Anti-inflammatory Compound Shows Therapeutic Safety and Efficacy against Flavivirus Infection

Flaviviruses comprise several medically important viruses, including Japanese encephalitis virus, West Nile virus, dengue virus (DENV), yellow fever virus, and Zika virus (ZIKV). A large outbreak of DENV and ZIKV occurred recently, leading to many cases of illness and death. However, despite decades of effort, we have no clinically specific therapeutic drugs against DENV and ZIKV. Previous studies showed that inflammatory responses play a critical role in dengue and Zika virus pathogenesis.

ABSTRACT

Flaviviruses comprise several medically important viruses, including Japanese encephalitis virus, West Nile virus, dengue virus (DENV), yellow fever virus, and Zika virus (ZIKV). A large outbreak of DENV and ZIKV occurred recently, leading to many cases of illness and death. However, despite decades of effort, we have no clinically specific therapeutic drugs against DENV and ZIKV. Previous studies showed that inflammatory responses play a critical role in dengue and Zika virus pathogenesis. Thus, in this study, we examined a series of novel anti-inflammatory compounds and found that treatment with compound 2d could dose dependently reduce viral protein expression and viral progeny production in HEK-293 and Raw264.7 cells infected with four serotypes of DENV and ZIKV. In addition, considering medication safety, compound 2d could not suppress cyclooxygenase-1 (COX-1) enzymatic activities and thus could prevent the side effect of bleeding. Moreover, compound 2d significantly inhibited COX-2 enzymatic activities and prostaglandin E₂ levels, associated with viral replication, compared to results with a selective COX-2 inhibitor, celecoxib. Furthermore, administering 5 mg/kg compound 2d to DENV-2-infected AG129 mice prolonged survival and reduced viremia and serum cytokine levels. Overall, compound 2d showed therapeutic safety and efficacy in vitro and in vivo and could be further developed as a potential therapeutic agent for flavivirus infection.

INTRODUCTION

Flaviviruses, single-stranded, positive-sense RNA viruses comprise several medically important viruses, including Japanese encephalitis virus, West Nile virus, dengue virus (DENV), yellow fever virus, Zika virus (ZIKV), and tick-borne encephalitis virus (1, 2). Among the mosquito-borne flaviviruses, DENV has four serotypes (DENV-1, -2, -3, and -4) that cause infection and can lead to illnesses ranging from dengue fever (DF) to severe and life-threatening dengue hemorrhagic fever (DHF) and dengue shock syndrome (DSS) (3). ZIKV was first identified in Uganda in 1947, and a large outbreak of ZIKV was reported in several countries in 2016 (4). ZIKV infection causes severe neurological complications in adults and microcephaly, congenital malformation, and fetal demise in fetuses (5–7).

With global climate change, these mosquito-borne disease epidemics seem to be more frequent and diverse. However, despite decades of effort, only one dengue vaccine, Dengvaxia, has been licensed and approved for use in only some countries of endemicity, but the overall vaccine efficacy was lower than expected (8). We have no clinically specific therapeutic drugs against DENV and ZIKV (9, 10). Effective anti-DENV and anti-ZIKV drugs are needed to ameliorate disease or reduce disease severity and fatalities during viral infection.

Inflammatory responses play a critical role in dengue virus pathogenesis and contribute to dengue disease severity (11, 12). Levels of inflammation mediators such as tumor necrosis factor alpha (TNF-α), gamma interferon (IFN-γ), interleukin-6 (IL-6), IL-8, and prostaglandin E₂ (PGE₂) are elevated in dengue patients and increased to higher levels in severe dengue cases (13–16). TNF-α is the main cytokine that induces vascular leakage, and its levels are correlated with disease severity during DENV infection (17). TNF-α-induced IL-6 and IL-8 production was reported to contribute to increased endothelial permeability in DHF patients (18). Moreover, PGE₂ overproduction could lead to bone pain in DENV-infected patients, and its level is associated with DENV replication and infectivity (19, 20). Therefore, reducing the inflammation response is a major key to addressing the severe forms of dengue disease.

