Discovery of novel West Nile Virus protease inhibitor based on isobenzonafuranone and triazolic derivatives of eugenol and indan-1,3-dione scaffolds

The West Nile Virus (WNV) NS2B-NS3 protease is an attractive target for the development of therapeutics against this arboviral pathogen. In the present investigation, the screening of a small library of fifty-eight synthetic compounds against the NS2-NB3 protease of WNV is described. The following groups of compounds were evaluated: 3-(2-aryl-2-oxoethyl)isobenzofuran-1(3H)-ones; eugenol derivatives bearing 1,2,3-triazolic functionalities; and indan-1,3-diones with 1,2,3-triazolic functionalities. The most promising of these was a eugenol derivative, namely 4-(3-(4-allyl-2-methoxyphenoxy)-propyl)-1-(2-bromobenzyl)-1H-1,2,3-triazole (35), which inhibited the protease with IC50 of 6.86 μmol L-1. Enzyme kinetic assays showed that this derivative of eugenol presents competitive inhibition behaviour. Molecular docking calculations predicted a recognition pattern involving the residues His51 and Ser135, which are members of the catalytic triad of the WNV NS2B-NS3 protease.


Introduction
The West Nile Virus (WNV) is a member of the same family as the Dengue virus (DENV), Zika virus (ZIKV) and Yellow Fever virus (YFV), the Flaviviridae family, Flavivirus genus. They are arboviruses that present RNA as a genome [1]. Diseases caused by Flavivirus are the major causes of fatality in poverty-stricken regions across Africa, Asia and some parts of the Americas. The combined potential health risk associated with arthropod-borne viruses like DENV, WNV, and ZIKV is enormous. These arboviruses are either emerging or re-emerging in many regions [2]. PLOS  Three WNV strains are known to be capable of causing unforeseen and large epidemics, leading to serious public health problems. Since 2004, lineages 1 and 3 have been circulating in Europe and, since 2010, beginning in a major epidemic in Greece, lineage 2 has been circulating in several European countries. [3,4]. The WNV crossed the Atlantic and reached the Western Hemisphere in 1999, when a group of patients with encephalitis was reported in the New York City metropolitan area. Within three years, the virus spread to Canada and Mexico, followed by animal cases in Central and South America [5,6]. Recently, the first human case of WNV was reported in Brazil, with the development of encephalitis. It is possible that sporadic cases or small groups of the WNV disease had already occurred in different regions of the country without being properly diagnosed [7].
WNV is a genetically and geographically diverse virus. Four or five distinct WNV genetic lines have been proposed, based on phylogenetic analyses of published isolates. Their genomes differ from each other by about 20-25%, and are well correlated with the geographic point of isolation [8][9][10]. They are enveloped viruses whose genome consists of single-stranded, positive-polarity RNA approximately 11 kb. This RNA contains a single open reading frame encoding a precursor polyprotein, which is processed by viral and host proteases, giving rise to three structural proteins: capsidial protein (C), envelope glycoprotein (E) and pre-membrane/ membrane protein (prM/M); and seven non-structural proteins, NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5, which are involved in the replicative cycle of the virus [11]. Viral protease performs the cleavage of some sites: NS2A-NS2B, NS2B-NS3, NS3-NS4A and NS4B-NS5. It also cleaves the signal sequences at the C-prM position and the NS4A-NS4B, within NS2A, and within the NS3 itself [12,13].
Despite the tremendous efforts invested in Flavivirus research, no clinically approved antiviral chemotherapeutics are available for humans, and disease treatment is limited to supportive care [13]. Inhibition of viral enzymes has proved to be one important approach toward the development of antiviral therapies [2,[13][14][15]. Non-structural proteins encoded by these RNA viruses are essential for their replication and maturation, and thus may offer ideal targets for developing antiviral drugs [2]. Flavivirus genomes are translated into a single polyprotein that needs to be cleaved by viral and host proteases. Because it processes most of the polyprotein cleavages, viral protease is necessary and essential for virus replication [16,17].

