Toll-like receptor ligands synergize through distinct dendritic cell pathways to induce T cell responses: Implications for vaccines

Abstract

Toll-like receptors (TLRs) may need to cooperate with each other to be effective in detecting imminent infection and trigger immune responses. Understanding is still limited about the intracellular mechanism of this cooperation. We found that when certain TLRs are involved, dendritic cells (DCs) establish unidirectional intracellular cross-talk, in which the MyD88-independent TRIF-dependent pathway amplifies the MyD88-dependent DC function through a JNK-dependent mechanism. The amplified MyD88-dependent DC function determines the induction of the T cell response to a given vaccine in vivo. Therefore, our study revealed an underlying TLR mechanism governing the functional, nonrandom interplay among TLRs for recognition of combinatorial ligands that may be dangerous to the host, providing important guidance for design of novel synergistic molecular vaccine adjuvants.

Development of innate and adaptive immunity critically depends on the engagement of pattern recognition receptors (PRRs), which specifically detect microbial components named pathogen- or microbe-associated molecular patterns (1–4). Among these receptors, TLRs represent an important group of PRRs that can sense the molecular patterns. They are widely expressed by various cells in the blood, spleen, lung, muscle, and intestines (5–9).

Ligation of the TLRs by their specific ligands results in conformational changes in the receptors, leading to downstream signal transduction that primarily involves MyD88- and TRIF-dependent pathways (3, 10). Except for TLR3, all other TLRs can signal through the MyD88 pathway to induce proinflammatory cytokines. The signaling involves activation of intracellular protein kinase cascades including IκB kinase (IKK)-NF-κB and extracellular signal-regulated protein kinase (ERK), c-Jun N-terminal kinase (JNK), and p38 mitogen-activation protein kinases (MAPKs) (3, 11, 12). The MyD88-independent, primarily TRIF-dependent, pathway is used by both TLR3 and TLR4 and mediates the induction of type I interferons (IFNs) as well as CD86 and IP-10 (13, 14).

Although TLR ligands individually may be ignored at certain low doses, the host may have evolved to recognize some together as a combinatorial assault and mount immune responses against these combinations in a synergistic manner. This strategy would allow the immune system to rapidly respond to infection. Studies have addressed the questions on what combinations of TLR ligands can synergistically increase the magnitude of cytokine production by DCs (15–22). However, in addition to discovering new synergistic combinations, questions still remain about whether there is a logic to the TLR synergistic patterns that can be identified, what the underlying mechanism is for TLR synergy within target cells, and from a translational point of view, what can be learned for vaccine design.

Here, we systematically studied combinations of TLR 2, 3, and 9 ligands at suboptimal doses (at which minimal immune responses were induced), and we discovered that there is a nonrandom synergy determined by unidirectional intracellular cross-talk, within DCs, between the MyD88-dependent and -independent signaling pathways. This strategic interplay leads to synergistic enhancement of MyD88-dependent cytokine production from DCs, resulting in synergistic T cell activation. These results suggest that this important combinatorial TLR-based host defense mechanism may provide a previously undescribed approach for vaccine development by using synergistic TLR ligand combinations as immune adjuvants.

Results

TLR Synergy Involves Two Distinct Intracellular Pathways and Confers T Cell Activation Through Amplification of the MyD88-Dependent DC Cellular Function.

