Eosinophil-derived neurotoxin acts as an alarmin to activate the TLR2–MyD88 signal pathway in dendritic cells and enhances Th2 immune responses

DC activation is characterized by the development of a mature phenotype (including increased expression of surface costimulatory and MHC molecules, production of proinflammatory cytokines, and conversion to a CCR7⁺ phenotype) and the acquisition of the capacity for antigen presentation (13, 14). We have previously reported that EDN, either purified from a natural source or recombinant, has the same capacity to up-regulate CD80, CD83, and CD86 expression and production of a variety of cytokines by DCs (11). To further determine its effect on DCs, we performed a thorough analysis of the phenotype and function of human monocyte–derived DCs upon treatment with recombinant EDN. The recombinant EDN used in this study displayed a single band on SDS-PAGE gel (not depicted) and did not contain a detectable level of LPS or peptidoglycan (PGN) as measured by the Cambrex QCL-1000 Chromogenic LAL Assay kit (Fig. 1 A) and the Wako SLP Reagent Set (Fig. 1 B), attesting to its purity. The capacity of the same batch of EDN to induce DC IL-6 production was previously shown to be destroyed by boiling (11). In addition, in comparison with EDN, human angiogenin (hAng) expressed in the same system did not induce DC IL-6 production (11). Therefore, the effect of EDN on DCs was unlikely to be due to LPS contamination. As shown by Fig. 1 C, human DCs incubated in the absence of any stimulant (sham-treated) for 48 h were positive for CD11c (CD11c⁺), negative for CD83 (CD83⁻), and expressed a very low level of CD80 (CD80^low) and an intermediate level of CD86 (CD86^int); these features are characteristic of an immature phenotype. DCs treated with EDN, similar to those treated with E. coli LPS, became CD83⁺, CD80^int, and CD86^high, characteristic of surface marker expression of mature DCs. Additionally, treatment of monocyte-derived DCs with 200∼2,000 ng/ml EDN for 48 h (Fig. 1 D, hatched bars) stimulated DC production of IL-6, IL-8, IL-12p70, and TNF-α in a dose-dependent fashion as compared with DCs incubated without a stimulant (Fig. 1 D, open bars). Of note, EDN at 2 μg/ml stimulated similar levels of IL-6, IL-8, and TNF-α, but much less IL-12p70 than 1 μg/ml of LPS (Fig. 1 D). Furthermore, we also examined whether EDN-treated DCs could migrate in response to an agonist for CCR5 (regulated upon activation, normal T cell–expressed, and secreted [RANTES]) or CCR7 (secondary lymphoid tissue chemokine [SLC]). As expected, sham-treated DCs migrated in response to RANTES, but not SLC (Fig. 1 E, white bars), whereas LPS-treated DCs migrated in response to SLC, but not RANTES (Fig. 1 E, hatched bars). DCs incubated with 1 μg/ml EDN for 24 h lost the capacity to migrate to RANTES, but gained the capacity to migrate to SLC, indicative of a switch in functional chemokine receptor expression from CCR5⁺ to CCR7⁺ (Fig. 1 E, solid bars). The ability of stromal cell–derived factor (SDF)-1α to induce the migration of all types of DCs provides a positive control and is in agreement with the reported expression of its receptor, CXCR4, by both immature and mature DCs (14). Thus, EDN induced phenotypic maturation of DCs.

To ensure that EDN-induced maturation of DCs was reflected at the functional level, EDN-treated human DCs were analyzed for their capacity to stimulate the proliferation of T cells in an allogeneic mixed lymphocyte reaction using sham-treated DCs as a control (Fig. 1 F). At the DC/T ratios tested (1:6,250∼1:250), sham-treated DCs did not stimulate proliferation of allogeneic T cells as judged by the failure to increase [³H]–TdR incorporation. In contrast, DCs treated with 1 μg/ml EDN, similarly to LPS (1 μg/ml) -treated DCs, stimulated the proliferation of allogeneic T cells when used at a DC/T ratio higher than 1:6,250, indicating that EDN treatment enhanced the capacity of DCs to stimulate T cell proliferation. Thus, EDN treatment results in full activation of DCs that not only acquire a mature phenotype (Fig. 1, C–E), but also become capable of antigen presentation and T cell stimulation (Fig. 1 F).

