Eosinophils Regulate Dendritic Cells and Th2 Pulmonary Immune Responses following Allergen Provocation

Allergic asthma is a chronic inflammatory disease thought to be initiated by innate pulmonary immune responses that subsequently activate the adaptive immune system against specific environmental assaults. This adaptive immunity and the characteristically persistent Th2 inflammation are often dominated by the presence of activated T cells (1) and eosinophils (2) in the lung even in mild cases of asthma. In this paradigm, eosinophils and T cells are generally considered disparate mediators of the inflammation, with T cells acting as immune regulators and eosinophils as end-stage destructive effector cells. However, recent studies suggest that this perspective of eosinophil effector functions is too narrow. Specifically, in mouse models of allergic respiratory inflammation, eosinophils appear to play a role during the secondary immune responses leading to the activation and proliferation of Ag-specific memory T cells (3, 4) and the subsequent recruitment of newly formed T effector cell populations to the lung (5, 6).

Dendritic cells (DCs) have long been classified as “professional” APCs fundamental for both T cell activation and immune tolerance. Specifically, several elegant studies have demonstrated that lung DCs have a unique capacity to engulf exogenous Ag and migrate to lung draining lymph nodes (LDLNs) to present Ags to T cells (7). These DCs are uniquely necessary for the induction of allergic pulmonary pathologies in mouse models of allergic respiratory inflammation (8) and are increased in the lungs of asthmatic patients (9). Mouse models that conditionally deplete DCs by injection of targeted toxins during allergen provocation completely abolished activation of T cells and pulmonary inflammation, indicating DCs are essential for induced allergen-specific pathologies (10, 11). This dependency is highlighted by adoptive transfer studies of myeloid DCs into the lungs of sensitized mice and their restoration of allergic pulmonary inflammation, including eosinophil recruitment to the lung (8, 10).

Leukocytes other than DCs have also been shown to have the ability to promote and/or modulate the activation and polarization of T cells (12), and thus represent particularly problematic observations for paradigms suggesting the singular importance of DC-mediated events. In particular, two independent studies using genetically engineered mouse models deficient in eosinophils demonstrated that it was possible to correlate the loss of eosinophils with significant reductions in Th2 pulmonary pathologies [e.g., PHIL (5) and Δdbl-GATA1 (6)]. Mice with a partial eosinophil deficiency, such as IL-5 knockout (13) and IL-5/eotaxin double-knockout (4) mice, have also been described to have impaired Th2 responses to allergen. In addition, eosinophils have been shown to have DC-like functions, such as the expression of MHC class II (MHC II) and costimulatory receptors (14–16), a capacity to traffic to LDLNs (14, 17), and the ability to stimulate Th2 cytokine production from Ag-specific CD4⁺ T cells (3, 18). Eosinophils are also capable of releasing immune modulators such as Th2 cytokines (e.g., IL-4 and IL-25), Th1 cytokines (e.g., IL-12), or suppressive mediators (e.g., indoleamine 2,3-dioxygenase and TGF-β) (19, 20). Collectively, these data suggest that the presence of these granulocytes may elicit underappreciated immune regulatory functions in the lung.

In this report, we suggest a solution to the paradox of the dual importance of both eosinophils and DCs in allergen-specific T cell responses. The studies presented capitalize on unique adoptive cell transfer approaches possible in the mouse as part of reductionist strategies defining mechanisms underlying immune responses. Specifically, we present data demonstrating that rather than mutually exclusive activities, eosinophils and DCs act in concert, particularly in the LDLNs where T cell activation and polarization primarily occur during allergen challenge in mice (21). Specifically, our studies demonstrate that eosinophils induce the accumulation of mature myeloid DCs to LDLNs during allergen challenge. The activation of T cells in response to this DC accumulation was independent of MHC II expression on eosinophils, suggesting eosinophils are not required APCs in the secondary immune responses. Instead, we demonstrate through adoptive transfer of activated myeloid DCs that the ability of these transferred cells to polarize pulmonary immune responses was a function of the presence of eosinophils. That is, in the presence of eosinophils (i.e., wild type recipient mice), transfer of myeloid DCs leads to characteristically Th2-polarized immune responses in the lung. In contrast, in the absence of eosinophils (i.e., PHIL recipient mice), the transferred myeloid DCs promote a mixed Th2/Th17/Th1 phenotype and the accumulation of pulmonary neutrophils. Significantly, the Th17 component and the induced neutrophilia of this response in PHIL recipient mice were suppressed by cotransfer of eosinophils during allergen challenge. These results imply an important function for eosinophils in modulating the balance between Th2 and Th17 pathways generated by DCs during allergen challenge and, in turn, the development of allergic respiratory inflammation.

