Zika Virus Targeting by Screening Inhibitors against NS2B/NS3 Protease

Zika flavivirus is suspected to cause Guillain-Barre syndrome in adults and microcephaly, along with other congenital abnormalities in infants. Presently, no vaccines or therapeutics are available. Here, we report novel compounds identified by high-throughput virtual screening of Maybridge chemical database and molecular docking studies. We selected viral enzyme NS2B/NS3 serine protease as the therapeutic target because of its important role in viral replication. We selected seven potential compounds as antiviral drug candidates because of their high GOLD fitness score, high AutoDock Vina score, or X-Score binding energy and analyzed the strength of molecular interactions between the active site amino acids and selected compounds. Our study also provides a foundation for similar studies for the search of novel therapeutics against Zika virus.


Introduction
Zika virus (ZIKV) is a reemerging mosquito-borne pathogen that belongs to the family Flaviviridae and genus Flavivirus [1]. e other members of this family include dengue virus, West Nile virus, Japanese encephalitis virus, and yellow fever virus [2]. ZIKV was first accidentally isolated in 1947 by the Uganda Viral Research team from a sentinel rhesus monkey in the Zika Forest while searching for yellow fever virus [3]. Later in 1948, the same virus was isolated from Aedes africanus mosquitoes [4]. Humans serve as the amplifying host for ZIKV, while Aedes mosquitoes are the vectors transmitting the pathogen. e first Zika virus infection was reported in 1954 in a young girl in Nigeria [5,6]. Later, the virus travelled to Asia and was first identified in Aedes aegypti mosquitoes in Malaysia in 1966 [7], while the first human infection was reported in Java, Indonesia, in 1977 [8]. Until 2007, ZIKV was not considered a significant human pathogen. However, an outbreak of fever and rash caused by this virus in the Yap Islands changed this notion.
is small outbreak had 49 confirmed cases and more than 50 probable cases of ZIKV infection in humans [9]. A couple of years later, a second larger outbreak took place in French Polynesia, where 333 cases were confirmed, and more than 19000-32000 suspected cases were observed. It was during this period that the relationship between ZIKV infection and Guillain-Barre syndrome was first noticed [10]. During subsequent outbreaks in Brazil in 2014-15, more evidence was collected pointing towards the possible connection of congenital brain anomalies and microcephaly in children exposed to ZIKV infection [11,12]. e mode of transmission of infection is either through the mosquito, Aedes, or blood transfusion from Zika-infected patients. In some cases, it was found that the Zika infection spread through sexual contact. It was found in the infected patients that the virus persisted strongly in semen longer than in blood, urine, and saliva [13][14][15]. Infected women may also transmit the virus to her growing fetus or to the baby during birth. erefore, there is strong evidence that the ZIKA infection can spread from human to human by different methods, such as blood transfusion, sexual contact, and from mother to fetus.
WHO announced ZIKV as a public health emergency of international concern in February 2016. Since then, researchers worldwide have tried to understand this virus and its pathogenesis more deeply. To this date, there are no vaccines or antiviral drugs to prevent or treat ZIKV infections, though many vaccine candidates are presently under development [16]. erefore, alternate therapeutics are the need of the hour to prevent and inhibit the spreading of infection, as well as for treatment [17].
Zika virus is an enveloped spherical virus with 10.7 kb single-strand positive-sense RNA genome. Genomic RNA of ZIKV consists of a single open reading frame (ORF), with untranslated regions (UTRs) at 3′ and 5′ ends. A single large polypeptide consisting of 3423 amino acids is encoded by viral ORF which is then cleaved by both viral and host proteases into three structural proteins, namely, envelope (E), premembrane (prM), capsid (C), and seven nonstructural proteins called as NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5 [18]. e structural proteins help in the creation of virus particles [19]. e C-reactive protein forms the nucleocapsid after binding to the RNA genome. e prM protein keeps a check at the untimely fusion of viral particles with the host-cell membranes [20]. e E protein is the key player in host recognition via receptor binding and fusion. e host's antibodies also mainly target the fusion loop region of ZIKV E protein [21]. ere is a small hidden region called M protein within the E protein. Both these proteins (E and M) form the icosahedral-shaped arrangement in 60 repeating units [19]. e nonstructural proteins of the virus take care of replication, evading host immune responses, assembly, etc. [22]. ere are two important nonstructural proteins playing a major role in viral replication: one is NS3, comprising serine protease at its N-terminal, which needs another viral protein NS2 for its activity. e NS3 protein contains RNA helicase and RNA triphosphatase activities at its C-terminal region, which engages in RNA synthesis as well as capping functions [23]. NS2B-NS3 protease has been known for cleaving 31 human proteins, which were identified by unbiased N-terminomics, particularly during Zika virus infection depletion of protein such as autophagy-related protein 16-1 (ATG16L1) and eukaryotic translation initiation factor 4 gamma 1 (eIF4G1), which mediates type-II interferon production and host-cell translation, respectively, possibly involve in immune system evasion and driving the Zika life cycle as well as in mediating neurotoxicity through cleavage of host proteins [24,25]. ZIKV-NS2A destabilizes the adherens junction complex and results in infection in the developing mammalian brain [26]. e two-protein component system of NS2B/NS3 presents an excellent target for the development of antiviral drugs against ZIKV. e enzyme complex NS2B/NS3 plays an important role in viral replication inside the host, as well as helps the virus in evading innate immunity [27]. NS3 has both a protease domain and an RNA helicase domain at its C-terminal. NS3 protein requires another membrane-bound protein NS2B for proper orientation and function of enzyme [28]. Hydroxychloroquine inhibits NS2B-NS3 protease activity, which in turn results in significantly decreasing Zika virus infection in placental cells [29]. e recent research studies on the development of therapeutics against ZIKV have targeted nonstructural proteins that are essential for the replication of viruses.
ese include development of inhibitors for NS2B/NS3 protease, in particular. A few researchers have reported different inhibitors against NS2B/ NS3 serine protease, though these findings are still in early lead phases [27,30,31].
Here, we report seven potential inhibitors against NS2B/ NS3 protease identified through high-throughput virtual screening of Maybridge database. Molecular docking studies are also conducted for the protein and selected compound complex. e nature of molecular interactions between the protein (active site amino acids) and the selected compounds is also explored. Our study provides a foundation for the development of new structure-based inhibitors as future drug candidates against Zika virus.

