The adaptor TRAF3 restrains the lineage determination of thymic regulatory T cells by modulating signaling via the receptor for IL-2.

The number of Foxp3+ regulatory T cells (Treg cells) must be tightly controlled for efficient suppression of autoimmunity with no impairment of normal immune responses. Here we found that the adaptor TRAF3 was intrinsically required for restraining the lineage determination of thymic Treg cells. T cell-specific deficiency in TRAF3 resulted in a two- to threefold greater frequency of Treg cells, due to the more efficient transition of precursors of Treg cells into Foxp3+ Treg cells. TRAF3 dampened interleukin 2 (IL-2) signaling by facilitating recruitment of the tyrosine phosphatase TCPTP to the IL-2 receptor complex, which resulted in dephosphorylation of the signaling molecules Jak1 and Jak3 and negative regulation of signaling via Jak and the transcription factor STAT5. Our results identify a role for TRAF3 as an important negative regulator of signaling via the IL-2 receptor that affects the development of Treg cells.

Tight regulation of the Foxp3+ regulatory T (Treg) cell population in immunity is crucial to avoid pathogenic autoreactivity while providing effective protection against infectious diseases and tumor cells[1]. Interleukin-2 receptor (IL-2R) mediated signaling is a major mechanism controlling Treg cell development and homeostasis, and has been widely investigated[2-4]. IL-2 binding to the IL-2R activates at least three distinct signaling pathways. Activation of Janus kinase (Jak) 1 and 3 associating with IL-2Rβ (CD122) and common γ chain (CD132) respectively, leads to phosphorylation of IL-2Rβ and the transcription factor STAT5[5,6]. Phosphorylated STAT5 binds to the promoter and first intron of the Foxp3 gene and is essential for initiating Foxp3 expression[7,8]. IL-2 also activates PI3K-Akt and Ras-MAPK signaling pathways. But in contrast to STAT5, which can be directly phosphorylated by Jak3, additional intermediate molecules, such as Shc, Syk, and Lck are required for activation of these pathways[7,9,10]. Several negative regulatory mechanisms are involved in restraining IL-2-mediated signaling. Suppressor of cytokine signaling 1 (SOCS1) and 3 play negative feedback roles in IL-2 signaling by associating with Jak1 and inhibiting its kinase activity[11,12]. The SH2 domain-containing protein phosphatase 1 (SHP-1) dephosphorylates Jak1 and negatively regulates IL-2R-Jak1 signaling[13]. T cell protein tyrosine phosphatase (TCPTP) can also directly interact with Jak1 and Jak3 and dephosphorylate these molecules upon IL-2 or interferon-γ (IFN-γ) stimulation[14]. As a tyrosine-specific phosphatase, TCPTP expression is ubiquitous, but it is expressed in higher amounts in cells of hematopoietic origin[15]. The important role of TCPTP in cytokine signaling is demonstrated in vivo by TCPTP-deficient mice, which show a severe pro-inflammatory phenotype and die at 3-5 weeks of age[16]. Notably, Treg cells are moderately increased in T cell specific TCPTP deficient mice[17].

TNF receptor associated factor 3 (TRAF3) is an adaptor molecule that participates in signaling by many members of the TNF receptor superfamily (TNFRSF), as well as innate immune receptors and the IL-17 receptor[18-20]. Previous studies indicate that the roles of TRAF3 are highly cell type- and receptor-dependent[21]. The functions regulated by TRAF3 in T cells have been less intensively examined than those in B cells. We reported that T cell-specific deficiency in TRAF3, while having no detectable impact on development of conventional T cells, causes decreased T cell effector functions and impaired T cell receptor (TCR) signaling in peripheral CD4+ and CD8+ T cells[22]. Deficiency of TRAF3 also results in both defective development and function of invariant Natural Killer T (iNKT) cells[23]. Another study indicates that Treg cell-specific TRAF3 expression is required for follicular Treg cell (TFR) induction[24]. Therefore, TRAF3 plays distinct roles in different T cell subsets.

In the current study, we examined the molecular mechanisms by which T cell-specific TRAF3 deficiency in mice results in a highly reproducible 2-3 fold increase of the Treg cell numbers. Our results establish TRAF3 as a critical factor in regulating IL-2R signaling to T cells, with important consequences for Treg cell development.

RESULTS

Cell-intrinsic TRAF3 impact on Treg cell development

Despite the ubiquitous expression of TRAF3, conventional CD4+ and CD8+ T cells appeared to develop normally in T cells deficient in TRAF3 (Cd4CreTraf3flox/flox, hereafter termed T-Traf3–/–) mice, although these cells have markedly reduced activation responses[22]. In contrast, the frequency of CD4+Foxp3+ Treg cells showed a highly reproducible 2-3-fold increase in frequency in all peripheral lymphoid tissues examined[22] (). The percentage and number of Treg cells, but not conventional CD4+ T (Tcon) cells in the thymus was also increased 2-3-fold in T-Traf3–/– mice compared to littermate controls (LMC) (). To determine whether this increased number was cell intrinsic, bone marrow (BM) chimeric mice were generated by transferring mixed wild-type (WT) (CD45.1+) and T-Traf3 (CD45.2+) BM at 1:1 or 20:1 ratios into lethally irradiated WT mice (CD45.1+ CD45.2+). Eight weeks after immune cell reconstitution, the percentage of Treg cells still showed a >2-fold increase in T cells derived from T-Traf3 BM compared to those derived from WT BM (), indicating that the increased Treg cell number in T-Traf3 mice is a cell-intrinsic effect. Additionally, T-Traf3 BM was transduced with control or TRAF3-expressing retroviruses, and used to produce BM chimeric mice. In these mice, TRAF3 over-expression drastically reduced the percentage of Treg cells compared to mice whose T cells were derived from T-Traf3 BM transduced with empty vector (). Moreover, in another T cell-specific TRAF3 deficient mouse strain, (LckCreTraf3flox/flox) mice, the percentage of Treg cells was also significantly increased (). These results indicate that TRAF3 is required for restraining Treg cell development in a cell-intrinsic manner.

