Lung dendritic cells induce migration of protective T cells to the gastrointestinal tract

We conducted a screen to examine the ability of mucosal and nonmucosal DCs to induce gut-homing integrin-α4β7 and CCR9 expression. CD11c⁺ cells were isolated from murine lung, spleen (SpDC), MLNs, and skin-draining LNs (SkDC) and co-cultured with CFSE-labeled T cells from OVA-reactive CD4⁺ T cells from OT-II mice at a ratio of 1:2 DC/T cells for 5–7 d. Unexpectedly, we observed that the expression of integrin α4β7 and CCR9 was significantly up-regulated by lung DCs, similar to MLN DCs, with α4β7 and CCR9 up-regulation most prominent on proliferating (CFSE^lo) cells (Fig. 1, a–c).

To test whether lung DCs can up-regulate α4β7 in vivo, we adoptively transferred CFSE-labeled CD45.1⁺ OT-II cells to naive CD45.2 mice and immunized the mice after 2 h with OVA protein and polyICLC delivered i.n., s.c., i.t., or orally (per os; p.o.). On day 4 after immunization, the percentage of CD45.1⁺Vα2⁺CD4⁺CFSE^lo cells expressing α4β7 was determined in the blood, lung, and spleen. As shown in Figs. 1 (d and e), a significantly higher percentage of CFSE^loα4β7⁺ cells were induced in the blood, lung, and spleen after i.n., i.t., and p.o. immunization than in the blood or spleen after s.c. immunization. Similar findings were observed in the case of CD8⁺OT-I cells (unpublished data).

We next examined the ability of lung DC subsets to induce α4β7. Lung MHC-II⁺CD11c^hi DCs (gating strategy shown in Fig. S1) were FACS-sorted into CD103⁺CD11b⁻ and CD11b⁺CD103⁻ DC populations and co-cultured with CFSE-labeled OT-II cells. Unexpectedly, both CD103⁺CD11b⁻ and CD11b⁺CD103⁻ lung DC subsets induced α4β7 integrin, with expression of α4β7 being even higher in the CD11b⁺CD103⁻ populations. As previously described (Jaensson et al., 2008), among the MLN DCs, CD103⁺CD11b⁻ DCs induced high levels of α4β7, whereas CD11b⁺CD103⁻ DCs did not (Fig. 2, a and b). Additionally, the lung CD11b⁺CD103⁻ population was sorted into CD24⁺CD64⁻ and CD24⁻CD64⁺ populations. We used these markers to distinguish between macrophages and DCs contained within the heterogeneous CD11b⁺ population (Langlet et al., 2012; Tamoutounour et al., 2012). Additionally, although not the focus of this study, a combination of MAR-1 and CD64 can also be used to distinguish between cDCs and MoDCs contained within the CD11b population (Plantinga et al., 2013). We observed striking differences between the CD24⁺ and CD64⁺ subpopulations, with the CD11b⁺CD24⁺ cells inducing both integrin α4β7 and CCR9. In contrast, the CD11b⁺CD64⁺ population was poor in inducing both T cell proliferation and gut-homing molecule induction (Fig. 2, c–e).

Targeting of lung DCs by i.n. delivery (Vermaelen et al., 2001) was confirmed in a time course experiment (unpublished data) where fluorescently labeled antibodies directed against CD11c were administered to C57BL/6 mice i.n. Both CD103⁺CD11b⁻ and CD11b⁺CD103⁻ DC subsets were targeted in the mediastinal LNs and lungs, with the peak labeling occurring 24 h after antibody administration. However, we could not detect antibody-labeled cells in the spleen, MLN, or the GI tissues even after 72 h. To rule out a possible role of gastric acid or GI proteases, cleaving the injected antibodies, orally delivered CD11c was administered to C57BL/6 mice. Peak labeling was detected in the small intestines after 16 h, targeting the CD11b⁺ population (unpublished data). Therefore, transport of labeled antibodies through the GI tract does not affect its ability to be used to stain gut DCs in vivo. To confirm this, we compared i.n. and intratracheal (i.t.) delivery of fluorescent antibodies and found that i.t. immunization targets DCs in the lung similar to those targeted by i.n. immunization (unpublished data).

We next conducted a time course experiment to study the kinetics of induction of integrin α4β7 and the migration of adoptively transferred cells into the GI tissues: SILP, colon, and MLNs. CFSE-labeled CD45.1⁺ OT-II cells were adoptively transferred to naive CD45.2 recipients that were immunized after 2 h with OVA protein and polyICLC-delivered i.n. We tested for the induction of integrin α4β7 on the transferred cells on days 1, 2, 3, 4, and 7 after transfer and quantified the frequency of transferred cells in GI tissues at the same time points (Fig. 3, a and b). Proliferation of the transferred cells was first detected in the mediastinal LNs on day 2 after transfer. On day 3, a sharp rise in the expression of integrin α4β7 was noted on the transferred cells in the mediastinal LNs and the lungs, peaking at day 4 and regressing by day 7. The frequency of transferred CD45.1⁺Vα2⁺CD4⁺ T cells showed a striking rise on day 4 in the GI tissues (SILP, colon, and MLNs) corresponding to the peak expression of integrin α4β7 on lung and mediastinal LN–resident transferred cells. Additionally, we did not detect proliferation of the CFSE-labeled CD45.1⁺ OT-II cells in the MLN on days 1, 2, and 3. The earliest site, where T cell proliferation was observed after i.n. immunization, was the mediastinal LNs (Fig. 3 c). In contrast, proliferating cells were detected in the MLNs by day 4, suggesting that T cell activation initially occurred in the mediastinal LNs and not the MLNs.

