|Home | About | Journals | Submit | Contact Us | Français|
Toll-like receptors (TLRs) play prominent roles in initiating adaptive immune responses to infection, but their roles in particular cell types in vivo are not established. Here we report the generation of mice selectively lacking the crucial TLR-signaling adaptor MyD88 in dendritic cells (DCs). In these mice, the early production of inflammatory cytokines, especially IL-12, was substantially reduced following TLR stimulation. Whereas, the innate interferon γ response of natural killer cells and natural killer T cells, and TH1 polarization of antigen-specific CD4 T cells were severely compromised after treatment with a soluble TLR9 ligand, they were largely intact following administration of an aggregated TLR9 ligand. These results demonstrate that the physical form of a TLR ligand affects which cells can respond to it and that DCs and other innate immune cells can respond via TLRs and collaborate in promoting TH1 adaptive immune responses to an aggregated stimulus.
The immune system of multicellular organisms must recognize the presence of infectious agents and direct effector mechanisms against those agents. In recent years, Toll-like receptors (TLRs), which recognize a variety of pathogen-associated molecular patterns of infectious agents, have emerged as critical for this recognition (Akira et al., 2006). In mammalian tissues, TLRs are highly expressed by resident immune cells, including DCs, tissue macrophages and mast cells, and to a lesser degree by other cell types including fibroblasts, epithelial cells and endothelial cells (Takeda et al., 2003). Upon binding ligands, all known TLRs except for TLR3 can activate downstream signaling cascades through the adaptor protein MyD88 to induce production of inflammatory cytokines by macrophages, DCs, and to a lesser extent by other cell types. These cytokines, including interferon-α (IFN-α), interferon-β (IFN-β), interleukin 12 (IL-12), tumor necrosis factor α (TNFα), interleukin 6 (IL-6) and interleukin 1 (IL-1) attract innate immune cells and/or promote the initiation and polarization of adaptive immune responses (Akira, 2006). Indeed, MyD88-deficient mice have defects in many innate and adaptive immune responses. For example, immunization of mice with TLR ligands, such as CpG, LPS, or with complete Freund’s adjuvant which contains ligands for TLR2 and TLR4, induces naïve CD4 T cells to differentiate into T cell helper type 1 (TH1) effector cells, and this is severely compromised in mice deficient in MyD88 (Iwasaki and Medzhitov, 2004).
DCs are known to be the main antigen-presenting cells that activate naïve T cells in all or most circumstances (Jung et al., 2002)(Itano and Jenkins, 2003), but how TLRs promote their maturation and ability to polarize CD4+ T cells to TH1 effector cells is less well understood. During infection, microbial TLR ligands can induce DCs to mature, a response characterized by up-regulation of surface MHC-peptide complexes and co-stimulatory molecules involved in activating T cells and of the chemokine receptor CCR7, and by migration to the T cell zones in the draining lymph node (Iwasaki and Medzhitov, 2004). Once in lymphoid tissue, mature DCs promote activation of antigen-specific T cells and secrete cytokines and other factors that can promote effector T cell differentiation. However, microbial TLR ligands can also activate other tissue resident cell types to secrete inflammatory mediators, such as IFN-α, IFN-β, and TNFα, which can also promote DCs to mature and enable them to promote T cell immune responses, providing an indirect mode of activation of DCs (Kapsenberg, 2003). In order to distinguish the role of direct and indirect modes of TLR activation of DCs, Sporri et al employed mixed bone marrow chimeric mice, which were treated with a synthetic TLR9 ligand. They found the indirect mode of stimulation was sufficient for maturation of DCs in their experimental system but was insufficient to promote a robust TH1 or TH2 effector cell response (Sporri and Reis e Sousa, 2005). This study suggested that direct TLR stimulation of antigen-presenting DCs, also called “TLR licensing” (Heath and Villadangos, 2005), is important for promoting a robust T cells response even in the context of inflammatory cytokines produced by neighboring cells. To address in more detail the role of TLR signaling in different cell types for immune responses, we created a conditional allele of the mouse myd88 gene. By crossing with a DC-specific Cre transgene, CD11cCre (Caton et al., 2007), we have generated highly DC-specific MyD88-deficient mice. Using these mice, we have found that MyD88-dependent signaling in DCs plays a very important role in innate cytokine production and TH1 polarization of antigen-specific CD4 T cells, but that in some circumstances non-DC cell types can cooperate with DCs to support these immune responses.
To create a conditional allele of mouse myd88, we introduced 34 base pair LoxP sites on either side of exon 3 of the gene by homologous recombination in embryonic stem cells, which were then used to generate chimeric mice by standard procedures (Supplemental Figure S1A and S1B). Splenic DCs from mice with two copies of this targeted allele expressed MyD88 protein at levels indistinguishable from wild type DCs (Supplemental Figure S1C), indicating that this allele is likely to function normally in the absence of Cre-mediated deletion.