Although synergistic activation of DCs and T cells through stimulation of TLRs in some combinations has been reported (15–19), we sought to understand the underlying intracellular mechanism of synergy as well as discover new synergies and examine the implications for vaccine design, such as against HIV, by using the BALB/c (H-2^d) mouse model and HIV epitopes immunogenic therein (23). Thus, based on preliminary data examining various TLR ligands, we systematically studied the bacterial protein macrophage-activating lipoprotein (MALP)-2 (or MP2 for TLR 2/6), poly(I·C) (or PIC for TLR3) as a model of viral dsRNA, and CpG-containing oligodeoxynucleotide (CpG-ODN) (for TLR9) as a model of bacterial DNA at suboptimal doses in pairwise combinations (see Materials and Methods). Initially, we examined which paired combinations are synergistic for T cell activation. First, direct stimulation of purified total T cells freshly isolated from naïve BALB/c mouse spleens with TLR ligands for 20 h barely induced CD8⁺ T cells to express the activation marker CD69 (Fig. 1A) (24–27). Stimulation of syngeneic macrophages generated from BALB/c mouse bone marrow with TLR ligands before coculture with naïve T cells also resulted in a certain degree of T cell activation, but was unable to induce any synergy (Fig. 1A). It appeared that only syngeneic bone marrow-derived DCs (BMDCs) activated T cells when pretreated with MALP-2 + poly(I·C) or poly(I·C) + CpG-ODN pairs were found to be synergistic (Fig. 1A). The enhancement was also seen when C57BL/6 mouse cells were used, indicating that the synergy described here is not strain specific (see below). In addition, simply doubling the dose of each when used singly did not produce the enhancement. Also, both CD4⁺ and CD8⁺ T cells could be activated in a synergistic way, and the responses demonstrated when the total T cell preparation was stimulated were not simply a bystander effect from one to the other because each purified subset could be activated independently by DCs [supporting information (SI) Fig. S1A]. Of further interest is the finding that the combination of MALP-2 + CpG-ODN showed little or no synergy. These results suggest that DCs, as professional antigen-presenting cells, may or may not enhance T cell responses depending on the combinations of TLR ligands recognized, suggesting nonrandomness in TLR synergy.

Fig. 1.

Combinatorial TLR ligands nonrandomly amplify the MyD88-dependent DC function that correlates with the T cell activation. Macrophages or DCs derived from BALB/c mouse bone marrow were incubated with MALP-2, poly(I·C), and CpG-ODN singly or pairwise for 20 h. (A) CD69 expression of syngeneic T cells after 24 h of coculture with TLR ligand stimulated cells. Excess ligands were washed off before the coculture. (B) DC activation after TLR ligand stimulation. Secreted IL-12p70 (sIL-12p70) and sIP-10 were measured in the supernatants with a multiplex cytokine system. Percentage of intracellular IL-12⁺ (iIL-12⁺) cells out of DCs and surface expression of CD86 expressed as geometric mean fluorescence intensity (GMFI) were analyzed by flow cytometry based on gated MHC class II⁺CD11c⁺ DCs. Value on bars (in this and subsequent figures) indicates a synergy between two ligands expressed as fold increase calculated by dividing the geometric mean of the increase (subtracting the no-ligand control) in the response to the paired ligands by the sum of the geometric means of individual ligands less controls. (C) Correlation between CD69-expressing responsive T cells and increases in the MyD88-dependent (IL-12 and TNFα) or MyD88-independent DC function (CD86 and IP-10). Responsiveness of T cells was determined as >2-fold increases in percentage of CD69⁺ out of CD8⁺ T cells. One-tenth of original values on increases in sTNFα are shown to fit in the same scale with the others.

We next explored whether the synergistic T cell activation was due to synergistic triggering of particular DC functions in response to the paired TLR ligands. Treatment of BALB/c BMDCs in vitro with either MALP-2 + poly(I·C) or poly(I·C) + CpG-ODN resulted in synergistically increased secretion of IL-12p70 (sIL-12) assayed in the culture supernatants (Fig. 1B). Intracellular cytokine staining showed that these TLR combinations increased the number of DCs producing IL-12p70/40 (iIL-12⁺) (Fig. 1B); sTNFα, sMIP-1α, and sIL-6 (Fig. S1 B and C) in the supernatant were also elevated, albeit to a higher level by the former combination than the latter. Production of IL-10 was enhanced as well (≈500 pg/ml) by the synergistic combination of MALP-2 and poly(I·C) (data not shown). Lipoteichoic acid (LTA), which, like MALP-2, uses TLR2/6, could also act synergistically with poly(I·C) to boost IL-12 production, in contrast to peptidoglycan (PGN) and PAM3CSK4 (PAM3), which use TLR2/1, which could not (Fig. S2A). Thus, it suggests that the substantial increase depends on the engagement of TLR2/6. Also, MALP-2 + CpG-ODN, which could not synergistically activate T cells, did not increase IL-12 production as compared with single ligands (Fig. 1B). Therefore, the TLR combinations can synergistically activate DCs and, again, in a nonrandom way.

