NIK signaling in dendritic cells but not in T cells is required for the development of effector T cells and cell-mediated immune responses

We have recently reported that NIK^aly/aly mice are resistant to the induction of EAE because of the loss of function of NIK within the hematopoietic compartment and more specifically because of a defect in T cell priming, independent of the lack of SLTs (Greter et al., 2009). Furthermore, NIK^aly/aly T cells have been reported to be defective in proliferation and secretion of IL-17, IL-2, and GM-CSF (Matsumoto et al., 2002; Sánchez-Valdepeñas et al., 2006; Jin et al., 2009). In line with these previous studies, we observed that in vitro, polyclonally activated NIK^aly/aly CD4⁺ T cells produced lower amounts of effector cytokines (IL-2, IFN-γ, and IL-17) than heterozygous NIK^aly/+ controls, whereas the production of IL-4 was unaffected (Fig. 1 A). This suggests a T cell intrinsic impairment of polarization conferred by the ablation of NIK signaling. To further investigate the requirement of NIK for antigen-specific T cell activation, we crossed NIK^aly/aly mice with TCR transgenic 2d2 mice, in which the TCR recognizes the immunodominant epitope of the myelin oligodendrocyte glycoprotein (MOG_35–55). As expected, NIK^aly/aly-2d2 T cells also failed to secrete effector cytokines upon encountering their cognate antigen (Fig. 1 B). To exclude the possibility that the observed defects were caused by the developmental malformations of SLTs in NIK^aly/aly mice, we generated BM chimeric mice (BMCs) in which WT mice were reconstituted with hematopoietic stem cells from either NIK^aly/aly-2d2 or NIK^aly/+-2d2 mice. We found that even if most of the CD4⁺ T cells carry the cognate antigen-specific TCR, NIK^aly/aly-2d2 → WT BMCs retained their EAE resistance upon MOG_35–55/CFA immunization (Fig. 1 C), emphasizing the critical role of NIK signaling for the development of T cell–mediated autoimmune responses.

Because T cells from NIK^aly/aly-2d2 → WT BMCs failed to acquire pathogenic properties, we addressed their behavior in antigen-independent homeostatic expansion in lymphopenic Rag1^−/− mice. After adoptive transfer of CD4⁺ T cells from NIK^aly/aly-2d2 → WT BMCs into Rag1^−/− mice, we observed a reduction in homeostatic expansion when compared with CD4⁺ T cells from NIK^aly/+-2d2 → WT BMCs (Fig. 1 D). Upon immunization of those mice with MOG_35–55/CFA, NIK^aly/aly-2d2 T cells further failed to respond to their cognate antigen, whereas control NIK^aly/+-2d2 T cells strongly expanded (Fig. 1 E). In addition, Rag1^−/− mice reconstituted with T cells from NIK^aly/aly-2d2 → WT BMCs remained completely resistant to EAE, suggesting that NIK^aly/aly-2d2 T cells cannot be primed by NIK-sufficient accessory cells (Fig. 1 F). Collectively, the data support the notion that NIK deficiency indeed leads to a T cell–intrinsic lesion.

Our data and a previous study (Jin et al., 2009) support a T cell–intrinsic role of NIK in cell-mediated immunity. However, we further aimed to identify the role of NIK in the accessory cell compartment. BMCs were generated by transferring a 4:1 mixture of Rag1^−/− and NIK^aly/aly-2d2 BM into Rag1^−/− mice. In those mice, NIK^aly/aly-2d2 T cell progenitors are developing within an NIK-sufficient accessory cell environment. Surprisingly, Rag1^−/− + NIK^aly/aly-2d2 → Rag1^−/− BMCs developed EAE comparable with NIK^aly/+-2d2 → Rag1^−/− BMCs (Fig. 2 A). This finding demonstrates that NIK plays a vital role in hematopoietic accessory cells rather than in T cells to develop autoimmune inflammation. Complementing this result, polyclonally in vitro activated CD4⁺ T cells from Rag1^−/− + NIK^aly/aly → WT BMCs were rescued in their ability to secrete IL-17 and IFN-γ (Fig. 2 B), suggesting that the production of effector cytokines by T cells requires intact NIK signaling in the accessory cell compartment but not the T cell compartment.