Activation of DCs in response to many stimuli, such as cytokines, CD40 ligand, and microbial products, converges in activating intracellular signaling transducers, such as NF-κB and MAPKs (15–18). To determine the effect of EDN, the levels of I-κBα and MAPKs in DCs treated with EDN were determined by Western blot using E. coli LPS as a positive control (Fig. 2). LPS caused down-regulation of I-κBα and activation of JNK, p38, and Erk MAPKs with distinct kinetics (Fig. 2, right), similarly to previous reports (17–19). EDN reduced the level of I-κBα in DCs in a time-dependent manner similarly to LPS, which began to decrease after 5∼10 min of treatment and became evident after 20 min of treatment, indicating the activation of NF-κB by EDN (Fig. 2, left). EDN treatment also activated all three MAPKs, including JNK, p38, and Erks, as shown by an increase in the levels of phosphorylated JNK, p38, and Erks (Fig. 2, left). Of note, EDN-induced activation of DC MAPKs showed different kinetics from those induced by LPS. EDN-induced phosphorylation of JNK only became evident after 20 min of incubation; however, LPS-induced JNK phosphorylation peaked at 5∼10 min and began to decrease thereafter. In addition, EDN induced more rapid phosphorylation of DC Erks (peaked at 1 min) than LPS (peaked at 10 min). Furthermore, the ratio of phosphorylated p44/42/p38 in EDN-treated DCs was relatively higher than that of LPS-treated DCs (Fig. 2).

To identify the mechanism of EDN-induced DC activation, we applied the gripNA transient gene-silencing technique (20) to investigate if blockade of MyD88 in human monocyte–derived DCs would inhibit EDN-induced IL-6 production. To this end, gripNA-hMyD88 was delivered into human monocyte–derived DCs, and the effect of EDN or IL-1α on IL-6 production by the resulting DCs was measured. Compared with DCs treated with Chariot II alone, DCs treated with Chariot II and gripNA^MyD88 produced much less IL-6 in response to EDN, suggesting that MyD88 might be involved in EDN-mediated DC activation (Fig. 3 A). IL-1α–induced IL-6 production in DCs was partially (∼37%) inhibited by gripNA^MyD88 treatment, suggesting incomplete blockade of DC MyD88 by gripNA^MyD88 transfection (Fig. 3 A). To determine whether MyD88 was required for EDN-induced DC activation, DCs derived from MyD88^+/+ and MyD88^−/− mice were used. When DCs generated from the bone marrow hematopoietic progenitor cells (HPCs) of littermate-matched MyD88^+/+ and MyD88^−/− mice were tested for their capacity to produce IL-6 in response to EDN, EDN, like Pam3 and LPS, induced IL-6 production by MyD88^+/+ DCs, but failed to induce IL-6 production by MyD88^−/− DCs, indicating that MyD88 plays a critical role in EDN-induced DC cytokine production (Fig. 3 B).

Among the many diverse DC-activating stimulants, only the IL-1 family of cytokines (including IL-18) and many TLR ligands activate DCs in a MyD88-dependent manner (15, 21). There is no obvious structural or functional resemblance between EDN and cytokines. On the other hand, certain AMPs have previously been reported to act as endogenous TLR ligands, such as the activation of TLR4 by mouse β-defensin-2 (22) and of TLR2 and TLR4 by HMGB1 (23, 24). Therefore, we investigated whether EDN could act as an endogenous ligand for a TLR by determining if EDN could stimulate NF-κB activation in HEK293 cells transiently transfected with a combination of individual TLR and an NF-κB–reporting luciferase construct. Human monocyte–derived DCs in our study expressed TLR1, 2, 3, 4, 5, 6, 7, 8, and 10 mRNAs, suggesting that TLR9 was unlikely to be involved (not depicted). In addition, the HEK293 cells used in our study, although expressing high levels of TLR5 and 6 mRNAs, and low levels of TLR1, 7, 8, and 10 mRNAs, did not show NF-κB activation in response to EDN (not depicted), suggesting that EDN did not activate TLR1, 5, 6, 7, 8, and 10. When HEK293 cells transiently transfected with a NF-κB–reporting luciferase construct together with plasmids expressing TLR2, TLR3, or a combination of TLR4 and MD-2 were measured for the luciferase activity in response to EDN, EDN induced an increase in luciferase activity only in cells expressing TLR2, but not in those expressing either TLR3 or TLR4, indicating that EDN interacted with TLR2 to activate NF-κB (Fig. 4 A). As expected, HEK293 cells expressing a particular TLR increased their luciferase activity in response to the corresponding agonists.

