|Home | About | Journals | Submit | Contact Us | Français|
Immunological tolerance is crucial to avoid autoimmune and inflammatory diseases, however the mechanisms involved are incompletely understood. To study peripheral tolerance to skin-associated antigens, we generated new transgenic mice expressing a membrane-bound form of ovalbumin in skin under the human K14-promoter (K14mOVA mice). In contrast to other transgenic mice expressing similar self-antigens in skin, adoptive transfer of antigen-specific T cells does not induce inflammatory skin disease in our K14-mOVA mice. OVA-specific T cells transferred into K14-mOVA mice are activated in lymphoid tissues, undergo clonal expansion and eventually acquire effector function. Importantly, these antigen-specific T cells selectively up-regulate expression of E-selectin ligand (ESL) in cutaneous lymph nodes but not in mesenteric lymph nodes and spleen, demonstrating that expression of endogenous self-antigens in skin dictates imprinting of skin tissue homing in vivo. However, an additional inflammatory signal, here induced by tape stripping, is required in K14-mOVA mice to induce T cell migration to skin and development of inflammatory skin disease. Depletion of regulatory CD4+ CD25+ T cells did not provoke homing of transferred T cells to skin under steady-state conditions, indicating that these cells are not the key regulators for inhibiting T cell homing in K14-mOVA mice. Both skin-derived and LN-resident CD8α+ DCs are responsible for antigen presentation in vivo and induce tolerance to skin antigens, as we show by selective depletion of Langerin+ and CD11c+ DCs. Taken together, controlled skin homing of T cells is critical for the maintenance of peripheral immune tolerance to epidermal self-antigens.
Dendritic cells (DCs) play an important role in the development of efficient and appropriate immune responses. They are highly efficient antigen presenting cells (APCs) that acquire antigen by continuously sampling their immediate surrounding tissues, including barrier tissues such as the skin and the intestines (1, 2). DCs internalize and process antigens for presentation as peptide/MHC complexes to naïve T cells, and thereby play a critical role in shaping immune responses. To this end, DCs need to display a correct profile of adhesion and tissue homing molecules allowing them to migrate from peripheral blood to tissues and subsequently from peripheral tissues to secondary lymphoid organs to encounter naïve T cells (3). Controlled DC migration is a crucial aspect of cutaneous and mucosal immune response to infection and a prominent feature in the development of inflammatory skin disorders such as psoriasis, atopic dermatitis, allergic contact hypersensitivity (CHS) and cutaneous T cell lymphomas. In addition to DC migration to skin, it is equally important that other leukocytes, such as T and B lymphocytes, macrophages, and neutrophils also participate in the local immune response and therefore acquire the ability to migrate to peripheral tissues. The extravasation of leukocytes into tissue through endothelia involves a multi-step cascade of molecular and cellular interactions, summarized as tethering, rolling, activation by chemoattractants, adhesion and diapedesis (4). The initial tethering and rolling steps are principally mediated by members of a family of type I membrane proteins called selectins (5). E-selectin is constitutively expressed on human endothelium (6) and is up-regulated, along with P-selectin, in response to inflammatory stimuli (7–9). E-and P-selectin ligands (ESL and PSL) are critical for T cell trafficking to inflammed skin and together are considered key skin homing receptors on leukocytes. In humans, skin homing effector/memory T cells express cutaneous lymphocyte-associated antigen (CLA), a carbohydrate epitope that serves as a ligand for E-selectin (10, 11), and that is recognized by the specific antibody HECA-452 (12). In mice, detection of ESL with a recombinant fusion protein has allowed the elucidation of the role of ESL in homing to inflamed skin (13, 14). In addition, the generation of mice deficient in E-and P-selectin has clarified their importance in mediating the control of skin inflammatory disorders (7, 15). Lymphocytes acquire their requisite tissue migratory capacity during antigen stimulation in a lymphoid tissue environment. Indeed, T cells activated in cutaneous lymph nodes (CLN) will display skin homing receptors, such as CCR4, ESL or PSL, whereas T cells activated in mesenteric lymph nodes (MLN) or Peyer's patches principally express high levels of CCR9 and α4β7 and home preferentially to the gut (16, 17). The imprinting of tissue homing receptors is dictated by local lymphoid organ DC subsets, as well as by the immunization route used (18, 19). Intracutaneous injection of antigen-pulsed bone marrow (BM) DCs can for example induce the expression of skin homing molecules on T cells (20). In vitro cultures of these skin-imprinted T cells with MLN-isolated DCs down-regulate ESL and induce CCR9 and α4β7 expression, demonstrating possible reprogramming of tissue-polarized T cells in presence of specific DC subsets (20). In addition to their role