A specific role for TLR1 in protective T_H17 immunity during mucosal infection

We hypothesize that the nature of the immune response induced by TLR2 depends on the partner with which it associates. This hypothesis was based on our finding that TLR2/TLR6 ligands promoted tolerogenic DCs and regulatory IL-10 T cell responses in vitro and in vivo, whereas TLR1 primarily induced IL-12p40 production in DCs in vitro (DePaolo et al., 2008). However, we were unable to demonstrate a role for TLR1 in vivo, so we sought to reassess the role using Y. enterocolitica. Because this species normally infects orally, we administered bacteria via gavage to TLR1^−/−, TLR6^−/−, and littermate control (TLR1^+/− and TLR6^+/−) mice (Fig. 1, A–C). In accordance with a previous study demonstrating that TLR2 is involved during oral Y. enterocolitica infection (Sing et al., 2002), we confirmed that TLR2^−/− mice had a defect in IL-10 production and a reduced bacterial burden in the mesenteric LNs (MLNs); however, overall mortality was not affected (not depicted). Furthermore, TLR6^−/− and TLR1^−/− had opposing effects; absence of TLR6 dramatically increased survival (Fig. 1 A) and decreased bacterial burden (Fig. 1 B), whereas the absence of TLR1 led to decreased survival (Fig. 1 A), higher bacterial burden in MLNs (Fig. 1 B), and faster systemic dissemination (Fig. 1 C).

Our previous work has demonstrated that TLR1 had no impact on the survival of mice during systemic infection with Yersinia pestis, the causative agent of bubonic plague (DePaolo et al., 2008). We sought to test whether this difference was caused by the genetic makeup of the bacteria or by the route of infection. We decided to test the latter possibility first, as the intestinal environment is unique in that it expresses high levels of TGF-β and the vitamin A metabolite retinoic acid (RA). In support of this hypothesis, we found that TLR1^−/− mice infected systemically with Y. enterocolitica displayed no phenotype as compared with infected littermate controls (Fig. 1, D–F), even at doses of infection as low as 10³ CFU (not depicted). These data demonstrate that TLR1 and TLR6 play opposing roles during oral Y. enterocolitica infection. Furthermore, they indicate that the host uses different innate receptors to mount protective immunity against a pathogen depending on the route of infection.

We next wanted to determine whether the differences we observed in TLR1 involvement were linked to differences in the T_H responses induced during oral and systemic Y. enterocolitica infection. Cytokine analysis of TLR1^−/− mice and littermate controls revealed that T_H17 polarizing innate cytokines (Fig. S1 and see Fig. 3 D) and T_H17 cell responses (Fig. 2 A) were dependent on the presence of TLR1 and were selectively induced during mucosal Y. enterocolitica infection. In contrast, IL-12p70 (Fig. S1) and T_H1 responses (Fig. 2 B) were observed during systemic Y. enterocolitica infection and conserved in the absence of TLR1. We have previously shown that IL-10 T cell responses were dependent on TLR6 expression during peripheral infection (DePaolo et al., 2008), and in this study we demonstrate that TLR6 is also required during oral infection with Y. enterocolitica (Fig. 2, A and B), indicating that TLR6 is a critical innate receptor with tolerogenic properties in and outside of the intestinal environment,

The discovery that T_H17 cells are important for protection against oral but not systemic infection is in agreement with observations that IL-6^−/− mice have a more severe disease after oral (Dube et al., 2004) but not i.v. (Matteoli et al., 2008) infection with Y. enterocolitica. Furthermore, our findings also explain seemingly contradictory studies showing that infection in IL-12p40^−/− mice results in increased bacterial burden and dissemination (Hein et al., 2001), whereas administration of bioactive IL-12p70 has no effect in mice orally infected with Y. enterocolitica (Bohn and Autenrieth, 1996). Our data also demonstrate that TLR1 is critical for the induction of T_H17 in response to oral Y. enterocolitica infections. IL-17 was previously shown to be induced homeostatically by segmented filamentous bacteria (Ivanov et al., 2009) in addition to having a critical role in the protection against intestinal bacterial pathogens such as Citrobacter rodentium (Mangan et al., 2006; Ishigame et al., 2009). However, the molecular mechanism behind the generation of T_H17 responses at mucosal surfaces has remained poorly understood to this point. The importance of TLR1 in promoting T_H17 immunity in response to pathogens other than Y. enterocolitica or commensal bacteria remains to be determined.

