Inhibition of N‐myristoyltransferase1 affects dengue virus replication

Abstract Dengue virus (DENV) causes dengue fever, a self‐limiting disease that could be fatal due to serious complications. No specific treatment is currently available and the preventative vaccine is only partially protective. To develop a potential drug target for dengue fever, we need to understand its biology and pathogenesis thoroughly. N‐myristoyltransferase (NMT) is an N‐terminal protein lipidation enzyme that catalyzes the covalent cotranslational attachment of fatty acids to the amino‐terminal glycine residue of a number of proteins, leading to the modulation of various signaling molecules. In this study, we investigated the interaction of dengue viral proteins with host NMT and its subsequent effect on DENV. Our bioinformatics, molecular docking, and far‐western blotting analyses demonstrated the interaction of viral envelope protein (E) with NMT. The gene expression of NMT was strongly elevated in a dependent manner during the viral replication phase in dendritic cells. Moreover, NMT gene silencing significantly inhibited DENV replication in dendritic cells. Further studies investigating the target cell types of other host factors are suggested.


| INTRODUC TI ON
Dengue virus (DENV) infection is a neglected tropical disease that is widespread in tropical and subtropical regions. Around 390 million infections are reported annually, with 500,000 severe cases resulting in ~25,000 deaths per year (Horstick, Tozan, & Wilder-Smith, 2015). Dengue fever is usually a self-limiting disease, but some cases can result in severe, life-threatening symptoms, such as dengue hemorrhagic fever, which causes bleeding that can lead to dengue shock syndrome, characterized by severe plasma leakage, prolonged shock, and multiple organ failure. Because the currently available vaccine is not completely effective and there is no specific treatment for dengue fever, treatment is supportive and includes fluid therapy.
The DENV genome encodes 10 viral proteins consisting of three structural proteins, namely capsid (C), premembrane/membrane (prM/M), and envelope (E) proteins; and seven nonstructural proteins-NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5 (Rustin et al., 2007). For viral replication to occur, a host element is needed to produce the new virion. Consequently, it is very important that the viral-host interaction be achieved during the viral life cycle (Noppakunmongkolchai et al., 2016). N-myristoyltransferase (NMT) belongs to the GCN5 N-acetyltransferase superfamily and is the key enzyme responsible for catalyzing the myristoylation process, which is the central switch of several cellular signaling pathways necessary for growth and cellular proliferation. Moreover, the fundamental role of NMT as being essential for cell survival makes it a potential drug target for the treatment of cancers, parasitic infections, and other infectious diseases. NMT has two isoforms, NMT1 and NMT2, encoded by different genes and with different specificities. NMT1 has exhibited numerous dominant functions, such as in vivo inhibition of tumor growth in NMT1 knockdowns and defective myelopoiesis in NMT1 knockdown mouse embryos (Kumar, Singh, Dimmock, & Sharma, 2011;Kumar & Sharma, 2015;Zhao & Ma, 2014). A previous study has shown that the disruption of hNMT1 long form, but not human NMT2, could inhibit HIV-1 replication in vivo (Takamune et al., 2008). Previous NMT1 studies have focused on cancer development, diagnostic biomarkers, and chemotherapeutic targets (Selvakumar et al., 2007;Shrivastav, Varma, Saxena, DeCoteau, & Sharma, 2007;Thinon, Morales-Sanfrutos, Mann, & Tate, 2016). Several studies have investigated the biology and pathogenesis of NMT from malaria and leishmania within its potential as a novel drug target by exploiting the inhibition effect observed when NMT expression is disrupted (Banerjee, Arora, & Murty, 2012;Bell et al., 2012;Brannigan et al., 2010;Tate, Bell, Rackham, & Wright, 2014). Furthermore, recognition of NMT as a cell wall target of Aspergillus fumigatus infection has also identified NMT as a potential therapeutic drug target (Fang et al., 2015;Valiante, Macheleidt, Foge, & Brakhage, 2015), and NMT has been identified as a novel antifungal target of Candida albicans (Sikorski et al., 1997;Wiegand et al., 1992). A study of the role of NMT in relation to human immunodeficiency virus proteins has revealed that the myristoylation of Gag and Nef proteins is a key for viral replication and virulence (Seaton & Smith, 2008). Concerning Flavivirus studies, an in silico analysis predicted a myristoylation site (Maurer-Stroh & Eisenhaber, 2014), while in silico studies focusing on the DENV-E protein also predicted posttranslational modification sites and investigated their role in pathogenesis (Ruhul Amin, Mahbub, Sikder, & Karim, 2010

| MATERIAL S AND ME THODS
We investigated the interaction between host NMT and DENV, particularly concerning virus replication, using the DENV Serotype 2 strain 16681 with human dendritic cells.

