Potential Role of Flavivirus NS2B-NS3 Proteases in Viral Pathogenesis and Anti-flavivirus Drug Discovery Employing Animal Cells and Models: A Review

Flaviviruses are known to cause a variety of diseases in humans in different parts of the world. There are very limited numbers of antivirals to combat flavivirus infection, and therefore new drug targets must be explored. The flavivirus NS2B-NS3 proteases are responsible for the cleavage of the flavivirus polyprotein, which is necessary for productive viral infection and for causing clinical infections; therefore, they are a promising drug target for devising novel drugs against different flaviviruses. This review highlights the structural details of the NS2B-NS3 proteases of different flaviviruses, and also describes potential antiviral drugs that can interfere with the viral protease activity, as determined by various studies. Moreover, optimized in vitro reaction conditions for studying the NS2B-NS3 proteases of different flaviviruses may vary and have been incorporated in this review. The increasing availability of the in silico and crystallographic/structural details of flavivirus NS2B-NS3 proteases in free and drug-bound states can pave the path for the development of promising antiflavivirus drugs to be used in clinics. However, there is a paucity of information available on using animal cells and models for studying flavivirus NS2B-NS3 proteases, as well as on the testing of the antiviral drug efficacy against NS2B-NS3 proteases. Therefore, on the basis of recent studies, an effort has also been made to propose potential cellular and animal models for the study of flavivirus NS2B-NS3 proteases for the purposes of exploring flavivirus pathogenesis and for testing the efficacy of possible drugs targets, in vitro and in vivo.


Structure and Role of NS3 and NS3 Protease Domain in Flavivirus Replication
The genetic similarity between the members of the flavivirus genus predicts the main features necessary for the viral replication cycle [50]. The NS3 is one of the major viral proteins possessing enzymatic function. It is found to be the most conserved among the viral proteins, and it exhibits approximately 65% sequence identity among JEV, WNV, DENV, YFV, and ZIKV ( Figure 2) [51]. As described previously, there are two major domains of this protein, which are the RNA helicase and protease domains, connected through a short linker (flexible). The three-dimensional structure of NS3 proteins has been well documented and resolved for various flaviviruses. However, depending on the virus replication stage, different conformations can exist [51,52]. For instance, the binding of RNA is one event that can induce a conformational change [53,54]. The N-terminal domain of the flavivirus NS3 protein consists of protease domains that contain four homologous sequences to serine protease. Three of them form catalytic domains, whereas the fourth helps in substrate binding [55]. It has been suggested that the specificity of the substrate binding is because of an aspartic acid residue, which is located in the lower portion of the binding pocket [56]. The exact site for the proteolytic cleavage depends on the cleavage site sequence, and it may vary among different members; however, the majority of these sites contain two basic residues, which are followed by a side chain within the viral polyprotein [57].

Structure and Role of NS2B and NS2B Hydrophilic Domain in Flavivirus Replication
NS2B consists of approx. 130 amino acids, and is a type of small integral membrane protein, having a molecular weight of 14 kD. It consists of three hydrophobic domains (which are supposed to be part of the transmembrane domain) and a central hydrophilic domain [58,59]. Studies have suggested that the central hydrophilic domain is required for the activation of NS3, and that any mutations in it can cause the defective protease activity of NS3, or may even cause NS3 instability, leading to faulty viral assembly [58,[60][61][62][63][64][65]. NS2B (H) (hydrophilic domain of NS2B) essentially acts as a cofactor for the protease activity of the NS3 protein. The initial characterization of the cofactor requirement for various flaviviruses has revealed that the minimal essential region for protease activity is positioned in a 40-50 residue central hydrophilic segment of NS2B (amino acid 45 to 95) [32,58,66,67]. NS2B contains a hydrophilic region, the central region of which contains a β-barrel, which folds around the β-barrel of the NS3 protease for its stability [55]. Upon substrate binding, conformational changes occur in the NS2B (C-terminal domain), which leads to the stability of βhairpin, which becomes the component of the active site [55,68]. The active NS2B (H)-NS3 protease is essential for the cleavage at the NS2A/NS2B, NS2B/NS3, NS3/NS4A, and NS4B/NS5 junctions [69,70]. Moreover, it has also been proposed that the cleavage of capsid protein may also be mediated through the NS2B-NS3 protease [71]. The NS2B-NS3 interaction may also cause the tethering of NS3 at the membrane, causing replicase complex anchoring at the compartment membranes [72]. Both NS2B(H) and NS3 are associated with the membrane structures (virus-induced) [73]. This suggests that the interaction of NS3 with NS2B (H) is mandatory for its membrane localization [63]. Residues located in four distinct substrate-binding pockets, i.e., S1, S2, S3, and S4, marked in orange, yellow, cyan, and green, respectively [55,74]. Nonconserved residues located at the binding pockets are marked in magenta arrowheads [75].

