Edited by: Chandrabose Selvaraj, Alagappa University, India
Reviewed by: Dhanachandra Singh Khuraijam, Lerner Research Institute, United States; Demetres D. Leonidas, University of Thessaly, Greece; Dongqing Wei, Shanghai Jiao Tong University, China
This article was submitted to Chemical Biology, a section of the journal Frontiers in Chemistry
†These authors have contributed equally to this work
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
The coronavirus disease 2019 (COVID-19) is a respiratory tract infection caused by the severe acute respiratory syndrome coronavirus (SARS)-CoV-2. In light of the urgent need to identify novel approaches to be used in the emergency phase, we have embarked on an exploratory campaign aimed at repurposing natural substances and clinically available drugs as potential anti-SARS-CoV2-2 agents by targeting viral proteins. Here we report on a strategy based on the virtual screening of druggable pockets located in the central β-sheet core of the SARS-CoV-2 Spike's protein receptor binding domain (RBD). By combining an
The coronavirus disease 2019 (COVID-19) is a respiratory tract infection caused severe acute respiratory syndrome (SARS)-CoV-2, a newly emerged coronavirus first identified in the city of Wuhan in China in December 2019 (Zhu et al.,
It is well-established that, similarly to SARS-CoV, SARS-CoV-2 enters the host cells by hijacking the human angiotensin converting enzyme receptor (ACE2) (Gui et al.,
Structural analysis has demonstrated that the N- and C- terminal portions of S1 fold as two independent domains, N-terminal domain (NTD) and C-terminal domain (CTD), with the latter corresponding to the receptor-binding domain (RBD) (Wang et al.,
The entry of SARS-CoV-2 in the host cells requires the cleavage of the S protein, a process that takes place in two steps. After binding to ACE2, the S protein is cleaved between the S1 and S2 subunits by a camostat-sensitive transmembrane serine protease, TMPRSS2 (Li et al.,
The spreading of the COVID-19 pandemic and the lack of effective therapies targeting the viral replication have prompted an impressive amount of investigations aimed at targeting several aspects of SARS-CoV-2 biology and viral interaction with ACE2. In this scenario, drug repurposing is a well-established strategy to quickly move already approved or shelved drugs to novel therapeutic targets, bypassing the time-consuming stages of drug development (Ghosh et al.,
One of such approaches is to prevent the S protein/ACE2 interaction as a strategy to prevent SARS-CoV-2 entry into target cells. Several virtual screening campaigns have already identified small molecules able to bind residues at the interface between the RBD of SARS-CoV-2 S protein and the ACE2 receptor (Ghosh et al.,
The electron microscopy (EM) model of SARS-CoV-2 Spike glycoprotein was downloaded from the Protein Data Bank (PDB ID:
The AutoDock4.2.6 suite (Morris et al.,
Each of the six selected RBD pockets were submitted to the AutoGrid4 tool, which calculates, for each bonding pocket, maps (or grids) of interaction, considering the different ligands and receptor-atom types through the definition of a cubic box. Subsequently, for each grid AutoDock4 calculates interaction energies (ADscore) that express the affinity of a given ligand for the receptor.