Current anti-inflammatory drugs such as acetylsalicylic acid, ibuprofen, or related nonsteroidal anti-inflammatory drugs (NSAIDs) aggravate the bleeding effect and enhance the syndrome of DHF owing to inhibition of cyclooxygenase-1 (COX-1) levels (21, 22); thus, these drugs have been prohibited for DENV-infected patients by the World Health Organization (WHO) (23, 24). Zika virus is also a mosquito-borne virus that produces illnesses clinically similar to dengue; treatment with classical NSAIDs has also been prohibited by the WHO (25). In addition, because dengue is frequently associated with hepatic effects, acetaminophen can be used only for dengue patients without liver failure (26). Furthermore, COX-2 expression has been found associated with DENV replication (16, 27), and the COX-2 inhibitor NS398 could reduce DENV-2 replication in hepatoma Huh-7 cells (16). However, NS398 treatment could not protect mice against DENV-2 challenge at 7 to 11 days postinfection (dpi) in the ICR suckling mouse model, and all infected mice died within 11 days although the survival rate at 6 dpi was 60% with 5 mg/kg NS398 (16). Another safer NSAID with selective inhibition of COX-2, celecoxib, was reported to decrease symptoms of severe systemic inflammation caused by influenza virus infection in mice (28). However, even though several studies revealed the association between COX-2 and antiviral abilities, COX-2 showed diverse expression profiles in different DENV-infected cell lines (29).

In this study, we aimed to evaluate the antiviral effects of a series of anti-inflammatory compounds against infection with the four serotypes of DENV and with ZIKV. Moreover, we examined the therapeutic safety and efficacy of the compound compared to results with celecoxib in vitro and in vivo.

RESULTS

A series of novel anti-inflammatory compounds exhibited antiviral potential without cytotoxic effects.

A series of 4H-chromenes and chromeno[2,3-b]pyridine derivative compounds were reported to have potent anti-inflammatory activities by suppressing the PGE₂ level in human chondrocytes and in the rat paw edema model (30). We first determined the cytotoxic effects of the compounds 1h, 2d, 2j, and 2l (Fig. 1A) by a WST-1 [2-(4-iodophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazoliumwater-soluble tetrazolium salt] cell proliferation assay in human kidney HEK-293 cells and murine macrophage Raw264.7 cells because macrophages are the main target of DENV (31). None of the tested compounds (1h, 2d, 2j, and 2l, at concentrations up to 50 μM) had significant cytotoxic effects in HEK-293 cells and Raw264.7 cells (Fig. 1B) at 48 h.

FIG 1.

Molecular structures and antiviral abilities of a series of anti-inflammatory compounds. (A) Structures of anti-inflammatory compounds 1h, 2d, 2j, and 2l. (B) HEK-293 cells or Raw 264.7 cells were treated with solvent or various concentrations of compounds 1h, 2d, 2j, and 2l for 48 h, and cell viability was determined by WST-1 assay. The percentage was obtained by comparison of results to those with a solvent control, set at 100%. (C and D) HEK-293 cells were infected with DENV-1 or DENV-2 (MOI of 1) and then treated with compound 1 h, 2d, 2j, or 2l at 20 μM for 48 h. Western blot analysis of viral protein levels of DENV-1 E (C) and DENV-2 NS3 (D) in cell lysates was performed, and the ratio of the viral E or NS3 protein level to the level of GAPDH or actin, respectively, was adjusted to that of the solvent control. Viral titers in culture supernatants were examined by plaque-forming assay. Data are means ± SD of 3 independent experiments. **, P < 0.01; ***, P < 0.001 (two-tailed Student's t test).

Next, we evaluated the antiviral activities of compounds 1 h, 2d, 2j, and 2l at 20 μM in HEK-293 cells infected with DENV-1 or DENV-2. Treatment with compound 2d significantly inhibited DENV-1 and DENV-2 viral protein expression and reduced viral titers by 1 to 2 orders of magnitude compared with levels for compounds 1 h, 2j, and 2l in HEK-293 cells during DENV-1 and DENV-2 infection (Fig. 1C and D). In addition, in DENV-2-infected HEK-293 cells (see Fig. S1A in the supplemental material) and Raw264.7 cells (Fig. S1B), there was a greater reduction in PGE₂ levels upon treatment with compound 2d than with compound 1h, 2j, or 2l. Therefore, compound 2d had potent antiviral ability and was selected for subsequent studies.

