OpenZika’s Work on World Community Grid is Complete

Progress on selecting and testing compounds

Continuing the progress we mentioned in our last project update earlier this year, we virtually selected and purchased 75 compounds--55 candidates for Zika virus (ZIKV) and 20 candidates for dengue virus (DENV--to be experimentally evaluated at Dr. Siqueira-Neto’s laboratory at the University ofCalifornia in San Diego. Dr. Siqueira-Neto’s lab performed cell-based assays in glioblastoma stem cells (hGSC) with ZIKV.

From the experimental results obtained so far, two compounds selected for NS2B-NS3 protease presented inhibitory activity against cells infected with ZIKV and low cytotoxicity. One of them presented activity in the nanomolar range and is therefore a promising candidate.

Moreover, from the seven compounds selected for NS3 helicase RNA site, three were for the ATP site and inhibited cells infected with ZIKV and possessed low cytotoxicity.

Two compounds were virtually selected for the NS5 methyltransferase and had inhibitory activity against cells infected with ZIKV (one for the active site and one for the SAM site) and had low cytotoxicity.

None of the compounds virtually selected for NS5 polymerase showed inhibitory activity against cells infected with ZIKV.

One compound virtually selected for the ZIKV envelope protein and was an inhibitor of cells infected with ZIKV and low cytotoxicity.

The enzymatic assays with the ZIKV NS3 helicase, NS2B-NS3 protease and NS5 polymerase proteins will be performed at the Physics Institute of Sao Carlos, University of Sao Paulo (Brazil), at Prof. Glaucius Oliva's laboratory, to validate the predicted enzymatic candidates' activities.

Two compounds virtually selected for DENV NS2B-NS3 protease were an inhibitor against cells infected with ZIKV and had low cytotoxicity (Figure 6). The active sites of DENV and ZIKV protease have more than 90% sequence identity. These compounds will be also tested against cells infected with DENV at the laboratory of Prof. Jose Modena, another OpenZika collaborator, at the University of Campinas(UNICAMP) in Brazil.

Two compounds virtually selected for DENV NS3 helicase were an inhibitor of cells infected with ZIKV and also had low cytotoxicity (Figure 7). The ATP and RNA sites of DENV and ZIKV protease are conserved and have more than 90% sequence identity. These compounds will be also tested against cells infected with DENV at Prof. José Modena’s laboratory at UNICAMP.

In another project conducted by the undergraduate student Paulo Ramos, we used integrative similarity analysis, docking and machine learning (ML) models for identifying new ZIKV NS5 inhibitors candidates guided by known DENV NS5 inhibitors. Comparing the primary sequence of NS5 ZIKV and NS5 DENV binding sites (SAM, GTP, RNA-site, catalytic site and N-Pocket), we found high sequence identity for all of them. We searched for DENV NS5 inhibitors in the PubChem and ChEMBL databases and screened through these ML models for Zika.

We found 145 compounds reported as DENV NS5 inhibitors that were screened through the ML filter (Figure 8). From this filter, 74 compounds were prioritized for the molecular docking calculations. The 32 compounds that presented free energy scores from the docking < -7.0 kcal/mol were searched for similarity to structures in the E-molecules commercial database which resulted in 6,053 similar compounds. The ML and docking process were repeated for these similar compounds. The compounds were also screened through Bayesian ML models for ZIKV and cytotoxicity, and 58 were prioritized. Finally, these compounds were screened through the STopTox¹ and Pred-hERG² servers and 38 compounds were predicted as non-toxic for the studied endpoints. These compounds will be purchased and submitted for experimental evaluation.

