Feline calicivirus- and murine norovirus-induced COX-2/PGE2 signaling pathway has proviral effects.

Cyclooxygenases (COXs)/prostaglandin E2 (PGE2) signaling pathways are known to modulate a variety of homeostatic processes and are involved in various pathophysiological conditions. COXs/PGE2 signaling pathways have also been demonstrated to have proviral or antiviral effects, which appeared different even in the same virus family. A porcine sapovirus Cowden strain, a member of genus Sapovirus within the Caliciviridae family, induces strong COX-2/PGE2 but transient COX-1/PGE2 signaling to enhance virus replication. However, whether infections of other viruses in the different genera activate COXs/PGE2 signaling, and thus affect the replication of viruses, remains unknown. In the present study, infections of cells with the feline calicivirus (FCV) F9 strain in the genus Vesivirus and murine norovirus (MNV) CW-1 strain in the genus Norovirus only activated the COX-2/PGE2 signaling in a time-dependent manner. Treatment with pharmacological inhibitors or transfection of small interfering RNAs (siRNAs) against COX-2 enzyme significantly reduced the production of PGE2 as well as FCV and MNV replications. The inhibitory effects of these pharmacological inhibitors against COX-2 enzyme on the replication of both viruses were restored by the addition of PGE2. Silencing of COX-1 via siRNAs and inhibition of COX-1 via an inhibitor also decrease the production of PGE2 and replication of both viruses, which can be attributed to the inhibition COX-1/PGE2 signaling pathway. These data indicate that the COX-2/PGE2 signaling pathway has proviral effects for the replication of FCV and MNV, and pharmacological inhibitors against these enzymes serve as potential therapeutic candidates for treating FCV and MNV infections.

FCV vaccines are available for cats [20]; however, there is a limit on its efficacy because of low or no preventive effect against FCVs with different antigenicity and short-lived immunity [13,20]. A lack of a robust and reproducible in vitro cultivation system for human noroviruses (HNoVs) has not been developed until recently, thereby hindering the development of effective interventions [21]. Therefore, our understanding of the HNoV life cycle was largely based on those surrogate viruses specifically the murine noroviruses (MNVs) [22]. Therapeutic candidates for FCV and MNV may be utilized for various FCV strains and HNoVs [13].
The cyclooxygenase (COX) enzymes, which convert arachidonic acid into prostaglandins (PGs), orchestrate a variety of homeostatic processes and participate in various pathophysiological conditions such as inflammation and immune responses [23][24][25][26][27][28]. Currently three isoforms of COXs have been identified, with COX-1 and COX-2 being the most studied [29]. COX-1 functions as a housekeeping isoform of COX and is expressed constitutively to perform functions such as control of renal blood flow, platelet aggregation, and provide protection against stomach ulcers [29]. COX-2 is activated in response to different extracellular or intracellular stimuli, which can lead to the accumulation of PGs including PGE 2 [30]. PGE 2 is commonly involved in cellular immunity and inflammation events, and serves as strong regulators of cell-cell interaction, cytokine production, antigen presentation, cell differentiation and survival, apoptosis, and cell migration [30].
Previously, we reported that infection of PSaV strain Cowden induced strong COX-2/PGE 2 signaling with only a transient COX-1/PGE 2 signal during late stage infection [34]. Pharmacological inhibitors or siRNAs against COX-1 and COX-2 significantly reduced PSaV replication, which was restored by the addition of PGE 2 , indicating that COXs/PGE 2 acts as a proviral signal [34]. However, the role of COXs/PGE 2 signaling is unknown for other members within the Caliciviridae family. This prompts us to investigate whether other members of caliciviruses, in particular the FCV and MNV strains, use the COX-2/PGE 2 signaling pathway in the regulation of their own replication. Here, we demonstrated that COX-2 expression was induced during FCV and MNV infections, thereby causing a surge in PGE 2 levels. Furthermore, pharmacological inhibitors and siRNAs against COX-1 and COX-2 enzymes significantly reduced PGE 2 production as well as MNV and FCV replication, which could be restored by the addition of PGE 2 . These results suggest that COX-2/PGE 2 pathway has proviral effects for FCV and MNV replication.

