These authors contributed equally to this work.
Steady state migratory RelB+ langerin+ dermal dendritic cells mediate peripheral induction of antigen-specific CD4+CD25+Foxp3+ regulatory T cells
Article first published online: 14 APR 2011
Copyright © 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
European Journal of Immunology
Volume 41, Issue 5, pages 1420–1434, May 2011
How to Cite
Azukizawa, H., Döhler, A., Kanazawa, N., Nayak, A., Lipp, M., Malissen, B., Autenrieth, I., Katayama, I., Riemann, M., Weih, F., Berberich-Siebelt, F. and Lutz, M. B. (2011), Steady state migratory RelB+ langerin+ dermal dendritic cells mediate peripheral induction of antigen-specific CD4+CD25+Foxp3+ regulatory T cells. Eur. J. Immunol., 41: 1420–1434. doi: 10.1002/eji.201040930
- Issue published online: 26 APR 2011
- Article first published online: 14 APR 2011
- Accepted manuscript online: 7 MAR 2011 09:22AM EST
- Manuscript Accepted: 23 FEB 2011
- Manuscript Revised: 20 JAN 2011
- Manuscript Received: 9 AUG 2010
- Deutsche Forschungsgemeinschaft (DFG)
- Interdisciplinary Center for Clinical Research Erlangen (IZKF)
- Collaborative Research Center. Grant Number: SFB643
- Transregio Collaborative Research Centre. Grant Number: TR52
- Graduate Program. Grant Number: GK520
- Cell trafficking;
- Immune regulation;
- Treg cells
- Top of page
- Materials and methods
- Supporting Information
Tolerance to self-antigens expressed in peripheral organs is maintained by CD4+ CD25+ Foxp3+ Treg cells, which are generated as a result of thymic selection or peripheral induction. Here, we demonstrate that steady-state migratory DCs from the skin mediated Treg conversion in draining lymph nodes of mice. These DCs displayed a partially mature MHC IIint CD86int CD40hi CCR7+ phenotype, used endogenous TGF-β for conversion and showed nuclear RelB translocation. Deficiency of the alternative NF-κB signaling pathway (RelB/p52) reduced steady-state migration of DCs. These DCs transported and directly presented soluble OVA provided by s.c. implanted osmotic minipumps, as well as cell-associated epidermal OVA in transgenic K5-mOVA mice to CD4+ OVA-specific TCR-transgenic OT-II T cells. The langerin+ dermal DC subset, but not epidermal Langerhans cells, mediated conversion of naive OT-II×RAG-1−/− T cells into proliferating CD4+ CD25+ Foxp3+ Tregs. Thus, our data suggest that steady-state migratory RelB+ TGF-β+ langerin+ dermal DCs mediate peripheral Treg conversion in response to epidermal antigen in skin-draining lymph nodes.
- Top of page
- Materials and methods
- Supporting Information
Immature DCs reside in peripheral tissues where they capture antigens of various types. Recognition of pathogenic structures then initiates DC maturation and migration to draining lymph nodes for induction of immunity against the transported pathogens 1. Also, under steady-state conditions immature DCs capture and present antigens in secondary lymphoid organs, presumably to induce tolerance 2, 3. These immature DCs are attached to the lymph node reticular conduit system where small antigens drained from peripheral tissues can be taken up from the fluid phase for presentation to T cells under steady-state conditions 4. Recently, it has been reported that continuous peripheral antigen delivery can induce Treg conversion in draining lymph nodes. This suggests that the reticular conduit system delivers exogenous peptides under non-immunogenic conditions, which can be mimicked by implanted osmotic pumps, to lymph node-resident immature DCs, which then present these peptides to induce Tregs in vivo 5.
Such mechanisms of antigen delivery may explain the induction of CD4+ T-cell anergy as shown for antigen presentation by immature DCs expressing only low levels of MHC II and costimulatory molecules 6. However, it does not explain how immature DCs may induce CD4+ Treg since B7-1/B7-2−/− and CD28−/− mice lack CD4+ CD25+ Tregs despite the presence of immature DCs 7. Splenic DCs converted Treg better than B cells, and CD80/86-deficiency of the DCs further increases the conversion. These conversion experiments were either in vitro studies or under a pathological situation (tumor) but not steady state 8. Later, the splenic CD8α+ DC subset was identified as mediating conversion after injection of a DEC205-targeting antibody 9. The physiological relevance of splenic DCs or DEC205 for Treg conversion with self-antigens remains open. A CD103+ DC population has been isolated from mesenteric lymph nodes that mediated TGF-β-dependent Treg conversion in culture, but the maturation stage of the DC has not been analyzed 10. In contrast, a CD103− DC subset within skin-draining lymph nodes was responsible for Treg conversion in vitro 11. This indicates that CD103 expression by DCs is not correlated with their capacity to convert Tregs and a specific phenotype for DCs inducing Tregs under steady-state conditions and in vivo remains to be identified 12.
Fluid phase soluble antigen transport also does not explain how lymph node T cells can be tolerized against cell-associated antigens. Therefore, the presentation of self-antigens by migratory DCs has been suggested as a mechanism of peripheral tolerance since these cells can be observed in draining lymph nodes of various organs 13, 14. The mature CCR7-expressing DC fraction within peripheral skin-draining lymph nodes of mice has been identified as consisting of migrated Langerhans cells (LCs) and dermal DCs 14, 15.
