Assessing the potential of NS2B/NS3 protease inhibitors biomarker in curbing dengue virus infections: In silico vs. In vitro approach

An increase in the occurrence of viral infectious diseases is a global concern for human health. According to a WHO report, dengue virus (DENV) is one of the most common viral diseases affecting approximately 400 million people annually, with worsening symptoms in nearly 1% of cases. Both academic and industrial researchers have conducted numerous studies on viral epidemiology, virus structure and function, source and route of infection, treatment targets, vaccines, and drugs. The development of CYD-TDV or Dengvaxia® vaccine has been a major milestone in dengue treatment. However, evidence has shown that vaccines have some drawbacks and limitations. Therefore, researchers are developing dengue antivirals to curb infections. DENV NS2B/NS3 protease is a DENV enzyme essential for replication and virus assembly, making it an interesting antiviral target. For faster hit and lead recognition of DENV targets, methods to screen large number of molecules at lower costs are essential. Similarly, an integrated and multidisciplinary approach involving in silico screening and confirmation of biological activity is required. In this review, we discuss recent strategies for searching for novel DENV NS2B/NS3 protease inhibitors from the in silico and in vitro perspectives, either by applying one of the approaches or by integrating both. Therefore, we hope that our review will encourage researchers to integrate the best strategies and encourage further developments in this area.


Introduction
Dengue virus (DENV), an RNA virus belonging to the family Flaviviridae and genus Flavivirus, is a fatal pathogenic arthropodborne virus (arboviruses). It is predominantly transmitted by Aedes aegypti and, to a lesser extent, Aedes albopictus. The disease is widespread in more than 110 countries, infects approximately 400 million people, and results in approximately 20,000 deaths annually (Liang Gao and Gould, 2015;World Health Organization, 2020). Over the past few years, the occurrence of dengue fever (DF), dengue hemorrhagic fever (DHF), and dengue shock syndrome (DSS) has significantly increased in major tropical regions, with alarming frequency, magnitude, and bearing dire consequences (Tiga-Loza et al., 2021). DENV has four antigenically distinct serotypes (DENV1-DENV4) with 65-70% identical genome sequences. Each DENV serotype comprises four-seven genotypes that differ by 10% at the amino acid level across the envelope protein. The four serotypes differ not only in sequence similarity but also in infection dynamics. For example, DENV-1 is the most common serotype, followed by DENV-2, which is more frequently associated with severe infections. However, the mechanisms underlying dengue infections, as well as the entire set of distinctions across serotypes, remain unknown. However, a few recent studies have investigated the differences between the serotypes (Delli Ponti and Mutwil, 2021;Katzelnick et al., 2021;Stica et al., 2022). In light of the above, we hope that the source, breadth, and impact of antigenic heterogeneity can be better understood, which will aid in the exploration of effective dengue inhibitors or vaccines.
One of the major milestones in combating dengue infection was the first licensed vaccine, CYD-TDV or Dengvaxia ® . Nevertheless, owing to some drawbacks and limitations in the ongoing trials, it was found that the vaccine increased the risk of developing a severe form of dengue infection in some receivers (Redoni et al., 2020;Tully and Griffiths, 2021). This has led researchers to accentuate the development of potent inhibitors that can curb infection. Therefore, it is crucial to explore drugs directed at viral targets or critical host mechanisms that can be used as prophylaxis or treatment for the disease. Drug efficacy in the effective amelioration of the disease or the reduction of disease severity and fatalities is needed to lower the burden of dengue (Low Gatsinga et al., 2018).
This study examined 105 studies published in the Scopus citation database, MEDLINE, PubMed, and Google Scholar from 2015 to 2022. The indexed articles focused on discovering potential DENV NS2B/NS3 protease inhibitors using in silico and in vitro approaches, either by integrating both or applying one of them. Hence, search strings tailored to each database were devised for the dengue NS2B/ NS3. Mendeley (Elsevier, London, England) was used to compile references for the identified articles, and duplicates were removed. All identified abstracts were examined and selected based on preset criteria. A systematic review of this paper began by tabulating significant potential inhibitors in silico and in vitro studies, followed by a Venn diagram illustrating the strategy distribution ( Figure 1). This review seeks to explore a better understanding of NS2B/NS3 proteases and their therapeutic inhibitory potential and thus enlighten researchers on integrating the best strategies in this area.