The most prominent accessory cells involved in the induction of cell-mediated immunity and antigen presentation are DCs. It has been reported that in vitro NIK^aly/aly DCs have defects in the expression of CD80, CD86, and MHCII as well as in antigen presentation and their ability to drive T cell expansion (Garceau et al., 2000; Tamura et al., 2006; Lind et al., 2008). Also the ability of T cells to secrete effector cytokines is largely dependent on the capacity of APCs to provide T cell instructive cytokines. In particular, the cytokines IL-12, IL-23, and IL-6 have a major impact on the polarization of effector T cells. We investigated the ability of splenic NIK^aly/aly DCs to secrete these factors after stimulation with anti-CD40, thereby mimicking T cell–APC interactions. Interestingly, activated NIK^aly/aly DCs secreted significantly lower levels of the proinflammatory cytokine subunit IL-12/IL-23p40 (Fig. 3, A and B; and Fig. S1, A and B). We could further observe a strong reduction of the transcripts for IL-12p35 and IL-23p19 (Fig. S1 C) and protein levels of IL-6 (Fig. 3 C). These data suggest that NIK signaling in DCs upon DC–T cell interaction via CD40 is critical for the capacity of DCs to secrete T cell-instructive cytokines.

To verify the reduced priming capacity of NIK^aly/aly DCs in vivo, we established a model based on diphtheria toxin (DTx)–mediated cell ablation. We created mice with an immune compartment containing both NIK-deficient and -sufficient T cells, whereas the vast majority of DCs carry the mutated NIK^aly/aly protein. To do so, a 1:1 mixture of CD11cDTR and NIK^aly/aly BM was transferred into irradiated WT recipients (Fig. 4 A). Upon injection of DTx, DCs of CD11cDTR origin (NIK^+/+) were depleted, whereas mutant NIK^aly/aly DCs were retained. The efficiency of DC ablation in CD11cDTR mice was >90% (Fig. 4 B). We observed a significant delay in disease onset (Fig. 4 C) with decreased central nervous system (CNS) infiltration of T cells (Fig. 4 D), and particularly IL-17–producing T cells (Fig. 4 E), in DTx-treated CD11cDTR + NIK^aly/aly → WT BMCs compared with DTx-treated control CD11cDTR + NIK^aly/+ → WT BMCs. This finding strongly supports our previous in vitro (Fig. 3) and in vivo (Fig. 2 A) data and demonstrates that NIK signaling in DCs is critical for their ability to instruct T cell polarization and effector function.

To ascertain a primary role of NIK signaling in the DC compartment, we generated mice in which NIK expression is restricted to DCs. R26Stop^FLNIK^WT mice express NIK^WT preceded by a loxP-flanked neo^R-Stop cassette and followed by an Frt-flanked IRES-eGFP within the ubiquitously active ROSA26 locus (Sasaki et al., 2008). Upon crossing to CD11c-cre mice, the neo^R-Stop is excised, and NIK^WT will be expressed in CD11c⁺ cells, which are mainly DCs (these mice are hereafter called DC^NIK). Upon further breeding those animals onto the NIK^−/− background, we generated mice that express NIK^WT in DCs, whereas other cells and tissues lack the ability to express NIK. To verify the cell type–specific targeting of the NIK^WT transgene, we analyzed GFP expression in different immune compartments, which confirmed the transgene expression in DCs (Fig. S2 A).

To manipulate the expression of NIK only within the hematopoietic compartment and to provide a normal lymphoid organ structure, we again generated BMCs by transferring BM of either DC^NIK-NIK^−/− or control DC^NIK-NIK^+/− and NIK^−/− mice into lethally irradiated WT mice. 6 wk after reconstitution, these BMCs were immunized with MOG_35–55/CFA and observed for the development of clinical disease. Strikingly, DC^NIK-NIK^−/− → WT BMCs were fully susceptible to EAE, even though most T cells are NIK^−/− (Fig. 5 A). Furthermore, the secretion of proinflammatory cytokines IL-12/IL-23p40 and IL-6 in DCs of DC^NIK-NIK^−/− → WT BMCs was largely restored (Fig. 5 B). Also, the restoration of NIK^WT in DCs rescued the secretion of T effector cytokines IL-17, IFN-γ, and GM-CSF (Fig. 5 C) after antigen restimulation. These findings demonstrate that NIK signaling in DCs is sufficient to generate autoaggressive CD4⁺ T cells, even if these T cells themselves do not express functional NIK.