TRAF3 deficiency and Treg cell properties

Treg cells exhibit several features that distinguish them from other T cell subsets. We thus explored whether TRAF3 deficiency affects the expression of signature proteins and functions of Treg cells. Expression of Foxp3, CTLA4, CD25, CD122, and GITR proteins were comparable or showed only slight differences in Treg cells from LMC vs. T-Traf3 mice (). The stability of Foxp3 expression upon in vitro TCR stimulation was similar to that seen in LMC Treg cells (). In addition, LMC and Traf3 Treg cells from splenocytes have similar baseline amounts of apoptosis, and these cells underwent apoptosis at the same rate when stimulated with anti-CD3 and anti-CD28 Abs in vitro (). To further explore whether TRAF3-deficient Treg cells display enhanced survival in vivo, splenic WT Treg cells (CD45.1+) were mixed with Traf3 Treg cells (CD45.2+) at a 1:1 ratio and transferred into WT (CD45.1+CD45.2+) mice. 3 weeks later, the ratio of transferred WT and Traf3 Treg cells was still 1:1 in the recipients’ spleens (). This result indicates that TRAF3 deficiency does not detectably alter Treg cell longevity.

The most important function of Treg cells is inhibition of immune responses. An in vitro suppressive assay showed that both LMC and Traf3–/– Treg cells efficiently suppressed the proliferation of conventional CD4+ T cells upon TCR stimulation (). We also found that both LMC and TRAF3-deficient Treg cells similarly suppressed the development of inflammatory bowel disease in a mouse model in vivo in Rag1–/– mice (). Thus, TRAF3 deficiency does not alter the basic biological properties of Treg cells.

Thymic origin of increased Treg cells

Treg cells are derived either from the thymus (tTreg) or arise in the periphery (pTreg). tTreg cells are a relatively stable population with sustained homeostasis, but pTreg cells are less stable and their number varies according to environmental stimuli[25,26]. Ki67 staining and BrDU incorporation to measure cell turnover showed that the frequency of Ki67+ and BrDU+ Treg cells was comparable in LMC and T- Traf3 mice (), indicating that TRAF3 deficiency does not alter Treg cell homeostatic proliferation. Foxp3 expression in tTreg cells is relatively stable, due to demethylation of specific conserved DNA regions[27]. Treg cells in LMC and T-Traf3 mice displayed similar demethylation status at conserved noncoding sequence 2 (CNS2) whereas as expected CNS2 in CD4+CD25– T cells was fully methylated (). Neuropilin-1 (Nrp1) and Helios have both been suggested as possible markers of tT cells[28-30]. Splenic Treg cells in LMC and T-Traf3 mice expressed indistinguishable levels of Nrp1 and Helios (). These results indicate that the increased Treg cell population in T-Traf3 mice is stable and thymus-derived.

NIK independent Treg cell development in T-Traf3 mice

TRAF3 deficiency in T cells results in enhanced basal activation of the non-canonical NF-κB2 pathway[22]. We thus investigated the importance of this pathway in the expanded Treg cell population in T-Traf3 mice, by breeding these mice with NF-κB inducing kinase deficient (Nik–/–, also known as Map3k14) mice, which lack NIK in all cell types. NIK deficiency substantially reduced Treg cell frequency in both Nik–/– and T-Traf3–/– mice, and the difference in Treg cell percentage between these mice disappeared (). This result seemed consistent with the hypothesis that enhanced NF-κB2 activation accounts for the increased Treg cell number in T-Traf3 mice. However, it has been shown that it is instead NIK deficiency in thymic accessory cells, in particular dendritic cells (DCs), that accounts for defective development of Treg cells and other CD4+ T cell lineages in Nik–/– mice, whereas T cell NIK deficiency is not involved[31,32]. Thus, we mixed Rag1–/– BM with Nik–/–, T-Traf3 or Nik–/– T-Traf3 BM at a 10:1 ratio and transferred this mixed BM into irradiated Rag1–/– mice. In this model, the majority of DCs are derived from Rag1–/– mice, which are NIK sufficient, and thymic epithelial cells are also NIK sufficient. We found that the percentage of thymic Treg cells was significantly increased in recipients of Rag1–/– :Nik–/– BM compared to recipients of Nik–/– BM mice, and was further increased in mice receiving Rag1–/– :Nik–/– T-Traf3 BM, to levels comparable to those seen in mice reconstituted with T-Traf3 BM (). Taken together, these results indicate that the elevated basal T cell NF-κB2 activation in T-Traf3 mice does not suffice to explain their increased Treg cell number.