in defining the homing properties of T cells during the induction of immune responses, DCs are involved in the maintenance of peripheral immune tolerance to self-antigens (21, 22). Under steady-state conditions, peripheral tolerance prevails by inactivation or deletion of autoreactive T cells, while in the context of infection and inflammation immunity may be initiated. New strategies now allow for the study of the mechanisms of peripheral tolerance in vivo to exogenous innocuous antigens. Such antigens can be selectively delivered to DCs in the absence of inflammatory signals, by use of chemically coupled exogenous proteins with antibodies targeting endocytic receptors expressed on DCs (23). The generation of transgenic (Tg) mouse models expressing self-antigens under a tissue-specific promoter also lead to great advances in this research field (24, 25). Several Tg mice expressing epidermal self-antigens were generated and allowed the study of immune responses to skin-associated antigens (26,30). Interestingly, although these transgenic lines differ in the promoter used or in the structure of the self-antigen expressed, they all developed inflammatory skin disease upon adoptive transfer of naïve antigen-specific T cells (26–28). We developed K14-mOVA mice, a new transgenic mouse model expressing OVA as a self-antigen in epidermis, to elucidate the imprinting of skin tissue homing characteristics on self antigen-specific T cells. In addition, we propose to study the outcome of such imprinting and especially its possible role in induction of tolerance or inflammatory skin disease. In these K14-mOVA mice, no skin disease is induced upon adoptive transfer of antigen-specific T cells. Importantly, these OVA-specific T cells selectively up-regulate expression of ESL in CLN but not in MLN and spleen (SPL), demonstrating that expression of endogenous self-antigens in skin can dictate imprinting of skin tissue homing in vivo. However, an additional inflammatory signal is required in K14-mOVA mice to induce the migration of transferred antigen-specific T cells to skin and development of inflammatory skin disease. We identified skin-derived DCs and LN-resident CD8α+ DCs as the central players in the induction of tolerance in these mice. Therefore, our new transgenic model allowed us to show that controlled skin homing of antigen-specific T cells is critical for the maintenance of peripheral immune tolerance to epidermal self-antigen.
C57BL/6, C57BL/6.PL, OT-I and OT-II Tg (The Jackson Laboratories), K14-mOVA (see below), CD11c-DTR (31), Langerin-DTR and Langerin-EGFP (32) mice were bred and maintained in the animal facility of Harvard Medical School (Boston, MA) under specific pathogen-free conditions. All mouse procedures were approved by the Harvard Medical School Standing Committee on the use of Animals in Research.
The membrane-bound form of OVA (mOVA) consists of a fusion protein containing amino acids 1–118 of the human transferrin transmembrane receptor fused to the amino acid sequence 139–386 of the chicken ovalbumin (33). A DNA fragment containing the sequence encoding mOVA was prepared from pBlue-RIP-tfrOVA (kindly provided by Paul A. Gleeson) and inserted into the BamHI site which is 3' of the human keratin 14 promoter in the plasmid K14-HGX as described (34). From the resulting plasmid, a 7kb fragment containing the K14 promoter, mOVA and the modified HGX polyadenylation signal from the human growth hormone gene was excised using restriction enzymes. After purification by agarose gel electrophoresis, the 7kb fragment was used to create Tg C57BL/6 mice by pronuclear injection. The transgenic allele was detected by PCR on genomic DNA prepared from tail biopsies. Seven founder mice were assessed for protein expression, and two founder lines were selected for breeding and further analysis.
Directly conjugated mAbs TCRβ-FITC (H57), MHC ClassII-FITC (M5/114.15.2), CD25-PECy5.5 (PC61), CD62L-APC (MEL-14), CD4-PECy7 (RM4-5), CD8α-PECy7 (53–6.7) and CD44-PECy7 (IM7) (eBiosciences), MHC ClassII-PE (AF6-120.1), CD90.1-PerCP (OX7), CD11c-APC (HL3) and CD4-APCCy7 (GK1.5) (BD Pharmingen), TCRβ-PE (H57) and B220-APC (RA3-6B2) (Biolegend) were used to perform 6-color flow cytometry using a FACSCanto Flow Cytometer (Beckton Dickinson) and data were analyzed using CellQuest software (Beckton Dickinson). All standard stainings were performed in 2.4G2 supernatant to block unspecific binding to Fcγ receptors. ESL staining was performed using recombinant mouse E-selectin-human Fc IgG chimera (R&D systems) in HBSS supplemented with Ca2+, Mg2+ and 10mM HEPES and detected with anti-human IgG Fcγ (Jackson ImmunoResearch Laboratories) as described (35). Each sample was subjected to 2mM EDTA treatment as control for ESL binding specificity. The absolute percentage of ESL+ cells was determined by subtraction of the percentage of ESL+ cells of the EDTA-treated sample. For DC sorting experiment, cells from CLN were first MACS purified according to the manufacturer's protocol using anti-CD11c magnetic microbeads (Miltenyi Biotech), then stained with MHC ClassII-FITC, CD11c-APC and CD8α-PECy7 mAbs and sorted using a FACSAria Flow Cytometer (Beckton Dickinson).