Synthesis
Solvents were purchased from Vetec (Rio de Janeiro, Brazil). Benzyl alcohols, pent-4-yn-1-ol, methanesulfonyl chloride, sodium azide, triethylamine, propargyl bromide, acetophenones, and indan-1,3-dione were procured from Sigma Aldrich (St. Louis, MO, United States) and used as received. Eugenol was extracted via hydrodistillation from cloves purchased in the local market in Viçosa, Minas Gerais state, Brazil, and subsequently purified by column chromatography (vide infra). 1 H-and 13 C-NMR spectra were recorded on a Varian Mercury 300 MNR Spectrometer (Varian, Palo Alto, CA, United States) at 300 MHz and 75 MHz, respectively, using CDCl 3 , C 6 D 6 or DMSO-d 6 as solvents. NMR data are presented as follows: chemical shift (δ) in ppm, multiplicity, the number of protons, and J values in Hertz (Hz). Multiplicities are shown as the following abbreviations: s (singlet), brs (broad singlet), d (doublet), d ap (apparent doublet), dd (doublet of a doublets), t (triplet), brd (broad doublet), ddt ap (apparent doublet of doublets of triplets), q (quartet), quint (quintet), and m (multiplet). Some signals in the 13 C NMR spectra were described as multiplets due to the 19 F-13 C coupling. IR spectra were obtained using a Varian 660-IR equipped with GladiATR (Varian, Palo Alto, CA, USA) scanning from 4000 to 500 cm −1 . Analytical thin-layer chromatography analysis was conducted on aluminum-backed, pre-coated silica gel plates using different solvent systems. TLC plates were visualized using potassium permanganate solution, phosphomolybdic acid solution and/or UV light. Flash column chromatography was performed using silica gel 60 (60-230 mesh). Melting points were determined using a MQAPF-302 melting point apparatus (Microquimica, Santa Catarina, Brazil) and are uncorrected. Solvents were dried using standard procedures described in the literature [42].
Extraction and purification of eugenol (19). Eugenol (19) was extracted via hydrodistillation from dried flower buds of Eugenia caryophyllata, commonly known as cloves, purchased in the local market in Viçosa, Minas Gerais state, Brazil. Thus, 60.0 g of cloves were mixed with 500 mL of distilled water in a round-bottom flask which was connected to the hydrodistillation apparatus. The mixture was heated for three hours. The obtained hydrolate was transferred to a separatory funnel and the aqueous layer was extracted with dichloromethane (3 x 30 mL). The organic extracts were combined and the resulting organic layer was dried over anhydrous sodium sulfate, filtered, and concentrated under reduced pressure. The resulting oil was submitted to column chromatography eluted with hexane-ethyl acetate (6:1 v/v). The described procedure afforded 7.12 g of eugenol (19), which corresponded to approximately a 12% yield in relation to the initial mass of cloves used in the extraction process.
Synthesis of 4-allyl-2-methoxy-1-(prop-2-yn-1-yloxy)benzene (20). A 50 mL roundbottom flask was charged with eugenol (19) (1.20 g, 7.32 mmol), sodium hydroxide (0.313 g, 7.83 mmol) and 25 mL of methanol. The resulting mixture was heated to 40 o C and magnetically stirred for 30 minutes. After this time, methanol was removed under reduced pressure and 10.0 mL of anhydrous ethanol was added for the removal of the residual water. The ethanol was removed under reduced pressure. Then, the round-bottom flask, under a nitrogen atmosphere, was charged with anhydrous acetonitrile (25.0 mL) and propargyl bromide (800 μL, 8.79 mmol) was added slowly. The mixture was magnetically stirred at room temperature for 18 hours. TLC analysis revealed the completion of the reaction after this time. The reaction mixture was concentrated under reduced pressure and the residue was partitioned between 25.0 mL of sodium hydroxide solution (0.1 mol L -1 ) and 25 mL of diethyl ether. The layers were separated and the aqueous phase was extracted with diethyl ether (2 x 25.0 mL). The organic extracts were combined and the resulting organic layer was washed with brine (25.0 mL), dried over anhydrous sodium sulfate, filtered and concentrated under reduced pressure. The residue was purified by column chromatography eluted with hexane-ethyl acetate (6:1 v/ v). The described procedure afforded 1.29 g (6.37 mmol, 87% yield) of compound 20.
Synthesis of pent-4-yn-1-yl methanesulfonate (21). Pent-4-yn-1-ol (1.68 g, 20.0 mmol) and dichloromethane (20 mL) were added to a 100 mL round-bottom flask under nitrogen atmosphere. The mixture was cooled to -50˚C and triethylamine (5.60 mL, 40.0 mmol) was added. After that, methanesulfonyl chloride was added slowly (2.32 mL, 30.0 mmol) to the reaction mixture under continuous stirring. The progress of the reaction was monitored by TLC. After completion of it, 10 mL of distilled water were added. The organic phase was washed with 1% HCl solution (3 x 15 mL) followed by saturated aqueous NaHCO 3 (3 x 5 mL), dried over anhydrous sodium sulfate, and concentrated under reduced pressure. The residue was purified by column chromatography eluted with hexane-ethyl acetate-dicloromethane (3:1:3 v/v) to give compound 21 in 92% yield (3.00 g, 18.0 mmol).
Compounds 24-51 were prepared using a procedure similar to that described for the synthesis of 23. The structures of these compounds were confirmed by NMR ( 1 H and 13 C), and IR analyses and are supported by the following data.
Compounds 52-63 were synthesized employing a procedure similar to that described for the preparation of 51. The structures of substances 52-63 are supported by the following data.

