Type II Cytokines Fine-Tune Thymic T Cell Selection to Offset Murine Central Nervous System Autoimmunity

Early thymic progenitors (ETPs), the bone marrow (BM)–derived stem cells that settle in the thymus are pluripotent (1, 2) and give rise to myeloid as well as adaptive and innate lymphoid cells (3–6). Recently, we demonstrated that ETPs expressing the IL-4Rα/IL-13Rα1 heteroreceptor (HR) give rise only to myeloid cells (7). This is because endogenous IL-4 and IL-13 use the HR to activate STAT1 and STAT6 and inhibit the ETPs’ T cell potential, leading to commitment to myeloid cells, the majority of which are dendritic cells (DCs) (8, 9). The biological significance of cytokine-driven maturation of HR⁺-ETPs to DCs remains a puzzle.

The thymus is the site for T cell development, a process that involves positive selection of maturing thymocytes and negative selection of self-reactive T cells (10). Both positive and negative selection of T cells require presentation of self-peptides by APCs (10). It is now clear that cortical thymic epithelial cells (cTEC) are the main APCs involved in positive selection, whereas medullary thymic epithelial cells (mTECs) and DCs are responsible for negative selection of self-reactive lymphocytes (11, 12). It is also known that mTECs sustain optimal expression of self-Ags in the thymus (13, 14), whereas DCs present self-peptides to the target T lymphocytes (15–17). This peptide presentation function has always been attributed to DCs generated from BM stem cells in sites peripheral to the thymus (18). The question that arises in this study is whether cytokines divert ETP maturation toward DCs to yield local APCs that would monitor developing thymocytes for self-reactivity and tighten negative selection and elimination of potentially harmful autoreactive T cells. To test this postulate, we developed animal models defective for T cell selection and used these tools to interrogate cytokine-induced ETP-derived APCs for restoration of central tolerance. The findings indicate that ETP-derived APCs, although unable to contribute to positive selection of maturing T cells, are effective in carrying out negative selection of self-reactive T cells and lessening the clinical signs of experimental autoimmune encephalomyelitis (EAE) in an autoimmune regulator (aire) gene (13, 14)–dependent fashion. These previously unrecognized observations suggest that IL-4 and IL-13 serve as pillars for control of ETP commitment and fine-tuning of central tolerance.

All animal experiments were done according to protocols approved by the University of Missouri Animal Care and Use Committee. C57BL/6, Aire^−/− C57BL/6 mice, and OT-1 and OT-II TCR-transgenic C57BL/6 mice were purchased from The Jackson Laboratory (Bar Harbor, ME). IL-13Rα1^+/+-GFP and IL-13Rα1^−/− C57BL/6 mice were previously described (7). CT2Akd/IRES/Ly5.1 mice deficient for MHC class II (MHCII) expression in mTECs were previously described (19). MHCII^−/− and MHC I^−/−/II^−/− C57BL/6 mice were purchased from Taconic (Hudson, NY). 2D2 C57BL/6 mice carrying the transgenic TCR specific for myelin oligodendrocyte glycoprotein (MOG) were previously described (20). Only 6- to 8-wk-old mice were used throughout the study.

MOG peptide (MOGp) corresponding to MOG 35–55 peptide and MOGp tetramer (MOG^tet) were previously described (21). Chicken OVA (OVA) aa 257–264 (SIINFEKL) and 323–339 (SQAVHAAHAEINEAGR) (OVAp) peptides were purchased from EZbiolab (Carmel, IN). Ig-OVA is an Ig chimera carrying OVAp (22), and Ig-p79 is carrying p79 peptide (23).

Abs were purchased from BD Biosciences (San Jose, CA), eBioscience (San Diego, CA) BioLegend (San Diego, CA) or Vector Laboratory (Burlingame, CA) and used according to the manufacturer. Sample reading used a Beckman Coulter CyAn (Brea, CA) and data were analyzed using FlowJo version 10 (Tree Star). Dead cells were excluded using 7-aminoactinomycin D (7AAD; EMD Biosciences) or Fixable Viability Dye (FVD) eFluor 780 (eBioscience).