Single cell suspensions from SPL and LN were prepared. OT-I and OT-II T cells (Thy1.1) were MACS purified according to the manufacturer's protocol using anti-CD8 and anti-CD4 magnetic microbeads (Miltenyi Biotech) and were checked for purity (>90%) by flow cytometry before transfer. Prior to CFSE staining, cells were washed twice with cold PBS and incubated at 10×106 cells/ml in PBS with 2μM CFSE (Molecular Probes, Eugene, OR) at 37°C for 10min. Cells were then washed twice with cold DMEM/10%FCS. 5×106 cells/mouse were injected i.v. For blocking experiments, FTY720 (Cayman Chemical, Ann Arbor, MI) (1mg/kg) was injected i.p 4h and 24h after adoptive transfer of CFSE-labeled OT-I and OT-II T cells. For DC depletion experiments, K14-mOVA-CD11cDTR and K14-mOVA-LangDTR mice were injected i.p with 0.25μg/mouse or 1μg/mouse DT (List Biological Laboratories Inc, CA) respectively 24h before adoptive transfer of CFSE-labeled OT-I and OT-II T cells. For BM chimeras experiments, eight weeks after reconstitution of lethally irradiated K14-mOVA mice (2×600rads with 3h interval) with 10×106 WT or β2-m−/− BM cells, CFSE-labeled OT-I T cells were adoptively transferred and mice were analyzed 48h later. For Treg depletion, anti-CD25 (PC61) mAb or isotype control (BioXCell, NH) (200μg/mouse) was injected i.p five and three days before adoptive transfer of OT-I T cells. Treg depletion was checked by flow cytometry using anti-CD25 (3C7, Biolegend) at the time of adoptive transfer and resulted in >85% depletion of CD4+ CD25+ Tregs. This treatment did not affect survival and proliferation of the transferred OT-I T cells.
WT splenocytes were either pulsed with SIINFEKL peptide (1μg/ml) for 1h at 37°C or left untreated. The peptide-pulsed splenocytes were labeled with 10μM CFSE, while the unpulsed splenocytes were labeled with 1μM CFSE. The two populations were mixed at a 1:1 ratio and 10×106 cells were injected i.v into syngeneic WT and K14-mOVA mice that had been adoptively transferred with OT-I T cells three days earlier. K14-mOVA recipient mice that did not receive OT-I T cells were used as control. After 5h, SPL were removed and the disappearance of the peptide-pulsed population was determined by flow cytometry. The percentage of specific lysis was calculated with the following formula: 100 − (((%pulsedsample/%unpulsedsample) / (%pulsedctrl/%unpulsedctrl)) * 100).
The dorsal and ventral ear halves were gently separated. For epidermal cell isolation, ear halves were floated dermal-side down on 0.25% Trypsin/2.21mM EDTA (Cellgro) for 30–45min at 37°C. The epidermis was then peeled off, cut into small pieces, further incubated in IMDM/10%FBS for 30min at 37°C and then filtered through a 70μm nylon filter as described (36). For in vitro culture, cells recovered from epidermal sheets from one ear were split into three wells and incubated with 200'000 T cells for three days in RPMI supplemented with 10% FBS (Sigma), 10mM HEPES, 10mM Penicillin Streptomycin and 50μM 2-β-mercaptoethanol (GIBCO). For isolation of whole ear skin cells, ear halves of five mice were pooled, cut into small pieces and incubated in HBSS/10%FBS with 5mM EDTA for 4h at 4°C. Cells recovered were then filtered through a 70μm nylon filter.
Recipient WT mice were anesthesized with 1.25% Avertin (Sigma) (250mg/kg) and shaved on the flank, after which a 1 cm2 piece of skin was removed. Back skin from shaved WT and K14-mOVA donor mice (sex-matched) was removed, cut into 1cm2 pieces and clipped to recipient skin using disposable skin staplers (3M, St-Paul, MN). A solution of 0.9% Sodium Chloride (Hospira Inc, Lake Forest, IL) (200μl/mouse) containing Banamine (Shering-Plough Animal Health) (1/250) was administered i.p twice after surgery and mice were monitored daily for five days. Ten days after engraftment, CFSE-labeled OT-I T cells were adoptively transferred into grafted mice and analyzed 48h later.
Mice were anesthesized with 1.25% Avertin. Each side of the ears was tape stripped 12 times with regular adhesive tape (Staples) and painted with 25μl acetone.
Skin samples were collected and fixed in 10% Formalin. Embedding, sections and H&E staining were performed at the Specialized Histopathology Service Longwood Core (Boston, MA). Representative images are shown for each experimental condition.
All graph data represent mean ± SD. Statistical analysis was performed using the two-tailed unpaired Student's t-test for comparison of two groups.
To study the mechanisms of peripheral tolerance to skin-associated self-antigens, we generated new Tg mice expressing a membrane-bound form of ovalbumin OVA139–386 (mOVA) under transcriptional control of the human keratin 14 (K14) promoter. This promoter drives expression in epithelial cells of the skin, as well as thymus, tongue and esophagus (37). In these K14-mOVA mice, keratinocytes in epidermis express mOVA, which contains epitopes for both MHC ClassI-restricted OVA-specific TCR Tg CD8 T cells (OT-I) and MHC ClassII-restricted OVA-specific TCR Tg CD4 T cells (OT-II) (38, 39), allowing the study of both CD8 and CD4 T cell responses in the same environment. However, it should be noted that OT-I and OT-II transgenic lines are not similar in their reactivity to peptide antigen and response to antigen molarity expressed in vivo.