Dengue Virus (DENV)
A deletion analysis of NS2B in DENV has demonstrated the sufficient role of the central hydrophilic region as a cofactor of NS3 [66,76,77]. The dengue virus (DENV) possesses a polyprotein that is needed to be processed, and that has been found to undergo cleavage at the rER of the host by NS2B-NS3 (cytoplasmic side) and by host cell peptidase (luminal side) [24]. NS2B (a.a. 1394 to 1440) is required as a cofactor for NS3 protease (a.a. 1476 to 1660) [77] and is also involved in the recognition of the substrate [78]. In the dengue virus, NS2B often acts as a cofactor of NS3 (protease domain), and it consists of 130 amino acids (15 kDa) [79]. The N and C terminal domains are located in the cytoplasm. It is proposed to have a helical bundle that consists of approximately four alpha-helix subunits (1)(2)(3)(4), which are short and transmembrane. Between the α2 and α3 subunits, it contains a central hydrophilic domain (consisting of 40 residues and that is highly conserved), which is responsible for its cofactor activity [80]. This domain leads to heterodimerization with the NS3 protease domain (noncovalently), and it results in the formation of a functional membrane-bound protease complex. This complex is needed for the appropriate localization and activation of the serine protease. There are suggestions that it is also needed for the trimerization of NS2B-NS3, although the exact mechanism remains to be elucidated [81]. In the open conformation of NS2B-NS3, the catalytic site is not wrapped by the cofactor, while in the closed conformation, it is needed for the appropriate recognition of the substrate, as well as for efficient proteolysis. Moreover, the latter is also the most predominant form of NS2B-NS3 in the solutions, whether it is ligand-bound or not [81,82]. The protease activity of the NS2B-NS3 leads to viral protein cleavage at NS2A/NS2B, NS2B/NS3 (through cis-cleavage), NS3/NS4A, and NS4B/NS5 (through trans-cleavage). This protease complex is also needed for the internal cleavage within the NS2A, NS4A, and NS3 helicases. Cleavage occurs at the dibasic motifs (RR, KR, RK) at P1 and P2, and at a short chain amino acid at P1 (A, G, or S). The protease complex also cleaves the C protein at the C terminus (at dibasic motifs, which are conserved) [63,82]. Interestingly, while studying the noncofactor roles of NS2B, it has been shown that there is colocalization of the NS2B with dsRNA, which indicates that it might be a part of the replication complex [83]. Moreover, it has also been implicated in viral replication, its assembly, and release, and thus may contribute towards the cytopathic effects (in combination with NS2A) [84]. In DENV, the oligomerization of the NS2B with the host cell membrane may be mediated by its alpha-helical TMD (transmembrane domain). In human red blood cells (RBCs), the DENV NS2B has been shown to destabilize and increase the membrane permeability that leads to pore formation [85]. NS2B mutations at the Trp 62 residue resulted in the complete elimination of the cis-cleavage ability of the NS2B-NS3 protease, while the substitution of alanine at Leu 75, Ile 77, and Ile 79 resulted in reduced proteolytic activity [86]. Recently, it has also been shown that NS2B (alone, or with NS3) interferes with type 1 interferon (IFN) production. This is conducted by specifically targeting the cyclic GMP-AMP synthetase (cGAS) for degradation. cGAS is required for binding with DNA (self or nonself) in the cytoplasm, and it activates a series of biochemical changes through signal transduction that ultimately results in STING activation, which is required for type 1 IFN generation. DENV NS2B causes the degradation of cGAS through the autophagy/lysosomal mediated pathway [87].