The type of antibody response to an antigen is linked to the nature of the T_H cells induced. In particular, T_H17 and T_H1 cells have been shown to promote IgA (Jaffar et al., 2009) and IgG2a antibody responses, respectively (Stevens et al., 1988). In accordance with the T_H cells induced (Fig. 2, A and B), Yersinia-specific IgA antibodies were found in the feces of orally infected mice, whereas IgG2a antibodies were prominently found in the serum of systemically infected mice (Fig. 2, C–E). IgG1 was induced in response to both types of infection. Supporting a role for TLR1 during oral infection in IgA-mediated immunity against Y. enterocolitica, TLR1^−/− mice did not generate Yersinia-specific IgA antibodies (Fig. 2, C–E). The absence of IgA antibodies was associated with an increase in IgG1 antibodies, whereas IgG2a antibodies overall remained unchanged (Fig. 2, C–E). As anticipated, absence of TLR1 had no significant effect on the antibody response during systemic infection (Fig. 2, C–E).

To directly assess the link between T_H17 responses and protective immunity against Y. enterocolitica, we treated mice with a neutralizing antibody against IL-17 or an IgG2a isotype control. Mice receiving neutralizing IL-17 antibody began losing weight earlier (day 2 vs. day 5; not depicted) and had increased mortality (P < 0.05; Fig. 2 F). In line with a role for IL-17 in mucosal IgA responses (Jaffar et al., 2009), mice treated with neutralizing anti–IL-17 antibody had a significant decrease in fecal levels of anti-Yersinia IgA antibodies (Fig. 2 G). Furthermore, anti–IL-17–treated mice also had higher bacterial burden and a more disseminated infection with 2-log more bacteria in the spleen (SPL) 7 d after infection as compared to mice receiving isotype control antibody (Fig. 2 H). As anticipated, blocking IL-17 had no effect during systemic Y. enterocolitica infection (Fig. 2, F–H). The discovery that Y. enterocolitica selectively induces T_H17-dependent IgA responses during oral infection fits from a teleological perspective given that T_H17 responses preferentially induce secretory IgA antibodies (Gaboriau-Routhiau et al., 2009; Ivanov et al., 2009; Jaffar et al., 2009), which are highly resistant to the proteolytic enzymes present in the intestinal environment because of their secretory component and thus are uniquely capable of neutralizing luminal antigens (Endt et al., 2010; Hapfelmeier et al., 2010). This strongly emphasizes the importance of the tissue microenvironment in determining which type of T_H response is induced.

Because TLR1 is expressed and functions similarly between mucosal and systemic DCs (unpublished data; Edwards et al., 2003) and the conditioning of DCs determines the type of T cell response induced, we hypothesized that intestinal but not SPL DCs would induce T_H17 cell differentiation when stimulated with Y. enterocolitica lysates. A distinctive feature of small intestinal DCs is that they operate in an environment rich in TGF-β and vitamin A and express ALDHA2 (aldehyde dehydrogenase-1 member A2), a retinal dehydrogenase involved in the conversion of retinal into RA (Coombes et al., 2007; Sun et al., 2007). Expression of TLR2 on CD4 T cells has been shown to promote T_H17 cells (Reynolds et al., 2010). To directly examine the effect of TLR1 deficiency in DCs, we used a co-culture system with wild-type naive CD4 T cells and DCs from TLR1-deficient mice. As anticipated, when stimulated with Y. enterocolitica lysates, MLN DCs from TLR1^+/− mice induced T_H17 cells in vitro, whereas SPL DCs induced T_H1 cell differentiation (Fig. 3 A). In accordance with a selective role for TLR1 in T_H17 cell differentiation in vivo (Fig. 2, A and B), TLR1^−/− MLN DCs induced significantly fewer T_H17 cells than littermate control MLN DCs, whereas no differences were observed for T_H1 polarization (Fig. 3 A). We next determined whether RA and/or TGF-β would be sufficient to confer the ability to induce T_H17 responses to SPL DCs. As anticipated by previous studies showing that under inflammatory conditions RA acquires adjuvant properties and has the ability to promote T_H1 and T_H17 responses (DePaolo et al., 2011; Hall et al., 2011), incubation of SPL DCs with RA significantly promoted T_H1 and, to a lesser degree, T_H17 cell differentiation (Fig. 3 B). When TGF-β was added (reproducing the conditions found in the intestinal environment), T_H1 polarization was lost and T_H17 polarization was observed (Fig. 3 B); however, neutralization of TGF-β in the cultures was able to restore T_H1 polarization (not depicted). These results emphasize the critical role played by the tissue in determining which type of immune response will be mounted against a given pathogen. It also reaffirms that RA loses its tolerogenic properties under inflammatory conditions and suggests that depending on the level of TGF-β, it may preferentially promote T_H1 or T_H17 immunity.

