A term “bone-breaking fever” is used in Chinese medicine to describe the symptoms of patients infected with dengue virus (DV). We examined the significance of the COX-prostaglandin pathway in human DC infected by DV. We show that DV infection induced the expression of COX-2 and the production of prostaglandin E2 (PGE2) in DC, and stimulated the DNA binding of NF-κB and the kinase activity of both IκBα kinase (IKK) α and β. DV infection also activated MAPK and AP-1 signaling. Both IκBα kinase-NF-κB and MAPK-AP-1 were upstream of COX-2 activation. Our investigation into the significance of COX-2-PGE2 pathway also revealed that DV infection enhances DC migration by inducing CC chemokine receptor 7 (CCR7) expression, and that blocking COX-2 or MAPK activity suppresses DV-induced DC migration. Our data also suggest that PGE2 can induce CCR7 expression on DC and that antagonists of the PGE2 receptors EP2 and EP4 suppress DV-induced DC migration. We further show that the increased CCR7 expression was observed in both DV-infected and bystander DC, suggesting the presence of secondary effects in inducing CCR7 expression. Collectively, this study reveals not only the pathways involved in COX-2 synthesis in DV-infected DC but also the autocrine action of PGE2 on the migration of DV-infected DC.
Dengue viruses (DV) are members of the mosquito-borne flaviviruses that infect humans and cause dengue fever with potential severe outcomes such as dengue hemorrhagic fever and dengue shock syndrome 1, 2. DV infection occurs in most tropical and subtropical areas of the world and has become one of the major concerns for those traveling to these areas 3. Epidemiological analysis reveals that worldwide the estimated occurrence is around 50–100 million cases for dengue fever yearly and for dengue hemorrhagic fever is about 250–500 thousand cases 4. Efforts aiming to developing a vaccine have not yet been successful 5. This may be due to the complicated mechanisms of immunopathogenesis 6, 7. Interestingly, although the members of flaviviruses vary, a recent observation by Moran et al.8 shows that a subset of residues within an epitope bound to a class II molecule are cross-reactively recognized by CD4+ TCR.
The term “bone-breaking fever” has long been used in Chinese medicine to describe the symptoms of DV-infected patients. Accordingly, the activation of COX-2-related signaling pathways may be involved in the pathogenesis of DV infection and thus contributes to the clinical manifestations. One of the products of this pathway, prostaglandin E2 (PGE2), has been shown to modulate inflammatory reactions 9 and regulate viral replication 10. Given the important roles of COX-2-PGE2 in many inflammatory responses on different host cells, it is surprising that the induction of COX-2-PGE2 has not been consistently observed in experimental viral infections 11–13.
The most probable natural targets for DV in humans are immature DC and Langerhans cells 14, 15. DC serve as the best professional antigen-presenting cells, playing critical roles in both innate and adaptive immune responses. After infection, DV activates DC to induce cytokine production, metalloproteinase production and cell maturation 15–17. Both maturation and activation processes drive the migration of DC to lymphoid tissues where the viral antigens are presented to T cells. The interaction between DC and T cells activates T cells and induces the production of both Th1 and Th2 cytokines 18. In this study, we investigated whether DV infection activates COX-2 and to what extent this event is significant in human monocyte-derived DC.
DV-induced COX-2-PGE2 expression
DC were established by culturing purified CD14+ monocytes from human peripheral blood with IL-4 and GM-CSF for 5–7 days as described previously 19. After mock or DV infection for 12, 24 or 32 h, the cells were collected and analyzed for COX-2 expression. In Fig. 1A, the increased expression of COX-2 mRNA is shown in DV-infected DC. DV infection also induced COX-2 protein expression (Fig. 1B). The UV-pretreated DV failed to induce COX-2 mRNA and protein expression (Fig. 1A and B). Immunocytochemistry studies revealed that DV infection induced COX-2 expression in individual DC with signal intensity comparable to LPS-stimulated DC (Fig. 1C). Following COX-2 activation, PGE2 production was also induced after viable but not UV-pretreated DV infection (Fig. 1D). Among other COX-2-derived products, while PGD2 was under detection limit, there was no significance difference of PGI2 production between mock and DV-infected DC (data not shown). Furthermore, different DV2 strains, including New Guinea C (NGC), 16681 and PL046 strains at MOI 1, all effectively induced the expression of COX-2 in DC (Fig. 1E). Consistently, the infection of DC by both DV2 strains 16681 and PL046 effectively induced the production of PGE2 (Supporting Information Fig. 1).