Yellow Fever Virus (YFV)
The yellow fever virus genome contains 10862 nucleotides, which encode a long precursor polyprotein. At the membranes of the ER, the generation of viral proteins occurs by the cleavage of the viral polyproteins. The cleavage of the viral structural proteins and NS4B (N-terminus) is mediated by signal peptidase, while the cleavage of the NS1-NS2A is mediated by the host protease (membrane-bound) in the host cell [88]. The cleavage of the remaining capsid protein (membrane-anchored), as well as the cotranslational cleavages, are mediated by the NS3 protease along with the NS2B cofactor [71,79,[89][90][91]. The various cleavage sites include consensus (C/virion C, 2A/2B, 2B/3, 3/4A, 4A/2K, and 4B/5) and alternative sites (aAα) [92]. The N-terminal of the NS3 protein possesses a trypsin-like serine protease domain that preferentially cleaves the two adjacent basic amino acids, e.g., RR or KR, or, in some cases, QR, QK in the consensus sequence of G/ARR2S/G [58,61,92].
The conserved central region of NS2B, and the amino-terminal region of the NS3B, together form the NS2B-NS3 protease complex. Just as in DENV, the NS2B-NS3(pro) constitutes a stable complex that mediates the polyprotein substrate cleavage, both in the cis and the transform [58]. However, charged amino acids are important for this protein cleavage, as it has been determined that the mutations involving charged-alanine replacement at NS2B-NS3181 have demonstrated that they affect polyprotein processing [93].

Zika Virus (ZIKV)
Structural studies have shown that the NS2B-NS3 protease of ZIKV exists in two forms: a closed form and an open form. In the presence of a substrate or inhibitor, it usually adopts a closed conformation, while in the absence of the substrate or inhibitor, it is in open conformation [94,95]. It has been shown that NS2B surrounds the NS3 in such a way that it leads to the formation of β-hairpin, which then makes an important contribution to the formation of the S2 pocket of NS3 [94][95][96]. The NS2B of the Zika virus exhibits a higher level of disorderliness, especially from the 62-98 residue region (37 residues) [97,98]. Ultimately, NS2B interacts with NS3 in such a way that it leads to the cleavage of the polyprotein into a variety of functional proteins, which are important in viral replication and maturation [55].

Japanese Encephalitis Virus (JEV)
Japanese encephalitis (JE) is a vaccine-preventable disease caused by the Japanese encephalitis virus (JEV), which is primarily prevalent in Asia. The JEV is classified into a single serotype, with five genetically distinct genotypes, i.e., I, II, III, IV, and V, having an 11 Kb genome, comprising three structural and seven nonstructural proteins [99][100][101][102]. In JEV, the N-terminal 1/3rd (180 residues) of the NS3 contains protease active sites, which include His 51, Asp 75, and Ser 135 [103]. Just as in other flaviviruses, NS2B acts as the cofactor of the NS3 serine protease [58,66]. NS2B-NS3 proteases have been involved in carrying out a variety of important phases, e.g., RNA replication (viral), polypeptide cleavage, and the processing and assembly of viral particles [104,105]. The protease activity of the JEV NS2B/NS3 leads to the viral polyprotein cleavage of the capsid (internal), NS2A/NS2B, NS2B/NS3, and NS3/NS4A sites [32]. Moreover, NS2B-NS3 proteases may also play an important role in the immune evasion by the virus [105]. It has been shown that NS2B-NS3 proteases have been involved in the cleavage of interferon stimulators. In mice, this ability was found to play a critical role in enhanced viral replication, as well as in enhanced virulence [106]. Researchers have also demonstrated that certain mutations in the NS2B-NS3 region (NS2B-99, NS3-78, and NS3-177) contribute to the enhanced infectivity of JEV (genotype I) in amplifying hosts [107]. In JEV, the residues, Ser 46 to Ile 60 (in particular Trp 53, Glu 55, and Arg 56), are essential for the NS3 protease activity (both cis-and transactivity), just as in DENV4 and YFV. The NS2B of JEV is found to exhibit 67% similarity with the WNV NS2B sequence, while it is found to exhibit 28-34% with other mosquito-borne flaviviruses [108].