To study immune responses induced to skin-associated self-antigens, we performed adoptive transfers of purified OT-I and OT-II Tg T cells (Thy1.1) into WT and K14mOVA (Thy1.2) mice. Because OVA is expressed in K14-mOVA skin and skin-derived antigens should be transported by local tissue-specific DCs to CLN (32), we anticipated that T cells would encounter the highest levels of presentation of OVA within the CLN. We first analyzed the accumulation of donor OT-I and OT-II T cells in CLN, MLN and SPL 48h and 72h respectively after transfer. Interestingly, the accumulation of OT-I T cells in K14-mOVA mice was comparable in each of these lymphoid tissues, while OT-II T cells preferentially accumulated in CLN rather than in MLN and SPL (Fig. 1A). We next monitored proliferation of donor T cells in K14-mOVA mice. To this end, we measured dilution of a cell-permeable fluorescent dye, 5,6-carboxyfluorescein diacetate succinimidyl ester (CFSE) by flow cytometry. We found that OT-I T cell proliferation was identical in CLN, MLN and SPL in K14-mOVA mice, while OT-II T cell proliferation was induced mainly in CLN and significantly less in MLN and SPL of K14mOVA mice (Fig. 1B). No proliferation was observed when cells were transferred into WT mice. Thus, in K14-mOVA mice OT-II T cells are preferentially stimulated in CLN, whereas OT-I T cells appear to be stimulated equally in CLN and other secondary lymphoid tissues. OT-I and OT-II T cells displayed activation markers when transferred into K14-mOVA mice, as shown by down-regulation of CD62L and up-regulation of CD44 and CD25 (Fig. 1C and data not shown). In addition, they eventually became functional effector cells. Indeed, three days after transfer into K14-mOVA mice, but not into WT mice, OT-I T cells acquired the ability to specifically kill cells displaying their cognate antigenic peptide, as shown with an in vivo CTL assay (Fig. 1E). Peptide-pulsed target cells transferred into K14-mOVA that did not receive OT-I T cells were unaffected. We observed that OT-I T cells in K14-mOVA mice had proliferated to the same extent in CLN, MLN and SPL, even though only the skin should express the transgene (Fig. 1B). This result prompted us to investigate whether OT-I T cells are preferentially stimulated in CLN after which they rapidly recirculate to other lymphoid organs or whether OT-I T cell priming occurs in other secondary lymphoid tissues besides CLN. To address these questions, we treated K14-mOVA mice with the sphingosine 1-phosphate receptor agonist FTY720 to block the egress of activated T cells from lymphoid organs (40). We predicted that if OVA-specific T cells were preferentially primed in the CLN, FTY720 treatment should trap stimulated T cells in CLN and prevent their migration to MLN and SPL. FTY720 treatment did not alter the profiles of proliferation of OT-I T cells in K14mOVA CLN, MLN and SPL (Fig. 1D), indicating that OT-I T cell priming is not restricted to CLN, but can occur in other secondary lymphoid organs as well. OT-II T cell priming occurred mainly in CLN in K14-mOVA mice, and treatment with FTY720 induced stronger proliferation in CLN but not in MLN and SPL (Fig. 1D), suggesting that priming of OT-II T cells only occurs in CLN. Taken together, these data indicate that epidermal self-antigens can reach CLN and be presented to T cells, but that the requirements for antigen presentation differ between CD8 and CD4 T cells. On the one hand, CD8 T cells are primed, proliferate and become functional in various lymphoid organs of K14-mOVA mice. On the other hand, CD4 T cells are primed only in CLN where the skin-migrating DCs arrive and present processed epidermal antigens. Since several subsets of DCs are present in the various lymphoid tissues, these data suggest that complementary DC subsets are involved in antigen presentation in K14-mOVA mice. In addition, it is possible that priming of OT-I and OTII T cells differs in the various lymphoid tissues tested because they respond differently to the same antigen dose. Indeed it is known that less peptide/MHC complexes are required to stimulate CD8 than CD4 T cells.
Elucidation of the APCs involved in tissue-specific T cell priming should yield knowledge about peripheral tolerance mechanisms. To assess antigen presentation in our K14-mOVA model, we compared the abilities of various subsets of DCs to present self-antigens in vitro. First, we prepared epithelial cell suspensions from ear epidermis or intestine from WT and K14-mOVA mice and cultured them with OT-I T cells. As shown in Fig. 2A, only skin epithelial cells from K14-mOVA mice, but not WT mice, induced OT-I T cell proliferation, as measured by H-thymidine incorporation and CFSE dilution. No proliferation was observed with epithelial cells from the intestine. These skin epidermal cell suspensions contain both keratinocytes, which express OVA and Langerhans cells (LCs), which presumably do not. As the preponderant population of epidermal APCs, it is probable that LCs retrieve antigens from keratinocytes, and upon maturation in culture present them to OT-I T cells. Second, we analyzed the ability of APCs extracted from various lymphoid organs to present epidermal self-antigens. Peripheral antigens from skin can be transported to CLN by migratory skin-resident DCs, such as LCs or dermal DCs (dDCs), that can directly present epidermal antigen-derived peptides to naïve T cells in CLN (29, 32). Alternatively, skin-derived DCs can transfer antigen to LN-resident DCs, which are particularly efficiently at presenting antigens (41). In addition, it is known that small amounts of antigen can traffic to CLN through lymphatics in a non cell-associated form and be acquired and presented by CLN-resident DCs. To characterize the APCs that present epidermal antigens in lymphoid organs of K14-mOVA mice, we cultured purified CD11c DCs from CLN, MLN and SPL from WT and K14-mOVA mice with OT-I T cells. Only DCs from CLN of K14-mOVA, and not DCs from MLN or SPL, induced OTI T cell proliferation in vitro (Fig. 2B); as expected, no proliferation was induced by WT DCs. We further investigated the CLN DC subset responsible for antigen presentation by culturing OT-I T cells with DC subsets sorted from CLN of K14-mOVA mice. Both skin-derived DCs (MHCClassIIhigh CD11clow CD8α−) and LN-resident CD8α+ DCs (MHCClassIIlow CD11chigh CD8α+) from K14-mOVA mice stimulated OT-I T cell proliferation, while LN-resident CD8α− DCs (MHCClassIIlow CD11chigh CD8α−) did not (Fig. 2C). Taken together, these data show that epidermal antigens from K14-mOVA mice can be presented by both skin-derived DCs - including LCs and dDCs - and LN-resident CD8α+ DCs, demonstrating that in the steady-state epidermal self-antigens are transported and presented by skin-derived DCs and that transfer can occur between tissue-derived and LN-resident DCs in K14-mOVA mice.