In addition to verifying the transgene expression in DCs, we thoroughly analyzed other leukocyte populations for inadvertent ectopic expression. Macrophages were found to be negative, and although microglia can express transient levels of CD11c upon activation, they can be excluded as effectors in this model because NIK^−/− → WT BMCs have NIK-sufficient microglia but remained EAE resistant (Fig. 5 A). In DC^NIK-NIK^−/− → WT BMCs, we detected a minor percentage of GFP⁺CD4⁺ and CD8⁺ T cells (Fig. S2 A). To ensure that this population of NIK-expressing T cells does not contribute to the EAE susceptibility of DC^NIK-NIK^−/− → WT BMCs, we further bred T^NIK-NIK^−/− mice by crossing R26Stop^FLNIK^WT mice to CD4-cre and NIK^−/− mice. T^NIK-NIK^−/− → WT BMCs were, similar to NIK^−/− → WT BMCs, completely EAE resistant (Fig. S2 B). Thus, the reintroduction of NIK^WT into the DC pool fully restored EAE susceptibility.

Our data thus far clearly showed that NIK signaling in DCs is critical for T effector function. However, if the restoration of NIK signaling in DCs alone reinstates T effector function, the fact that adoptively transferred mature NIK^aly/aly T cells into NIK-sufficient recipients fail to acquire effector function represents a contradiction (Fig. 1). Thus, we hypothesized that NIK signaling in thymic DCs is required to enable developing thymocytes to exit the thymus as fully functional T cells. Therefore, the thymic DC compartment of NIK^aly/aly and NIK^aly/+ mice was analyzed in detail. All three thymic DC subsets (Wu and Shortman, 2005; Proietto et al., 2008a; Li et al., 2009), namely migratory and resident conventional DCs (cDCs; CD11c⁺CD172α⁺ and CD172α⁻, respectively) and plasmacytoid DCs (pDCs; CD11c^intCD45RA⁺), were found in comparable numbers in NIK^aly/aly and NIK^aly/+ mice (Fig. 6 A). However, both cDCs and pDCs in NIK^aly/aly mice expressed reduced levels of CD80 and CD86 and even more drastically MHCII (Fig. 6 B), with the strongest reduction in the resident DC subset. This indicates some degree of functional impairment of APC properties, which we further assessed by co-culturing NIK^aly/aly and NIK^aly/+ thymic DCs with 2d2 CD4⁺ single-positive (SP) thymocytes in the presence of MOG_35–55. NIK^aly/aly thymic resident DCs elicited reduced proliferation in 2d2 CD4⁺ SP thymocytes (Fig. 6 C).

To further investigate the phenotypic properties of NIK^aly/aly thymic DCs, we performed quantitative RT-PCR (qRT-PCR) analysis for various chemokine ligands and receptors involved in thymic DC function (Proietto et al., 2008a). NIK^aly/aly thymic DCs revealed a strong reduction in the expression of CCL17, CCL19, and CCL21 (Fig. 6 D), which are crucial for the migration of developing thymocytes (Ueno et al., 2004; Proietto et al., 2008a). Furthermore, the analysis of chemokine receptors revealed an overall reduction in the levels of CCR2, CCR5, CCR6, and CCR7 but increased expression of CCR9 and TLR9 (Fig. S3). Collectively, we found that NIK^aly/aly thymic DCs are phenotypically and functionally distinct from those in NIK^aly/+ mice, suggesting a causative link between the altered function of thymic DCs and the subsequent loss of T effector function.

Thymic DCs have primarily been implicated in mediating negative selection (Gallegos and Bevan, 2004). However, there is increasing evidence that thymocyte development not only selects the T cell receptor repertoire but also imprints effector function onto thymic emigrants. In particular, NIK has been suggested to be involved in the expansion of CD25⁺CD4⁺ T cells (Lu et al., 2005; Tamura et al., 2006). We found a 50% reduction in FoxP3⁺ T_reg cells in both developing and mature T cells (Fig. 7, A and B). Given the reduced effector cytokine expression of NIK^aly/aly T cells, we speculated whether the observed decrease of natural occurring nT_reg cells in NIK^aly/aly thymi is the result of altered licensing of T cells during their development. Recently, thymic T cell lineage commitment has been expanded to other lineages including T_H17 cells (Marks et al., 2009). Therefore, we analyzed the gene expression of lineage-specific transcription factors in CD4⁺ SP cells of NIK^aly/aly thymi (Fig. 7 C) and spleens (Fig. 7 D) and found that in addition to Foxp3, the expression of RORγt and Tbet was decreased, whereas GATA3 was not affected. We further found that the percentage of IL-17– and IFN-γ–producing CD4⁺ SP thymocytes was strongly reduced in naive NIK^aly/aly mice (Fig. 7 E). Interestingly, a very recent study described that RelB, which is one of the target molecules of NIK, is essential for LTβR-dependent thymic commitment of γδ T cells toward IL-17 production (Powolny-Budnicka et al., 2011). In line with this, our observations suggest that thymic commitment toward the T_H1 or T_H17 lineage might also occur for αβ T cells and that NIK signaling in thymic DCs is crucial for this process.