West Nile Virus (WNV)
In the West Nile virus, just as in other flaviviruses, NS2B (25 kDa) consists of a transmembrane protein (hydrophobic) that is involved in the replication of the genome, the formation of the membranous structure, and the assembly of virions [109,110]. In order to obtain the association of the protease complex into virus-induced membranes, the domains at both the N and C terminals (residues at 59-62 and 75-87, respectively) play an important role [111]. Another study has shown that the mutation in NS2B at D(80)DD and G83 results in a reduction in the viral NS2B-NS3 protease activity, as well as replication [112]. The unwinding activity of RNA by NS3 is likely made possible after the association of NS2B with NS3 [73,113]. The exact mechanism by which NS2B acts as a cofactor is not completely understood; however, several studies have revealed that, in the presence of NS2B, there is a substantial rearrangement in the NS3 [55,68]. Crystal structures have shown that the NS2B (residue 49-88) tends to form a belt that surrounds the NS3 protease domain. This interaction then forces the NS3 to adopt active conformation [55,77,109]. The NS3 protein is a highly conserved protein that possesses serine protease activity at the N-terminal domain. As this protein lacks a transmembrane domain, after its cleavage from polyprotein, it either goes in the cytoplasm, or remains retained in the ER, where its enzymatic domains are needed [114,115]. Just as with other flaviviruses, it is only active in the presence of the NS2B cofactor, and, in the case of its absence, the NS3 protease domain remains inactive [66,79,116]. This complex (NS2B(H)-NS3) then cleaves the viral polyprotein into a variety of structural and nonstructural proteins [111,117]. The complex of NS2B-NS3 proteases has been found to localize within the convoluted membranes (CM) or para crystalline (PC) arrays, which suggests the possible involvement of the membranes in the proteolytic cleavage [78]. In WNV, the proteolytic activity of the NS3 (Pro), in association with NS2B (hydrophilic region; residue , has been demonstrated by employing an E. coli expression system [118]. Recently, crystal studies involving DENV and WNV NS2B-NS3 proteases have demonstrated that the residues, 51-57 and 82-85 of the NS2B, are important for the stabilization of the NS3 protease and the substrate recognition activity, respectively [55]. Sequence analysis and mutation studies have revealed that the determinants of the flavivirus NS2B protein (except in JEV), which control NS3 protease activation and activities, are located at the positions: Glu52-Leu53-Lys54-Lys55 of YFV [62,93]; Trp62, Leu75-Ser76-Ile77-Thr78-Ile79, and Glu89-Glu90-Glu91-Glu92 of DENV-2 [86,119]; and Trp60, Gly68, Gln77, Gly81, and Val88 of Alkhurma virus (ALKV) [120]. It has also been reported that the NS2B-NS3 proteases were responsible for the apoptosis in human medulloblastoma cells through the activation of caspase-3 and the mitochondrial mediated pathway [121]. The cleavage sites, which are proteolytically processed by the NS2B-NS3 proteases in the polyproteins of various flaviviruses, are summarized in Figure 3. . Cleavage sites proteolytically processed by NS2B-NS3 proteases in polyproteins of various flaviviruses: Cleavage sites proteolytically processed by Japanese encephalitis virus NS2B-NS3 proteases are shown by red arrows [32]; West Nile virus NS2B-NS3 cleavage sites are shown by blue lightning [67,118,122,123]; Yellow fever virus NS2B-NS3 cleavage sites are shown by yellow stars [69,70,124]; Dengue virus NS2B-NS3 cleavage sites are shown by green arrows [70,76,79,82,92,125,126]; and Zika virus NS2B-NS3 cleavage sites are shown by pink Xs [127].