Anti-inflammatory compound 2d exhibited antiviral activity against four DENV serotypes and ZIKV infection.

To investigate the antiviral activities of compound 2d against other DENV serotypes and ZIKV, we preinfected HEK-293 cells with each virus and then treated cells with compound 2d at 10 to 30 μM. Western blot analysis, a plaque assay, and an immunofluorescence assay demonstrated that, at 48 h postinfection, compound 2d dose dependently reduced viral protein expression and viral progeny production in HEK-293 cells infected with DENV-1, -2, -3, and -4 and ZIKV (Fig. 2A and B). In addition, we determined the selectivity index (SI) of compound 2d for four DENV serotypes and ZIKV in HEK-293 cells (Table 1 and Fig. S2). The 50% inhibitory concentration (IC₅₀; calculated as the concentration of drug at which virus yield was inhibited by 50%) of compound 2d against all DENV serotypes and ZIKV in HEK-293 cells at 48 h ranged from 6.9 to 7.6 μM based on virus titer levels (Table 1 and Fig. S2B to F), whereas the 50% cytotoxic concentration (CC₅₀; calculated as the concentration that resulted in 50% cellular cytotoxic effect) of compound 2d in uninfected HEK-293 cells was 112.3 μM at 48 h (Table 1 and Fig. S2A). Thus, the SIs (SI = CC₅₀/IC₅₀) were 14.8, 14.8, 15.8, 15.6, and 16.3 for DENV-1, DENV-2, DENV-3, DENV-4, and ZIKV, respectively (Table 1), thereby suggesting a broad antiviral effect of compound 2d. Furthermore, because macrophages are important target cells during natural DENV infection, we used macrophage Raw264.7 cells to further examine the antiviral effects of compound 2d. Compound 2d at the indicated doses shown in Fig. S3 inhibited viral protein expression and reduced viral titers by 1 to 2 logs. Hence, compound 2d has a broad antiviral effect against Flavivirus members, so it could be a potential therapeutic drug against four serotypes of DENV and ZIKV infection.

FIG 2.

Antiviral activities of compound 2d against four DENV serotypes and ZIKV infection in HEK-293 cells. (A) HEK-293 cells were infected with DENV-1 to DENV-4 or ZIKV without (solvent) or with compound 2d for 48 h. Viral protein levels were determined by Western blot analysis, and actin or GAPDH was used for a loading control; the ratio of the viral NS3 or E protein level to actin or GAPDH, respectively, was adjusted to the level of the solvent control. Viral progeny production in culture supernatants was measured by plaque-forming assay. Data are means ± SD of 3 independent experiments. *, P < 0.05; **, P < 0.01 (by two-tailed Student's t test). (B) Immunofluorescence assay of DENV-2 NS3 protein level (green) and nuclei (blue) in HEK-293 cells with or without compound 2d treatment.

TABLE 1.

Inhibitory concentrations and cytotoxic concentrations of compound 2d at 48 h in HEK-293 cells

Virus	CC50 (μM)a	IC50 (μM)b	SI (CC50/IC50)c
DENV-1	112.3	7.6	14.8
DENV-2	112.3	7.6	14.8
DENV-3	112.3	7.1	15.8
DENV-4	112.3	7.2	15.6
ZIKV	112.3	6.9	16.3

^aCC₅₀ is the concentration that produced a 50% cellular cytotoxic effect in uninfected HEK-293 cells.

^bIC₅₀ is the concentration that inhibited 50% of virus titers in HEK-293 cells infected with the indicated viruses (MOI of 1) and treated with various concentrations of compound 2d. Virus yields in the supernatants were determined at 48 h postinfection by focus-forming assay.

^cSI, selectivity index.

Inhibition of DENV-2 and ZIKV infection by compound 2d is not related to a virucidal effect.

To evaluate a possible virucidal effect of compound 2d, DENV-2 or ZIKV was preincubated with an equal volume of the different concentrations of compound 2d for 1 h at 37°C and then titrated to determine the remaining infectivity. The results showed no significant reduction in DENV-2 (Fig. 3A) or ZIKV (Fig. 3B) infectivity compared to the level for the solvent control, thus indicating that compound 2d does not exhibit a virucidal effect against DENV-2 or ZIKV.