To assess the role of skin-derived DCs in transporting epidermal antigens to CLN in vivo, we transplanted skin from WT and K14-mOVA mice onto the flanks of WT mice. After 10 days of engraftment, CFSE-labeled OT-I T cells (Thy1.1) were transferred into grafted WT(Thy1.2) mice and their accumulation and proliferation were monitored in CLN draining the graft and in CLN from the contralateral side of the graft, as well as in MLN. OT-I T cells preferentially accumulated in CLN draining the K14-mOVA skin graft, and not in CLN from the contralateral side (Fig. 3A). Mice grafted with WT skin served as control to show that increased accumulation of OT-I T cells to the CLN draining the graft was not caused by surgical trauma. In addition, proliferation of OT-I T cells occurred in CLN draining the graft but not in CLN from the contralateral side, as measured by CFSE dilution (Fig. 3A). The observation that OT-I T cells failed to accumulate or proliferate in MLN shows that epidermal self-antigens were not transported from skin to MLN, but only to CLN. Taken together, these data suggest that epidermal antigens are transported by skin-resident DCs - either epidermal LCs or dDCs - to CLN and presented to T cells to induce immune responses in vivo. However, these experiments do not exclude the possibility that a small amount of antigen reaches CLN by trafficking without DC involvement through the lymphatic network. Most importantly, skin-derived antigens are not transported to MLN.
To define the DC subsets that present epidermal antigens in vivo in this system, we reconstituted lethally irradiated K14-mOVA mice with bone marrow (BM) from β2-microglobulin (β2-m−/−) deficient mice (42). In these mice only radioresistant cells, such as LCs and Langerin dermal DCs, are capable of presenting epidermal antigens (43, 44). K14-mOVA mice reconstituted with WT BM were generated as control; in these mice self-antigens can be presented by both radioresistant and radiosensitive APCs. Eight weeks after reconstitution, CFSE-labeled OT-I T cells were transferred into BM chimeras and their proliferation was monitored by flow cytometry after 48h. As shown in Fig. 3B, proliferation was comparable in chimeras reconstituted with β2-m−/− or WT BM, demonstrating that the APCs that present epidermal antigens to OT-I T cells in K14mOVA mice are radioresistant. The major class of professional APCs that are radioresistant are skin-derived DCs, such as LCs and dermal DCs, however certain radioresistant stromal cells can also present MHC ClassI-restricted antigens (45). Langerin+ DCs include skin-resident epidermal LCs and dDCs in skin, as well as skin-derived migratory LCs and dDCs and blood-derived DCs in CLN. They can be depleted by injection of diphteria toxin (DT) in Langerin-DTR/EGFP knock-in mice (LangDTR) (32, 46). To further elucidate the role of skin-derived DCs in presenting epidermal self-antigens in vivo, we performed adoptive transfers of OT-I T cells into K14-mOVA-LangDTR previously treated with DT or left untreated. DT injection resulted in 70% depletion of Langerin DCs in CLN and >90% in epidermis at 24 hours post-injection ((32) and data not shown). Proliferation of OT-I T cells was comparable in Langerin depleted and non-depleted K14-mOVA mice (Fig. 3C). These data indicate that Langerin+ cells are not required to present epidermal self-antigens to CD8 T cells, suggesting that another APC population in CLN beside LCs also mediates CD8 T cell stimulation in response to skin-derived self-antigens in vivo. Therefore, we next investigated whether LN-resident DCs could present epidermal antigens in K14-mOVA mice in vivo. To this end, we took advantage of CD11c-DTR Tg mice in which CD11c DCs can be depleted by DT injection (31). We performed adoptive transfer experiments of OT-I and OT-II T cells into K14-mOVACD11cDTR mice previously treated with DT or left untreated. Injection of DT into these mice resulted in 70% depletion of CD11c+ DCs ((31) and data not shown); in addition, since LCs express low levels of CD11c, they were largely unaffected by treatment with DT (Supplementary Fig. 1). As shown in Fig. 3D, absence of CD11c+ DCs had no significant effect on OT-I T cell proliferation, however it significantly impaired proliferation of OTII T cells, indicating that LN-resident DCs play an important role in vivo in CD4 T cell priming in K14-mOVA mice. Taken together, these data revealed that in vivo Langerin+ cells are not the only APC population capable of presenting epidermal antigens to CD8 T cells, suggesting that LN-resident DCs might also play a role in that process; the absence of either Langerin+ DCs or CD11c DCs did not impair CD8 T cell stimulation. In contrast, presentation of epidermal antigens to CD4 T cells significantly involves LN-resident DCs, suggesting that the role of LCs in this case is minor.