Thymic selection in T-Traf3mice

Early TCR signaling events in mature T cells lacking TRAF3 are markedly reduced[22]. To explore whether thymic selection is also impacted, and could account for the increased Treg cells in T-Traf3 mice, positive and negative selection of thymocytes was examined. Frequencies and numbers of thymocyte populations were comparable between LMC and T-Traf3 mice (). No differences in expression of TCR-β, CD5 and CD69 were seen when comparing double positive (DP) thymocytes from LMC and T-Traf3 mice (). In addition, the frequency and numbers of DP and CD4 single positive (CD4SP) cells were similar between LMC OTII and T-Traf3 OTII mice (). The percentage and expression of Vβ5 (TCR β chain of OTII mice) were also indistinguishable ().

To evaluate thymic negative selection, LMC OTII and T-Traf3 OTII DP thymocytes were stimulated with OTII peptide-pulsed APC in vitro and stained for Annexin V and propidium iodide (PI). Similar numbers of these cells undergoing apoptosis were observed, irrespective of TRAF3 status (). Additionally, expression of Nur77, a marker of TCR signaling strength[33], was not different in Treg cells or Treg precursors from LMC vs. T-Traf3 mice (), indicating that TRAF3 deficiency does not detectably alter TCR signaling in early T cell development.

Our previous report implicates TRAF3 as promoting TCR and CD28-mediated signaling[22]. Given the important role of CD28 signaling in tTreg cell development, T-Traf3–/– mice were bred with CD28 deficient (Cd28–/–) mice, to explore whether CD28 signaling is required for the development of increased Treg cells in T-Traf3–/– mice. Flow cytometric analysis showed that although CD28 deficiency decreased the Treg cell percentage in both LMC and T-Traf3–/– mice, there were still 2-3-fold more Treg cells in Cd28–/–T-Traf3–/– mice than in Cd28–/– LMC mice (), indicating that altered CD28 signaling is not responsible for the increased tTreg cell population in T-Traf3–/– mice.

According to a two-step model of Treg cell development, TCR signaling is required for Treg precursor selection, while signaling induced by common γ chain cytokines, particularly IL-2, is essential for the upregulation of Foxp3[2,3]. Although Treg cell numbers were increased 2-3 fold in the T-Traf3–/– thymus, the percentage of Treg precursors was comparable in LMC and T-Traf3–/– mice (defined as CD4+CD8–Foxp3–CD25+ in , or CD4+CD8–Foxp3–CD25+GITR+ in ). These results further indicate that thymic selection is not affected by the absence of TRAF3 in the T cell compartment. This was also true in Cd28–/– mice (). Thus we conclude that alterations in thymic selection are not responsible for increased Treg cells in T-Traf3–/– mice.

TRAF3 inhibition of Treg precursor to Treg cell transition

Unaltered numbers of Treg precursors, in addition to the normal homeostasis and survival of Traf3–/– Treg cells, prompted us to hypothesize that the transition from Treg precursor to Foxp3+ Treg cell is more efficient in the absence of TRAF3. To address this hypothesis, T-Traf3–/– mice were bred to Foxp3-GFP mice and Treg precursors were sorted for in vitro culture in the presence of IL-2. Results show that transition from Treg precursor to Treg cell with the addition of IL-2 was twice as efficient in the absence of TRAF3 (). Consistent with this finding, blocking CD25 in fetal thymus organ culture (FTOC) resulted in disappearance of the advantage of over-development of Treg cells in T-Traf3–/– thymus (). These results suggested that IL-2R signaling was enhanced in Traf3–/– Treg precursors. Indeed, IL-2-activated phosphorylation of STAT5 in Traf3–/– Treg precursors was markedly elevated in comparison to LMC Treg precursors (), with no discernible change in the MAPK p- Erk or p-Akt (). Additionally, the expression of CD25, CD122 and CD132 in Treg cell precursors was not impacted by the absence of TRAF3 (). Thus, TRAF3 specifically regulates IL-2-induced activation of STAT5 in Treg precursors.

Histone deacetylases (HDACs) are required for induction of STAT5-dependent gene transcription, including that of Foxp3[34]. To investigate whether the observed enhanced phosphorylation of STAT5 is responsible for more efficient Treg cell conversion from precursors in T-Traf3–/– mice, sorted Treg precursors were cultured with IL-2 and HDAC inhibitors Trichostatin A (TSA) or apicidin. These inhibitors completely blocked the transition in both LMC and Traf3–/– Treg precursors (), further indicating the crucial role of enhanced p-STAT5 in the transition process.

Generation of Treg cells in the thymus is characteristically delayed by several days compared to T effector cells[35]. Foxp3+ Treg cells emerged at day 1 after birth in T-Traf3–/– mice and the trend of increased percentages of Treg cells was observed from this time onward (). The early appearance of Treg cells in the thymus is similar to findings in a mouse expressing transgenic STAT5 that is constitutively active at a low level[2], indicating that enhanced IL-2 signaling can accelerate Treg cell development. TRAF3 deficiency thus results in enhanced IL-2R signaling in Treg precursors, and this leads to increased efficiency of the transition from precursor to Foxp3+ Treg cells.