To test the ability of the Cre recombinase to delete exon 3 of myd88, we generated mice carrying both the myd88flox allele and the Vav-Cre transgene, which is expressed in all cells of the hematopoietic lineage (de Boer et al., 2003). Splenocytes from myd88flox/floxVav-Cre mice exhibited nearly complete deletion as judged by loss of MyD88 protein expression (Supplemental Figure S1C).
To study the role of MyD88 in DCs, we crossed mice with the myd88flox allele to mice carrying the CD11c-Cre transgene, which is preferentially expressed in DCs (Caton et al., 2007). The numbers of DCs and of their subsets were unchanged in the spleen and lymph nodes of the myd88flox/floxCD11c-Cre mice (Supplemental Table S2 and data not shown). The deletion of the myd88flox allele was measured by a quantitative PCR assay, which detected the amount of residual myd88 exon 3 sequence in genomic DNA. We found that over 94% of the myd88flox allele was deleted in the conventional DCs (cDCs) and around 80% was deleted in plasmacytoid DCs (pDCs), whereas significant deletion was not detected in splenic macrophages/neutrophils, or in naïve or activated CD4+ T cell (Table and Supplemental Table S1). These result were corroborated by measuring MyD88 protein in purified cells (Supplemental Figure S1D). NK cells also express CD11c, although at lower levels than DCs (Laouar et al., 2005). We found no significant deletion in purified NK cells (Table). These results demonstrate that there is highly selective deletion in DCs of the myd88flox/floxCD11c-Cre mice, and therefore, these mice are referred to here as DC-MyD88 KO mice.
TLR stimulation induces DCs to produce cytokines, including IL-12, which promotes IFN-γ production and polarization of activated CD4 T cells to the TH1 effector type. To test the importance of MyD88-dependent signaling in DCs for this response, we injected DC-MyD88 KO mice and control mice intravenously with CpG-containing oligodeoxynucleotide ODN1826, a B/K-type phosphorothioate oligodeoxynucleotide (CpG), a TLR9 ligand which signals solely through a MyD88-dependent pathway. One hour later, we examined splenic DCs for IL-12 and/or IL-23 production by staining intracellular IL-12 p40. The splenic CD8α+ DCs in wild type mice had a robust induction of IL-12p40, but this response was completely defective in the DC-MyD88 KO mice (Figure 1A). This result confirmed that ablation of MyD88 function in CD8α+ DCs had been achieved and that this rapid response requires MyD88 signaling within DCs. Next, we examined the contribution of DCs to systemic IL-12 production in vivo. Two hours after intravenous administration of CpG, the levels of IL-12 p70 in the serum of DC-MyD88 KO mice were below detection limit, whereas the levels in the control mice increased significantly (Figure 1B), indicating that the DCs are the major cell type to rapidly produce functional IL-12 in vivo following TLR9 stimulation via the blood stream. Next, we examined the IL-12p40 response of these mice at 1 hour after stimulation with ligands for TLR1/2, TLR4, TLR5, TLR7 or TLR9. We found that the levels of IL-12p40 mRNA induced in total splenocytes in response to stimulation by these TLR ligands administrated systemically were all greatly reduced (5–30×) in the DC-MyD88 KO mice compared to the control mice (Figure 1C). These results indicate that DCs are the major cell type in the spleen to rapidly produce IL-12p40 in response to the stimulation of most TLRs.
Although DCs are generally thought to be major producers of IL-12, other splenic cell types such as macrophages are also capable of producing substantial amounts of inflammatory cytokines, including IL-1β, IL-6, and TNFα in response to TLR stimulation. Therefore, we examined the induction of several proinflammatory cytokines after systemic administration of CpG. Surprisingly, we found that DC-MyD88 KO mice exhibited substantially reduced mRNA induction of most tested cytokines in the spleen (Figure 1D), as well as greatly decreased serum cytokine levels (Supplemental Figure S2 and data not shown). Similar results were also found with ligands for TLR1/2, TLR4, TLR5, and TLR7 (data not shown). These results indicate that DCs play a surprisingly major role in early innate immune cytokine responses upon systemic administration TLR ligands.