Cells and viruses
RAW264.7 mouse macrophage and Crandell Rees feline kidney (CRFK) cells were purchased from the American Type Culture Collection (ATCC, Manassas, VI, USA) and were routinely grown in Dulbecco's modified Eagle's minimal essential medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin. The FCV F9 strain was obtained from ATCC and was propagated in CRFK cells. The MNV strain CW-1 was a kind gift from Dr. H.W. Virgin, Washington University School of Medicine, USA. Cesium chloride (CsCl) density gradient ultracentrifugation was performed to purify each mass-cultured strain as described previously [61][62]. The viral titer for MNV and FCV strains was determined by median tissue culture infectious dose (TCID 50 ) in units per milliliter and immunofluorescence assay, respectively, as described below.

Chemicals and antibodies
NS-398 and SC-560 were purchased from Cayman Chemical (Ann Arbor, MI, USA). Indomethacin and dimethyl sulfoxide (DMSO) were from Sigma-Aldrich (St. Louis, MO, USA). COX-1 and COX-2 siRNAs, which were comprised of three targets for each gene, and scrambled siRNA were purchased from Santa Cruz Biotechnology, Inc. (Dallas, TX, USA). Synthetic PGE 2 was from Tocris Bioscience (Ellisville, MO, USA). Polyclonal antibody against rabbit COX-2 and monoclonal antibody against mouse COX-1 were acquired from Abcam (Cambridge, MA, USA). Mouse monoclonal antibody against FCV capsid protein was from Santa Cruz Biotechnology, Inc. The rabbit polyclonal MNV VPg antibody was a kind gift from Dr. I. Goodfellow, Cambridge University, UK. The secondary antibodies used were horseradish peroxidase-conjugated goat immunoglobulin against rabbit IgG (Cell Signaling, Beverly, MA, USA) and mouse IgG (Santa Cruz), and fluorescein isothiocyanae (FITC)-conjugated goat immunoglobulin against rabbit IgG and mouse IgG (Jackson Immuno Research Laboratory, West Grove, PA, USA).
Briefly, monolayers of RAW264.7 and CRFK cells grown in 96-well plates were incubated for 24 h with different concentrations of chemicals. After removal of the medium and washing twice with phosphate buffered saline (PBS, pH 7.4), 200 μl of MTT solution was added in each well and incubated for 4 h at 37˚C in a CO 2 incubator. Next, 150 μl of DMSO was added to each well and incubated for 10 min at room temperature. Optical density (OD) was determined in an enzyme-linked immunoabsorbent assay (ELISA) reader at 570 nm. Calculations for the percent cell viability was determine using the following formula: [(OD sample -OD blank )/ (OD control -OD blank )] × 100. Nontoxic concentrations of chemicals were utilized in this study.

Treatment of cells with inhibitors
The inhibitors and chemicals used in this study were diluted in DMSO to a stock concentration of 10 mM and were subsequently diluted in media to make working solutions. Monolayers of RAW264.7 and CRFK cells grown in 6-or 12-well plates were treated with chemicals or inhibitors as described previously [34]: mock-treatment, pre-treatment, post-treatment, and pre-post-treatment. Briefly, confluent cells pretreated with various concentrations of the inhibitors for 24 h were infected with MNV (MOI, 1 TCID 50 /ml) or FCV (MOI, 1 FFU/ml) strains for the pre-treatment groups. For the post-treatment groups, cells were adsorbed with MNV (MOI, 1 TCID 50 /ml) or FCV (MOI, 1 FFU/ml) strains and then treated with different concentrations of inhibitors. For the pre-post-treatment groups, cells were pretreated with different concentrations of inhibitors for 24 h, adsorbed with MNV (MOI, 1 TCID 50 /ml) or FCV (MOI, 1 FFU/ml) strains, and treated again with the same concentration of inhibitors described above.