Next, we wanted to confirm that the transferred CD45.1⁺Vα2⁺CD4⁺ T cells detected in the GI tissues after i.n. immunization were being generated in the lung-draining LNs and not gut-draining LNs as a result of inadvertent swallowing of i.n.-delivered antigen. We used FTY-720, an inhibitor of lymphocyte egress from LNs (Fig. 4) whereby CFSE-labeled CD45.1⁺ OT-II cells were adoptively transferred to naive CD45.2 recipients that were immunized after 2 h with OVA protein and polyICLC-delivered i.n. On days 0–3, recipient mice were administered FTY-720 (1 µg/g mouse). The mice were sacrificed on day 4, and the frequency of transferred CD45.1⁺Vα2⁺CD4⁺ T cells was quantified in the mediastinal LNs, MLNs, SILP, and colon. An increase in the frequency of transferred cells was noted in the mediastinal LNs. In contrast, we noted a decrease in the frequency of transferred cells in the MLNs, suggesting that antigen exposure and expansion of the transferred OT-II cells was occurring in the mediastinal LNs and not MLNs. As expected, a significant decrease in the frequency of transferred cells was noted in SILP and colon due to their arrest within the mediastinal LN (Fig. 4, a and b). Furthermore, as shown in Fig. 4 c, in the absence of gut exposure, the frequency of CFSE^lo antigen-specific T cells was dramatically lowered in the gut compartments in the FTY-treated group. Thus, based on the in vivo labeling, time course, and FTY experiments, we were able to conclude that after i.n. delivery of antigen, integrin α4β7 was induced on transferred T cells within the mediastinal LNs and lungs and that inadvertent swallowing of i.n.-delivered antigen was not a factor in our experimental system.

Next, we sought to compare the efficiency of lung DC–primed and skin DC–primed T cells to migrate to the GI tract. Mice were immunized i.n. or s.c. (2 h after adoptive transfer of CFSE-labeled CD45.1⁺ OT-II cells to naive CD45.2⁺ mice) with OVA protein and polyICLC. On day 7, after immunization, the frequency of CD45.1⁺Vα2⁺ cells in the small intestines and colon was 3–10 fold higher after i.n. immunization than after s.c. immunization, demonstrating that lung DC–primed congenic OT-II cells migrated to the lamina propria with a significantly greater efficiency than skin DC–primed OT-II cells (Fig. 5, a and b).

We next sought to examine the effect of blocking integrin α4β7 on lung DC–primed, α4β7-expressing cells to the GI tract. After adoptive transfer of CFSE-labeled CD45.1⁺OT-II cells and i.n. immunization with OVA/polyICLC, half of the mice received 100 µg of anti-α4β7 antibody delivered i.p. on days 0 and 3, with the other half receiving PBS as a control. As shown in Fig. 5 (c and d), anti-α4β7 antibody administration led to a significant reduction in the migration of CD4⁺ T cells to the small intestinal and colonic lamina propria, as well as the intraepithelial compartments. Conversely, an increase in the percentage of CD45.1⁺Vα2⁺CD4⁺ T cells was noted in systemic sites such as the blood, spleen, and lung where α4β7 is not required for cellular migration. This demonstrated that i.n. immunization-induced OT-II cell migration to the GI tract was mediated by α4β7.

RA has an unequivocal role in inducing α4β7 (Iwata et al., 2004) and CD103⁺ DCs from PPs, and MLNs are capable of metabolizing vitamin A into RA (Johansson-Lindbom et al., 2005). To investigate the mechanisms mediating integrin α4β7 induction by lung DCs, we started by determining the presence of retinal to RA-oxidizing enzymes, aldehyde dehydrogenases (ALDHs) in lung DCs, using a fluorescent ALDH substrate called ALDEFLOUR(Yokota et al., 2009). Both CD103⁺CD11b⁻ and CD103⁻CD11b⁺ lung DC subsets demonstrated the presence of ALDH at levels comparable to MLN CD103⁺CD11b⁻ DCs (Fig. 6 a). This is in concurrence with data generated by the Immgen Consortium (Heng and Painter, 2008) and a similar study by Guilliams et al. (2010), although the ability of lung DCs to induce integrin α4β7 was not examined in either of those studies. Next, we investigated the impact of inhibiting retinoic acid receptor (RAR) signaling on α4β7 induction. Adding 1 µM RAR-β inhibitor LE540 to DC/OT-II cultures resulted in a significant reduction of α4β7 expression (Figs. 6, b and c). To test for RAR and TGF-β signaling in vivo, we used a dominant-negative RAR (dnRAR) mouse crossed to OT-II on a RAG background (Rajaii et al., 2008; Pino-Lagos et al., 2011), as well as a TGF-βR^def OT-II mouse, respectively (see Materials and methods). WT OT-II cells, dnRAR-OT-II cells, and TGF-βR^def OT-II cells were transferred into congenic mice, and OVA/polyICLC was administered i.n. We found that α4β7 induction was attenuated on TGF-βR^defOT-II cells and virtually absent on dnRAR-OT-II cells, suggesting important roles for TGF-β and RAR signaling in lung DC-mediated α4β7 induction (Fig. 6 d).

Because low levels of integrin α4β7 can be induced by cellular activation, we examined the role of adjuvant in α4β7 induction. After CD45.1⁺OT-II cell transfer into naive CD45.2⁺ hosts, the recipient mice were immunized i.n. with OVA alone, OVA/LPS, OVA/polyICLC, and polyICLC alone. PolyICLC alone did not induce α4β7, whereas OVA alone did, demonstrating the need for antigen presentation. Additionally, OVA/LPS also induced α4β7, suggesting that α4β7 induction was not adjuvant specific (Fig. 6 e).