FIG 3.

Evaluation the virucidal activity of compound 2d. To assay virucidal activity of compound 2d, DENV-2 (A) or ZIKV (B) was preincubated with compound 2d at different concentrations for 1 h at 37°C. Then, the remaining infectivity in each sample was determined by plaque-forming assay. Data are means ± SD of 3 independent experiments.

Compound 2d showed therapeutic safety and efficacy during DENV and ZIKV infection compared to results with celecoxib.

Classical NSAIDs, such as aspirin, could inhibit viral protein and COX-1 expression in DENV-2-or ZIKV-infected HEK-293 cells (Fig. 4A and B), but they are prohibited for DENV- or ZIKV-infected patients because a COX-1 inhibitor could lead to a bleeding side effect owing to suppression of COX-1 enzymatic activity (23–25). Therefore, we further examined the therapeutic safety of compound 2d. We detected the COX-1 mRNA level in DENV-2-infected HEK-293 cells and found that compound 2d could not inhibit COX-1 levels (Fig. 4C), so the compound could have therapeutic safety in DENV-infected patients. Because suppressing COX-2 levels could inhibit DENV-2 replication (16), we compared the therapeutic efficacies between compound 2d and celecoxib, an NSAID without bleeding side effects because it selectively inhibits the COX-2 level. Compared to results with celecoxib in DENV-2- or ZIKV-infected HEK-293 cells, compound 2d significantly reduced the mRNA levels of COX-2 (Fig. 4D) and its metabolite PGE₂ (Fig. 4E) as well as inhibiting viral protein expression and viral titers (Fig. 4F and G).

FIG 4.

Therapeutic safety and efficacy of compound 2d compared to results with celecoxib. (A and B) HEK-293 cells were infected with DENV-2 (A) or ZIKV (B) without (solvent) or with aspirin at 2 or 4 mM for 48 h. Western blot analysis of DENV-2 NS3, ZIKV E, and COX-1 protein levels in DENV-2- or ZIKV-infected HEK-293 cells. (C to G) HEK-293 cells were infected with DENV-2 or ZIKV without (solvent) treatment or with compound 2d or celecoxib treatment for 48 h. The levels of COX-1 (C) and COX-2 (D) were measured by qRT-PCR, and the PGE₂ level (E) in cell culture supernatants was detected by ELISA. The percent change of COX-1, COX-2, and PGE₂ expression was obtained by comparison with that of DENV-2-infected HEK-293 cells, set to 100%. Western blot analysis of COX-2 and DENV-2 NS3 (F) and ZIKV E (G) protein levels and a plaque-forming assay of viral titers were performed in DENV-2- or ZIKV-infected HEK-293 cells with or without compound 2d and celecoxib treatment. (H to K) HEK-293 cells were infected with DENV-2 or ZIKV, as indicated, at an MOI 1 without treatment (solvent) or with treatment with compound 2d or celecoxib or aspirin for 24 h. The percent change of COX-2 or COX-1 activity, as indicated, was obtained by comparison to that of the solvent control, set to 100%. Data are means ± SD of 3 independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001 (by two-tailed Student's t test).

In addition, to further explore whether the antiviral mechanism of compound 2d directly and selectively inhibited COX-2 enzymatic activity, we measured COX-1 and COX-2 enzymatic activities by using a COX activity assay kit. Compared to levels of activity with celecoxib or aspirin, compound 2d significantly inhibited COX-2 enzymatic activity (Fig. 4H and I) in DENV-2- or ZIKV-infected HEK-293 cells without suppressing COX-1 enzymatic activity (Fig. 4J and K), which indicates that the antiviral activity of compound 2d was via direct and selective inhibition of COX-2 enzymatic activity. Overall, compound 2d showed therapeutic safety and efficacy compared to results with celecoxib during flavivirus infection.