It is no surprise that poly(I·C) alone could activate DCs to up-regulate CD86 expression and produce IP-10; however, addition of MALP-2 or CpG-ODN did not result in a further significant increase (Fig. 1B). Even when the poly(I·C) concentration was titrated down in the presence of MALP-2 at the same concentration used, synergy for CD86 and sIP-10 was still marginal, in contrast to the strong synergy in IL-12 production (Fig. S2B). Similarly, other TLR2 ligands did not show synergy with poly(I·C) in this regard (Fig. S2A). Thus, the changes in poly(I·C)-induced CD86 and IP-10 DC functions did not seem to parallel T cell activation as well as the IL-12 and TNFα production described above.

TLR signaling involves MyD88-dependent or MyD88-independent, TRIF-dependent pathways (3, 28, 29). To quantitatively analyze the two signaling pathways in terms of DC functions on T cell activation, we calculated the Pearson correlation coefficient (r) between each of several DC activities and T cells activated to express CD69. The increase in CD69⁺ (of CD8⁺ cells) was strongly correlated with elevated DC production of MyD88-dependent cytokines sIL-12 (r = 0.776; P = 0.005) and sTNFα (r = 0.670; P = 0.003) (Fig. 1C). A stronger correlation was seen with the increase in number of iIL-12⁺ DCs (r = 0.945; P = 0.0001) (Fig. 1C). In contrast, T cell activation was poorly correlated with MyD88-independent production of IP-10 (r = 0.272; P = 0.476) and CD86 expression (r = 0.023; P = 0.948) by DCs (Fig. 1C). This finding implies that amplification of MyD88-dependent cytokines/chemokines contributes to the TLR-mediated synergistic activation of T cells.

MyD88-cJun Signaling Pathway Is Essential for Synergistic TLR-Ligand-Mediated DC to Activation of T Cells.

To understand the molecular mechanism of the TLR synergy, we studied the role of the MyD88 and TRIF signaling pathways by using KO mice. Because these KO mice were on a C57BL/6 background, we also had to use a different antigen system, the SIINFEKL epitope of ovalbumin. BMDCs derived from MyD88-deficient mice pretreated with the synergistic TLR combinations failed to enhance T cell CD69 expression in vitro compared with their C57BL/6 WT counterparts (Fig. 2A). In parallel, production of MyD88-dependent cytokines IL-12 (Fig. 2A), TNFα, MIP-1α, and IL-6 (Fig. S3 A–C) was diminished in these MyD88^−/− DCs. In contrast, MyD88-independent up-regulation of CD86 and IP-10 was barely changed (Fig. 2B), confirming that the MyD88-independent pathway is not directly associated with the DC function for TLR-mediated synergistic T cell activation. However, the enhancement of the MyD88-dependent DC function requires stimulation of the MyD88-independent pathway by poly(I·C), which works through TRIF (Fig. 2C), because synergistic IL-12 production was abrogated in the TRIF^−/− DCs (Fig. 2C) and these DCs were unable efficiently activate T cells (Fig. S3D).

Fig. 2.

Both MyD88- and TRIF-dependent pathways are essential in the synergy, but TRIF is required for the enhancement of IL-12 production. Cells from C57BL/6 WT mice were used as syngeneic controls for those from MyD88^−/− or TRIF^−/− mice. (A) Percentage of iIL-12⁺ DCs after TLR ligand treatment and percentage of CD69⁺CD8⁺ WT spleen T cells after 24 h of coculture with MyD88^−/− or WT DCs pretreated with TLR ligands. (B) Expression of surface CD86 by DCs and secreted IP-10 in the supernatant of DC culture after 20 h of TLR stimulation. (C) TRIF^−/− and WT BMDCs were treated with poly(I·C) and/or CpG-ODN for 20 h and iIL-12⁺ DCs (Left) or CD86 (Right) was measured.