In the experiments described above, we used a synthetic TLR9 ligand that acts in monomeric form (Klinman, 2004). However, many natural TLR ligands exist in complexed forms, which could interact differently with host immune cells. To make a complexed form of a TLR ligand, we encapsulated the TLR9 ligand CpG with a cationic lipid, 1,2-dioleoyloxy-3-trimethylammonium-propane-methylsulfate (DOTAP) (hereby referred to as CpG/DOTAP). It has been shown that responses to CpG/DOTAP, like responses to uncomplexed CpG, are TLR9- and MyD88-dependent (Yasuda et al., 2005)(Honda et al., 2005), a point that we confirmed with myd88−/− mice (data not shown). Using a fluorescently labeled CpG, we found that DOTAP enhanced up-take of CpG in splenic cell types such as CD11b+ DCs, pDCs, macrophages and monocytes (Figure 2A). Complexing CpG with DOTAP also changed the profile of the responding cell types. For example, injecting CpG/DOTAP intravenously induced both splenic CD8α+ DCs and CD11b+ DCs to express IL-12p40 intracellularly in WT mice (Figure 2B), a feature that is similar to LPS, a lipid ligand for TLR4 that also forms aggregated membrane-like structures (Skelly et al., 1979) but unlike soluble CpG (Figure 2B). As expected, the induction of intracellular IL-12p40 was still dependent on DC-intrinsic MyD88 signaling as it was abolished in both CD8α+ DCs and CD11b+ DCs of the DC-MyD88 KO mice (Supplemental Figure S2A). Interestingly, we also found that CpG/DOTAP induced intracellular TNFα in CD11b+ DCs and in an additional population of cells that were CD11b+, F4/80+, Ly6C+, side light scatter (SSC)low, CD11c−, NK1.1−, B220−, consistent with a monocyte phenotype (Figure 2C), suggesting that DOTAP also enhances the responses to CpG of other cell types in addition to CD11b+ DCs. This TNFα response was not impaired in the monocytes, but was ablated in DCs in the DC-MyD88 KO mice (Figure 2C and Supplemental Figure S2B), confirming the DC-specificity of MyD88 ablation by CD11c-Cre.
Next, to examine a broader range of inflammatory cytokine responses to this complexed form of TLR9 ligand, we measured inflammatory cytokine mRNA induction in the spleen after injecting the mice i.v. with CpG/DOTAP. Interestingly, the inductions of many cytokines, including IL-23 p19, IL-6 and TNFα were attenuated to a much lesser degree in the DC-MyD88 KO mice stimulated in this way than in mice stimulated with CpG alone (Figure 1D), suggesting that CpG/DOTAP stimulated TLR9 on non-DC cell types in the spleen more effectively than uncomplexed CpG. In agreement with a previous report (Honda et al., 2005), the CpG/DOTAP complex strongly induced IFN-β and IFN-α4 mRNA, whereas soluble CpG did so poorly (Figure 2D). By measuring the mRNA induction in different cell types, we found that pDCs were the major type 1 IFN-producing cell type in the spleen in response to CpG/DOTAP, but other cell types including DCs and macrophages could also produce these cytokines (Supplemental Table S2). This type 1 IFN response was abolished in DCs and was significantly reduced in the pDCs in the DC-MyD88 KO mice (Supplemental Table S2), but the overall induction in the spleen was still substantially higher than in the WT mice stimulated with CpG alone (Figure 2D).
In the early inflammatory cytokine response to systemic administration of TLR agonists, we consistently observed strongly attenuated IL-12 p40 expression in the DC-MyD88 KO mice. As IL-12 is a potent inducer of IFNγ production, we next asked how IFNγ production in response to TLR stimulation was affected by ablation of MyD88 selectively in DCs. To identify the cellular source of IFNγ after TLR stimulation, we stained splenocytes for intracellular IFNγ at different time points after injecting mice with CpG. Consistent with a previous report (Laouar et al., 2005), we found that NK cells were the major IFNγ producing cells in the spleen (Supplemental Figure S2), starting only a few hours after TLR ligand administration. NKT cells also expressed IFNγ, but to a significantly lesser degree both in terms of the number of IFNγ-positive cells in the spleen and of the amount of IFNγ detected per cell (Supplemental Figure S4). In contrast, CD4+ T cells, CD8+ T cells, and DCs did not express a significant amount of IFNγ in response to CpG at these early time points. To examine the role of MyD88 signaling in DCs for this in vivo innate IFNγ response, we measured IFNγ production by NK cells and NKT cells after TLR stimulation in the DC-MyD88 KO mice. Interestingly, we found that the IFNγ response was totally dependent on MyD88 function in DCs when the mice were stimulated with CpG (Figure 3A and data not shown for NKT cells). Similar results were seen with Pam3CSK4 (TLR1/2 agonist) and Imiquimod (TLR7 agonist) (data not shown). As expected, this IFNγ response was abolished in mice deficient in IL-12p35 (Figure 3B). These results suggest that the response of NK cells or NKT cells to direct TLR stimulation was not sufficient to produce IFNγ, and instead these cells needed to be stimulated by cytokines such as IL-12 produced by DCs. When the response to i.v. injection of CpG/DOTAP was examined, the number of IFNγ-positive cells was substantially enhanced compared to the response to CpG alone (Figure 3A and data not shown for NKT cells). This enhanced IFNγ response was fully MyD88-dependent and was to a large part dependent on type 1 IFN signaling (Figure 3B). Interestingly, ablating MyD88 in DCs only partially reduced the IFNγ response to CpG/DOTAP administrated i.v. (Figure 3A), indicating that MyD88 signaling in cell types other than DCs plays a substantial role in this response. Stimulation with LPS also induced a strong IFNγ response from NK cells in the control mice (Figure 3A, data not shown for NKT cells). This IFNγ response was also only partially reduced in the DC-MyD88 KO mice (Figure 3A), but was almost completely abolished in the MyD88 KO mice and in the myd88flox/floxVav-Cre mice (Supplemental Figure S5).
