Full correspondence: Dr. Baohua Zhou, Department of Pediatrics and Department of Microbiology and Immunology, Indiana University School of Medicine, HB Wells Center for Pediatric Research, 1044 West Walnut Street, Room 272, Indianapolis, IN 46202, USA
Thymic stromal lymphopoietin (TSLP) plays important roles in the pathogenesis of allergic diseases. Whether and how TSLP is involved in the initial priming of T helper type-2 (Th2) differentiation against harmless antigen remains unclear. Using an intranasal sensitization protocol with OVA and LPS, we showed that TSLP signaling is required for low-dose LPS-induced Th2 inflammation, but not for high-dose LPS-induced Th1 immunity. We further demonstrated that low-dose LPS-activated bone marrow-derived dendritic cells expressed relatively high Tslp but low Il12a, and were able to prime naïve DO11.10 T cells to differentiate into Th2 cells in a TSLP-dependent manner. After transfer into wild-type recipient mice, the low-dose LPS-activated OVA-loaded dendritic cells (DCs) induced airway eosinophilia, but primed neutrophil-dominated airway inflammation when TSLP-deficient DCs were used. These studies demonstrate that TSLP released by DCs in response to a low concentration of LPS plays a role in priming Th2 differentiation and thus may serve as a polarizing third signal, in addition to antigen/MHC class II and co-stimulatory factors, from antigen-presenting DCs to direct effector T-cell differentiation.
Asthma is a chronic pulmonary disease of dysregulated immune responses against commonly inhaled innocuous antigens, which may be caused by a combination of environmental and genetic factors . Studies of patients and animal models demonstrated that T helper type-2 (Th2) effector cells producing cytokines IL-4, IL-5, and IL-13 contribute to many of the pathophysiological features of asthma, including airway inflammation, mucus hypersecretion, and airway hyperresponsiveness (AHR) .
Adaptive immune responses depend on signals from innate immune cells, particularly the professional antigen-presenting dendritic cells (DCs). In addition to providing TCR ligands and costimulation, DCs are able to secrete IL-12 upon activation of pattern recognition receptors by pathogen-associated molecular patterns, priming naïve CD4+ T cells to differentiate into Th1 cells [2, 3]. How DCs control Th2 differentiation, however, is not well understood. IL-4 is the most important cytokine for Th2 differentiation in vivo and in vitro [4-7]. Unlike the Th1 polarizing cytokine IL-12, the early source of IL-4 that initiates Th2 differentiation has been debated. Natural killer T cells  and basophils [9, 10] produce large amounts of IL-4 when activated and are essential for some allergic responses in vivo [11, 12], though the importance of these cells in Th2-cell induction or merely orchestrating/mediating the effector response in vivo is unclear.
Recent studies showed that thymic stromal lympho-poietin (TSLP) is critical for allergic inflammation in humans  and mice . TSLP is highly expressed in acute and chronic atopic dermatitis lesions and airways of allergic asthma patients [13, 15, 16]. Overexpression of TSLP in the lung leads to eosinophilic airway inflammation and hyperactivity , while TSLP overexpression in the skin results in skin inflammation characteristic of atopic dermatitis [17, 18]. TSLP receptor-deficient (Tslpr–/–) mice are protected from developing allergic airway inflammation [14, 19] and allergic skin inflammation . Human TSLP strongly activated peripheral blood derived DCs in vitro to upregulate MHC class II and other costimulatory molecules lead to Th2 differentiation of allogeneic CD4+ T cells in vitro [13, 21]. In mice, TSLP is able to directly act on naïve CD4+ T cells to promote Th2 differentiation and/or IL-4 secretion in vivo and in vitro [11, 20, 22]. Transferring wild-type (WT) CD4+ T cells into Tslpr–/– mice restores their allergic responses in murine asthma and atopic dermatitis models [19, 20], supporting a function of TSLP through a direct action on CD4+ T cells. However, TSLP is believed to be mainly produced by epithelial cells of peripheral tissues [20, 23-26]. Thus, it is not clear whether and how TSLP is involved in primary immune response (i.e. sensitization), which would require the presence of TSLP in the draining lymph nodes at the time of T-cell activation.