Several Tg mouse models expressing skin-associated antigens develop skin diseases, GVHD and autoimmunity, which involve the homing of activated T cells to the skin (26,28, 30). We therefore investigated whether OVA-specific T cells transferred into K14mOVA mice were able to home to skin after antigen encounter and activation in CLN. To migrate to skin, T cells are required to display the proper skin homing receptors, such as ESL and PSL. We therefore analyzed expression of ESL on OT-I and OT-II T cells in various lymphoid organs upon transfer into WT and K14-mOVA mice. Up-regulation of ESL on OT-I and OT-II T cells was observed only when transferred into K14-mOVA mice, and not into WT mice, and specifically on T cells primed in CLN and not in MLN or in SPL (Fig. 4A,4B and 4D). EDTA treatment was used in all samples to determine the specificity of ESL staining, since binding of ESL chimeric molecule is Ca dependent. In addition, OT-I T cells in CLN draining K14-mOVA skin transplants, but not in CLN draining the contralateral side, also up-regulated ESL; in contrast, OT-I T cells that were found in MLN of such transplant-recipient WT mice, expressed the gut-associated homing molecule α4β7 (data not shown). This is in accordance with previous reports showing that tissue-specific homing properties on T cells are dictated by to the localization of priming and the origin of the APCs (16, 17). Interestingly, CD8 T cells started the up-regulation of ESL 48h after transfer, while CD4 T cells required 5 days (Fig. 4A and 4D). In addition, the percentage of ESL+ CD8 T cells was greater than the percentage of ESL+ CD4 T cells in K14-mOVA mice (Fig. 5A and and6A).6A). These results are not surprising since CD8 and CD4 T cells present intrinsic differences in terms of proliferation, differentiation and survival. All together, these data show that imprinting of skin tissue homing is dictated in vivo in presence of endogenously expressed epidermal self-antigens. In addition, our data indicate that CD8 T cells up-regulate ESL much faster and more efficiently than CD4 T cells and suggest that homing to skin might be more efficient for CD8 than CD4 T cells. Since expression of skin homing receptors is required to allow homing to skin, we addressed whether up-regulation of ESL on OVA-specific T cells in CLN of K14-mOVA mice is sufficient to mediate T cell homing to skin under steady-state conditions. Five days after transfer, no OT-I T cells – or at least no detectable numbers of OT-I T cells – could be detected in K14-mOVA skin (Fig. 4C). As expected OT-I T cells transferred into WT mice that did not up-regulate ESL did not home to the skin. Similar results were found with OT-II T cell transfers (data not shown). Due to the presence of a non-negligible proportion of OVA-specific CD4+ CD25+ regulatory T cells (Tregs) in K14-mOVA mice, we have examined the possible role for Tregs in the inhibition of effector/memory T cell homing to skin under steady-state conditions. After two injections of anti-CD25 mAb, we found that OT-I T cells did not migrate to skin (Fig. 5D). In addition, histology examination did not reveal any mononuclear cell infiltration or inflammation in these conditions (Fig. 5C). All together, these data indicate that despite significant up-regulation of ESL on T cells, OVA-specific T cells do not migrate to skin under steady-state conditions, resulting in the maintenance of immune tolerance to peripheral self-antigens. However, peripheral tolerance in these mice does not involve Tregs.