Determination of IC 50
The inhibitory enzymatic activity of compound 35, the one most active against the WNV NS2-NB3 protease, was evaluated at eight different concentrations (66 μmol L -1 -0.5 μmol L -1 ) using the protease assay as described above. Fluorescence was measured in triplicate wells at intervals of 30 s for 5 min in three independent experiments. IC 50 values were calculated using GraphPad Prism software 6 (GraphPad Software Inc., San Diego, CA, United States), using four-parameter nonlinear regression analysis (Hill slope method).

Determination of K i
Three different concentrations (2, 4 and 8 μmol L -1 ) of inhibitor 35 and five different concentrations of substrate pERTKR-AMC (20,40,60,80, 100 mmol L -1 ) were tested in vitro against the WNV protease (37.04 nmol L -1 protein, 1 ng μL -1 ). Fluorescence was measured in triplicate wells at an interval of 30 s. The velocity values (RFU/minute) were then calculated for each substrate/inhibitor pair. K i values were calculated with GraphPad Prism software 6 (GraphPad Software Inc., San Diego, CA, United States) with non-linear regression in the competitive inhibition mode of enzyme-kinetics.

Cytotoxicity assay
The cytotoxicity of compound 35 was assessed using an MTT assay [44]. VERO cells (5 x 10 4 cells) were seeded in 96-well plates. Each well contained 100 μL of each compound solution at different concentrations (1000, 250, 125, 63, 32, 17, 8 and 4 μmol L -1 ). The compound was diluted in MEM medium with 2% FBS and 1% DMSO. After 24 h of incubation at 37 o C, 100 μL of the MTT solution (5%) was added to the wells. After 4 h at 37˚C, the MTT solution was removed and 100 μL/well of DMSO was added to solubilize the formazan. Absorbance was measured at 550 nm in a microplate reader (Multiskan™ GO Microplate Spectrophotometer-ThermoFisher 1 , Waltham, MA, United States). The data were analyzed and CC 50 was determined using GraphPad Prism 6.

Virucidal assay
The virucidal assay was performed as described by Oliveira et al. [41].

Molecular modeling studies
Ligands were prepared in the Ligprep program (Schrödinger, New York, NY, United States) [45] employing the OPLS_2005 force field, with protonation states predicted using Epik at pH 9.5 ± 2.0. The West Nile Virus protease NS2B-NS3 PDB code 2IJO [46], chosen as the receptor, was prepared with the Protein Preparation Wizard (Schrödinger, New York, NY, United States), with removal of all waters and addition of hydrogens based on PROPKA calculations at pH 9.5. Docking calculations were performed with the Glide software (Schrödinger, New York, NY, United States) [47,48], employing the Induced Fit docking methodology [49] and Glide SP. All the software packages used are part of the Schrödinger Release 2016-2 package (Schrödinger, New York, NY, United States) [50]. A spherical grid with 12 Å radius was centered in the Isoleucine 123 residue so that the active site was fully included within the grid. All residues within 5 Å from the center were considered flexible. Docking results were ranked based on their docking score and the top ranking poses for each compound were analyzed with the Maestro 10.6 software (Schrödinger, New York, NY, United States) [51].
Then, the CuAAC reactions (click reactions) between benzyl azides (ArCH 2 N 3 ) and terminal alkynes 21 and 22 led to the formation of triazolic derivatives 23-49 (Fig 4). The click reactions, in general, took less than one minute for their completion.
It should be mentioned that the azides, used in the preparation of compounds 23-49 and 51-63, were obtained via the methodology previously described in the literature [52]. Once prepared, the synthesized compounds were submitted to biological assays to evaluate their inhibitory effects against the NS2-NB3 protease of WNV.

