Molecular Characterization and Identification of Potential Inhibitors for ‘E’ Protein of Dengue Virus

Dengue is an arthropod-borne acute febrile illness caused by Dengue Virus (DENV), a member of Flaviviridae. Severity of the infection ranges from mild self-limiting illness to severe life-threatening hemorrhagic fever (DHF) and dengue shock syndrome (DSS). To date, there is no specific antiviral therapy established to treat the infection. The current study reports the epidemiology of DENV infections and potential inhibitors of DENV ‘E’ protein. Among the various serotypes, DENV-2 serotype was observed more frequently, followed by DENV-4, DENV-1, and DENV-3. New variants of existing genotypes were observed in DENV-1, 2, and 4 serotypes. Predominantly, the severe form of dengue was attributable to DENV-2 infections, and the incidence was more common in males and pediatric populations. Both the incidence and the disease severity were more common among the residents of non-urban environments. Due to the predominantly self-limiting nature of primary dengue infection and folk medicine practices of non-urban populations, we observed a greater number of secondary dengue cases than primary dengue cases. Hemorrhagic manifestations were more in secondary dengue in particularly in the pediatric group. Through different computational methods, ligands RGBLD1, RGBLD2, RGBLD3, and RGBLD4 are proposed as potential inhibitors in silico against DENV-1, -2, -3, and -4 serotypes.


Introduction
Dengue Virus (DENV) is a member of Flaviviridae, mainly spread by Aedes mosquitoes [1]. Currently, there are four DENV serotypes (DENV1-4), sharing 65% genetic homology. DENV infection manifests in various clinical presentations including acute febrile illness, dengue hemorrhagic fever (DHF), dengue shock syndrome (DSS), etc. [2]. Despite DENV posing a global threat, leading to several deaths every year, there is no specific antiviral drug to treat the infection. Several studies have reported inhibitors against dengue envelope protein; however, nearly all studies built the molecules based on the monomer of the envelope protein.
In reality, the dengue envelope is in dimer form, which is an active form of DENV. Only in this dimer state is the dengue virus is capable of infecting its host cells. Moreover, the inhibitor(s) should be stable and exist as an active form, even at high temperatures, to work on the virus at high temperatures of febrile illness. The information about the inhibitor activity at high temperatures is not available, and inhibitors have been studied at a standard temperature in silico. The current study reports the epidemiology of circulating DENV infections. An active form of the envelope protein of DENV-1-4 was built in silico, potential inhibitors were identified, and their stability at febrile condition was studied.

Materials and Methods
Peripheral venous blood (3 mL) sample was collected by a clinician or a trained phlebotomist within seven days after onset of fever from all eligible patients (as per dengue case definition) [3] included in the study. Patients were attended the hospital from Chittoor, Kadapa, Nellore, Anantapur, East-Godavari, and Guntur districts of Andhra Pradesh state, South India. The serum was aliquoted and stored at −70 • C for later use [4]. Serum was tested for the presence of dengue NS1 antigen using Panbio dengue Early enzyme-linked immunosorbent assay (ELISA) kit (Standard Diagnostics, Inc., Suwon city, Korea). The samples that tested positive for NS1 antigen were tested for dengue IgM (IgM capture ELISA kit, NIV, Pune, India) and IgG (Microlisa, J.Mitra & Co. Pvt. Ltd. New Delhi, India) antibodies following the manufacturer's instructions.

Dengue Serotyping by rRT-PCR
QIAamp Viral RNA mini kit (Qiagen, Hilden, Germany) was used to purify viral RNA from 140 µL of NS1 antigen-positive serum samples, and RNA was eluted in 60 µL of Buffer AVE. The extracted RNA was reverse transcribed and real-time polymerase chain reaction (rRT-PCR) was performed using Superscript-III one-step RT-PCR kit (Thermo Fisher Scientific Inc., Waltham, MA, USA) using CDC DENV-1-4 Real-Time RT-PCR Kit on Agilent AriaMx Real-Time PCR System (Santa Clara, CA, USA) [5]. The thermal cycling conditions were Stage 1; 30 min at 50 • C, Stage 2; 2 min at 95 • C, and Stage 3; 15 s at 95 • C followed by1 min at 60.0 • C for 45 cycles.
The acquired sequences were checked for sequence similarity using nucleotide BLAST. The serotypes were interpreted based on the BLAST results. The sequenced amplicons were trimmed using MEGA7 software [6] and submitted to NCBI GenBank. Accession numbers were obtained after submission. The sequences were checked for dengue genotypes and phylogenetic analysis was performed using DENGUE VIRUS TYPING TOOL [7].