We next assessed the importance of MyD88 signaling in DCs for their ability to prime naïve CD4+ T cells. DC maturation in the spleen was examined after injecting the mice intravenously with CpG or CpG/DOTAP. In control mice, both CpG and CpG/DOTAP induced increased expression of CD86, CD40 and to a lesser extent, class II MHC molecules in splenic cDCs, but the response to CpG/DOTAP was much stronger (Figure 4A). These responses were attenuated in the DC-MyD88 KO mice, but in the mice stimulated with CpG/DOTAP, a subset of DCs in the spleen, including both CD8α+ and CD11b+ DCs (data not shown), induced expression of CD86 to a similar level to that of the DCs from the stimulated MyD88-expressing mice (Figure 4B). These results indicate that cytokines from other cell types surrounding DCs can partially compensate for the loss of direct TLR stimulation to induce some DCs to mature in response to this potent TLR stimulus. It should be noted that the response to CpG/DOTAP remained fully MyD88-dependent because there was no DC maturation in myd88−/− mice (Figure 4B).
To examine the effect of defective TLR signaling in DCs on the activation of adaptive immune responses, we adoptively transferred CFSE-labeled ovalbumin (OVA)-specific T cell receptor (TCR)-transgenic OT-II T cells into DC-MyD88 KO and control mice, and then immunized the mice with OVA using either CpG or CpG/DOTAP as adjuvants. The proliferation and effector polarization of OT-II T cells in the spleen were analyzed 4 and 7 days later. We found that both immunization protocols led to accumulation of increased numbers of OT-II T cells in the spleen of wild type recipient mice compared with mice immunized with OVA alone (data not shown). Most of the transferred OT-II T cells had divided multiple times by day 4 after immunization with OVA plus either CpG or CpG/DOTAP, as indicated by CFSE dilution (Figure 5A). However, we consistently found that more OT-II T cells had accumulated in the spleen of the control mice immunized with CpG/DOTAP than in the control mice immunized with CpG (Figure 5B), correlating with the fact that more DCs had matured in the CpG/DOTAP-treated mice (Figure 4). This expansion and survival of antigen-specific naïve CD4+ T cells was clearly decreased in the DC-MyD88 KO mice immunized with OVA + CpG compared to the control mice (Figure 5B). Similar results were seen on day 7 (data not shown). In contrast, the response was not compromised by MyD88-deficiency in DCs in mice immunized with OVA + CpG/DOTAP (Figure 5B).
In addition, OT-II T cell effector function was evaluated 4 and 7 days after immunization by intracellular staining of IFNγ after in vitro re-stimulation for 4 hours. We found that selective deletion of MyD88 in DCs impaired the TH1 effector differentiation of OT-II T cells when CpG was used as adjuvant, as indicated by a reduced number of IFNγ-producing cells on day 4 (Figure 5), and similarly on day 7 (data not shown). In contrast, no such difference was found when mice were immunized with CpG/DOTAP (Figure 5), suggesting that TH1 differentiation was largely intact in this circumstance where MyD88 signaling in cells other than DCs plays a more prominent role in cytokine production and in induction of DC maturation. Interestingly, mice immunized with OVA + LPS also exhibited strong expansion of OT-II CD4+ T cells, and differentiation to TH1 cells in both control and DC-MyD88 KO mice (data not shown).