TRAF3-mediated restraint of IL-2R signaling in Tcon cells

IL-2 signaling is essential for the homeostasis and maintenance of Treg cells. Data presented here show that homeostatic proliferation and survival of Traf3–/– Treg cells were indistinguishable from LMC Treg cells (). In contrast to findings of enhanced IL-2R signaling in Treg precursors, p-STAT5 in TRAF3-deficient mature Treg cells was at most slightly increased compared to LMC Treg cells (). Thus, TRAF3 appears to play different roles in IL-2R signaling in Treg precursors vs. Treg cells. It was thus of interest to determine whether TRAF3 also affects IL-2R signaling in conventional CD4+CD25– T cells. Phosphorylation of STAT5, Jak1 and Jak3 was robustly elevated in Traf3–/– Tcon cells, but phosphorylation of Erk and Akt was not, similar to the pattern observed in Treg precursors (). Consistent with enhanced IL-2 signaling, increased STAT5 association with two of its known target genes, IL-2Rα chain and Cytokine inducible SH2-containing protein (Cis) was detected in TRAF3 deficient T cells by chromatin-immunoprecipitation (ChIP) assay (). The difference in IL-2 signaling in Treg cells and Tcon cells was further validated with sorted CD4+Foxp3GFP+ and CD4+Foxp3GFP– splenocytes (). These results indicate that TRAF3 differentially regulates IL-2R signaling in Treg cells, Treg precursors and Tcon cells, principally by altering Jak-STAT5 signaling pathways. Consistent with this finding, protein expression of TRAF3 in Treg cells are much lower than in Tcon cells (). Validating findings in mouse T cells, STAT5 phosphorylation was also augmented upon IL-2 stimulation in normal human CD4+ T cells, when TRAF3 was depleted with siRNA (). Collectively, these results show TRAF3 is intrinsically involved in IL-2R signaling.

Interaction between TRAF3 and TCPTP in IL-2R signaling

IL-2 binding induces oligomerization of IL-2R components and recruitment of Jak1 and Jak3. This receptor complex initiates activation of downstream signaling pathways[5]. The elevated phosphorylation of Jak1 and Jak3 in TRAF3-deficient T cells suggests that TRAF3 exerts regulatory effects upon the IL-2R complex. Immunoprecipitation of Jak3 showed that TRAF3 interacts with Jak3 upon IL-2 stimulation (). Jak1 and Jak3 were also detected in TRAF3 immunoprecipitates from stimulated T cells (). In human CD4+ T cells, we also observed interaction between TRAF3 and Jak3 (). Thus, TRAF3 was recruited to the IL-2R complex upon IL-2 signaling in both mouse and human T cells. It has been reported that the phosphatase TCPTP associates with Jak1 and Jak3 to negatively regulate IL-2 and IFN-γ signaling[14], which was also demonstrated to be true in human T cells (). To explore whether TRAF3 affects the binding of TCPTP to the IL-2R complex, Jak3 was immunoprecipitated from both LMC and Traf3–/– Tcon cells. TCPTP clearly interacted with Jak3 in LMC CD4+ T cells, but the association was negligible in the absence of TRAF3 (). Notably, when CD122 was immunoprecipitated, recruitment of Jak1, Jak3 and SHP-1 to the IL-2R was not significantly changed in Traf3–/– T cells compared to LMC T cells (), indicating that TRAF3 deficiency specifically affects recruitment of TCPTP to the IL-2R complex. There are two splice variants of TCPTP. The nuclear form (45kD) can access both nuclear and cytoplasmic substrates, and interact with Jak1 and Jak3 to dephosphorylate them in response to IL-2 or IFN-γ stimulation[14]. An alternative cytoplasmic form (48kD), unique to human T cells, is targeted to the endoplasmic reticulum[15]. To explore whether TRAF3 can interact with TCPTP, HA-TRAF3, both forms of TCPTP (45kD and 48kD) and their substrate-trapping mutants (D182A) were overexpressed in 293 epithelial cells. In HA immunoprecipitates, only the 45kD form of TCPTP and its substrate-trapping mutant interacted with TRAF3, but not the 48kD form and its mutant (), consistent with the important role of the nuclear form of TCPTP in IL-2 signaling. Further supporting the interaction between TRAF3 and TCPTP, IFN-γ-induced STAT1 phosphorylation was also enhanced in conventional CD4+ T cells in the absence of TRAF3 (). To test which domain of TRAF3 is required for the interaction between TRAF3 and TCPTP, either WT TRAF3, or mutant TRAF3 was co-transfected with the 45kD form of TCPTP into 293 cells. Results show that deletion of both the RING and Zinc finger domains was needed to abrogate TRAF3 binding to TCPTP (). Although TRAF3 has been reported to be an E3 ubiquitin ligase in certain settings, and its RING domain is required for this activity[36], overexpression of ubiquitin, TRAF3 and TCPTP in 293 cells did not result in detectable ubiquitination of TCPTP (). Taken together, these results clearly indicate that interactions between TRAF3 and TCPTP facilitate TCPTP recruitment to the IL-2R complex upon IL-2 stimulation.

