CD28 and c-Rel-dependent Pathways Initiate Regulatory T Cell Development

Abstract

Regulatory T cell (Treg)⁴ development proceeds via a two step process in which naïve CD4⁺ thymocytes are first converted into CD4⁺CD25⁺CD122⁺GITR⁺Foxp3⁻ Treg progenitors followed by a second step in which IL2 converts these Treg progenitors into CD4⁺Foxp3⁺ Tregs. The costimulatory molecule CD28 is required for efficient Treg development. However, the stage at which CD28 affects Treg development remains undefined. Herein we demonstrate that Cd28^-/- mice lack Treg progenitors. Furthermore, the P¹⁸⁷YAP motif in the cytoplasmic tail of CD28, which links CD28 to Lck activation, is required for this process. In contrast, the Y¹⁷⁰MNM motif, which links CD28 to PI3K activation, is not required for Treg progenitor development. Finally, the CD28/Lck pathway has been shown to activate the NFκB family of transcription factors. We demonstrate that c-Rel but not NFκB1 promotes the development of Treg progenitors. Thus, a CD28/c-rel-dependent pathway is involved in initiating Treg development.

Introduction

Regulatory T cells (Tregs) that develop in the thymus do so via a two step-process (1, 2). The first step involves the differentiation of CD4⁺ thymocytes into CD4⁺CD25⁺CD122⁺GITR⁺Foxp3⁻ Treg progenitors. In the second step these Treg progenitors are then converted into CD4⁺Foxp3⁺ Tregs by stimulation with IL2. The first step of this differentiation process is thought to be governed by TCR-dependent signals (2-4). An important question is whether the co-stimulatory molecule CD28 also plays a role in this process.

Previous studies have demonstrated that Cd28^-/- mice have a greatly reduced population of CD4⁺CD25⁺ T cells (5). The effect of CD28 on Tregs could be due to a role for CD28 in promoting IL2 production by effector T cells, or due to a cell intrinsic effect. Work by Tai et al using Cd28 transgenic mice indicated that a PYAP motif within the cytoplasmic domain of CD28 is required for IL2 production and the development of CD4⁺CD25⁺ T cells (6). However, by using WT versus Cd28^-/- bone marrow chimeric mice they demonstrated that the effect of CD28 on Treg development was independent of IL2. Thus, CD28 plays a cell intrinsic role in development of CD4⁺CD25⁺ T cells.

The above studies strongly suggested that CD28, acting via the PYAP cytoplasmic motif, was critical for regulatory T cell development. However, since these studies pre-dated the discovery of thymic Treg progenitors we do not know at what stage CD28 affects Treg development. In addition, these previous studies relied largely on CD25 to identify Tregs and thus may not accurately reflect the development of CD4⁺Foxp3⁺ Tregs.

Moreover, the transgenic expression of CD28 used in the above studies is likely different from that expressed by the endogenous Cd28 gene. This latter point is important, as even 2-fold differences in CD28 expression have been shown to have significant effects on CD28 function (7). Finally, the signaling pathways downstream of CD28 that entrain Treg development have also not been identified. To address this issue we made use of Cd28^-/- mice, as well as two distinct Cd28 knock-in mutants with defects in CD28-dependent PI3K (Y170F mutant) or Lck (AYAA mutant) activation, respectively (7, 8). We demonstrate that CD28 is involved at the first stage of Treg differentiation as Cd28^-/- mice lack Treg progenitors. In addition, we demonstrate that CD28 drives Treg development via the PYAP motif, a motif that has been show to activate the NFκB pathway (9-12). Finally, we demonstrate that the NFκB family member c-Rel, but not NFκB1, promotes the development of Treg progenitors. Thus a CD28/c-Rel dependent pathway is critical for initiating Treg differentiation in the thymus.

Materials and Methods

Intracellular Foxp3 flow cytometry

Thymii were harvested and single cell suspensions were prepared, stained and analyzed as previously described (1, 13). Antibodies used were obtained from eBioscience (San Diego, CA): CD3 (2C11), CD4 (RM4-5), CD8 (53-6.7), CD122 (5H4), CD25 (PC61.5), GITR (DTA-1), CD45.1 (A20) and Foxp3 (FJK-16s).

Mice

nfκb1^-/-, Cd28^-/-, c-rel^-/-, nfκb1^-/- x c-rel^-/-, Cd28 AYAA and Cd28 Y170F KI mutant mice have been previously described (7, 8, 14-17). All mice were on the C57Bl/6 background.