DV-induced COX-2 expression is mediated by upregulation of NF-κB
One of the major transcription factors that regulates the expression of COX-2 is NF-κB 20. We asked if the activity of NF-κB in DV-infected DC was increased. As shown in Fig. 2A, DV infection increased DNA-binding activity of NF-κB. By competition and supershift mobility assays, we confirmed that the shifted band was indeed NF-κB specific and both p65 and p50 were in the complex (Fig. 2B). DV infection induced p65 accumulation in the nucleus of DC (Fig. 2C). Although we were not able to show marked reduction of IκBα in DV-infected DC compared with mock-infected cells (data not shown), DV infection effectively induced kinase activities of both IκBα kinase (IKK) α and β (Fig. 2D). When NF-κB activation was blocked by the p50 antagonist peptide SN50 the DV-induced COX-2 activity was greatly suppressed (Fig. 2E). Furthermore, the pre-treatment with an inhibitor of IκBα phosphorylation (E)-3-(4-methylphenylsulfonyl)-2-propenenitrile (BAY-11-7082) also successfully reduced DV-induced COX-2 expression (Fig. 2F).
DV-activated MAPK-AP-1 pathway, upstream of COX-2
We determined whether DV infection could activate AP-1 signaling pathway, a mechanism critical in inflammatory reactions. We show that DV infection increased DNA-binding activity of AP-1 (Fig. 3A). The upstream MAP kinases that regulate the activation of AP-1, including JNK, ERK and p38 were also induced after DV infection (Fig. 3B). The pre-treatment with JNK inhibitor SP600125, p38 inhibitor SB203580 or ERK inhibitor PD98059, effectively reduced the expression of COX-2 in DV-infected DC in a dose-dependent manner (Fig. 3C–E).
DV infection-induced DC migration was COX-2 dependent
Infection of DC by DV causes cell maturation and probably enhances cell migration to lymphoid organs for interaction with T cells 15, 18. The significance of activation of COX-2-PGE2 signaling pathway in DC migration was examined. As shown in Fig. 4A, after infection by DV, significant numbers of virus-infected DC migrated from the upper to the lower chamber containing the chemokine MIP-3β. PGE2 treatment alone mildly increased DC migration and the effect was additionally enhanced after DV infection (Fig. 4B). Pre-treatment with the COX-2 inhibitor celecoxib resulted in suppression of DV-induced DC migration (Fig. 4C). Similar observations were demonstrated examining another chemokine, CCL21 (Fig. 4D). Under the concentrations examined, although the highest concentration of celecoxib only marginally reduced cell viability (Supporting Information Fig. 2), cell migration was largely inhibited (Fig. 4C and D). Furthermore, all MAPK inhibitors, at non-cytotoxic dosages, effectively suppressed DV-induced DC migration toward different chemokines (Fig. 4E and F). In contrast, the JAK inhibitor AG490 had no significant effect. As shown in the Supporting Information Fig. 3, all MAPK inhibitors and celecoxib suppressed DV-induced PGE2 production to basal levels.
DV-induced CC chemokine receptor 7 expression that was susceptible to COX-2 inhibition
Enhanced DC migration after DV infection by COX-2 and MAPK inhibitors suggests that CC chemokine receptor 7 (CCR7), the receptor for both MIP-3β and CCL21, may be induced on DC. As shown in Fig. 5A, the expression of CCR7 was indeed induced after DV infection. In addition, the results show that the pre-incubation of DC with anti-CCR7 antibody effectively suppressed DV-induced DC migration toward MIP-3β and CCL21 (Fig. 5B). Given that the suppressive effect by anti-CCR7 antibody treatment appeared to be not absolute, additional mechanisms might be involved in DV-induced DC migration. We examined whether the treatment with COX-2 inhibitors may affect DV-induced CCR7 expression. The pooled data examining six different donor DC suggest that both COX-2 inhibitors celecoxib and NS398 could suppress DV-induced CCR7 expression on DC (Fig. 5C).
PGE2 through receptors EP2 and EP4 enhanced CCR7 expression in both DV-infected and bystander DC
We investigated the role of PGE2 in DV-induced CCR7 expression. DC were mock or DV infected and exogenously PGE2 was added. As shown in Fig. 6A, PGE2 could effectively enhance CCR7 expression on mock-infected DC and the DV-induced CCR7 expression was further enhanced. To further confirm the effect of PGE2 on DV-induced DC migration, DC were treated with PGE2 receptor EP2 or EP4 antagonist or in combination after DV infection and migration assays were performed. We observed that the DV-induced DC migration was successfully blocked by PGE2 receptor EP2 or EP4 antagonist and there was an additive effect by blocking both EP2 and EP4 receptors (Fig. 6B and C). The relatively weaker suppressive effects by PGE2 receptor antagonists compared with that of COX-2 inhibitors suggest that other COX-2-derived products might also be involved in DV-induced DC migration. The previous studies demonstrate different fates of virus-infected DC and bystander cells after DV infection 21. We thus determined which population of cells after DV infection expressed CCR7 molecules. The results (Fig. 6D) suggest that the increased CCR7 expression was observed in both DV-infected DC and bystander DC.