TH1 cells secrete cytokines including IFNγ to induce class switch to IgG2c and IgG2b by responding B cells. As an alternative readout for TH1 development, we examined antibody responses after immunizing mice with OVA plus either CpG or CpG/DOTAP. Although roughly similar amounts of anti-OVA IgM were induced in control and DC-MyD88 KO mice (Figure 6 upper panel), total IgG levels were significantly reduced in DC-MyD88 KO mice immunized with OVA + CpG. In addition, there was even more drastically reduced production of IgG2c and IgG2b anti-OVA antibody in the DC-MyD88 KO mice. In contrast, immunization with OVA + CpG/DOTAP induced comparable levels of total IgG and IgG1, as well as substantial amounts of IgG2c and IgG2b anti-OVA antibodies in the DC-MyD88 KO mice, although the titers were slightly reduced compared to control mice (Figure 6 lower panel). Similar results to those seen with OVA + CpG/DOTAP were seen when mice were immunized with OVA+LPS (data not shown). Thus, as in the case of effector TH1 differentiation, the T cell-dependent IgG response to OVA + CpG was largely dependent on MyD88 function in DCs. However, changing the physical form of the TLR ligand, in this case CpG, to a more aggregated form enhanced cytokine production from other cell types, which likely contributed to the observed T cell-dependent antibody response.
In the experiments described here, we have examined the role of TLRs on different immune cell types for the rapid production of inflammatory cytokines and for the activation of the adaptive immune response. Experiments employing a new conditional allele of the gene encoding the key TLR signaling adaptor molecule, MyD88, together with DC-specific expression of the Cre recombinase revealed a critical role for TLR signaling in DCs for many responses to TLR ligands. These experiments also revealed situations where other cell types can contribute importantly to the production of certain inflammatory cytokines and relieve the requirement for DC-intrinsic TLR signaling in order to stimulate a vigorous TH1 response.
These experiments provide evidence for the view that the direct recognition of microbial ligands by TLRs on DCs plays a prominent role for the initiation of the adaptive immune response and for directing polarization to a TH1 response. When mice were immunized with ovalbumin and soluble CpG, TH1 polarization of ovalbumin-specific CD4+ T cells was greatly diminished by deletion of the myd88 gene selectively in DCs. This result is consistent with a previous study using wild type and MyD88-deficient bone marrow chimeras (Sporri and Reis e Sousa, 2005). However, in contrast to that study, we found that DCs lacking MyD88 had substantially compromised maturation in response to soluble CpG administrated i.v., as indicated by the reduced induction of co-stimulatory molecules. A major difference between those experiments and the ones described here is that in the bone-marrow chimeric mice, there was a mixture of MyD88-expressing DCs and MyD88-deficient DCs, whereas in the experiments described here, the vast majority of the DCs were MyD88-deficient and other cell types retained MyD88 expression. Therefore, the indirect maturation observed in the bone marrow chimeric mice probably reflects the action of cytokines produced primarily by neighboring DCs, acting in a paracrine manner. This interpretation is supported by observations reported here that DCs were the major cell types producing cytokines in response to soluble TLR ligands. Taken together, our results and those of Sporri and Reis e Sousa (2005) demonstrate the importance of TLR signaling in DCs for CD4+ T cell responses, at least in the context of immunization with soluble TLR ligands
MyD88 function in DCs was found to be especially important for IL-12 production in response to TLR ligand stimulation. IL-12 is known to simulate innate immune cell types such as NK cells to express IFNγ, and to promote the TH1 polarization of CD4+ T cells (Magram et al., 1996), both of which can be enhanced by IL-18 (Takeda et al., 1998). Indeed, we found that the early IFNγ production from NK cells and NKT cells in response to soluble CpG was totally dependent on IL-12. IFNγ from NK cells can initiate TH1 polarization of antigen-stimulated CD4 T cells (Martin-Fontecha et al., 2004) by inducing the fate-determining transcription factor T-bet (Afkarian et al., 2002). In addition, IL-12 from DCs promotes fate stabilization of T cells polarizing to TH1 cell by promoting their secretion of IFNγ. Given the critical roles of IL-12 and the fact that its induction by TLR ligands was highly dependent on direct TLR stimulation in DCs, it is very likely that IL-12 was an essential cytokine for promoting TH1 polarization under these circumstances. Thus, TLR-MyD88 signaling in DCs likely contributes to adaptive TH1 immune responses to immunization with soluble antigen and TLR ligands in two main ways: by contributing to DC maturation and by inducing IL-12 production.