Homology Modelling
The nucleotide sequences were translated into amino acids and protein sequences were made using Molecular Evolutionary Genetics Analysis 7 (MEGA7) [6]. A consensus protein sequence was drawn using MEGA7 for each serotype (DENV-1, -2, -3, -4) from the processed sequences after genotyping. The envelope protein was constructed using a template that was searched through National Center for Biotechnology Information (NCBI) (NCBI Resource Coordinators) Basic Local Alignment Search Tool protein (BLASTp) [8] against protein data bank (PDB) [9] based on query coverage and identity. Target template pairwise sequence alignment was performed using CLUSTALX v2.1 [10] and 3D structures for envelope protein of dengue serotypes (homodimer consensus) were built using Modeller v9.21 [11]. Among the generated models for each dengue envelope (DENV) serotype, the best model was selected based on DOPE score. The best models of DENV serotypes were validated through PROCHECK [12], protein structural analysis (ProSA) [13] and protein quality (ProQ) [14] analysis to define the stereo-chemical quality and overall quality of the protein model, respectively.

Protein Processing
The best three-dimensional model structures of DENV-1, -2, -3, -4 active forms (homodimers) were imported to Maestro v11.1 [15] preprocessed by Prime [16] optimized by Epik [17] and energy minimized by using optimized potentials for liquid simulations (OPLS_3) [18] force field with default parameters of protein preparation wizard options of Schrodinger Software Suite. A grid was generated on the interfacial residues on each DENV serotype.

Ligand Processing and Docking of Published Inhibitors
Published inhibitors of DENV were retrieved from literature and were drawn using Marvinsketch 19.3 and/or retrieved from PubChem [19][20][21][22][23]. LigPrep application with inbuilt Epik and OPLS_3 forcefield module of Schrodinger was applied for ligand preparation. Grid-based ligand docking with energetics (GLIDE) [24] extra precision (XP) docking procedure has opted for docking prepared ligands over grids to analyze the binding affinity between the protein and ligand.

Virtual Screening and Docking of Analogs
Each best docked published inhibitor of DENV serotypes based on XPG score was screened for analogues against an in-house library containing more than 28 million compounds. Analogues obtained were docked on the same grids and a similar docking protocol was implemented and the best-scored molecules better than published inhibitors were proposed as leads [25].

Molecular Dynamics Simulations
The interactions and stability of the leads over published inhibitors with each DENV protein subtypes were further assessed by running 100 ns molecular dynamics simulations in Dipalmitoyl phosphatidylcholine (DPPC) membrane for transmembrane residues and simple point charge (SPC) water solvent model system at the febrile condition of 313.15 K were carried by using Desmond's default protocol [26]. Atomic coordinate data were recorded for every 100 ps as a trajectory and system energies were logged for every 2 fs as a trajectory thereby several molecular dynamics parameters were assessed between published inhibitors and leads with DENV serotypes.

Results
A total of 3926 patients clinically suspected to have dengue fever (as per the dengue case definition) were interviewed to obtain relevant clinical data and blood samples.