We have shown previously that in vitro generated and TNF-matured DCs differentiated into a semi-mature phenotype could tolerogenic 16 and, in other systems such as inflammation-mediated DC maturation, were unable to differentiate immunogenic CD4+ T helper cells 17. Therefore, semi-mature DCs with tolerogenic potential can be distinguished from fully matured DCs with immunogenic functions 13. The demonstration of the presence and function of such a partially mature and tolerogenic DC phenotype in vivo, however, remained open.
Three subsets of migratory DCs have been identified in skin-draining lymph nodes of mice, which consist of epidermal langerin+ LCs and two subsets of langerin+ and langerin− dermal DCs 18–20. Recent evidence suggests that in the K5-mOVA mouse model cross-tolerance of OVA-specific CD8+ OT-I T cells to this epidermal neo-self-antigen is mediated predominantly by the CD103+ langerin+ dermal DCs but not the other subtypes 21, 22.
Here, we found no clear correlation for CD103 as a marker for steady-state migratory DCs but that extrathymic self-antigen presentation by skin-derived RelB+ langerin+ dermal DCs results in the de novo generation of Foxp3+ Tregs in the periphery.
- Top of page
- Materials and methods
- Supporting Information
Skin-draining lymph nodes contain resident immature and migratory semi-mature DCs in the steady state
As reported by others before 14, 23 skin-draining lymph nodes of WT but not CCR7−/− mice contained CD11c+ DCs expressing high levels of CD40 (CD40hi) as a sign of maturity. This was not observed in mesenteric lymph nodes or spleen, whereas DCs expressing low levels of CD40 (CD40low) were found in all three organs (Fig. 1A). Consequently, only CD40hi DCs but not CD40low DCs expressed the lymph node homing receptor CCR7 on their surface (Fig. 1B). Also, in agreement with previous findings 14, 24, CCR7−/− mice showed smaller lymph nodes (Supporting Information. Fig. 1). The CD103 marker has been used to identify tolerogenic, Tregs-converting DCs in the mesenteric lymph nodes and lamina propria 10, 25. We found similar proportions of CD103 expression by both resident CD40low and migratory CD40hi DC subsets (Supporting Information Fig. 2).
Steady-state and immunogenic migratory DCs show different maturation stages
It is under debate how tolerogenic antigen uptake and presentation of self-antigens by DCs are handled during infections. To investigate how steady-state migratory DCs might be influenced under inflammatory conditions, we compared their surface profiles of MHC II, CD80 and CD86 with migratory DCs that have been matured by FITC-painting, known to induce contact hypersensitivity responses. After 1 day, FITC was detected transiently until day 6 in a fraction of CD40hi migratory DCs but not in CD40low resident DCs, suggesting that the hapten was transported from the skin to the draining lymph nodes by CD40hi migratory DCs. In the same lymph node CD40hi FITC− DCs, representing steady-state migratory DCs, were also constantly present throughout the sensitization phase (Fig. 2). Interestingly, the steady-state migratory DCs retained their partially mature phenotype during sensitization, while the FITC+ migratory DC fraction expressed transiently much higher levels of CD86, CD80 and MHC II molecules (Fig. 2 and Supporting Information Fig. 3). These findings support the concept that two different maturation stages of DCs can migrate in parallel to provide steady-state and inflammatory/pathogen-derived antigen transport.
Steady-state migratory CD40hi DC express RelB, which is translocated into the nucleus
Within the Rel/NF-κB transcription factor family, tissue-specific and inducible members have been described. Since RelB was identified as the constitutively active κB-binding activity in lymphoid tissues under steady-state conditions 26, we investigated expression levels and subcellular distribution of RelB in sorted DC subsets. Analyses of peripheral lymph node DCs revealed that only the CD11c+ CD40hi migratory DCs expressed intracellular RelB (Fig. 3A). This expression was equally strong in EpCAMhi DCs, which represent epidermal LC and langerin+ dermal DC subsets 22, and EpCAMlow DCs, reported to be langerin− dermal DCs 22, but was not restricted to CD103+ DCs that represent a cross-tolerizing DC subtype 21 (Fig. 3B). Thus, steady-state migration of DCs correlates with RelB expression but not with a specific skin DC subset.
When the intracellular distribution of RelB was determined by confocal microscopy only very low levels of RelB were detected in the cytoplasm of resident DCs, but RelB expression was increased in steady-state migratory DCs. A substantial amount localized to the nucleus, similar to what we observed for FITC− DCs of contact-sensitized mice, which represent steady-state migratory DCs under FITC treatment (Fig. 3C and Supporting Information Fig. 4). In FITC+ DCs RelB expression in the nuclei remained high. In contrast, c-Rel could not be detected in the nuclei of resident, steady-state migratory, FITC− or FITC+ DCs, while nuclear RelA (p65) staining appeared in all four subsets (Fig. 3C and Supporting Information Fig. 4). Thus, nuclear translocation of RelB is observed not only under immunogenic conditions but also in migratory DCs during the steady state.
