Replication of Marek's Disease Virus Is Dependent on Synthesis of De Novo Fatty Acid and Prostaglandin E2

Disturbances of the lipid metabolism in chickens infected with MDV contribute to the pathogenesis of disease. However, the role of lipid metabolism in MDV replication remained unknown. Here, we demonstrate that MDV infection activates FAS and induces LD formation. Moreover, our results demonstrate that MDV replication is highly dependent on the FAS pathway and the downstream metabolites. Finally, our results reveal that MDV also activates the COX-2/PGE2 pathway, which supports MDV replication by activating PGE2/EP2 and PGE2/EP4 signaling pathways.

acids, cholesterol, cholesterol esters, squalene, phospholipids, and triacylglycerol. Furthermore, excess lipid biosynthesis triggers cellular deposition of lipid droplets in MDV-infected cells (4,5). Despite these intriguing observations, the role of lipid metabolism in MDV-infected cells remained unknown.
In fatty acid synthesis (FAS), acetyl-coenzyme A (CoA) is converted to malonyl-CoA and subsequently to fatty acids. The first step toward FAS is the conversion of citric acid into acetyl-CoA by direct phosphorylation of ATP-citrate lyase (ACLY). The subsequent committed step involves the conversion of acetyl-CoA into malonyl-CoA by acetyl-CoA carboxylase (ACC). The final step involves committed elongation by utilizing both acetyl-CoA and malonyl-CoA coupled to the multifunctional fatty acid synthase (FASN) to generate fatty acids (6). Dengue virus (DENV) (7), West Nile virus (WNV) (8) and hepatitis C virus (HCV) (9) have been shown to preferentially enhance FASN activity and fatty acid synthesis. Fatty acid can contribute to several key biological functions, such as fatty acid oxidation (FAO), ␤-oxidation, posttranslational modification of proteins, and generation of very-long-chain fatty acids. Prostaglandin E 2 (PGE 2 ), a potent lipid modulator, is derived from enzymatic activity of inducible COX-2 on arachidonic acid (AA). A direct association between induction of COX-2 activity and enhancement of human cytomegalovirus (HCMV) replication has been reported (10,11).
Here, we demonstrate that MDV infection activates both FAS and COX-2/PGE 2 pathways, which are crucial for MDV replication. Interestingly, exogenous malonyl-CoA and palmitic acid completely restore the inhibitory effects of FAS inhibitors on MDV replication, suggesting that the synthetized fatty acids are crucial for MDV replication. Intriguingly, addition of PGE 2 partially restored the inhibitory effects of FAS inhibitors on MDV replication, indicating that the two pathways are connected. Taken together, our results demonstrate that FAS and PGE 2 synthesis contribute to MDV replication.
(This article was submitted to an online preprint archive [12].)

Increased levels of lipid metabolites in MDV-infected cells.
Relative production of a panel of metabolites was determined in mock-and MDV-infected primary chicken embryo fibroblasts (CEFs) at 48 and 72 h postinfection (hpi) using two-dimensional gas chromatography with mass spectrometry (GCϫGC-MS). The 16 lipid metabolites were detected and analyzed. Eight lipid metabolites, including palmitic acid, were increased at 72 hpi ( MDV replication depends on fatty acid synthesis. Here, we examined whether MDV infection increases FAS. For a better understanding of FAS, a schematic illustration of the FAS pathway is shown in Fig. 2A. Gene expression of the cellular enzymes involved in the FAS pathway was analyzed in mock-and MDV-infected CEF cells using quantitative PCR (qPCR). The results demonstrate that acetyl-CoA carboxylase (ACC) and fatty acid synthase (FASN) genes were highly upregulated at 72 hpi (Fig. 2B). Analysis of ACC protein levels in mock-and MDV-infected cells was performed using Western blotting. The results demonstrate that the expression of ACC subunit 1 (62 kDa) (Fig. 2C) and FASN proteins (Fig. 2D) is increased in MDV-infected cells. Specific chemical inhibitors for ACC (TOFA) and FASN (C75) were utilized to determine the role of the FAS pathway in MDV replication. Nontoxic concentrations of TOFA, C75, or TOFA and C75 (T/C) were determined based on viability and confluence of the treated CEFs (data not shown). At 72 hpi, viral titers were determined in the presence of the pharmacological inhibitors with addition of the exogenous downstream metabolites (malonyl-CoA and palmitic acid). The inhibitors of ACC (TOFA; 1.54 M) and FASN (C75; 5.9 M) significantly reduced MDV titer by 27 (Fig. 2E)-and 28 (Fig. 2F)-fold, respectively. MDV titer was further reduced when the cells were treated with nontoxic concentrations (data not shown) of TOFA (1.54 M) and C75 (5.9 M) combined (T/C) compared to those of C75 or TOFA alone (Fig. 2G). Therefore, we used T/C in some experiments to observe maximum inhib-itory effects. Addition of malonyl-CoA to the culture in the presence of TOFA restored MDV replication (Fig. 2H). Similarly, palmitic acid restored virus replication in the presence of C75 (Fig. 2I). Treatment of the cells with C75 or TOFA did not reduce plaque sizes (data not shown). To confirm the role of the FAS pathway in MDV replication, we also analyzed MDV genome copy numbers at 72 hpi in the cells treated with TOFA or C75 using qPCR. The results demonstrated that TOFA and C75 also reduce virus copy numbers (Fig. 2J), suggesting that reduction in virus titer by TOFA or C75 is associated with virus copy numbers. Taken together, the results demonstrate that MDV activates the FAS pathway and blocking ACC and FASN decreases MDV titer, which can be rescued by the downstream metabolites.