We found that the physical form of the TLR ligand had a large effect on the ability of different cell types to contribute to the immune response in vivo. CpG presented in an aggregated form by complexing it with the cationic lipid DOTAP induced a potent type 1 IFN response compared to CpG alone, as reported previously (Honda et al., 2005). In contrast to the strong dependence of MyD88 function in DCs for the cytokine responses to soluble CpG, MyD88 function in both DCs and non-DC cell types made important contributions to this response. Indeed, whereas when soluble CpG was injected i.v., it was primarily CD8α+ DCs in the spleen that made cytokines initially, when CpG/DOTAP was injected i.v., both CD8α+ and CD8α− DC subsets responded rapidly, as did a F4/80+ cell type that may be the inflammatory monocyte. Interestingly, the uptake of CpG in these experiments was substantially enhanced when CpG was complexed with DOTAP. The altered spectrum of responding cell types may relate to changes in the mechanism of cell uptake, as it has been reported that CpG/DOTAP complexes enter cells through the endocytic pathway (Zabner et al., 1995). Interestingly, interaction of self-DNA with a cationic amphipathic antimicrobial peptide LL37 has recently been implicated in the pathogenesis of psoriasis. This complex, which may be similar in its action to the CpG/DOTAP aggregate studied here, was found to greatly enhance activation of pDCs in the affected skin of patients with psoriasis (Lande et al., 2007). It is likely that many microbial and endogenous TLR ligands exist in aggregated or particulate forms, so the ability of both DCs and other myeloid cell types to respond to TLR ligands presented in these complex forms may be relevant to many biological situations.
The ability of CpG/DOTAP to induce large amounts of type 1 IFNs from pDCs, as well as from other non-DC cell type is likely to explain its ability to induce strong IFNγ production from NK cells, robust DC maturation, and a vigorous TH1 response in the absence of MyD88 expression in DCs. Other have reported that type 1 IFNs can synergize with IL-18 to induce IFNγ even in IL-12-deficient splenocyte cultures (Freudenberg et al., 2002), and we found that the NK cell IFNγ response to CpG was largely dependent on the expression of type 1 IFN receptors. In addition, type 1 IFNs are known to be able to induce maturation of DCs (Hoebe and Beutler, 2004). Indeed, it has been shown that co-administration of IFNα with antigen induces delayed-type hypersensitivity (Gallucci et al., 1999), as well as IgG2a antibody production (Le Bon et al., 2001), both of which are typical TH1 responses. Thus, the robust production of type 1 IFNs by cell types other than DCs is likely to explain why MyD88 function in DCs was not necessary for DC maturation or for a vigorous TH1 response in DC-MyD88 KO mice immunized with OVA + CpG/DOTAP.
In our experiments, LPS had a behavior that was very similar to that of CpG/DOTAP. LPS is an amphipathic component of gram-negative bacterial cell walls that forms large aggregates in solution, so it is not surprising that is has the ability to induce robust cytokine responses from DC subsets and also from additional myeloid cell types similarly to CpG/DOTAP. However, it is worth noting that LPS, like CpG/DOTAP, neither induced IFNγ from NK cells and NKT cells (Supplemental Figure S3) nor promoted TH1 polarization in MyD88 KO mice (Schnare et al., 2001). Although TLR9 and TLR4 both signal via MyD88, TLR4 also signals via the TRIF adaptor and this pathway induces abundant type 1 IFNs. Indeed, previous work showed that LPS stimulation of MyD88 KO mice failed to promote TH1 responses despite a vigorous type I IFN response and clearly evident DC maturation (Pasare and Medzhitov, 2004). Thus, type I IFN production alone is insufficient to drive TH1 responses. These results suggest that one or more MyD88-dependent cytokines in addition to type 1 IFNs coming from cells other than DCs are also required for the innate IFNγ and TH1 response seen in DC-MyD88 KO mice immunized with CpG/DOTAP or LPS. IL-18 is known to synergize with type 1 IFNs or IL-12 to induce IFNγ (Freudenberg et al., 2002; Nakanishi et al., 2001) and therefore is a strong candidate for the additional MyD88-dependent cytokine required for these responses.
In any case, our results clearly show that the functional fate of a T cell is not only affected by the DC that the T cell is interacting with, but also by surrounding cells responding via TLRs and MyD88. These surrounding cells include other DCs and non-DC cell types in the infected tissues. This trans effect may allow cooperation between cell types expressing different sets of TLRs. For example, CD8α+ DCs, which do not express TLR7 (Edwards et al., 2003), can receive important cytokine signals from pDCs, which secrete type 1 IFNs after TLR7 stimulation, allowing them to activate T cells. Other recent studies have also provided evidence for regulatory effects of macrophages on DCs for polarizing T cell differentiation. For example, lamina propria macrophages from the gut were found to express anti-inflammatory cytokines even after TLR stimulation in vitro and to promote development of FoxP3+ regulatory T cells, which restrain immune responses to commensal microbes and dietary antigens, whereas DCs from the same location responded to the same stimulus by producing proinflammatory cytokines and promoting IL-17 producing T cell responses (Denning et al., 2007). Clearly, much remain to be learned about how different types of innate immune cells communicate with one another and combine to direct the nature of the adaptive immune responses.