Finally, to test for involvement of polyICLC-induced RIG-I– and TLR3-mediated DC maturation on α4β7 induction, we transferred CD45.1⁺OT-II cells into WT and MDA-5^−/−TLR3^−/− hosts (McCartney et al., 2009). No significant difference was noted between the MDA-5^−/−TLR3^−/− and WT mice (unpublished data). Thus, we show that expression of α4β7 on T cells after i.n. immunization is not induced in a nonspecific fashion by polyICLC alone.

To test for the role of DCs in the induction of α4β7 after i.n. immunization, we used two different experimental systems. First, we used CD11c-DTR mice (Jung et al., 2002). CD11c-DTR chimeras (CD11c-DTR bone marrow into WT mice) were generated (see Materials and methods) to avoid the lethality of DT treatment in CD11c-DTR mice (Zammit et al., 2005). 24 h after diphtheria toxin (DT) administration, we transferred CD45.1⁺ OT-II cells and immunized with OVA/polyICLC. CD11c-DTR mice administered PBS served as controls. Significantly lower levels of α4β7 were induced on the transferred Vα2⁺CD45.1⁺CD4⁺CFSE^lo cells after DT-mediated ablation of DCs (Fig. 7, a and b). Because CD11c is also expressed on various cells, including activated monocytes, macrophages, and plasmacytoid DCs (pDCs), the CD11c-DTR model cannot definitively distinguish the role of classical DCs (cDCs) from activated monocytes and macrophages (Probst et al., 2005; Zammit et al., 2005; Bennett and Clausen, 2007) in α4β7 induction. To discern the role of lung cDCs in α4β7 induction, we used the recently described zDC-DTR mice (Meredith et al., 2012a,b). In these mice, a zinc finger transcription factor, Zbtb46, which is specific to cDCs, is conditionally ablated, thus distinguishing the role of cDCs from other cells types expressing CD11c. Again, zDC-DTR chimeras (zDC-DTR bone marrow into WT mice) were generated (see Materials and methods) to avoid the lethality of DT treatment in zDC-DTR mice (Meredith et al., 2012a). CD45.1⁺OT-II cells were transferred into zDC-DTR chimeras 24 h after DT ablation, and the mice were immunized with OVA/polyICLC delivered i.n. zDC-DTR mice administered PBS served as controls. Lung DC depletion after DT administration was confirmed (unpublished data). Significantly lower levels of α4β7 were induced on the transferred Vα2⁺CD45.1⁺CD4⁺CFSE^lo cells after DT-mediated ablation of cDCs (Fig. 7, c and d). Thus, using two different methods of DC depletion, we confirmed that lung DCs mediated the induction of integrin α4β7 in vivo.

Contrary to the MLN, where only CD103⁺ DCs (and not CD11b⁺ DCs) up-regulate gut-homing phenotype (Johansson-Lindbom et al., 2005), we have found that both CD103⁺ and CD11b⁺ lung DC subsets express ALDH (Fig. 6) and that both lung DC subsets up-regulated α4β7 and CCR9 in vitro (Fig. 1). Here, we wanted to test the effect of ablating specific lung DC populations on the induction of α4β7 in vivo. To deplete CD11b⁺ lung DCs, we used CD11b-DTR mice (Duffield et al., 2005). CD11b-DTR chimeras were created (CD11b-DTR bone marrow into WT mice). Two doses of DT (25 ng/g) were administered on days 0 and 1. On day 3, CD45.1⁺ OT-II cells were adoptively transferred, and the mice were immunized with OVA and polyICLC. 4 d later, we examined the transferred cells for α4β7 induction. CD11b-DTR chimera that received PBS instead of DT served as controls. As shown in Fig. 8 (a and b), the α4β7 level on transferred Vα2⁺CD45.1⁺CD4⁺CFSE^lo cells in the blood, lung and mediastinal LN were significantly lower in the DT injected mice compared with mice that received PBS. Additionally, we examined the transferred CD45.1⁺ T cells in the spleen and MLN of recipient mice and observed similar attenuation of α4β7 induction (unpublished data). We tested multiple doses of DT and found that two doses of 25 ng/g mouse, 1 d apart, were optimal in effecting depletion of CD11b⁺ lung DCs and mediastinal LN DCs (Fig. 8, c and d). One dose of DT resulted in monocyte depletion in the blood, but not in lung tissue, and three doses of DT were lethal after i.n. administration of PolyICLC (unpublished data).

To examine the effect of depleting CD103⁺ lung DCs, we used two systems: temporal (Langerin DTR; Kissenpfennig et al., 2005) and constitutive (Batf3^−/−; Hildner et al., 2008). In the Langerin DTR mice, 1 d after administration of DT, CD45.1⁺OT-II cells were adoptively transferred and the mice were immunized with OVA and polyICLC. 4 d later, we examined the transferred cells for α4β7 induction. Langerin DTR mice administered PBS served as controls. There was no difference in the levels of α4β7 between PBS- and DT-treated mice (Fig. 9, a and b). Next, we compared Batf3^−/− and WT mice. Again, after adoptive transfer of congenic OT-II cells and immunization with OVA/PolyICLC, no difference was observed between the levels of α4β7 induced on the adoptively transferred cells (Fig. 9, c and d). Together, these data suggest that in vivo, lung CD11b⁺ DC subsets are more proficient in inducing α4β7 than the CD103⁺ DCs.