DENV polyproteins
Morphologically, dengue viruses are approximately 50 nm in diameter with an open reading frame (ORF) of over 10,000 bases (Hahn et al., 1990). Upon infection, the positive-sense single-stranded RNA genome is replicated and translated in the endoplasmic reticulum (ER), where host ribosomes translate RNA into polyproteins. These nascent proteins are further broken down by host and viral proteases into structural and nonstructural (NS) proteins ( Figure 2). The ORF is flanked by two untranslated regions (UTRs) that contain structural and functional elements essential for viral translation and replication. The UTRs are translated into polyproteins that are processed co-and post-translationally by the host and DENV proteases to produce ten mature viral proteins. From the N-terminal region, three structural proteins are encoded in the Nterminal region: capsid protein (C, 11kDa), membrane protein (M, 8kDa), and envelope proteins (E, 53 kDa) (Hosseini et al., 2018). NS proteins are essential for viral replication and are retained in all DENV serotypes (Ahmad and Poh, 2019). Hence, these proteins are important components of the DENV genome replication machinery. Table 1 briefly describes each protein and its relevance to viral pathogenicity. Venn diagram illustrating the distribution of the in silico, in vitro and approach integrating both strategies.

DENV NS2B/NS3 protease as drug target
NS3 is a large multifunctional protein with serine protease (with NS2B as a cofactor), 5′-RNA triphosphatase (RTPase), nucleoside triphosphatase (NTPase), and helicase activity (Wengler and Wengler, 1991;Warrener et al., 1993;Li et al., 1999). The Nterminal 170 amino acids of NS3 have protease activity and a hydrophobic core of approximately 40 amino acids within NS2B that provides an essential cofactor function (Hahn et al., 1990;Chambers et al., 1991;Falgout et al., 1991). NS3 protease (NS3 pro) is a trypsin-like serine protease with a classic serine protease catalytic triad consisting of His51, Asp75, and Ser135 residues (Bazan and Fletterick, 1989). All four DENV serotypes have approximately 65-74% amino acid sequence homology and a common substrate preference (Li et al., 2005). The C-terminal b-hairpin of NS2B in its catalytically active form wraps around the active site of NS3 ( Figure 3) (Erbel et al., 2006). Consistent with the important structural role of the C-terminal b-hairpin of NS2B, structural comparisons indicated that the amino acids within the N-terminal portion displayed similar conformations in all structures, regardless of the presence or absence of inhibitors.