Dengue Serotyping by Real-Time Reverse Transcriptase Polymerase Chain Reaction (rRT-PCR)
All 354 dengue NS1-positive samples were subjected to serotyping by Centre for Cisease Control (CDC, Atlanta, GA, USA) rRT-PCR kit. Multiple serotypes were detected in 18.4% of cases and a total of 449 dengue serotypes were obtained from 354 dengue NS1-positive samples. DENV-2 was the major serotype (191 (42.5%)) followed by DENV-1 (102 (22.7%)), DENV-4 (96 (21.4%)), and DENV-3 (60 (13.4%)). Nearly one-fifth (18.4%) of cases had coinfections with more than one serotype. DENV-1 and -2 serotypes were predominantly observed in co-infections. Notably, 10 pediatric cases had co-infections with all four serotypes and all were primary dengue cases; no such findings were observed in adult cases. Multiple infections were more common among non-urban residents. In secondary dengue, no co-infections other than the co-infection with 1 or 2 were detected ( Table 2). All four serotypes were co-existing in the community throughout the year. However, seasonal peaks were observed during rainy months (July and October) ( Figure 1).

Clinical Manifestations
As febrile illness was the main inclusion criteria to recruit cases, it was found in all cases. Chills and myalgia were most common manifestations after a febrile illness. Hemorrhagic manifestations were observed in 142 (41.5%) of 354 NS1-positive patients. There was no significant difference in hemorrhagic manifestations between adults (20.6%) and pediatric populations (20.9%). In primary dengue hemorrhagic manifestation was high in the adult population with 23.5% cases, whereas in the pediatric population, it was 14.8%. In secondary dengue hemorrhagic manifestations were high in the pediatric population with 26.0% cases, whereas in adults, it was 18.2% (Table 3a,

Clinical Manifestations
As febrile illness was the main inclusion criteria to recruit cases, it was found in all cases. Chills and myalgia were most common manifestations after a febrile illness. Hemorrhagic manifestations were observed in 142 (41.5%) of 354 NS1-positive patients. There was no significant difference in hemorrhagic manifestations between adults (20.6%) and pediatric populations (20.9%). In primary dengue hemorrhagic manifestation was high in the adult population with 23.5% cases, whereas in the pediatric population, it was 14.8%. In secondary dengue hemorrhagic manifestations were high in the pediatric population with 26.0% cases, whereas in adults, it was 18.2% (Table 3a, .017 * all cases (10) 100

Dengue Envelope Gene Characterization
Two sets of primers were used to amplify the dengue envelop gene (~1657 bp; coordinates; 853-2509 bp). Target was applied from all samples successfully and around 920 bp (907-939 bp) bands were observed after 2% agarose gel electrophoresis. The purified product was used for bidirectional sequencing by a sanger sequencer. All samples were successfully sequenced, and the resulting DNA sequence (forward and reverse) and chromatogram were analyzed for any errors/mismatches. The sequences were submitted in the GenBank database for public access (Supplementary File S6).

Homology Modelling
Hundred models for each of the four consensus DENV serotype dimeric forms were built based on the target-template alignment and atomic coordinates. The best models of DENV serotypes with the least Discrete Optimized Protein Energy (DOPE) score were selected and validated. PROCHECK analysis revealed that~96.5% of the best model DENV serotypes were in allowed regions of Ramachandran plot, ProSA 'Z' score analysis stated that the best models generated were of good quality with~−5.6, and the predicted LG score of ProQ analysis inferred that the best models were of extremely good models with −11.153 (Table 5 and Figures 2-5).

Protein Preparation
The three-dimensional structures of the DENV-1, -2, -3, -4 best models were pre-processed by adding the parameters such as added hydrogen atoms, bond order and formal

Protein Preparation
The three-dimensional structures of the DENV-1, -2, -3, -4 best models were preprocessed by adding the parameters such as added hydrogen atoms, bond order and formal charge corrections, removed atomic clashes, tautomeric alterations, and ionization states of the protein.

Molecular Dynamics Simulations
The MD simulations studies revealed the conformational stability of RGBLD1-DENV-1; RGBLD2-DENV-2; RGBLD3-DENV-3; RGBLD4-DENV-4 was much more consistent than Agnuside-DENV-1; Rhodioloin-DENV-2; Chlorogenic acid-DENV3; and NITD448-DENV-4 envelope proteins. Molecular dynamic parameters such as root mean square deviations, root mean square fluctuations, energies (total and potential energies of the system), and protein-ligand contacts are much more favorable to the proposed leads than the existing ligands with DENV serotypes (Figures 2-5 and Tables 6-9). Table 6. Comparison of scores obtained in the molecular dynamic's simulation parameters for DENV-1 reference molecule and novel analog molecule.