Plasmid constructs
The cDNA from a partial part of the polymerase of the FCV F9 strain and a part of the N-terminal of the open reading frame (ORF) 1 of the MNV strain were amplified by conventional PCR using the primers listed in Table 1. The DNA fragments were gel purified using the Wiz-ard1 SV Gel and PCR Clean-Up system (Promega, Fitchburg, Wisconsin, USA) and ligated to pGEM-Teasy vector (Promega). Samples were transformed in homemade DH5α and positive colonies were picked by white and blue screening. The selected colonies were grown in LB-medium containing ampicillin. The plasmids were purified using Hybrid-Q Plasmid Rapidprep kit (GeneAll, Songpa-gu, Seoul, South Korea) and concentrations were determined by spectrophotometry (BioPhotometer plus, Eppendorf, Hamburg, Germany). The plasmids for FCV and MNV strains were named pTA-FCV and pTA-MNV-1, respectively.

RNA extraction
Mock-or virus-infected, mock-or inhibitor-treated, or siRNA-transfected cells were washed thrice with cold PBS, scraped, and collected in clean microtubes. Harvested samples were centrifuged at 2,469 × g for 10 min, and then total RNA was extracted using the PureLink RNA minikit (Ambion Life Technologies, Carlsbard, CA, USA) following the manufacturer's instructions. To quantify the genome copy numbers of FCV or MNV strains, the cells treated with inhibitors, infected with each strain, or transfected with each siRNA described above were freeze-thawed three times, and cell debris were removed by centrifugation at 2,469 × g at 10 min at 4˚C. Samples were immediately processed or stored at -80˚C until use. Total RNA was isolated from supernatants using a Viral RNA extraction kit (Bionner, Daejeon, South Korea) following the manufacturer's instructions. Concentrations of extracted RNAs were calculated by spectrophotometry at 260 nm using the BioPhotometer plus (Eppendorf).

Quantitative real-time PCR
A one-step real-time RT-PCR (qRT-PCR) assay with primer pairs specific to the polymerase of FCV and ORF1 of MNV was performed as described previously [61,[64][65][66]. The primer pairs used are listed in Table 1. Briefly, total RNA was isolated as described above from harvested samples that were subjected to three times freeze-thaw cycles. All reactions were carried out using a Corbett Research Rotor-Gene Real-time Amplification system (Corbett Research, Mortlake, Australia) and SensiFast™ SYBR1 Lo-ROX One Step Kit (Bioline, London, UK). Next, 25 μl qRT-PCR reactions were prepared, which were composed of 5 μl RNA template, 10 μl SensiFast one-step mixture, 1 μl of forward and reverse primers (final concentration of each primer: 10 pmole), 0.2 μl reverse transcriptase, 0.4 μl of RiboSafe RNase inhibitor, and 7.4 μl of RNase free water. The amplification profile for FCV was as follows: reverse transcription at 45˚C for 10 min, polymerase activation at 95˚C for 2 min, and 45 cycles of amplification consisting of denaturation at 94˚C for 10 s, primer annealing at 55˚C for 20 s, and extension at 72˚C for 10 s. For MNV-1, the reverse transcription was carried out at 45˚C for 10 min, followed by polymerase activation at 95˚C for 2 min, an initial denaturation at 95˚C for 5 min, and 45 cycles of amplification consisting of denaturation at 94˚C for 15 s, primer annealing at 57˚C for 30 s, and extension at 72˚C for 20 s. The copy numbers of the FCV and MNV-1 genes were calculated using 10-fold dilutions of a known concentration of in vitro transcribed complementary RNA using the pTA-FCV and pTA-MNV-1 plasmids to make the standard curve. Quantifications of RNA from samples were calculated using the Rotorgene 30001 (Corbett Research, Mortlake, Australia).
To determine the expression levels of COX genes in mock-, virus-, chemical-, or siRNAtreated samples, cells were harvested and total RNA were extracted as described above. RNA concentrations were determined, and cDNA was prepared using 1 μg of RNA and reverse transcribed using random hexamers (Roche, Basel, Switzerland). The specific primer pairs for qPCR were designed based on the published sequences for COX-1 and COX-2 as listed in Table 1. Reactions were carried out in a total volume of 25 μl containing 10 pmole of forward and reverse primers, cDNA, and TOPreal qPCR 2X Premix (Enzynomics, Daejon, South Korea). The amplification of COX-1 and COX-2 were as follows: initial denaturation at 95˚C for 5 min, followed by 40 cycles of denaturation at 95˚C for 10 s, primer annealing at 57˚C for 20 s, and extension at 72˚C for 10 s. The relative expressions of COX-1 and COX-2 were calculated via 2 -ΔΔCT using the Rotorgene 30001 software (Corbett Research, Mortlake, Australia) as described previously [34]. To normalize the samples, the β actin gene (Table 1) was quantified.