| Dendritic cell isolation and generation
The buffy coat was diluted 1:1 with sterile phosphate-buffered saline (PBS), and peripheral blood mononuclear cell (PBMC) isolation was performed using Ficoll ® -Paque PREMIUM density gradient medium (GE Healthcare, Little Chalfont, UK) according to the manufacturer's recommendations. Isolated PBMCs were cultured in T75 flasks for monocyte selection and stimulated to monocyte-derived dendritic cells with recombinant human IL-4 and granulocyte-macrophage colony-stimulating factor (Luplertlop et al., 2006). An African green monkey kidney cell line (Vero) was kindly provided by Dr A.
The titer for virus viability and serotype specificity was confirmed by plaque titration assay and nested RT-PCR, respectively (Baer & Kehn-Hall, 2014;Lanciotti, Calisher, Gubler, Chang, & Vorndam, 1992). Throughout this experiment, virus infection was performed at a multiplicity of infection (MOI) of 1 PFU/cell. The virus was added to the host cells and incubated at 37°C in 5% CO 2 in a humidified incubator for 90 min, to allow host-cell infection. Then, the cells were washed twice with 1× PBS to remove excess virus and serum-free medium was added for further incubation. The cells were harvested at specific experiment time-points according to the viral life cycle processes, consisting of adsorption, viral fusion, protein translation/ genome replication, viral assembly, viral maturation, and viral release (Mukhopadhyay, Kuhn, & Rossmann, 2005). We harvested the cells at 1, 12, and 36 hr postinfection, as adjusted from previously described studies (Barth, 1992;Mosso, Galvan-Mendoza, Ludert, & del Angel, 2008;Shrivastava, Sripada, Kaur, Shah, & Cecilia, 2011;Thepparit, Phoolcharoen, Suksanpaisan, & Smith, 2004).

| In silico analysis
To preliminarily investigate the possibility of NMT interaction with DENV protein, the computational bioinformatics tool ExPASy ScanProsite (https://prosite.expasy.org/scanprosite/) was used to screen for the N-myristoylation prediction site (Prosite PS0008) from complete whole protein sequences of DENV Serotypes 1-4 accord-  (Grand, 1989;Hulo et al., 2006;Towler, Gordon, Adams, & Glaser, 1988). The DENV strains were selected from those available on the NCBI database (http://www.ncbi.nlm. nih.gov/protein/) by random sampling from different countries of origin, gathering 10 complete protein sequences from each serotype (Appendix Table A1). The data were analyzed utilizing a percentage calculation for each N-myristoylation prediction site within the DENV genome, and a heatmap was generated using RStudio version 1.1.423. To confirm our preliminary results, a protein-protein docking analysis was performed to screen the overall interaction and compare the two most possible candidate DENV proteins, using protein x-ray crystallization structures available on Protein Data Bank (PDB) (http://www.rcsb.org/). The candidate proteins were DENV Type 2 envelope protein (PDB ID: 1OAN) and full-length NS5 protein of DENV Type 3 (PDB ID: 5CCV). Analysis was performed using the ZDOCK online server (http://zdock.umassmed.edu) (Pierce et al., 2014;Vakser, 2014) with human myristoyl-CoA:protein NMT (PDB ID: 1RXT) and BIOVIA Discovery Studio Visualizer 2.5.

| Far-western blot analysis
To confirm the best single candidate protein result from the molecular docking investigation, we performed a far-western blot analysis (Rudtanatip, Withyachumnarnkul, & Wongprasert, 2015) with re-

| Pull-down assay
The NMT recombinant protein (Abnova) and lysate protein from DENV-2 strain 16681 infected the induced dendritic cells (iDCs) cell for 48 hr using pull-down lysis buffer. This experiment was performed with a Pierce™ GST Protein Interaction Pull-Down Kit (Thermo Scientific, MA). Briefly, as a bait protein the rNMT was immobilized with agarose beads to allow the protein to bind with the affinity ligand for 30 min at 4°C. Then it was washed five times and the DENV-infected cell lysate was allowed to bind with the bait protein overnight at 4°C. Then, the product was washed five times to remove nonspecific proteins and elute the interaction protein using Glutathione elution buffer; western blot analysis was performed to detect protein using antibody specific for NMT (Santa Cruz Biotechnology) and E glycoprotein (Abcam).