Discussion
In India, the first dengue fever (DF) case was reported from Madras, Tamil Nadu, in 1946 [28]; later in 1963, an outbreak of DHF was reported from Calcutta, West Bengal [29]. About 33% of global dengue infections were represented by India [30]. Despite frequent outbreaks, there is a paucity in the information concerning circulating genotypes/serotypes from Andhra Pradesh. This is the first study from the state of Andhra Pradesh after the 1965 Visakhapatnam dengue outbreak [31,32].
A total of 1316 samples were included in the study. Of these, 354 samples tested reactive to dengue NS1 antigen. The proportion of male pediatric cases from rural areas was observed to be high with a median age of 15 years (6.7-29.2 years). Of these, 354 dengue NS1-positive cases, 162 (45.8%) were primary infections, and 192 (54.24%) were secondary infections. Among primary infections, the pediatric group was affected most commonly, with a median age of 16 years (7-31 years). The primary cases were reported more in females and rural areas. In secondary infections, pediatric age group and rural areas were also most commonly affected with a median age of 15 years (6-27 years). Males were infected more in secondary infections. In primary cases, DENV-2 was more common, followed by DENV-3, DENV-1/2, DENV-1, and DENV-1/2/3/4. Among the secondary cases, DENV-2 was observed to be more common followed by DENV-4, DENV-1, DENV-3, and DENV-1/2 ( Table 2).
In the current study, the pediatric age group was most commonly affected which is in discordance with the other reports from India [33][34][35]. Though the difference was not significant, minimal male preponderance was noted in our study, which is in line with other studies from India [33,[36][37][38]. The most common symptom in all cases was fever; myalgia, arthralgia, diarrhea, and vomiting were the other clinical symptoms. Hemorrhagic fever was observed in all serotypes except in coinfections of DENV-1/2/3, DENV-1/3/4, DENV-2/3 with the highest in DENV-2 followed by DENV-4. Among all infections, the DENV-2 infections were more followed by DENV-1, DENV-4, and DENV-3 in decreasing order. In single infections, DENV-2 was the commonest followed by DENV-4, DENV-1, and DENV-3 in declining order. The pediatric age group, males, and rural populations were predominantly contributed to the dengue positive rate in the present study.
The prevalence of dengue subtypes changes from time to time; during 2003-2009, DENV-3 was the major cause of dengue followed by DENV-2 in 2010-2011 and DENV-1 in 2012, which were reported from the northern states of India [35,37]. A similar trend was observed from other parts of India with no or very low prevalence of DENV-4 [29,30,33,38,[40][41][42][43]. In contrary to the above reports, DENV-4 was reported as a predominant serotype from the state of Telangana during the 2007 dengue outbreak [44]. Very few studies were conducted on concurrent infections [34,38,45], and the data are scarce concerning serotypes in Andhra Pradesh despite regular outbreaks [30]. In our study, all four serotypes were reported with DENV-2 (41%) preponderance and all possibilities of coinfections were found (Table 2).
A large multicentric study (2018-2019) [46] from India reported a high proportion of secondary dengue cases (65.0%) from southern states, while very low prevalence (<10%) from the northern states of India. The present study also reports relatively high proportion of secondary dengue cases (54.2%) but less prevalence as compared to the above study. This could be due to the inclusion of only NS1 positive samples for analysis in the present study. Five decades back, different dengue serotypes were limited to particular geographies. In recent times, due to the increase in transportation, the migration and evaluation of dengue virus to adapt to new mosquito species outbreaks and co-infections with multiple dengue serotypes have become endemic and have been reported from almost all parts of India [47][48][49]. Compared to North India (more winter months), South India climate is favorable to mosquito breeding and development.