Furthermore, because proinflammatory cytokine production in DENV infection was found associated with disease severity in DENV-infected patients (11, 12), we next determined whether compound 2d could reduce levels of inflammatory cytokines. TNF-α is the main cytokine that induces vascular leakage in DENV-infected patients, and IL-6 and IL-8 levels were found to be elevated in DHF (13). We measured levels of TNF-α, IL-6, and IL-8 in DENV-2- or ZIKV-infected HEK-293 cells with or without compound 2d treatment. Compound 2d dose dependently reduced the production of TNF-α, IL-6, and IL-8 during infection with DENV-2 (Fig. S4A to C) and ZIKV (Fig. S4D to F), which suggests that it has potential for severe dengue diseases such as DHF and DSS. We also demonstrated that exogenous IL-6 addition could dose dependently diminish the antiviral effect of compound 2d in DENV-2- or ZIKV-infected HEK-293 cells (Fig. S4G and H), which suggests that the antiviral mechanism of compound 2d is associated with inflammatory responses.

Compound 2d protected against DENV-2 infection in the AG129 mouse model and reduced inflammatory cytokine expression.

To determine whether compound 2d has therapeutic potential in vivo, we challenged AG129 mice with 10⁷ focus-forming units (FFUs) of DENV-2 or 1 FFU of ZIKV and then orally inoculated them with 1 or 5 mg/kg compound 2d once daily for 10 days. The survival rates with compound 2d were 33% and 66% with 1 and 5 mg/kg, respectively, in DENV-2-infected AG129 mice (Fig. 5A). However, all groups of AG129 mice died within 15 days when challenged with only 1 FFU of ZIKV, likely because the ZIKV PRVABC59 strain is highly virulent in immunodeficient mice, as described previously (32) (Fig. 5B). Next, we measured viremia levels of DENV-2 or ZIKV in serum from AG129 mice treated with compound 2d on day 3 postinfection. Viremia levels were dose dependently lower in AG129 mice inoculated with compound 2d (Fig. 5C and D). Collectively, compound 2d could improve survival and reduce viremia levels with DENV-2 and ZIKV infection in vivo.

FIG 5.

Compound 2d had protective efficacy in vivo. Groups of AG129 mice were first challenged with 10⁷ FFU of DENV-2 or 1 FFU of ZIKV and then inoculated orally with 1 or 5 mg/kg compound 2d once daily for 10 days. Serum samples were collected from AG129 mice treated with compound 2d on day 3 postinfection. Survival rates of mice infected with DENV-2 (A) or ZIKV (B) were monitored daily for 30 days. Viremia levels of DENV-2 (C) or ZIKV (D) from serum samples were measured by focus-forming assay. Levels of PGE₂ (E), TNF-α (F), IL-6 (G), and IL-8 (H) from DENV-2-infected serum samples were examined by ELISA. The percent change in viremia/PGE₂/cytokine secretion was compared to that of the PBS control, set to 100%. Data are means ± SD of 3 independent experiments. **, P < 0.01; ***, P< 0.001 (by two-tailed Student's t test).

Furthermore, we detected levels of PGE₂, TNF-α, IL-6, and IL-8 in serum from DENV-2-infected AG129 mice; levels were dose dependently lower in AG129 mice inoculated with compound 2d (Fig. 5E to H), so compound 2d could inhibit DENV-2 replication and reduce cytokine levels in vivo and in vitro.

DISCUSSION

We have no specific antiviral agents against DENV and ZIKV and only symptomatic treatments clinically, so effective anti-DENV and anti-ZIKV drugs are needed to ameliorate disease or reduce disease severity and mortality during viral infection. Inflammatory responses play critical roles in viral virulence and pathogenesis, and in this study, we screened a series of anti-inflammatory compounds (Fig. 1). Compound 2d could effectively inhibit four DENV serotypes and ZIKV in HEK-293 cells and Raw264.7 macrophages (Fig. 2 and Fig. S3 in the supplemental material). In addition, compound 2d suppressed DENV-2 infection in HEK-293 cells with an SI of 14.8 (Table 1 and Fig. S2), whereas the SI values for the antiviral drug ribavirin, the antibiotic doxycycline, and the NSAID mefenamic acid against DENV-2 were 9.3, 3.1, and 4.7, respectively (33, 34). Thus, with the improved SI of compound 2d compared to values for antiviral drugs, compound 2d might be a promising antiviral candidate for further evaluation. Overall, compound 2d could have potent therapeutic efficacy and could be used for combination therapy assessment and application in the future.