LPS, a cell-wall component of Gram-negative bacteria, is ubiquitously present in the environment, and known to induce DC maturation and production of IL-12 and IFN-γ to drive Th1 immunity . Several studies demonstrated that airway Th1 or Th2 responses to inhaled antigen are determined by levels of LPS [26, 28, 29]. While inhalation of high concentration of LPS (HiLPS) induces IL-12 expression leading to Th1 response, animals sensitized with low-dose LPS (LoLPS) fail to induce IL-12 in DCs resulting in Th2 airway inflammation after challenge. In this study, we showed that TSLP signaling was only required for LoLPS-induced Th2-biased airway inflammation, but not HiLPS-induced Th1 inflammation. Our data further demonstrated that LPS activated bone marrow-derived dendritic cells (BMDCs) to express TSLP, which provided a polarizing signal to prime Th2 differentiation in vitro. When transferred intranasally (i.n.) to naïve mice, the LoLPS-activated, OVA-loaded DCs primed Th2 sensitization dependent on TSLP derived from the DCs. Thus, in addition to an established role for DCs in Th1 polarization, our data suggest that DC-derived TSLP could act as a polarizing cytokine to initiate Th2 differentiation and allergic sensitization in vivo, and further define the Th1/Th2 paradigm of airway mucosal immunity in response to the natural adjuvant LPS.
TSLP is required for LoLPS-induced Th2 sensitization
Several studies have shown that generation of Th1- or Th2-mediated inflammatory responses to inhaled antigen is dependent on the concentration of LPS administered with antigen [26, 28, 29]. Specifically, HiLPS induces a Th1 response and LoLPS induces Th2. Since TSLP is critical for Th2-mediated allergic inflammation , we set out to determine whether TSLP plays a role in LoLPS induced allergic airway inflammation. BALB/c mice and Tslpr–/– mice were sensitized i.n. with 100 μg OVA + 20 ng LPS for 3 days. After i.n. OVA challenge, BALB/c mice developed airway inflammation with significantly increased total cell numbers (Fig. 1A) and eosinophils (Fig. 1B) in the bronchoalveolar lavage (BAL) as compared with nonsensitized mice (OVA only without LPS at sensitization). The sensitized WT mice also exhibited AHR (Fig. 1C). In contrast to WT mice, Tslpr–/– mice showed no significant airway inflammation and AHR, compared with nonsensitized mice (Fig. 1A to C). Increased serum IgE, whose synthesis heavily relies on combined effects of Th2 cytokines , is closely related to allergic or atopic diseases. Consistent to their much reduced allergic airway inflammation, Tslpr–/– mice showed no increase in total serum IgE compared with control mice (Fig. 1D). We next isolated cells from lung draining lymph nodes and stimulated with anti-CD3/CD28. Only cells from WT animals produced significantly increased IL-4 (Fig. 1E).
Since even WT animals sensitized with 20 ng LPS showed only a modest increase in total BAL cells (Fig. 1A) and eosinophils (Fig. 1B), we increased the LPS dose used at sensitization. When sensitized with 100 μg OVA in the presence of 50 ng LPS, Tslpr–/– mice indeed showed significant airway inflammation with similar total cell counts in BAL to that seen in WT mice (Fig. 2A). Surprisingly, we found that types of airway inflammation in Tslpr–/– mice and WT mice were completely different. While WT mice displayed typical eosinophilic airway inflammation, Tslpr–/– mice showed neutrophil-dominated infiltration in the airways with significantly higher neutrophils than eosinophils in the BAL (Fig. 2B). These data suggested that instead of the Th2-biased immune response in WT mice, 50 ng LPS at sensitization primed a Th1-biased immune response when TSLP signaling was interrupted. To further confirm the Th2 to Th1 shift in Tslpr–/– mice, we examined cytokine expression by restimulated cells isolated from lung draining lymph nodes. Consistent with the composition of airway infiltrating leukocytes, cells from WT animals produced more IL-4 (Fig. 2C) while Tslpr–/– cells secreted a higher concentration of IFN-γ (Fig. 2D).