With evidence of COX-2 induction and the related signaling cascades in DV-infected DC, we screened DV-susceptible cell lines and observed increased expression of COX-2 after DV infection in a lung cancer cell line A549 but not in a hepatoma cell line HepG2 22. Such an observation raises the concern of studying immortalized cells and clearly strengthens the importance of examining primary cells in studying the consequences of DV infection. By different approaches, we demonstrate that DV infection resulted in activation of COX-2 and led to the production of PGE2. The mechanisms involved in the activation of COX-2 included the activation of IKK-NF-κB and MAPK-AP-1 signaling pathways. Through the activation of COX-2 and the secreted PGE2, DV infection enhanced DC migration and CCR7 expression, both of which could be suppressed by COX-2 inhibitions and PGE2 receptor antagonists. Blocking MAPK activities also suppressed DV-induced DC migration. Furthermore, we showed that the increased expression of CCR7 was observed in both DV-infected DC and bystander DC. These results thus illustrate the signaling pathways participating in COX-2 synthesis in DV-infected DC as well as the autocrine action of PGE2 on the migration of DV-infected DC.
Although in most situations both IKKα and IKKβ preserve similar activity in response to different stimuli, a few reports indicate the potential difference between these two kinases in some physiological and pathological conditions 23. In our study, we showed that the protein complex-binding κB site in DV-infected DC contained at least both p65 and p50. These results define the involvement of IKKs-NF-κB-COX-2 signaling in DV infection of human DC.
The understanding of signaling cascades leading to COX-2 activation is important for both the disease immunopathogenesis and the therapeutic approaches. Several COX-2 upstream signaling molecules have been suggested as useful regimens for treatment or control of virus infection. For example, the blockade of ERK upstream MKK1/2 kinase activity causes suppression of virus infectivity by inhibiting early gene expression in human cytomegalovirus infection 24. In addition, the suppression of p38 is shown to inhibit human cytomegalovirus DNA replication and late gene expression 25. The blockade of JNK serves as a potential therapy for the prevention of persistent infection in SARS-CoV-infected cells 26 as well as for the treatment of varicella-zoster virus and hepatitis C virus infection 27, 28. A report from Caposio et al. 29 demonstrates that the replication of human cytomegalovirus in endothelial cells can be inhibited by suppressing IKKβ activity. It is therefore anticipated that targeted suppression of molecules involved in sequential signaling cascades of virus infection may provide a powerful therapeutic option in preventing subsequent detrimental events after DV infection.
Fully functional DC require their interaction with other immune effector cells after processing viral antigens 30. The first step necessary for DC to accomplish this process is to migrate from peripheral tissues to lymphoid organs 31. In this study, we demonstrate that DV infection and exogenously added PGE2 enhanced DC migration, a process likely to be mediated through enhanced expression of CCR7, a receptor molecule essential for DC migration to the T-cell area of draining lymph nodes 32, 33. Importantly, such a migration process appeared to be dependent on fully functional COX-2-PGE2 and MAPK signaling pathways. This is different from the example of vaccinia virus infection where DC migration is suppressed 34. The finding that PGE2 was capable of inducing CCR7 expression is consistent with the other reports 35, 36. It is surprising that although CCR7 was greatly induced in the presence of PGE2 in mock-infected DC, PGE2 at 1 μg/mL (2.84 μM) only marginally enhanced DC migration. According to Scandella et al.35, soluble CD40 ligand- and polyI:C-treated DC express comparable levels of CCR7 as cytokine cocktail does but gain only marginal effect on enhancing DC migration. The evident enhancing effect of PGE2 on migration of DV-infected DC compared with that of mock-infected DC may be related to the maturation status of the cells 15, 35. Given that both EP2 and EP4 receptor antagonists were capable of suppressing DV-induced DC migration, our results were also in agreement with other studies examining the receptors participating in PGE2-triggered migration of monocyte-derived DC 36.