To confirm the requirement of TLR2 for EDN-induced activation of DCs, we compared EDN-induced production of IL-6 by DCs derived from the bone marrow progenitors of TLR2^−/−, TLR2^+/+, C3H/HeN, and C3H/HeJ (harboring a point mutation that renders TLR4 nonfunctional) mice (Fig. 4 B). TLR2^+/+ DCs produced considerable levels of IL-6 in response to EDN, Pam3, and LPS. However, TLR2^−/− DCs failed to produce IL-6 in response to either EDN or Pam3, a known TLR2 ligand, but did produce IL-6 in response to E. coli LPS, a TLR4 ligand. Additionally, EDN induced a similar amount of IL-6 production by both C3H/HeN and C3H/HeJ DCs, suggesting that EDN does not trigger TLR4. E. coli LPS induction of similar levels of IL-6 from all but C3H/HeJ DCs ensured that C3H/HeJ DCs used in these experiments had no functional TLR4. These data clearly demonstrate that EDN is an endogenous ligand of TLR2. This conclusion is supported by the dependence of EDN-induced DC IL-6 production on MyD88 (Fig. 3) because MyD88 is the exclusive signal transducer for TLR2 (15). Furthermore, the observation that EDN induced much less IL-12p70 than LPS (Fig. 1 B) activated Erks MAPK more than p38 MAPK in DCs (Fig. 2) is also consistent with the usage of TLR2 by EDN, as other TLR2 ligands have also been reported to have these properties (19).

TLR2 often heterodimerizes with TLR1 or TLR6 to respond to many TLR2 ligands (21, 25–27). To determine whether EDN activation of DCs was also dependent on TLR1 or TLR6, we assessed whether EDN-induced production of TNF-α and IL-6 by DCs derived from the bone marrow progenitors of TLR1^−/−, TLR2^−/−, TLR6^−/−, and WT C57BL/6 mice using Pam3, a ligand for TLR2/1 (25), fibroblast-stimulating lipopeptide (FSL)-1, a ligand for TLR2/6 (27), and E. coli LPS as controls (Fig. 5). As expected, LPS stimulated the production of similar amounts of TNF-α and IL-6 by all types of DCs. EDN, although capable of inducing comparable amounts of cytokines in WT, TLR1^−/−, and TLR6^−/− DCs, did not induce TNF-α and IL-6 in TLR2^−/− DCs. The TLR1^−/−, TLR2^−/−, and TLR6^−/− DCs showed the expected responses because (a) Pam3 and FSL-1 did not induce TNF-α and IL-6 production in TLR2^−/− DCs, (b) Pam3 induction of TNF-α was partially reduced in TLR1^−/− DCs, and (c) FSL-1–induced TNF-α and IL-6 were greatly reduced in TLR6^−/− DCs. These data demonstrate that EDN acts only on TLR2, not on a heterodimer of TLR2/1 or TLR2/6.

To determine if EDN could enhance antigen-specific immune responses, C57BL/6 mice were immunized with OVA in the absence or presence of EDN on day 1 and boosted on day 14 with OVA alone. The levels of anti-OVA IgG antibody in the serum of mice taken on days 10 and 21 were measured for primary and secondary anti-OVA immune response. EDN dramatically enhanced both the primary (Fig. 6 A) and secondary (Fig. 6 B) anti-OVA antibody responses at both doses, demonstrating the capacity of EDN to promote an antigen-specific immune response. Alum, as a positive control, also enhanced anti-OVA antibody responses as expected (Fig. 6, A and B). hAng, another member of the RNase superfamily, which is unable to induce either migration or maturation of DCs (10, 11), did not enhance the primary or secondary anti-OVA antibody responses (Fig. 6, A and B). Assaying the subclass of anti-OVA IgG antibodies revealed that coimmunization with EDN selectively enhanced the production of OVA-specific IgG1, but not IgG2 or IgG3, during the primary immune response (Fig. 6 C). During the secondary immune response, booster with OVA caused an enhancement of all subclasses of anti-OVA IgG antibodies in EDN-treated mice; however, the OVA-specific IgG1 level was still higher than that of OVA-specific IgG2a, IgG2b, or IgG3 (Fig. 6 D). Alum, as expected, promoted the generation of more anti-OVA IgG1 than IgG2a in both primary and secondary antibody responses (Fig. 6, C and D). Because predominant generation of IgG1 antibody is an indicator of Th2 response, the data suggest that EDN may stimulate Th2 polarization.

