in the entry of these inflammatory cells into the lung during COVID-19. Thus, if a drug is able to reduce monocyte adhesion, it would be expected to be therapeutic. Figure-11 shows there was an increase in monocyte adhesion by LPS by about 48%. While Homoharringtonine did not completely inhibit LPS induced monocyte adhesion the increase was less than that found with LPS alone (30% versus 48%). These data suggest that this drug may modulate monocyte adhesion beneficially. Our work described here has identified Homoharringtonine as a potential drug candidate for repurposing for COVID-19 treatment. We find reasonable correlation between the predicted in silico activity of the drugs towards inhibition of RBD binding to ACE-2 with in vitro, suggesting that a rational drug design approach is possible for this disease. While Homoharringtonine appeared to have the greatest number of desirable properties for therapeutic intervention, Ritonavir and Chloroquine which have been tested in clinical studies showed much less activities [9] . Neither drug was able to completely inhibit binding of RBD to ACE-2 and did not possess some other properties as well. Our data would have predicted that what has been observed clinically, i.e. that these drugs are unlikely to be very effective in treatment of this disease (9) . Our data substantiates the observation of others with Homoharringtonine. Another report has recently shown that this drug inhibits in vitro replication of the virus in-Vero cells [10] . In addition, most recently it has been found to be active in a live virus induced mice model of COVID-19 [11] . The in vivo study reported a severe reduction in viral load upon dosing with homoharringtonine. The mechanism(s) of this efficacy were not probed in that study but in part can be attributed to its effect on viral replication and as shown by our work here potential prevention of viral entry into host cells. Since most animal models of COVID-19 represent early stage disease, our ability to demonstrate anti-thrombogenic and anti-adhesive properties in a human 3D model become very relevant in terms of the massive inflammation and micro thrombi observed in human lungs. We propose that our data with homoharringtonine using the 3D model adds to the potential mechanism by which this drug may be useful in treatment of COVID-19. In conclusion, we propose based on our own and others work, Homoharringtonine can be tested in human trials as a nasal spray to maximize its distribution into the nasal and pulmonary tissue which are the major points of viral attachment and entry. We would also like to recognize the utility of our 3D human vascular lung model to study critical features of human disease such as thrombogenicity/clotting which is part of human disease and is also now seen as a serious adverse event with mRNA based COVID-19 vaccines. A systematic review of SARS-CoV-2 vaccine candidates Cheminformatics-Based Identification of Potential Novel Anti-SARS-CoV-2 Natural Compounds of African Origin Evidence of a wide gap between COVID-19 in humans and animal models: a systematic review DrugBank: a comprehensive resource for in silico drug discovery and exploration Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE-2 receptor Lead Finder, version Antiviral treatment of COVID-19 Repurposed antiviral drugs for COVID-19-interim WHO SOLIDARITY trial results Remdesivir, lopinavir, emetine, and homoharringtonine inhibit SARS-CoV-2 replication in vitro Homoharringtonine (HHT)-A highly effective drug against coronaviruses and the potential for large-scale clinical applications