Dendritic cell expression of the transcription factor T-bet regulates mast cell progenitor homing to mucosal tissue

We have recently shown that CD4⁺ T cell trafficking in T-bet^−/− mice is dramatically reduced (20). Our interest in MC trafficking then led us to investigate whether MCp homing to the mucosal tissues is impaired in T-bet^−/− mice. We measured the total number of MCp in the intestine and the lung, as well as in the BM and the spleen. T-bet deficiency resulted in a nearly total loss of the MCp reservoir normally present in the small intestine and in a significant decrease in the lung, whereas MCp numbers in the spleen and the BM were comparable to those in WT mice. This finding was consistently observed in both BALB/c (Fig. 1 A) and C57BL/6 (Fig. 1 B) mice strains lacking T-bet. Consistent with previous findings (13), the number of MCp found in the tissues examined in the BALB/c strain was greater than that in C57/BL6 mice. Histochemical analysis of tissue sections prepared from intestine and spleen of adult, age-matched T-bet^−/− and WT mice also showed that fewer MCs were detected in the intestine of T bet^−/− compared with WT mice. In the spleen, which showed no differences in MCp, MCs were also diminished in the T-bet^−/− mice (Fig. 2). These data indicate that under basal conditions in the absence of T-bet there is a tissue-specific homing defect in MCp homing to mucosal tissues.

Having observed a lack of homing of MCp to the intestine, which is the largest reservoir of committed MCp in the mouse (22), we postulated that T-bet^−/− MCp might exhibit diminished α4 integrin–dependent adhesive interactions with endothelial cell adhesion molecules VCAM-1 and MAdCAM-1, because these endothelial ligands have previously been shown to be essential for MCp homing to the intestine (13). To assess this further, we measured adhesion of WT and T-bet^−/− BMMCs to immobilized, recombinant mouse VCAM-1 or MAdCAM-1 under conditions of physiological laminar shear flow. Because the number of MCp in the BM or in circulating blood of adult mice is low (1, 23), it is impossible to isolate quantities sufficient to perform these adhesion assays. Therefore, we and others (24, 25) have used BMMCs generated in vitro as a surrogate model to study the different properties and adhesion pathways of MCs. T-bet^−/− BMMCs displayed markedly reduced binding to both VCAM-1 and MAdCAM-1 across a range of shear flow (Fig. 3). Incubation of WT and T -bet^−/− BMMCs with blocking antibodies against the α4 integrin chain abolished BMMC binding to VCAM-1 (unpublished data). These data suggest that T-bet affects the adhesive interactions between MCp and intestinal endothelial VCAM-1 and MAdCAM-1. Because MCp express receptors for keratinocyte-derived chemokine (KC/CXC chemokine ligand [CXCL] 1) and chemokines coimmobilized with adhesion molecules are known to trigger leukocyte stable arrest under shear flow conditions (26, 27), we examined whether this or other chemokines (stromal cell–derived factor 1α [SDF-1α]) coimmobilized with VCAM-1 or MAdCAM-1 would overcome the adhesion defect. Neither WT nor T-bet^−/− MCp showed an increase in adhesion (Fig. S1, A–C) nor resistance to detachment after adhesion to MAdCAM-1 (Fig. S1 D), suggesting that MCs and T cells do not exhibit the same recognition response in this assay. We next tested whether surface expression of α4 integrins (CD49d), which bind to VCAM-1 and MAdCAM-1, were altered. FACS analysis revealed very low levels of surface expression of α4 and β7 integrins and the chemokine receptor CXCR2 in BMMCs generated from either WT or T-bet^−/− mice (Fig. S2), in agreement with previous reports (24, 28, 29). Thus, the defect is not explained by a loss of α4β7 integrin expression or a lack of chemokine-induced arrest under flow conditions.

Because this defect in adhesion suggests a direct role for T-bet in controlling MCp adhesion, the next set of studies determined T-bet expression in cultured BMMCs or other cells along the developmental pathway that we have previously described (1). Western blot analysis revealed that T-bet is not expressed in either T-bet^−/− or WT BMMCs under basal conditions or after stimulation with nonphysiologic (PMA + ionomycin) or physiologic (LPS and IgE cross-linking) stimuli (Fig. 4 A). In contrast, T-bet was easily detected in T cytotoxic type 1 cells (Tc1), analyzed in parallel as a positive control for T-bet expression.