Interaction of Flavivirus NS2B-NS3 Proteases with Cellular Proteins
The Flaviviral RNA tends to replicate on the membrane of the ER, leading to the formation of a replication complex. Many cellular and viral factors participate and are pivotal for the formation of this complex. Therefore, several NS proteins (including NS2B/NS3 proteases) of the flaviviruses act together to retain the replication assembly at the ER. Owing to the larger genome of the DENV, extensive interactions are needed between DENV and the host cells. It has been reported that the NS3 protein of DENV redirects the fatty acid synthase (FASN) on the ER (the replication site for DENV). It was also seen that DENVinfected cells demonstrated the increased synthesis of the fatty acids during infection [128]. Moreover, it was also found that Rab 18 (GTPase located in the ER and that is responsible for vesicle trafficking) helps in the DENV replication by recruiting FASN to the sites where the virus is replicating, and by facilitating its interaction with NS3 to trigger fatty acid synthesis [129]. Recently, it has been reported that the NS3 of DENV (full-length isolated helical and protease domains of NS3) also interacts with the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) enzyme, and this results in enhanced NS3 ATPase activity and reduced glycolytic activities [130]. The nonspecific functions of GAPDH are mRNA translation and stability [131,132]. Therefore, it may be postulated that the interaction between NS3 and GAPDH may result in the unwinding of double-stranded (ds) RNA, as well as vesicle formation (vesicle-induced), which is ultimately needed for virion assembly [130]. Recently, it has been found that the JEV NS3 protein also interacts with the isoforms of the 14-3-3 protein (14-3-3 ε and 14-3-3 η) to block the translocation of the RIG-1 and MDA-5 from the cytosol to the mitochondria, thereby suppressing the host immune response, leading to enhanced viral replication in the cells. The researchers further postulated that the 14-3-3 protein is well conserved among insects, humans, and mice, and that targeting it may thereby facilitate viral replication in multiple hosts [133,134].
In order to maintain homeostasis, cells perform the process of reticulophagy (in which they degrade ER). FAM134B is one of the important host cell restriction factors located on the ER. A study has shown that the NS3 proteases of WNV, DENV, and ZIKV can cause the cleavage of FAM134B, thereby suppressing the reticulophagy pathway, leading to enhanced viral replication, presumably by utilizing the ER membrane for efficient viral budding [135]. The ZIKV NS2B/NS3 protease is also involved in interactions with many other cellular proteins, which include the cleavage of the cytoskeletal factor, Septin-2 (at residue R306), which results in slow cell division, enhanced apoptosis, multipolar spindles in the mitotic defects, and delayed cytokinesis in the neural progenitor cells (NPCs). These changes are likely to produce microencephalopathy [136]. Another study recently indicated that the DENV NS2B-NS3 protease is involved in the cleavage of the DDX21 protein. DDX21 is an RNA helicase enzyme that is involved in the host cell defense against a myriad of viral infections. In cells infected with DENV, there is a translocation of DDX21 from the nucleus to the cytosol. This causes the activation of IFN-β, and thus inhibits the DENV replication during early viral replication. Thus, DENV NS2B-NS3 proteases cause the subversion of the host cell immune system in order to facilitate enhanced viral replication [137]. While the interactions of NS3 with many cellular target proteins result in enhanced viral replication, many other proteins interact with the virus in a way that results in reduced viral replication. A heat shock protein, (Hsp 40) DNAJB6, interacted with NS3, which resulted in reduced viral propagation [138].
Mitochondria and mitochondrial-associated membranes (MAMs) are also known to play an important role in several processes that are pivotal for viral replication, i.e., ATP generation, lipid synthesis, and the induction of cellular apoptosis [139]. Flaviviruses also interact with mitochondria and MAMs and can regulate (up or down) these processes, causing the disturbance in cellular homeostasis. A recent study has demonstrated that the DENV NS2B3 protease interacts with mitochondria and results in the cleavage of MAMs and microfusion (MFN1 and 2) that ultimately leads to the fragmentation of the mitochondria, which can contribute to disease pathogenesis [140]. Keeping this in view, another study was designed to investigate the NS3 protease location in mitochondria. It was found that the N-terminal of the NS3 protease bears a mitochondrial signal sequence, and this facilitates its localization in the matrix of the mitochondria. Upon viral entry into the mitochondria, it was found that the NS3 pro and NS3 pro helicases both resulted in the cleavage of the GrpEL1 protein; the finding was also observed in the samples of clinically infected patients. GrpEL1 protein functions as a cochaperon of the Hsp-70 protein, which implies that the cleavage of the GrpEL1 protein may lead to the dysfunction of the Hsp-70 protein. The exact consequences of this dysfunction are yet to be elucidated; however, based on the correlation between the cellular level of the GrpEL1 protein and the platelet count, the possible dysfunction of the mitochondria was postulated, which leads to thrombocytopenia [141].