Having observed that lung DCs can license T cells to migrate to the GI tract, we wanted to examine whether protective gut immunity can be induced after targeting of lung DCs by i.n. immunization. We first transferred CD45.1⁺ OT-II cells into naive mice and immunized i.n. or s.c., with unimmunized mice serving as controls. After gavage of pathogenic Salmonella typhimurium expressing OVA, all the unimmunized or s.c.-immunized mice died between days 6–9. In contrast, survival in the i.n. immunized mice was prolonged by 2–3 d (unpublished data). Transferred Vα2⁺CD45.1⁺CD4⁺ T cells were enriched in the small and large intestinal lamina propria and intraepithelial compartments in the i.n.-immunized mice compared with the s.c.-immunized animals (unpublished data). Having observed that i.n. immunization extended the survival of Salmonella-challenged animals in the transgenic OVA TCR model, we asked whether a protective immune response could be generated from the endogenous immune repertoire. WT mice were immunized with OVA/polyICLC (100 µg and 50 µg, respectively) delivered i.n. or s.c. in a prime-boost fashion or p.o. with OVA/Cholera Toxin (50 mg/ 2 µg; a booster dose of the vaccine was given 4 wk after priming), and challenged orally with Salmonella-OVA. Survival was significantly greater in the i.n.-immunized mice compared with unimmunized mice, with 36% of the mice demonstrating long term survival (>40 d, P < 0.05; Fig. 10 a).Pathological examination of the liver, spleen, and MLN demonstrated significant disruption of hepatic, splenic, and LN architecture, vasculitis, coagulative necrosis, and intravascular thrombi in the challenged, unimmunized mice compared with WT mice, similar to previous studies (Jones and Falkow, 1996; Griffin and McSorley, 2011). In contrast, the liver, spleen, and MLN of i.n.-immunized mice were indistinguishable from WT mice (Fig. 10 b). This is also reflected in the inflammation score (Wu et al., 2005) for each of the vaccinated groups (Fig. 10 c). Stool cultures showed a significantly higher level of Salmonella in the stool of i.n. immunized mice on days 2 and 4 after challenge (unpublished data), perhaps because of luminal sequestration of the bacteria. Interestingly, the i.n.-immunized mice showing long term survival continued to demonstrate detectable levels of Salmonella in the stool even at day 40 (unpublished data). Finally, we also immunized naive C57BL/6 mice with heat and paraformaldehyde inactivated Salmonella typhimurium, delivered i.n. or s.c. Significantly lower levels of Salmonella CFUs were observed in the spleen of i.n.-immunized mice compared with unimmunized mice (Fig. 10 d). Thus, using three different models of protection against pathogenic Salmonella, we found that i.n. immunization extends survival, and is associated with a striking absence of pathology in the liver and spleen in the immunized compared with unimmunized mice.

In this study, we propose a novel pathway of lung DC–mediated T cell migration to the GI lamina propria via induction of gut-homing integrin α4β7. Several lines of evidence pointed us in this direction. First, lungs harbor extrahepatic stores of vitamin A (Okabe et al., 1984; Dirami et al., 2004). Second, lungs are anatomically juxtaposed to the liver (the largest store of vitamin A in the body), directly receiving hepatic blood. Third, lung DCs express the enzyme RALDH (Guilliams et al., 2010) and are capable of oxidizing Vitamin A into RA. Thus, lung DCs have access to the necessary substrate (Vitamin A) and are equipped with the enzymatic machinery required to metabolize it and generate RA. These facts led us to hypothesize that lung DCs may up-regulate integrin α4β7 and induce a gut-homing phenotype on T cells.

We have used multiple experimental approaches to test our hypothesis. First, lung DCs co-cultured with OVA-specific OT-II cells in the presence of OVA express integrin α4β7 at levels comparable to those induced by MLN CD103⁺ DCs. Second, integrin α4β7 is induced on adoptively transferred OT-II cells after i.n. immunization. Third, T cells can be tracked to the GI lamina propria after i.n. immunization and fourth, T cell migration can be blocked by anti-α4β7 antibody, causing the transferred T cells to accumulate in systemic compartments.

It may be argued that i.n. immunization may lead to inadvertent swallowing of antigen and targeting of GI-resident DCs that are known to induce gut migration of T and B cells. To exclude this possibility, we first used a time course experiment to identify the populations targeted by i.n. delivery of fluorescent, DC-specific antibodies. We could readily detect targeting of mediastinal LN- and lung-resident DCs, but not GI-resident DCs. Because this does not exclude migration of antigen to the GI tract, we performed experiments to block the egress of T cells from LNs using FTY-720. In doing so, we observed an accumulation of transferred cells in the mediastinal LNs and, in contrast, a drop in transferred cell frequency in MLNs in mice given FTY-720, thereby confirming that i.n. immunization-mediated expansion of transferred cells was occurring in the former and not the latter lymphoid sites. Finally, we observed early (day 2) proliferation of transferred cells within the mediastinal LNs and the appearance of integrin α4β7 on mediastinal LN- and lung-resident cells (day 3), followed by the migration of transferred cells to GI compartments (day 4), which could be blocked with an anti-α4β7 antibody. These data extend the work of Ciabattini et al. (2011), who recently demonstrated that after i.n. immunization, antigen-specific T cells trafficked to distal skin-draining LNs in a CD62L-dependent fashion and to the MLNs in an α4β7-dependent fashion. However, unlike in the present study, mechanisms of induction of α4β7, i.e., the role of lung DCs or the physiological relevance of these pathways was not explored by Ciabattini et al. (2011).