TSLP is not required for HiLPS induced Th1 sensitization
Next, we tested the role of TSLP in a HiLPS-primed Th1 response. Mice were sensitized i.n. with 100 μg OVA + 10 μg LPS and challenged i.n. with 50 μg OVA. Consistent with published data [26, 28, 29], high-dose LPS during sensitization primed a Th1-biased airway immune response dominated by neutrophils regardless of TSLP signaling (Fig. 3A and B). Intracellular staining of total BAL cells demonstrated that CD4+ T cells in the airways were Th1 polarized in both WT and Tslpr–/– mice (Fig. 3C). Consistent with intracellular staining of the BAL cells, anti-CD3 restimulation of cells isolated from lung draining lymph nodes also showed a low IL-4 (Fig. 3D) and high IFN-γ (Fig. 3E) cytokine expression profile in HiLPS induced allergic response in both WT and Tslpr–/– mice.
DC-derived Tslp expression promotes Th2 differentiation in vitro
Airway DCs are essential for controlling effector T-cell response in allergen sensitization [3, 28]. To delineate the mechanism by which TSLP primes Th2 differentiation, we examined the ability of BMDCs to express TSLP when stimulated by LPS. Consistent with a recent finding that TSLP can be produced by mouse and human DCs [32, 33], we also found that BMDCs were able to express TSLP in response to LPS stimulation. In response to 10 ng/mL LPS, BMDCs quickly upregulated Tslp expression, reaching a peak around 4-8 hr (Fig. 4A), a time-course similar to expression of the IL-12 components Il12a and Il12b expression (Fig. 4A and data not shown). Surprisingly, unlike Il12a whose expression increased with increasing dose of LPS, Tslp expression followed a bell curve pattern and its peak expression was seen between 10 – 100 ng/mL LPS (Fig. 4B).
Next, we examined the role of DC-derived TSLP in Th2 differentiation in vitro. BMDCs were pulsed with 10 ng/mL LPS and 100 μg OVA overnight, washed and cultured with WT or Tslpr–/– DO11.10 CD4+ T cells in the presence of 100 μg OVA or 10 μg OVA323-339 peptide for 5 days. After the culture, helper T-cell polarization was determined by intracellular cytokine staining. As shown in Fig. 5, WT DO11.10 CD4+ T cells showed strong Th2 polarization with IL-4+/IFN-γ+ cell ratio of 3.4 ± 0.6 whereas Tslpr–/– cells exhibited modest Th1 polarization with IL-4+/IFN-γ+ cell ratio of 0.7 ± 0.2. To rule out the possibility that Tslpr–/– cells were developmentally predisposed to Th1 differentiation rather than the role of DC-derived TSLP, we cultured WT DO11.10 CD4+ T cells with conditioned BMDCs as above in the presence of anti-TSLP neutralization antibody  and observed that neutralizing TSLP reduced IL-4+ cells and increased IFN-γ+ cells in the culture (Fig. 5C).
Priming of Th2 response by LoLPS-conditioned DCs relies on DC-derived TSLP in vivo
Based on our in vitro results (Figs. 4 and 5), we hypothesized that BMDCs stimulated with 10 ng/mL LPS, expressing high TSLP and low IL-12, would prime Th2 sensitization when transferred into recipient mice. To test this hypothesis, BMDCs were pulsed overnight with 10 ng/mL or 1 μg/mL LPS in the presence of 100 μg/mL OVA, washed three times and i.n. transferred into WT BALB/c animals (2 × 106 cells/mouse). Ten days after the cell transfer, mice were challenged i.n. with OVA for three times and airway inflammation examined. We observed that transferring BMDCs matured with 10 ng/mL LPS (LoLPS-DCs) or 1 μg/mL LPS (HiLPS-DCs) induced similar airway inflammation after challenge (Fig. 6A). However, differential cell counts indicated that LoLPS-DCs induced eosinophilic airway inflammation while transfer of HiLPS-DCs led to neutrophilic airway inflammation (Fig. 6B). Consistent with the BAL cell composition, CD4+ T cells in BAL were polarized into Th2 or Th1 determined by LPS doses used to mature BMDCs (Fig. 6C). These results indicated that varying concentrations of LPS acting on DCs alone are sufficient to determine Th1/Th2 polarization in the airways.