Interactions of Flavivirus NS3 with Host Cell NPC and Nucleus
The majority of macromolecular transport between the nucleus and the cytoplasm is mediated mainly through the nuclear pore complex (NPC). The NPC is a disk-like structure (500 nm × 100 nm) that consists of multiple copies of 30 different proteins, which are termed "nucleoporins" (Nups). The NPC and its associated machinery play a pivotal role in the regulation of many cellular pathways. Different viruses have evolved a variety of strategies in order to manipulate the NPC in such a way that ultimately leads to the favoring of viral replication in cells [142]. Altering the NPC integrity is also one of the major activities carried out by viral-encoded proteases to facilitate the viral entry into the nucleus, thus favoring viral replication. This phenomenon has not only been observed in viruses replicating in the nucleus, but also in viruses that replicate in the cytoplasm. The flaviviral proteins are known to interact with the NPC and the associated proteins to disrupt the nucleocytoplasmic trafficking, and to gain entry into the nucleus [139]. The latter strategy may be adopted so that the NPC changes result in the reduced trafficking of mRNA or other transcription factors, which can result in a suppressed immune response against that viral infection [143]. Recent studies have also demonstrated the ability of flaviviral NS2B-NS3 to affect the integrity and distribution of nucleoporins (Nups). Nup62, Nup98, and Nup153 have been found to be disrupted by DENV, whereas Nup98 and Nup153 were affected by the ZIKV NS2B-NS3 proteases [144]. These studies indicate that the NPC and the associated factors in host cells are manipulated as the targets for Flaviviridae replication.
The NS3 of ZIKV tends to locate itself in the perinuclear regions of the infected cells, and causes alterations in the nuclear lamina structure, which leads to the formation of extrusion sites. This may affect the function of centromeres [145]. It has also been observed that NS3 tends to deposit itself on the concave surface of the nucleus (kidneyshaped altered nuclei) and may also be involved in changing the other components of the nuclear envelope [146]. Other studies have indicated that the NS3 of DENV is located on the nucleus of infected cells at an earlier time (8-12 h) than on cytoplasm (16-24 h), postinfection [147,148].

Characterization of Flavivirus NS2B-NS3 Proteases
In order to design an appropriate flavivirus inhibitor, the very first approach is to design an appropriate substrate and optimize the in vitro reaction/working conditions for viral NS2B/NS3 proteases. Various substrate profiling studies have shown that the WNV protease preferentially cleaves at the K/R motifs. The presence of bulky residues, e.g., Tyr, Trp, or Phe at positions P1 or P2, can be well tolerated by the DENV protease as long as the Gly occupies the other position [108,149]. The sequences of amino acid required for polyprotein processing in DENV, WNV, and YFV are homologues; however, minor differences exist among them. In DENV, the hydrolysis sites exist after a pair of basic residues, e.g., Lys-Arg, Arg-Arg, or Arg-Lys at positions P1 and P2 [69]. In WNV, the majority of cleavage sites possess Lys and Arg sequences at positions P2 and P1, and Gly at P1 [150]. Importantly, the YFV polyprotein processing sites contain a pair of Arg-Arg, followed by Gly, Val, or Ser [89]. The substrate sites/sequences susceptible to cleaving by various flavivirus NS2B-NS3 proteases are summarized in Table 1.  [161,162] The pH, the buffers, and the reaction temperature are crucial to characterizing flaviviral NS2B/NS3 proteases [76,77]. Several researchers have optimized these conditions to efficiently determine the in vitro proteolytic activities of NS2B-NS3 proteases, which are compiled in Table 2.