We next assessed the in vitro contribution of TLR1 in the induction of innate cytokines involved in T_H1 and T_H17 polarization in the presence of Y. enterocolitica lysates (Fig. 3 C). We found that TLR1 signaling in DCs was involved in the production of IL-6 and IL-23 (Fig. 3 C) but not TGF-β (not depicted). These results are consistent with observations that TLR1-deficient mice displayed decreased levels of IL-6 and IL-23 (Fig. 3 D) but normal levels of TGF-β (not depicted) in MLNs 3 d after oral infection. In accordance with the data showing normal T_H1 responses in TLR1-deficient mice during systemic Y. enterocolitica infection (Fig. 2 B), we found no role for TLR1 in IL-12p70 induction in vivo (Fig. 3 D). However, the in vitro experiments indicated a possible role for TLR1 in IL-12p70 induction under conditions in which TGF-β was absent (Fig. 3 C).

Together, these data demonstrate that although TLR1 is important for the induction of IL-6 by DCs, the combinatorial effect of tissue-derived TGF-β and IL-6 is necessary for driving anti-Yersinia T_H17 responses. Finally, having demonstrated that TLR1 was required for induction of IL-6 and for T_H17 polarization, we also found that IL-6 was required for mucosal T_H17 but not for peripheral T_H1 responses during Y. enterocolitica infection (Fig. 3 E).

Given that we observed a more drastic and specific role for TLR1 in the induction of T_H17 responses in vivo than in vitro, we tested whether TLR1 had an additional role in vivo during oral Y. enterocolitica infection. Taking into account previous findings demonstrating that TLR2/TLR1 ligands imprint a gut-tropic phenotype on T cells (Wang et al., 2011) and low levels of IL-6 and IL-23 in MLNs of TLR1–deficient mice, we hypothesized that TLR1 may also have an effect on DC migration. In accordance with this hypothesis, mice deficient for TLR1 had a 50% reduction in percentage and total DC number in MLNs compared with littermate control mice (Fig. 4 A). In contrast, the absence of TLR1 did not affect the number of DCs in the SPLs of i.v. infected mice (Fig. 4 B). These data suggest that TLR1 controls induction of T_H17 responses during oral Y. enterocolitica infection by inducing IL-6 and Il-23 in DCs and allowing their migration to MLNs. We have preliminary evidence that TLR1 may be capable of regulating the production of chemokines necessary for DC recruitment (not depicted).

Altogether our study emphasizes that TLR2 is unique in that it possesses dual and opposing immunological functions in the intestinal mucosa when associated with TLR1 or TLR6. Given that TLR2/TRL1 and TLR2/TLR6 ligands are triacylated and diacylated lipopeptides, respectively (Omueti et al., 2005; Jin et al., 2007; Kang et al., 2009), there is the intriguing possibility that enzymes involved in their acylation may dictate whether they will act as pro- or antiinflammatory mediators. Alterations in the acylation status of LPS with similar consequences have been reported (D’Hauteville et al., 2002; Montminy et al., 2006). These observations suggest that by fine-tuning the activity of enzymes mediating acylation, bacteria may have the means to positively or negatively modulate the host immune response, thus allowing for controlled colonization of the host. This is in line with the more general concept that bacteria have evolved ways to promote both pro- and antiinflammatory responses in a concerted manner for their own benefit (Sansonetti and Di Santo, 2007).

Our findings also have implications in the field of vaccine development, as these data highlight the importance of accounting for both the type of pathogen and the site of action when selecting an immune response to maximize protection while avoiding collateral tissue damage. In particular, they stress that the route of vaccination is critical in determining the nature of the T_H immune response induced. Finally, in line with other studies (DePaolo et al., 2011; Hall et al., 2011), our observation that RA promotes T_H1 and T_H17 cell polarization in vitro in the presence of Y. enterocolitica lysates suggests that RA may be a useful adjuvant for vaccines that comprise proinflammatory innate ligands.

TLR1^−/− and TLR6^−/− mice were generously provided by S. Akira (Osaka University, Suita, Osaka, Japan) and then bred to C57BL/6 mice. TLR1^+/− and TLR6^+/− were used as controls, unless otherwise noted in figure legends. All mice were maintained at the University of Chicago, and all animal experiments were performed in accordance with institutional guidelines following experimental protocol review and approval by the Institutional Biosafety Committee and the Institutional Animal Care and Use Committee.

Y. enterocolitica strain 8081 was grown overnight at 26°C in tryptic soy broth (BD), and the culture was refreshed the next morning for 2 h. Bacterial density was measured by OD at 600 nM and diluted to 10⁶ CFU/ml in sterile PBS. Mice were administered 100 µl by intragastric gavage or i.v. In some experiments, mice were treated i.p. with 100 µg anti–IL-17 or anti-IgG2a (R&D Systems) or 100 µg anti–IL-6 or anti-IgG1 (eBioscience) beginning at the time of infection and then every other day for 5 d as previously described (Meng et al., 2009). Bacterial burden was determined by performing serial dilutions of MLNs or SPLs on tryptic soy agar plates (BD). Plates were grown for 48 h at 26°C.