All studies were performed with mice on the C57BL/6J background. Eosinophil-deficient PHIL mice (22) and IL-5^−/− mice (13) were generated from established institutional colonies. MHC II knockout mice (B6.129S2-H2^dlAb1-Ea/J) were purchased from The Jackson Laboratory (Bar Harbor, ME). MHC II^−/− mice were crossed with IL-5–expressing transgenic mice (NJ.1638) (23) to generate MHC II^−/− IL-5 transgenic mice to isolate MHC II ^−/− donor eosinophils. Mice were maintained in ventilated microisolator cages housed in the specific pathogen-free animal facility at the Mayo Clinic Arizona. Protocols and studies involving animals were performed in accordance with National Institutes of Health and Mayo Foundation institutional guidelines.

Unless indicated otherwise, mice were sensitized with 100-μl injections (i.p.) of 400 μg/ml OVA grade VI (Sigma-Aldrich) and 2.25 mg Imject Alum (Thermo Scientific) on days 0 and 14 of the protocols. In studies involving no adoptive transfers or only eosinophil adoptive transfers, mice were challenged with aerosolized OVA (grade VI) at a concentration of 1% (w/v) for 25 min on days 24–26 and assessed on day 28. Time-course studies are indicated as hours after first OVA challenge. I.p. transfer of eosinophils was completed by adoptively transferring 4 × 10⁷ eosinophils on day 24 before OVA challenge. Experiments including DC adoptive transfer were completed as described in previously published studies (8, 24). In brief, intratracheal (i.t.) transfer of 2–3 × 10⁶ million bone marrow-derived, OVA-pulsed myeloid DCs (30 μl saline) into OVA-sensitized mice (or nonsensitized where indicated) were performed immediately before 1% OVA challenge on day 24. Continued OVA challenges of these mice occurred on days 25–27 (20 min each day), and end-point assessments were performed on day 29. Where indicated, some mice received both 4 × 10⁷ eosinophils by i.p. transfer and 2–3 × 10⁶ DCs by i.t. transfer on day 24, with 1% OVA challenges on days 24–27; end-point assessments were performed on day 29. Controls include saline-sensitized mice, saline-challenged mice, and mice receiving saline transfers rather than cell transfers. Each of these OVA protocols and adoptive cell transfer strategies are outlined in Supplementary Fig. 1.

Four hours before harvesting LDLNs of OVA-treated (i.e., OVA-sensitized/challenged) mice on day 28, the animals were injected (i.p.) with BrdU (3 mg/mouse; Sigma-Aldrich). BrdU incorporation into T cells was assessed by colabeling cells with Abs to TCR-β and following manufacturer’s recommendations for the FITC BrdU Flow Kit (BD Biosciences).

Bronchoalveolar lavage fluid (BAL) assessments were completed as described previously (5). Cell differentials of each sample were performed on ≥300 cells.

Histopathologic changes of the airways were assessed as described previously (5). Formalin-fixed and paraffin-embedded sections of mouse lungs were stained with H&E or periodic acid–Schiff.

Total pulmonary leukocytes were recovered from lungs perfused with PBS-EDTA before removal, placed in complete RPMI-1640 media with 175 U/ml collagenase type I (Invitrogen), 2 mM CaCl₂, and digested for 45 min at 37°C. Single-cell suspensions of lung and lymph nodes were obtained by homogenization with frosted glass slides followed by passing through a 40-μm nylon filter to remove larger aggregates of cells per tissue. RBCs were lysed by brief exposure to PharmLyse (BD Biosciences). Cell counts were completed after the last wash before staining for flow cytometry or lymph node culture assays.