The library of FDA approved drugs has been obtained both from DrugBank (2106 compounds) (Drugbank,
MD simulations were performed using the CUDA version of the AMBER18 suite (Lee et al.,
The Dynamical Network Analysis was performed on 500 ns long MD trajectories of the RBD domain using the plugin Carma ver. 0.8 (Glykos,
OCA, BAR704, BAR501, and BAR502 were synthesized as previously described (Festa et al.,
We tested the selected compounds (UDCA, T-UDCA, G-UDCA, CDCA, G-CDCA, OCA, BAR501, BAR502, BAR704, betulinic acid, oleanolic acid, glycyrrhetinic acid, potassium canrenoate) using the ACE2: SARS-CoV-2 Spike Inhibitor Screening Assay Kit (BPS Bioscience Cat. number #79936) according to the manufacturer's instructions. All compounds were tested at different concentrations in a range from 0.01 to 100 μM. In addition, a concentration-response curve for the Spike protein (0.1–100 nM) was constructed to confirm a concentration-dependent increase in luminescence. A spike concentration of 5 nM was used for the screening of the compounds. Briefly, thaw ACE2 protein on ice and dilute to 1 μg/ml in PBS. Use 50 μL of ACE solution to coat a 96-well nickel-coated plate and incubate 1 h at room temperature with slow shaking. Wash the plate 3 times and incubate for 10 min with a Blocking Buffer. Next, add 10 μL of inhibitor solution containing the selected compound and incubate for 1 h at room temperature with slow shaking. For the “Positive Control” and “Blank,” add 10 μL of inhibitor buffer (5% DMSO solution). After the incubation, thaw SARS-CoV-2 Spike (RBD)-Fc on ice and dilute to 0.25 ng/μL (~5 nM) in Assay Buffer 1. Add the diluted Spike protein to each well, except to the blank. Incubate the reaction for 1 h at room temperature, with slow shaking. After 3 washes and incubation with a Blocking Buffer (10 min), treat the plate with an Anti-mouse-Fc-HRP and incubate for 1 h at room temperature with slow shaking. Finally, add an HRP substrate to the plate to produce chemiluminescence, which then can be measured using FluoStar Omega microplate reader.
In another experimental setting, we have tested the selected compounds using the ACE2: SARS-CoV-2 Spike Inhibitor Screening Assay Kit with a slight modification to the protocol. In particular, tested compounds were pre-incubated for 2 h with the Spike-RBD, and immediately afterwards the mix was incubated with ACE2 coated on the 96-well plate.
To confirm the validity of the assay used in this study, five remnants of plasma samples used to test levels of anti-SARS CoV2 IgG in post COVID-19 patients were used. The original samples were collected at the blood bank of Azienda Ospedaliera of Perugia from post COVID-19 donors who participate to a program of plasma biobanking. An informed and written consent was signed by donors recruited in this program. The program's protocol included the quantitative analysis of the anti-SARS-CoV-2 IgG antibodies directed against the subunits (S1) and (S2) of the virus spike protein. IgGs were therefore measured by chemiluminescence immunoassay (CLIA) technology (LIAISON®SARS-CoV-2 IgG kit, DiaSorin®, Saluggia, Italy). Leftovers of five samples from this assay of ≈ 40–50 μL whose destiny was to be discharged were used to validate the SARS-CoV-2/ACE2 assay used in our study. While donors have provided a written informed consent for plasma donation as mentioned above, and no blood samples were taken specifically for this study, we (SB and DF) have contacted the five donors whose serum leftovers were used in this study by a phone call and asked the permission to use the sample remnants. The permission was granted by all five donors. We wish to thank all of them for the kind collaboration.
With the aim to identify chemical scaffolds capable of inhibiting ACE2/Spike interaction by targeting the RBD of the S1 domain of the SARS-CoV-2 (
First, these pockets were used for the virtual screening of 2,906 FDA-approved drugs from the DrugBank and the Selleckchem websites, using the AutoDock4.2.6 program (Morris et al.,
These studies allowed the identification of several compounds with steroidal and triterpenoid scaffold, including glycyrrhetinic acid, betulinic acid and the corresponding alcohol (betulin), canrenone and the corresponding open form on the γ-lactone ring as potassium salt (potassium canrenoate), spironolactone and oleanolic acid, showing robust binding selectivity toward the RBD's pocket 1 (
Results of the screening of FDA approved drugs on the RBD region of the Spike protein of SARS-CoV-2 with the Autodock 4.2.6 program.