Another important issue to consider for drug development against DENV and ZIKV infection is therapeutic safety. The classical NSAIDs acetaminophen and ibuprofen were prohibited in DENV-infected patients by the WHO (23, 24). In this study, we found that there is no virucidal effect of compound 2d against DENV-2 or ZIKV (Fig. 3). Moreover, compound 2d could directly and selectively inhibit COX-2 enzymatic activity (Fig. 4H and I), which is associated with viral replication, without suppressing COX-1 enzymatic activity and thus prevent the bleeding side effect (Fig. 4J and K). Moreover, compound 2d significantly reduced DENV-2 and ZIKV infection in vitro (Fig. 4F and G) and in vivo (Fig. 5A, C, and D), likely because compound 2d decreased PGE₂ levels more (Fig. 4E and 5E). Furthermore, there was a greater reduction in PGE₂ levels with compound 2d than with compound 1h, 2j, or 2l in DENV-2-infected HEK-293 cells (Fig. S1A) and Raw264.7 cells (Fig. S1B). These findings imply that anti-inflammatory activities may be highly associated with antiviral abilities (compare Fig. 1C and D with Fig. S1A and B; also compare Fig. 4E and F and Fig. 5A and E) and provide an important perspective for drug treatment against flaviviruses.

A previous study showed that different doses and dosing regimens could affect antiviral efficacy in vivo (35); thus, to increase survival rates in the ZIKV-infected AG129 mice (Fig. 5B), the optimal doses and dosing regimens of compound 2d and the optimal challenge doses of ZIKV in AG129 mice will need to be further investigated. In addition, this previous study also demonstrated that viremia profiles peaked on day 3 after infection, but levels of proinflammatory cytokines, including TNF-α and IL-6, increased from day 3 postinfection in DENV-2-infected AG129 mice (35). Therefore, in our study, we collected serum samples on day 3 postinfection, which revealed that treatment with compound 2d could significantly reduce viremia and inflammatory cytokine levels (Fig. 5C, F, G, and H). However, the interplay between viremia and inflammatory cytokine levels needs further detailed study by multiple comparisons of serum samples collected at different times with various doses and dosing regimens of compound 2d in DENV-2-infected AG129 mice.

TNF-α, IL-6, and IL-8 expression is highly related to the pathogenesis of DHF/DSS (36). In this study, we found that compound 2d could inhibit TNF-α, IL-6, and IL-8 expression in vitro (Fig. S4A to F) and in vivo (Fig. 5F to H). In addition, exogenous IL-6 addition could restore viral protein expression in DENV-2- or ZIKV-infected HEK-293 cells (Fig. S4G and H). These findings give another perspective on the use of compound 2d in severe dengue cases. However, because antibody-dependent enhancement of DENV infection is complicated, whether compound 2d could reduce antibody-dependent enhancement is another important issue.

The link between COX-2/PGE₂ expression and DENV infection was previously described (16, 27); however, COX-2/PGE₂ expression has differential effects in various cell lines (29), and all DENV-2-infected ICR suckling mice treated with the COX-2 inhibitor NS398 died within 11 days although survival was 60% at 6 dpi (16, 27). Thus, in this study, because macrophages are the main target of DENV infection, we used Raw264.7 macrophages (31) and the DENV-infected AG129 mouse model to show elevated COX-2/PGE₂ expression with DENV-2 infection; treatment with the anti-inflammatory compound 2d could significantly reduce PGE-2 expression (Fig. 5E and Fig. S1B) and protect the AG129 mouse against DENV-2 challenge for up to 30 days after infection (Fig. 5A). Moreover, viremia levels could be reduced in DENV-2- or ZIKV-infected AG129 mice inoculated with compound 2d (Fig. 5C and D). However, the detailed antiviral mechanisms of compound 2d could be further elucidated in vitro and in vivo.

In conclusion, we demonstrated that compound 2d displays therapeutic efficacy and safety in virus infection and could be a promising antiviral candidate for medically important flaviviruses.

MATERIALS AND METHODS

Cell lines, viruses, and drugs.