To determine whether EDN enhancement of antigen-specific immune response depends on its capacity to activate TLR2 and to evaluate the nature of EDN-enhanced antigen-specific immune response, TLR2 knockout (TLR2^−/−) and littermate-matched control (TLR2^+/+) mice were immunized with OVA in the absence or presence of EDN or LPS, and the subsequent immune responses were measured. Splenocytes from TLR2^+/+ mice immunized with OVA in the presence EDN or LPS incorporated significantly more [³H]-TdR than cells of mice immunized with OVA alone, indicating the enhancement of OVA-specific T cell–proliferative response by EDN or LPS (Fig. 7 A). The splenocytes of TLR2^+/+ mice immunized with OVA in the presence of EDN produced predominantly IL-5, IL-6, IL-10, and IL-13, whereas splenocytes of TLR2^+/+ mice immunized with OVA in the presence of LPS produced predominantly IFN-γ, suggesting that EDN and LPS polarized in vivo OVA-specific Th2 and Th1 T cell responses, respectively (Fig. 7 B). The production of OVA-specific IL-4 was not significantly enhanced by simultaneous in vivo administration of either EDN or LPS (Fig. 7 B). The anti-OVA IgG responses in TLR2^+/+ mice immunized with OVA in the presence of EDN were very similar to those in C57/BL6 mice (Fig. 6), with significant enhancement of total anti-OVA IgG (Fig. 7 C, a and b) and higher levels of IgG1 than IgG2a (Fig. 7 C, c and d) in both primary (Fig. 7 C, a and c) and secondary (Fig. 7 C, b and d) antibody responses. TLR2^+/+ mice immunized with OVA in the presence of LPS also showed enhanced production of anti-OVA IgG (Fig. 7 C, a and b); however, the predominant subclass was IgG2a (Fig. 7 C, c and d), echoing the capacity of LPS to polarize Th1 immune response (Fig. 7 B). Importantly, there was no enhancement of OVA-specific T cell and antibody responses by EDN in TLR2^−/− mice, whereas LPS similarly enhanced OVA-specific T cell and antibody responses in both TLR2^−/− and TLR2^+/+ mice (Fig. 7, A–C). Thus, the capacity of EDN to enhance antigen-specific Th2-polarized immune response was dependent on TLR2.

To determine the role of DC activation in EDN enhancement of antigen-specific Th2 immune response, we investigated whether OVA-loaded DCs that were activated by EDN in vitro could promote OVA-specific Th2 immune response in vivo. To this end, DCs of WT or TLR2^−/− mice were pulsed with OVA, treated with EDN (or E. coli LPS as a control), and injected into the peritoneal cavity of C57BL/6 mice. The mice were immunized i.p. with OVA on day 5 and killed on day 12 for the determination of OVA-specific splenocyte production of cytokines. WT DCs in response to EDN produced a substantial amount of IL-1β, IL-6, IL-10, and TNF-α, a low (nevertheless statistically significant) amount of IL-12p70, and a little IL-23 (Fig. 8 A). LPS enhanced DC production of all cytokines measured, with much less induction of IL-10 and much greater induction of IL-6, IL-12p70, and IL-23 in comparison with EDN treatment (Fig. 8 A). TLR2^−/− DCs did not produce any significant cytokine in response to EDN (Fig. 8 A). The splenocytes of C57BL/6 mice immunized with OVA-loaded and EDN-treated WT DCs generated predominantly IL-5, IL-6, IL-10, and IL-13, but no IFN-γ when stimulated with OVA (Fig. 8 B). In contrast, the splenocytes of mice immunized with LPS-treated DCs predominantly generated IFN-γ, some IL-6 and IL-10, but no IL-5 or IL-13 upon OVA stimulation (Fig. 8 B). The splenocytes of mice immunized with EDN-treated TLR2^−/− DCs did not produce any Th1 or Th2 cytokines (Fig. 8 B). These data, in addition to further strengthening the conclusion that EDN activates DCs by triggering TLR2, also demonstrate that DCs contribute to EDN enhancement of OVA-specific Th2 immune response.