To verify that DC-restricted expression of NIK^WT is capable of restoring early lineage commitment, we again generated DC^NIK-NIK^−/− → WT, DC^NIK-NIK^+/− → WT, and NIK^−/− → WT BMCs and analyzed the expression of transcription factors in CD4⁺ SP thymocytes. As shown in Fig. 8 (A and B), DC-restricted NIK^WT expression fully restored RORγt, Tbet, and also Foxp3 expression in CD4⁺ SP thymocytes.

NIK is widely held as a central mediator of noncanonical NF-κB signaling and the activation of NF-κB2. Indeed, both NF-κB2^−/− and NIK^aly/aly mice show impaired T and B cell responses, while displaying lymphocyte infiltration into various organs similar to that of Aire^−/− mice (Anderson et al., 2002; Liston et al., 2003; Kajiura et al., 2004; Zhu et al., 2006). However, the autoimmune phenotype in NIK^aly/aly and NF-κB2^−/− mice seems to originate from the stromal compartment, as transplantation of NIK^aly/aly or NF-κB2^−/− thymi into WT mice was sufficient to induce the breakdown in self-tolerance, which is mediated by the loss of Aire function in mTECs (Kajiura et al., 2004; Zhu et al., 2006). In contrast, the impairment of T cell responses in NIK^aly/aly mice resulted from disrupted NIK signaling in hematopoietic cells (Greter et al., 2009). Also, NIK^aly/aly mice have a defect in the generation of T_reg cells, which is not observed in NF-κB2^−/− mice (Zhu et al., 2006). These and other observations (Ramakrishnan et al., 2004; Speirs et al., 2004; Zarnegar et al., 2008; Sasaki et al., 2011) suggest that the signaling cascade by which NIK executes its function in cell-mediated immunity may not be exclusive to the p52-dependent noncanonical NF-κB pathway.

As it was widely believed that NIK signaling is predominately involved in the formation of SLTs, the apparent immunodeficiency of the NIK^aly/aly strain was held as evidence for the requirement of SLTs in the formation of cell-mediated immunity (Hofmann et al., 2010). In recent years, it has however become evident that NIK signaling is also applied by other cell types such as B cells, osteoclasts, cancer cells, and also by DCs and T cells, suggesting a role of noncanonical NF-κB signaling in adaptive immune responses (Matsumoto et al., 2002; Ishimaru et al., 2006; Tamura et al., 2006; Lind et al., 2008; Greter et al., 2009; Jin et al., 2009). We have recently shown that the inability of NIK^aly/aly mice to mount cell-mediated immunity is independent of the developmental lymphoreticular malformations but a result of the interrupted NIK signaling in hematopoietic cells (Greter et al., 2009). Yet the mechanistic consequences of lesioned NIK signaling in DCs and T cells remained poorly understood or might have been wrongly interpreted.