Detection of PGE 2 in cell culture supernatant by ELISA
Levels of PGE 2 were determined via the PGE 2 EIA kit (Cayman Chemical, Ann, Arbor, MI, USA) following the manufacturer's instructions for clarified supernatants collected in different time points from the cells mock-infected or infected with each virus, mock-treated or treated with each chemical, or scrambled siRNA-transfected or transfected with siRNAs against COX-1 or COX-2. Using an ELISA reader, the absorbance was read at 405 nm and concentrations of PGE 2 were calculated in comparison to a standard curve.

TCID 50 assay
The median tissue culture infectious dose (TCID 50 ) of MNV and FCV strains in the samples obtained from the above experiments were determined as described previously [34]. Briefly, 10-fold serial dilutions of clarified supernatants were prepared. Confluent RAW264.7 and CRFK cells grown on 96-well plates were inoculated with 200 μl of each diluted sample and then incubated at 37˚C in a 5% CO 2 incubator. After 3-5 days postinfection, viral titers were calculated using the Reed and Muench method [67] and expressed as TCID 50 /ml.

Silencing of COX-1 and COX-2 genes by transfection of siRNA
The siRNAs targeting COX-1 and COX-2 genes or scrambled siRNA were transfected to confluent CRFK and RAW 264.7 cells grown on 6-or 24-well plates using Lipofectamine 30001 reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions [34]. Cell were then mock-infected or infected with FCV (MOI, 1 FFU/ml) and MNV (MOI, 1 TCID 50 /ml) strains. Cells were subjected to qPCR, TCID 50 /ml, and Western blot analyses at the appropriate time points.

Cell lysate preparation and Western blot analysis
Cells from the above experiments were washed, harvested, and lysed at different time points using a cell extraction buffer (Invitrogen) supplemented with protease and phosphatase inhibitors (Roche). Denatured cell lysates were resolved in sodium dodecyl sulfate (SDS) polyacrylamide gels and then transferred to nitrocellulose blotting membranes (Amersham Protran, GE Healthcare Life Science, Germany). Membranes were immunoblotted with primary antibodies against COX-1, COX-2, glyceraldehyde 3-phosphate dehydrogenase (GAPDH), MNV VPg, or FCV capsid. Secondary antibodies specific for rabbit or mouse IgG were added for 1 h at room temperature and were developed using an enhanced chemiluminescence reaction kit (DoGen, Nowon-gu, Seoul, South Korea). Images were captured on the Davinch-Western imaging system (Young Ltd., Kang-Nam, Seoul, South Korea). Samples were normalized to the corresponding GAPDH level in the same samples. The intensity of COX-1 and COX-2 proteins relative to GAPDH was determined by densitometric analysis.