−8.1 | 1 | |
Betulinic Acid | ||
−7.4 | 1 | |
Betulin | ||
−8.6 | 1 | |
Glycyrrhetinic acid | ||
−8.2 | 1 | |
Oleanolic acid | ||
−7.9 | 1 | |
Canrenone | ||
−6.9 | 1 | |
Potassium Canrenoate | ||
−6.2 | 1 | |
Spironolactone |
Pocket 1, located on the β-sheet in the central core of the RBD, is the less conserved among the screened, presenting five conservative (R346K, S438T, L440I, S442A) and two non-conservative (G445T and L451K) mutations from SARS-CoV-2 to SARS-CoV.
Glycyrrhetinic acid, the best compound according to the AD score, binds the pocket through both hydrophobic and polar interactions. The triterpenoid scaffold relied between the hydrophobic side of the β-sheet core of RBD, defined by W436, F374 and the side chain of R509, and L441 on the other side, engaging hydrophobic contacts. In addition, the binding is reinforced by ionic contacts between the carboxyl group with R509, and by hydrogen bonds between the carbonyl group with N440 and the hydroxyl group with S375. Oleanolic acid and betulinic acid showed similar binding modes with the main difference in the carboxylic groups oriented toward the solvent. Finally, potassium canrenoate showed a different orientation of the steroidal system within the binding site, with the carboxylic function weakly bonded to S375 (3.1 Å), and the π-system of rings A and B stacked between W436 and L441 (
Graphical representation of the binding mode of the best compounds resulting from the screening in pocket 1. The RBD region is represented in transparent surface colored by residues hydrophobicity. Color codes are: dodger blue for the most hydrophilic regions, white, to orange-red for the most hydrophobic.
Because the above mentioned triterpenoids have been identified as natural ligands for two bile acid activated receptors, the Farnesoid-X-Receptor (FXR) and G protein Bile Acid Receptor (GPBAR)-1 (Sepe et al.,
Bile acids are steroidal molecules generated in the liver from cholesterol breakdown (Fiorucci and Distrutti,
Taking into account the structural similarity and the ability to bind the same receptor systems, we have carried out an in-depth docking analysis of natural bile acids and their semisynthetic derivatives currently available in therapy or under pre-clinical and clinical development (De Marino et al.,
Results of the screening of natural bile acids on the RBD region of the Spike protein of SARS-CoV-2 with the Autodock 4.2.6 program.
−7.0 | 5 | |
Ursodeoxycholic acid (UDCA) | ||
−7.0 | 5 | |
Tauro-ursodeoxycholic Acid (T-UDCA) | ||
−7.3 | 5 | |
Glyco-ursodeoxycholic Acid (G-UDCA) | ||
−7.3 | 5 | |
Chenodeoxycholic acid (CDCA) | ||
−7.6 | 5 | |
Glyco-chenodeoxycholic acid (G-CDCA) | ||
−7.6 | 5 | |
Obeticholic acid (OCA) | ||
−7.2 | 5 | |
BAR704 | ||
−6.9 | 5 | |
BAR501 | ||
−7.3 | 5 | |
BAR502 |
As shown in
Graphical representation of the binding mode of the best compounds resulting from the screening in pocket 5. The RBD region is represented in tan cartoon, while the pocket 5 residues as transparent surface colored by residues hydrophobicity. Color codes are: dodger blue for the most hydrophilic regions, white, to orange-red for the most hydrophobic.
In the binding mode of UDCA, the carboxylic group on the side chain is positioned between K378 and R408 and the steroidal scaffold is placed in a hydrophobic surface defined by the side chains of K378, T376, F377, Y380 and P384. Additionally, the 3β-hydroxyl group on ring A forms H-bonds with the backbone carbonyl of C379. The corresponding glycine and taurine-conjugated derivatives (G-UDCA and T-UDCA, respectively) showed the same ionic interactions of their negatively charged groups with K378 and R408. Albeit the greater length of the side chain, the H-bond with the backbone carbonyl of C379 induces a shift of the steroidal system toward T376, and an additional π-interaction between the electron density of the glycine amide region and the guanidine moiety of R408. This results in a better score for G-UDCA, and a reduction in the case of T-UDCA, likely due to a non-optimal arrangement of the taurine moiety within the binding pocket. CDCA showed a very similar binding mode, with the only difference that it formed an additional H-bond with the backbone carbonyl of F377 due to the modification in the configuration of the C-7 hydroxyl group (α-oriented in CDCA and β-oriented in UDCA). As for G-UDCA, also G-CDCA established the same H-bonds network of the parent CDCA, while the steroidal core slightly shifted as described for G-UDCA. Interestingly, AD scores of G-UDCA and G-CDCA clearly indicated that the H-bond between the hydroxyl group at C-7 and F377 does not contribute significantly to the binding mode.