To further investigate whether tolerance can be broken to induce T cell homing to skin in K14-mOVA mice, we used tape stripping as a method to induce skin inflammation. Tape stripping induces skin damage by disrupting the cutaneous epithelial barrier, avoiding the administration of exogenous agents and resulting in the production of a specific profile of inflammatory cytokines and adhesion molecules, such as TNFα, IL-8, IFNγ, TGFβ, IL1α, IL-1β, as well as E-selectin and VCAM-1 (47, 48), and suffices to induce migration of LCs from epidermis to CLN (49). Indeed, we hypothesize that in K14-mOVA mice tape stripping would induce activation and migration of skin-derived DCs to CLN under inflammatory conditions. Accordingly, skin-derived DCs would prime T cells in an immunogenic manner eliciting a skin homing phenotype that would induce T cell homing to skin. To test this hypothesis, we adoptively transferred OT-I and OT-II T cells into WT and K14-mOVA mice and either tape stripped the ears or left them untreated. We then analyzed characteristics of skin homing and inflammation. Upon tape stripping, we observed an increase in the total cellularity of auricular LN (ALN); in addition, we found increased numbers of OT-I T cells in ALN in K14-mOVA mice compared to non-treated K14-mOVA mice (Fig. 5A). The effects of tape stripping on ALN cellularity and OT-I T cell accumulation in WT skin were not significant. We further investigated the effect of tape stripping on ESL expression and homing to skin. Tape stripping induced significant up-regulation of ESL expression on OT-I T cells transferred into K14-mOVA, but not on OT-I T cells transferred into WT mice (Fig. 5A). In addition, tape stripping induced the migration of OT-I T cells to skin in K14-mOVA mice, as measured by flow cytometry of cells extracted from ear skin (1.4% versus 3.9% of CD8 T cells were Thy1.1+) (Fig. 5B), indicating that tolerance can be broken under these conditions. Histology examination revealed abundant mononuclear cell infiltration in tape stripped K14-mOVA skin, but not untreated K14-mOVA skin (Fig. 5C), indicating inflammation and T cell recruitment upon treatment of K14-mOVA mice. Tape stripping of WT skin did not induce inflammation. These data suggest that tape stripping induced inflammation mediates antigen-specific T cell homing to skin in K14mOVA mice. Adoptive transfer of OT-II T cells into K14-mOVA mice provoked modest, but significant increases of ESL expression on these cells. In contrast to OT-I cells, tape stripping did not yield further increase in ESL expression on OT-II cells (Fig. 6A). Tape stripping of K14-mOVA skin however resulted in changes in ear thickness, and flow cytometry of cells extracted from ear skin showed inflammation and infiltration into skin (0.6% versus 1.6% of CD4 T cells were Thy1.1+) (Fig. 6B and 6C). This result suggests that CD4 T cell tolerance to skin-derived antigens could also be reversed by skin tape stripping. Taken together, these data indicate that tolerance is established in K14-mOVA mice, thereby inhibiting T cell homing to skin and autoimmunity. When tolerance is broken by tape stripping, migration to skin occurs and results in an inflamed skin phenotype.
We here report for the first time that imprinting of a skin homing phenotype on T cells is dictated in vivo in mice in which the model antigen OVA is endogenously expressed in skin. Indeed, the skin homing molecule ESL is induced on T cells upon transfer into K14mOVA mice. Most importantly, we show that ESL up-regulation occurs specifically and solely in lymphoid tissues draining the skin and not in other secondary lymphoid organs in K14-mOVA mice, suggesting that skin tissue imprinting is critically dependent on skin-derived DCs that carry the antigen from skin to CLN. This is in accordance with previous studies in which exogenous antigen was administered via the skin in vivo or that used DCs from skin-and gut-associated tissues ex vivo (16, 18, 19). However, the molecular mechanisms that mediate imprinting of leukocyte tissue-specific homing are not fully understood but may involve vitamin metabolites through their ability to induce or suppress expression of tissue-associated homing molecules, such as retinoic acid or 1,25(OH)2D3, the active form of vitamin D3 (50–52). While under steady-state conditions the CLN microenvironment in K14-mOVA mice induces up-regulation of ESL expression on T cells, this is not sufficient to induce trafficking of T cells to the skin. Although we have identified the presence of transferred OVA-specific T cells in all lymphoid tissues of K14-mOVA mice, we could hardly detect any of these T cells in the skin of K14-mOVA mice under steady-state conditions. Since dissemination of transferred T cells to non-lymphoid tissues is universal and irrespective of a specific antigen tissue distribution (53), we believe that low levels of OVA-specific T cells do enter the skin of K14-mOVA mice in the absence of inflammation but that the numbers are below the limits of detection with the methods we have used in this study. The lack of extensive antigen-driven T cell homing to skin in K14-mOVA mice contradicts other studies that used comparable mouse models expressing self-antigens in skin – K5-mOVA, K14-mOVA or K14-OVAp mice – in which transfer of OT-I T cells induces inflammatory skin disease due to effector T cell infiltration of the skin (26–28). In our new line of K14-mOVA mice, tolerance predominates: transferred OT-I T cells do not induce skin disease unless tape stripping is applied. There are several features that might account for the discrepancy between our model and those utilized in the previous studies. The constructs used to generate mOVA fusion proteins for Tg mice differ in composition resulting in structurally different expressed antigens; indeed, the existing K14-mOVA mice express OVA fused to PDGFR (27), while the K14-mOVA mice presented here express OVA fused to TfR. The level of transgene expression probably also differs between strains, since the different transgenic lines employ distinct promoters (K5 versus K14) and are likely affected by their loci of integration. In addition, K14OVAp mice express only the peptide recognized by OT-I T cells and not the cognate peptide for OT-II T cells. Therefore a plausible explanation for why K14-OVAp mice develop severe inflammatory skin disease upon OT-I T cell transfer could be that immune