In a collaborative project with Profs. Scott Laster and Frank Scholle from North Carolina State University, we performed docking calculations for natural compounds from Ashitaba (Angelica keiskei), which is used in Asiatic medicine. These compounds were described in the literature as viral inhibitors. We screened the extracts from Ashitaba through cell based and cytotoxic assays (Figure 9). They also presented antiviral activity against ZIKV. Thus, we isolated the compounds and three chalcones presented anti-ZIKV activity with IC50s in the range of 5-20 µM and low cytotoxicity in mammalian cells (CC50 of 100 µM). Target prediction was performed, and we predicted that protease is the most probable target of these compounds. We also performed molecular docking studiesagainst all ZIKV proteins in the OpenZika project. Agreeing with the target prediction, docking suggested that ZIKV protease is the main target. Protease enzymatic assays showed that the compounds can inhibit ZIKV protease, validating the in silico predictions.

Accepted and upcoming publications

Our paper entitled “A diarylamine derived from anthranilic acid inhibits ZIKV replication” was accepted at the Scientific Reports journal. This study was a collaborative work with Prof. Ana Carolina Jardim, from the Federal University of Uberlandia, Brazil, and it reports a series of analogs from anthranilic acid screened by cell-based assays. Then, docking calculations performed in the OpenZika project suggested that NS3 helicase was the most probable target, which was experimentally validated by enzymatic assays.

We recently submitted a paper entitled “Natural products from Angelica keiskei with activity against Zika vírus NS2B-NS3 protease” to Antiviral Research. This study describes three compounds extracted from Angelica keiskei, commonly used in Asian medicine, that presented anti-ZIKV activity at cell-based assays. Target prediction and molecular docking suggested NS2B-NS3 protease as the most probable target. Enzymatic assays validated the in silico predictions.

Papers being written – to be submitted soon:

We are preparing the following papers regarding the exciting results of OpenZika project:

• One paper reporting the results (virtual and experimental) of the first round of compounds selected against ZIKV NS3 helicase that presented anti-ZIKV activity in human glioblastoma stem cells (hGSCs);
• One paper reporting the results from the virtual screening and experimental evaluation of natural products from Pterogyne nitens, which presented anti-ZIKV activity, inhibiting ZIKV proteins protease and helicase;
• One paper reporting the results for approved drugs/clinical collections compounds, that have anti-malarial and anti-ebola activity, that presented anti-ZIKV activities against ZIKV helicase protein, candidates for drug

These papers will be soon submitted to high-impact scientific journals.

Past publications and outreach

We published a review entitled “High Throughput and Computational Repurposing for Neglected Diseases” in the journal Pharmaceutical Research. This paper describes the many drug repurposing efforts that have been going on in different labs around the world to try to find treatments for the many tropical diseases, including Zika and dengue infections.

Our Drug Discovery Today keynote review “The A–Z of Zika drug discovery” was published on June 20 2018. This is a comprehensive review of the recent advances in ZIKV drug discovery efforts, highlighting drug repositioning and computationally guided compounds, including recently discovered viral and host cell inhibitors. Promising ZIKV molecular targets are also described and discussed, as well as targets belonging to the host cell, as new opportunities for ZIKV drug discovery. All this knowledge is not only crucial to advancing the fight against the Zika virus and other flaviviruses, but it will also help the scientific community prepare for the next emerging virus outbreak to which we will have to respond.

Our paper “Computational drug discovery for the Zika virus” was published in a special issue of the Brazilian Journal of Pharmaceutical Sciences. In this paper, we summarize current computational drug discovery efforts and their application to the discovery of anti-ZIKV drugs. We also present successful examples of the use of computational approaches to ZIKV drug discovery, including our OpenZika project.

Our PLoS Neglected Tropical Diseases paper, "OpenZika: An IBM World Community Grid Project to Accelerate Zika Virus Drug Discovery," was published on October 20 2016, and it has already been viewed over 5,200 times. Anyone can access and read this paper for free. Another research paper, “Illustrating and homology modeling the proteins of the Zika virus,” was published in F1000Research and viewed > 4,200 times.