With respect to CDCA, the introduction of the ethyl group at the C-6 position as in OCA and in BAR704 improves the internal energy of the ligand (−0.27 for CDCA vs. −0.59 and −0.60 kcal/mol for OCA and BAR704, respectively) and further favors the binding (
BAR501, a neutral UDCA derivative, with an alcoholic side-chain end group and the ethyl group at C-6 β-oriented showed a very similar binding mode compared to the parent compound, with the side chain hydroxyl group H-bonded to R408. Finally, BAR502, with a one carbon less on the side chain positioned the steroidal core as for G-CDCA, thus allowing the C-23 OH group H-bonding with the side chain hydroxyl group of T376.
To support our hypothesis about the allosteric inhibitory potential of the identified pockets, we performed a dynamical network and community map analysis on 500 ns of molecular dynamics (MD) simulations of the RBD domain. Overall, the network analysis found 12 communities (Com1-Com12) (
Community network representation of the RBD domain and community residue members of
Community map distribution of the RBD domain, retrieved after 500 ns-long MD simulation.
Com1 | 14 | N334; C361; V382; P384; T385; L387; D389; V524-K529; | Blue |
Com2 | 1 | V445 | Ice-blue |
Com3 | 1 | G476 | Dark-gray |
Com4 | 18 | V350; G416; D420; G446-R454; F456; F490-S494; | Orange |
Com5 | 7 | C336; E340; F342; A344 | Yellow |
Com6 | 9 | Y495-G502; G504 | Tan |
Com7 | 33 | L455;R457-A475; S477-Y489 | Light-Gray |
Com8 | 20 | T376; K378; C379; R408; I410-T415; I418; A419; Y421; Y423-P426; D428; T430; V511 | Green |
Com9 | 9 | A363-Y369; S371; S383 | White |
Com10 | 30 | L335; R355-N360; V362; Y380; G381; K386; L390-V395; D427-F429; L513-T523 | Pink |
Com11 | 38 | V341; N343;T345; R346; Y351-N354; N370; A372-S375; F377; Y396-F400; N422; G431-A435; N437; N439-L441; S443; P507-V510; V512 | Cyan |
Com12 | 16 | V401-V407; Q409; K417; W436; S438; D442; K444; V503; Y505; Q506 | Purple |
Given the results of the virtual screening, we have then investigated whether the agents mentioned in
The ACE2:SARS-CoV-2 Spike Inhibitor Screening assay was performed as described in Material and Method section. Data shown are:
Because these data demonstrate that betulinic acid and oleanolic acid were effective in inhibiting the binding of the S protein RBD to ACE2, and the two triterpenoids were known for their ability to modulate GPBAR1, we then tested whether natural GPBAR1 bile acids ligands were also effective in reducing the SARS-CoV-2-ACE2 interaction. As illustrated in
The ACE2:SARS-CoV-2 Spike Inhibitor Screening assay was performed as described in Material ad Method section. Natural bile acids (A) UDCA, (B) TUDCA, (C) GUDCA, (D) CDCA, (E) GCDCA (0.1, 1 and 10μM) and (F) a combination of GCDCA + UDCA (100μM), were tested to evaluate their ability to inhibit the binding of Spike protein (5 nM) to immobilized ACE2, by using the ACE2:SARS-CoV-2 Spike Inhibitor Screening assay Kit. Luminescence was measured using a Fluo-Star Omega fluorescent microplate reader. Luminescence values of Spike 5 nM were arbitrarily set to 100%. Results are expressed as mean ± standard error. *
Continuing the
The ACE2:SARS-CoV-2 Spike Inhibitor Screening assay was performed as described in Materials and Methods section. The semi-synthetic bile acid receptor agonists OCA, BAR704, BAR502, and BAR501, were tested at different concentration (0.1, 1, and 10 μM) to evaluate their ability to inhibit the binding of Spike protein (5 nM) to immobilized ACE2, by using the ACE2:SARS-CoV-2 Spike Inhibitor Screening assay Kit. Luminescence was measured using a Fluo-Star Omega fluorescent microplate reader. Luminescence values of Spike 5 nM were arbitrarily set to 100%. Results are expressed as mean ± standard error. *