It has been well established that TLRs can recognize infectious “nonself” by sensing pathogen-associated molecular patterns and triggering antimicrobial immune responses of the host (28, 29); however, emerging evidence indicates that certain endogenous molecules can also activate TLRs. Several endogenous ligands have been reported to activate TLR4, such as fibronectin extra domain A (30), fibrinogen (31), lung surfactant protein A (32), heparan sulfate (33), hyaluronan fragments (34, 35), and mouse β-defensin-2 (22). Mammalian RNA and DNA in the form of immune complexes can act as endogenous ligands for TLR7 and TLR9, respectively (36, 37). Although heat shock proteins and HMGB1 have been reported to trigger both TLR2 and TLR4 (23, 24, 38–40), no endogenous ligand acting exclusively on TLR2 has been reported previously. We have shown that EDN is capable of inducing DC maturation, cytokine production, and activation of NF-κB and multiple MAPKs in DCs (Figs. 1 and Figs.2). In addition, EDN induction of NF-κB activation in HEK293 cells is mediated by TLR2, and EDN-induced DC cytokine production is abrogated in TLR2^−/− DCs, but not in TLR1^−/−, TLR4^−/−, or TLR6^−/− DCs (Figs. 4, Figs.5, and Figs.8). Furthermore, EDN enhancement of OVA-specific immune responses also relies on TLR2 (Figs. 7 and Figs.8). Thus, EDN appears to be a TLR2-specific endogenous ligand with the capacity to enhance antigen-specific immune responses.

EDN enhancement of OVA-specific immune responses can be achieved by immunizing mice either with OVA in the presence of EDN (Figs. 6 and Figs.7) or with EDN-activated OVA-loaded DCs (Fig. 8). The antigen-specific immune response selectively promoted by EDN is predominantly of the Th2 type, which is indicated by preferential promotion of antigen-specific IgG1 production in both the primary and secondary antibody responses (Figs. 6 and 7 C), as well as generation of T cells that are capable of secreting elevated levels of IL-5, IL-6, IL-10, and IL-13, but not of IFN-γ or TNF-α (Figs. 7 B and 8 B). This notion is in good agreement with data demonstrating that EDN, in comparison with E. coli LPS, induces much less IL-12p70 and considerably more IL-10 production by both human and mouse DCs (Figs. 1 D and 8 B). Conventional Th2 T cells, such as those generated in response to Schistosoma infection or DCs conditioned with Schistosoma mansoni egg antigen treatment, are characterized by simultaneous production of IL-4, IL-5, IL-10, and IL-13 (41, 42). In contrast, OVA-specific Th2 cells promoted by EDN differ from conventional Th2 cells in lacking production of IL-4 (Figs. 7 B and 8 B). It has also been documented that thymic stromal lymphopoietin can also condition DCs for the induction of Th2 cells that are distinct from conventional Th2 cells in producing IL-4, IL-5, IL-13, and TNF-α, but not IL-10 (43, 44). It thus appears that different stimulants can induce the generation of Th2 with a distinct cytokine profile. It remains to be determined why EDN leads to the generation of Th2 cells with this particular profile of cytokine production.

EDN stimulates DCs to generate an abundant amount of IL-6 and IL-1β (Figs. 1 and Figs.8), and eosinophil can produce TGF-β (45). Because IL-6, IL-1β, and TGF-β are known to contribute to the differentiation of CD4 T cells into Th17 lineage (46), it is conceivable that eosinophils and/or EDN may promote the generation of antigen-specific Th17 cells. However, the splenocytes of mice immunized with EDN and OVA did not produce a significant amount of IL-17 (not depicted). Therefore, EDN could not promote the generation of Th17 cells, at least in the mouse models used in the present study.