Production of IFN-β was also enhanced by the synergistic TLR ligand combinations (Fig. S3E); however, blocking either IFN-α or IFN-β or both cytokines did not significantly abrogate the synergistic production of IL-12 (Fig. S3F) or T cell activation (data not shown). These results suggest that the synergistic effect of TLR combinations on DC and subsequent T cell activation critically depends on the amplification of the MyD88-dependent signaling pathway and does not require the presence of type I IFNs.

The MyD88-dependent signaling pathway is known to be primarily mediated through NF-κB and MAPKs (3, 11, 12). We observed that after synergistic TLR triggering, both IL-12 secretion (Fig. 3A) and iIL-12⁺ DCs (Fig. 3B and Fig. S4F) were inhibited in the presence of NF-κB and JNK inhibitors. Production of TNFα, MIP-1α, or IL-6 was also blocked by the JNK inhibitor and selectively by other inhibitors as well (Fig. S4 A–C). In contrast, none of these inhibitors blocked up-regulation of IP-10 and CD86 (Fig. S4D and Fig. S4E). However, only the JNK inhibitor effectively blocked the ability of DCs to activate T cells (Fig. 3C and Fig. S4G). Therefore, signaling through the JNK pathway is necessary for TLR synergistic activation of DCs to stimulate T cells.

Fig. 3.

Signaling through JNK is essential in the TLR-mediated DC functional synergy for T cell activation. Inhibitors for P38, ERK, JNK, or NF-κB (see SI Materials and Methods) at doses indicated (× 10 μM) were added to BALB/c BMDC cultures 1 h before the addition of TLR ligands; sIL-12p70 (A) and iIL-12⁺ DCs (B) were measured after 20-h stimulation with MALP-2 + poly(I·C). (C) Percentage of syngeneic CD69⁺CD8⁺ T cells after 24-h coculture with DCs pretreated with MALP-2 + poly(I·C) in the presence of inhibitors.

The Mechanism for Synergistic DC Activation by TLR Combinations Governs the T Cell Response to Vaccines.

It is critical to determine whether the above TLR synergy mechanism shown in vitro is meaningful and could be beneficial in vaccine design. Accordingly, we conducted immunization experiments by using as immunogens either the HIV CTL antigenic peptide P18I10 in BALB/c mice (H-2D^d) or the ovalbumin peptide SIINFEKL in C57BL/6 mice (H-2K^b; to use gene-KO mice). When naïve BALB/c mice were injected s.c. in the footpads with DCs pretreated with the synergistic TLR ligands and pulsed with P18I10, synergistically enhanced P18I10-specific CD8⁺ T cell responses were detected in the draining popliteal lymph nodes (LNs), assayed at day 5, by H-2D^d-P18I10 tetramer staining (Fig. 4A). In contrast, single ligands or the ineffective MALP-2 + CpG-ODN combination showed limited effect (Fig. 4A). The data indicate that the specific TLR synergy pattern observed in DC-mediated T cell activation in vitro translates to similar synergy in vivo.

Fig. 4.

DCs synergistically activated by TLR ligands can prime antigen-specific T cell in vivo, dependent on the MyD88-cJun intracellular pathway. Depending on the mouse genetic background, either the P18I10 or SIINFEKL antigen system was used together with MALP-2, poly(I·C), and CpG-ODN singly or pairwise. Mice were immunized in the footpad either 3 consecutive days with antigen peptides and TLR ligands or once with peptide-pulsed syngeneic BMDCs pretreated with these ligands. Cells were isolated from paired popliteal LNs and stained with tetramers for flow cytometry. (A) Total number of H-2D^d-P18I10 tetramer positive (tet⁺) CD8⁺ T cells at day 5 after immunization with BALB/c DCs pulsed with P18I10. (B) Total number of H-2K^b-SIINFEKL tet⁺CD8⁺ T cells in popliteal LNs of WT C57BL/6 mice at day 5 after immunization with MyD88^−/− or WT syngeneic DCs pretreated with TLR ligands and pulsed with SIINFEKL peptide. (C) Total number of H-2K^b-SIINFEKL tet⁺CD8⁺ T cells in the popliteal LNs of MyD88^−/− or WT C57BL/6 mice after the first immunization with SIINFEKL and TLR ligands. (D) Total number of popliteal LN H-2D^d-P18I10 tet⁺CD8⁺ T cells from WT BALB/c mice after immunization with P18I10-pulsed DCs pretreated with JNK inhibitors (×10 μM). N, not available.