It is interesting to note that whereas migration to GI sites was blocked using an anti-α4β7 antibody, migration to the systemic sites was not. In fact, there was a nonsignificant increase in the frequency of transferred cells in other sites examined, including the lungs. This lends specificity to our experimental approach, as migration of T cells to the lungs has been shown to be independent of α4β7/MAdCAM but dependent on l-Selectin/PNAd, α4β1/VCAM-1 and LFA-1 (Xu et al., 2003).

To examine the role of lung DCs in the in vivo induction of integrin α4β7, we used two different conditional knockout models. Use of the CD11c DTR leads to the depletion of 85–90% of CD11c⁺ DCs (Bennett and Clausen, 2007). However, because this model cannot distinguish between classical DCs, activated monocytes, and alveolar macrophages, all of which express CD11c and are therefore depleted (Probst et al., 2005; Zammit et al., 2005), we also used the newly described zDC-DTR model (Meredith et al., 2012a,b). Significantly attenuated numbers of transferred congenic cells expressing α4β7 were seen in both models, demonstrating that lung DCs induced integrin α4β7 after i.n. vaccination. These results need to be interpreted with the caveat that both DT-induced depletion of DCs and nasal administration of polyICLC are likely associated with lung inflammation (Tittel et al., 2012). We speculate that local inflammation is responsible for the variability observed in the levels of α4β7 expressed on transferred cells isolated from various compartments, the expression being high on cells isolated from the lungs compared with the cells isolated from the mediastinal LN and blood, or from the spleen and MLN (unpublished data), where a more profound attenuation of α4β7 was seen.

Thus, based on the data generated by different experimental approaches, we propose that after i.n. immunization, lung and mediastinal LN-resident DCs are targeted which in turn induce the expression of integrin α4β7 on T cells, resulting in their localization to the GI tract. Herein, we provide novel functional evidence of mucosal cross talk mediated by DCs. This provides a mechanistic insight to the observation of T cell responses in the GI lamina propria after i.n. challenge (Esplugues et al., 2011; Masopust et al., 2010) or after the appearance of antigen-specific T cells in the MLN after i.n. immunization (Ciabattini et al., 2011).

Although the mechanisms underlying α4β7 induction appear to be conserved between lung- and GI-resident DCs, the DC subsets involved show striking differences. For example, consistent with studies of α4β7 induction by gut DCs (Iwata et al., 2004; Kang et al., 2011), our results show that RA and TGF-β are required in the lung DC–mediated induction of α4β7. Notably, in addition to an in vitro culture system, we have used novel, dominant-negative RAR-OTII Rag^−/− mice and TGF-β receptor^fl/fl-CD4^CRE OT-II Rag^−/− mice to address these questions. Splenic OT-II cells from the dominant-negative RAR-OTII Rag^−/− mice and TGF-β receptor^fl/fl-CD4^CRE OT-II Rag^−/− mice were used as donor cells, so local lung DC subsets in these mice are not relevant here. However, these mice will be described in detail in future studies.

In contrast to the existing literature showing that CD103⁺ MLN DCs induce integrin α4β7, whereas the CD11b⁺ MLN DCs do not, we have found that both CD103⁺ and CD11b⁺ lung DC populations induced integrin α4β7, the expression being higher in the lung CD11b⁺ DC induced co-cultures in vitro. Because these differences were novel and unexpected, we examined them by using in vivo DC depletion models. To deplete the CD103⁺ DC populations, we used Langerin DTR (Bennett et al., 2005) mice and could not detect any significant differences in the level of α4β7 induction between the langerin depleted and replete mice. To further confirm the effect of CD103⁺ DC depletion, we used the Batf3^−/− mice, in which the deletion of the transcription factor Batf3 results in ablation of cross presenting, CD103⁺ tissue DCs (Hildner et al., 2008). Again, no significant differences in the induction of α4β7 were observed between WT and Batf3^−/− animals. Finally, to deplete the CD11b⁺ lung DCs, we used the CD11b DTR mice. Both CD11b-expressing alveolar macrophages and DCs are depleted in this model; however, the use of zDC-DTR demonstrates the role of classical DCs. Attenuated levels of α4β7 were observed in the mice in which CD11b-expressing cells were depleted. It may be noted that although the dose of DT administered to the Langerin DTR and CD11b DTR mice was different, we were guided by previous studies on the dose required to deplete the respective DC subsets (Duffield et al., 2005; Kissenpfennig et al., 2005). Additionally, in this study, we confirmed the depletion of respective lung DC populations. Therefore, in contrast with MLN DCs, both CD103⁺ and CD11b⁺ lung DC populations induced integrin α4β7 in vitro, although in vivo, the CD11b⁺ populations appeared to have a more dominant effect on the induction of α4β7. Furthermore, within the heterogeneous CD11b⁺ population, we observed a striking difference between the CD24⁺ and CD64⁺ subsets, where the CD24⁺CD11b⁺ DCs induced robust levels of α4β7 and CCR9, whereas CD64⁺CD11b⁺ cells did not. We speculate that whereas the migratory gut CD103⁺ DCs use dietary and biliary sources of vitamin A to generate RA (Jaensson-Gyllenbäck et al., 2011), the lung DCs either use hematogenous vitamin A, released from hepatic stores, or local stores of vitamin A (Dirami et al., 2004; Okabe et al., 1984). Thus, the predominance of CD103⁺ DC in mediating gut-homing, as in the case of MLN, is not observed in case of lung DCs. This further allows us to speculate that specific DC populations, i.e., CD103⁺ DCs, may not be preconditioned to metabolize vitamin A but rather, they may acquire this property based on local environmental factors like the availability of vitamin A, TLR ligands, or other inflammatory stimuli. Detailed mechanistic studies are underway to further address this question, including a study of the relative expression of RALDH and the effect of adjuvants on RALDH expression by lung DC subsets.