| Gene silencing and viral replication assay in NMT1-knockdown cell
The antisense 2′-deoxy-2′-fluoroarabino nucleic acid oligonucleotides (FANA oligos) of NMT1 and a negative control that was used to perform the gene silencing experiment were designed and synthesized by AUM BioTech and prepared for transfection into the iDCs via gymnotic delivery (Fazil et al., 2016;Souleimanian et al., 2012

| Western blot analysis
We performed western blots to confirm that the negative experiment of the far-western blot analysis was successful, and that iDC silencing had been achieved after FANA antisense oligonucleotide F I G U R E 1 Heatmap shows the dendrogram arrangement of N-myristoylation site prediction catalyzed by N-myristoyltransferase in each dengue viral protein of dengue virus (DENV) Serotypes 1-4 (Appendix Table A1). The envelope protein was highly shared at the predicted N-myristoylation site, followed by NS5, NS3, and NS1, whereas NS4A showed the lowest predicted site of N-myristoylation (a). The normal arrangement of the DENV viral protein is shown in part b treatment. The iDC supernatant was lysed with radioimmunoprecipitation assay buffer to obtain the total protein.  Quantum™ CCD camera.

| Plaque titration assay
The supernatant containing the virus was diluted from a 1:10 dilution to fourfold dilutions which were used to infect a monolayer of LLC-MK2 host cells in 24-well plates. Following 90 min incubation at 37°C with 5% CO 2 , excess virus was discarded and a carboxymethylcellulose (CMC) substrate was added for further incubation for 7 days, depending on the observation of cytopathic effects under an inverted microscope. The CMC was removed and 3.7% paraformaldehyde was added for 20 min to fix the cells. Then, the cells were stained with 0.1% crystal violet in 20% ethanol. The plate was washed twice with PBS and tap water until the plaques were clearly visible and were counted as plaque-forming units (PFUs) (Hamel et al., 2015).

| Immunogold staining by transmission electron microscope
Immunogold staining was used to confirm the silencing of NMT1 related with envelope protein synthesis, to imply the replication ability of DENV in dendritic cells in the NMT silencing condition.
The cell pellet was fixed with 2.5% glutaraldehyde for 1 hr and washed three times with glucose phosphate buffer before processing and gold staining with the desired antibody, NMT1 antibody (Santa Cruz Biotechnology) and anti-Flavivirus E-glycoprotein antibody (Abcam). The sample was processed for viewing under a model HT7700 transmission electron microscope (TEM) (HITACHI Ltd., Japan).

| Statistical analysis
All experiments were performed in triplicate and the data were presented as mean ± SD. The two-tailed Student's t test was used to assess the point of statistical significance between the test groups (p < 0.05), indicated by an asterisk (*).

| E protein is the most frequently predicted Nmyristoylation site: in silico analysis
The dendrogram arrangement of the heatmap from the protein sequences of several strains of DENV Serotypes 1-4, which F I G U R E 2 Protein-protein interaction of (a) hNMT: 1RXT (tetramer ribbon protein complex) and dengue virus (DENV) 2 envelope protein: 1OAN (green) with the binding pocket (white) and (b) hNMT: 1RXT (tetramer ribbon protein complex) and DENV NS5 protein: 5CCV (green) with the binding pocket (white). The envelope protein showed two interaction sites, while NS5 showed only one interaction site were available on the NCBI database, was analyzed with ExPasy ScanProsite to determine the N-myristoylation site, catalyzed by NMT. The heatmap and dendrogram results revealed that most N-myristoylation prediction sites in complete DENV viral protein sequences are in the DENV-E protein, followed by NS5, NS3, NS1, NS4B, NS2A, PrM/M, NS2A, NS4B, and C proteins, respectively ( Figure 1a). The DENV protein arrangement sequence is indicated in Figure 1b. The two most likely DENV viral protein candidates, DENV-E and NS5, were to be confirmed using protein-protein interaction docking methods in further experiments.