Single-cell suspensions were stained for 25 min on ice with cell type-specific Abs after blockade of Fc receptors using 1 μg/μl Fc blocker (CD16/32; BD Biosciences). Abs used for staining unique cell types were as follows: Abs used to characterize T cell populations were CD4 (RM4-5; eBiosciences), TCR-β (H57-597; BD Biosciences), and CD44 (IM27; eBiosciences). Eosinophils were identified by Abs to CCR3 (83101; R&D Systems) and Siglec-F (E50-2440; BD Biosciences). Macrophages were stained with F4/80 (BM8; BD Biosciences). DC subtypes and activation status were stained with CD11c (HL3; BD Biosciences), Gr1 (RB6-8C5; eBiosciences), B220 (RA3-6B2; eBiosciences), MHC II (AF6-120; BD Biosciences), and CD11b (M1/70; eBiosciences). Flow cytometry was performed on a cytofluorimeter (Cyan; DAKO). Data acquisition and analysis were performed using Summit (version 4.3; Dako) software. Calculations of cell numbers are completed by counting total cellularity of whole-lung or LDLN single-cell suspensions and multiplying with the percentage of the population as determined by flow cytometry.

Mouse IL-17, IL-13, and IFN-γ levels were assessed using immunoassay kits (R&D Systems) according to the manufacturer’s instructions. The limits of detection for each ELISA assay were 5–10 pg/ml.

Single-cell suspensions of LDLNs were plated for 72 h on a 96-well plate at 5 × 10⁵/well with either complete RPMI-1640 media alone or OVA (grade VI; Sigma-Aldrich) that had been previously treated with 200 μg/ml Detoxi-Gel endotoxin removing gel (Thermo Scientific). Supernatants were analyzed by ELISA.

Bone marrow cells were isolated from naive 6- to 12-wk-old C57BL/6J wild-type mice, and RBCs were lysed by PharmLyse treatment followed by washing with cold MACS buffer (PBS, 0.5% [w/v] BSA, 2 mM EDTA). Lymphocytes were removed from the bone marrow by negative selection with magnetic beads conjugated with Abs to CD45R/B220 and CD90/Thy1.2 as per the manufacturer’s recommendations (Miltenyi Biotech). The negative fraction was plated (2 × 10⁵ cells/ml) in complete RPMI media supplemented with GM-CSF (30 ng/ml; R&D Systems) and IL-4 (10 ng/ml; R&D Systems) for 2–3 d. Sequentially lower doses of GM-CSF were used with 20 ng/ml on day 3 and 10 ng/ml on day 6, as described previously (25). On day 8 or 9, cells were resuspended in fresh complete RPMI-1640 media containing 10 ng/ml GM-CSF and 200 μg/ml Detoxi-Gel–treated OVA grade VI (Sigma-Aldrich). Sixteen to 18 h later, the cells were collected and stained with Abs to select myeloid DCs (F4/80⁻CD11c⁺B220⁻) for adoptive transfers. Cells were sorted by FACSAria (BD Biosciences), washed in cold PBS twice, and resuspended for i.t. transfer into mice. No significant differences in results were seen between sorted and unsorted bone marrow-derived OVA-pulsed DCs (data not shown).

Eosinophils were isolated from IL-5–expressing transgenic mice (NJ.1638) or NJ.1638/MHC II^−/− as previously described (5). In brief, peripheral blood pooled from mice was layered onto single-step Histopaque 1119 gradient (Sigma-Aldrich). The eosinophil-containing interface was then treated briefly (<10 s) with ice-cold distilled water to lyse any remaining RBCs before adding 1/10 volume of 10× PBS (i.e., cells are returned to a final concentration of 1× PBS). No differences in T cell proliferation were found when cells were lysed with the ammonium chloride-based PharmLyse solution. Eosinophils of >99% purity were isolated according to the manufacturer’s recommendations (Miltenyi Biotech) by negative selection with magnetic beads conjugated with Abs to CD45R/B220 and CD90/Thy1.2.

All data are derived from at least three independent experiments each with cohort sizes of one to six mice (error bars, SEM; *p < 0.05, **p < 0.01, ***p < 0.001, unpaired two-tailed Student t test).