We next considered the possibility that T-bet was not expressed in BMMCs because of a loss of expression under the culture conditions required for successful in vitro differentiation. Because previous studies have reported that the Lin⁻/c-kit⁺/Sca1⁻ CD34⁺ cell population from adult BM gives rise to MCs (1, 30), we used a high-speed FACS approach on the BM of WT adult mice to isolate these cells and probe the isolated cells for T-bet expression by quantitative real-time PCR, using Th1 cells as a positive control (Fig. 4 B). T-bet was present in the WT BM, but at very low levels. The cell populations sorted from the BM as Lin⁻ or Lin⁺ cells, which correspond to uncommitted or committed BM progenitors (of MCs and other lineages), respectively, also expressed very low levels of T-bet (Fig. 4 B, inset plotted with expanded y axis). The expression of T-bet by whole BM cells is most likely caused by Lin⁺ cells, because T-bet was below detectable levels in the population (Lin⁻) that leads to MCp production. We also tested for the expression of T-bet in other hematopoietic progenitors present in the BM and committed to the myeloid pathway, such as the common myeloid progenitor, the granulocyte monocyte progenitor (GMP), the recently described basophil MC progenitor (BMCP) (1), and the basophil progenitor (BaP), as well as in MCp derived in vitro from GMP and mature peritoneal MCs. None of these cells, progenitors or mature MCs, expressed T-bet. In contrast, T bet was easily detected in positive-control Th1 cells (Fig. 4 C). We conclude that T-bet is not expressed in committed MCp, its immediate precursors, or BMMCs derived in vitro, and hence postulate that T-bet plays an indirect role in the homing of MCp to the mucosal tissues.

In view of the inability of our assays to detect T-bet expression in MCp or cultured BMMCs, we were forced by the observed striking defect of MCp pools in certain mucosal tissues of T-bet–null mice to pursue other approaches to understanding how T-bet exerts this effect. To define whether the observed homing defect was due exclusively to a defect in the T-bet^−/− BM (including MCp or other BM precursors expressing T-bet) or, alternatively, to a defect of the local intestinal microenvironment by itself, we depleted the hematopoietic precursors in WT mice by sublethal irradiation and then reconstituted the animals by adoptive transfer of either WT or T-bet^−/− BM cells (SIBR). The number of MCp found in the intestine of WT-SIBR mice reconstituted with T-bet^−/− BM was very low compared with that in WT-SIBR mice reconstituted with WT BM. In contrast, the numbers of MCp in the spleen and BM were not significantly different in WT mice reconstituted with T bet^−/− BM from those reconstituted with WT BM. (Fig. 5 A). On the other hand, sublethally irradiated T-bet^−/− mice reconstituted with BM from WT mice showed total reconstitution of intestinal MCp, as well as of BM and spleen MCp (Fig. 5 B). These results demonstrate that the MCp that arise from the T bet^−/− BM show a defect in homing to the intestine despite normal numbers of precursors in the spleen or BM, whereas the MCp arising from the WT BM and transferred into a T-bet^−/− host show adequate tissue homing of MCp, including homing to the intestine. These data implicate a defect or loss of T-bet–expressing cells in the BM compartment that is responsible for the lack of MCp adhesion and intestinal-specific homing in the T-bet^−/− mice and rule out the possibility of a defect in the intestinal compartment of T-bet^−/− mice.

Because the findings in Fig. 5 demonstrated that a defect in the BM compartment of T-bet^−/− mice is responsible for the lack of MCp homing to the intestine, we performed experiments to identify the BM cell type responsible. Based on the literature (18), DCs are good candidates because their development, differentiation, and activation is independent of T-bet, but DCs can up-regulate T-bet and secrete IFN-γ (18). We derived DCs from WT BM by culture in GM-CSF for 7 d and transferred these BMDCs into WT or T-bet^−/− mice. Strikingly, T–bet^−/− mice given WT BMDCs showed levels of MCp in the intestine comparable to WT mice given BMDCs in parallel when evaluated 14 d later. Transfer of BMDCs into WT mice did not significantly increase the number of MCp (Fig. 6 A). In contrast, T-bet^−/− animals that did not receive BMDCs exhibited a dramatic reduction of MCp in the intestine as compared with WT mice.