Prospective developments
Considering the global threat of DENV and the urgent need for effective drugs, several efforts have been made to identify potential protease inhibitors. The development of NS2B/NS3pro inhibitors began with the structure-activity relationship of NS2B-NS3pro, inferred from the well-established cleavage sites of the DENV polyprotein by NS2B-NS3pro. This led to the discovery of two tetrapeptides, Bz-Nle-Lys-Arg-Arg and Bz-Nle-Lys-Thr-Arg, that have been shown to have high affinities for NS2B-NS3pro (Ki¼12.42 and 33.9 mM, respectively) (Yin et al., 2006;Othman et al., 2007;da Silva-Júnior and de Araújo-Júnior, 2019). Subsequently, efforts have been made to design peptidomimetics that have the ability to mimic the natural substrate (Gibbs et al., 2018;da Silva-Juńior and de Araujo-Juńior, 2019;Drazǐćet al., 2020).
The latter group of inhibitors has long been recognized as an invaluable component of medicine, and many targeted therapies have focused on these small-molecule drugs. These low molecular weight (less than 900 Da) organic compounds help to control biological targets, such as enzymes, channels, or receptors, to alter the disease cycle (Phanthanawiboon et al., 2014;Lenci and Trabocchi, 2019). At present, 90 percent of the therapeutics in the pharmaceutical market are small-molecule drugs. These include ten clinically available human immunodeficiency virus 1 (HIV-1) protease inhibitors and hepatitis C virus (HCV) protease inhibitors (De Clercq, 2009;Manns and Von Hahn, 2013). These facts also suggest that protease inhibitors of the dengue virus could be clinically effective. In the last decade, the development of small molecule NS2B/NS3pro inhibitors has involved high-throughput screening (HTS) of the  Illustration of DENV structural and NS proteins on the ER with its cleavage sites. Modified from Uno & Ross, 2018. natural product (Kiat et al., 2006;de Sousa Wu et al., 2015), and synthesis of rational drug design (Liu et al., 2014;Viswanathan et al., 2014;Raut et al., 2015a) with virtual screening using computer-aided drug design (CADD) being in-process (Cabarcas-Montalvo et al., 2016). This review highlights the recent development of DENV inhibitor successors, mainly small molecules. Owing to advances in bioinformatics in drug discovery, non-peptide antiviral activity evaluation has been explored in vitro, as well as in silico and in HTS (Kanakaveti et al., 2020). Weighing the benefits of both approaches provides greater knowledge and an understanding of the anti-DENV drug development pipeline. Summarizing our findings on the methods used in developing NS2B/NS3pro inhibitors, this study highlights methods that are relevant to this co-protein only. Methods were classified into three cohorts: studies focusing on in silico methods, in vitro methods, and both. By subdividing these approaches, we hope this analysis will promote further progress in discovering potent inhibitors of fatal arbovirus infections.

In silico approach
Various computational tools have been used to identify small target molecules for dengue drug discovery. Structure-based drug design (SBDD) methods, namely molecular dynamics, fragmentbased drug design, pharmacophore modelling, and most importantly, molecular docking, have provided information about many molecules, including DENV protein targets, such as NS2B/ NS3pro. Among the above mentioned methods, molecular docking is the most popular for searching for potential NS2B/NS3pro inhibitors. The aim of molecular docking is to determine the best ligand-binding positions in the NS2B/NS3pro binding pocket and estimate the affinity of the ligand for the protein (Jakhar et al., 2019). To date, 13 crystal structures of DENV NS2B/NS3pro with different PDB codes have been solved for all dengue serotypes (Table 2).
However, ligands mostly originate from virtual libraries comprising thousands to millions of compounds. The most commonly used docking software in NS2B/NS3pro studies are AutoDock, AutoDock Vina, and Molecular Operating Environment (MOE). These platforms have an algorithm for identifying the NS2B/ NS3pro active site by allowing small drug-like molecules to bind to different parts of the protein. The best ligand-protein affinity and binding positions were then observed (Jakhar et al., 2019). Nevertheless, it is essential to observe the hydrogen bonding and optimize the hydrophobic interactions, as they are the key players in obtaining stable energy-favored ligands at the interface of a protein structure and help in modifying the binding affinity for the drug's effectiveness. Studies that have applied only in silico-based approaches to explore the interaction between NS2B/NS3pro and its possible inhibitor candidates in recent years are tabulated in Table 3.
In conclusion, this approach determines the best-fitting ligand positions in the NS2B/NS3pro binding pocket and estimates the affinity of the ligand to the protein. The in silico approach uses crystal structures of the DENV NS2B/NS3pro protein with various PDB codes as well as ligands from virtual libraries containing hundreds to millions of chemicals. The ideal ligand-protein affinity and binding location can be determined using the software. However, the crucial point is that many chemical compounds and peptides have shown significant in silico binding affinity towards viral targets, but their affinity has yet to be evaluated using in vitro methods in many cases. Hence, the mechanism underlying the inhibition of most peptides remains unknown.