| The E and NS5 DENV proteins bind with hNMT1
We performed a protein-protein interaction analysis on the two predicated DENV candidate proteins, DENV-E and NS5, determined from in silico analysis. The results indicated that both proteins interacted with hNMT1. The tetramer structure of hNMT1 (PDB: 1RXT) protein had two binding pockets on DENV-E (PDB: 1OAN) protein (Figure 2a), while DENV NS5 (PDB: 5CCV) protein had one binding pocket for the tetramer of hNMT1 protein (Figure 2b).

| Decreased DENV viral titer in hNMT1 silencing iDC
The Following 48 hr incubation at 37°C in a humidified incubator with 5% CO 2 , the supernatant was collected to investigate virus viability and production by plaque titration assay. Immunogold staining showed a high density of NMT-1 in the absence of NMT-1 silencing (Figure 5b), related to the high amount of DENV envelope protein presented (Figure 5d). The NMT-1 silencing condition showed low amounts of NMT-1 than without the NMT-1 silencing condition (Figure 5c), which also showed a lower amount of DENV envelope protein (Figure 5e). The protein was extracted from the cells to perform western blot analysis. The results revealed the DENV-E protein downregulation, not complete inhibition, when compared to negative silencing and the GAPDH loading control Figure 5f). Hence, we determined that the titer of viable virus from the plaque titration assay was significantly downregulated in the iDC silencing treatment when compared to negative silencing (Figure 5g).

| D ISCUSS I ON
Numerous studies have identified significant viral-host interactions in several kinds of viruses, and DENV-host interaction has been established as necessary for the viral replication cycle, antiviral response, and as a possible approach for antiviral targets (Acosta, Kumar, & Bartenschlager, 2014). In this study, we examined the role and efficiency of hNMT1 during DENV replication. First, we performed in silico analysis to further test the DENV-host interactions related to the replication cycle in dendritic cells, the primary sentinel human target cell and the primary target for DENV replication and mediated immunity (Marovich et al., 2001;Schmid, Diamond, & Harris, 2014). We included DENV Serotype 2 strain 16681 in the in silico analysis (Figure 2), which showed that DENV-2 had a high abundance of predicted N-myristoylation sites and that DENV-2 strains F I G U R E 4 The triplicate gene expression experiment of hNMT1 in dengue virus (DENV) infected induced dendritic cells used qRT-PCR relative and normalized with uninfected cells and beta actin as a housekeeping gene. The significant upregulation of hNMT-1 in the DENV-infected cells showed in 1 h.p.i. with slight downregulation in 12 h.p.i. and marked upregulation in 36 h.p.i. associated with the membrane involved. The difference in multiplicity of infection (MOI) (0.1, 1, 10) with the level of NMT1 expression was MOIdependent were associated with severe dengue cases and most frequently found in worldwide epidemics (Nunes et al., 2016;Wei & Li, 2017).
DENV strain 16681 was isolated from a patient with a severe case of dengue fever and has been widely used to study DENV pathogenesis (Okamoto et al., 2012). The in silico results showed that Nmyristoylation sites were abundant in DENV serotypes, correlating F I G U R E 5 The western blotting of (a) induced dendritic cell ( biosynthesis (Heaton et al., 2010;Salazar, del Angel, Lanz-Mendoza, Ludert, & Pando-Robles, 2014). This is compatible with the theory of the myristoylation process (Resh, 1999).
We used the best two candidate proteins from the in silico analysis, E and NS5 proteins to perform the protein-protein interaction analysis with hNMT, aimed to screening the overall interaction to provide the information platform for the experimental study. Our results found more interaction pockets in the DENV-E protein than  (Moriya et al., 2013), and viral exocytosis ( Figure 6).
To verify the precise role of NMT in DENV infection, we knockeddown hNMT1 in iDCs to determine DENV replication. Our results indicated DENV-E protein downregulation when compared with the control, as detected by immunogold TEM and western blotting, the viral viability titer and immunogold staining were significantly decreased, without complete inhibition, in the hNMT1 knockdown iDCs detected by plaque titration assay. It is important to note that hNMT1 might only be one of many host factors that facilitate the replication of DENV and is also involved in membrane-associated viral entry into host cells.
In conclusion, the results of this study suggest that NMT1 may act as a host factor that facilitates DENV replication through the interaction of a membrane-involved viral life cycle, especially adsorption, viral assembly, and exocytosis, as summarized in Figure 7.

CO N FLI C T O F I NTE R E S T S
The authors declare they have no competing interests.