The increased expression of CCR7 could be detected in both the DV-infected DC and the bystander cells (Fig. 6D). As DV infection induces secretion of many different cytokines 15, 21, 37, it is possible that the induction of CCR7 expression on both virus-infected and bystander DC is different; for example, the increased expression of CCR7 on bystander DC might be part of the effects induced by secondarily secreted cytokines from infected cells as suggested by Palmer et al. 21. Further studies on this aspect may provide useful messages in clarifying the different roles of virus-infected and bystander cells in DV-mediated immunopathogenesis.
Materials and methods
Culture medium and reagents
The culture medium consisted of RPMI 1640 (Gibco-BRL) supplemented with 10% fetal bovine serum, 2 mM glutamine and 1000 U/mL penicillin–streptomycin. Recombinant GM-CSF and IL-4 were purchased from R&D. The COX-2 inhibitors celecoxib and NS398 were obtained from Pfizer and Calbiochem, respectively. JNK, p38, ERK, IκBα phosphorylation, JAK and p50 inhibitors were purchased from Calbiochem. The Ab against p65, p50, IκBα or upstream stimulatory factor-2 was purchased from Cell Signaling or Santa Cruz. Anti-CCR7 Ab and chemoattractants, including MIP-3β and CCL21 were purchased from R&D. PGE2 was purchased from Cayman. Both PGE2 receptor EP2 and EP4 antagonists were purchased from Sigma-Aldrich. Unless specified, the rest of reagents were purchased from Sigma-Aldrich.
Establishment of DC
DC were established from purified CD14+ monocytes as described previously 19. In brief, buffy coat (equivalent to 500 mL of whole blood for each) from blood bank was mixed with Ficoll-Hypaque, after centrifugation, the layer of PBMC was collected. PBMC were then incubated with anti-CD14 microbeads at 4–8°C for 15 min. After wash, the CD14+ cells were isolated using a MACS cell isolation column (Miltenyi Biotech). The purified monocytes were cultured in the medium containing 800 U/mL GM-CSF and 500 U/mL IL-4 at a cell density of 1×106 cells/mL. The culture medium was changed every other day with 300 μL of fresh medium containing 2400 U GM-CSF and 1500 U IL-4 and the cells after 5–7 days of culture with purity higher than 92% were used for experiments 19.
The preparation of DV has been described previously 19. In brief, DV2 New Guinea C (NGC, ATCC), a WT local Taiwanese strain PL046 and 16681 strains, were propagated in C6/36 mosquito cells in RPMI containing 5% heat-inactivated FBS and maintained at 28°C for 7 days. To prepare mock-infected supernatants, all of the procedures were identical except that buffered saline was substituted for the virus inoculation. The virus titers in supernatants were determined by plaque-forming assays and stored at −70°C until use 19.
DC were infected with mock or DV at different MOI for 4 h at 37°C 19. After viral adsorption, cells were then washed and cultured in 6-well plates (Costar) with culture medium in the absence of exogenously added cytokines.
Total RNA was isolated after lysing cells by Trizol (Gibco). After reverse transcription of RNA to cDNA, samples were subjected to PCR reactions. The consensus primers for COX-2 were 5′-TTCAAATGAGATTGTGGGAAAATTGCT-3′ and 5′-AGATCATCTCTGCCTGAGTATCTT-3′; for β-actin were 5′-ATGGCCACGGCTGCTTCCAGC-3′ and 5′-CATGGTGGTGCCGCCAGA-3′.
After treatment, DC suspensions were adhered onto a polylysine-coated slide and cells were fixed by immersing in 4% paraformaldehyde for 20 min. Cells were permeablized by adding 1% v/v Triton X-100 for 20 min and then 1% BSA was added to block non-specific binding. Subsequently, the anti-COX-2 Ab (1:100, Santa Cruz) was added and incubated for 2 h. After wash with PBS, biotinyl-donkey-anti-goat Ab (1:200, Jackson Laboratory) was added and incubated for 1 h. The cells were stained with streptavidin–phycoerythrin (1:200, BD Pharmingen) for another 1 h. Finally, after adding 20 μL of mounting oil, DC were covered with cover slips and visualized under fluorescence microscope.
Flow cytometry analysis
The determination of single expression or co-expression of both cell surface (CCR7) and intracellular molecules (viral E protein) has been described 19. For dual stainings, the anti-CCR7 mAb conjugated with phycoerythrin was added to DC suspensions and the mixture was incubated at 4°C for 30 min. Cells were then permeablized with 0.25% saponin (Sigma). After incubation for another 20 min, the anti-DV E protein Ab was added. After wash, the goat anti-mouse Ab conjugated with fluorescein isothiocyanate was added and incubated for another 30 min. Finally, the samples were analyzed by flow cytometry.