Immunofluorescence assay
The viral titer for FCV F9 strain was determined by immunofluorescence assays as described previously [12,13]. Briefly, confluent CRFK cells grown on 96-well plates were inoculated with 200 μl of each diluted sample and then incubated at 37˚C in a 5% CO 2 incubator. After 4 hpi, cells were fixed with cold acetone. Cells were washed with PBS containing new fetal bovine serum (PBS-FBS) before incubating with anti-FCV capsid (1:100 dilution) overnight at 4˚C. The cells were washed three times with cold PBS, and then FITC-conjugated goat anti-mouse IgG (Molecular Probes, Eugene, OR, USA) secondary antibody was added and incubated for 1 h at room temperature. Afterwards, slides were washed three times with cold PBS and mounted with 60% glycerol in PBS (pH 8.0). FCV titer was calculated using the Reed and Muench method [67] and expressed as fluorescence focus unit (FFU)/ml. Infection inhibitory effects of each inhibitor or siRNA were determined by immunofluorescence assays as described previously [68]. Briefly, cells grown on 8-well chamber glass slides were treated with various inhibitors or transfected with siRNAs before or after infection of FCV (1 FFU/ml) or MNV (MOI, 1 TCID 50 /ml) strains as mentioned above. Then, cells were washed three times with cold PBS before being fixed with 4% formaldehyde in PBS. Cells were permeabilized with 0.02% Triton X-100, incubated at room temperature for 10 min, washed with PBS containing new fetal bovine serum (PBS-FBS) before incubating with anti-MNV VPg (1:100 dilution) or anti-FCV capsid (1:100 dilution) overnight at 4˚C. The cells were washed three times with cold PBS, and then anti-rabbit or anti-mouse Alexa Fluor 488 (Molecular Probes, Eugene, OR, USA) secondary antibodies were added. Afterwards, slides were mounted with SlowFade Gold antifade reagent (Life Technologies, Eugene, OR, USA) containing DAPI (4',6-diamidino-2-phenylindole) solution for nuclear staining and was visualize using the LSM 510 confocal microscope and analyzed using LSM software (Carl Zeiss, Oberkochen, Germany).

FCV and MNV infections induce COX-2 expression that leads to the production of PGE 2
We previously reported that infection of LLC-PK cells with PSaV strain Cowden strongly activated the COX-2/PGE 2 signaling pathway but transiently induced COX-1/PGE 2 signaling pathway during the late stage of infection [34]; hence, we examined whether both FCV and MNV strains also have the ability to activate the COXs/PGE 2 signaling pathway during their replication. To investigate this, we first evaluated the influence of FCV and MNV infections on COX gene and protein expressions. Expression levels of COX-2 mRNA and protein gradually increased from 1 h post-infection (hpi) (Fig 1A and 1C) in parallel with increased FCV viral RNA synthesis (Fig 1B), while COX-1 protein levels remained unchanged (Fig 1A and 1C). Similarly, the timedependent increase in the expression of COX-2 mRNA and protein levels was observed in association with an increase in MNV mRNA level starting at 4 hpi (Fig 2A-2C). The expression level of COX-1 mRNA and protein was also unaffected during MNV infection (Fig 2A and 2C). Since PGE 2 is one of the main products of the COX enzymes [29], we also assessed the expression levels of PGE 2 during FCV and MNV infections. Concurrent with the increased COX-2 levels during MNV and FCV infections, the production of PGE 2 was also elevated in a time-dependent manner starting from 1 hpi for FCV ( Fig 1D) and from 4 hpi for MNV (Fig 2D).
To further confirm the correlation of increased PGE 2 with the FCV-and MNV-mediated induction of COX-2, the influence of selective and nonselective COX inhibitors on PGE 2 expression was investigated. Non-toxic doses were used for this study as determined by the MTT assay (S1 Fig). The selective COX-1 inhibitor SC-560, selective COX-2 inhibitor NS-398, and nonselective COX-1 and -2 inhibitor indomethacin hampered FCV mediated-PGE 2 induction in a dose-dependent manner when added after the removal of the virus inoculum (post-treatment) or added before and after the virus adsorption step (pre-post-treatment) (Fig 3A-3C). Similarly, addition of COX inhibitors after removal of the MNV inoculum or during the entire course of the MNV infection reduced PGE 2 levels (Fig 3D-3F). However, the pretreatment of cells and consequent removal of inhibitors before FCV or MNV infection did not lower PGE 2 levels due to the reversible nature of the inhibitors (Fig 3A-3F).
To further corroborate the influence of COX genes in the production of PGE 2 during FCV and MNV infections, the effects of siRNAs were also examined. RAW 264.7 and CRFK cells transfected with siRNAs against COX-1 or COX-2 genes showed a reduction in the expression levels of their corresponding target genes (Fig 3G and 3H). Transfection of siRNAs targeting COX-1 or COX-2 in RAW 264.7 and CRFK cells following FCV and MNV infections, respectively, significantly reduced the amount of PGE 2 released (Fig 3G and 3H). These results confirmed that induction of COX-2, together with the basal expression of COX-1, accounted for the increase in PGE 2 in FCV and MNV-infected cells.