Discussion

A multilayered mechanism is required for controlling Treg cell development[1,4,26]. Although the lineage determination from Treg precursors to Foxp3+ Treg cells is limited due to easily saturable niches[37], our results indicate that unrestrained IL-2R signaling can lead to an abnormal accumulation of the tTreg cell population. We demonstrate TRAF3 to be an important factor controlling tTreg cell numbers by negatively regulating IL-2R signals to Treg precursors. TRAF3 was required for TCPTP recruitment to the IL-2R complex, an event that downregulates IL-2 signaling. Downstream of this enhanced signaling in the absence of TRAF3, the transition from Treg precursors to Treg cells was more efficient, resulting in 2-3 fold more tTreg cells in T-Traf3–/– mice.

In T cell-specific TRAF3-deficient mice, TCR signaling in peripheral T cells is impaired, yet it suffices to allow normal development of Tcon cells[22]. Here, we found no evidence of altered thymic selection in T-Traf3–/– mice, ruling out this potential reason for increased Treg cell number in these mice. The comparable percentage of Treg precursors in LMC and T-Traf3–/– mice also supports the conclusion of unaltered thymic selection. Although iNKT cell development in T-Traf3–/– mice is impaired, early stages of iNKT cell development which require TCR signaling are normal, and only later stages are compromised[23]. It thus appears that TRAF3 plays more important roles in TCR signaling in mature T cells than in developmentally immature T cells. Our current findings that IL-2R signaling in Treg precursors, conventional T cells and Treg cells was differentially regulated by TRAF3, further strengthen the concept that TRAF3 plays multifaceted roles in serving different cell types, including distinct developmental stages, as well as distinct receptors within the same cell type[21].

IL-2-induced signaling differs in distinct T cell subsets. Although IL-2 activates S6 kinase in both CD8+ and CD4+ T cells, much higher activation was found in the former[38]. Compared to CD4+ conventional T cells, IL-2 fails to activate the PI3K-Akt pathway in Treg cells, due to high expression of phosphatase and tensin homolog protein (PTEN)[39,40]. Foxp3 expression in Treg cells mediates a unique gene expression profile, which may contribute to regulation of IL-2 signaling[41]. In addition, the constitutive activation of IL-2 signaling in Treg cells due to high expression of CD25 may activate a negative feedback loop[38,39], to alter IL-2 signaling. Consistent with our results that IL-2 signaling and homeostasis of mature Treg cells show little change in the absence of TRAF3, another study found that Treg cell number is only modestly increased in Traf3flox/floxFoxp3GFP-hCre mice in which TRAF3 is depleted from mature T cells[24]. This report further supports our observation that TRAF3 controls Treg cell number specifically at the precursor stage. Decreased expression of TRAF3 in mature Treg cells shown here also suggest its minimal role in this population. As TRAF3 is involved in many TNFRSF-mediated signaling pathways, it might also impact Treg cell features to some extent through these signaling pathways.

IL-2R signaling is initiated by ligand-induced oligomerization of three IL-2R components which recruit Jak1 and Jak3 and ignite a signaling cascade. TCPTP interacts with Jak1 and Jak3 and dephosphorylates them, restraining IL-2R signaling[14]. In the current study, TRAF3 deficiency caused defective recruitment of TCPTP to Jak1 and Jak3 and selectively affected phosphorylation of STAT5, but not Erk and Akt, indicating that these three signaling pathways are differentially regulated. Although overlapping mechanisms also exist, STAT5 can be directly phosphorylated by Jak3, while activation of PI3K-Akt and Ras-Erk requires additional intermediate molecules, e.g. Shc, Lck and Syk[6,9,10]. It is thus not entirely surprising that these pathways are less affected by loss of TRAF3. Although TRAF3 was not previously considered as regulating IL-2R signaling, TRAF6 can compete with Jak1 for binding to the IL-2Rβ chain and negatively regulate Jak1-Erk signaling. However, PI3KAkt and Jak3-STAT5 pathways were not explored[42]. TRAF6 and TRAF3 appear to play different roles in IL-2 signaling, as they do in a variety of other signaling pathways in immune cells.

TCPTP can directly dephosphorylate p-STAT5 in the nucleus upon prolactin stimulation, independent of receptor recruitment[43]. However, the elevated p-Jak1 and p-Jak3 observed here suggest that TRAF3 facilitates TCPTP effects upstream of IL-2-induced STAT5 phosphorylation. Indeed, the recruitment of TCPTP to the IL-2R complex is defective in the absence of TRAF3. The nuclear form of TCPTP plays critical negative roles in cytokine signaling, but the role of the cytoplasmic form of TCPTP is less understood. We found that only the nuclear form of TCPTP can interact with TRAF3, consistent with its role in IL-2R signaling, and both WT TCPTP and its substrate-trapping mutant can equally interact with TRAF3, suggesting that TRAF3 may act as a scaffold molecule to recruit TCPTP to the IL-2R complex, rather than as a TCPTP substrate. Similar to our observations, the nuclear form of TCPTP associates with TRAF2 in TNFR signaling, and specifically regulates TNF-induced MAPK but not NF-κB activation[44]. As nuclear TCPTP translocates to the cytoplasm upon stimulation, it is quite possible that it interacts with TRAF3 or TRAF2 in the cytoplasm. However, the possibility that TRAF3 and/or TRAF2 also play roles in the process of TCPTP translocation cannot be excluded.