ETPs were isolated as previously described (7). In brief, thymic cells were depleted of Lin⁺ thymic cells and the HR⁺ETPs (cKit⁺CD44⁺CD25⁻) were sorted from IL-13Rα1^+/+-GFP reporter mice on the basis of GFP (IL-13Rα1) expression. HR⁻ETPs were sorted from Lin⁻ thymic cells of IL-13Rα1^−/− mice on the basis of CD44, cKit, and CD25 (cKit⁺CD44⁺CD25⁻).

Thymi were harvested, and CD3⁺CD4⁺CD8⁺ cells were sorted from either 2D2 TCR-transgenic or MHCII^−/− C57BL/6 mice to isolate monoclonal and polyclonal double-positive (DP) thymocytes.

Thymic epithelial cells (TECs) were isolated from the thymus as described (24) with a slight modification. Briefly, thymi from 6- to 8-wk-old MHCII^+/+ C57BL/6 mice were treated with 0.005% (weight/volume) Liberase TH and 100 U/ml DNase I (Roche Diagnostics, Indianapolis, IN) to release epithelial cells from the thymi. cTECs were sorted as CD45⁻EpCAM⁺MHCII⁺UEA⁻Ly51⁺ cells, whereas mTECs were isolated as CD45⁻EpCAM⁺MHCII⁺UEA⁺Ly51⁻ cells.

Sorted ETPs were cultured on OP9 stromal cells as previously described (25) with 10 ng of IL-4, and myeloid progeny were sorted on day 7 as CD45⁺CD11c⁺.

Sorting was performed on a Beckman Coulter MoFlo XDP (Brea, CA) cell sorter. Only sorts with a purity of ˃95% were used in this study.

ETPs, DP thymocytes, cTECs, and mTECs were resuspended in 30 μl of PBS and injected into isoflurane-anesthetized mice through the skin between the third and fourth rib of the thoracic cavity using a 0.3-ml, 31-gauge, 8-mm insulin syringe.

HR^+/+ C57BL/6 mice (CD45.1) were given (intrathymically [i.t.]) HR⁺ETPs (5 × 10⁴ cells per mouse) from HR^+/+ C57BL/6 donors (CD45.2), and thymic cells were harvested on day 12 or 16 posttransfer. The day 12 cells were used to analyze expression of CD11b, CD11c, and CD8α on CD45.2 gated cells, whereas the day 16 cells served to analyze expression of CD11b, CD11c, and CD3 markers.

HR^+/+ C57BL/6 mice deficient for MHCII (MHCII^−/−) or MHC class I and MHCII (MHC I^−/−II^−/−) were given (i.t.) HR⁺ETPs (15 × 10³ cells per mouse), or cTECs (10 × 10³ cells per mouse) twice (7 d apart). The hosts were sacrificed at different time points, and their thymic and peripheral blood cells were assessed for single-positive (SP) CD4 and CD8 T cells. Negative control mice received PBS with no cells (NIL).

C2TAkd mice (CD45.1) were lethally irradiated (900 rad) and given BM cells (10 × 10⁶ cells per mouse) from MHCII^−/− mice. After 2 wk of reconstitution, the mice were injected (i.t.) with unselected DP CD4⁺CD8⁺ monoclonal 2D2 TCR-transgenic or polyclonal (CD45.2) MHCII^−/− thymocytes. In parallel, the hosts were given HR⁺ETPs (12 × 10³ cells per mouse) from MHCII^+/+ mice and negative selection was measured at different time points by assessing the number of SP CD4⁺ T cells by flow cytometry.

The mice were given (i.t.) HR⁺ETPs (15 × 10³ cells per mouse) from aire^+/+ C57BL/6 mice twice, 7 d apart. Thymic cells were harvested at different time points, and the number of SP CD4⁺ or CD8⁺ T cells were analyzed by flow cytometry. Positive control mice were given (i.t.) mTECs (15 × 10³ cell per mouse) from aire^+/+ C57BL/6 mice, and the number of SP CD4⁺ and CD8⁺ T cells in thymus were analyzed on day 6 posttransfer by flow cytometry.