Proposing Role of STING in Development of In Vitro and In Vivo Models for Studying Flavivirus Pathogenesis and Antiviral Drug Screens
One of the major strategies used to develop a vaccine or antiviral drug against flavivirus is through studying disease by employing animal models (Table 4). However, it is difficult to use such models to study flaviviral pathogenesis and disease control measures. For instance, various studies have reported using humanized mice for studying the clinical infection of DENV, with several limitations associated with their use [204][205][206][207][208][209][210][211][212]. There is also a paucity of information on using them to successfully test DENV and ZIKV vaccines [213]. The lack of animal models against flaviviruses has hampered a deep understanding and the development of novel therapeutics/vaccines against most of the flaviviruses. To be used for vaccine or novel therapeutic testing, animal models must exhibit immune competency and viremia (reproducible) against the virus. Moreover, it is also required that the animal model inoculated with a particular virus must demonstrate the same signs as in natural infection. For example, in the case of DENV, none of the humanized mice exhibited the classical features of hemorrhage and the leakage of plasma [214]. That is why a combination of several different models is needed to test the therapeutic efficacy of a novel antiviral or vaccine candidate against different flaviviruses.
Finding a cellular protein that acts particularly as a substrate for some enzymes greatly increases the mechanistic specificity for that protein. Studies in the past have demonstrated a potential new cellular target, the STING (the stimulator of interferon gene) protein, which may allow researchers to develop some appropriate animal models to design novel therapeutics against flavivirus NS2B-NS3, as it has been found that all flavivirus NS2B-NS3 (except YFV) preferentially cleaves to the STING as a substrate [215].
The STING is a multipass protein that resides on the ER, and it plays a pivotal role in inducing the innate immune signaling upon intracellular infection [216][217][218][219]. Originally, it was proposed that it is activated on the intracellular binding of cytosolic DNA species, such as viral DNA, [217,220]. However, later studies have demonstrated that it may also be activated by viral RNA infection [221]. Because of its important role in innate immunity and interferon (IFN) production, several viruses possess proteins that can degrade the STING [215,[222][223][224][225]. The NS2B-NS3 proteases of flaviviruses (WNV, ZIKV, JEV, and DENV; but not YFV), for instance, effectively cleave the STING in human cells, leading to the lower production of type I IFN by those cells, resulting in enhanced intracellular viral replication [215,222,225,226]. Moreover, DENV is also known to play a critical role in the degradation of STAT2, another player in the host immune response [227][228][229][230]. However, the mice STING is resistant to degradation by flavivirus proteins, which results in strong interferon responses and protects them from flavivirus infection [222,225,227,231]. For this reason, they are unable to be used as an effective model for experimental flavivirus infection. Expression of G protein in Vaccinia virus (DENV-2). [233] DNA vaccine (encoding Pr-M and E) of DENV-2. [234] DENV-I vaccine. [235] Tetravalent vaccine expressed in Adenovirus. [236] Tetravalent DNA vaccine (chimeric). [237] Mutant DENV (live attenuated) vaccine. [238] Inactivated DENV (tetravalent). [239] DNA vaccine. [240] Cynomolgous macaques Live attenuated and recombinant vaccine comparison. [241] Chimeric DENV1/2 vaccine. [242] Recombinant DENV. [243] Recombinant protein (DENV 1-4). [244] Tetravalent DENV vaccine (chimeric). [245] Tetravalent DENV vaccine (live attenuated). [246] DENV-2 virus-like particles. [247] Mice A/J DENV-2 caused thrombocytopenia. [248] AG129 (do not have type I and II Interferon receptors) DENV caused neurological manifestations leading to death. [249] DENV infection caused systemic infection and vascular leakage, leading to death. [250] DENV infection resulted in splenomegaly. [251] IFNAR−/− (Lack of IFN type I receptors; background of C57BL/6 mice) DENV-2 infection resulted in viral growth in small intestine, liver, and bone marrow, resulting in death. [252] Cardif −/− DENV infection in mice resulted in viral growth in lymph nodes, bone marrow, and spleen. [253] STAT 1 −/− DENV infection resulted in viral growth in kidney, liver, and small intestine; however, the mice survived.  In order to study the flavivirus NS2B-NS3 proteases in vitro, it is very important to develop cellular models that have functional STINGs that may be cleaved by flavivirus NS2B-NS3 proteases, as it occurs under clinical circumstances in humans. A variety of cells from the human lineage may be used for this purpose. A recently conducted study on ZIKV demonstrated the ability of ZIKV-associated NS2B-NS3 proteases to cleave the STING in fibroblasts derived from humans, as well as nonhuman primates (NHPs) [215]. The results from this study make it possible to use NHP-derived fibroblasts as a possible cell-based model to study and develop novel antiviral drugs/vaccines against flavivirus NS2B-NS3 proteases in vitro.