The Indian climate conditions varies widely from very cold Northern hilly states (−14 • C) to very hot Western Indian states (maximum (+)45 • C), predominantly dependent on the Indian monsoon system. The extrinsic incubation period (EIP) of dengue virus largely depends the climatic conditions and reported as short as 5.6 days at +35 • C and as long as 96.5 days at 0 • C. Moderate-to-hot temperate zones (south, central, and western Indian states) favor the mosquito breeding rate and shortening the virus incubation time, thereby increasing the dengue risk and rate of transmission [48]. In the present study and studies from other part of India reported seasonal peaks during the July to October months. During this season, the average climatic temperatures were maintained around 30 to 35 • C with intermittent rains. Furthermore, open drainage system, stagnant of water during rainy seasons, and storage of water in wide mouth earthen pots or reservoirs without lid (act as a breeding ground) could have been contributed to the high prevalence of dengue in this part of the country. In line with the findings of this study, a recent study from India has compiled all dengue outbreaks in India since the last 50 years, and the authors reported that most dengue outbreaks occurred predominantly during the monsoon (June to September) or post-monsoon (October to December) period [41]. These findings support the relation between dengue seasonality and roles of both rainfall and ambient temperature.
In line with the above findings, recent reports from southern states observed high disease burden and has led to an increase in secondary dengue cases. Due to the open drainage system, poor mosquito prevention, and hygiene practices and awareness, the present study observed high prevalence of dengue cases in rural areas. This is an important information required for implementing effective prevention and clinical management protocols in the hotspot regions [50].
The important amino acids which showed stable interactions in the RGBLD1/DENV-1envelope protein were Phe90 and Asp235 in the A chain and Ala88, Phe90, Gln234, Asp235 in the B chain. Stable interactions were seen between RGBLD2/DENV-2, Asn10, Ile24, Glu31, Leu32, Val432, and Gly433 in A chain. In RGBLD3/DENV-3, there were stable interactions with Arg231 in A chain and with Glu83, and His92 in the B chain. In RGBLD4/DENV-4, His230, Glu84, and Tyr90, there were stable interactions throughout the simulations. These analogues showed interactions with both the chains of the dengue envelope protein dimer. The pharmacological properties of the best leads were correlated favorably with more than 95% of approved drug molecules.
The study focused on 93 published inhibitors to dock with the respective DENV serotype envelope protein obtained from the study samples; as a result, the four best molecules were shortlisted for further use. Among 93 published inhibitors, Agnuside was found to be the best inhibitor for DENV-1, Rhodiolin for DENV-2, Chlorogenic acid for DENV-3, and NITD448 for DENV-4. Agnuside and Rhodiolin were reported elsewhere as the best molecules for dengue virus protease (NS2B-NS3pro), helicase (NS3 helicase), methyltransferase (MTase), and RdRp of DENV serotypes [51].
Similarly, Ribavirin was found to be effective against RdRp of DENV and other flaviviruses [52]. Chlorogenic acid is another broad-spectrum antiviral molecule found to be active against influenza A (H1N9) virus RdRp and neuraminidase (NA), and chikungunya virus glycoprotein (E3-E2-E1) and protease (nsP2) [53]. In our study, during in silico analysis, these molecules showed the best binding scores when docked with our protein construct. Using these structures as a reference molecule, we searched for the novel and potent analogs against 285 million compounds. Based on the scores obtained, RGBLD1 (Agnuside analog), RGBLD2 and RGBLD4 (Ribavirin analogs), and RGBLD3 (Chlorogenic acid analog) were found as the best analogs, and all new molecules scored better than its analogs. Hence, the new analogs (RGBLD-1 to 4) would be a potential and broad-spectrum antiviral candidate against dengue serotypes and other similar viruses, as shown for its reference molecules. However, further in vitro and in vivo model studies are required to validate these new molecules.
Altogether, the best leads showed satisfactory interactions with a greater number of amino acids with two chains of the envelope protein, interactions fraction, and pharmacological properties. Strong interactions between the dengue envelope proteins and leads constrict the movements of homodimers of the envelope, thus stopping the viral entry into the host. Therefore, it can be proposed that the best leads can be potential inhibitors of the dengue virus.