The TRAF domain of TRAF3 is thought to be mostly involved in self-association and interaction with upstream receptors. However, the N-terminus of TRAF3, but not the TRAF domain is required for interactions between NIK and TRAF3, although the TRAF domain is indispensable for association with CD40 or BAFFR in B cells[45]. In addition, the N-terminus of TRAF3 is required for interaction between NIK and TRAF3 in 293 epithelial cells[46]. Therefore, the N-terminus of TRAF3 can also be involved in protein-protein interactions. Our findings are consistent with these reports, in that the N-terminus is required for TRAF3 to associate with TCPTP.

Results of the present study emphasize that IL-2R signaling in the transition stage from Treg cell precursors to Foxp3+ Treg cells is the critical checkpoint for controlling Treg cell numbers. Our observations shed light on the molecular regulation of IL-2R signaling and may provide novel avenues for manipulation of Treg cell numbers.

ONLINE METHODS

Mice

Traf3flox/flox mice were previously described[47] and backcrossed with C57BL/6 mice for at least 10 generations. Traf3flox/flox mice were bred with Cd4Cre mice (Taconic Farms, Hudson, NY) and LckCre mice (Jackson Labs, Bar Harbor, ME). CD45.2+ C57BL/6 and congenic CD45.1+ C57BL/6 mice (Jackson Labs) were bred to generate CD45.2 and CD45.1 double positive mice. Cd28–/– mice were purchased from Jackson Labs. Rag1–/– mice were provided by Dr. Fayyaz Sutterwala (University of Iowa, Iowa City, IA). Foxp3GFP mice were originally generated by Dr. Alexander Rudensky (Memorial Sloan-Kettering Cancer Center, New York, NY) and provided by Dr. Thomas J. Waldschmidt (University of Iowa). OTII transgenic mice were provided by Dr. Annette Schlueter (University of Iowa). Nik–/– mice were originally generated by Dr. Robert Schreiber (University of Washington at St. Louis, St. Louis, MS) and provided by Dr. David Parker (Oregon Health Science University, Portland, OR). Mice of 6-12 weeks of age were used for all experiments except when indicated. All mice were maintained under specific pathogen-free conditions at The University of Iowa and were used in accordance with National Institutes of Health guidelines under an animal protocol approved by the Animal Care and Use Committee of the University of Iowa.

Retrovirus transduction and bone marrow chimeras

For virus packaging, the mouse Traf3 gene was cloned and inserted into the retrovirus backbone pMIG. pMIG or pMIG-Traf3 and helper vector pCLECO were co-transfected into 293T epithelial cells using lipofectamine (Invitrogen, Grand Island, NY). Supernatant was harvested after 48hr. Lineage negative BM cells were purified using a Miltenyi kit (Cambridge, MA), and were stimulated overnight with a cytokine combination (IL-6, IL-3, SCF) (Peprotech, Rocky Hill, NJ), and transduced with viral supernatant. Recipients Rag1–/– mice were prepared by sublethal irradiation with 500 rads γ-ray and rested overnight. 0.5×106 transduced BM cells were transferred by i.v. injection. The resulting chimeras were analyzed 8 wks later[48].

Recipient CD45.1+CD45.2+ congenic C57BL/6 mice were given 1100 rads γ-irradiation 16 hr before transfer. BM cells harvested from the tibiae and femurs of WT (CD45.1+) and T-Traf3–/– (CD45.2+) mice were depleted of B220+ and CD3+ cells by Miltenyi magnetic bead separation and mixed at a 1:1 or 20:1 ratio. 10×106 BM cells were injected intravenously into recipient mice. Mice were euthanized 8 wks later for experiments.

In another set of experiments, BM cells isolated from Rag1–/– mice were mixed with BM cells from WT, Nik–/–, T-Traf3–/–, Nik–/– Traf3–/– mice at a 10:1 ratio. 10×106 BM cells were injected intravenously into sublethally irradiated Rag1–/– mice. Mice were euthanized 8 wks later for experiments. Mouse irradiation was conducted at the Free Radical & Radiation Biology Core, University of Iowa.

Flow cytometry

Single-cell suspensions were prepared from thymus, spleen, lymph nodes, liver and Peyer’s Patch, and erythrocytes were lysed. For flow cytometry staining, cells were blocked with anti-mouse CD16/CD32 mAb and stained with fluorescently labeled antibodies against CD3 (145-2C11), CD4 (GK1.5), CD8 (53-6.7), Foxp3 (FJK-16s), B220 (RA3-6B2), CD122 (5H4), CD132(TUGh4), GITR (DTA-1), CTLA4 (UC10-4B9), CD25 (eBio7D4), Ki67 (SolA15), Helios (22F6), Neuropilin (761705), Nur77 (12.14), CD24 (M1/69), CD5 (53-7.3), CD69 (H1.2F3), TCR-β (H57-597), CD45.2 (104) and CD45.1 (A20). For intracellular phosphoprotein staining, cells were fixed immediately after treatment, then permeabilized and stained with fluorescently labeled Abs against surface markers and anti- p-STAT5 (D47E7) or p-Erk (D13.14.4E) or p-Akt (D25E6) Ab, followed by anti-rabbit–APC secondary Ab. All antibodies were purchased from eBioscience (San Diego, CA), BD Biosciences (San Jose, CA), R&D System (Minneapolis, MN) or Cell Signaling Technology (Danvers, MA). For Treg precursor cell sorting, CD8+ T cells were depleted with Miltenyi beads and the remaining cells were stained for CD4, CD8 and CD25. CD4 single positive Foxp3GFP–CD25+ cells were sorted as Treg precursor cells. Flow cytometric analysis and cell sorting were performed using a BD FACS LSRII or Aria at The University of Iowa Flow Cytometry Facility. Results were analyzed with FlowJo software (TreeStar).