The role of T-bet⁺ BMDCs in controlling MCp intestinal homing was further investigated by testing whether co-culture of WT BMDCs with T-bet^−/− BMMCs restored T-bet BMMC binding to VCAM-1 and MAdCAM-1. Indeed, a 6-h co-culture of T-bet^−/− BMMCs with WT BMDCs completely rescued binding to VCAM-1 and MAdCAM-1, as these cells exhibited no binding under conditions of shear flow that was comparable to WT BMMCs (Fig. 6, B and C). BMMC adhesion was abolished by anti–α4 integrin antibody. These data strongly suggest that T-bet in the DCs plays a role in homing of MCp to the intestine by regulating the α4β7-dependent binding to the endothelial cell ligands VCAM-1 and MAdCAM-1, which are necessary for the appropriate homing of MCp to the intestine (13).

Infection with T. spiralis stimulates a large increase in the number of MCs residing in the small intestine (8). The rapid expulsion of T. spiralis is associated with and dependent on the pronounced mastocytosis, which is turn is mediated by a Th2-type response (31–34). Lack of the intestinal MCs and MCp in β7 integrin–deficient mice resulted in a delayed MC hyperplasia and delayed rejection of the T. spiralis from infected mice (35). Thus, we assessed the intestinal MC response of WT and T-bet^−/− mice to infection with T. spiralis. Despite the low number of resident MCs and reduced MCp reservoir in the intestine of T-bet^−/− mice, a robust mucosal MC response was noted in the T-bet–deficient mice. The response was even greater than that observed in WT mice infected in parallel and resulted in an enhanced elimination of the worms from the intestine (Fig. 7) regardless of the strain background. The enhanced clearance of infection was observed with T-bet^−/− mice in a Th2 cell–type background strain (Fig. 7 A, BALB/c mice) and with T-bet^−/− mice in a Th1 cell–type background strain (Fig. 7 B, C57BL/6 mice). The local MC hyperplasia in response to the infection was also markedly increased in the T-bet^−/− mice, even more so than in WT mice (Fig. 7, C and D). These results are consistent with the view that the T-bet–null mouse is polarized toward the induction of a Th2 cell inflammatory milieu (19), and T. spiralis, which provides a potent stimulus, provokes a more vigorous MC proliferation response to this pathogen.

We report on the unexpected and novel finding that the transcription factor T-bet expressed in DCs indirectly regulates MCp homing to mucosal tissues. The involvement of T-bet in Th1 T cell polarization, IFN-γ production by DCs, and terminal maturation of NK cells has been documented (16–19), whereas the role for T-bet in leukocyte trafficking is an emerging area. Recent reports showed that T-bet was necessary for trafficking of CD4⁺ T cells (20), CD8⁺ T cells (36), and NKT cells (21) to sites of immune reactions. Our interest in understanding the factors that control MC trafficking (1, 3) led us to study the effect of T-bet deficiency on MCp homing to tissues.

The initial observation of a dramatically lower number of MCp in mucosal tissues in T-bet^−/− animals (Fig. 1), specifically in the small intestine as measured by limiting dilution bioassay, was very surprising. That this reduction in MCp was physiologically relevant was corroborated by the observed 50–70% reduction in mature tissue MCs, as detected by histological analysis of the small intestine and spleen (Fig. 2). As some mature MCs in the spleen have been shown to be derived from the intestine, at least after infection with T. spiralis (37), the decrease in splenic MC numbers could reflect the decreased numbers of intestinal MCs in the T-bet^−/− mice or a maturational defect within the spleen.

From earlier studies, we knew that T-bet was necessary for both the biosynthesis of functional selectin ligands and the expression of chemokine receptors in CD4⁺ T cells (20), which guided us to consider that MCp adhesion by itself was defective. Because the hematopoietic pool of committed MCp in mice is estimated to be <30,000 cells (Fig. 1) (1, 23), we used BMMCs cultured in vitro as a surrogate for MCp cells to study their adhesive properties under physiological shear flow conditions. These experiments revealed that T-bet–deficient BMMCs adhered poorly to both VCAM-1 and MAdCAM-1 (Fig. 3). Furthermore, coimmobilized chemokines such as SDF-1α/CXCL12 or KC/CXCL1 were not sufficient to reconstitute or augment adhesion. Despite the decreased adhesion, no loss of α4β7 integrin or CXCR2 expression was detected (Fig. S2), perhaps implicating a defect in α4 integrin–dependent or chemokine receptor signaling.