We have also published another research paper entitled “Molecular Dynamics simulations of Zika Virus NS3 helicase: Insights into RNA binding site activity” in a special issue on Flaviviruses for the journal Biochemical and Biophysical Research Communications. This study of the NS3 helicase system helped us learn more about this promising target for blocking Zika replication. The results will help guide how we analyze the virtual screens that we performed against NS3 helicase, and the Molecular Dynamics simulations generated new conformations of this system that we have been using as targets in new virtual screens that we performed as part of OpenZika.

Dr. Sean Ekins presented a poster at Cell Symposia: Emerging and Re-emerging Viruses, on October 1-3, 2017, in Arlington, VA, USA, entitled “OpenZika: Opening up the discovery of new antiviral candidates against Zika virus.”

The OpenZika researchers Dr. Melina Mottin, Dr. Roosevelt Silva, Msc. Bruna Sousa and Paulo Ramos presented posters at the 9^th BrazMedChem 2019 conference, the major medicinal chemistry conference in Latin America, which was organized by the OpenZika PI, Prof. Carolina Horta Andrade.

Dr. Melina and Bruna presented the studies related to natural compounds: “Discovery of flavonoids from Pterogyne nitens with potent activity against Zika virus protease and helicase” and “Discovery of new Zika virus candidates: natural products from Angelica keiskei with activity against NS2B-NS3 protease,” respectively. Paulo presented the work "Integrative Similarity analysis, Docking and Machine Learning models for identifying new Zika NS5 hits guided by Dengue NS5 inhibitors.”  Paulo also presented this work at the 16th CONPEEX (Research, Teaching and Extension Congress of Federal University of Goiás - UFG) and the work was selected for the UFG Undergraduate award.

Dr. Roosevelt presented the ZIKV NS1 molecular dynamics poster: “Dynamic Behavior of Dengue and Zika viruses NS1 protein reveals monomer-monomer interaction mechanisms and insights to rational drug design."

Dr. Sean Ekins also participated in the 9th BrazMedChem conference as the Keynote Speaker, presenting the talk “The Next Era of Pharmaceutical Research: From Bayesian Models to Deep Learning” regarding updated results of the OpenZika Project as well as his work on Chagas disease and Ebola.

Dr. Melina Mottin recently presented updates on the OpenZika project as oral presentation at RENORBIO 2019: II Meeting Biotechnology of Northeast in Fortaleza, Brazil, on November 28, 2019.

Recently, Prof. Carolina Horta Andrade gave the opening conference at the III Young Medicinal Chemist Workshop in Salvador, BA, Brazil, on November 21, 2019. In her lecture,she showed the OpenZika Project and our latest and exciting results.

Status of the calculations

In total, we have submitted almost 9.29 billion docking jobs, which involved 427 different target sites. Our initial screens used an older library of 6 million commercially available compounds, and our current experiments utilize the new ZINC15 library of 30.2 million compounds.

The 80,000 volunteers who donated their spare computing power to OpenZika gave us 92,696 CPU years' worth of docking calculations! Thank you all very much for your help!!

We have received all of the results for our experiments that involve docking 30.2 million compounds versus NS1,NS helicase (both the RNA binding site and the NTPsite), NS5 RNA polymerase (NTP and RNA pocket), NS5 methyltransferase(SAM and GTP site), NS2B/NS3 protease, capsid (binding pockets 1 and 2) and envelope protein.

We are incredibly grateful for all the volunteers who donated their unused computing time to this project! Thank you very much!!

References

1. STOPTOX 1.0. LabMol, 2016. Available at: http://stoptox.labmol.com.br/. Access on: Dec 02. of 2019.
2. Pred-hERGA novel web-accessible computational tool for predicting cardiac toxicity. Braga, R.C.; Alves, V.M.; Silva, M.F.B.; Muratov, E.; Fourches, D.; Liao, L.M.; Tropsha, A.; Andrade, C.H. Mol. Inf.2015,34, 698-701.10.1002/minf.201500040