MDV infection increases the formation of neutral LDs. Lipid droplets (LDs) are endoplasmic reticulum (ER)-derived organelles that consist of a neutral lipid core with a phospholipid bilayer, and an increase in the number of LDs is an indication of lipogenesis. Here, the role of FAS on lipid metabolism was evaluated in MDV-infected cells based on LD formation. The presence of lipid droplets was examined in mock-and MDV-infected CEFs using oil red O staining at 72 hpi (Fig. 3A). An increased accumulation of LDs was observed in the MDV-infected cells (Fig. 3Ai) compared to that of the mock-infected cells (Fig. 3Aii). To quantify the number of LDs, pRB1B UL35-GFP-infected CEFs (488 nm) were stained with Red LipidTOX (568 nm), and LD formation was analyzed using confocal microscopy. Larger numbers of LDs per cell were observed in MDV-infected cells than in the mock-infected cells, as determined using the IMARIS software (P ϭ 0.0001) ( Fig. 3B and C). Treatment of CEFs with TOFA or C75 decreases (P ϭ 0.0001) the LD numbers per cell in MDV-infected cells (Fig. 3D). Altogether, these data showed that MDV infection induced the accumulation of LDs and FAS inhibitors reduced LD formation, indicating that there was a direct link between FAS and LD formation in MDV infection.
MDV activates the COX-2/PGE 2 pathway. LDs are reservoirs of COX-2 and the site of PGE 2 synthesis (13,14) by which arachidonic acid (AA) can be converted into eicosanoids, including PGE 2 , thromboxane A 2 (TXA 2 ), prostaglandin D 2 (PGD 2 ), prostaglandin I 2 (PGI 2 ), and prostaglandin F 2 alpha (PGF 2␣ ) (Fig. 4A). AA is converted from elongation/desaturation of 18:3 -6 fatty acid, an essential fatty acid which can be freed by lipases. In our experiments, we demonstrate that AA levels were elevated in the MDV-infected CEFs (P ϭ 0.003) at 48 and 72 hpi (Fig. 1A). No significant difference in COX-1 mRNA transcript expression levels was observed between the mock-and  (Fig. 4D). Higher levels of PGE 2 were detected in the supernatant of MDV-infected cells than in mock-infected cells at 72 hpi using a specific enzyme-linked immunosorbent assay (ELISA) (Fig. 4E). Treatment of the cells with a COX-2 inhibitor, SC-236, reduced MDV titer in a dose-dependent manner (Fig. 4F). To confirm the role of COX-2 in MDV replication, we also demonstrated that short hairpin   (Fig. 4H). Exogenous PGE 2 , TXA 2 , PGD 2 , PGI 2 , or PGF 2␣ was added to the cells treated with SC-236, and the results demonstrated that PGE 2 was the only prostanoid that rescued the inhibitory effects of the COX-2 inhibitor on MDV titer (Fig. 4I). Exogenous PGE 2 (0.1 to 5 g/ml) rescued the inhibitory effects of SC-236 on MDV titer (Fig. 4J), while it did not alter MDV titer in the absence of the COX-2 inhibitor (Fig. 4K). Strikingly, exogenous PGE 2 also restored the inhibitory effects of TOFA (Fig. 4L) or C75 (Fig. 4M) on virus titer, suggesting that the inhibitory effect of FAS pathway inhibitors was at least partially dependent on inhibition of PGE 2 synthesis. Exogenous PGE 2 recovered an MDV titer in the MDVinfected cells treated with low (0.31 M), intermediate (0.77 M), or high (1.54 M) concentrations of TOFA (Fig. 4L). Similarly, exogenous PGE 2 rescued the inhibitory effects of low (1.97 M) and high (5.9 M) concentrations of C75 on MDV titer (Fig. 4M). To demonstrate that PGE 2 does not block the function of TOFA and C75 on lipid metabolism, the cells were treated with T/C in the presence or absence of PGE 2 (5 g/ml), and the fold reduction in the numbers of lipid droplets per cell were analyzed using confocal microscopy. The results demonstrated that the presence of PGE 2 did not affect the ability of T/C to reduce lipid droplet numbers (Fig. 4N). PGE 2 supports MDV replication through EP2 and EP4 receptors. In mammals, PGE 2 exerts its biological activities through prostaglandin receptors EP1 to EP4 (Fig. 5A). Chicken EP2, EP3, and EP4 receptors have