Identification of WNV protease inhibitors
The Flavivirus protease is essential for the processing of the polyprotein which generates the viral proteins required for viral replication and maturation of infectious virions. Therefore, the protease is an ideal target for the discovery of antivirals against Flavivirus [53]. In the present investigation, we evaluated the inhibitory activity of fifty-eight compounds against the WNV NS2B-NS3 protease (eighteen 3-(2-oxo-2-aryl)-isobenzofuran-1(3H)-ones, compounds 1-18; twenty-seven derivatives of eugenol with triazole rings, compounds 23-49; thirteen derivatives of indan-1,3-diones presenting triazole rings, compounds 51-63, Fig 2). In this regard, a purified preparation of WNV NS2B3NS3pro and the fluorogenic peptide substrate pERTKR-AMC was utilized.
Our primary screen resulted in the identification of eighteen compounds (~31% of the evaluated compounds) that exhibited inhibitory effects on protease activity (Fig 6).
Subsequently, we took these eighteen compounds and conducted a secondary screening with further validation based on their relative strengths of inhibition (at least 50%). This resulted in the selection of compound 35, a eugenol derivative which was utilized in further experiments.
A derivative of the natural product eugenol capable of significantly inhibiting the activity of the WNV NS2B-NS3 protease was identified. Taking the eugenol derivatives into Discovery of novel West Nile Virus protease inhibitor consideration, we made variations in the size of the carbon chain that links eugenol moiety and the triazole ring. Also, the substitution pattern of the aromatic ring attached to the triazolic portion was varied. These modifications afforded a group of eugenol triazolic derivatives from which was identified the very active compound 35. It is important to mention that further chemical modifications can be planned regarding the eugenol and triazole fragments so that new derivatives with improved activity may be obtained. Considering that there is no Flavivirus protease inhibitor approved for pre-clinical trial [54], the exploitation of the new scaffold herein identified, namely triazolic derivatives of eugenol, would be of considerable importance.

Determination of IC 50 and Ki values of the enzymatic inhibitory activity of compound 35
The enzymatic inhibition was evaluated in the presence of varying concentrations (0.5 μmol L -1 to 66 μmol L 1 ) of compound 35. A dose-response inhibition was noticed with IC 50 value of 6.86 μmol L -1 (Fig 7).
The enzyme kinetic assay was conducted under five different substrate concentrations and three varying concentrations of compound 35. We used the Michaelis-Menten equation to find the values of V MAX and K M ; and with these, the Lineweaver-Burk graph was built. Compound 35 showed a decrease of enzymatic V MAX and an increase in K M , presenting the behaviour of a competitive inhibitor (Fig 8A and 8B). The K i value was 3.06 (± 0.38) μmol L -1 .
For compound 35, the enzyme kinetic data showed an increase in the K M value, a typical behaviour of competitive inhibitors. In addition, this compound displayed a low K i value. In a recent investigation, Balasubramanian and collaborators [53] found eight promising flavivirus protease inhibitors presenting K i values within the 0.22 to 6.9 μmol L -1 range. However, the K i value may not be analyzed individually. In the study by Balasubramanian and collaborators, a compound presenting low K i value (0.22 μmol L -1 ) but displaying a CC 50 of 29.16 μmol L -1 was identified [53].
For the development of new drugs, the World Health Organization (WHO) strongly recommends the use of compounds with pharmacological effects already described and substances already approved for clinical use, so that several steps can be abbreviated in the long validation process [55]. Eugenol (19) has been safely used in in vivo experiments, and it has Discovery of novel West Nile Virus protease inhibitor been recognized as a safe, effective and inexpensive anesthetic for fish, amphibians and rats [56,57]. Also, the analgesic effect of eugenol (19) in different models of pain has been well documented [58][59][60][61][62]. A recent work with BALB/c mice used a derivative of eugenol (19) to treat visceral Leishmaniasis. This derivative presented low cytotoxicity for macrophages as well as for naive mice with immune-stimulatory activity. Moreover, no biochemical alterations in Discovery of novel West Nile Virus protease inhibitor hepatic and renal enzymes were noticed [63]. Eugenol (19) is generally non-allergenic for humans, although in sensitized individuals it may cause a range of tissue reactions from lowgrade local to systemic. Low concentrations of eugenol (19) are well known to exert local antiinflammatory, antiseptic, and anesthetic effects on dental pulp. Also, eugenol (19) may have antibacterial effects that are beneficial for dental hygiene, being included in materials such as toothpastes and mouthwashes [64][65][66]. All these features make eugenol (19), as well as its derivatives, very interesting compounds to be explored in drug development.