Recently, it was reported that NIK^−/− T cells adoptively transferred into Rag2^−/− mice failed to develop encephalitogenic properties and to secrete proinflammatory cytokines (Jin et al., 2009). The authors concluded that NIK^−/− T cells are intrinsically defective and that NIK signaling in T cells is vital for the acquisition of an effector phenotype. This assumption is corroborated by a previous study showing that NIK^aly/aly T cells secrete reduced levels of IL-2 and GM-CSF (Matsumoto et al., 2002). Indeed, we confirmed a reduction in the secretion of proinflammatory cytokines by NIK^aly/aly T cells and their failure to homeostatically expand. In addition we found that NIK^aly/aly T cells were anergic toward their cognate antigen and thus failed to acquire pathogenic properties in the context of autoimmune disease. However, our observation that T cell function was impaired as a result of the loss of NIK in hematopoietic accessory cells and more specifically in DCs challenges the concept of a T cell–intrinsic defect in NIK^aly/aly mice. Three independent experimental setups demonstrated that the T cell defects were the result of a DC-intrinsic utilization of NIK: first, the presence of NIK^aly/aly DCs together with WT T cells in vivo was sufficient to significantly diminish EAE development (Fig. 4); second, NIK^−/− T cells could differentiate into fully functional, autoaggressive T cells when NIK was restored in accessory cells only (Fig. 2); and third, when the expression of NIK was transgenically restricted to DCs via the CD11c promoter, cell-mediated immunity was fully restored even if T cells were NIK^−/− (Fig. 5). One caveat is that CD11c-Cre was also active in a small population of T cells. This small population of transgenic NIK-expressing T cells could potentially be involved in the rescue of immune function in DC^NIK-NIK^−/− → WT BMCs. However, this is most unlikely because (a) we did not observe a preferential accumulation of NIK-expressing GFP⁺ T cells in the inflamed CNS at peak disease in DC^NIK-NIK^−/− → WT BMCs (not depicted), (b) T cell function of mixed Rag1^−/− + NIK^aly/aly → Rag1^−/− BMCs was fully restored even though the entire T cell compartment was NIK deficient, and (c) the transgenic expression of NIK^WT in T cells of T^NIK-NIK^−/−-2d2 → WT BMCs did not render mice EAE susceptible (Fig. S2 B). However, preliminary data suggested that ectopic expression of NIK^WT in T^NIK mice might affect additional aspects of T cell function.

Interestingly, adoptively transferred adult NIK^aly/aly T cells failed to acquire pathogenicity but rather displayed an anergy-like state, although they were primed by NIK-sufficient accessory cells (Fig. 1). Only upon undergoing thymic development in the presence of NIK-sufficient accessory cells could CD4⁺ T cells give rise to pathogenic effector cells. This suggests that against current belief, NIK is largely dispensable within T cells. In contrast, we suggest that the anergic T cell phenotype observed in NIK^aly/aly and NIK^−/− mice is caused by defective T cell development caused by dysfunctional thymic DCs. We therefore propose that NIK signaling is critical in DCs to license T cells during thymic development and avoid anergy.

A role for noncanonical NF-κB signaling in DCs has already been suggested, but the precise impact on T cell function has not been addressed before. Although it was claimed that CD40-mediated activation of DCs is not dependent on NIK (Garceau et al., 2000; Yamada et al., 2000; Andreakos et al., 2003), others have demonstrated that peripheral NIK^aly/aly DCs show reduced APC capacity, which results in diminished T cell proliferation (Tamura et al., 2006). Furthermore, NIK^aly/aly DCs showed a decreased ability to induce the expansion of CD25⁺ CD4⁺ T cells in vitro (Tamura et al., 2006) and were unable to cross-prime CD8⁺ T cells to exogenous antigen, involving multiple defects in antigen-processing pathways (Lind et al., 2008).

We demonstrate that splenic NIK^aly/aly DCs produced lower levels of IL-12, IL-23, and IL-6. Furthermore, we show that NIK^aly/aly DCs were hampered in the priming of CD4⁺ autoreactive T cells in vivo (Fig. 4). However, the relevance of NIK signaling in thymic DCs has until now not been addressed, most likely because of the incomplete understanding of the function of thymic DCs in general.

Currently, three phenotypically distinct thymic DC subsets have been described, namely pDCs, CD172α⁺CD11b⁺CD8α^−/lo migratory cDCs, and CD172α⁻CD11b⁻CD8α^hi thymic resident cDCs (Wu and Shortman, 2005). One important function of thymic DCs appears to be negative selection at the CD4⁺ SP stage in T cell development (Brocker et al., 1997; Dakic et al., 2004; Gallegos and Bevan, 2004; Bonasio et al., 2006). In this context, death caused by high-affinity interaction or by no interaction (death by neglect) is not the only fate of developing thymocytes, but also the generation of nT_reg cells or induction of anergic T cells (Ramsdell et al., 1989; Ramsdell and Fowlkes, 1990; Bendelac, 2004). The mechanistic underpinnings of these processes are up to today poorly understood. Although the effect of the complete lack of DCs on T cell development and negative selection is discussed controversially (Birnberg et al., 2008; Ohnmacht et al., 2009), a diverted function of thymic DCs is expected to have consequences on T cell development.