We used mice that are eosinophil-sufficient (wild type), eosinophil-low (IL-5^−/− mice) (13), and eosinophil-null (PHIL mice) (5, 22) to test the hypothesis that eosinophil levels and their specific tissue localization effect T cell accumulation/activation during allergen provocation. Wild type, IL-5^−/−, and PHIL mice were sensitized with i.p. injections of OVA/Alum (days 0 and 14) and challenged via the airways with a 1% OVA aerosol (OVA-treated) on days 24, 25, and 26 (control animals received saline alone; Supplemental Fig. 1A). The partial (IL-5^−/−) and complete (PHIL) absence of eosinophils in these mice were each accompanied by a loss of the OVA-induced CD4⁺ T cell accumulation in the lung as compared with OVA-treated wild type animals (Fig. 1A). However, injection of these OVA-treated mice with BrdU (4 h before end-point assessments on day 28) to measure T cell proliferation in LDLNs showed that only the complete loss of eosinophils (i.e., PHIL mice), and not the partial (85%) loss (i.e., IL-5^−/− mice; data not shown), blocked T cell proliferation in the lymphatic compartment (Fig. 1B). Moreover, the failure to induce proliferation in the LDLNs of PHIL mice was not unique to the OVA allergen because sensitization and challenge with a more environmentally relevant allergen, ragweed, produced similar results (Supplemental Fig. 2). To bypass the complete absence of peripheral eosinophils in PHIL mice and promote their recruitment to LDLNs after OVA provocation, we adoptively transferred eosinophils by i.p. injection; this strategy has been shown to promote leukocyte recruitment promptly to LDLN within hours of injection (3, 11). PHIL mice received 4 × 10⁷ eosinophils by i.p. transfer 1 h before the first OVA challenge (Supplemental Fig. 1B). This was sufficient to return eosinophil percentages in the LDLNs to levels similar to those observed in OVA-treated wild type mice (Fig. 1C). It is noteworthy that as previously reported (3), transfer of eosinophils via the peritoneal cavity restored eosinophil accumulation only to the lymphatic compartments of PHIL mice and not in the lung itself, where eosinophil levels remained >98% lower relative to wild type animals. In addition, unlike the trafficking of mononuclear cells such as dendritic cells (26) or T cells (27), eosinophil movement through lymphatic circulation occurred independently of CCR7 expression (Supplemental Fig. 3). Concomitant with eosinophil accumulation to the LDLNs of OVA-treated PHIL mice, CD4⁺ T cell numbers in the LDLNs increased to levels equivalent to that observed in OVA-treated wild type animals (Fig. 1D). Similar to OVA-treated IL-5^−/− mice, the accumulation was limited to the LDLNs; that is, i.p. transfer of eosinophils into OVA-treated PHIL mice did not induce the accumulation of CD4⁺ T cells into the lungs (Fig. 1E). These data demonstrate that the localization of eosinophils to the LDLNs was necessary for, and limited to, T cell accumulation in LDLNs in response to airway allergen challenge of OVA-treated PHIL mice.

Eosinophils have been demonstrated to express MHC II molecules and potentially mediate Ag presentation in vivo as part of immune responses leading to the T cell activation (14, 15, 18). We adoptively transferred (i.p.) MHC II-deficient eosinophils into OVA-treated PHIL mice to determine whether the T cell activation/proliferation occurring in the LDLNs of recipient animals was a consequence of MHC II-dependent Ag presentation by eosinophils. These studies showed that migration of MHC II^+/+ or MHC II^−/− eosinophils to LDLNs was equivalent (Fig. 2A). More importantly, transfer of either MHC II^+/+ or MHC II^−/− eosinophils into OVA-treated PHIL recipients promoted comparable levels of CD4⁺ effector T cell accumulation in LDLNs (Fig. 2B), showing that eosinophils do not require expression of MHC II to promote the activation of T cells in the LDLNs after allergen challenge.