Integrins may be induced via nonspecific stimuli, such as cellular activation (Laudanna et al., 2002). Because our in vivo experimental system involved use of the TLR-3 agonist, polyICLC, we tested its effect on the induction of integrin α4β7. Initially, we used polyICLC alone (without antigen) but could not observe any significant induction of α4β7, whereas OVA alone induced α4β7 on transferred OTII cells. In addition, using the MDA5^−/−TLR3^−/− mice, we were able to confirm that i.n. immunization-induced α4β7 was not a nonspecific effect of adjuvant. Additionally, we have tested the use of different adjuvants in the induction of α4β7 after i.n. immunization and observed no significant differences between them (unpublished data). Combined, we speculate that in the right cellular context, i.e., in the presence of local availability of RA and TGF-β, the use of an adjuvant amplifies the induction of α4β7 because of DC and T cell activation, as well as DC migration.

Having shown that lung DCs induce α4β7 and induce cellular migration to the GI tract, we wanted to study if pathogen-specific intestinal immunity can be induced by targeting lung DCs. We chose Salmonella typhimurium, a pathogen of global significance (Levine, 2006) that is transmitted across the intestinal mucosa and causes a spectrum of diseases ranging from localized intestinal infection to severe systemic illness (Griffin and McSorley, 2011). Notably, although Salmonella is not a lung pathogen, we used this model as a tool to examine i.n. vaccine induced protection against Salmonella invasion across the GI tract. To confirm protection against enteric challenge with Salmonella , we have used three different experimental systems: (1) passive immunization using transferred OT-II cells followed by i.n. immunization, where we demonstrated a modest increase in survival; (2) active prime-boost immunization with OVA showing a significantly increased survival and lack of systemic pathology in the i.n.-immunized mice; and (3) active prime-boost immunization using inactivated Salmonella showing reduced systemic burden in the i.n.-immunized mice. Thus, using multiple experimental systems, we were able to show that lung DCs targeted by i.n. immunization are able to induce protective immunity within the GI tract. Notably, in mice that received s.c. vaccine, a nonsignificant increase in survival was also observed. We hypothesize this to be due to the systemic immunity generated by s.c. vaccine and plan to study this further with the use of IgA-deficient, CD103^−/− and B7^−/− mice.

To conclude, in this manuscript we challenge the dogma that only CD103⁺ gut-resident DCs can recruit T cells to the GI tract. Rather, we provide compelling evidence that lung DCs targeted by i.n. immunization are capable of inducing robust levels of gut-homing integrin α4β7 and licensing T cells to migrate to the GI tract. Pathways of recruitment of antigen-specific immune cells to the gut are therefore much more promiscuous than previously appreciated. There appears to be considerable, albeit hitherto unrecognized, DC-orchestrated, mucosal cross talk that can inform the rational design of novel vaccines.

C57BL/6 (B6), C57BL/6-Tg(TcraTcrb)425Cbn/J (OT-II), and C57BL/6-Tg(TcraTcrb)1100MjbJ (OT-I) mice were purchased from Taconic Farms or bred at The Rockefeller University. 6–8-wk-old CD11c-DTR (B6.FVB-Tg(Itgax-DTR/EGFP)57Lan/J) mice were purchased from The Jackson Laboratory. CD4^CRETGF-βR^fl/fl OT-II mice and the dnRAR-OT-II mice were provided by the Mucida Laboratory (The Rockefeller University, New York, NY). In brief, CD4^CRE mice were crossed with TGF-βR^fl/fl mice (provided to the Mucida laboratory by the National Cancer Institute, Rockville MD) to specifically deplete TGF-βR from CD4 T cells. To avoid autoimmune disease and premature death in CD4^CRE x TGF-βR^fl/fl mice, these animals were backcrossed to OT-II Rag^−/− animals to obtain TGF-βR^def OT-II (TGF-βR^def) mice (these mice will be described elsewhere). To analyze RA signaling in CD4⁺ T cells, CD4^CRE mice were crossed with dnRAR^fl/fl mice(Pino-Lagos et al., 2011; Rajaii et al., 2008) to specifically express the dominant-negative form of RAR in CD4⁺ T cells. To address the importance of RA for mucosal T cell migration after specific T cell activation, these animals were backcrossed to an OT-II Rag^−/− background. MDA-5^−/−TLR3^−/− mice (McCartney et al., 2009) were provided by M. Colonna (Washington University, St. Louis, MO).

Bone marrow chimeras were created by transferring bone marrow from CD11c-DTR, zDC-DTR, and CD11b DTR mice into lethally irradiated WT C57BL/6 mice to avoid lethality associated with DT treatments of CD11c-DTR (Jung et al., 2002; Zammit et al., 2005), zDC-DTR (Meredith et al., 2012a), or CD11b DTR. In brief, C57BL/6 hosts underwent two treatments of 500 rads in an animal γ-irradiator 3-h apart, and 2.5 × 10⁶ bone marrow cells from CD11c-DTR, zDC-DTR, and CD11b-DTR donor, respectively, were transferred i.v. The resulting mice (referred to herein as CD11c-DTR, zDC-DTR, and CD11b-DTR) were housed under immunocompromised mouse conditions and treated regularly with 2 mg/ml neomycin from Sigma-Aldrich in their drinking water. The chimeric mice were fully reconstituted and ready for experimental use after 6 wk.

Mice were maintained in specific pathogen–free conditions. Protocols were approved by the The Rockefeller University Animal Care and Use Committee.