Mice were induced for EAE with 60 μg of MOGp, as previously described (21). The mice were scored daily for clinical signs of EAE as follows: 0, no clinical signs; 1, loss of tail tone; 2, hind limb weakness; 3, hind limb paralysis; 4, forelimb weakness; 5, forelimb paralysis; and 6, moribund or death. The cumulative disease score was calculated by adding the daily scores that the mice received during the monitoring period divided by the number of mice per group. The mean maximal disease score (mmds) represents the average of the highest score received by each mouse during the monitoring period.

Data were analyzed using either an unpaired, two-tailed Student t test, one-way ANOVA, or Mann–Whitney U test as indicated. All statistical analyses were performed using Prism software version 4.0c (GraphPad).

In earlier studies, we have shown that HR⁺ETPs cultured in vitro on OP9 stromal cells in the presence of IL-4 or IL-13 cytokines give rise to CD11c⁺ DCs (9). In this study, we asked whether these HR⁺ETP–derived DCs can function as APCs. The results show that DCs derived in vitro from HR⁺ETP by culture on OP9 stromal cells in the presence of IL-4 function present Ag on MHC class I and MHCII molecules (Fig. 1). Indeed, the DCs were able to present free K^b-restricted SIINFEKL peptide to OT-I CD8 T cells as measured by CFSE dilution (Fig. 1A). Moreover, the DCs are able to cross-present whole-OVA protein and induced CFSE dilution of OT-1 CD8 T cells when the native OVA was loaded into the DCs by osmotic shock in hypertonic but not isotonic media (Fig. 1A). Similarly, the HR⁺ETP–derived DCs were able to present class II–restricted free OVAp to OT-II CD4 T cells (Fig. 1B). In addition, when OVAp was delivered to the DCs in the form of Ig-OVA, endocytic presentation was operative and the T cells were able to proliferate and dilute CFSE (Fig. 1B). The control p79 peptide and Ig-p79 did not induce proliferation of OT-II CD4 T cells. In all, DCs derived from HR⁺ETPs by stimulation with IL-4 function as APCs in vitro.

The CD11c⁺ DCs derived from ETPs comprise CD8α⁺DCs that express IRF-8 but not SIRPα markers (9). These nonmigratory CD8α⁺SIRPα^- DCs are commonly found in mice sufficient for the HR (9) perhaps because IL-4 is readily available in the thymic environment (26). To ensure that HR⁺ETP maturation is restricted to the myeloid lineage and give rise to DCs in vivo, the cells were sorted from the thymi of CD45.2 IL-13Rα1^+/+GFP reporter donor C57BL/6 mice, transferred i.t. into CD45.1 IL-13Rα1^+/+ C57BL/6 hosts and their in vivo maturation to both myeloid and lymphoid lineages was analyzed. The results show that HR⁺ETPs do not give rise to T cells but rather to myeloid CD11c⁺cells that would include both CD11c⁺CD11b⁺ and CD11c⁺CD11b⁻ cells (Fig. 1C). Data compiled from several experiments show that the findings are statistically significant despite that the incubation period was extended to 16 d, an optimal time point for CD3 expression (27). Myeloid cells are diverse in nature and can be broadly classified into CD11b⁺ monocyte/macrophage cells or CD11c⁺ DCs. Because conventional DCs are specialized APCs and contribute to thymic T cell selection (10) we determined the cellular make-up of the in vivo ETP-derived myeloid population. The results show that 17.7 ± 1.2% of the CD45.2 cells expressed only CD11b, a profile for monocyte/macrophage cells (Fig. 1D). However, 61.9 ± 2.7% of the cells had CD11c but not CD11b phenotype, which represents a conventional DC phenotype (Fig. 1D). Interestingly, most of the DCs (78.7 ± 2.7) expressed the CD8α DC subset–specific marker. In terms of cell number, data collected from several experiments indicated that 50 × 10³ HR⁺ETPs gave rise to 400 × 10³ myeloid cells on average. Of these myeloid cells, 70.8 × 10³ (17.7%) were CD11b⁺ CD11c⁻ cells and 247.6 × 10³ (61.9%) were CD11b⁻ CD11c⁺ DCs. Of these CD11b-CD11c⁺ cells, 193 × 10³ (78%) were CD8α⁺ DCs. Overall, HR⁺ETPs give rise to myeloid but not T cells in vivo and a significant number of these cells are CD8α⁺CD11c⁺ DCs.