Inhibition of both COX enzymes reduce FCV and MNV replication
To assess the effect of COX stimulation on the replications of FCV and MNV strains, we assessed the impact of COX inhibitors on FCV and MNV replication. RAW 267. 4

Fig 1. Feline calicivirus (FCV) infection induces COX-2 mRNA and protein expression leading to the production of PGE 2 . (A and B)
The mRNA expression levels of COX-1 and COX-2 genes (A), and viral RNA level (B) in CRFK cells infected with or without FCV (MOI, 1 FFU/cell) at the indicated time points were determined by quantitative real time PCR. For COX-1 and COX-2, mRNA levels were normalized to β-actin mRNA and are illustrated as a fold induction against that of the mock-infected cells. (C) Monolayers of CRFK cells were infected with or without FCV (MOI, 1 TCID 50 /ml) for the indicated time points, and the levels of the COX-1, COX-2, and GAPDH proteins were analyzed by Western blot analysis. GAPDH was used as a loading control. The intensity of each target protein relative to GAPDH was determined by densitometric analysis and is indicated above each lane. (D) PGE 2 levels in supernatants harvested from cells infected with or without FCV (MOI, 1 FFU/cell) at the indicated time points were analyzed by ELISA. The levels of PGE 2 in supernatants were compared between mock-and FCV-infected groups. All data shown were from three independent experiments and are presented as means and standard errors of mean. Statistical differences were evaluated by one-way analysis of variance. Ã p < 0.05; ÃÃ p < 0.001; ÃÃÃ p < 0.0001. https://doi.org/10.1371/journal.pone.0200726.g001 cells were pre-treated (Pre), post-treated (Post), or pre-post-treated (Pre-Post) with COX inhibitors, and viral replication was monitored at 8 hpi for FCV and 24 hpi for MNV by examining the viral titers and RNA levels. The selective COX-1 inhibitor SC-560 reduced the levels of infectious virus and RNA~3 log 10 for FCV (Fig 4A and 4B) and by~4 log 10 for MNV ( Fig  5A and 5B) when added after infection (post-treatment) or maintained throughout the virus infection (pre-post-treatment). Conversely, the addition of NS-398 (selective COX-2 inhibitor) following the removal of the virus inoculum (post-treatment) or during the entire progression of infection (pre-post-treatment) caused a more significant reduction in FCV (~4.5 log 10 ) in both infectious virus and viral RNA (Fig 4C and 4D), while MNV infectious viral titer and genome copy numbers were reduced~3 log 10 (Fig 5C and 5D). Indomethacin, a nonselective COX inhibitor, showed more significant effects on MNV infection, leading to a~5 log 10 reduction in virus yield and RNA synthesis (Fig 5E and 5F), while FCV infection elicited a~3 log 10 decrease in infectious viral titer and genome copy numbers (Fig 4E and 4F) when added after virus adsorption or maintained throughout the course of the infection. There was no significant effect on FCV or MNV replication in cells pretreated with COX inhibitors (Figs 4A-4F and 5A-5F). We also assessed the effect of COX-specific siRNAs on FCV and MNV replication. FCV genome copy numbers and titers were more strongly reduced (~4 log 10 ) in cells transfected with siRNA against COX-2 than that in cells transfected with COX-1 siRNA (~2 log 10 ) (Fig 4G and 4H). Transfection of COX-1-or COX-2-specific siRNAs resulted in similar substantial reduction in MNV replication, causing a~4 log 10 decrease in infectious viral titer and RNA levels (Fig 5G and 5H).
To further assess the effect of COX inhibitors on FCV and MNV replication, we also analyzed viral protein levels in the pre-and post-treatment groups. Consistent with the above results, a reduced production of FCV capsid protein was observed with the post-treatment of COX inhibitors in a dose-dependent manner, in which NS398 showed stronger anti-FCV effects against other inhibitors (Fig 6A). The expression level of MNV VPg protein decreased  The inhibitory effects of each drug on virus titer or genome copy number were compared between mock-and drug-treated groups. (G and H) COX-1, COX-2, or scrambled (Scram) siRNAs were transfected in CRFK cells before infection with FCV (MOI, 1 FFU/cell). The samples were harvested at 8 hpi, and then the viral titer and genome copy number were determined by TCID 50 /ml (G) and qRT-PCR (H) analyses. Three independent experiments were conducted and presented as due to the COX inhibitors in a dose-dependent manner, particularly with the post-treatment of indomethacin (Fig 7A). Transfection of COX-1 and COX-2 siRNAs resulted in lower protein induction in both FCV-and MNV-infected cells (Figs 6B and 7B). Furthermore, treatment of CRFK cells with NS-398 or transfection with COX-1 and COX-2 specific siRNAs caused a significant decreased in the number of FCV antigen-positive cells (Fig 6C). Similarly, a reduced number of MNV antigen-positive cells was also observed in RAW264.7 cells treated with NS-398 or transfected with COX-1 or COX-2 specific siRNAs ( Fig 7C). Together, these results suggested that both COX-1 and COX-2 augment FCV and MNV replication possibly in relation with the increased levels of PGE 2 .