In vitro approach
According to Lim et al., virtual hits derived from in silico docking require further validation by in vitro methods. These methods can verify on-target effects in cells (Lim, 2019). The in vitro assays are commonly performed to investigate the inhibitory properties of candidates against NS2B/NS3pro, as briefly described in Figure 4. Plaque, cytotoxicity, and immunofluorescence (IF) assays are examples of cell-based assays that provide substantial information on various cellular responses to compound exposure. Therefore, choosing the right cell type based on the target biology is critical. Among the cell types used in recent dengue inhibition studies, Vero, X-ray crystal structure of catalytically active conformation of DENV NS2B/NS3 pro (PDB code: 2FOM). Grey ribbon: NS3 structure; red ribbon: NS2B cofactor; yellow ribbon: S1-ß-hairpin; and blue ribbon: Erbel et al. (2006).  29   (Phanthanawiboon et al., 2014). According to the Guidance for Industry-Antiviral Product Development by the US Food and Drug Administration (FDA), the specific antiviral activity was quantitatively measured by calculating the replication of the virus in the presence of increasing drug concentrations as opposed to replication in the absence of the drug. Therefore, to evaluate drug potency, the inhibition concentration (IC 50 ) and effective concentration (EC 50 ) must be measured (FDA, 2006). Nearly all recent studies have reported the IC 50 of the tested compounds and their activities against the dengue enzyme. Quantification was performed using a protease inhibition assay, which measures the inhibitory activity of the drug candidates and the catalytic activity of the proteolytic enzyme.
As mentioned previously, in vitro drug potency measurements are essential for drug discovery. We reviewed recent studies that evaluated the EC 50 of inhibitors through cytotoxicity tests (Chu Lee et al., 2015;Li et al., 2015;Lim et al., 2020) or plaque assays, such as In vitro approach in DENV NS2B/NS3 pro inhibition study.
the time of drug addition (Li et al., 2018;Yao et al., 2018), viral plaque reduction (Brecher et al., 2017;Li et al., 2018), and viral titer reduction assay (Brecher et al., 2017). In addition to identifying an effective drug, it is crucial to determine the cytotoxic potential of the tested compounds in the drug discovery process (Slater, 2001). Briefly, the cytotoxic concentration (CC 50 ) of the compounds that caused a reduction in cell viability was measured using a dilution assay. In recent studies, MTT (Raut et al., 2015b;Cabarcas-Montalvo et al., 2016;Beesetti et al., 2018;Li et al., 2018) or other cytotoxicity assays have been used to observe their effects on virus-infected host cells. The antiviral activity of the compounds was tested at different concentrations. Hence, developing potential inhibitors with lower cytotoxic concentrations is recommended (Alagarasu et al., 2022). In conclusion, many DENV NS2B/NS3 pro-inhibitor candidates have yet to be subjected to cytotoxicity investigation, making these products uncertain for further development.
Protease assays are another key pre-clinical assay in investigating protease inhibitors and their activity. For most inhibitor candidates, the target enzymatic activity was quantitatively determined to test their efficacy against the NS2B/NS3 pro-enzyme (Raut et al., 2015b;Brecher et al., 2017;Osman Idris et al., 2017;Beesetti et al., 2018;Euanorasetr et al., 2019;Hariono et al., 2019;Lim et al., 2020;Saleem et al., 2019;Sulaiman et al., 2019). The activity was determined if the tested compounds modulated the DENV NS2B/NS3 pro-enzyme function. Here, we highlight the recent five-year studies that applied only the in vitro approach (Table 4) and a combination of in vitro and in silico approaches (Table 5) to determine protease enzymatic activity.
By using the in vitro methods, the on-target effects in cells can be verified using in vitro methods. The candidate inhibitory activities against NS2B/NS3pro were evaluated using a protease inhibition assay. Moreover, plaque, cytotoxicity, and IF assays can provide valuable information on the diverse cellular responses to compound