suppression mediated by OVA-specific CD4+ CD25+ Tregs is lacking. In our K14-mOVA mice however Tregs seem unlikely to play a critical role in suppression of OVA-specific T cell homing to skin and in induction of skin inflammation. Our study however revealed a significant antigen-dependent T cell migration to skin when we induced mild inflammation of the skin by tape stripping. We conclude that tape stripping of K14-mOVA skin induces antigen-experienced skin-resident DCs to undergo maturation, migrate and present self-antigens in an immunogenic manner in CLN. In contrast to other models, this additional signal of inflammation is absolutely required in K14-mOVA mice to break the established peripheral tolerance and allow homing of T cells to skin and induction of tissue damage. Leukocyte migration to inflamed tissue has been widely described and is clearly dependent on ESL expression (3, 13). We here demonstrate that low level of inflammation in K14-mOVA mice induces a significant increase in ESL expression on adoptively transferred OT-I T cells, but no such increase is seen on OT-II T cells. The level of endogenous OVA expressed in K14-mOVA skin may be a limitation to induce extensive expression of ESL on stimulated CD4 T cells, explaining also their lower efficiency to migrate to skin compared to CD8 T cells. We believe that restrained expression of homing receptors accounts for the maintenance of peripheral tolerance in K14-mOVA mice. A remaining issue regarding the K14-mOVA mice presented in this study is to identify the cells involved in antigen presentation. The skin graft experiments allowed us to show that skin-expressed antigens are transported from epidermis to CLN by skin-derived DCs in K14-mOVA mice. Although we cannot rule out that the antigen traffics to some extent to CLN by the lymphatics and conduits network in a non cell-associated form; however this scenario is considered less likely since tissue-derived DCs are critically important in sustaining IL-2 receptor expression on T cells (54). Using mice that allow specific depletion of DC subsets in vivo, we also show that both skin-derived DCs and LN-resident CD8α+ DCs present epidermal antigens to T cells – this was also supported by our in vitro experiments –, however depletion does not alter T cell activation, suggesting that neither of these populations is absolutely required in this model. Our BM chimera experiments furthermore confirm that both radio-resistant and radio-sensitive populations contribute to antigen presentation in K14-mOVA mice. Taken together, we conclude that skin-derived DCs and LN-resident DCs play a pivotal role in the maintenance of peripheral tolerance to skin-expressed self-antigens in K14-mOVA mice, through restrained induction of tissue homing receptor expression on antigen-specific effector/memory T cells to skin. Our study does not rule out the possibility that another non-hematopoietic population additionally contributes to the induction of tolerance. Indeed, the contribution of the stroma in secondary lymphoid tissues is a new previously unappreciated population involved in tolerance to self-antigens (45, 55). In contrast to other mouse models expressing epidermal self-antigens in the skin in which OVA-specific T cell proliferation occurs only in CLN, we observed proliferation of OT-I T cells in all lymphoid organs of K14-mOVA mice. Although the K14 promoter should drive expression of the transgene mainly in skin and thymus and not in other lymphoid tissues, we cannot exclude the possibility that self-antigens are presented by stromal cells in K14-mOVA mice, resulting in the indistinct proliferation of OT-I T cells in all lymphoid tissues. However, a role for imprinting tissue-homing receptor expression on T cells by stromal cells has not been described yet. Interestingly, in contrast to OT-I T cells, OT-II T cells are mainly primed in CLN and to a much lesser extent in MLN and SPL; there may be several explanations for the difference of priming of CD4 and CD8 T cells in K14-mOVA mice. First, due to T cell intrinsic differences, the frequency of displayed antigenic peptide/MHC complexes in lymphoid tissues other than CLN might be too low for priming CD4 T cells, while it is sufficient to induce CD8 T cell stimulation. Second, stromal cells have been implicated in the activation and subsequent deletion of CD8 T cells to promote tolerance, but no such information is known to date about CD4 T cells. Third, the tissue distribution of the DC subsets presenting the antigen may result in a difference in priming location as well. Overall, our study supports the idea that induction of tolerance and immunity can be driven by the same DC subsets but depends on the context where and how the antigen is presented. Taken together, we conclude that skin-derived DCs and LN-resident DCs are both involved in the maintenance of peripheral tolerance to a skin specific antigen in K14-mOVA mice. Our data obtained in K14-mOVA mice shows that DCs control the levels of ESL on antigen-specific effector/memory T cells, and thereby restrict their homing to skin.
We are particularly grateful to Dr. B. Malissen for the gift of LangDTR/EGFP mice. We thank Dr. R. Krzysiek, Dr. L Liu and Dr. M. Bai for technical advice and support, Dr. P. Stoitzner for expert advice on LC isolation and all the members of the Boes laboratory for helpful discussions.
This research was supported by The Swiss National Science Foundation (PA00A3-119493, TB); the Howard Hughes Medical Institute (HHMI) research training fellowship for medical students (LP); Crohn's and Colitis Foundation of America (MAW); the Harvard Skin Disease Research Center (TK); and NWO-Veni grant and RO1-AR052810 (MB).
Publisher's Disclaimer: “This is an author-produced version of a manuscript accepted for publication in The Journal of Immunology (The JI). The American Association of Immunologists, Inc. (AAI), publisher of The JI, holds the copyright to this manuscript. This version of the manuscript has not yet been copyedited or subjected to editorial proofreading by The JI; hence, it may differ from the final version published in The JI (online and in print). AAI (The JI) is not liable for errors or omissions in this author-produced version of the manuscript or in any version derived from it by the U.S. National Institutes of Health or any other third party. The final, citable version of record can be found at www.jimmunol.org.”