We then asked whether nuclear RelB is of functional relevance for the appearance of steady-state migratory DCs in the lymph nodes. Since homozygous RelB-deficient mice lack lymph nodes 27, we tested heterozygous relB+/− mice for their proportions of DC subsets in peripheral lymph nodes. Although lymph nodes from relB+/− mice appeared to be slightly smaller than their WT counterparts, the relative number of resident CD40low DCs remained unchanged whereas the frequency of steady-state migratory CD40hi DCs was significantly decreased within relB+/− skin-draining lymph nodes (Fig. 3D). To further confirm a specific role of the alternative NF-κB pathway, the RelB-binding partner p52 was investigated. The lymph nodes of p52−/− mice were much smaller than those of WT or p50−/− mice (Fig. 3E) due to a known specific loss of B220+ B cells in these mice 28. Peripheral lymph nodes of p52−/− mice showed a dramatic reduction in the frequency of migratory but only partial effects on resident DCs (Fig. 3E). To investigate whether the classical NF-κB pathway through RelA/p50 would contribute to steady-state DC migration, we analyzed p50−/− mice. The binding partner RelA could not be screened this way since relA−/− mice die at the embryonic stage 27. Peripheral lymph nodes of p50−/− mice showed equally reduced numbers of resident and migratory DCs. This indicates that p50 affects DCs at the immature stage but this defect is not enhanced upon partial maturation as observed for the steady-state migratory DCs (Fig. 3E). Thus, activation of the alternative NF-κB pathway through RelB/p52 correlated with the partial maturation state of DCs and was required for a normal frequency of steady-state migratory DCs in skin-draining lymph nodes.
Only migratory DCs transport and present epidermal OVA
K5-mOVA transgenic mice express membrane-bound (cell-associated) OVA in epidermal keratinocytes as well as in epithelial cells within the thymus and esophagus under the control of the keratin-5 promoter 29. Recently, we reported that the epidermal neo-self-antigen OVA was transported and cross-presented in K5-mOVA mice by migratory DCs into skin-draining lymph nodes leading to deletion of OVA-specific OT-I CD8+ T cells 30. To investigate which populations within peripheral lymph nodes were able to present OVA to CD4+ T cells we sorted CD11c+ and CD11c− cells prom peripheral lymph nodes of K5-mOVA WT mice or crossed with CCR7−/− mice in which migratory DCs are missing. The different DCs were then cultured with BO97 OVA-specific hybridoma T cells. The results indicated that BO97 cells only responded to CD11c+ cells from mice containing migratory DCs and mice heterozygous for CCR7 show intermediate responses (Fig. 4A). Then resident CD40low or migratory CD40high expressing CD11c+ DCs from peripheral lymph nodes of K5-mOVA mice were cultured with BO97 cells. While resident DCs could not stimulate the BO97 cells, migratory DCs were able to present OVA (Fig. 4B).
We further investigated whether DCs from skin-draining lymph nodes of K5-mOVA mice would also stimulate fresh OVA-specific CD4+ T cells. Migratory CD40hi and resident CD40low DCs were sorted from pooled skin-draining lymph nodes of either K5-mOVA or WT mice. These DCs were incubated with anti-CD40 and LPS to achieve full maturation and with either purified CD4+ OT-II×RAG-1−/− or CD8+ OT-I T cells. After 60 h, OVA-specific responses of CD4+ or CD8+ T cells were detected only by steady-state migratory CD40hi DCs from K5-mOVA mice but not with CD40low DCs from K5-mOVA mice or DCs from WT controls (Fig. 4C). While cell division of OT-I cells was low but readily detectable, OT-II cells showed clear CD69 but little CD25 upregulation and the proliferation remained very low (Fig. 4C). Thus, in vitro steady-state migratory DCs but not resident DCs contained low amounts of endogenous OVA. The transported epidermal self-antigen OVA allowed substantial cross-presentation but hardly conventional MHC II presentation. Importantly, resident DCs could not present or cross-present OVA under steady-state conditions, ruling out transfer from migratory to resident DC subtypes.
Steady-state migratory DCs mediate Treg conversion in vitro by using endogenous TGF-β
To address whether OT-II cell activation by steady-state migratory DCs would enable TGF-β-dependent Treg conversion, the differential DC subsets were first stained on their surface for the TGF-β-associated molecule latency-associated peptide (LAP), which has been shown to correlate with Treg conversion by human DCs 31. The highest expression was found on CD40high DCs, while it was lower on CD40low DCs both from peripheral lymph nodes and at intermediate levels on CD11c+ DCs from mesenteric lymph nodes (Fig. 5A). To test whether this differential LAP expression would reflect different Treg conversion rates, in vitro co-cultures of WT DCs and purified CD4+ CD25− OT-II cells in the presence of OVA were performed for 5 days and then tested for CD25 and Foxp3 expression. While all conditions led to CD25 upregulation, only CD40high DCs showed a substantial conversion that could be blocked by anti-TGF-β (Fig. 5B). Resident CD40low DCs were unable to generate Foxp3+ cells, but addition of exogenous TGF-β allowed some Foxp3+ Treg induction (Fig. 5B). Total CD11c+ DCs from mesenteric lymph nodes were similarly able to perform this in vitro conversion (data not shown) as reported before 10. Addition of exogenous porcine TGF-β did not further increase the conversion rate of CD40high DCs but rather seemed to compete with the endogenous TGF-β, leading to lower conversion rates.
Thus, steady-state migratory DCs but not lymph node resident DCs have the capacity to convert CD4+ T cells into Tregs by using endogenous TGF-β/LAP complexes on their surface.
Treg conversion in vivo occurs through langerin+ dermal DCs
The minipump system was used before to show Treg conversion in vivo 5. To investigate whether steady-state migratory DCs can transport and present soluble self-antigen to T cells in vivo also in our system, we transferred CFSE+ OVA-specific transgenic OT-II cells into mice implanted with an osmotic minipump secreting continuously low amounts of OVA327–339 peptide. Activated CD4+ CD25+ Foxp3− and CD4+ CD25+ Foxp3+ Tregs of OT-II origin accumulated in the skin-draining lymph nodes of OVA peptide pump-implanted mice similarly as described before 5. Control mice implanted with a PBS pump accumulated neither activated nor regulatory OT-II T cells (Fig. 6A, upper row).