Mixed bone marrow chimeras

Bone marrow cells from mutant (CD45.2) or WT (CD45.1) mice were harvested and mature hematopoietic cell subsets were depleted using biotinylated antibodies (CD3, CD4, CD8, CD25, B220 (RA3-6B2), CD19 (eBio1D3), Ter119 (TER-119), Gr-1 (RB6-8C5) and NK1.1 (PK136)) and Miltenyi magnetic beads. Cells from distinct mice were mixed at a 1:1 ratio and introduced into sublethally irradiated rag2^-/- x γc^-/- host mice by tail vein injection. Mice were analyzed 8 weeks after bone marrow reconstitution. To normalize for differences in the amount of stem cells injected from WT versus mutant bone marrow we first calculated the ratio of WT to mutant cells in DP thymocytes, which are the direct progenitors of both CD4⁺CD25⁻Foxp3⁻ thymocytes and CD4⁺CD25⁺Foxp3⁻ Treg progenitors. We then calculated the ratio of WT to mutant in the non-Treg, Treg progenitor, and mature Treg compartments and divided those ratios by the ratio found in the corresponding DP compartment. This allows us to directly compare the relative contribution of mutant versus WT cells in these distinct cell populations.

Results and Discussion

CD28 is required for Treg progenitor development

To determine whether CD28 is required for development of Treg progenitors we used flow cytometry to identify this population in WT littermate control and Cd28^-/- mice (Fig. 1A). Cd28^-/- mice exhibit a clear reduction in CD4⁺CD25⁺CD122⁺GITR⁺Foxp3⁻ Treg progenitors as well as CD4⁺Foxp3⁺ Tregs (Fig. 1B). We next examined Treg progenitors in two distinct CD28 knock-in (KI) mutant mice. The first CD28 KI involves mutation of a P¹⁸⁷YAP sequence within the cytoplasmic tail to A¹⁸⁷YAA; this mutation prevents CD28 interaction with Lck and results in loss of CD28-dependent IL2 production (referred to as AYAA mutant). The second CD28 KI involves mutation of tyrosine-170 to phenylalanine and prevents CD28-induced activation of the PI3K pathway (referred to as Y170F mutant) (7, 8). As shown in figure 1B, the AYAA mutant mice exhibited defective development of Treg progenitors in the thymus; in contrast, Y170F mutant mice had no defect in Treg progenitor differentiation. Similar results were seen when examining the more mature CD4⁺Foxp3⁺ Treg population in the thymus (Fig. 1B).

Fig. 1.

The CD28 Lck binding motif P¹⁸⁷YAP is required for Treg progenitor development. (A) The gating strategy to identify Tregs and Treg progenitors is shown. The CD122⁺GITR⁺ gate used to identify Treg progenitors was based on staining observed for these markers on CD4⁺Foxp3⁺ Tregs. Shown are CD4⁺Foxp3⁺CD122⁺GITR⁺ Tregs (top right) and CD4⁺CD25⁺CD122⁺GITR⁺Foxp3⁻ Treg progenitors (bottom right) from the thymus of a C57Bl/6 mouse. (B) The frequency of CD4⁺CD25⁺CD122⁺GITR⁺Foxp3⁻ Treg progenitors (left) and CD4⁺Foxp3⁺ Tregs (right) within the CD3⁺CD4⁺CD8⁻ thymocyte subset is shown. Error bars represent standard error of the mean. Data is representative of three independent experiments with 6 mice per group. p ≤ 0.05; determined by Anova using Bonferroni correction.

The CD28-Lck binding motif is required in a cell-intrinsic manner for Treg progenitor development