The expression of various chemokine receptors and chemokines in thymic DC subsets has been analyzed and compared with expression profiles of splenic DCs (Proietto et al., 2008a). We also profiled thymic DCs of NIK^aly/aly mice and found strongly reduced gene expression of CCR2, CCR5, CCR6, and CCR7. It has been proposed that these chemokine receptors are important for the localization and migration of DCs into lymphoid tissues in general including the thymus (Bonasio et al., 2006). We further observed a strong reduction in the expression of the chemokines CCL17, CCL19, and CCL21 in thymic DCs of NIK^aly/aly mice, which are important for intrathymic migration of positively selected thymocytes from the cortex to the medulla (Ueno et al., 2004; Proietto et al., 2008a).

Thymic DCs have been proposed to induce FoxP3 expression and the formation of nT_reg cells (Proietto et al., 2008b). Recently, it became evident that also other αβ T effector cell lineages such as T_H17 cells can at least partially be licensed already during thymic development (Marks et al., 2009). Of note, for γδ T cells, a similar commitment toward IL-17 or IFN-γ production has been shown previously (Ribot et al., 2009), and a very recent study suggested both RelA and RelB to be the critical factors for this process (Powolny-Budnicka et al., 2011). We found that thymic licensing of αβ CD4⁺ T effector lineages at least to a certain extent relies on the function of NIK and that the expression of NIK^WT in DCs alone could rescue not only FoxP3 but also Tbet and RORγt expression in developing NIK-deficient thymocytes. Therefore, we propose that NIK signaling in thymic DCs is crucial to imprint developing T cells to subsequently acquire full effector capabilities and to avoid progression into an anergic state.

C57BL/6 (WT) and Rag1^−/− mice were purchased from the Jackson Laboratory and bred in house. Alymphoplasia mice (Map3k14^aly mice here depicted as NIK^aly/aly) were obtained from Clea Laboratories and bred in house. NIK^−/− mice on 129Sv/Ev background were provided by R.D. Schreiber (Washington University in St. Louis, St. Louis, MO) and bred onto C57BL/6 background in house for 10 generations. Both NIK^aly/aly and NIK^−/− mice were maintained by heterozygous breedings. NIK^aly/+ and NIK^+/− mice are haplo-sufficient (Miyawaki et al., 1994; Yanagawa and Onoé, 2006), justifying the use of heterozygous animals as littermate controls. Furthermore, NIK^aly/aly and NIK^−/− mice are identical in several aspects of their structural and functional impairments (Yin et al., 2001; Greter et al., 2009; Jin et al., 2009). In all breedings with CD4-cre and CD11c-cre as well as R26NIK^WT mice, which are pure C57BL/6, we used NIK^−/− animals to avoid the usage of mixed genetic backgrounds. Nonetheless, to ensure consistency between the different strains used, we analyzed NIK^aly/aly and NIK^−/− mice and heterozygous controls as well as NIK^+/+ mice for expression of effector cytokines and the proportion of nT_reg cells (Fig. S4). 2d2 (MOG-TCR transgenic) mice were provided by V. Kuchroo (Harvard Medical School, Boston, MA). CD11cDTR mice were provided by S. Jung (Weizmann Institute of Science, Rehovot, Israel). CD11c-cre mice were provided by B. Reizis (Columbia University, New York, NY). R26Stop^FLNIK^WT and all other mice were maintained under specific pathogen-free conditions.

BMCs were generated as described previously (Becher et al., 2002, 2003). In brief, mice were lethally irradiated with a split dose of 1,100 rad. Femur, tibia, and pelvis of donor animals were flushed with PBS to obtain BM stem cells. 10 × 10⁶ cells were injected i.v. per mouse. Mice were treated with 0.2% BORGAL in drinking water for 3 wk to prevent bacterial infections. All experiments involving animals were approved by the Swiss Cantonal Veterinary Office (13/2006, 55/2009; Zurich, Switzerland).

EAE was induced as described previously (Gutcher et al., 2006; Gyülvészi et al., 2009). In brief, mice were immunized s.c. with 200 µg MOG_35–55 (MEVGWYRSPFSRVVHLYRNGK; GenScript) emulsified in CFA (Difco) and two i.p. injections of 200 ng pertussis toxin on days 0 and 2. BMCs did not receive pertussis toxin. For EAE experiments with DTx treatment, mice were injected i.p. with 400 ng DTx (EMD) 1 d before immunization and then 200 ng DTx every second day for the entire length of the experiment.