Commonly accepted paradigms on DC functions suggest an independence between eosinophils and LDLN T cell activation. However, our results demonstrating that eosinophils are required for T cell proliferation, as well as studies describing eosinophil-mediated effects on DCs (28, 29), suggested a possible eosinophil dependence of DC activities in secondary immune responses after allergen challenge. To test this hypothesis, LDLNs in OVA-treated wild type versus OVA-treated PHIL mice were examined with and without cotransfer of eosinophils to determine the relative effects of eosinophils on activated MHC II^hi DC accumulation in the LDLNs. Significantly, OVA-treated PHIL mice have reduced numbers of DCs in the LDLNs as compared with OVA-treated wild type mice (Fig. 3A). This reduction occurred without changes in basophil levels in the LDLN, thus suggesting that eosinophils alone and not basophils are correlated with DC levels (Supplemental Fig. 4). Furthermore, the induced DC accumulation in the LDLN of OVA-treated PHIL mice was restored by i.p. adoptive transfer of MHC II^+/+ or MHC II^−/− eosinophils (Fig. 3A). Significantly, the presence, although reduced, of eosinophils in the LDLNs of OVA-treated IL-5^−/− mice was also sufficient to induce accumulation of activated MHC II^hi DCs.

Myeloid dendritic cells, in particular, have been suggested to be key agonists of pathways linked with pulmonary allergic immune responses (7). Assessments of the DC subtypes present in the lungs and LDLNs of OVA-treated wild type and PHIL mice 5 d after the first OVA challenge revealed that OVA-treated PHIL mice have equivalent (i.e., relative to wild type) percentages of lymphocytic DCs (F4/80⁻/CD11b⁺/Gr1⁺/CD11c⁺), plasmacytoid DCs (F4/80⁻/CD11b⁻/Gr1⁺/CD11c⁺), and monocytic inflammatory DCs (F4/80⁻/CD11b⁺/Gr1⁺/Cd11c⁻) in both the lungs and LDLNs (Fig. 3B). However, myeloid DC (F4/80⁻/CD11b⁺/Gr1⁻/CD11c⁺) levels were lower in both the lung and LDLNs of OVA-treated PHIL mice relative to control OVA-treated wild type animals (Fig. 3B). Moreover, the total number of myeloid DCs were significantly reduced in the lungs of OVA-treated PHIL mice as compared with wild type mice (0.47 × 10⁶ ± 0.7 × 10⁵ versus 1.02 × 10⁶ ± 1.5 × 10⁵, respectively; p = 0.007). In addition, the number of myeloid DCs in the LDLNs (2.56 × 10⁵ ± 0.3 × 10⁵ versus 5.90 × 10⁴ ± 0.8 × 10⁵, OVA-treated PHIL versus wild type, respectively; p = 0.0007), mirrored that of LDLN MHC II^hi DCs. Thus, myeloid DC accumulation in both the lungs and LDLNs of mice after allergen provocation appears to be a function of one or more eosinophil activities.

Several studies have demonstrated that myeloid DCs migrate from the lung to the LDLNs within ∼10 h after aerosolized allergen challenge, peaking at ∼40 h (30). Thus, we assessed DC levels at various time points after allergen challenge to determine whether OVA-treated PHIL mice had recruited DCs at a rate similar to OVA-wild type mice and simply failed to accumulate these cells in draining pulmonary lymphoid compartments because of a change(s) in their survival (i.e., LDLN half-life). The LDLNs of OVA-treated wild type mice accumulated eosinophils and MHC II^hi activated DCs within 20 h of OVA challenge (Fig. 3C); myeloid DC (F4/80⁻/CD11b⁺/Gr1⁻/CD11c⁺) populations mirrored MHC II^hi DCs (MHC II^hi/CD11c⁺). In contrast, OVA-treated PHIL mice failed to accumulate DCs within 20 h of allergen challenge, suggesting a deficiency in recruitment of these lung DCs to the LDLN. These data demonstrated that the presence of eosinophils in the LDLN is necessary to promote myeloid DC recruitment to LDLNs and, in turn, T cell activation during OVA provocation.