At this juncture, the picture emerging was that T-bet expression in MCp selectively controlled homing to mucosal tissues (but not development in BM or spleen) via the α4β7 integrin–VCAM-1/MAdCAM-1/CXCR2 adhesion pathway. An unexpected finding was that T-bet transcripts were below detectable levels in mature MCs isolated from the peritoneal cavity, in BMMCs or in committed intestinal MCp obtained by immunoisolation and high-speed FACS sorting (Fig. 4). We did detect T-bet expression in Lin⁺ BM cells, which are precursors of many other cell types, but not in the uncommitted Lin⁻ population or in other early myeloid progenitors, such us common myeloid progenitor, GMP, and BMCP, or the committed progeny of the BMCP, the MCp or BaP. From these collective findings, we concluded that T-bet cannot directly regulate MCp homing via α4β7 integrin/CXCR2–dependent adhesion. To further understand the role of T-bet, we addressed whether the defect resided in other cellular components of the BM and/or in the intestinal milieu by reconstituting sublethally irradiated WT or T-bet^−/−–null animals with BM from WT or T-bet^−/− mice. These studies indicated that the defect lies in the BM compartment and not in the intestinal microenvironment (Fig. 5), because T-bet–null mice reconstituted with normal BM cells had similar numbers of intestinal MCp to WT mice reconstituted with WT BM cells. In contrast, WT mice reconstituted with T-bet^−/− BM cells showed decreased MCp homing to the intestine. Because the genotype of the BM was critical and not that of the intestine, these results indicate that T-bet expression in BM cells, other than the MCp or its immediate precursors, is critical to proper homing of MCp to tissues. Although surprising, this is similar to the recent observation that TNF-α production by CD11b⁺ BM cells was critical to development and expansion of BMMCs (38) and supports the concept that MC development along the myeloid pathway requires interaction with other leukocytes. Given the restricted expression of T-bet within leukocytes of the BM, the most likely T-bet⁺ cell found in this environment is the DC.

We have tested this hypothesis by adoptive transfer of WT DCs into T-bet^−/− mice and found that these animals now exhibited normal homing of MCp to intestine. Furthermore, T-bet^−/− BMMCs co-cultured with WT DCs recover the ability to bind to MAdCAM-1 and VCAM-1 at levels that are comparable to WT BMMCs. We conclude that the presence of DCs expressing T-bet in the BM compartment is required for normal α4β7 integrin–dependent binding to MAdCAM-1 and VCAM-1 and MCp intestinal homing.

We further probed whether such a defect in the MCp pool increased the susceptibility of these animals to a serious outcome when challenged with a helminth infection. Based on studies in β7 integrin–deficient mice, which lack MCp and MCs in the intestine and show delayed rejection of an intestinal helminth, we anticipated a similar finding in T-bet–null mice, even though these mice are polarized toward Th2 responses (35, 13). Unexpectedly, the rejection of the worms by T-bet mice was equal to or even better than that by WT mice (Fig. 7), consistent with previous reports of an association of a rapid expulsion of T. spiralis with a pronounced mastocytosis mediated by a Th2-type response and the findings that T-bet–null mice are Th2 cell polarized (19, 35). Our finding suggests that like in the β7 integrin–null mice, the adhesion defect in α4β7 integrin/CXCR2–mediated signaling for homing is compensated by other pathways elicited by the potent Th2 cell–mediated immune response that overcomes the initial dependence on the former molecules.

We have yet to identify the specific DC T-bet–dependent mechanisms in BM that are responsible for the defect in the α4β7 integrin–CXCR2 adhesion pathway. We do not suspect that reduced α4 integrin/CXCR2 expression is responsible, because BMMCs derived from WT and T-bet–deficient mice showed equivalent, albeit low, levels of surface-expressed α4 and β7 integrins and CXCR2. The loss of the α4 integrin during protracted in vitro culture that we observed (unpublished data) has also been reported previously (28, 29). Another possibility is that T-bet^−/− cells have a defect in α4 integrin or CXCR2 signaling that leads to the observed loss of adhesion. We note that one should interpret these results cautiously, because BMMCs are not MCp; they are only a tool, the best in vitro surrogate cells currently available for studying MCp.