Moreover, K5-mOVA mice that received OT-II cells for the same period of time presented cell-associated self-antigens in skin-draining lymph nodes as indicated by the appearance of activated CD25+ Foxp3− and regulatory CD25+ Foxp3+ OT-II cells. To clarify whether activation and Treg induction in both systems was dependent on migratory DCs, OT-II cells were transferred into OVA peptide pump-implanted mice or K5-mOVA mice lacking CCR7, the chemokine receptor known to be required for DC migration to lymph nodes 14. In the absence of migratory DCs in CCR7−/− mice, both activation and Treg induction were completely aborted (Fig. 6A, lower row). This suggests that low doses of soluble and cell-associated antigens require transport by DCs to draining lymph nodes to accumulate activated and Tregs.
To control whether fluid phase transport of antigen is intact in CCR7−/− mice, CFSE-labeled OT-II cells were transferred into CCR7−/− mice that were then injected s.c. with titrated amounts of OVA327–339 peptide. In both WT and CCR7−/− mice OT-II cells proliferated dose-dependently, indicating an intact conduit system and antigen presentation by the resident DC subset in CCR7−/− mice (Supporting Information Fig. 5). The proliferation rates of OT-II cells were slightly reduced in CCR7−/− mice, which may reflect the contribution of migratory DCs to the antigen presentation in the lymph node.
The skin is populated by epidermal LCs and two subsets of langerin+ and langerin− dermal DCs 18–20, which are all migratory and may transport antigens from osmotic pumps or cell-associated epidermal OVA in K5-mOVA mice. To test whether langerin+ DCs were involved in the transport of keratinocyte-associated epidermal OVA, we crossed K5-mOVA mice with mice expressing the diphtheria toxin receptor (DTR) under the langerin promoter (langerin-DTR mice), which allows ablation of LCs and langerin+ dermal DCs by diphtheria toxin injection 19. Analysis of transferred OT-II cells in these mice showed no activation or Treg induction (Fig. 6B). These results indicate that migratory langerin+ dermal DCs or LCs transport epidermal self-antigen to the lymph nodes.
To further discriminate between these two DC subtypes bone marrow chimera experiments were performed. K5-mOVA mice were irradiated and reconstituted with MHC II−/− bone marrow cells. In the skin of these mice only the radioresistant LCs remain MHC II+ while both dermal DC subsets appear MHC II− (Supporting Information Fig. 6). Transfer of OT-II cells into these chimeras abrogated Vβ5+ OT-II cell proliferation and their conversion into Foxp3+ Tregs whereas WT chimeras remained unaffected (Fig. 6C). Thus, only the langerin+ dermal DC subset appears competent for Treg appearance in K5-mOVA mice. These data are consistent with findings that OVA cross-presentation to CD8+ OT-I cells in K5-mOVA mice is also performed by this DC subset 21, 22.
Steady-state migratory DCs induce naive T-cell conversion, expansion and regulation of Tregs
In the previous experiments bulk OT-II cells were adoptively transferred, which contained both naive conventional CD4+ T cells and Treg. To determine whether the appearance of CD4+ CD25+ Foxp3+ OT-II Tregs in peripheral lymph nodes from K5-mOVA mice was due to the trapping or expansion of pre-existing Tregs or due to peripheral conversion of naive T cells, we used purified CD4+ T cells from OT-II×RAG-1−/− mice for the adoptive transfer into K5-mOVA mice. In OT-II mice on a RAG-deficient background only naive conventional CD4+ T cells were released from the thymus but not Foxp3+ Tregs (Supporting Information Fig. 7). Analysis of skin-draining and mesenteric lymph nodes as well as the spleen indicated that after 13 days CD4+ CD25+ Foxp3+ Tregs appeared predominantly in peripheral lymph nodes (Fig. 7A). Since the adoptively transferred cells were also labeled with CFSE, we were able to follow their proliferation. Analysis of the CD4+ CFSE+ population indicated that these cells clearly divided and, among those, the proportion of CD25+ Foxp3+ cells was higher in dividing than non-dividing cells (Fig. 7B). Although expression of Foxp3 in murine CD4+ T cells is strictly linked to functional suppression, the regulatory capacity of induced OT-II Tregs was tested by OVA injection into WT or K5-mOVA mice that were either adoptively transferred by CD25− OT-II cells 10 days before to allow their Treg conversion or were left untreated. Then, CFSE-labeled CD25− OT-II responder cells were transferred into all mice at the day of OVA-injection. The proliferative capacity of the CFSE+ OT-II responder cells was inhibited in WT mice by 18%, most likely due to the activity of CD4+ CD25− Foxp3+ Tregs that have been transferred (data not shown). The inhibition of proliferation in K5-mOVA mice was clearly higher by 30%, indicating an increased regulatory activity (Fig. 7C). These data suggest that conversion of naive CD4+ OT-II cells into 2–3% proliferating Tregs by steady-state migratory DCs that could be observed after 14 days in the draining lymph nodes of K5-mOVA mice led almost to a doubling in the suppressive capacity from 18 to 30% within the same time period.