Because BM-derived DCs migrate to the thymus and participate in thymic selection of T cells (18, 28, 29), one would envision that HR⁺ETPs give rise to CD8α⁺SIRPα⁻ thymic resident DCs to assist in T cell development. To test this premise, experimental models were set up to assess the contribution of HR⁺ETP–derived APCs to positive, as well as negative, T cell selection. Accordingly, MHCII^−/− mice, in which CD4 T cell selection is not operative, were given i.t. HR⁺ETPs and analyzed for development of CD4⁺ SP T cells in the thymus and the periphery. The result shows that the frequency of CD4⁺ T cells did not increase in either the blood or thymus 15 d postintrathymic injection, as the numbers of cells were similar to mice that did not receive HR⁺ETPs (Supplemental Fig. 1A). In contrast, MHCII^−/− mice recipient of MHCII⁺ cTECs that support positive T cell selection (10) showed an increase in the percentage of CD4⁺ SP T cells in both the thymus and in the periphery (Supplemental Fig. 1B, upper panels). Data compiled from several experiments show that the increases in CD4⁺ SP T cells were statistically significant (Supplemental Fig. 1B, lower panels). Furthermore, when MHCI^−/−/II^−/− mice were given HR⁺ETPs and tested for positive selection of both CD4 and CD8 T cells, there was no significant increase of either T cell type in the thymus (Supplemental Fig. 2A). Again, mice recipient of MHCI⁺II⁺ cTECs had significantly increased percentages of both CD4 and CD8 T cells in the blood and thymus (Supplemental Fig. 2B, 2C). Data compiled from several experiments show that the increases in CD4⁺ and CD8⁺ SP T cells were statistically significant (lower panels in Supplemental Fig. 2B, 2C). Overall, these findings indicate that HR⁺ETP–derived APCs are unable to support thymic-positive selection of T cells.

The fact that HR⁺ETPs do not contribute to positive selection of T cells does not necessarily preclude contribution to negative T cell selection. To test this premise, we set up a chimeric mouse model suitable for this investigation as illustrated in Fig. 2A. Accordingly, C2TAkd mice (19), which lack MHCII on mTECs but not cTECs, were lethally irradiated and reconstituted with BM from MHCII^−/− mice yielding a chimeric host devoid of MHCII expression in APCs and mTECs but not cTECs. The thymic configuration in this host supports positive but not negative selection. This chimera was then given HR⁺ETPs from MHCII^+/+ mice to provide thymic resident MHCII-sufficient APCs that could serve in negative selection. The hosts were then given unselected CD4⁺CD8⁺ DP 2D2 TCR_MOGp monoclonal (20) or CD45.2 C57BL/6 polyclonal thymocytes to serve as targets for negative selection. After 7–21 d the thymi were analyzed for CD4⁺CD8⁻ SP T cells as a measure of thymic-negative T cell selection. The results show that both monoclonal (MOG^tet+) and polyclonal (CD3⁺) CD4⁺ SP T cells were significantly lower in number in mice recipient of DP thymocytes and HR⁺ETPs relative to those given DP thymocytes alone (Fig. 2B, 2C). Furthermore, when the polyclonal CD45.2 CD4⁺ SP T cells were stained with Nur77 Ab and FVD, there was a gradual increase over time in the number (Fig. 2D) and percentage (Fig. 2E) of cells binding FVD, a marker for apoptosis, and expressing Nur77, a marker for negatively selected thymocytes (30), in the mice recipient of DP thymocytes and HR⁺ETPs relative to those given the DP thymocytes alone. These results indicate that HR⁺ETP–derived APCs support negative selection of CD4 T cells. HR⁺ETP–derived APCs were not tested for negative selection of CD8 T cells because a similar animal model could not be devised because of the lack of mice with MHC class I deficiency in mTECs.