been cloned and characterized (15,16), while EP1 receptor has not yet been identified in chickens. Analysis of gene expression in the mock-and MDV-infected CEFs demonstrated that the EP2 and EP4, but not EP3, receptors were upregulated upon MDV infection (Fig. 5B), suggesting that PGE 2 supported MDV replication through an EP2-and/or EP4-mediated mechanism. Receptor antagonists for EP1 (SC-51322), EP2 (TG 4-155), and EP4 (ER-819762) were utilized to examine the role of different EP receptors in MDV replication in vitro. As expected, the EP1 receptor antagonist (SC-51322; 0.1, 1, 5, and 10 M) had no effect on MDV replication (Fig. 5C). In contrast, TG 4-155 (2 and 4 M) (Fig. 5D) and ER-819762 (0.1, 0.5 and 1 M (Fig. 5E) reduced MDV titer (P ϭ 0.0001), suggesting that PGE 2 supports MDV replication through the EP2 and EP4 receptors. To confirm these data, we also used an shRNA system to downregulate the EP2 and EP4 receptors in CEFs (Fig. 4F). The results demonstrate that RNA silencing of EP2 and EP4 by shRNAs and chemical inhibitors (TG 4-155, 4 M; ER-819762, 1.0 M) significantly reduced total numbers of MDV plaques (Fig. 5G). Taken together, our data indicate that virus-induced PGE 2 synthesis supports MDV replication through the EP2 and EP4 receptors.

DISCUSSION
MDV is an Alphaherpesvirus that infects chickens and causes a deadly lymphoproliferative disease. In addition to transformation of CD4 ϩ T cells, MDV causes atherosclerosis by disturbing the lipid metabolism in infected chickens (17), which can be inhibited by vaccine-induced immunity (4,18). MDV is a cell-associated virus, and cell-free virus can only be produced by feather follicle epithelial cells in vivo. MDV is transmitted to other chickens via inhalation of virus. No cell-free virus can be generated in vitro, and surprisingly, little is known about the mechanism of MDV entry, replication, and egress in vitro or in vivo. After infection of chickens, MDV can be detected in both immune and nonimmune cells; however, the virus can only be detected in less than 2% of CD4 ϩ T cells in the spleen. As there is no specific marker for identification of MDV-transformed T cells (3) and the majority of these cells undergo apoptosis even following in vitro T cell activation, which can potentially modulate lipid metabolism, Recently, an in vitro infection system for primary B and T cells has been developed that we will also include in future studies (19). Our analysis of MDV-infected CEFs demonstrates that MDV infection enhances the expression of genes involved in FAS and increases the levels of AA, the prostaglandin precursor. Our data suggest that MDV hijacks host metabolic pathways to provide essential macromolecular synthesis to support infection and replication.  results demonstrated that blocking ACC and FASN activity significantly reduces MDV replication, suggesting that MDV preferentially modulates the FAS pathway to generate a variety of lipids that contribute to several key cellular processes. Our data confirm that the inhibitory effects of FAS inhibitors on MDV replication could be overcome by the addition of palmitic acid, a metabolite downstream of FASN in the FAS pathway. This indicates that inhibition of MDV by FAS inhibitors is not simply detrimental to the cell but is essential for the production of infectious virus. There is very limited information on the mechanism involved in alteration of cellular metabolism by viruses. There are only two examples where the viral gene products necessary for changes in each specific metabolic pathway have been identified (reviewed in reference 24). Similarly, there is no information on the mechanism involved in activation of FAS by MDV. Interestingly, MDV encodes a secreted glycoprotein, vLIP, which has homology with lipase but does not have any enzymatic activity (25). Further research is required to determine the role of MDV genes, including vLIP, in activation of lipid metabolism.