Using a combination of methods
In summary, incorporating in silico and in vitro approaches to determine the potency of dengue inhibitors can lead to the development of more potential drug candidates. Furthermore, integrating in vitro methods with in vivo assessments will reduce the number of physiologically relevant potential candidates and evaluate their characteristics simultaneously. It will also evaluate drug-drug interactions (DDI) and help comprehend the underlying mechanisms of drug candidates. Additionally, combining these approaches will help verify the relevance of in vitro results. Thus, substantiating the extrapolation of in vitro outcomes to the clinical phase of the drug development pipeline.
In addition, we would like to highlight the sources of DENV NS2B/NS3 pro candidates. In our study, the small-molecule or nonpeptide candidates explored were either synthetic (25 studies) or derived from natural sources (16 studies). Currently, most medicines used in clinical practice are synthetically formulated and include chemical processes (reactions) and phytochemicals. The four anti-DENV drugs under clinical trials, celgosivir, UV4B, chloroquine, and balapiravir (Anasir et al., 2020), are small synthetic molecules developed from natural sources. As synthetic drugs have benefits such as chemical purity, a simple and cost-effective preparation process, and higher quality, more effective and safer drugs can be prepared by altering the chemical structure of the drug prototype.
Alternatively, using natural sources is a well-established method for discovering new substances with possible therapeutic effects. This class of drugs comprises new bioactive compounds that are essential for the production of modern medicines (Kumar et al., 2019). It also provides information on different classes of bioactive lead compounds for the discovery and development of novel drugs. Recent studies have focused on the bioactive compounds present in plants, such as Carica papaya (Ghosh and Talukdar, 2019;Farooq et al., 2020), Azadirachta indica (Dwivedi et al., 2016), Ganoderma lucidum (Bharadwaj et al., 2019), Ganoderma lucidum var. antler (Lim et al., 2020), Curcuma longa (Balasubramanian et al., 2019), Endiandra kingiana (Sulaiman et al., 2019), Cynodon dactylon (Chandani et al., 2019), Dryobalanops aromaticum , Acorus tatarinowii Schott (Yao et al., 2018), and Syzygium aromaticum (Saleem et al., 2019). The above mentioned studies included the extraction of crude plants (or plant parts) in solvents, mainly methanol, before investigating its activity against DENV NS2B/NS3 pro. Nonetheless, from our observation, both cohorts led to potent inhibitors with promising activity against the DENV NS2B/NS3 proenzyme, which has the potential to progress to the next anti-DENV drug development phase.

Conclusion
The number of hits, particularly those obtained from in silico docking, should be verified using an in vitro approach. It is also important to fully characterize the hits identified from the compound libraries. Furthermore, to produce a promising DENV antiviral inhibitor, verifying its activity using in vitro methods is crucial. To further ascertain the outcome of the two approaches, incorporating in vivo assessments can be beneficial, as they can substantiate the in vitro outcomes to the clinical phase in the drug development pipeline. These combined approaches can lead to promising antiviral candidates that may curb dengue infection. Additionally, along with small drug-like molecules, the search for dengue inhibitors should focus on using peptides. As signaling molecules, this possible approach exhibits complex biological roles with high selectivity and comparatively safe criteria.
Furthermore, we emphasize that a DENV inhibitor must be effective against all four DENV serotypes, as these serotypes cocirculate in highly endemic regions [95]. Nevertheless, it is essential to remember that the plausibility of dengue serotypes, together with other factors, such as secondary infection by a heterologous serotype, age, comorbidity, poor clinical prognosis, diagnosis, virulence, and the host immune response, contribute to the development of severe dengue infection (Puc et al., 2021). Finally, considering the recent attempts to identify DENV NS2B/NS3pro inhibitors, a range of antiviral targets display antiviral intervention potential. Although small-molecule inhibitors require clinical approval, promising dengue antivirals will be possible soon.

Author contributions
LS designed the study, HN carried out the data collection, HN, KE, LH, data analysis and interpretation. LS. and HN, drafted the article. HN, RV, RA and AH edited the article. All authors read and approved the final article. Authors contributed equally for the preparation of this review.