We performed adoptive transfer experiments of activated (i.e., OVA-pulsed) myeloid DCs similar in design of other studies (8, 10, 31) to circumvent the limitation of eosinophil-dependent DC activation/recruitment. This allowed for the assessment of activated myeloid DCs to induce immune responses in the absence of eosinophils. Specifically, bone marrow-derived, OVA-pulsed myeloid DCs were adoptively transferred i.t. on the first day of allergen challenge into either OVA-sensitized PHIL or nonsensitized PHIL mice (Supplemental Fig. 1C); OVA-sensitized wild type animals served as controls. These adoptively transferred OVA-pulsed DCs accumulated in the LDLNs in all mice receiving DCs, including nonsensitized PHIL mice, demonstrating that recruitment of activated DCs to LDLNs is eosinophil independent in this system (Fig. 4A). Despite the presence of DCs in the LDLNs of nonsensitized PHIL mice after adoptive transfer, production of effector T cells is deficient (Fig. 4B), demonstrating that OVA-specific memory T cells are required for their activation into effector T cells. Furthermore, the increase in effector T cells that occurred in OVA-treated PHIL mice receiving activated DCs was equivalent to the levels observed in OVA-treated wild type mice receiving DCs.

OVA-treated PHIL mice receiving OVA-pulsed DCs (i.t.) displayed histopathologic changes, including goblet cell metaplasia/airway epithelial cell mucin accumulation that was indistinguishable from OVA-treated wild type control animals (Fig. 5A). OVA-treated PHIL mice transferred with DCs (i.t.) and eosinophils (i.p.) were comparable in their histopathology to OVA-treated PHIL mice transferred only with DCs. Marked elevations in BAL total cellularity also occurred in OVA-treated PHIL mice receiving OVA-pulsed DCs relative to nonsensitized PHIL animals (Fig. 5B). However, a significant change occurred in the composition of the BAL cellularity. In OVA-treated PHIL DC-recipient mice, the OVA-induced BAL cellularity was predominantly a neutrophilic and lymphocytic infiltrate. The significance of this airway neutrophilia is that it was Ag dependent (i.e., memory T cell dependent) because nonsensitized PHIL DC-recipient mice (i.e., mice with no OVA-specific memory T cells) failed to develop airway inflammation (Fig. 5A). In comparison, OVA-treated wild type mice receiving OVA-pulsed DCs developed both airway inflammation and a significant eosinophilic infiltrate in the airways with no increase in neutrophil levels (Fig. 5B). Surprisingly, despite the induced neutrophilia in the OVA-treated DC-recipient PHIL mice, the BAL cytokine milieu was Th2 polarized with elevated expression levels of cytokines such as IL-13 and undetectable BAL levels of either IFN-γ or IL-17 (Fig. 5C). Thus, in the absence of eosinophils, OVA-mediated immune responses were mixed with a decidedly Th2 polarized cytokine response in the lung that nonetheless was accompanied by an induced airway neutrophilia. Significantly, OVA-treated PHIL mice receiving cotransfer of eosinophils with the OVA-pulsed DCs resulted in a significant reduction in the percentage of neutrophils in the BAL as compared with PHIL mice transferred with OVA-pulsed DCs alone (Fig. 5B), suggesting that eosinophils are key mediators of modulating the pulmonary cellular infiltrate after OVA challenge.

The increased presence of neutrophils and lymphocytes as part of the BAL cellular infiltrate of OVA-treated PHIL mice after the transfer of OVA-pulsed DCs suggested that in the absence of eosinophils, a change in the character of the OVA-induced memory T cell responses occurred. The role of eosinophils in modulating the unique phenotype of LDLN effector T cell populations was determined by coculturing OVA with LDLN cells from each of these groups of mice and assessing their expression of the cytokines IL-17 (Th17), IFN-γ (Th1), and IL-13 (Th2). LDLN cells from OVA-treated wild type mice after adoptive transfer (i.t.) of OVA-pulsed DCs displayed a distinct Th2 polarized phenotype, expressing only IL-13 (Fig. 6). In contrast, OVA-treated PHIL animals receiving OVA-pulsed DCs had a mixed effector T cell phenotype with increased levels of IFN-γ (Th1), IL-13 (Th2), and IL-17 (Th17). Strikingly, adoptive transfer of eosinophils into OVA-treated PHIL recipients that also received OVA-pulsed DCs selectively resulted in the suppression of the induced Th17 responses (coinciding with a reduction of the airways neutrophilia; Fig. 5B) in these mice and, to a lower, yet not significant degree, the Th1 responses; this leads to the emergence of a Th2 polarized immune response (Fig. 6). These data demonstrate a unique role for eosinophils in suppressing Th17 and possibly Th1 events after transfer of OVA-pulsed DCs that changes the character of the induced pulmonary immune response(s).