Next, we transferred low dose LPS (10 ng/mL) conditioned, OVA-pulsed WT and TSLPR-deficient BMDCs into WT BALB/c mice and challenged these mice with OVA three times. The mice exhibited similar airway eosinophilia and Th2 polarization of CD4+ T cells in BAL (Fig. 6D–F). The fact that both WT and TSLPR-deficient BMDCs primed Th2 sensitization suggest that Th2 differentiation of CD4+ T cells was not determined by the ability of antigen-presenting DCs to respond to TSLP in this experimental model.
To directly examine the role of DC-derived TSLP in priming Th2 sensitization in vivo, we generated TSLP knockout (Tslp–/–) mice (C57BL/6 background) with targeted embryonic stem cells purchased from the Knock-Out Mouse Project (KOMP). BMDCs from these mice expressed Il12a but no detectable Tslp expression, after culture for 8 h with 20 ng/mL LPS (Fig. 7A). After being pulsed with 20 ng/mL LPS and 100 μg/mL OVA, WT or Tslp–/– DCs were washed three times and transferred i.n. into WT C57BL/6 mice. Since C57BL/6 mice were used in this experiment, they were challenged six times with 50 μg OVA 10 days after DC transfer. Under these conditions, both WT DCs and Tslp–/– DCs were able to induce mild airway inflammation without significant difference in total BAL cell counts (Fig. 7B). While WT DCs induced eosinophilic airway infiltration with significantly higher eosinophils than neutrophils in BAL, Tslp–/– DC transfer induced neutrophilic airway inflammation with more neutrophils than eosinophils in BAL (Fig. 7C). The composition of the inflammatory cells infiltrating the airways suggested that the ability of low LPS-stimulated DCs to prime Th2 differentiation in vivo was dependent on DC-derived TSLP.
Antigens presented by mature DCs result in the activation of naïve CD4+ T cells, which differentiate into helper T-cell subsets in the presence of polarizing cytokines. Although the DC-derived polarizing signals such as IL-12 involved in Th1 differentiation are well defined, much less is known about those required for the generation of Th2 responses. DCs activated and matured by LPS or other TLR ligands do not produce the primary Th2 polarizing cytokine IL-4. Thus, it has been proposed that Th2 development occurs by default in the absence of IL-12 from DCs activated by low dose of LPS . In this report, we present evidence showing that DC-derived TSLP in response to low LPS might serve as a polarizing cytokine to induce Th2 differentiation during primary immune responses.
While the requirement for TSLP in the pathogenesis of allergic airway inflammation has been documented [14, 19], the use of peritoneal OVA/alum to initiate strong Th2 responses is not the optimal model for examining T cell–APC interactions. Sensitization with antigen and LPS through the respiratory system, the anatomic site that would normally encounter environmental allergens and LPS, provides a particularly relevant model to investigate immune responses in the pathogenesis of asthma . Here we demonstrate that TSLPR-deficient mice failed to initiate a Th2 response when i.n. sensitized with OVA + 20 ng LPS, but exhibited normal Th1 immunity in response to OVA + 10 μg LPS sensitization. The skewing from Th2 to Th1 immunity by Tslpr–/– mice in response to 50 ng LPS strongly supports a critical role for TSLP in promoting Th2 polarization in vivo.