The following reagents were purchased from BD: anti-CD3 (500A2), anti-CD3 (145–2011), anti-CD103 (M290), anti-CD4 (RM4-5), and anti-CD8 (53–6.9). The following reagents were purchased from eBioscience: anti-MHCII (M5/114.15.2), anti-IFN-γ (XMG1.2), anti-CD4 (RM4-5), and anti-CD8 (53–6.9). The following antibodies were obtained from BioLegend: anti-CD45 (30-F11), anti-CD45.1 (A20), anti-CD45.2 anti-CD11c (N418), anti-CD11b (M1/70), anti-IFN-γ (XMG1.2), anti-CD103 (M290), anti-LPAM-1 (DATK32), anti-CD24 (M1/6a), anti-CCR9 (CW12), anti-CD64 (X54-51711), and anti-TCRVα2 (B20.1). AQUA (L34957) was obtained from Invitrogen, and CFSE was purchased from Sigma. Anti-CD205 (NLDCC) was produced locally, as previously described (Trumpfheller et al., 2006).

OT-I or OT-II mice were used as cell donors for adoptive transfer into syngeneic recipient animals. Lymphocytes were isolated from the spleen and labeled with 5 µM CFSE (Invitrogen) for 10 min at 37°C. After washing twice with PBS containing 3% FCS, 7 × 10⁶ cells per mouse were injected via the retroorbital venous plexus.

CFSE-labeled OT-I or OT-II mice were adoptively transferred into syngeneic age- and gender-matched WT mice and controls. 2 h later, the recipient mice were immunized i.n., i.t., or s.c. with 100 µg OVA (LPS-free; Seikagaku Corp.) and 50 µg of poly ICLC (Oncovir, Inc.) using Isoflurane as a short-acting inhalant anesthetic. Additionally, recipient mice were immunized p.o. with 50 mg of OVA (Grade III; Sigma-Aldrich) and 2 µg of cholera toxin (Sigma-Aldrich). All animals were given food and water ad libitum.

Mice were euthanized and blood was obtained from the experimental mice by cardiac puncture. Lung mononuclear cell isolation was performed as previously described (Vermaelen et al., 2001). In brief, after sacrificing the animals, the pulmonary circulation was perfused with saline to remove the intravascular pool of cells. Lungs were carefully separated from thymic and cardiovascular remnants and removed in toto. Organs were thoroughly minced using iridectomy scissors and incubated for 60 min in digestion medium containing collagenase D in a humidified incubator at 37°C and 5% CO₂. In the last 5 min, 10 mM EDTA was added. Tissue fragments were disrupted mechanically by pipetting and passed through a 70-µm cell strainer. Mononuclear cells were washed twice in RPMI before use in subsequent experiments as described below.

The spleen and LN were harvested in RPMI media with 5% FCS, mashed between two sterile glass slides, washed with RPMI complete media, and then filtered through a 70-µm filter. Spleen but not LN samples went through subsequent ACK lysis, were washed twice, and counted. LN single-cell suspensions were prepared by incubation with 400 U/ml collagenase D at 37°C for 30 min.

Intestinal lymphocytes were isolated and prepared as previously described (Mucida et al., 2007). In brief, small and large intestines were removed and placed in cold HBSS media containing 5% FCS. The intestines were carefully cleaned from the mesentery and flushed of fecal content. Intestines were opened longitudinally and then cut into 1-cm pieces. The intestinal tissue was incubated with 1.3 mM EDTA (Cellgro) in HBSS at 37°C for 1 h. The supernatants containing intestinal epithelial cell (IEC) with some superficial villi cells, referred to as the “IEC fraction,” were collected and layered over a discontinuous 40/80% Percoll gradient, and centrifuged at 2,800 rpm for 30 min. Cells from the 40/80% interface were collected, washed, and resuspended in complete RPMI media. These purified cells constituted the intraepithelial lymphocyte (IEL) population. To isolate the lamina propria lymphocytes (LPL), the remaining intestinal tissue was minced and transferred to conical tubes. The minced pieces were resuspended in 20 ml of complete RPMI containing 0.125 mg/ml of collagenase (Sigma-Aldrich) and shaken at 200 rpm for 50 min at 37°C. The tissue suspension was collected and passed through a 70-µm cell strainer and the cells were pelleted by centrifugation at 1,200 rpm. The cells were then resuspended and layered onto a 40/80% Percoll gradient, centrifuged, and processed as described above for the IEL preparation.

Single-cell suspensions were isolated from the murine lung, spleen, mesenteric, mediastinal, and skin-draining LNs as described above. CD11c⁺ cells were isolated as previously described (Ing et al., 2006) using CD11c magnetic beads (Miltenyi Biotec). In brief, the cells were washed in 1 ml of MACS buffer (Miltenyi Biotec), before 30-min incubation on ice with CD11c⁺ beads. The cellular suspensions were washed twice in MACS buffer and passed through a magnetic column. CD11c⁺ cells were isolated by positive selection, washed, resuspended in complete RPMI medium, and counted before co-culture with T cells, as described in the following section.

Lung mononuclear cells were isolated as described above. Using magnetic beads, CD11c⁺ cells were enriched and sorted on FACSAria (BD) as B220⁻MHC II⁺CD11c^hiCD103⁺CD11b⁻ and B220⁻MHC II⁺CD11c^hiCD103⁻CD11b⁺ subsets.