HR⁺ETP–derived APCs contribute negative selection of self-reactive T cells and reduce the frequency of myelin-specific T lymphocytes in the periphery. This process likely impacts autoimmunity and impedes the development of EAE. To test this premise, HR^−/− mice in which the frequency of thymic resident CD8α⁺SIRPα⁻ DCs is diminished (9), and negative T cell selection would be less effective, were induced for EAE with MOGp and their daily disease severity scores were compared with HR^+/+ mice. Indeed, the results show that HR^−/− mice begin to develop signs of clinical EAE on day 5 postdisease induction, whereas HR^+/+ mice had the initial disease scores on day 9 (Fig. 3A, left panel). Also, the pattern of paralysis was more severe in HR^−/− relative to HR^+/+ mice (Fig. 3A, right panel). Furthermore, the mmds was 4.0 ± 0.0 in HR^−/− mice compared with 2.8 ± 0.4 in HR^+/+ animals. The cumulative disease score was 43.8 ± 2.7 in HR^−/− mice which is much higher than the 24.4 ± 1.8 in HR^+/+ mice. These observations suggest that the increased frequency of ETP-derived DCs plays a critical role in the development of EAE. This statement is supported by data showing that HR^−/− recipients of HR⁺ETPs develop milder EAE relative to HR^−/− mice that did not receive HR⁺ETPs prior to disease induction (Fig. 3B). Indeed, although the onset of EAE was similar in both experimental groups, the pattern of paralysis is milder in the mice recipient of HR⁺ETPs. In fact, the mmds decreased from 4.0 ± 0.0 in HR^−/− mice with no ETP transfer to 3.0 ± 0.0 in HR^−/− mice recipient of HR⁺ETPs. In addition, the cumulative disease score, which was 45.0 ± 2.4 in HR^−/− mice, decreased to 30.3 ± 1.2 in the mice recipient of HR⁺ETPs. In all, the data indicates that HR⁺ETP–derived APCs support thymic-negative selection of self-reactive T cells leading to reduced susceptibility to EAE induction.

Thymic-negative T cell selection relies on the expression of the aire gene by mTECs (13, 14) and their cooperation with thymic APCs (10). Thus, we sought to determine whether HR⁺ETP–derived APCs would require the function of aire gene to contribute negative T cell selection and protect against EAE. To this end, aire^−/−HR^+/+ hosts were given HR⁺ETPs or mTECs from aire^+/+ mice and then tested for negative selection of T cells and resistance to EAE induction. The results show that the hosts, in which both endogenous mTEC and HR⁺ETPs lack aire expression, had no significant decrease in numbers of either CD4 or CD8 SP T cells when given aire⁺HR⁺ETPs or PBS (NIL) (Fig. 4A). In contrast, hosts recipient of aire⁺mTECs showed significant reduction in both frequency and numbers of CD4 or CD8 SP T cells relative to NIL hosts (Fig. 4B, 4C), indicating cooperation between endogenous aire⁻HR⁺ETP–derived APCs and exogenous aire⁺mTECs. These findings parallel with data presented in Fig. 2 showing cooperation between HR⁺ETP–derived APCs and MHCII^-aire⁺mTECs of the C2TAkd hosts. In addition, given that APCs derived from HR⁺ETPs, whether in vitro or in vivo, have little aire mRNA relative to mTECs (Fig. 4D), these results suggest that negative selection of T cells mediated by HR⁺ETP–derived APCs require aire expression in mTECS. This statement is supported by findings showing that HR⁺ETPs are able to lessen the clinical signs of EAE in HR^−/− mice sufficient for aire (Fig. 4E). Indeed, the clinical signs EAE in HR^−/−Aire^−/− mice were similar whether the hosts are given HR⁺ETP or PBS (NIL) control (Fig. 4E, left panel). However, the severity of EAE was significantly lower in HR^−/−Aire^+/+ hosts given HR⁺ETP in comparison with HR^−/−Aire^+/+ mice given PBS (NIL) (Fig. 4E, right panel). Given that the host mice are deficient for the HR and do not have endogenous HR⁺ETPs, the differential patterns of EAE is because of cooperation between aire and the transferred HR⁺ETPs. In all, HR⁺ETP–derived APCs require the function of aire gene to contribute central tolerance and impact the development of autoimmunity.