Spleens were removed under sterile conditions. Each spleen was injected with a cocktail of 1 mg/ml Liberase and 0.5 mg/ml DNaseI (Roche) in medium and incubated at 37°C for 20 min. Single cell suspensions were prepared by homogenizing the tissue between glass slides and filtering through 70-µm cell strainers followed by erythrocyte lysis. Splenic DCs were isolated with CD11c⁺ positive magnetic selection according to the manufacturer’s instructions (Miltenyi Biotech).

DCs were plated at a concentration of 10⁶ cells/ml in RPMI 1640 medium supplemented with 10% FCS, 1% l-glutamine, and 1% penicillin-streptavidin (Invitrogen) and stimulated for 6–24 h at 37°C and 5% CO₂ with 10 µg/ml α-CD40 (FGK 4.5; BioXCell) and 20 ng/ml IFN-γ (PeproTech). Supernatants were analyzed for IL-6 and IL-12/IL-23p40 using ELISA according to the manufacturer’s instructions (BD).

Thymic DCs were isolated as previously described (Wirnsberger et al., 2009). In brief, thymi were digested in IMDM containing 2% FCS, 25 mM HEPES, 0.4 mg/ml Collagenase D (Roche), and DNase for 40 min at 37°C. Afterward, high-density cells were separated from low-density cells by using a discontinuous Percoll density gradient (ρ = 1.115 and ρ = 1.055; GE Healthcare). After removal from the gradient, cells were washed and stained with antibodies against CD11c, CD8, CD172a, and CD45RA, followed by sorting into migratory DCs (CD11c⁺CD172a⁺), resident DCs (CD11c⁺CD172a⁻), and pDCs (CD11c^intCD45RA⁺) on a FACSAria (BD). Purity was routinely >95%.

For in vitro culture, 20,000 DCs were cultured with 100,000 sorted and CFSE-labeled CD4⁺ thymocytes or peripheral T cells in the presence of 10 µg/ml MOG_35–55 and 10 ng/ml IL-7. Analysis was performed after 72 h of culture.

Splenocyte single cell suspensions were prepared as described in Isolation of splenic DCs and in vitro stimulation. CD4⁺ T cells were purified with CD4⁺ negative magnetic selection according to the manufacturer’s instructions (Miltenyi Biotech). The purity was routinely >95% as confirmed by flow cytometry.

For in vitro stimulations of splenic CD4⁺ T cells, 3 × 10⁶ CD4⁺ T cells/ml were cultured in RPMI 1640 medium supplemented with 10% FCS, 1% l-glutamine, and 1% penicillin-streptavidin. Polyclonal CD4⁺ T cell activation was performed with 5 µg/ml plate-bound α-CD3 and α-CD28 for 48 h. For antigen-specific CD4⁺ T cell activation, whole 2d2 splenocytes were stimulated with 20 µg/ml MOG_35–55 and 5 µg/ml soluble α-CD28 for 48 h. Supernatants were harvested, and concentrations of IFN-γ, IL-17, GM-CSF, IL-2, and IL-4 were quantified by ELISA according to the manufacturer’s instructions (BD).

For adoptive transfer experiments, 2 × 10⁶ purified splenic CD4⁺ T cells in PBS were injected i.v. into Rag1^−/− mice. Homeostatic expansion of T cells in Rag1^−/− mice was monitored weekly by tail bleeding and flow cytometry, and the percentage of CD4⁺ T cells of the lymphocyte gate was calculated.

For cell surface staining, we used the following antibodies: CD11c, I^ab, CD80, CD86, CD172a, CD45RA, CD4, CD8, IL-12/IL-23p40, Vα3.2, and Vβ11 (BD). Intracellular FoxP3 staining was performed according to the manufacturer’s instructions (eBioscience). Cells were incubated with antibodies at the optimal concentration for 20 min at 4°C, and cells were analyzed either on FACS Canto II or LSRII Fortessa (BD). Postacquisition analysis was performed with either FACS Diva or FlowJo (Tree Star) software. Cytofluorometric analysis of CNS-invading lymphocytes has been described previously (Gutcher et al., 2006). For intracellular cytokine staining, cells were treated with GolgiPlug (BD) for the last 4 h of culture. T cells were additionally stimulated with PMA and ionomycin for the last 4 h of culture. After surface staining, cells were permeabilized with Cytofix/Cytoperm (BD) according to the manufacturer’s recommendations and stained intracellularly with IL-12/IL-23p40–specific antibody (BD) or FoxP3-, IL-17 (eBioscience)–, and IFN-γ–specific antibody (BD).