Nuclear extract preparation
Nuclear extracts were prepared as described previously 19. Briefly, the treated cells were left at 4°C in 50 μL of buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl2, 1 mM DTT, 1 mM PMSF and 3.3 μg/mL aprotinin) for 15 min with occasional gentle vortexing. Swollen cells were centrifuged at 15 000 rpm for 3 min. After removal of the supernatants (cytoplasmic extracts), the pelleted nuclei were washed with 50 μL buffer A and subsequently cell pellets were resuspended in 30 μL buffer C (20 mM HEPES, pH 7.9, 420 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 25% glycerol, 1 mM DTT, 0.5 mM PMSF and 3.3 μg/mL aprotinin) and incubated at 4°C for 30 min with occasional vigorous vortexing. The mixtures were centrifuged at 15 000 rpm for 20 min, and the supernatants were used as nuclear extracts.
The EMSA was performed as described previously 19. The oligonucleotides containing NF-κB and AP-1 binding sites were purchased and used as DNA probes (Promega). The probes were radiolabeled with [γ-32p]ATP using T4 kinase (Promega). For the binding reaction, the radiolabeled NF-κB or AP-1 probe was incubated with 5 μg of nuclear extracts. The binding buffer contained 10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 0.5 mM EDTA, 1 mM DTT, 1 mM MgCl2, 4% glycerol and 2 μg poly(dI-dC). Binding reaction proceeded for 20 min at room temperature. If un-radiolabeled competitive oligonucleotides or mAb for supershift assays were added, they were preincubated with nuclear extracts for 30 min before adding radiolabeled probes.
ECL western blotting (Amersham) was performed as described previously 19. Briefly, equal amounts of proteins were analyzed on 10% SDS-PAGE and transferred to the nitrocellulose filter. For immunoblotting, the nitrocellulose filter was incubated with non-fat milk buffer for 1 h, and then blotted with antisera against individual proteins overnight. After washing with milk buffer, the filter was incubated with secondary Ab conjugated to horseradish peroxidase for 1 h. The filter was then incubated with the substrate and exposed to an x-ray film.
Immunoprecipitation kinase assay
The immunoprecipitation kinase assay was performed as described previously 38. The GST-IκBα and GST-c-Jun fusion proteins were used as substrates for both IKKα and IKKβ and JNK, respectively. Myelin basic protein, a substrate for ERK and p38, was purchased from Sigma. The Ab for kinase assays were purchased from Cell Signaling (anti-JNK and anti-p38) or Santa Cruz (anti-ERK, anti-IKKα and anti-IKKβ). To perform immunoprecipitation kinase assays, the cellular extract 50–100 μg was mixed with 5 μL of specific Ab in incubation buffer containing 25 mM HEPES (pH 7.7), 300 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 0.1% Triton-X-100, 20 mM β-glycerophosphate, 0.1 mM Na3VO4, 2 μM leupeptin and 400 μM PMSF overnight. After precipitation with protein A beads and subsequent wash, the beads were resuspended in 40 μL kinase buffer containing cold ATP (30 μM), substrates and 10 μCi of [γ-32P]ATP. The mixture was incubated at 30°C with occasional gentle mixing for 30 min. The reaction was then terminated by resuspending in 1% SDS solubilizing buffer and boiled for 5 min and analyzed in SDS-PAGE.
The chemotaxis assays were performed according to the report 39. In brief, DC treated under different conditions as indicated migrated through a polycarbonate filter of 5 μm pore size in 24-well transwell chambers (Corning Costar). In the lower chamber of a transwell cassette, it contained serum free RPMI 600 μL with or without 500 ng/mL MIP-3β or other chemoattractant (R&D). DC 1×105 in 100 μL serum-free medium were loaded in the upper chamber and incubated for 2 h at 37°C. Then cells migrating from the upper chamber to the lower chamber were counted by flow cytometry. The acquired events for a fixed time period of 60 s in a FACScan were determined using CellQuest software.
The results are presented as the mean±SD from at least three independent experiments. Paired Student's t-test was used to determine the difference and p<0.05 was considered as significant.
The authors thank Pfizer for providing the drug compound celecoxib as well as the helpful comments from Dr. Lin YL. This work is supported in part by a grant from the National Health Research Institutes (NHRI-EX95-9408SI), the National Science Council (NSC-96-2314-B-016-015-MY3) and the Chen-Han Foundation for Education, Taiwan.
Conflict of interest: The authors declare no financial or commercial conflict of interest.