Addition of PGE 2 restores the inhibitor effect of COXs on FCV and MNV replication
To further study the proviral effect of COXs/PGE 2 induction during FCV and MNV infections, we assessed whether the addition of exogenous PGE 2 restores the inhibitory effect of COX inhibitors on viral replication. The selective COX-2 inhibitor NS-398 was used for FCV, while the nonselective inhibitor indomethacin was used for MNV, both of which showed strong inhibitory effect compared with other COX inhibitors. Non-cytotoxic doses of COX-inhibitors in combination with exogenous PGE 2 were used in this experiment. The addition of PGE 2 lead to the restoration of both FCV and MNV progeny production and viral RNA levels in a dosedependent manner (Fig 8A-8D). Collectively, these results confirmed that PGE 2 , the final product of both COX enzymes, mediated the proviral effects on FCV and MNV replication.

Discussion
Depending on the viruses, the COXs/PGE 2 signaling pathway was used differently: COX-1/ PGE 2 for PRRS arterivirus [35], COX-2/PGE 2 for many viruses [36, 37, 39-42, 46-49, 50-53], or both for pseudorabies herpesvirus [54]. In the present study, both FCV and MNV strongly activated the COX-2/PGE 2 pathway from an early replication time point. These results are different from our previous finding where the PSaV strain Cowden, belonging to the same Caliciviridae family, induced late activation of the COX-2/PGE 2 signaling pathway [34]. This suggested that the activation time of the COX-2/PGE 2 signaling pathway was different even in the same Caliciviridae family. In addition, the PSaV Cowden strain transiently activated COX-1/PGE 2 signaling during the later stage of infection [34]. Interestingly, no increases in the COX-1 mRNA and protein levels were observed during the entire FCV and MNV infection. However, silencing of COX-1 by specific siRNAs as well as the inhibition of COX-1 by the specific inhibitor SC-560 decreased the production of PGE 2 and replication of both viruses. Although the detailed mechanism of these results is unknown, it is presumed that the inhibition of the basal expression level of COX-1 by its specific inhibitor and siRNA suppressed PGE 2 production and thereby the replication of both viruses. Based on these results, it could be concluded that unlike the PSaV-induced late activation of both COX-1 and COX-2/PGE 2 signaling pathways, both FCV and MNV activate only COX-2/PGE 2 signaling pathway from early replication, but COX-1 might be an important factor for the successful replication of FCV and MNV.
https://doi.org/10.1371/journal.pone.0200726.g004 COX-2/PGE 2 pathway enhances FCV and MNV replication The inhibitory effects of each drug on the viral titer or genome copy number were compared between mock-and study, the inhibition or silencing of COX-2 reduced replication of FCV and MNV. Moreover, reduced viral replication by COX inhibitors was restored by addition of PGE 2 . Similar results have been recently reported in a PSaV infection study [34]. These data indicate that like PSaV [34], FCV-and MNV-induced activation of the COX-2/PGE 2 signaling pathway acts as a proviral mechanism for the replication of FCV and MNV.