Allergen exposure to the airways of sensitized mice results in a cascade of events that have been characterized as Th2-dominated immune responses leading to histopathologies and lung dysfunction pathognomonic of asthma. Although eosinophil accumulation in the lung has been considered a hallmark part of these Th2-dominated pulmonary immune responses [occurring in both patients (32) and animal models (33)], the definition of specific eosinophil-mediated mechanisms within the immune cascade associated with allergen provocation still has remained unclear. The seemingly incongruent observation that both dendritic cells (8, 10, 11) and eosinophils (5, 6) are necessary for Th2 pulmonary immune responses suggest that previously unrecognized (and/or poorly understood) nodes exist whereby these two cells interact as part the development of pulmonary immune responses. The data presented in this study now provide evidence for these nodes.

Our studies demonstrated that mice deficient in eosinophils were unable to promote the Th2 polarization of the lung microenvironment after allergen provocation because of a T cell activation/proliferation deficiency in LDLNs. This observation was linked, in part, to a failure of DCs to accumulate in the LDLNs of OVA-treated PHIL mice, suggesting a previously underappreciated mechanism by which the steady-state presence of circulating/lymphatic eosinophils are necessary for pulmonary DC emigration to the LDLNs after allergen challenge. The mechanism of eosinophil-induced recruitment/accumulation of DCs is unknown but may rely on release of eosinophil mediators to influence maturation/activation (29) and/or chemoattraction (29, 34). Significantly, adoptive transfer of eosinophils into PHIL mice only modulated the accumulation of the DC subtype known to induce pulmonary inflammation (i.e., MHC II^hi myeloid DCs) and did not elicit the accumulation of Ag tolerance-inducing plasmacytoid DCs (35). In addition, adoptive cell transfer studies of MHC II-deficient eosinophils into PHIL recipient mice demonstrated that MHC II expression on eosinophils was not required for LDLN T cell proliferation in this allergen challenge model system. Thus, although eosinophils have the potential to act as APCs (14–16), it appears that the multitude of overlapping pathways offered by the various professional APCs available in the pulmonary compartment (in particular, dendritic cells) are both necessary and sufficient providers of the APC activities needed for immune responses to allergen provocation.

The adoptive transfers of activated OVA-pulsed DCs into wild type versus PHIL mice suggest a larger than previously suspected role for eosinophils in the Th2-associated pulmonary immune responses that occur after allergen challenge. In particular, these data suggested a reciprocal correlation exists between eosinophils and T cell immune responses such that in the absence of eosinophils, allergen-dependent T cell immunity is skewed and becomes a mixed Th2/Th17/Th1 collection of responses that now induces a pulmonary neutrophilia. This role for eosinophils in DC-mediated T cell polarization provides a parsimonious explanation for observations from several earlier studies investigating the activation and polarization of T cells in allergen-mediated pulmonary immune responses. For example, two recent studies (31, 36) demonstrated that in the absence of DCs (or the targeted knockout of MHC II expression on DCs), LDLN T cell proliferation occurs in mice, but the Th2 polarization of these T cells after allergen challenge is dependent on another (i.e., non-DC) MHC II-expressing cell(s). As a consequence, mice deficient in non-DC MHC II-expressing cells experience development of a neutrophilic and IFN-γ–dependent inflammation, although the status of Th17 was unknown in these models. An additional example of the proposed importance of one or more non-DC MHC II-expressing cells on the skewing of Ag-induced immune responses is provided in studies depleting basophils with MAR-1 (FcεRIα) Ab in a lung inflammation model (37). Significantly, this Ab also reduced eosinophil levels in these mice (31). In doing so, this leaves open the possibility that eosinophils and not basophils were the contributors skewing the immune responses in that study.