Total RNA was isolated according to the manufacturer’s instructions (RNeasy Mini Plus kit; QIAGEN). RT was performed using random hexamer primers and Moloney murine leukemia virus RT (splenic DCs) or Superscript II (thymic DCs; Invitrogen). cDNA was analyzed by quantitative real-time PCR (qRT-PCR; Bio-Rad Laboratories) in duplicates using SYBR Green PCR Mastermix (Invitrogen) or hydrolysis TAMRA probes (Roche). The expression level of each gene was normalized to HPRT or DNA Polymerase II. The following primers purchased from Operon Technologies were used: HPRT (5′-GACCGGTCCCGTCATGC-3′ and 5′-TCATAACCTGGTTCATCATCGC-3′), DNA Polymerase II (5′-CTGGTCCTTCGAATCCGCATC-3′ and 5′-GCTCGATACCCTGCAGGGTCA-3′), IL-12/IL-23p40 (5′-GACCATCACTGTCAAAGAGTTTCTAGAT-3′ and 5′-AGGAAAGTCTTGTTTTTGAAATTTTTTAA-3′), IL-12p35 (5′-TACTAGAGAGACTTCTTCCACAACAAGAG-3′ and 5′-TCTGGTACATCTTCAAGTCCTCATAGA-3′), IL-23p19 (5′-AGCGGGACATATGAATCTACTAAGAGA-3′ and 5′-GTCCTAGTAGGGAGGTGTGAAGTTG-3′), TLR9 (5′-GCCTTCGTGGTGTTCGATAAGG-3′ and 5′-GAGGTTCTCGAAGAGCGTCTGG-3′), CCL17 (5′-TACTTCAAAGGGGCCATTCCT-3′ and 5′-GCCTTGGGTTTTTCACCAATCT-3′), CCL19 (5′-GGCCTGCCTCAGATTATCTGCCAT-3′ and 5′-GGAAGGCTTTCACGATGTTCC-3′), CCL21 (5′-GGACCCAAGGCAGTGATGGAG-3′ and 5′-CTTCCTCAGGGTTTGCACATAG-3′), CCR2 (5′-ACAAGCACTTAGACCAGGCCAT-3′ and 5′-AAACTGGGCACTGTTTGC-3′), CCR5 (5′-ACTGCTGCCTAAACCCTGTCA-3′ and 5′-GTTTTCGGAAGAACACTGAGAGATAA-3′), CCR6 (5′-TCCATCATCATCTCAAGCCCTACA-3′ and 5′-AGGGGTGAAGAACCCAAAGAACA-3′), CCR7 (5′-ACCATGGACCCAGGGAAC-3′ and 5′-GGTATTCTCGCCGATGTAGTCAT-3′), and CCR9 (5′-TGGCTTGTGTTCATTGTGGGCA-3′ and 5′-ATCCATTGACCAGCAGCAGCAA-3′).

Fig. S1 shows in vitro stimulated splenic DCs and their expression of IL-12/IL-23p40, IL-12p35, and IL-12p19 as described in Fig. 3. Fig. S2 shows the transgenic GFP expression of DC^NIK-NIK^−/− → WT BMCs and the EAE score of T^NIK-NIK^−/− → WT BMCs. Fig. S3 summarizes the expression of various chemokine receptors of thymic DC subsets as described in Fig. 6 D. Fig. S4 shows a phenotypic comparison of T cells from NIK^aly/aly, NIK^aly/+, NIK^−/−, NIK^+/−, and NIK^+/+ mice, in particular the expression of FoxP3 and effector cytokines.

We thank R.D. Schreiber for NIK^−/− mice, S. Jung for CD11cDTR mice, B. Reizis for CD11c-cre mice, and V. Tosevski, R. Höppli, D. Haefeli, A. Reiter, J. Jaberg, and S. Hasler for technical assistance.

The project was supported financially by a grant from the Swiss National Science Foundation (to B. Becher), the Swiss Multiple Sclerosis Society (to B. Becher), the David and Betty Koetser Foundation (to B. Becher and J. Hofmann), and a junior research fellowship from the University of Zürich (to F. Mair).

The authors declare no competing financial interests.