It is known that fatty acids are essential components for the initiation of key cellular processes, such as membrane lipid synthesis, generation of LDs, and eicosanoid synthesis (26). LDs are classically defined as organelles with stored neutral lipids, and some reports suggest that some viruses promote LD formation, which is involved in virus replication and assembly (27)(28)(29). Our data demonstrate that MDV infection also increases LD formation; however, the exact role of LDs in MDV infection is still unknown. LD formation in MDV-infected cells is dependent on the FAS pathway, as FAS inhibitors reduced the numbers of LDs. This organelle is a significant source of triglyceride-derived arachidonic acid (AA), reservoir of COX-2 and site of PGE 2 synthesis (13,14,30). Our results demonstrate that MDV infection increases the levels of FAS, AA, COX-2, and PGE 2 synthesis; however, further studies are required to establish whether PGE 2 synthesis occurs in LDs of MDV-infected cells. Furthermore, our results confirm previous observations on the role of FAS in LD formation, replication of pathogens, and PGE 2 synthesis (13). In MDV-infected cells, both metabolic and nonmetabolic factors could be involved in COX-2 activation. Upregulation of COX-2 in the MDV-infected cells could be induced by several mechanisms, including transforming growth factor beta (TGF-␤) (31, 32), cyclic AMP (33), and activation of NF-B (34). Further studies are required to support the role of these factors in the activation of COX-2 during MDV infection. COX-2-inducing factors such as TGF-␤ (32), which also activates fatty acid synthesis (31,35), might be involved in activation of COX-2 in MDV-infected cells. In fact, we have recently shown that MDV infection increases the induction of TGF-␤ in vivo (36); however, further studies are required to examine the role of MDV-induced TGF-␤ in the induction of COX-2 during MDV infection. It has been shown that COX-2 in LDs, but not in cytoplasm, is involved in PGE 2 synthesis following infection or activation of the cells (14). Further experiments are required to establish the role of FAS-induced LDs in the activation of the COX-2/PGE 2 pathway by MDV. Finally, we demonstrated that PGE 2 promotes MDV replication via EP2 and EP4 receptors, which are upregulated in MDV-infected cells. PGE 2 can contribute to virus replication by inhibition of nitric oxide production (37), type I interferon production (38), and modulation of immune cells, including antigen-presenting cells and T cells (39). We have recently reported that soluble factors released from MDV-transformed T cells inhibit the function of chicken T cells in a COX-2-dependent manner (36). The proposed model for the role of FAS and COX-2/PGE 2 pathways in MDV infection is summarized in Fig. 6. This study has assisted us in the identification of potential pathways involved in MDV pathogenesis and may eventually lead to the development of control measures. However, we do not anticipate the use of FAS inhibitors as an antiviral agent in poultry because of their systemic side effects and potential residue in meat and eggs. Taken together, our results demonstrate that MDV activates FAS and COX-2/PGE 2 pathways, and replication of MDV is dependent on PGE 2 synthesis, which supports MDV infectivity through EP2 and EP4 receptors.

MATERIALS AND METHODS
harboring plasmid was performed per the manufacturer's instructions. For silencing purpose, CEFs were transfected with the silencing and scrambled plasmid backbone containing green fluorescent protein using Lipofectamine stem reagent (Life Technologies, Warrington, UK). The transfection efficiency was observed at 24 h posttransfection using fluorescence microscopy, and cells showing more than 70% transfection efficiency were used for MDV infection. A standard curve generated for both Meq (10-fold serial dilutions prepared from plasmid construct with Meq target) and the ovo gene (10-fold serial dilutions prepared from plasmid construct with ovo target) were used to normalize DNA samples and to quantify MDV genomes per 10 4 cells. All reactions were performed in triplicates to detect both Meq and the chicken ovotransferrin (ovo) gene on an ABI7500 system (Applied Biosystems) using standard conditions. MDV genomes were normalized and are reported as viral genome per 10 4 cells.