Library Selection.
We have selected the Maybridge database for virtual screening. e Maybridge database (over 53,000 compounds) is a commercial library of small hit-like and lead-like compounds. Out of these compounds, the HitCreatorTM Collection (14,400 compounds) represents the diversity covering the drug-like chemical space. Maybridge also has a fragment library (30,000 fragments), a hit-to-lead building block collection, and a rule of 3 at least 2500 diversity fragment library with a Tanimoto similarity coefficient of 0.6.6.

Target Preparation and Virtual Screening.
e crystal structure of ZIKA NS2B-NS3 (PDB-ID: 5GXJ) was retrieved from the Protein Data Bank [32]. Protein is prepared using the removal of water molecules followed by addition of hydrogen atoms. e amino acids His1051, Gly1133, r1134, Ser1135, and Tyr1150 were used as the active sites for docking experiments. Virtual screening has been performed by using the AutoDock Vina (http://vina.scripps. edu/download.html) and GOLD programs. GOLD (Genetic Optimization for Ligand Docking) is a commercially available docking program produced from the collaboration between the University of Sheffield, GlaxoSmithKline plc, and Cambridge Crystallographic Data Center [33]. It is an automated ligand-docking program that uses a stochastic method with a genetic algorithm to explore the full range of ligand conformational flexibility with partial flexibility of the protein, and it satisfies the fundamental requirement that the ligand must displace loosely bound water on binding. Best scored compounds from the GOLD and AutoDock software were selected for their mode of interaction analyses. Best scored and conformation of compounds from the GOLD and AutoDock software were selected for their mode of interaction analyses. Selecting the promising conformation from the docking study is crucial for determining accurate accuracy of the ligand within the active site of the target [34,35]. e default setting available in the GOLD and AutoDock software was used to select the best conformation of the ligand.

Postdocking
Analysis. X-Score, a consensus scoring function, uses the negative logarithm of the dissociation constant of the ligand to the protein,ÀlogKd, as the average of three scoring functions (HP Score, HM Score, and HS Score) [36]. X-Score was known to have an accuracy of ±2.2 kcal/mol relative to the actual binding energies. For analyzing the interactions of docked protein-ligand complexes, the LIGPLOT programme was used to check the hydrogen bond and hydrophobic interactions between the receptor and the ligand [37]. Also, PyMOL (V-1.3) [38] and Discovery Studio Visualizer (http://accelrys.com/products/ collaborative-science/biovia-discovery-studio/visualizationdownload.php) were used to visualize the interactions and prepare figures for top-ranked molecules.