Foxp3 expression stability assay

CD4+CD25+ Treg cells were purified from spleen with Miltenyi beads, and cultured in RPMI-1640 (Gibco, Grand Island, NY) supplemented with 10% FBS (Hyclone, Logan, UT), 10 mM HEPES (Sigma, St. Louis, MO), 1 mM sodium pyruvate (Sigma), 50 μM 2-ME (Gibco), 100 U/ml penicillin (Sigma) and 100 μg/ml streptomycin (Sigma). Cells were stimulated with 1ug/ml anti-CD3 Ab and 2ug/ml anti-CD28 Ab (eBioscience) and/or 50U rmIL-2 (Peprotech) for 72 hrs. Frequency and expression level of Foxp3 were measured by flow cytometry.

Treg cell death assay

To measure in vitro cell death, enriched splenic CD4+CD25+ Treg cells were stimulated with 1μg/ml anti-CD3 and 2 μg/ml anti-CD28 Abs. Samples were taken at various time points and stained for Annexin V (BD Bioscience) and Foxp3 and analyzed by flow cytometry. To detect ex vivo Treg cell death, freshly isolated splenocytes were stained for Annexin V, CD4, CD8 and Foxp3 and analyzed by flow cytometry.

Bromodeoxyuridine (BrDU)in vivo incorporation assay

Mice were administered 2 mg BrdU (Sigma) i.p. 24 hr prior to analysis. To detect BrdU incorporation, cells were first stained for Foxp3 followed by intracellular BrdU staining with anti-BrdU Ab (clone PRB-1; eBioscience) using a BrdU flow cytometry detection kit (BD Biosciences) according to the manufacturers’ instructions. Samples were analyzed using flow cytometry.

In vitro Treg cell suppressive assay

CD4+CD25– CD62Lhi T cells (5×104) were isolated from LMC spleen. They were cultured for 72 hrs with irradiated splenocytes (2×105) and anti-CD3 Ab (2μg/ml) in the presence or absence of various numbers of T cells. [3] H-thymidine was added for the last 16 hrs. Cells were harvested and CPM were measured on a β-counter (LS6500, Beckman Coulter, CA, USA).

In vivo Treg cell suppressive assay

Naïve CD4+CD45RBhiCD25– T cells (4×105) were sorted from LMC spleen. They were transferred into Rag1–/– mice alone or in combination with LMC or T-Traf3–/– T cells (4×105). After T cell reconstitution, mice were weighed weekly and monitored for signs of disease. Mice were euthanized at 8 weeks after T cell transfer and their colons were used for histopathology analysis. Histological score was performed according to previous description[49]. Briefly, grade 0, no changes were observed; grade 1, minimal inflammatory infiltrates present in the lamina propria; grade 2, mild inflammation in the lamina propria, minimal to mild mucosal hyperplasia and mucin depletion; grade 3, mild to moderate inflammation in the lamina propria and moderate mucosal hyperplasia and mucin depletion; grade 4, marked inflammatory infiltrates commonly transmural with ulceration, marked mucosal hyperplasia and mucin depletion, and multifocal crypt necrosis; grade 5, marked transmural inflammation with ulceration, wide-spread crypt necrosis and loss of intestinal glands.

In vivo Treg cell survival assay

CD4+CD25+ Treg cells were isolated from spleen in WT (CD45.1+) and T-Traf3–/– mice (CD45.2+), and mixed at a 1:1 ratio. 2×106 cells were transferred into WT recipient mice (CD45.1+CD45.2+) by intravenous injection. 21 days after transfer, splenocytes were harvested for staining and flow cytometry analysis.

In vitro thymic negative selection

T cell- depleted splenocytes were irradiated and pulsed with 10 μg/ml OVA peptide 323-339 (ISQAVHAAHAEINEAGR) (AnaSpec, San Jose, CA). 2×105 LMC and T-Traf3–/– OTII thymocytes were stimulated with 1×106 OTII peptide-pulsed antigen presenting cells for different time point. They were stained for CD4, CD8, Annexin V and propidium iodide (PI) and analyzed by flow cytometry.

Treg cell induction from precursors

Sorted Treg precursor cells (CD4+Foxp3GFP–CD25+) were seeded into 96-well plates in the presence of medium alone or medium containing rmIL-2 (50U/ml). In another set of experiments, cells were treated with 100 nM Trichostatin A (Sigma, St. Louis, MO), 800 nM Apicidin (Sigma, St. Louis, MO), or DMSO (0.2%) and rmIL-22. 24 hrs later, cells were analyzed by flow cytometry.

DNA methylation analysis

CD4+CD25+ Treg cells and CD4+CD25– conventional T cells were sorted from spleen. Only male mice were used. Genomic DNA was isolated and methylation analysis was performed by bisulfite conversion of genomic DNA using the EZ DNA Methylation-Gold™ Kit (Zymo Research, Irvine, CA) following the manufacturer instructions. PCR were performed as previously described[27]. The PCR product was cloned using the TOPO TA Cloning kit (Invitrogen).