To further confirm our results, additional
The ACE2:SARS-CoV-2 Spike Inhibitor Screening assay was performed as described in Material ad Method section. The selected compounds were tested at 10 μM to evaluate their ability to inhibit the binding of Spike protein (5 nM) to immobilized ACE2, according the ACE2:SARS-CoV-2 Spike Inhibitor Screening assay Kit instructions or with a modified protocol in which we have performed a pre-incubation of these compounds with Spike-RBD (2 h). Tested compounds were:
To confirm the concept that binding the pockets in the central β-sheet core of Spike RBD effectively prevents its interaction with the consensus of ACE2 receptor, we then carried out a set of control experiments using remnants of the plasma samples from five donors that have recovered from COVID-19. These donors had a slightly different title of anti SARS-CoV-2 antibodies (See Material and Methods,
Percentage of inhibition of the Spike:ACE2 binding.
1 | 96.6 AU/mL | 98.6 | 99.5 | 99.6 |
2 | 170 AU/mL | 99.3 | 99.4 | 99.3 |
3 | 89.4 AU/mL | 98.1 | 99.3 | 99.4 |
4 | 125 AU/mL | 98.8 | 99.3 | 99.4 |
5 | 146 AU/mL | 95.7 | 96.9 | 97.3 |
In this study we report the results of a virtual screening campaign designed to identify natural and clinically available compounds that might have utility in the prevention/treatment of the SARS-CoV-2 infection. In the light of the need of effective therapies to be rapidly tested for preventing or treating COVID-19, we initiated an
Our
Our
Hydrophobicity is also the main determinant of the activity of the bile acids and their semisynthetic derivatives, as demonstrated by CDCA, the corresponding glyco-conjugated derivative (G-CDCA) and its semisynthetic derivatives OCA, BAR704, and BAR502. Indeed, comparing the binding mode and the inhibition efficacy of CDCA and OCA with the related 6-ethyl derivative BAR704 highlighted the critical effect of the 6α-ethyl group in the inhibition activity and the negligible contribution of the 3β-hydroxyl group. The above positive effect could be explained considering the internal energy contribution of these ligands to the AD score, as well as the possibility of engaging more hydrophobic contacts. Indeed, the AD score internal energy contribution, significantly higher for the 6-ethyl derivatives, represents a measure of the conformational energy of the bound
Moreover, also the comparison of the binding modes for G-CDCA and G-UDCA supported the hypothesis that the main determinant for the activity should be related to the network of hydrophobic interactions more than to the lack of a punctual hydrogen bond. Indeed, unlike the weakly active UDCA, the steroid core of G-UDCA is shifted to T376, and the resulting binding mode looks very similar to G-CDCA's. Finally, the better inhibitory efficacy of BAR501 with respect to UDCA, further confirmed the not-essential effect of the charged group on the side chain in terms of inhibition activity. Interestingly, the analysis of the binding mode of BAR501 also suggested that the stereochemistry of the ethyl group at C-6 is not pharmacophoric, being the 6β-ethyl group still able to potentially interact with P384 and Y369.