T-bet controls a multitude of genes in several cell types, making it difficult to discern how T-bet is indirectly controlling MCp homing to mucosal tissues. Our experiments provide evidence that the transcription factor T-bet expressed in DCs within the BM indirectly imprints a program in developing MCp that ensures appropriate homing to the mucosal tissues. Regardless of the fact that the IFN-γ gene is one of the main direct targets of T-bet (16, 18), we were not able to increase the MCp intestinal homing in T-bet^−/− mice by intraperitoneal injection of IFN-γ, and more strikingly, the MCp present in the WT intestine were depleted after IFN-γ treatment, probably due to apoptosis of MCs induced by IFN-γ, as described by other authors (unpublished data) (39). Thus, it seems most likely that IFN-γ is not directly responsible for MCp homing and that some other factor causes the homing defect in T-bet–deficient mice. Nonetheless, these findings are remarkable, as it is the first indication that another innate effector cell directly influences MC development in a way that affects its trafficking in the absence of overt inflammation.

In summary, these observations provide further evidence that the transcription factor T-bet is involved in leukocyte trafficking and show, for the first time, a role for T-bet expression in DCs for the constitutive homing of MCp cells to mucosal tissues.

Dulbecco's PBS and RPMI 1640 were purchased from Invitrogen. The following reagents were obtained from the indicated sources: mouse MAdCAM-1 chimeric protein (T. Yednock, Elan Corporation, San Francisco, CA), recombinant mouse VCAM-1-Fc chimera (R&D Systems), and recombinant mouse IL-3, stem cell factor (SCF), SDF-1α, GM-CSF, and KC/CXCL1 (PeproTech). The monoclonal antibodies to mouse PSGL-1 (clone 2PH1), CD62L (MEL-14), α4 integrin (clones R1-2 and 9C10), CXCR4 (clone 2B11), MHC class II (anti–H-2K^d; SF1-1.1–FITC), and CD11c (HL3-APC) were purchased from BD Biosciences; anti–mouse CXCR2 was an anti-peptide antibody made in rabbits, as previously described (14). Anti–T-bet antibody (clone 4B10) was obtained from Santa Cruz Biotechnology, Inc. IgE-DNP, PMA, ionomycin, LPS, fetal calf serum, and polyclonal antibody to β-actin were purchased from Sigma-Aldrich. Gr1 (anti-Ly6G/C; RB6-8C5–PE) was obtained from eBioscience.

Previously described 6–8-wk-old WT and T-bet^−/− mice (19) on C57/BL6 and BALB/c backgrounds were purchased from the Jackson Laboratory. Mice were maintained in a specific pathogen-free barrier unit at the Harvard New Research Building facilities or at the Dana Farber Cancer Institute animal facilities. All experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee in accordance with policies of the Public Health Service.

BM MCs were generated as previously detailed (7). 0.5 × 10⁶ BMMCs/ml were cultured for 4 wk in T-25 culture flasks (Corning) under standard conditions in the presence of 10 ng/ml IL-3 and 10 ng/ml SCF.

BMDCs were generated as previously detailed (40, 41). BMDCs from WT mice were cultured in the presence of 10ng/ml GM-CSF, with supplementation on days 0, 3, and 6. On day 7, they were analyzed by FACS. The BMDC purity was 65–75% CD11c⁺, class II⁺, and Gr1⁻, with contaminating cells predominantly Gr-1⁺ cells (Fig. S3). This Gr-1 population is likely to contain a mixture of mature granulocytes, monocytes, and macrophages, none of which express T-bet, as previously described (18). 2 × 10⁶ cultured BMDCs/mouse were injected intravenously into T-bet^−/− mice. Also on day 7, they were stimulated with 10 U/ml IFN-γ for 1 h and co-cultured with BMMCs (DC/MC = 1:1.5) for 6 h. These co-cultures were tested for adhesion to VCAM-1 and MAdCAM-1 under shear flow conditions.

Glass coverslips (25-mm diameter; Carolina Biological Supply) were coated with 10 μg/ml VCAM-1 or 20 μg/ml MAdCAM-1. BMMC adhesive interactions were examined under conditions of fluid shear stress using a video microscopy system, as previously described (42). Cell accumulation was determined after the initial minute of each flow rate by counting the number of adherent cells in four different fields using a 20× phase-contrast objective (Plan Apo; Nikon) and recorded with videomicroscopy connected to VideoLab software (Ed Marcus Laboratories). Data are the mean ± SEM (n = 3 experiments). P ≤ 0.5 or less was considered statistically significant using the paired t test or one-way analysis of variance for multiple groups.