BHK-21 cells, the baby hamster kidney cell line (ATCC CCL-10), and C6/36 cells, the mosquito cell line (ATCC CRL-1660), were cultured in RPMI 1640 medium containing 5% fetal bovine serum (FBS; HyClone). Vero cells, the African green monkey kidney cell line (ATCC CCL-81), were grown in minimal essential medium (MEM) containing 10% FBS. HEK-293 cells, the human embryonic kidney cell line (ATCC CRL-1573), and Raw264.7 cells, the mouse macrophage cell line (ATCC TIB-71), were grown in Dulbecco modified Eagle medium (DMEM) containing 10% FBS. Four serotypes of DENV (DENV-1, Hawaii strain; DENV-2, 16681 strain; DENV-3, H87 strain; DENV-4, H241 strain) were propagated in C6/36 cells, and viral titers were determined by focus-forming assay in BHK-21 cells as described previously (37). The ZIKV PRVABC59 strain was amplified in C6/36 cells, and viral titers were measured by plaque-forming assays in Vero cells as described previously (38, 39). The anti-inflammatory compounds 1h, 2d, 2j, and 2l were synthesized as described previously (30) and named as follows: 2-amino-4-(3,4,5-trimethoxyphenyl)-4H-benzo[h]chromene-3-carbonitrile, compound 1h; 2,4-diamino-5-(1-hydroxynaphthalen-2-yl)-5H-chromeno[2,3-b]pyridine-3 carbonitrile, 2d; 9,11-diamino-12-(4-(dimethylamino)-2-hydroxyphenyl)-12H-benzo[5,6]chromeno[2,3-b]pyridine-10-carbonitrile, 2j; and 9,11-diamino-12-(3-hydroxynaphthalen-2-yl)-12H-benzo[5,6]chromeno[2,3-b]pyridine-10-carbonitrile, 2l. The 95% purity of these compounds was assessed by using high-performance liquid chromatography with UV and mass spectrometry (HPLC-UV/MS). Aspirin (COX-1 inhibitor) and celecoxib (COX-2 inhibitor) were from Sigma-Aldrich. Compounds were dissolved in dimethyl sulfoxide (DMSO) and stored at −20°C until use.

Drug cytotoxicity assay.

HEK-293 or Raw264.7 cells were treated with doses of the compounds 1h, 2d, 2j, and 2l for 48 h and analyzed by using the cell proliferation reagent WST-1 (Roche). Sample absorbance from WST was determined by use of an enzyme-linked immunosorbent assay (ELISA) reader (Molecular Devices) at 450 and 490 nm.

Western blot analysis.

To investigate the antiviral effects of compounds 1h, 2d, 2j, and 2l and celecoxib, HEK-293 or Raw264.7 cells were infected with DENV-1, -2, -3, or -4 or with ZIKV, with or without compound treatment. After infection for 48 h, cell lysates were analyzed by Western blot analysis with the antibodies anti-DENV NS3 (GeneTex), anti-DENV E (GeneTex), anti-ZIKV E (GeneTex), anti-COX-1 (ARG23725), anti-COX-2 (abcam), anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Millipore), and anti-actin (Millipore). Then, membranes were probed with the secondary antibody horseradish peroxidase-conjugated goat anti-mouse IgG antibody (Jackson ImmunoResearch). The signals were developed by enhanced chemiluminescence (Millipore) and photographed using a Luminescent Image Analyzer (LAS-3000; Fujifilm).

Immunofluorescence assay.

To investigate the antiviral effects of compound 2d, HEK-293 cells were infected with DENV-1, -2, -3, or -4 or with ZIKV, with or without the compound at the indicated dose (Fig. 2B). Cells were fixed by using an acetone-methanol mixture for 5 min, stained with anti-DENV-2 NS3 or anti-ZIKV E antibody, and incubated overnight at 4°C. After cells were washed with phosphate-buffered saline (PBS), they were stained with Alexa Fluor 594-conjugated goat anti-mouse IgG (Thermo). Cells were then covered with 4′,6′-diamidino-2-phenylindole (DAPI; 300 nM) for nuclear staining. Images were captured under an inverted fluorescence microscope.

Virucidal activity assay.

Virucidal activity of compound 2d was investigated by incubating DENV-2 (5 × 10⁵ PFU) or ZIKV (1 × 10⁵ PFU) with an equal volume of compound 2d from 5 to 30 μM for 1 h at 37°C in the culture medium. Next, samples were diluted and tested for virus survival by plaque assay in BHK-21 cells for DENV-2 and in Vero cells for ZIKV.