TSLP is able to directly act on naïve CD4+ T cells to promote Th2 differentiation and/or IL-4 secretion [11, 20, 22]. TSLP treatment leads to immediate, direct Il4 gene transcription and drives Th2 differentiation in the absence of exogenous IL-4 . However, IL-4 blockade inhibited TSLP-mediated Th2 differentiation, demonstrating that endogenous IL-4 is involved via an autocrine mechanism. Yet, it is not clear whether and how TSLP is involved in the initiation of Th2 sensitization in vivo since TSLP is produced by epithelial cells of peripheral tissues in response to TLR agonists and proinflammatory cytokines , [23-26]]. Studies by Sokol et al.  showed that basophils activated by papain were recruited to draining lymph nodes and expressed TSLP and IL-4. Unlike DCs that were recruited into draining lymph nodes 18 h after papain immunization, significant basophil numbers were recruited to draining lymph nodes 3 days after immunization when substantial IL-4-eGFP+ Th2 cells already existed in the draining lymph nodes . Thus, TSLP and IL-4 expressing basophils are most likely augmenting, rather than initiating, Th2 differentiation in vivo. Such a notion was supported by a recent study showing that basophils did not interact with antigen-specific T cells in lymph nodes and were not required for Th2 priming in vivo .
Airway DCs are essential for controlling effector T-cell response in sensitization, the primary immune response [3, 28]. DCs isolated from OVA + HiLPS sensitized mice induced Th1 polarization and DCs isolated from OVA + LoLPS sensitized mice induced Th2 polarization . The fact that LoLPS induced stronger eosinophilic airway inflammation in mice with intact TLR4 on hematopoietic cells than in mice with intact TLR4 on lung stromal cells  supports the existence of the LoLPS induced signal(s) on airway DCs to prime Th2 sensitization in vivo. A recent study showed that various types of DCs produced TSLP in response to TLR agonist and allergen house dust mite extract [32, 33]. Importantly, lung DCs expressed even higher TSLP than lung epithelial cells . We show in this report that lower concentrations of LPS activated DCs expressed relatively higher Tslp but lower Il12a. Although Tslp mRNA expression peaked at 2–10 h and decreased to a relatively low level (approximately tenfold increase over nonactivated DCs) 16 h after LPS treatment, the LoLPS-primed DCs were capable of inducing Th2 polarization in a TSLP-dependent manner in vitro. Arguably, these antigen-loaded, TSLP-expressing DCs would be able to prime Th2 sensitization when they reach the draining lymph nodes. Indeed, mature TSLP-expressing DCs were detected in the draining lymph nodes after immunization . By i.n. transfer of OVA-loaded DCs, we demonstrated that the dose of LPS used to induce DC maturation determines Th1/Th2 sensitization of the recipient mice. More importantly, the ability of LoLPS-matured DCs to induce Th2 sensitization in vivo was dependent on the ability of the DCs to express TSLP. Consistent to our finding that DC-derived TSLP directly acts on CD4+ T cells to direct T helper cell differentiation, a recent study demonstrated that DC-derived TSLP directly acts on CD4+ T cells to suppress Th17 differentiation while fostering Foxp3+ Treg cells in the gut .
Taken together, our data suggested parallel pathways for initiating Th1/Th2 polarization involving IL-12/IFN-γ and TSLP/IL-4 action and the balance of DC-derived IL-12 and TSLP, depending on the dose of LPS, determines the nature of airway immune responses against harmless antigens. In sensing HiLPS, DCs express relatively high IL-12 and low TSLP, which stimulate naïve CD4+ T cells to differentiate into IFN-γ-producing Th1 cells . Autocrine action of endogenous IFN-γ would further amplify Th1 development . In the LoLPS condition, DCs express relatively high TSLP but low IL-12, which could stimulate naïve CD4+ T cells to differentiate into IL-4-producing Th2 cells [11, 22]. Autocrine action of endogenous IL-4 would further amplify Th2 development .
Materials and methods
BALB/c, C57BL/6 WT mice and DO11.10 TCR transgenic mice were purchased from Jackson Laboratories (Bar Harbor, Maine, USA). Tslpr-deficient (Tslpr–/–) mice were described previously . The Tslp targeted embryonic stem cells used to generate Tslp-deficient (Tslp–/–) mice were purchased from Knockout Mouse Project (KOMP). All mice were housed in specific pathogen-free conditions and all experiments were performed as approved by the Indiana University Institutional Animal Care and Use Committee.