The procedure for determining the number of MCp was essentially as previously described (13, 22). Individual tissues from two mice were pooled and processed to isolate MNCs. MNCs were serially diluted in 96-well flat-bottom microtiter plates, and 100 μl of γ-irradiated (30 Gy) spleen-derived feeder cells in media containing 20 ng/ml IL-3 and 20 ng/ml SCF were added. MNCs were cultured for 12 d before MC colonies were identified and counted with inverted microscopes (CK, Olympus; TMS, Nikon. The MCp concentration is expressed as MCp/tissue and is derived by multiplying the concentration of MCp per 10⁶ MNCs by the MNC yield/organ. For the BM, the number represents the total MCp per femur.

BMMCs from WT mice were stimulated for 24 h with 2 μg/ml LPS, 50 ng/ml PMA plus 2 μM ionomycin, or IgE cross-linking (1μg/ml of mouse αDNP-IgE for 16 h plus 10 ng/ml DNP-OVA). Immunoblot analysis was performed by using the 4B10 mAb, as previously described (16).

Sorting of myeloid progenitors was as previously described (30). Myeloid progenitors were sorted as Lin⁻ Sca-1⁻ c-kit⁺ CD34⁺, as previously described (43). BaP or MCp sorting was as previously described (1). Total RNA was extracted from the sorted cells or from in vitro–generated mouse Th1 cells or Tc1 cells (provided by N. Grabie, Brigham and Women's Hospital, Boston, MA) as positive controls using TRIzol solution (Invitrogen) and reverse transcribed using a cDNA synthesis kit (iScript; Bio-Rad Laboratories). The amount of amplicon generated was monitored with an ABI Prism 7700 (Applied Biosystems). A specific probe labeled with both reporter and a quencher dye was added into the Taqman PCR mix at the beginning of the reaction. The cycle number was normalized to β-actin.

For semiquantitative PCR, the following primers were used to amplify T-bet and β-actin: T-bet (forward), 5′-GCCAGGGAACCGCTTATATGTC-3′; T-bet (reverse), 5′-CTGTGAGATCATATCCTTGGGCTG-3′; β-actin (forward), 5′-GTGGGCCGCTCTAGGCACCA-3′; and β-actin (reverse), 5′-CGGTTGGCCTTAGGGTTCAGGGGGG-3′.

Thin sections (1.5-μm thickness) of tissue were fixed with 4% paraformaldehyde and subjected to the chloroacetate esterase reactivity cytochemistry procedure for MC histology analysis, as previously described (8). The results are expressed as the number of MCs per cross-sectional area of the intestine or spleen.

Trichinella larvae were obtained as detailed previously (8). Mice were infected with ∼450 larvae suspended in 200 μl PBS by direct gastric installation. Worm burden was determined after 13 d, as previously described (44). Each experimental group comprised six animals.

Animals were irradiated (GammaCell-40; Atomic Energy of Canada) for 4.5 min, 5 gray (500 rad). 2 h after irradiation, mice were given intravenous injections of 10⁷ BM cells that had been isolated from femurs and tibias of syngenic or congenic animals. 14 d after irradiation, mice were killed, and MCp limiting dilution assays were performed.

Data are the mean ± SEM (n = 5 separate experiments with two mice per group in each experiment), unless indicated otherwise. P ≤ 0.05 was considered statistically significant by the paired t test or one-way analysis of variance for multiple groups using Prism software (version 4; GraphPad).

The online supplemental material contains additional information about adhesion studies under flow conditions, as well as characterization of DCs and BMMCs. Fig. S1 evaluates the adhesion of BMMCs under shear flow conditions to VCAM-1 and MAdCAM-1, coimmobilized with the chemokines KC and SDF-1α. Fig. S2 shows expression of CXCR2 and α4 and β7 integrins in BMMCs by FACS analysis. Fig. S3 shows the phenotypes of the DCs used in our assays.

We wish to thank Dr. Ravi Rao for his helpful discussions regarding T-bet function in MC biology, Dr. Nir Grabie for his gift of mouse Tc1 lymphocytes, Dr. Gail Newton for help with designing primers for T-bet RT-PCR, and Drs. Daniel Friend and Yuhui Xu for MC histological analysis.

These studies were supported by grants from the Fulbright-Spanish Ministry of Education and Science (to P. Alcaide), the P.E. Lindahl Fund from the Royal Swedish Academy of Sciences (to J. Hallgren), the National Institutes of Health (HL36028, HL53993, and HL56985 to F.W. Luscinskas; CA112663 and AI56296 to L.H. Glimcher; and HL 036110 and AI 031599 to M.F. Gurish), and the Ellison Medical Foundation (L.H. Glimcher).

The authors have no conflicting financial interests.