The culture medium used for DC/T cell cultures was RPMI (Invitrogen) supplemented with 10% heat-inactivated FCS, 2 mM l-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, and 5 mM β-mercaptoethanol (all from Sigma-Aldrich). Allogeneic T cells were isolated from spleens of C57BL/6 mice by excluding B220⁺, F4/80⁺, CD49b⁺, and I-Ab⁺ cells with anti–rat IgG Dynabeads (Invitrogen). T cells were added to round-bottom microtest wells at 10⁵/well and mixed with isolated DCs at a DC/T cell ratio of 1:2, 1:5, and 1:10 in the presence of OVA (25 µg/ml final). After 5 d, the expression of α4β7 on proliferating T cells was evaluated by FACS. For the retinoic acid inhibition assays, the DC/T cell co-cultures were incubated with 1 µM of RAR-β antagonist, LE540 (Wako; dissolved in DMSO at a stock concentration of 1 mM), and added to cultures at final concentration of 1 µM.

Cells were isolated as described in the preceeding section. Before staining, cells were washed and resuspended in staining buffer containing 1× PBS, 2% BSA, 10 mM EDTA, and 0.01% NaN₃. To block nonspecific staining, the 2.4G2 anti-CD16/32 antibody was added. Antibodies for cell surface markers were added and cells were incubated for 25 min at 4°C. After the staining, the cells were washed twice and analyzed immediately or fixed in PBS containing 1% paraformaldehyde and 0.01% NaN₃, and analyzed later on an LSR II (BD) using multiparameter flow cytometry.

For intracellular cytokine staining, after surface staining, cells were resuspended in Fixation/Permeabilization solution (Cytofix/Cytoperm kit; BD), and intracellular cytokine staining was performed according to the manufacturer’s protocol. Flow cytometric data were analyzed with FlowJo software (Tree Star). For analysis of the adoptively transferred OT-I and OT-II populations, the transferred cells were identified by forward and side scatter characteristics and dead cells were excluded using Live-dead fixable Aqua (Invitrogen), followed by sequential gating on CD45.1⁺, CD3⁺, CD4⁺Vα2⁺ double-positive populations followed by the expression of α4β7 on CFSE^lo cells.

ALDH activity was determined using the ALDEFLUOR staining kit (STEMCELL Technologies) according to the manufacturer’s instructions, with some modifications. In brief, LN and lung mononuclear cells were resuspended at 10⁶ cells/ml in ALDEFLUOR assay buffer containing activated ALDEFLUOR substrate (final concentration, 1.5 µM) with or without the aldehyde dehydrogenase inhibitor DEAB (final concentration, 45 µM) and incubated for 30 min at 37°C. For immunophenotyping of ALDEFLUOR-reactive cells, the cells were subsequently stained with PE-, PE-Cy7–, A700-, PerCP Cy5.5-, and Pacific blue–conjugated mAbs in ice-cold ALDEFLUOR assay buffer. ALDEFLUOR-reactive cells were detected in the FITC channel using multi-parameter flow cytometry, as described above.

DT purchased from Sigma-Aldrich was prepared in a sterile solution of PBS at a concentration of 1 mg/ml. Transient DC depletion was effected in CD11c-DTR chimeras by a single i.p. injection of 200 ng DT; in zDC-DTR mice, a single 100-ng i.p. injection of DT was given i.p.; in CD11b-DTR mice, two 500-ng injections were given on days 0 and 1 i.p.; and in langerin DTR mice, a single dose of 100 ng of DT was given i.p.

Recipients of adoptively transferred OT-II cells, mice immunized i.n. or s.c. with OVA, and mice immunized with heat inactivated, paraformaldehyde-fixed Salmonella (10⁶ CFU) were inoculated orally by gavage with WT Salmonella-OVA (10⁶ CFU/animal). Mice were monitored with recording of daily weights. Stool culture was performed by culturing preweighed stool pellets on days 2, 4, 8, 12, and 40 after challenge (day 40 stool cultures were performed in the long-term survivors to detect for persistence of Salmonella in the stool) and expressed as Salmonella CFU/g of stool tissue. For splenic cultures, the spleen was harvested from the Salmonella-challenged mice on day 6 after challenge, and then weighed, ground in LB broth, and inoculated on Macconkey’s Agar in serial dilutions. Kaplan-Meier curves were used to determine the frequency of survival among the challenged mice.

Statistical significance was evaluated using a two-tailed Student’s t test with a 95% confidence interval. Results are expressed as means ± SD. Analysis was performed with a Prism 4 program (GraphPad Software, Inc.).

Fig. S1 shows the gating strategy used to identify and sort lung DCs.

This work is dedicated to the memory of Dr. R.M. Steinman. We would like to thank Dr. M. Nussenzweig for his valuable comments and suggestions, J. Adams for help with the figures and M. Nulty and J. Chiappetta for administrative support. PolyICLC was kindly provided by A.M. Salazar.

This work was supported by grants from the American Gastroenterology Association Elsevier Award (S. Mehandru), The Rockefeller University Clinical and Translation science award pilot project from National Institutes of Health (NIH)/National Center for Research Resources (5UL1RR024143-05; S. Mehandru), the Leona M. and Harry B. Helmsley Charitable Trust, Crohn’s & Colitis Foundation of America Senior Award and Ellison Medical Foundation New Scholars Award in Aging (D. Mucida), and NIH/National Institute of Allergy and Infectious Disease Exploratory/Development grant (1R21AI082331-01; Y. Do), Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2012-0003632; Y. Do) and, finally, C. Cheong was supported by the Canadian Institutes of Health Research (MOP-125933). Lastly, we would like to thank Irish Research Council for Science Engineering and Technology (IRCSET-RS/2010/2207; D. Ruane).

The authors have no further conflicts of interest.