- Top of page
- Materials and methods
- Supporting Information
Immature DCs and CD4+ CD25+ Foxp3+ Tregs represent major tolerogenic immune cells 32–34. Therefore, immature DCs were also believed to be major interaction partners and inducers of Tregs 12, 35. However, for Treg homeostasis CD28 costimulation is required 7, which cannot be provided by immature DCs. Mature DCs highly express costimulatory molecules but mainly act immunogenic 32. Surprisingly, mature DCs have also been shown to be superior to immature DCs in the expansion of pre-formed Tregs in mice 36–39. It remained unclear, however, which DC maturation stage was responsible for conversion of naive CD4+ T cells into Foxp3+ Tregs under steady-state conditions in vivo. We and others proposed earlier that a semi-mature DC stage, such as after stimulation with TNF or E-cadherin disruption, may lead to optimal induction of Foxp3− IL-10+ T cells 13, 40, 41. For human thymic “more mature” DCs the conversion into Foxp3+ Tregs has been shown 42. Thus, a certain extent of DC maturation but not immature DCs may induce de novo Treg conversion under physiological conditions.
Here we showed that steady-state migratory DCs in peripheral lymph nodes display a semi-mature DC phenotype, characterized by intermediate surface expression of MHC II, CD80, CD86, CD40, nuclear RelB and surface LAP/TGF-β complexes. Functionally they transport CCR7-dependent peripheral self-antigens to the draining lymph nodes and present them with some costimulation. They convert naive CD4+ T cells into Tregs. In the case of cell-associated epidermal OVA, the langerin+ dermal DC subset was required to mediate this effect.
Several authors observed that peripheral lymph nodes of mice contain a fraction of mature CD40hi DCs 14, 43–45. Here steady-state migratory DCs already show nuclear activity of RelB but have only a partially mature phenotype since costimulatory molecule expression on FITC+ DCs was still higher. Our data suggest that nuclear activity of RelB and p52 are needed for inducing or maintaining the phenotype of the steady-state migratory DCs. RelB has previously been described to be expressed in the T-cell areas of lymph nodes 46. RelB-deficient mice lack peripheral lymph nodes and the CD8αneg myeloid DC lineage, indicating that this transcription factor is important for organogenesis of certain secondary lymphoid organs and myeloid DC development 47, 48. These mice also develop T-cell-dependent multi-organ inflammation, which further points to a role of RelB as an anti-inflammatory factor in the steady state 49. RelB may bind to its partner p52, since the p52−/− mice also showed a severe reduction of migratory DCs. In contrast, p50−/− mice displayed no specific effect on migratory DCs and both resident and migratory DC populations of the peripheral lymph nodes were equally affected. This points to a role of p50 already at the immature DC stage since the effect was not further increased after maturation into steady-state migratory DCs. Together, our data suggest that steady-state migratory DCs partially depend on the alternative NF-κB pathway, mediated through RelB/p52, whereas inflammatory DC maturation may occur through the classical NF-κB pathway as suggested by the requirement of p50 for IL-12p70 and c-Rel for IL-27 production by DCs 50, 51.
Naturally occurring Tregs, which develop in the thymus, play a major role for the maintenance of peripheral tolerance. However, peripherally induced Tregs have been reported only recently. The chronic delivery of peptides by implantation of osmotic pumps led to de novo induction of Tregs in the pump-draining peripheral lymph nodes after 2 wk 5. Also targeting the DC receptor DEC205 but not 33D1 of splenic DCs induced Foxp3+ Tregs in the steady state in vivo 9. In models of oral tolerance, it was shown that a population of CD103+ DCs isolated from mesenteric lymph nodes 10 mediated Treg conversion in vitro. Similarly, a CD103+ DC was identified in the lamina propria to mediate Treg conversion in vitro or under lymphopenic conditions in vivo 25. Since exogenous peptide or anti-CD3 had to be added to demonstrate Treg conversion, it remained unclear whether these DCs could transport orally administered antigens from the gut to mesenteric lymph nodes or lamina propria. In skin-draining lymph nodes CD103− DCs were responsible for Treg conversion in vitro 52. In vivo Treg conversion data for a skin-derived self-antigen are still lacking. We found similar CD103 levels on a subpopulation (∼20%) of migratory and resident DCs from peripheral or mesenteric lymph nodes, suggesting that this marker may not generally serve to identify CCR7+ migratory DCs with the ability to transport self-antigens and convert Tregs. Comparing the efficiency of CD4+ T-cell activation and Treg conversion in vitro by exogenously added peptide or in vivo by endogenous OVA from the DCs it appears that the in vitro and in vivo response to endogenous OVA is rather low. The poor efficacy of Treg conversion in our system may reflect the physiological situation where only small amounts of self-antigens are presented by few steady-state migratory DCs. It is of note that our system only relies on endogenous antigen, and not high doses of orally applied OVA, lymphopenic conditions or antibody targeting. The OVA pump control experiments and in vitro Treg conversion rates showed a poor Foxp3+ Treg induction, which may also indicate an intrinsically poor capacity of the OT-II TCR to be activated/converted in this system.
We found that besides Treg conversion also a substantial activation of conventional T cells occurred, appearing as a CD4+ CD25+ Foxp3− subset. This was not observed by others using the same minipump system but different TCR-transgenic T cells 5. We cannot explain the differential behavior of the transgenic T cells but the activated conventional T cells in our system may produce IL-2 and thus explain the proliferation of Tregs.
Treg conversion requires TGF-β to induce the transcription factor Foxp3 and retinoic acid counteracts costimulation on the converting APCs 8, 10, 53. Our data indicate that steady-state migratory but not resident DCs of peripheral lymph nodes directly activate naive CD4+ T cells and use endogenous TGF-β for Treg conversion, which was detectable by LAP staining at the cell surface, similar as described for splenic CD8α+ DEC205+ DCs ex vivo 54 or by in vivo targeting to DEC205 9, 55. However, a contribution of splenic DCs to the repertoire of peripherally induced Tregs remains to be shown. Although successful antibody targeting of DEC205 has been shown for all three skin DC subsets 56 and this pathway opens a wide array of therapeutical implications 57, physiological ligands of this receptor and its contribution for tolerance induction are not known. Whether the TGF-β is produced by the DC itself or captured form environmental sources such as shown for intestinal epithelial cells remains open 58. Although we tested retinoic acid inhibitors our results remained unclear and require further investigations, since retinoic acid can also be produced by epithelia 58.