Quantitative reverse transcription-PCR (RT-qPCR).

To examine COX-1 and COX-2 expression, HEK-293 cells were infected with DENV-2 or ZIKV with or without treatment with compound 2d and celecoxib. Total cellular RNA was extracted by using a GeneJET RNA purification kit (Thermo) according to the manufacturer’s instructions and transcribed to cDNA with a Maxima H Minus First Strand cDNA Synthesis kit (Thermo). cDNA samples were analyzed by using a Maxima SYBR green/ROX quantitative PCR (qPCR) master mix (Thermo). All samples were run in triplicate on a StepOne Real-Time PCR System (Applied Biosystems). The primer pairs for qPCR were as follows: human COX-1 sense, 5′-CTCTTGCGGTACTCATTGAAG-3′, and antisense, 5′-GAGCTGCTGTTCGGTGTC-3′; human COX-2 sense, 5′-CCCTTGGGTGTCAAAGGTAA-3′, and antisense, 5′-GCCCTCGCTTATGATCTGTC-3′; human GAPDH sense, 5′-GTCTTCACCACCATGGAGAA-3′, and antisense, 5′-ATGGCATGGACTGTGGTCAT-3′.

COX activity assay.

COX-1 and COX-2 activity assays were performed using a COX activity assay kit (Cayman Chemical) according to the manufacturer’s instructions. HEK-293 cells were infected with DENV-2 or ZIKV at a multiplicity of infection (MOI) of 1 with or without compound 2d or celecoxib or aspirin for 24 h. Then, HEK-293 cells were lysed by sonication in a cold buffer (0.1 M Tris-HCl, pH 7.8, containing 1 mM EDTA), and after centrifugation at 10,000 × g for 15 min at 4°C, the supernatants were kept on ice for detection of COX-1 and COX-2 activity. The peroxidase activities of COX were measured colorimetrically by monitoring the appearance of oxidized N,N,N′,N′-tetramethyl-p-phenylenediamine (TMPD) at 590 nm. The specific enzyme activities were calculated using isozyme-specific inhibitors included in the assay kit for distinguishing COX-1 activity from COX-2 activity.

Cytokine quantification by ELISA.

ELISA cytokine kits were used according to the manufacturer’s instructions (all, R&D Systems, CA, USA) to measure levels of TNF-α, IL-6, IL-8, and PGE₂ in cell supernatants from DENV-2- or ZIKV-infected cells with or without treatment with compound 1h, 2d, 2j, or 2l or celecoxib. The absorbance was measured by using an ELISA reader at 450 nm.

Mouse model.

To examine the protective efficacy of compound 2d against DENV-2 or ZIKV in vivo, AG129 mice were divided into three groups for treatment, as follows (n = 6 per group): intraperitoneal (i.p.) injection with 10⁷ focus-forming units (FFU) of DENV-2 or 1 FFU of ZIKV and PBS (vehicle control); i.p. injection with 10⁷ FFU of DENV-2 or 1 FFU of ZIKV and then oral inoculation with 1 mg/kg compound 2d; i.p. injection with 10⁷ FFU of DENV-2 or 1 FFU of ZIKV and then oral inoculation with 5 mg/kg compound 2d. Thereafter, once a day, mice received the same dose for up to 10 days. The survival of these mice was monitored daily for 30 days. For viremia, PGE₂, and cytokine detection, serum samples were collected on day 3 postinfection. Then, viral titers were determined by focus-forming assay, and levels of serum PGE₂, TNF-α, IL-6, and IL-8 were detected by ELISA as described above.

Ethics statement.

All animal experiments were carried out according to the guidelines outlined by Council of Agriculture, Executive Yuan, Republic of China, in adherence with the Declaration of Helsinki and U.S. National Institutes of Health guidelines for the treatment of laboratory animals. The animal protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of the National Defense Medical Center (permit no. IACUC-18-158).

Statistical analysis.

GraphPad Prism, version 5.0 (GraphPad Software, San Diego, CA), was used for data analysis. The data were analyzed by using an unpaired t test and are presented as means ± standard deviations (SD). Survival curves were analyzed by a log rank test. A P value of <0.05 was considered statistically significant.