The results shown in figure 1 could reflect a requirement for CD28-dependent IL2 production by effector T cells needed for survival of Treg progenitors. Alternatively, these results could be due to a cell intrinsic effect of the CD28/Lck pathway on the conversion of naïve CD4⁺CD25⁻Foxp3⁻ thymocytes into Treg progenitors. To examine this issue, we isolated bone marrow from Cd28^-/-, AYAA, or Y170F mutant mice, mixed them at a 1:1 ratio with CD45.1+ WT congenic mice and injected them into sublethally irradiated rag2^-/- x γc^-/- host mice to generate mixed bone marrow chimeras. We then analyzed these mice by flow cytometry to assess the relative contribution of WT and mutant –derived cells to distinct thymocyte subsets. Both CD4⁺CD8⁺ double positive thymocytes and CD4⁺CD25⁻Foxp3⁻ single positive thymocytes showed no obvious skewing of the ratio between WT and mutant progenitors from the initial input of ~1:1 (Fig. 2 and data not shown). In contrast, both Cd28^-/- and AYAA-derived cells exhibited a significant decrease in their ability to contribute to Treg progenitor populations in these mixed bone marrow chimeras, although the magnitude of the defect for AYAA mutant cells was modestly but consistently less than that observed for Cd28^-/- cells. A similar result was observed when examining the contribution of Cd28^-/- and AYAA-derived cells to the CD4⁺Foxp3⁺ Treg subset. Finally, Y170F-derived cells were able to contribute to Treg progenitor and Treg subsets with equal efficiency as that seen for CD45.1+ WT–derived cells (Fig. 2). These results demonstrate that CD28 is required in a cell-intrinsic manner for development of Treg progenitors and that this is largely dependent on the CD28 P¹⁸⁷YAP, but not the Y¹⁷⁰MNM, motif.

Fig. 2.

The CD28 Lck binding motif is required for Treg progenitor development in a cell-intrinsic manner. (A) Thymii were isolated and stained as described in the methods. Contour plots depict mutant (CD45.1−) and WT (CD45.1+) cells gated on CD4⁺CD25⁻ Foxp3⁻ thymocytes (top row), CD4⁺CD25⁺CD122⁺GITR⁺Foxp3⁻ Treg progenitors (middle row) and CD4⁺Foxp3⁺ Tregs (bottom row). (B) Comparison of the ratio of WT to mutant cells between mixed chimeras in CD4⁺CD25⁻Foxp3⁻ thymocytes (left), Treg progenitors (middle), and CD4⁺Foxp3⁺ Tregs (right). Ratios were normalized by comparison to the ratio of WT to mutant cells in double positive thymocytes to control for differences in the initial input ratio of WT to mutant cells. Error bars represent standard error of the mean. Data is representative of three independent experiments with 11-15 recipient mice per group. p ≤ 0.002 determined by unpaired t-test.

c-Rel but not NFκB1 is required for Treg progenitor development

The CD28-PYAP motif is known to govern a Lck/PKCθ pathway that subsequently regulates activation of the CARMA1/Bcl10/Malt1 complex (18). The CARMA1 complex in turn governs activation of both Jnk2 and the IKK complex leading to activation of the classical NFκB pathway (18). Previous work has demonstrated that mice lacking both NFκB1 and c-Rel have defects in CD4⁺CD25⁺ T cells suggesting a role for this pathway in CD4⁺Foxp3⁺ Treg homeostasis and/or function (17). Likewise, mice expressing a NFκB1^SSAA knock-in mutation, which leads to impaired activation of NFκB1, RelA and c-Rel have also been shown to have reduced numbers of mature Foxp3⁺ Tregs (19). To explore the role of the NFκB pathway in the development of Treg progenitors, and to more precisely identify which NFκB family members are required for this process, we examined these populations in nfκb1^-/-, c-rel^-/- and nfκb1^-/- x c-rel^-/- mice. Compared to WT controls, nfκb1^-/- x c-rel^-/- mice were devoid of Treg progenitors and mature Tregs (Fig. 3). A similar result was seen in c-rel^-/- mice. In contrast, nfkb1^-/- mice showed no defect in Treg progenitors although they did exhibit a significantly smaller population of mature CD4⁺Foxp3⁺ Tregs in the thymus (Fig. 3).

Fig. 3.

c-Rel but not NFκB1 is required for Treg progenitor development. (A) Shown are contour plots gated on CD4⁺CD25⁺CD122⁺GITR⁺Foxp3⁻ Treg progenitors from WT control (far left), nfκb1^-/- x c-rel^-/- (middle left), nfκb1^-/- (middle right) and c-rel^-/- (far right). (B) The frequency of Treg progenitors (left) and CD4⁺Foxp3⁺ Tregs (right) in CD3⁺CD4⁺CD8⁻ thymocytes is shown. Data is representative of two independent experiments with three mice per group. p ≤ 0.05; determined by Anova using Bonferroni correction.