It has been shown that CD80/86 costimulation by DCs counteracts Treg conversion and therefore suppression of CD80/86 expression by retinoic acid may favor DC immaturity and thereby Treg conversion 8, 54. In contrast, in vitro conversion stimulated through antibodies instead of APCs requires also CD28 antibodies in conjunction with anti-CD3 and TGF-β 59, indicating that some CD80/86 costimulation may be required or it can be substituted by other molecules expressed on DCs such as PD-L1 or GITR 54. DCs also needed to express CD80/CD86 to convert Tregs efficiently in vitro by using splenic DC subsets because this costimulation induced conventional T cells to produce IL-2, which was required for the conversion 39, 60. Our data suggest that the semi-mature phenotype of steady-state migratory DCs provides endogenous TGF-β, detected as LAP on their surface, and an intermediate optimal dose of CD80/86 for Treg conversion.
One would expect that epidermal LCs in K5-mOVA mice capture epidermal cell-associated OVA antigen much better than dermal DCs. However, we did not observe Treg induction in draining lymph nodes from bone marrow chimeric mice in which only radioresistant LCs remained capable of MHC II presentation to OT-II cells. The results from K5-mOVA×langerin-DTR mice, where both epidermal LCs and langerin+ dermal DCs were depleted, provide strong evidence that steady-state migratory langerin+ dermal DCs represent the major responsible DC population for OVA transport and presentation in our system. Similar data were obtained for the induction of cross-tolerance in K5-mOVA mice 21, 22, although a minor role for cross-presentation by LCs could not be fully excluded in other experimental settings. When OVA-specific CD8+ OT-I T cells were transferred into LC-ablated K5-mOVA mice they still proliferated, indicating that epidermal self-antigen was transported and cross-presented by both LCs and dermal DCs in the skin-draining lymph node 30 (and data not shown). Models of how dermal DCs may access epidermal antigens are discussed elsewhere 15.
After adoptive transfer of TCR-transgenic Tregs their proliferation could be observed in lymph nodes draining the immunization site with mature DCs or adjuvant 60–62. However, in situations where Treg induction was followed under steady-state conditions, as s.c. installed osmotic pumps, proliferation of Tregs has not been reported 5. Here, we found that transgenic OVA that was transported by steady-state migratory DCs not only converted naive CD4+ T cells into Tregs, but also stimulated their proliferation, although to a moderate extent of three to four divisions in 2 wk. Despite proliferation their suppressive potential was maintained and the proportion of Foxp3+ cells was increasing with the number of cell divisions. Suppression of secondarily transferred OT-II cells was stronger in K5-mOVA mice than WT controls. Since we hardly observed proliferation of OT-II cells after implantation of OVA-secreting pumps (data not shown), proliferation of Tregs may be induced only by cell-associated but not by soluble antigens under steady-state conditions. By comparing directly the osmotic pump system with the K5-mOVA mice providing cell-associated antigens, we found a similar rate of Treg conversion in WT but not in CCR7−/− mice, which lack migratory DCs. The functionality of the reticular conduit system in CCR7−/− mice was shown by the presence of WT OT-II T cells in T-cell areas by OT-II cell proliferation after injection of high doses of soluble peptides. For very low doses of soluble antigen in the peripheral tissue, as provided by the pump system, transport by CCR7+ migratory DC was required.
In conclusion, our data indicate that low levels of soluble or cell-associated neo-self-antigens in the skin require transport and presentation by CCR7+ RelB+ steady-state migratory DCs and cannot be mediated by fluid phase transport to immature lymph node-resident DCs. In K5-mOVA mice steady-state migratory langerin+ dermal DCs are the major subset in converting adoptively transferred naive CD4+ T cells into proliferating Foxp3-expressing Tregs. Only migratory but not resident DCs can mediate Treg conversion by endogenous TGF-β. Thus, we identified and characterized the antigen presenting cell type, which is responsible for the de novo Treg induction under physiological conditions in skin-draining lymph nodes in vivo.
Materials and methods
- Top of page
- Materials and methods
- Supporting Information
All mice were bred and maintained at the Universities of Erlangen and Würzburg. K5-mOVA mice were obtained from Department of Dermatology, Osaka University 29; and OT-I and OT-II mice were kindly provided by Francis Carbone, Melbourne, Australia. OT-II mice were crossed with RAG-1−/− mice (gifted from Thomas Winkler, Erlangen, Germany). CCR7−/− mice 24 and langerin-DTR-EGFP transgenic mice 63, each on a C57BL/6 background, were crossed with K5-mOVA. C57BL/6 and 129 mice were purchased from Charles River. MHC II−/− mice by Horst Bluethmann, F. Hoffmann-La Roche, Basel, Switzerland.
For bone marrow chimera experiments K5-mOVA mice were irradiated with two doses of 4.5 Gy each with an interval of 4 h. After another 4 h mice were reconstituted with 5×106 bone marrow cells from MHC II−/− or WT mice as a control. All animal experiments were performed in accordance with institutional guidelines and permission (Regierung von Unterfranken 55.2-2531.01-73/07) with age- and sex-matched animals.