In the present study, we have developed a strategy to target the interaction of SARS-CoV-2 S protein RBD with the ACE2 receptor. As described in the introduction, SARS-CoV-2 enters the target cells by binding the carboxypeptidase domain of the ACE2 receptor, exposing a cleavage site, a hinge region between S1 and S2, to TMPRSSS2, which in turn allows the S2 subunit of the Spike protein to bind with the cell membrane, leading to the virus/host cells membrane fusion and SARS-CoV-2 penetration in to host cells.
The two pockets we have identified in the β-sheet core of the Spike RBD appear to be targetable by steroidal molecules and, importantly, we found that both naturally occurring bile acids and their metabolites in humans reduce the binding of Spike's RBD to ACE2. Of interest, natural bile acids, such as UDCA, T-UDCA, CDCA, and G-CDCA, exerted a greater inhibitory effect when preincubated with Spike reaching ~45-50% of binding inhibition. Importantly, we found that most of the agents tested in this study were agonists of two main bile acid activated receptors, i.e., the Farnesoid-x-Receptor (FXR) and a cell membrane receptor known as GPBAR1. Thus, betulinic acid and oleanolic acid, along with UDCA and its metabolites, BAR501 and BAR502 are effective ligands for GPBAR1. In contrast, glycyrrhetinic acid, CDCA, G-CDCA and T-CDCA, OCA and BAR704 are known for their ability to bind FXR (Festa et al.,
Of interest, some of these agents have been reported for the potential use as anti-HIV agents (Rezanka et al.,
One important observation we have made in this study is that, while two different pockets of Spike RBD are potentially druggable, these are contiguous, and indeed, when we attempted drug combinations, none of these combinations effectively increased the anti-adhesive efficacy in comparison to the single agent.
This study has several limitations. First of all, we observed that the anti-adhesive efficacy of hyperimmune plasmas obtained from donors who have recovered from COVID-19 and containing high titles of neutralizing antibodies, in inhibiting the Spike RBD/ACE2 interaction, is close to 99%. This percentage is significantly higher than what we measured with our compounds. One possible explanation of this different efficacy can be found in terms of difference in affinity of our compounds with respect to the antibodies but could also be related to the mechanism of allosteric connections suggested by dynamical network and community map analysis. Indeed pockets 1 and 5 resulted tightly connected with the loop G496-Y505, and weakly with the larger loop F456-F490. This suggests that small molecules binding the hydrophobic pockets are less effective than a neutralizing antibody. This also suggests that our pharmacological approach will likely be poorly effective in the presence of a high viral load, and the approach we have developed might have some efficacy only in the case of low viral load. Nevertheless, the mild inhibition efficacy showed by bile acids and their derivatives could pave the way for a further optimization of the binding mode in order to identify additional potential interactions, particularly in pocket 5, which has been demonstrated the least exposed to mutations.
Another limitation is that we have not tested the effect of these treatments on viral replication and further studies are needed to clarify this point.
In conclusion, in this paper, we report the identification of several potential binding sites in the RBD of the SARS-CoV-2 S protein. Several triterpenoids, such as glycyrrhetinic and oleanolic acids, and natural bile acids and their semisynthetic derivatives have been proven effective in reducing the Spike RBD's adhesion to its ACE2 consensus
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study.
SB and DF provided serum samples. BF, FM, and BC performed virtual screening and analyzed the data. CF and VS performed chemical synthesis. AC, SM, and MB generated the
This paper was supported by a research grant by BAR Pharmaceuticals S.r.L. to the Department of Pharmacy of the University of Napoli Federico II and to the Department of Surgical and Biomedical Sciences, University of Perugia. The authors declare the following competing financial interest(s): SF, AZ, and BC have filed an Italian patent application no.102020000011092 in the name of BAR Pharmaceuticals S.r.L. on the compounds described in this paper. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Authors wish to thank all the donors for the kind collaboration. This manuscript has been released as a PrePrint (Carino et al.,