Intranasal sensitization and challenge
Mice were anesthetized with ketamine/xylazine, and then sensitized i.n. with 100 μg OVA (Worthington Biochemicals) plus 20–50 ng LPS (Sigma-Aldrich) (LoLPS) or 10 μg LPS (HiLPS) in 40 μL PBS on days 0, 1, and 2. Mice received OVA were used as controls. Mice were challenged i.n. on day 14, 15, 18, and 19 with 50 μg OVA . Mice were then analyzed for AHR by whole body plethysmography  and sacrificed on day 21 for further analysis.
Evaluation of lung inflammation
Mice were euthanized, and bronchoalveolar lavage was performed as described previously . Briefly, lungs were washed three times with 1 mL cold PBS. BAL fluid fractions were centrifuged at 1400 × g for 5 min at 4°C. Pellets were resuspended and counted. Cytospin preparation were stained with modified Wright-Giemsa stain, and differential cell counts were evaluated by counting at least 200 cells for determination of relative percentage of each cell type in the BAL .
Isolation and restimulation of cells from lung draining lymph nodes
Mediastinal lymph node was dissected from sensitized and challenged mice, and single-cell suspensions were prepared and stimulated in vitro either with plate-bound 2 μg/mL anti-CD3 (17A2; Biolegend) for 24 h. Cell-free supernatants were analyzed for IFN-γ and IL-4 using commercially available ELISA kit (Biolegend and eBioscience). Lower detection limits were 1.0 pg/mL (IL-4) and 4.0 pg/mL (IFN-γ).
Generation of BMDCs
BMDCs were generated as described . RPMI 1640 medium was supplemented with 5% fetal bovine serum, 1 mM sodium pyruvate, 10 mM HEPES buffer, 100 U/mL penicillin, 100 μg/mL streptomycin, 50 μM 2-ME, and 20 ng/mL recombinant mouse granulocyte macrophage-colony stimulating factor (GM-CSF; Bio-legend, San Diego, CA). At day 9, they were pulsed in vitro overnight with OVA and LPS for in vivo experiment or directly used for in vitro experiment.
Asthma model using OVA + LPS-pulsed DCs
BMDCs were matured with 10–20 ng/mL (LoLPS-DCs) or 1 μg/mL (HiLPS-DCs) LPS in the presence of 100 μg/mL OVA. After wash, 2 × 106 BMDCs were transferred i.n. into the airways of naive anesthetized mice, as described previously . Ten days after immunization, mice were challenged i.n. with 50 μg OVA for three times (BALB/c mice) or 6 times (C57BL/6 mice).
Coculture of BMDCs with naïve CD4+T cells
Naïve CD4+CD62L+ T cells were isolated from spleens and lymph nodes of DO11.10 or Tslpr–/– DO11.10 mice (MACS isolation system; Miltenyi Biotec, Auburn, CA). BMDCs were activated by 10 ng/mL LPS plus 100 μg/mL OVA. After overnight incubation, the cells were seeded (5 × 104 cells/well in 1 mL) in the complete medium with naïve T cells (1 × 105 cells/well) for 5 days in 48-well plates in the presence of 10 μg/mL OVA323-339 peptide (GenScript) or 100 μg/mL OVA protein . After the coculture, CD4+ T cells were stained for expression of IL-4 and IFN-γ.
Intracellular cytokine staining and Abs
The following mAbs were purchased from Biolegend: CD4 (clone GK1.5), DO11.10 TCR (clone KJ1-26), IL-4 (clone 11B11), and IFN-γ (clone XMG1.2). Rat IgG1 (clone RTK2071), rat IgG2a (clone RTK2758), and rat IgG2b (RTK4530) were included as isotype controls. For the detection of intracellular cytokines, cells were stimulated with 50 ng/mL PMA and 500 ng/mL ionomycin for 6 h. Four hours prior to harvesting, monensin (Biolegend) was added to cultures to retain cytokines in the cytoplasm. Cells were then stained for IL-4 and IFN-γ production as described .
This work was supported by NIH grants R21 AI072617, R01 AI085046 to B.Z.
Conflict of interest
The authors declare no financial or commercial conflict of interest.