Preparation and cell sorting of DCs from lymph nodes
Skin-draining lymph nodes (cervical, axillar, brachial and inguinal) were cut into small pieces and digested for 20 min at room temperature with 1 mg/mL DNase I (Sigma) and 1 mg/mL collagenase type III (Worthington) in RPMI 1640 containing 10% FCS, 50 μM 2-ME, 2 mM L-glutamine, 100 U/mL Penicillin (Sigma) and Streptomycin 100 μg/mL (Sigma). Lymph node tissue was then incubated in the same media for another 5 min at room temperature by adding 0.01 M EDTA to disrupt T cell–DC complexes. Then, the suspensions were passed through a 70 μm cell strainer to remove debris and cells were resuspended in PBS containing 5% FCS and 1 mM EDTA. From this step onwards, cells were always kept on ice or 4°C. Cells were stained with mAB against CD11c (HL-3) and CD40 (3.23). Cells were washed and resuspended in PBS containing 2% FCS, 1 mM EDTA and 1 μg/mL DAPI. CD40hi and CD40low cells from CD11c+ DAPI− population were sorted with a MoFlo high-speed sorter (Cytomation).
Implantation of osmotic pumps secreting OVA327–339 peptide
Osmotic minipumps (Alzet ♯1002) were filled with an OVA327–339 peptide solution in PBS to secrete 10 μg/day for 14 days or PBS only, as previously described 5. The pump was inserted into the s.c. cavity of recipient mice after a small incision in the back, and the wound was closed by the AUTOCLIP system (Becton Dickinson, BD). Two days after implantation, OT-II cells were adoptively transferred.
FACS analysis of DCs
Surface staining was performed using the following mAbs purchased from BD if not otherwise indicated: CD11c-FITC, -PE or -eFlour450 (eBioscience), CD40-biotin, -FITC or -APC (Miltenyi), CCR7-biotin (eBioscience), EpCAM-PE (eBioscience), CD103-biotin, MHC II-FITC or -PE, CD80-FITC, CD86-FITC, CD205 (Serotec), CD4-PE and CD8-PerCP and LAP-PE (R&D Systems). Biotinylated antibodies were detected by incubation with either PerCP- or APC-conjugated streptavidin (BD). As isotype controls the following fluorochrom-conjugated or biotinylated mAbs were used (all BD): mouse IgG1, rat IgG1, rat IgG2a, rat IgG2b, armenian hamster IgG2 and armenian hamster IgG1. Intracellular RelB staining (polyclonal rabbit IgG, Santa Cruz), was performed after 2% formaldehyde fixation and ice cold 90% methanol permeabilization. For detection a secondary goat-anti-rabbit IgG F(ab)2 FITC-conjugate (Dianova) was used. Cells were measured with a FACSScan, FACSCalibur or FACSCanto II (BD) and analyzed with CellQuest (BD) or FlowJo software (Stanford University).
Adoptive transfer of OT-II cells
Cervical, axillary, brachial, inguinal and mesenteric lymph nodes and spleens were isolated from conventional OT-II or OT-II×RAG-1−/− mice and single-cell suspensions prepared. CD4+ cells were enriched by using the “CD4 negative isolation kit” (purity >90%, Dynal). Briefly, cells were labeled with rat-anti-CD8, -CD11b, -CD16/32 (2.4G2), -CD45R (B220), -TER119 mAbs, followed by magnetic depletion with sheep-anti-rat Ig-Dynabeads conjugates (Dynal Biotech/Invitrogen). The untouched fraction was collected, and labeled with 5 μM CFSE (CFDA SE, Molecular Probes/Invitrogen) at 37°C for 10 min. Cells were washed with PBS and injected into the tail vein of recipient mice. After 3–13 days, recipient mice were sacrificed and cell suspensions were prepared from lymph nodes and spleen. Extracellular staining was performed using CD4-PerCP and CD25-APC mAbs (BD). Intracellular Foxp3 staining was conducted with the murine Foxp3-PE staining kit (eBioscience) following the manufactor's instructions. Isotype control stainings included the monoclonal rat IgG1-APC and rat IgG2a-PE (BD).
Immunization of mice with converted Tregs
K5-mOVA or WT control mice were left untreated or reconstituted with 6×106 CD4+ CD25− OT-II cells. After 10 days all mice were adoptively transferred with 2×106 CFSE+ CD4+ CD25− OT-II cells and injected with 100 μg OVA peptide s.c. into the foot pads. Three days later the popliteal and inguinal lymph nodes were removed and the single-cell suspension tested by FACS analysis for CD4, Vβ5 and CFSE.
In vitro Treg conversion assay
Naive CD25− CD4+ T cells from lymph nodes and spleen of OT-II mice were separated using first a mouse CD4+ T Cell Enrichment Kit (CD4+ purity >90%, StemCells) and second CD25 MACS Micro Beads (CD25− CD4+ purity >90%, Miltenyi). CD40high CD11c+ and CD40low CD11c+ DC from skin-draining lymph nodes were isolated and sorted as described above. A total of 20 000 CD25− CD4+ OT-II T cells were cultured in round bottom 96-well plates with 6 000 DCs for 5 days in the presence of 100 ng/mL OVA327–339 peptide with or without 2 ng/mL porcine TGF-β (R&D Systems) as described previously. A blocking anti-TGF-β (clone 1D11, R&D Systems) or mouse IgG1 control (clone 11711, R&D Systems) was added at a final concentration of 20 μg/mL. After 5 days cultures were stained with CD4-PacificBlue, Vβ5-FITC, CD25-APC (BD) and Foxp3-PE (eBioscience) and analyzed by FACS.