To our knowledge, this study reports new insights on the function of IL-4 and IL-13 in ETP maturation and its significance to central T cell tolerance and the development of CNS autoimmunity. IL-4/IL-13 were primarily identified as major players in allergic reactions and parasite immunity (31). The cytokines can also play anti-inflammatory functions against autoimmunity by promoting peripheral tolerance (32, 33). Recently we reported that IL-4/IL-13 signaling through the HR guides ETP maturation toward myeloid cells, the majority of which belong to the DC population (9). This previously unappreciated function positions the cytokines as factors that assist ETP commitment to a specific lineage and reinforce the contribution of the thymic microenvironment to ETP fate decision (34, 35).

The other intriguing aspect in the process of cytokine guided ETP maturation relates to the biological significance associated with the shifting of ETP maturation toward myeloid cells, specifically DCs. It is well known that DCs are professional APCs specialized in induction of immunity as well as peripheral tolerance (36). Lately, it has been reported that a specific population of BM-derived DCs, namely the CD8α⁺CD11c⁺ subset is specialized in thymic selection of T cells (18, 28). Interestingly, IL-4/IL-13 driven signaling supports HR⁺ETPs to give rise to DCs that are able to function as APCs and present Ag via MHC class I and MHCII classical pathways. The ETP-derived DCs are also able to cross-present Ag to CD8 T cells, a functional attribute of CD11c⁺CD8α⁺ DCs that contribute to thymic-negative T cell selection. These observations prompted us to test whether the APC function of HR⁺ETP–derived DCs contribute to thymic T cell selection (10). The findings indicate that HR⁺ETP–derived APCs, although unable to support positive selection of T cells because this is a defined function for cTECs (10), were able to negatively select self-reactive T cells including 2D2 TCR-transgenic myelin-specific T lymphocytes. In fact, HR⁺ETP–derived APCs were able to lessen severity of EAE as the mice had milder clinical signs of disease when compared with animals in which negative selection is not operative. In addition, both negative selection of T cells and lessening of EAE by HR⁺-derived APCs were dependent on the function of the aire gene (14), suggesting that the DCs present self-peptide generated from mTECs as was previously defined (13). Although it has been reported that EAE severity is slightly reduced in young aire^−/− mice compared with their wild-type counterparts (37), this was not the case in young aire^−/− HR^−/− mice, and the disease pattern was similar to young aire^+/+ HR^−/− mice, further supporting our prior observation that HR deficiency increases susceptibility to EAE (32). These observations point to new attributes for IL-4 and IL-13 whereby signaling through the HR guides ETPs to give rise to APCs that tighten negative selection of self-reactive T cells and reduce susceptibility to autoimmunity. The significance of these findings is 2-fold. On one hand, stimulators of IL-4 and IL-13 secretion would be able to influence ETP maturation and impact central tolerance as well as autoimmunity. On the other hand, the environment would also control this process under circumstances that could favor or deter the development of autoimmunity. For instance, parasitic infections and allergens that stimulate type II cytokines would support the generation of ETP-derived APCs, which foster negative T cell selection and limit the generation of self-reactive T cells yielding a lymphocyte repertoire devoid of self-reactivity and thus, beneficial against autoimmunity. From this perspective, a clean environment would foster susceptibility to autoimmunity (38, 39). In all, these observations add insight as to the environmental factors that would control the development of autoimmune diseases and assert the hygiene hypothesis (40).

We thank Chyi-Song Hsieh and Ludger Klein for providing the CT2Akd IRES Ly5.1 mice.

The authors have no financial conflicts of interest.