To confirm that MyD88 is also essential in the in vivo synergy, MyD88^−/− DCs (C57BL/6 background), after in vitro treatment with MALP-2 + poly(I·C) and pulsing with SIINFEKL peptide, were used to immunize C57BL/6 WT mice in the footpad. These immunized animals failed to develop SIINFEKL-specific CD8⁺ T cells (H-2K^b-SIINFEKL tetramer⁺) in the popliteal LNs, in contrast to those immunized with WT DCs (Fig. 4B and Fig. S5A). This result rules out carry-over of TLR ligands for activation of host DCs. Indeed, when immunized directly with SIINFEKL peptide together with MALP-2 + poly(I·C), mice deficient in MyD88 also failed to show enhanced antigen-specific T cell responses compared with WT recipients (Fig. 4C and Fig. S5B). This result suggests that MyD88 is required as well for the in vivo enhancement of antigen-specific T cells by the synergistic TLR combinations.

To verify that the JNK pathway is involved in the in vivo synergy, P18I10-pulsed WT BALB/c BMDCs were treated with the synergistic TLR ligands in the presence of JNK inhibitors. We found that these DCs were unable to efficiently prime antigen-specific T cells in the draining LNs of syngeneic mice after footpad injection (Fig. 4D). Therefore, MyD88-dependent JNK-dependent DC activation is also critically involved in in vivo T cell priming.

To further determine whether in vivo recognition of TLR ligands directly resulted in a similar DC activation pattern in vivo, popliteal LN cells were recovered at 36 h after immunization of BALB/c mice. Administration of the synergistic TLR combinations dramatically increased the total numbers of functioning LN DCs as assayed by ex vivo intracellular staining of IL-12 (P < 0.01) (Fig. 5A and Fig. S6A), with a slight increase in total DC counts (Fig. S6A). However, there was no further increase in up-regulation of CD86 compared with poly(I·C) alone (Fig. 5A). Thus, the in vivo synergistic amplification pattern for DC activation correlates with what was seen in vitro.

Fig. 5.

Combinatorial TLR ligands activate DCs in vivo in the same pattern as seen in vitro and locally prime functioning T cells against vaccines. BALB/c mice were immunized in the footpads with the TLR ligands indicated and PCLUS3–18IIIB (containing P18I10) for 3 consecutive days except where indicated. (A) Total numbers of iIL-12⁺MHC class II⁺CD11c⁺ DCs and CD86 expression were evaluated 36 h after injection (two times). (B) Total number of H-2D^d-P18I10 tet⁺CD8⁺ T cells in the popliteal LNs 5 days after the first immunization. (C) Time course of H-2D^d-P18I10 tet⁺CD8⁺ T cells (Left) and IFN-γ-producing (IFN-γ⁺) tet⁺CD8⁺ T cells (Right) after immunization.

To define T cell responses after vaccination, PCLUS3–18IIIB peptide, which contains P18I10 linked to a T-helper peptide (PCLUS3) (23), was injected into BALB/c mice together with the TLR ligands. As assayed at day 5 after immunization, the synergistic combinations effectively primed antigen-specific T cells in a synergistic way (Fig. 5B), and the synergistic activation of antigen-specific and functional CD8⁺ T cells persisted through the entire priming phase in the draining LNs (Fig. 5C). The MALP-2 + CpG-ODN combination was again found ineffective in vivo (Fig. 5 A and B). The facts that use of synergistic combinations of TLR ligands as adjuvants results in the same synergistic activation of DCs as well as antigen-specific T cells in vivo, and that immunization with peptide-pulsed DCs treated with the same synergistic combinations of TLR ligands is sufficient to synergistically induce a specific T cell response implies that the mechanism of adjuvant function of the synergistic combinations of TLR ligands in vivo is through activation of DCs, and that the mechanism in vivo is the same as that worked out in vitro. Together, our results identified the mechanism involved in TLR nonrandom interplay (paralleled both in vivo and in vitro and different strains of mice and antigen systems), providing important implications for the design of vaccines by choosing appropriate combinations of TLR ligands as adjuvants.