TRAF3 and TCPTP depletion by siRNA

Human lymphocytes were obtained from the DeGowin Blood Center at the University of Iowa. Blood from healthy donors aged 18-55 years gave written consent for their blood to be used in research projects, in compliance with the University of Iowa's Institutional Review Board. Human lymphocytes were removed from whole blood using leukocyte reduction cones. CD4+ T cells were purified with Milteyni beads and stimulated with 5μg/ml anti- human CD3 and CD28 Abs (eBioscience) for 24 hrs. Traf3 or Tcptp Trilencer-27 siRNA was transfected into cells with siTran 1.0 (Origene, Rockville, MD), according to the manufacturer's instructions. 20U/ml IL-2 was added during transfection. Cells were transfected again after 24 hrs. After another 24 hr incubation, cells were washed and rested in regular medium without IL-2 twice with a 6 hr interval. They were stimulated with rhIL-2 (Peprotech) for indicated times and Western blots were performed for p-STAT5 detection.

Fetal thymic organ culture

For fetal thymic organ culture, E14-E15 fetal thymuses were harvested, bisected and cultured in trans-well plates. Pairwise comparisons of thymic lobes with 20 μg/ml of isotype control Ab or CD25 blocking Ab (3C7) (ebioscience) were set up. Medium with antibodies was changed on day 5. Single cell suspensions were stained with fluorescently-labeled anti-CD4, anti-CD8, anti-CD25, anti-Foxp3 Abs on day 10 and analyzed by flow cytometry.

Chromatin Immunoprecipitation assay

1×107 CD4+CD25– cells isolated from spleen were stimulated with 500U/ml rmIL-2 for 1 hr. Cells were crosslinked and nuclei were subjected to chromatin shearing according to the manufacturer's instructions (truCHIP Chromatin Shearing Reagent Kit, Covaris, Woburn, MA). STAT5 was immunoprecipitated overnight at 4 °C with anti-STAT5 Ab (Cell Signaling Technology). The precipitated DNA was quantified by real-time PCR to determine the relative abundance of STAT5 associated target DNA. Specific primers for analysis of the chromatin binding to STAT5 are for IL-2Rα chain: 5’-GCATGATATGATGTGCAGTTTCTTC-3’ and 5’-TCAGGACTGGTGGTTGGTTG-3’; and for Cis: 5’-GTCCAAAGCACTAGACGCCTG-3’ and 5’-TTCCCGGAAGCCTCATCTT-3’.

Immunoprecipitation (IP) and Western blots

CD4+CD25– T cells or CD4+CD25+ Treg cells were treated with IL-2 for different time points. Whole cell lysates were separated by SDS-PAGE and transferred to PVDF membranes for Western blots. Alternatively, whole cell lysates were first pre-cleared with magnetic beads, then the relevant Ab was added and incubated for 4 hrs with rotation. Magnetic beads were added and incubated with rotation overnight. IP products were used for Western blots. To measure direct interactions between TRAF3 and TCPTP, the pMIG-HA-Traf3, or pMIG-HA-Traf3 mutant (Δ1-113) (RING domain deleted), or pMIG-HA-Traf3 mutant (Δ 1-258) (both RING and Zinc Finger regions deleted), or pMIG-HA-Traf3 mutant (Δ 382-568) (TRAF domain deleted) plasmid was co-transfected with a TCPTP-encoding plasmid into 293 cells. 48 hrs later, 293 cells were harvested and lysed. Anti-HA Ab (HA-7, Sigma) was used for TRAF3 IP. For TCPTP ubiquitination assay, HA-tagged ubiquitin, FLAG-tagged TRAF3 and TCPTP were transfected into 293 cells. 48 hrs later, cell lysates were denatured by boiling in 1% SDS for 1 min. 1:10 diluted cell lysates were immunoprecipitated with anti-TCPTP Ab. Western blotting was performed and anti-HA antibody was used for ubiquitination detection[36]. Plasmids carrying 45KD and 48KD forms of TCPTP and their 182D/A substrate-trapping mutants were kindly provided by Dr. Nicholas Tonks (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY)[50]. Abs used for IP and Western blots include: Anti-Jak1 (6G4), p-Jak1, Jak3 (D1H3), CD122 (C-2), SHP-1 (C14H6), TCPTP (252294), TCPTP (6F3), STAT5 (3H7), p-STAT5 (D47E7), p-Erk (D13.14.4E), p-Akt (D25E6), β-actin (EP1123Y), p-Y (4G10), FLAG (M2), TRAF3 (M20) and TRAF3 (H122) Abs. They were purchased from Cell Signaling Technology, Santa Cruz Biotech, R&D System, Sigma, EMA Millipore (Billerica, MA), or Medimabs (Montreal, Quebec, Canada). Densitometry analysis was done using ImageJ software (National Institutes of Health) for Western blots.

Statistical analysis

Results are presented as mean values ± SEM. Statistical differences between two means were evaluated using the two-tailed unpaired Student's t-test. For comparisons of multiple groups, two-way ANOVA was used. Statistical significance was set at a p value of <0.05. All values were calculated with Prism software (GraphPad). Sample size was not specifically predetermined, but the number of mice used was consistent with prior experience with similar experiments.