EDN uses the TLR2–MyD88 signal transduction pathway for the activation of DCs and enhances OVA-specific Th2-polarized immune responses in a TLR2-dependent manner. Our data are consistent with several previous reports showing that DCs activated by TLR2 agonists exhibit more pronounced Erk activation, produce less IL-12p70, and preferentially induce antigen-specific Th2-polarized immune responses (19, 47–49). Is MyD88 or TLR2 crucial for Th2 polarization? Ample evidence indicates that signaling through MyD88 does not determine Th1/Th2 polarization. For example, MyD88 is not required for Schistosoma mansoni egg antigen–induced Th2 polarization (50). Moreover, although both TLR2 and TLR9 transduce signals exclusively through MyD88 (21, 28, 29), their agonists, PGN and synthetic stimulatory oligodeoxynucleotide, enhance Th2- and Th1-type immune responses, respectively (48). The TLR2 ligands, such as Pam3, PGN, and EDN, upon coadministration with antigen in the primary immunization, promote distinct Th2-type antigen-specific immune responses (19, 47–49, and this study), suggesting that signaling through TLR2 has the potential to promote Th2 polarization under certain conditions. However, TLR2 agonists, such as Pam3, lipoteichic acid, and macrophage-activating lipopeptide 2, are capable of suppressing previously established allergic Th2 responses (51–53). Propionebacterium acnes, albeit activating DCs via TLR2, not only induce a Pa-specific Th1 immune response (54), but also stimulate DCs to generate a negative signal for Th2 development in a MyD88-dependent manner (55). Compared with most other TLR2 agonists, zymosan is reported to induce IL-10–producing antigen-specific Tr1 without generation of either Th1 or Th2 T cells (56). Thus, the effect of TLR2 activation on the polarization of Th responses may be influenced by multiple factors, such as the type of TLR2 ligands and the means and timing of administration, and it is still in need of further investigation.

The recruitment and activation of eosinophils are most often seen in tissues subject to parasitic infection and allergy (7, 41, 57). EDN is likely to be generated and present at sites of inflammation associated with such disorders. EDN is present in the serum and nasal lavage fluid of healthy individuals at ∼20 ng/ml (58, 59), but its concentration in the nasal lavage fluid of patients with allergic rhinitis can reach as high as 880 ng/ml (60), suggesting that EDN in tissues subject to allergic reaction or parasitic infection can reach a concentration (100∼1,000 ng/ml) necessary for the activation of DCs and enhancement of immune response described in the present study. Therefore, the capacity of EDN to activate DCs and to enhance a Th2-polarized immune response may provide a basis for the induction and/or maintenance of a Th2-polarized immune response during parasitic infection, allergy, or atopic reaction. However, a recent report showing that there is no deficit in the generation of IL-4 and IL-5 in eosinophil-ablated mice in response to helminth infection (61) suggests that eosinophil, and consequently eosinophil-generated EDN, may not be required for the generation of Th2-type immune response. How can this report be reconciled with data of the present study? Because EDN can also be produced by cells other than eosinophils, such as neutrophils, macrophages, and certain placental epithelial cells (7, 11, 62), it is possible for EDN to be present at sites of infection even in the absence of eosinophils. It is also possible that factors other than EDN provide the Th2-polarizing signal in the absence of eosinophils.

Alarmins are defined as endogenous mediators rapidly released by cells of the host innate immune system in response to infection and/or tissue injury, which possess the dual activities of recruiting and activating DCs and consequently are capable of enhancing antigen-specific adaptive immune responses (5, 6). EDN is an endogenous mediator rapidly released from eosinophils, neutrophils, and by activated macrophages (7, 11, 12, 62). In addition, EDN can selectively induce the recruitment of DCs to the site of inflammation (10). Furthermore, EDN has DC-activating activity and enhances antigen-specific immune response in vivo (this study). Thus, EDN can be considered a bona fide alarmin that favors an alternative type of Th2 immune response.

Recombinant human and mouse GM-CSF, IL-4, SDF-1α, RANTES, and SLC were purchased from PeproTech. Recombinant EDN was generated and purified as described previously (11). The purity of this batch of EDN was analyzed by SDS-PAGE. The potential levels of contaminating LPS and PGN in the sample were measured using QCL-1000 Chromogenic LAL Assay kit (Cambrex) and SLP Reagent Set (Wako) according to the manufacturers' protocols. hAng was purchased from R&D Systems.

MyD88^−/−, TLR1^−/−, TLR2^−/−, and TLR6^−/− mice generated by S. Akira (Osaka University, Osaka, Japan) (21, 25, 26, 63, 64) were backcrossed eight or more generations onto the C57BL/6 background, and then intercrossed to obtain the knockout genotypes. C57BL/6, C3H/HeN, and C3H/HeJ mice were obtained from Charles River Laboratories. All mice were kept under specific pathogen-free conditions, with water and food given ad libitum. All experiments with mice were performed in compliance with the principles and procedures outlined in the NIH Guide for the Care and Use of Animals and were approved by the NCI-Frederick Animal Care and Use Committee.