Discussion

The present study uncovered the underlying mechanism involved in TLR synergy and demonstrated the implications for vaccine development, as well as discovering previously undescribed synergies. We first found that the MyD88-dependent and MyD88-independent DC functions are divergent in response to nonrandom, synergistic TLR ligands. MALP-2- and CpG-ODN-associated MyD88-dependent DC function in producing IL-12, TNFα, and IL-6 was remarkably augmented when poly(I·C) was included. In contrast, poly(I·C)-associated MyD88-independent DC function such as CD86 and IP-10 was readily up-regulated by poly(I·C) alone but could not be further increased by addition of either MALP-2 or CpG-ODN. Although cytokines were seen to be synergistically produced as a result of one such case (21), we demonstrated that the magnitude of MyD88-dependent, but not MyD88-independent, DC function was correlated with DC-induced T cell activation, in which signaling through JNK is essential. When the poly(I·C) concentrations were reduced further, the magnitude of CD86 and sIP-10 elevation reflected only a weak synergy, which was little more than an additive effect, compared with the much greater synergy for IL-12 production. The stronger synergistic amplification of the MyD88-dependnent signaling pathway suggests unidirectional intracellular cross-talk between the two signaling pathways, mediated by the MyD88-independent pathway through TRIF. An ongoing stimulation through the TRIF-dependent pathway is required for the enhancement of the MyD88-dependent DC function and for the T cell activation. We consider that the immune system has designed a delicate system aimed at effective operation of TLR recognition, conferring on DCs the ability to distinguish danger signals from innocuous stimuli (i.e., nonrandom) and induce T cell-mediated adaptive immune responses to these dangers. This mechanism represents a substantial difference from macrophages, which react to TLR ligands by mediating direct antimicrobial innate effector responses (30–32).

This previously undescribed unidirectional cross-talk is reminiscent of a previous hypothesis about DC signals (33, 34). The MyD88-independent pathway corresponding to signal 2 (costimulation) increases the likelihood of T cell priming, but the MyD88-dependent pathway as signal 3 (cytokines) determines a T cell response. The cross-talk may be established to avoid induction of unnecessary activation of immune responses while allowing an immune response against infection if it is likely to occur. Also, the MyD88-dependent cytokine amplification may be self-controlling through enhanced production of regulatory factors such as IL-10. How the TLR-mediated proinflammatory response is fine-tuned merits further study.

Based on the synergy mechanism described here, we reasoned that TLR4, because it uses both signaling pathways (13, 35), would be capable of self-synergizing in DCs. We confirmed this hypothesis by showing that LPS, MPL, and E6020 alone had an efficacy equivalent to poly(I·C) in up-regulation of CD86, as well as equivalent to the synergistic combinations in inducing IL-12 (Fig. S2A). Also, addition of poly(I·C) to the TLR4 stimulation did not further enhance IL-12 production (Fig. S2A). Thus, our synergy mechanism may possibly also fit with what is known about the TLR4-associated hypersensitivity immune response. Effective recognition by cooperative TLRs may be a critical surrogate for TLR4 when endotoxins are unseen or absent from some microbes and vice versa, thereby complementing each other to ensure effective pathogen recognition.

Other than TLR3 binding, dsRNA also binds to and activates PKR, involving NF-κB activation and up-regulation of cytokines such as IFNs (36). Intracellular RIG-1 and MDA5 have been recently shown to recognize dsRNA (37, 38). Whether and how these binding receptors are involved in synergy in addition to TRIF, which we found to be necessary by using TRIF-KO mice, warrants further investigation. Previous studies showed that JNK could be also activated through triggering of TLR3 (39), either PKR dependent (40) or independent (41). Based on our results, we speculate that JNK may be an essential intracellular transduction point through which the synergy is generated. In other words, we envision that intracellular activation and unidirectional cross-talk from the TRIF pathway to the MyD88 one, rather than poly(I·C)-induced surface CD86 or extracellular IP-10 and type I IFNs, is essential for the induction of TLR synergy for DC-mediated T cell activation.