Coculture of DC subpopulations with T cells in vitro
As indicated in the respective figure legends total CD11c+ or CD11c− cells or sorted migratory CD11c+ CD40hi or resident CD11c+ CD40low DCs were sorted from skin-draining lymph nodes of WT or K5-mOVA mice or the indicated homozygous or heterozygous crossings with CCR7−/− mice. The sorted cell populations were matured with LPS (1 μg/mL, SIGMA) plus anti-CD40 (3/23, 5 μg/mL, BD) and cultured at titrated amounts with OVA323–339-specific 5×104 BO97 hybridoma T cells (BO97.10.5, OVA-specific and I-Ab-restricted, was a gift from Philippa Marrack, Jewish Medical Center, Denver, CO, USA). After 24 h the culture supernatants were collected and tested for their IL-2 content by ELISA (BD). OT-I CD8+ T cells and OT-II×RAG-1−/− CD4+ T cells were purified by magnetic cell sorting (Dynal) and labeled with CFSE. In total, 30 000 DCs and T cells were mixed at a 1:1 ratio in the presence of anti-CD40 (5 μg/mL, BD) and LPS (1 μg/mL, Sigma) to achieve full DC maturation and cultured for 60 h at 37°C, 5% CO2. As a positive control, OVA257–264 or OVA327–339 peptides were added to the T-cell cultures with WT migratory DCs. FACS staining included counterstainings for CD69 and CD25 (BD).
Visualization of FITC-transporting migratory DCs
FITC isomer I (Sigma) was dissolved in DMSO (Stock concentration 0.5 mg/mL) and mixed with 1:1 acetone/dibutylphtalate (Sigma) to a final FITC concentration of 0.5% w/v as described 45 and modified 64. A volume of 100 μL of this 0.5% FITC solution was painted on the shaved abdomen of the mice. After 1, 2, 4 or 6 days, the draining axillary, brachial and inguinal lymph nodes were colleted from those mice. Cells were digested and separated as described above. For FACS analysis, cells were stained with CD11c-APC (HL-3) and CD40-biotin (3.23) followed by streptavidine-PerCP (BD) for 30 min on ice.
Immunofluorescence and confocal microscopy
To detect nuclear translocation of Rel/NF-κB transcription factors, DCs were isolated from peripheral lymph nodes of untreated or FITC-painted mice as described above and FACS-sorted according their expression of CD11c and CD40low or CD40high. Cytospin preparations of the isolated DCs were dried overnight at room temperature, fixed in 4% formaldehyde and permeabilized with 0.2% Triton X-100 followed by blocking in 1:20 diluted donkey serum for 20 min. For immunofluorescence staining anti-mouse RelB Ab (Santa Cruz, rabbit polyclonal, 1:50 dilution), anti-mouse RelA Ab (Santa Cruz, rabbit polyclonal, 1:50 dilution) or anti-mouse cRel (Santa Cruz, rabbit polyclonal, 1:50 dilution) were used, followed by an anti-rabbit Cy3 Ab (1:500, Jackson Immunoresearch), each 30 min at room temperature. Slides were mounted with Fluoromount-G (Southern Biotechnology Associates) containing DAPI and images were taken with a confocal microscope (Leica TCS SP2, Wetzlar, Germany). To determine the MFI of nuclear RelB, RelA and cRel, a distinct area of the nuclei (40–47 μm2) from 20–30 cells per condition was analyzed by Leica LCS software.
To detect MHC II in skin sections from the indicated bone marrow chimeras or WT control mice anti-MHC II staining was combined with hematoxylin staining. Briefly, cryostat sections (9 μm) were fixed with 4% paraformaldehyde, incubated with 10% BSA/PBS to block unspecific binding of immunoglobulins and stained with a pure rat MHC II (clone 2G9, BD) or a rat IgG2a isotype control mAb (BD) followed by an rat IgG-biotin mAb (BA-4001, Vector, U.S.), a streptavidin–AB complex (DAKO) and development with 3,3′-diaminobenzidine substrate (Fluka). Sections were counterstained with hematoxylin, dehydrated in a graded series of ethanol (76–100%) and embedded with permanent mounting medium (Eukitt, Merck).
The paired Student's t-test (Microsoft Excel software) was used for determining the significance of experiments. p<0.05 were considered as statistically significant.
- Top of page
- Materials and methods
- Supporting Information
Transgenic and knockout mice or reagents were kindly provided by Horst Bluethmann, Manfred Kopf and Francis Carbone. We thank Gerold Schuler and Thomas Hünig for their generous support of this project; Khash Khazaie, Ludger Klein and Francis Carbone for critical reading of the manuscript and helpful comments; Susanne Rößner and Melanie Schott for expert technical assistance; Katrien Pletinckx and Isabell Senft for their help on mouse preparations, Martin Väth for his help with confocal microscopy and Uwe Appelt and Christian Linden for cell sorting. This work was supported by the Deutsche Forschungsgemeinschaft (DFG) through the Interdisciplinary Center for Clinical Research Erlangen (IZKF) for N.K. and M.B.L., the Collaborative Research Center (SFB643) for H.A. and M.B.L., the Transregio Collaborative Research Centre (TR52) for A.D., M.L., F.B.S. and M.B.L. and the Graduate Program (GK520) for A.D. and M.B.L.
Conflict of interest: The authors declare no financial or commercial conflict of interest.
- Top of page
- Materials and methods
- Supporting Information
- Top of page
- Materials and methods
- Supporting Information
Detailed facts of importance to specialist readers are published as ”Supporting Information”. Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors.
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.