Cytotoxicity assay
The cytotoxicity of compound 35 on Vero cells was investigated via the colorimetric MTT assay and the determined CC 50 value was 327.20 μmol L -1 (Fig 9). Luo and collaborators [54] reported the best parameters for the most promising compounds against the protease of Flavivirus. They highlighted a compound with low cytotoxicity presenting CC 50 superior to 300 μmol L -1 . Eugenol derivative 35 presented a similar CC 50 , and it can be considered a compound with low cytotoxicity. Therefore, although a very potent inhibitor, this compound presented considerable cytotoxicity.

Molecular modeling
Multiple targets and inhibitory mechanisms have been proposed for Flavivirus proteases so far [53,54,67,68], which have provided insightful structural information regarding the NS2B-NS3 catalytic site, helping computational-aided drug design efforts even further. For instance, by solving NS2B-NS3's tridimensional structure bound to a peptide-like inhibitor, Erbel and coworkers [69] provided fundamental information regarding the protease fold, catalytic site structure and inhibition mechanisms. As found by others [46,[70][71][72], inhibitors often bind to His 51 or to nearby residues such as Asp 75 , Asp 129 , Gly 153 and Tyr 161 (S1 Fig), hampering the bond of the substrate to the catalytic site.
In order to perform our own molecular docking calculations, the crystallographic structure of the NS2B-NS3 protease was obtained from PDB 2IJO, in which NS2B-NS3 is co-crystallized with the WNV protease inhibitor aprotinin. All molecules derived from eugenol were analyzed for their probable three-dimensional binding conformation, binding energy, chemical groups involved, profile of binding and identity of the involved amino acids, as observed in table A in S1 File. The predicted recognition mechanism for compound 35 relies on interactions with His 51 , Thr 134 , Ser 135 and Tyr 161 , as represented in Fig 10A, and the ligand occupies pocket S1 of the catalytic site. The NH + of the His 51 ring interacts with the triazole ring of 35 in cation-π and π-π type interactions, and the Tyr 161 ring interacts with the phenyl ring of 35 via π-π type interaction, while a halogen bond can be formed between the Br atom and the OH from Ser 135 or the NH from Thr 134 . It is important to highlight that both His 51 and Ser 135 are members of the catalytic triad of the NS2B-NS3 protease, which might explain the observed competitive inhibitory activity of 35.
In addition, compound 36 showed no inhibitory activity, despite the fact that the only difference from compound 35 is the position of a bromo substituent (orto in 35 and meta in 36). Our calculations suggest that the formation of a halogen bond for 36 with bromo in meta might induce a more stable conformation for the ligand in which the triazole ring interacts with Gly 153 instead of His 51 and, consequently, leaves the S1 pocket free (Fig 10B). These results suggest that the bromo substituent in orto is pivotal for a proper recognition of compound 35, along with the cation-π interaction between His 51 and the triazole ring, while the methoxyphenyl ring region might be a suitable target for further improvements.

Virucidal assay
To determine whether compound 35 would have antiviral activity for Flaviviruses, a virucidal assay was performed with four serotypes of DENV, due to its close evolutionary proximity to WNV and due to the highly conserved NS2B-NS3 fold and sequence. This assay was performed by prior incubation of the compound with each viral strain, followed by its addition to the cell layer for virus adsorption and internalization. Subsequently, the compound-virus solution was removed and antiviral action was observed through the formation of lysis plates. The concentration of test compounds that inhibited 50% of the viral infection (EC 50 ) was obtained by nonlinear regression, leading to the calculation of the Selectivity Index (SI).The observed results are presented in Table 1.
The compound had good SI values for DENV-1-3 and a higher value for DENV-4. All together, the results indicate that compound 35 has significant antiviral efficacy and is a promising antiviral candidate for Flavivirus.

Conclusion
In this work, a small library of fifty-eight synthetic compounds (isobenzofuran-1(3H)-ones and triazolic derivatives of eugenol and indandione) were screened against the WNV NS2B-NS3 protease. By modifying the structure of the natural product eugenol to produce 1,2,3-triazolic derivatives, a compound presenting low cytotoxicity and considerable inhibitory protease activity was identified. Compound 35 corresponds to 4-(3-(4-allyl-2-methoxyphenoxy)propyl)-1-(2-bromobenzyl)-1H-1,2,3-triazole. In addition, molecular docking calculations suggested that the inhibition mechanism relies on interactions between His 51 , Thr 134 , Ser 135 and Tyr 161 . The virucidal assay with DENV-1-4 strains indicates that compound 35 is a promising lead compound for antiviral activity against Flavivirus. Taken together, our results provide insightful information for further development of Flavivirus protease inhibitors via rational drug design. Efforts towards this end are under way in our laboratories.