TLR2 recognizes its ligands by forming heterodimers with either TLR1 or TLR6 depending on whether the ligands are triacylated or diacylated, respectively. MALP-2 and LTA (binding TLR2/6) are diacylated lipoproteins, whereas PGN and PAM3CSK4 (binding TLR2/1) are triacylated. This difference in acylation might account for the differences in the level of DC activities observed between the two subgroups. PGN and PAM3CSK4 did not synergize with poly(I·C) to up-regulate IL-12, whereas MALP-2 and LTA did (Fig. S2A). PAM3CSK has been found not to favor IL-12 production (42, 43). Thus, depending on heterodimer formation, TLR2 might signal differently to more selectively determine priming or tolerance (42, 44).

Recent studies demonstrated that human myeloid DCs also express TLR9 and respond to CpG-ODN, although the expression detected was somewhat weak and its capacity to induce type I IFNs is not as dramatic as in plasmacytoid DCs (45, 46). These studies showed that monocyte-derived DCs are not necessarily incompetent to respond to CpG-ODN and may carry out specific functions different from those mediated by plasmacytoid DCs. Because type I IFNs are dispensable in the synergy, including CpG-ODN would still be applicable for vaccine design for human use.

The synergistic activation of DC and T cells was established when TLRs detected dsRNA in the presence of either surface molecules or genomic DNA from bacteria. This situation is reminiscent of coinfections of viruses and bacteria, such as influenza with bacterial pneumonia or HIV with TB, which cause severe pathology and could be life-threatening. In vaccine development, to our knowledge, covaccination with viral and bacteria components has hardly ever been studied. Future studies may need to determine whether viral–bacterial covaccines are potentially more effective than each alone.

In summary, we have elucidated what intracellular molecular mechanism accounts for the synergistic immune response initiated within DCs in response to multiple TLR ligands. The previously undescribed unidirectional intracellular cross-talk between the two TLR signaling pathways provides important insight into the host defense in response to combinatorial microbial components that alert the host to infection. Such an immune activation mechanism may lead to a new rationale in the design of more effective vaccines by using multicomponent immune adjuvants, without the need for endotoxin or high doses of single TLR ligands.

Materials and Methods

See SI Materials and Methods for details.

Animals and Reagents.

Female BALB/c and C57BL/6 mice were purchased from the Frederick Cancer Research Center or Taconic Farms. MyD88^−/− and TRIF^−/− mice (C57BL/6 H-2^b background) were generated as previously described (47). Peptides include PCLUS3–18IIIB, P18-I10, and SIINFEKL. TLR ligands include LTA, MALP-2, Zym, Pam3, poly(I·C) or PIC, LPS, MPL, E6020, and CpG-ODNs. Dosage of MALP-2, poly(I·C), and CpG-ODN was, unless otherwise indicated: 0.1, 20, 3 μg/ml in vitro or 0.1, 30, 5 μg in vivo. Inhibitors for P38, ERK, JNK, and NF-κB were purchased from EMD Biosciences.

Cell Isolation, Purification, and Coculture.

BMDCs were generated as previously described (48). Popliteal LN cells were isolated after DC or peptide immunization in the footpad. T cells for culture were purified from mouse spleens.

Immunization.

For peptide immunization, TLR ligands were included and given by footpad injection. For DC immunization, DCs were pretreated with TLR ligands in vitro for 20 h and pulsed with peptide for 2 h.

Flow Cytometry and Cytokine Measurements.

Antibodies for flow cytometry were purchased from eBioscience or BD Biosciences. IFN-γ production in T cells was measured after restimulated with proper peptides. Intracellular cytokines was measured in DCs after stimulation with TLR ligands (49). ELISA or multiplex systems were used to determine cytokine production in DC supernatants.

Statistical Analysis.

Comparisons between groups were analyzed by Student's t test. P values <0.05 were considered statistically significant. The pairwise correlation between DC functions and T cell activation was estimated by r.