In summary, our results indicate that MyD88-dependent signaling in both DCs and non-DC cell types can support TH1 differentiation depending on the type of TLR stimulation. Whereas direct TLR stimulation is likely to be the most efficient way for activating DCs and for activating adaptive responses, we have found that other cell types stimulated with TLR ligands in complex forms secrete substantial amount of cytokines that can make important contributions to both innate and adaptive immune responses. It should be very interesting to use the mice described here to dissect further the role of TLR signaling in different cell types for activation of adaptive immune responses in more complicated and biologically important situations, such as infections with pathogens and autoimmune diseases.
A conditional allele of myd88 was created in mouse E14 embryonic stem (ES) cells following standard procedures (see Supplemental Experimental Procedures for details). Out of 400 ES cell clones screened, 5 had recombined into the endogenous myd88 locus. Three independent lines of the homologously recombined ES cells were injected into blastocysts from B6 mice by the UCSF mouse genetic core facility, and animals with high levels of incorporation of the 129 ES cells into the embryo were obtained. The mice were bred and germ line transmission of the targeted allele was obtained in mice originating from the three ES cell lines. The offspring were then bred to ACTB-FLPe mice, which express FLP recombinase (Rodriguez et al., 2000), to remove the neomycin resistance cassette. This leaves the conditional allele with two loxP sites and one residual frt site.
B6 (000664; C57BL/6J) and B6-Thy1.1 (001317; B6.Cg-Igha Thy1a Gpi1a/J) mice were from Jackson Laboratory. ACTB-FLPe mice (Rodriguez et al., 2000) were obtained from G. Martin (UCSF). CD11c-Cre mice have been described (Caton et al., 2007), and the mice used in this study were backcrossed to B6 for at least 6 generations. Vav-Cre mice (de Boer et al., 2003) were a gift of D. Kioussis (National Institute for Medical Research, London, GB). myd88−/− mice (Adachi et al., 1998) were originally from S. Akira (Osaka University, Osaka, Japan), and were backcrossed to B6 for 10 generations in our colony. IL-12p35−/− mice (Mattner et al., 1996) were obtained from R. Locksley (UCSF). Ifnar−/−mice (Muller et al., 1994) were obtained from M. Matloubian (UCSF). OT-II mice (Barnden et al., 1998) were bred to B6-Thy1.1 mice, and the F1 male mice were used as the donors of TCR-transgenic CD4 T cells.
The myd88flox mice used in this study were backcrossed to B6 for at least 6 generations, and then crossed to CD11c-Cre mice or Vav-Cre transgenic mice. In this study, myd88flox/flox or myd88flox/null mice containing the CD11c-Cre transgene (referred to as DC-MyD88 KO) or the Vav-Cre transgene were used, and myd88flox/flox or myd88flox/null mice were used as controls.
All experimental mice were used at 8–12 weeks of age and were sex-matched and age-matched (within 2 weeks) within experiments. All animals were housed in a specific pathogen free animal facility at UCSF under conditions that meet institutional animal care and use committee (IACUC) and NIH guidelines.
CpG oligodeoxynucleotide 1826 containing a phosphorothioate backbone (CpG) and Cy5.5-labeled CpG were purchased from Integrated DNA Technologies (IDT). To make the CpG/DOTAP complex, 25 μg of CpG was diluted in 75 μl of 20 mM HEPES-buffered saline (HBS) and then mixed with 50 μg of Dioleoyloxy-trimethylammonium-propane-methylsulfate (DOTAP) (Roche) diluted to 75 μl with HBS for 15 minutes before injection. Ultra pure LPS (E.coli 0111:B4), Pam3CSK4 and imiquimod (R837) were purchased from Invivogen. Salmonella typhimurium flagellin was purified from a fljB−/fliC+ strain (TH4778, kindly provided by K. Hughes (University of Utah, UT)), following a protocol provided by K.D. Smith (University of Washington, WA). Chicken ovalbumin was purchased from Sigma-Aldrich, and endotoxin was removed by Triton X-114 treatment as described (Aida and Pabst, 1990). All reagents were free of endotoxin as determined by the Limulus Amebocyte Lysate Test (BioWhittaker).
For assessing the deletion efficiency in particular cell types of the DC-MyD88 KO mice, genomic DNA was extracted from FACS-purified cells, and the residual amount of the “floxed” region was quantified by Taqman PCR (see Supplemental Experimental Procedures for details). The amount of flox allele of each mouse was normalized to Edg-1. The genomic DNA from myd88flox/flox mice was used for the no-deletion control.
For quantifying cytokine induction after systemic administration of TLR ligands, mouse spleens were harvested and snap-frozen in liquid nitrogen. Total RNA was extracted using the RNeasy kit (Qiagen) with on-column DNase digestion. cDNA was transcribed from total RNA using the iScript cDNA Synthesis Kit (Bio-Rad). Transcripts were quantified by PCR using iTaq SYBR Green Supermix with ROX (Bio-Rad), and the levels of cytokine transcripts were normalized to the levels of HPRT mRNA. The induction of cytokine mRNA was expressed as a ratio between the mRNA level of the TLR ligand-treated mice and that of the vehicle-treated control mice. All primers (Supplemental Table S3) were obtained from IDT.