Nonsteroidal anti-inflammatory drugs (NSAIDs) targeting COX enzymes are one of the most widely prescribed drugs due to their analgesic effects and their potent anti-inflammatory and anti-pyretic properties in the western world [26,70]. NSAIDs have been reported as potential therapeutic drugs for some viral infection. For example, pharmacological use of a COX-2 inhibitor in combination with neuraminidase inhibitors enhanced the survival of mice infected with H5N1 influenza [71]. The selective COX-2 inhibitor NS-398 was shown to protect mice from succumbing to dengue virus-2 infection [41]. In this study, we determined the potency of different COX inhibitors against FCV and MNV infections. In the FCV infection, the COX-2 selective inhibitor NS398 exerted~4 log 10 suppression in the FCV titer compared to other inhibitors. Among the COX inhibitors tested against MNV, indomethacin, a nonselective COX inhibitor, showed stronger anti-MNV effects than other inhibitors, leading to ã 5 log 10 decrease in MNV titer. These results suggest that COX inhibitors are potential drug-treated groups. (G and H) COX-1, COX-2, and scrambled (Scram) siRNAs were transfected in RAW264.7 cells before infection with MNV (MOI, 1 TCID 50 /ml). The samples were harvested at 24 hpi, and virus titer and genome copy number were determined by TCID 50 /ml (G) or qRT-PCR (H) analyses. The inhibitory effects of each siRNA on viral titer and genome copy number were compared between the mock-and each siRNA-transfected group. Three independent experiments were conducted and presented as means and standard errors of the mean. Statistical differences were determined using one-way analysis of variance. ÃÃ p < 0.001; ÃÃÃ p < 0.0001.
https://doi.org/10.1371/journal.pone.0200726.g005 candidates for the treatment of FCV infections, which are particularly serious problems for cats [12,14]. Future studies are needed to determine whether in vivo treatments with COX inhibitors are effective in treating FCV and MNV infections. Nitric oxide (NO) has been shown to exert antiviral effects during virus infections [72][73][74][75][76]. It has also been shown that the COX/PGE 2 signaling pathway also regulate inducible nitric oxide synthase, the enzyme responsible for the production of NO [77]. In vitro and in vivo vesicular stomatitis virus infection has been shown to activate the COX-2/PGE 2 signaling pathway, which in turn restrict NO production [37,46]. In the same way, we demonstrated that PSaV-induced activation of COX-2/PGE 2 signaling pathway inhibited NO production [34] to favor PSaV replication. A similar response is expected in FCV and MNV infections, which is the ongoing goal of future studies.
In a few cases, viral proteins could activate COX/PGE 2 signaling, either directly or indirectly as a transcriptional transactivator of COX-2 gene expression [42,[78][79][80]. For example, the HCV proteins such as NS3 and NS5A are known to enhance the COX-2/PGE 2 pathway by activating multiple signaling pathways [81,82]. In severe acute respiratory syndrome coronavirus infection, viral nucleocapsid protein activates COX-2/PGE 2 pathway through direct binding to regulatory elements for NF-κB and CCAAT/enhancer binding protein [80]. In our previous report, PSaV VPg and ProPol proteins significantly enhanced the expression of COX-2/PGE 2 signaling pathway [34]. Likewise, some FCV and MNV proteins are anticipated to activate the COX-2/PGE 2 signaling pathway, the mechanism of which will be investigated in a future study.
In conclusion, our results denote a crucial role for the COX-2/PGE 2 signaling pathway in the successful replication of FCV and MNV by creating a cellular environment suitable for