The mechanisms by which LDLN eosinophils suppress the production of Th17/Th1 cells are unknown; however, they appear to modulate DC-induced memory T cell polarization as they proliferate during their transition into effector T cells. That is, the presence of eosinophils in the immune environment where DC-mediated T cell activation/polarization is occurring provides (or in the case of PHIL mice, fails to provide) additional signals to suppress the differentiation of Th17, and to some extent Th1, cells (38). Potential mechanisms by which eosinophils may modulate DC–T cell interactions abound. For example, Yang and colleagues (29) demonstrated that human eosinophil granule proteins can stimulate DCs through a TLR2-dependent pathway, which has been shown to be an important pathway to suppress Th17 responses (39) and induce Th2 pulmonary inflammation (40). Moreover, other studies have demonstrated that eosinophils are also a potentially significant source of the DC-modulating cytokine IL-25 that augments Th2 responses (41). Still other studies have shown that eosinophils express additional cytokines/enzymes with the potential ability(ies) to alter the relative balance of immune responses such as Th2 (IL-4) versus Th1 (IL-12) versus immunosuppressive responses characterized by expression of TGF-β and indoleamine 2,3-dioxygenase (19, 20). Finally, eosinophils may selectively activate or suppress T cells derived from a mixed pool of Th17/Th1/Th2 memory T cells that are plastic in phenotype (42).

The potential significance of eosinophil–DC–T cell interactions as part of cellular/molecular mechanism occurring in subsets of asthma patients offers unique and/or alternative explanations for confounding observations often associated with the care of these patients. Three examples are of particular interest. First, viral infections or postviral complications represent significant events linked with asthma exacerbation leading to hospital visits (43). The assumption has been that viral-mediated lung pathology simply synergized with the difficulties already experienced by asthmatics leading to exacerbation events in this population (44). In addition, studies have demonstrated increased susceptibility to viral infections and extended/greater viral load, suggesting deficiencies in production of Th1 cytokine responses because of augmented Th2 cytokine production (45, 46). The hypothesis articulated in this article suggests that in asthmatic patients experiencing a viral infection, some DCs arriving in LDLNs encounter activated eosinophils that may differentially suppress Th17/Th1 events. This, in turn, would reduce targeted antiviral immune responses, increasing the likelihood of infection and/or its severity (47). Moreover, the suppression of Th17/Th1 immune responses by activated eosinophils may also lead to the disproportional production of viral Ag-specific Th2 effector cells that upon recruitment to the lung enhance allergic respiratory inflammation in these patients (48). Second, the paradigm of eosinophil–DC–T cell interactions suggests that inhaled corticosteroids (ICS) may attenuate local immune responses in the lung via a novel eosinophil-dependent pathway. Specifically, in addition to the ability of ICS to eliminate eosinophil activities through cytocidal effects (49), ICS treatment of patients would disrupt eosinophil-mediated positive regulatory loops promoting the production of Th2 effector cells in LDLNs and their subsequent recruitment to the lung. It is noteworthy that this same eosinophil-mediated disruption of immune-regulatory loops may also provide an explanation for the observation that the most significant effect of anti–IL-5–based treatment of asthma patients (i.e., mepolizumab) (50, 51) is not acute improvement of lung function with the resolution of symptoms, but the reduction of exacerbation events over time. Third, the proposed Th17-suppressive character of eosinophils may also provide a mechanism to explain the symptoms/pathologies experienced by patients with neutrophil-dominated severe asthma. In particular, corticosteroid administration that reduces pulmonary and LDLN-associated eosinophils may also have the additional consequence of inducing/exacerbating severe and neutrophilic asthma (52) through increased Th17 production in these patients (53).

We suggest that the respective activities and importance of pulmonary and LDLN eosinophils may be greater than previously suspected. In particular, the roles of eosinophils as unique mediators of local immune responses provide new insights as to the larger contributory role of this leukocyte in the complex inflammatory cascades leading to disease pathology. We would also suggest that these roles are likely to extend beyond asthma and allergic diseases, and may provide explanations for the involvement of eosinophils in other inflammatory diseases that have been linked with these cells, including helminth infection, gastrointestinal diseases, cancer, and organ transplant rejection (54).

We acknowledge the invaluable contribution of numerous individuals, including Dr. Sergei Ochkur, Ralph Pero, Alfred Doyle, and Dr. Michael McGarry. We also acknowledge the tireless efforts of the Mayo Clinic Arizona medical graphic artist Marv Ruona and Joseph Esposito of Research Library Services. In addition, we express our gratitude to the Lee Laboratories administrative staff, Linda Mardel and Charlie Kern, without whom we could not function as an integrated group.

The authors have no financial conflicts of interest.