Human PBMCs were isolated by Ficoll density gradient centrifugation as described previously (65). Monocytes were purified (>95%) from human PBMCs with MACS CD14 monocyte isolation kit (Miltenyi Biotec). Human T cells were isolated from human PBMCs by using CD3 Enrichment Columns (R&D Systems). Mouse HPCs were isolated as reported previously (66) from the bone marrow of knockout and littermate-matched WT mice.

Human and mouse DCs were generated as described previously (66). In brief, monocytes or mouse bone marrow progenitors were incubated at 2∼5 × 10⁵/ml in G4 medium (RPMI 1640 containing 10% FBS [Hyclone], 2 mM glutamine, 25 mM Hepes, 100 U/ml penicillin, 100 μg/ml streptomycin, 50 ng/ml of human or mouse GM-CSF, and 50 ng/ml of human or mouse IL-4) at 37°C in a CO₂ (5%) incubator for 5∼7 d to generate immature DCs (iDCs), with 50% of the culture medium replaced with prewarmed fresh G4 medium on days 3 and 5. Subsequently, iDCs were incubated in fresh G4 medium in the absence or presence of EDN, Pam3 (N-palmitoyl-S-[2,3-bis(palmitoyloxy)-(2RS)-propyl]-[R]-cys-[S]-serl-[S]-lys [reference 4] trihydrochloride; InvivoGen), FSL-1 (InvivoGen), or LPS (E. coli O55:B5; Sigma-Aldrich) at specified concentrations for 24∼48 h in a CO₂ (5%) incubator before analyzing their phenotype and function. In some experiments, mouse DCs were pulsed with antigen by incubating with 100 μg/ml OVA for 24 h before treatment with EDN or E. coli LPS.

For the transient silencing of DC MyD88, monocyte-derived iDCs were transduced without or with gripNA^MyD88 (5′-CCTGCAGCCAYDDCGGGC-3′; Active Motif) at 10 nmol/10⁶ cells using Chariot II (Active Motif) according to the manufacturer's instructions. After overnight incubation, both sham-transfected and gripNA^MyD88-transfected DCs were harvested, counted, and treated with EDN or IL-1α for 24 h before the collection of the culture supernatants.

Multiple cytokines in the culture supernatants of human DCs were measured using SearchLight microassay (Thermo Fisher Scientific). Mouse IL-6 and TNF-α in the supernatants of mouse DC culture were measured using ELISA kits (R&D Systems). DC migration was assessed using a 48-well microchemotaxis chamber assay, as described previously (67). The expression of surface molecules was determined by flow cytometry using a FACScan (Becton Dickinson) after staining DCs (5 × 10⁵/sample) with FITC- or PE-conjugated mouse monoclonal antibodies (all from BD Biosciences) against human CD11c (IgG1, κ, clone B-Ly6), CD80 (IgG1, κ, clone L307.4), CD83 (IgG1, κ, clone HB15e), CD86 (IgG1, κ, clone 2331), or isotype-matched control antibodies. The capacity of DCs to stimulate the proliferation of allogeneic T cells was evaluated by mixed lymphocyte reaction (68) with minor modifications. In brief, purified allogeneic T cells (10⁵/well) were cultured with different numbers of DCs treated in the absence (sham) or presence of EDN in a 96-well flat-bottom plate for 6 d at 37°C in humidified air with 5% CO₂. The proliferative response of T cells was examined by pulsing the culture with 0.5 μCi/well of [³H]-TdR (New England Nuclear) for the last 18 h before harvesting. [³H]-TdR incorporation was measured with a microbeta counter (Wallac).

DCs starved in serum-free G4 medium for 6 h were incubated at 37°C in the absence or presence of 1 μg/ml EDN or LPS for a period of time as specified. Treated cells were lysed in SDS sample buffer (62.5 mM Tris-HCl, pH 6.8, at 25°C, 2% wt/vol SDS, 10% glycerol, 50 mM dithiothretol, 0.01% bromophenol blue) at 10⁷/ml and loaded (20 μl/lane) on a 4–12% NuPAGE Bis-Tris Gel using 1× NuPAGE MES SDS Running Buffer as the electrode buffer and using SeeBlue Plus2 (Invitrogen) as molecular size markers. After transfer onto a piece of Immobilon membrane (Millipore), I-kBa was detected by Western blotting using rabbit anti–I-κBa (Cell Signaling Technology) as the first antibody and horseradish peroxidase–conjugated anti–rabbit IgG (Cell Signaling Technology) as the second antibody. The blot was developed using ECL Plus Western Blotting Detection System (GE Healthcare) and exposed to a piece of BioMax x-ray film (Kodak). Subsequently, the same membrane was stripped and probed consecutively for activated MAPKs in the same manner using rabbit anti-phosphorylated JNK, anti-phosphorylated p38, and anti–phospho-p44/42 (all from Cell Signaling Technology) as the first antibodies.