Results and Discussion
e in silico approach of high-throughput virtual screening has become one of the most popular techniques for drug discovery, currently. With this approach, not only very large amounts of data can be analyzed quickly, but also an immense amount of energy, time, and costs related to drug discovery research can be saved [39].
ZIKV has long been neglected since its first identification, assuming its low/mild pathogenicity in humans. But the recent suggestions of strong links between ZIKV infection with Guillain-Barre syndrome in adults and congenital defects and microcephaly in newborns have brought the attention of scientific fraternity to this virus. Presently, there are no effective treatments available to cure ZIKV infection, and hence, development of novel antiviral therapeutics is urgently needed.
In the present study, we used in silico virtual screening and docking approach to identify potential chemical molecules active against the viral protein NS2B/NS3 serine protease complex. ese identified molecules may pave way for the development of future antiviral drugs against Zika virus. e Maybridge chemical database was used for virtual screening of lead-like compounds. Docking simulation was performed by using the AutoDock Vina and GOLD docking software. GOLD allows full flexibility of the ligand and partial flexibility of the receptor [40]. e docked compounds were assessed on the basis of the GOLD fitness score. To reduce the chances of false results, only compounds with higher fitness scores were chosen for further studies. To understand likely interactions between lead compounds and proteins, we analyzed the docking results through the GOLD fitness score, constructive binding, and strong interactions of important amino acids of the protein and the compounds. e best compounds selected on the basis of the GOLD fitness score were further analyzed with X-Score. It predicts the binding energies by calculating the negative logarithm of the dissociation constant of the compounds to the protein. It predicts the binding energies with the accuracy of ±2.2 kcal/ mol [41]. Boronate inhibitor (PubChem ID: 16740933) was used as a reference compound. Recently, this compound has been used as reference molecules for screening the database against the NS2B-NS3 target [42].
On the basis of the docking program's binding affinity, 7 compounds were selected from the Maybridge database having high GOLD fitness scores and binding energies. ese compounds have been found to have better binding energy (-6.9 kcal/mol) and have shown strong interactions with active site amino acids ( Figure 1). Compounds (except HTS03171 and CD03173) were also found to have drug-like properties [43]. HTS03171 and CD03173 were found to have one Lipinski rule of violation. ese compounds have been listed in Table 1. Figure 2 represented the 2D structures of best-selected compounds. Drug-like properties of the compounds are listed in Table 1. e highest GOLD fitness scores of 72.01 and 71.21 were shown by CD03173 and JFD01698, respectively (Table 2). CD03173 was found to have binding energy from AutoDock Vina. Compounds HTS07252 and HTS04601 also displayed fitness scores of 69.68 and 69.59, respectively. Compounds CD11575, HTS03171, and KM10383 also formed stable complexes with proteins with fitness scores of 68.67, 67.81, and 62.45, respectively. Moreover, binding energies from AutoDock Vina for these compounds were -8.1, -6.9, and -6.8 kcal/mol, respectively.
Within the protein-ligand complex, hydrogen bonds play an important role in determining its specificity and affinity [44]. To evaluate the affinity of the selected compounds with the enzyme, we also investigated the amino acids of proteins involved in forming hydrogen bonds with the compounds and the strength of these bonds. Amino acids involved in hydrophobic interactions between proteins and ligands were also studied. Hydrogen bonds and hydrophobic interactions play a crucial part in giving shape and stabilizing the protein-ligand complexes [45]. ese molecular interactions may be supportive in enhancing proximity of ligand in the active site of protein, thus helping in its biological activity [46]. us, we also did a comprehensive analysis of the molecular interactions, hydrogen bonds, and hydrophobic interactions, in particular, within the active site of NS2B/NS3 protein, for the selected compounds (Table 2 and Figure 3).
ese inhibitors are commercially available and can be used for further experimental verification. e information obtained from this study can be used in the future for the development of more inhibitors against Zika virus.

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare that they have no conflicts of interest.