To investigate whether c-Rel or NFκB1 is required in a cell intrinsic manner for Treg progenitor or mature Treg development, we made mixed bone marrow chimeras with CD45.1+ WT versus CD45.2+ nfkb1^-/-, c-rel^-/-, or nfkb1^-/- x c-rel^-/- bone marrow. Consistent with previous studies, nfκb1^-/-, c-rel^-/- and nfκb1^-/- x c-rel^-/- derived cells showed no defect in the ability to generate CD4⁺CD25⁻Foxp3⁻ thymocytes (Fig. 4A, top row). In contrast, both c-rel^-/- and nfκb1^-/- x c-rel^-/- cells were unable to efficiently generate Treg progenitors and mature Tregs (Fig. 4A, middle and bottom rows). Nfκb1^-/- derived cells did not have a defect in the ability to generate either Treg progenitors or mature Tregs in these mixed bone marrow chimeras (Fig. 4A-B). This latter result suggests that the reduced number of mature Tregs in the thymus of nfkb1^-/- mice reflects a role for NFκB1 in trans, most likely due to effects of NFκB1 in promoting IL2 production by effector T cells.

Fig. 4.

c-Rel but not NFκB1 is required in a cell intrinsic manner for Treg progenitor development. (A) Contour plots depict the ratio of mutant (CD45.1−) versus wild-type cells (CD45.1+) gated on CD4⁺CD25⁻Foxp3⁻ thymocytes (left), CD4⁺CD25⁺CD122⁺GITR⁺Foxp3⁻ Treg progenitors (middle) and CD4⁺Foxp3⁺ Tregs (bottom). (B) Comparison of the ratio of wild-type to mutant cells between mixed chimeras in CD4⁺CD25⁻Foxp3⁻ thymocytes (left), Treg progenitors (middle), and CD4⁺Foxp3⁺ Tregs (right). Ratios were normalized by comparison to the ratio of WT to mutant cells in double positive thymocytes to control for differences in the initial input ratio of WT to mutant cells. Error bars represent standard error of the mean. Data is representative of two independent experiments with 8 nfκb1^-/- vs. WT, 9 nfkb1^-/- x c-rel^-/- vs. WT, and 4 c-rel^-/- vs. WT mixed bone marrow chimeric mice. p < 0.02 determined by unpaired t-test.

Our results demonstrate that CD28 and c-Rel-dependent pathway plays a critical role in initiating the development of Treg progenitors. This most likely reflects a role for c-Rel homodimers (or possibly c-Rel:RelA or c-Rel:NFκB2 heterodimers), as NFκB1 is not required for this process. Activation of the c-Rel pathway may involve signals initiated by the TCR although previous studies have shown that this is greatly enhanced by CD28 co-stimulation (9). Moreover, substantial evidence exists tying CD28 and the P¹⁸⁷YAP motif to c-Rel activation (9-12, 18). Taken together with the findings reported here, this suggests a critical role for a CD28/c-Rel dependent pathway in initiating the first step in Treg development.

c-Rel likely initiates Treg development via two distinct mechanisms. First, NFκB family members have been shown to play a role in regulating CD25 expression on T cells (20). Thus, c-Rel probably allows developing Tregs to respond to IL2 by inducing high-level expression of the IL2R complex on Treg progenitors. However, this cannot be the only mechanism by which c-Rel governs Treg development as crossing Carma1^-/- mice, which are defective in c-Rel activation, to mice expressing a constitutively active form of STAT5 (STAT5b-CA) does not rescue Treg development (13). Since Treg development is restored in IL2Rβ^-/- x STAT5b-CA mice, the failure to rescue Tregs in Carma1^-/- x STAT5b-CA mice suggests that c-Rel affects Treg differentiation via mechanisms other than just inducing IL2R expression. An intriguing possibility is that c-Rel may bind to and prime the foxp3 locus for subsequent cytokine-dependent transcription. We envision that c-Rel binding to the foxp3 gene would result in epigenetic changes that would permit subsequent IL2/STAT5-dependent signals to initiate foxp3 transcription. Supporting this hypothesis, conserved NFκB binding sites exist in both the first and second introns of the foxp3 gene (data not shown). Our results support a model in which CD28/c-Rel-dependent signals promote the development of Treg progenitors by inducing expression of the IL2R complex on these cells and possibly by priming the foxp3 gene, in an as yet undefined manner, for subsequent transcription (step I). These primed Treg progenitors would then be converted into mature Tregs by IL2/STAT5-dependent signals that lead to actual foxp3 transcription (step II), thereby completing Treg differentiation.