Similarly, in most of the NHPs, DENV cannot degrade the STING because of a small variation in the STING sequence of nonhuman primates [226]. This small variation may demonstrate the reason for better DENV replication in humans. Further studies have shown that the STING can effectively be degraded in three species of rodents and apes, each indicating the possibility of using these species as an effective model of flavivirus replication in the nonhuman host [226]. The results of this study are promising and provide new hope for the use of these animal hosts as models for studying pathogenesis, and for designing novel therapeutic products against flavivirus NS2B-NS3 proteases.
Moreover, testing drugs on animals prior to humans is one of the preliminary requirements. The abovementioned studies may also pave the path towards finding a possible use for NHPs as animal models for studying the pathogenesis of different flaviviruses. Therefore, in the future, different NHPs (Old and New World monkeys, great apes) may be tested for different flavivirus replications, thus allowing for the use of in vivo models for flavivirus replication, understanding pathogenesis, and devising novel antiviral treatments.

Conclusions
In this review, the structure, optimized reaction/working conditions, potential antiviral targets, and possible cellular and animal models are proposed to study the NS2B-NS3 proteases of various flaviviruses. One approach to treat flavivirus infection is through developing enzyme inhibitors. This approach involves finding compounds that can interact and disorient the enzymatic active site in such a way that it is no longer capable of carrying out its specific function/reaction. Thus, this approach often serves as the starting point for selecting an antiviral inhibitor, whose binding affinity for the active site often resembles, or even exceeds, the normal substrate. Using this approach, many promising compounds have been discovered, as described previously. However, it is also important to mention here that, despite more than two decades of research, not much success has been observed in developing NS2B-NS3 protease inhibitors, and there are several reasons. Firstly, the flat and hydrophobic nature of the enzyme's active site greatly hinders the strong binding affinity of the inhibitor with its active site. Secondly, from a toxicological point of view, the structure of the active site of NS2B-NS3 proteases greatly resembles the host serine proteases, and, thus, the use of such NS2B-NS3 protease inhibitors may lead to severe damage in the host cells. Therefore, prolonged studies must be conducted at the cellular level and with experimental animals before its consideration for use in humans. Moreover, as the active site exhibits great affinity towards positively charged substrates/inhibitors, the use of such compounds may have some negative effects on the bioavailability of the compounds.
The latest crystallographic studies of NS2B-NS3 proteases with substrate-bound and unbound forms have provided some mechanistic evidence of the enzymatic mode of action, which may help in the future for developing a potential safe inhibitor of NS2B-NS3 proteases. Both in silico and high-throughput screening (HTS) methods may be deployed initially for shortlisting the inhibitors of NS2B-NS3 proteases, which may later be confirmed through crystallographic studies. These studies may help in identifying the more allosteric sites in flavivirus NS2B-NS3 proteases, and may lead to the discovery of more effective, potent, and safe flavivirus NS2B-NS3 protease inhibitors. Moreover, as the structures of NS2B-NS3 proteases exhibit great similarity in different flaviviruses, efforts must be made to find an antiviral agent that can be used effectively to inhibit proteases from different flaviviruses, and that thus exhibit a broad range of antiviral activities.