96-well plates were coated overnight at 4°C with 10 µg/ml Y. enterocolitica lysate prepared as previously described (DePaolo et al., 2008). In brief, strain 8081 was grown as described in the previous section. The bacteria was pelleted and reconstituted in 0.5 ml sterile PBS. Silicon beads were added to the bacteria, and the solution was lysed using a BeadBeater (BioSpec) for 5 min. Protein determination was performed on the lysate, and 100 µl lysate was plated to make sure all the bacteria were killed. Fecal samples were prepared by collecting fresh fecal pellets and dissolving in 1 ml lysis buffer containing soybean trypsin. Serum was prepared via terminal bleed of the retroorbital plexus. Undiluted fecal supernatants and serum (diluted 1:2) were added to bacteria-coated plates for 2 h at room temperature. Plates were washed and then incubated with HRP-conjugated anti–mouse IgA (Santa Cruz Biotechnology, Inc.) or rabbit anti–mouse IgG1, rabbit anti–mouse IgG2a, or rabbit anti–mouse IgM followed by anti–rabbit HRP (BD). TMB substrate (Dako) was used for detection, and absorbance was read at 495 nm OD. Antibody titers were detected by ELISA and calculated according to the formula: (OD_450nm sample − OD_450nm of blank) × serum dilution.

MLN and SPL cells (0.5–10⁶) were incubated overnight with 10 µg/ml heat-killed Y. enterocolitica (HKY) and the next day were stimulated for 2 h with PMA and ionomycin in the presence of Golgi plug (BD). The cells were washed and incubated with anti–mouse FcRII/III (CD16/32 clone 2.4G2; BD) for 15 min at 4°C to prevent nonspecific Fc binding. Cells were washed in PBS with 10% BSA (Sigma-Aldrich) and labeled directly with monoclonal antibodies specific for lymphocyte surface markers CD4 PeCy5 (L3T4; BD), IFN-γ PE (XMG1.2; eBioscience), IL-17 PE (17B7; eBioscience), and IL-10 APC (JES5-16E3; eBioscience) at 4°C for 15 min. For CD11c experiments, MLNs and SPLs were digested as described in In vitro DC and T cell co-cultures. Cells were stained with CD45 APC-Cy7 (3011; BD), CD11c FITC (HL3; BD), CD3e PE (145-2C11; BD), and CD19 PE (ID3; BD). Expression was determined on a FACSCanto flow cytometer (BD) using FlowJo software (Tree Star).

MLNs or SPLs were homogenized in 1 ml PBS, and ELISA was performed according to product instructions from BD OptEIA kits (IL-10, IL-12p40, and IL-12p70) and R&D Systems kits (IL-23 and TGF-β).

MLN DCs and SPL DCs were isolated as previously described (DePaolo et al., 2011). In brief, MLNs or SPLs were dissociated using collagenase IV (Sigma-Aldrich), and monocytes were purified using a density gradient. CD11c⁺ cells were purified using CD11c-microbeads (Miltenyi Biotec). 5 × 10⁴ DCs were stimulated with 10 µg/ml Y. enterocolitica lysate generated through sonication, and 10⁵ CD4 T cells were harvested from SPLs of wild-type mice via CD4-positive selection (autoMACS; Miltenyi Biotec). Cultures were then stimulated with 10 nM RA or 10 nM RA plus 2 ng/ml TGF-β and in some experiments 1 µg/ml anti–TGF-β (clone 9016; R&D Systems), and supernatants were analyzed after 48 h for cytokines IFN-γ and IL-17 using ELISA.

Paired and unpaired Student’s t tests were used when noted in figure legends. Tests that had an interaction of P < 0.05 were considered significant. Wilcoxon log-rank test was used for analysis of survival.

Fig. S1 shows the levels of innate cytokines from TLR1^−/−, TLR6^−/−, or littermate control mice infected orally or i.v.

We thank Benjamin Sally for critical reading of the manuscript.

This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases K-award DK8275 (to R.W. DePaolo) and Crohn’s and Colitis Foundation of America Senior research award 2831 (to B. Jabri).

The authors declare no competing financial interests.

Author contributions: R.W. DePaolo conceived the idea, supervised all investigations, designed and performed experiments, analyzed data, and wrote the manuscript; S. Khakpour, Y. Sugiura, and W. Wang provided technical assistance; B. Jabri provided input into the conceptual development of the experiments and wrote the manuscript; K. Kamdar performed experiments and helped write the manuscript.