Real-Time PCR. Total RNA was extracted from mock-and MDV-infected CEFs using TRIzol (Thermo Fisher Scientific, Paisley, UK) according to the manufacturer's protocol and treated with DNA-free DNase. Subsequently, 1 g of purified RNA was reverse transcribed to cDNA using a Superscript III first-strand synthesis kit (Thermo Fisher Scientific, Paisley, UK) and oligo(dT) primers according to the manufacturer's recommended protocol. The resulting cDNA was diluted 1 : 10 in diethyl pyrocarbonate-treated water. Quantitative real-time PCR using SYBR green was performed on diluted cDNA using a LightCycler 480 II (Roche Diagnostics GmbH, Mannheim, GER) as described previously (36). Data represent means from 6 biological replicates. The fold decreases in the mRNA transcript for EP2, EP4, and COX-2 genes were detected by one-step RT-PCR performed with a Luna Universal one-step RT-qPCR kit (NEB) on an Applied Biosystems 7500 qPCR system using primers outlined in Table 2.
Western blotting. Mock-and MDV-infected CEFs were lysed in radioimmunoprecipitation assay buffer and quantified by spectrophotometry. The lysate was suspended in the Laemmli's sample loading buffer (Sigma-Aldrich, Dorset, UK) and loaded in 10% SDS-PAGE. After semidry transfer of SDS-PAGE on a nitrocellulose membrane, the membrane was blocked in 5% skim milk powder for 2 h. The membranes were incubated with primary antibody for 12 h at 4°C, washed, and incubated with secondary antibodies. Finally, the blots were probed with the Odyssey CLx imaging system (Li-Cor, USA), and bands were quantified with Image Studio Lite software.
Prostaglandin E 2 ELISA. PGE 2 was quantified in supernatant of mock-and MDV-infected cells using a colorimetric assay (R&D Systems, Abingdon, UK) based on competition between unlabeled PGE 2 in the sample and a fixed amount of conjugated PGE 2 . The assay was performed according to the recommendations of the assay kit manufacturer. In brief, CEFs were either mock infected or infected with RB1B, and the levels of PGE 2 in the supernatant were determined at 72 hpi or the assay results were measured using optical density at 450 nm. The concentrations of PGE 2 were determined against a standard curve.
Oil red O staining. For analysis of lipid droplets, cell monolayer was washed with phosphatebuffered saline (PBS) and fixed with 4% formaldehyde for 30 min. Cells were subsequently stained with oil red O solution for 30 min, followed by a wash in PBS. Cells were counterstained with hematoxylin for 3 min, followed by a wash with super Q water. Plates were visualized and imaged using a light microscope, and the pictures were processed using Adobe Photoshop software.

Gene and target
Oligonucleotide sequence a COX-2(PTGER2) siRNA 1 5=-ACCTCGATTGACAGCCCACCAACATATCAAGAGTATGTTGGTGGGCTGTCAATCTT-3= 5=-CAAAAAGATTGACAGCCCACCAACATACTCTTGATATGTTGGTGGGCTGTCAATCG-3= a The boldface at 5' ends represents sequences for duplex stability, which provide resistance to RNase degradation. Boldface in the middle of a segment represents the site for hairpin loop folding. Boldface at the 3' region represent oligonucleotide overhangs, which are resistant to nuclease degradation.
Fluorescence confocal microscopy. CEFs were seeded in 24-well plates that contained 12-mmdiameter round coverslips at a rate of 1.0 ϫ 10 5 cells per well. At 72 h after mock infection or infection with the pRB1B UL35-GFP virus (the virus was kindly provided by V. K. Nair, The Pirbright Institute), the samples were prepared for imaging. In brief, mock-or MDV-infected CEFs were fixed with 4% formaldehyde for 30 min and incubated with HCS LipidTOX Red neutral lipid stain, and then nuclei were labeled with 4=,6-diamidino-2-phenylindole (DAPI). Cells were viewed using a Leica SP2 laser-scanning confocal microscope. The data are presented as maximum projections of z-stacks (23 to 25 sections; spacing, 0.3 mm), which were analyzed using IMARIS (Bitplane Scientific Software). Ninety MDV-infected cells and 40 mock-infected cells were analyzed, and their LipidTOX-labeled neutral lipid-containing organelles were detected with the spot function of IMARIS. Images were processed using Adobe Photoshop software.
Statistical analysis. All data are presented as means Ϯ standard deviations (SD) from at least three independent experiments. Quantification was performed using Graph Pad Prism 7 for Windows. The differences between groups, in each experiment, were analyzed by nonparametric Wilcoxon tests (Mann-Whitney) or by Kruskal-Wallis test (one-way analysis of variance, nonparametric). Results were considered statistically significant at a P value of Ͻ0.05 (*).