The levels of cytokines in serum were analyzed by using standard sandwich ELISA kits (BD Biosciences).
Relative levels of OVA-specific immunoglobulin isotypes in serum were determined by ELISA with horseradish peroxidase (HRP)-conjugated anti-mouse IgM, total IgG, IgG1, IgG2b or IgG2c reagents (Southern Biotech) to detect immunoglobulin bound to the OVA-coated plates. Antibody titers were determined as the reciprocal of the dilution that gave an optical density value (450–570 nm wavelength) that was more than ten times of the standard deviation above the mean value of the negative control wells.
For surface staining of DCs, single cell suspensions were prepared from spleens with the Digestion medium (see Supplemental Experimental Procedures). The cells were then stained in ice-cold FACS buffer (PBS supplemented with 2μM EDTA, 1% heat-inactivated FCS, and 0.02% sodium azide). Anti-CD16/CD32 antibody (Ab) (2.4G2, BD PharMingen) was used to block non-specific antibody binding. All fluorochrome-conjugated monoclonal antibodies were purchased from BD PharMingen or eBioscience.
To detect in vivo IL-12, TNFα, or IFNγ expression in particular cell types, splenocytes were prepared and surface markers were stained as described above with the exception that all media contained 10μg/ml brefeldin A (Sigma-Aldrich). Then the cells were fixed and permeabilized using the Cytofix/Cytoperm kit (BD Biosciences). Intracellular IL-12 was detected by phycoerythrin- or allophycocyanin-labeled anti-IL12p40/p70 Ab (C15.6, BD PharMingen), TNFα was detected by phycoerythrin-labeled anti-TNFα Ab (MP6-XT22, BD PharMingen), and IFNγ was detected by allophycocyanin-labeled anti-IFNγ Ab (XMG1.2 BD PharMingen).
To assess ex vivo T cell IFNγ expression, single cell suspensions were prepared from the spleen, and 5×106 cells/ml were cultured for 4 hours in complete RPMI-1640 medium (10% heat-inactivated FCS, 25mM HEPES, 1mM L-glutamine, 50μM 2-mercaptoethanol) containing 100pg/ml PMA and 1ng/ml ionomycin. 10μg/ml brefeldin A was added for the last 2 hours of culture. Then the intracellular IFN-γ was stained as described above.
All data were collected on a LSRII flow cytometer (Becton Dickinson) and were analyzed with FlowJo software (TreeStar).
5×105 purified OT-II T cells were labeled with 5μM carboxyfluorescein diacetate succinimidyl ester (CSFE) (Molecular Probes) for 8 minutes at 37oC, and then transferred into sex-matched recipient mice by intravenous injection 1 day before immunization.
For in vivo stimulation, mice were injected with TLR ligands either intravenously (CpG 25 μg, Cy5.5-labeled CpG 25μg, LPS 25μg, Pam3CSK4 50μg, flagellin 20μg or CpG/DOTAP complex 25μg/50μg) or intra-peritoneally (imiquimod 150μg).
Mice were immunized with 50μg OVA mixed with 25μg CpG or with CpG/DOTAP (25μg/50μg) intraperitoneally for antibody response, or intravenously for T cell responses. Sera and lymphoid organs were collected at the indicated times.
Statistical significance was calculated with an unpaired Student’s t-test or Mann-Whitney U-test. All P values of 0.05 or less were considered significant.
We thank N. Killeen, V. Nguyen, and A. Kuroda for assistance in creating the myd88flox allele; W. Hazenbos for providing the Vav-Cre mice; S. Watson, J. Jarjoura and C. McArthur for assistance with cell purification by flow cytometry; L. Kuzmich and J. Lyandres for assistance with the mouse colonies; Y. Xu for helping with RT-PCR; L. Lee for providing low LPS flagellin; A. Abbas, J. Bluestone, Z. Hua, L. Lanier, and R. Locksley for helpful discussions; A. Gross for comments; C. Lowell and L. Fong for critical reading of the manuscript.
This research was supported by awards from the Academic senate of UCSF, the Sandler Family Supporting Foundation, and the National Institute of Health (R01 AI072058) to A.L.D., and from the Sandler Foundation and the National Institutes of Health (AI067804) to B.R. B.H. is a recipient of an Arthritis Foundation Postdoctoral Fellowship.
COMPETING INTERESTS STATEMENT
The authors declare they have no competing financial interests.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.