For the measurement of NF-κB activation, HEK293 cells (10⁴/well) were transfected using the calcium phosphate technique for 3 h in 96-well plates, with 12 ng IgkB-luciferase, 12 ng pSV-β-galactosidase, 2.5 ng pcDNA-hTLR2 or pcDNA-hTLR3, a combination of 2.5 ng pcDNA-hTLR4 and 6 ng pEF-boss-hMD2, or control plasmids (22). 24 h after transfection, the cells were stimulated with the indicated concentrations of PGN (Sigma-Aldrich), polyI:C (Sigma-Aldrich), LPS, or EDN at 37°C for a 24-h lysis of cells for the measurement of firefly and Renilla luciferase activities using the DualLuciferase Reporter Assay System (Promega) and for β-galactosidase (Tropix) on a Lumat LB9501 (Berthold). Firefly luciferase activity of individual transfections was normalized against Renilla luciferase activity. Data reflect the luciferase relative luciferase unit (RLU) divided by the control β-galactosidase RLU.

8-wk-old C57BL/6, TLR2^+/+, or TLR^−/− mice (3∼5 mice/group) were injected i.p. on day 1 with 0.2 ml PBS containing 50 μg OVA (Sigma-Aldrich) in the presence or absence of alum (Sigma-Aldrich), EDN, hAng, or LPS. On Day 14, all mice were booster immunized by i.p. injection of 0.2 ml PBS containing 50 μg OVA. Blood samples were taken on days 10 and 21 for the preparation of serum samples used for the measurement of anti-OVA antibody. On day 21, immunized mice were killed to remove spleens. Alternatively, 8-wk-old C57BL/6 mice (three mice/group) were injected i.p. with OVA-loaded DCs (5 × 10⁶/0.2 ml PBS/mouse) that had been treated in the absence or presence of 1 μg/ml EDN or LPS for 24 h. Mice were immunized by i.p. injection of OVA (50 μg/0.2 ml PBS/mouse) on day 5 and killed for the removal of spleens.

Total and subclass OVA-specific IgG antibodies were measured by ELISA as described previously (67). For the detection of total anti-OVA IgG, mouse anti-OVA IgG antibody (AntibodyShop) was used as the standard, and the concentration of total OVA-specific IgG in the serum was presented as ng/ml. For the measurement of subclass anti-OVA antibody, rabbit anti–mouse IgG1, IgG2a, IgG2b, or IgG3 (BD Biosciences) was used as the secondary antibody, and the level of anti-OVA IgG1, IgG2a, IgG2b, or IgG3 was illustrated as the reciprocal logarithmic value of the highest dilution that gave an absorbance ≥0.1 above the negative control.

Spleens of each group of mice were collected and pooled to make single splenocyte suspension. OVA-specific splenocyte proliferation and/or cytokine production were measured as described previously with minor modifications (69). In brief, splenocytes (5 × 10⁵/well) were seeded in triplicate in round-bottomed 96-well plates in complete RPMI 1640 medium (0.2 ml/well) and incubated in the presence or absence of the indicated concentration of OVA at 37°C in a CO₂ incubator for 60 h. The cells were pulsed with [³H]-TdR (1 μCi/well) for the last 18 h for the measurement of splenocyte proliferation. Alternatively, pooled splenocytes of each group were cultured in complete RPMI 1640 in 24-well plates (5 × 10⁶/1 ml/well) with the indicated concentrations of OVA for 48 h before the culture supernatants were harvested for the measurement of cytokines (SearchLight multiplex ELISA; Thermo Fisher Scientific).

The authors are grateful for Drs. S. Akira (Osaka University, Osaka, Japan) and S.M. Rybak (National Cancer Institute, Frederick, MD) for sharing MyD88^−/−, TLR1^−/−, TLR2^−/−, and TLR6^−/− mice, and the EDN-expressing construct, respectively.

This project has been funded in whole or in part with federal funds from the National Cancer Institute, NIH, under contract N01-CO-12400. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. This Research was supported in part by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.

The authors have no conflicting financial interests.