TRAF3 regulates the effector function of regulatory T cells and humoral immune responses

To study the function of TRAF3 in T_reg cells, we generated T_reg-conditional TRAF3 KO mice by crossing the Traf3-floxed mice with Foxp3-Cre mice (Fig. 1 A). Ablation of TRAF3 did not reduce, but even moderately increased, the frequency of T_reg cells in the spleen (Fig. 1 B). To examine the effect of TRAF3 deficiency on the immunosuppressive function of T_reg cells, we performed in vitro suppression assays using T_reg cells purified from the spleen of WT and Traf3^Treg-KO mice. The WT and TRAF3-deficient T_reg cells had comparable efficiencies in suppressing the proliferation of naive T cells (Fig. 1 C), suggesting a dispensable role for TRAF3 in regulating the in vitro function of T_reg cells.

We next examined the in vivo function of T_reg cells using a well-defined adoptive transfer model, which involved cotransfer of CD45RB^hi naive CD4⁺ T cells with T_reg cells purified from the WT and Traf3^Treg-KO mice. As expected, adoptive transfer of CD45RB^hi naive T cells into Rag1-KO mice induced severe body weight loss of the recipient mice, coupled with massive expansion of the transferred T cells, and these pathological phenotypes were efficiently blocked when the naive T cells were cotransferred with WT T_reg cells (Fig. 1 D). Compared with the WT T_reg cells, the TRAF3-deficient T_reg cells exhibited a reduced ability to suppress the expansion and pathological activity of the naive T cells (Fig. 1, E and F). Although the mutant T_reg cells remained fully functional in inhibiting the production of Th17 cells from the cotransferred naive CD4⁺ T cells, their ability to suppress the generation of Th1 cells was partially impaired (Fig. 1 F). After the adoptive transfer, a lower number of TRAF3-deficient T_reg cells was recovered compared with that of the WT T_reg cells (Fig. 1 G), suggesting reduced expansion or survival of the mutant T_reg cells. Furthermore, they showed a reduced level of IL-10 production (Fig. 1 H). Collectively, these data suggest that T_reg cell–specific TRAF3 deficiency does not compromise the homeostasis or in vitro function of T_reg cells but partially impairs the in vivo functions of T_reg cells.

To assess the function of the TRAF3-deficient T_reg cells under nontransferred conditions, we examined the homeostasis of CD4⁺ T cells in peripheral lymphoid organs. The spleens and lymph nodes of the Traf3^Treg-KO mice contained an enhanced frequency of CD4⁺ T cells with memory-like markers (CD44^hiCD62L^lo; Fig. 2, A and B). Among the memory-like CD4⁺ T cells in the peripheral lymphoid organs, there was a significant increase in the frequency of IFN-γ–producing Th1 cells in the Traf3^Treg-KO mice, although the frequency of Th17 cells was generally low and not significantly different between the WT and mutant mice (Fig. 2 C).

T_reg cells play an important role in maintaining T cell homeostasis in the intestine, an organ with dynamic interactions between the immune system and microbiota (MacDonald et al., 2011). The T_reg cell–specific TRAF3 deficiency resulted in a moderate increase in the frequency of CD4⁺ T cells, but not the CD8⁺ T cells, in the lamina propria of the small and large intestines (not depicted). Among the CD4⁺ T cells, there was a significantly increased frequency of Th1 cells in the lamina propria of Traf3^Treg-KO mice (Fig. 2 D). Thus, the loss of TRAF3 in T_reg cells altered the homeostasis of T cells, a finding which was consistent with the partially compromised function of the TRAF3-deficient T_reg cells (Fig. 1, D–F). It is important to note, however, that the overall phenotype of the Traf3^Treg-KO mice in T cell homeostasis was considerably milder compared with that observed in mutant mice lacking Foxp3 or its major regulators that mediate the development and core functions of T_reg cells (Josefowicz et al., 2012). Moreover, the Traf3^Treg-KO mice only displayed relatively mild tissue inflammation at older ages, mostly seen in the lung and the liver (not depicted). These results suggest that TRAF3 may regulate a specific function, instead of mediating the core suppressive activity, of T_reg cells.

In addition to maintaining T cell homeostasis and preventing autoimmune diseases, T_reg cells play an important role in modulating immune responses to foreign antigens. To examine the role of TRAF3 in mediating this latter function of T_reg cells, we analyzed the effect of T_reg cell–specific TRAF3 deficiency on the induction of antibodies by well-characterized protein and nonprotein antigens. Upon immunization of a protein antigen, sheep RBCs (SRBCs), the young Traf3^Treg-KO and WT control mice produced comparable levels of IgM (not depicted). However, the Traf3^Treg-KO mice produced significantly more IgG and its subtypes (Fig. 3 A). Similar results were obtained when the mice were immunized with another protein antigen composed of the hapten 4-hydroxy-3-nitrophenylacetyl (NP) conjugated to the carrier protein KLH (NP-KLH; Fig. 3 B). ELISA assays using NP9-BSA (for high affinity) and NP26-BSA (for global affinity) suggested that the T_reg cell–specific TRAF3 ablation particularly enhanced the production of high-affinity antibodies (Fig. 3 B and not depicted). In response to immunization by a nonprotein antigen, NP-Ficoll, the production of high-affinity IgG, mainly IgG2b, was only moderately enhanced in the Traf3^Treg-KO mice (Fig. 3 C).

We also examined the effect of TRAF3 deficiency on the basal level of antibodies in different ages of mice. At a young age, the WT and Traf3^Treg-KO mice had comparable serum concentrations of antibodies (not depicted). However, older Traf3^Treg-KO mice (18–20 wk old) had an elevated basal level of serum IgG, particularly IgG1 and IgG2b (Fig. 3 D). These results suggest that expression of TRAF3 in T_reg cells is important for controlling the magnitude of humoral immune responses, especially the production of high-affinity IgG antibodies.

To further assess the role of T_reg cell–specific TRAF3 in mediating humoral immune responses, we used a model of influenza infection, which is known to trigger strong T cell–dependent antibody responses (Elsner et al., 2012). As expected, the WT mice experienced bodyweight loss after the intranasal infection with the A/PR8 (H1N1) strain of influenza virus (Fig. 3 E). Importantly, under the same infection conditions, the Traf3^Treg-KO mice had much less bodyweight reduction compared with the WT mice (Fig. 3 E). Moreover, the Traf3^Treg-KO mice also produced significantly higher levels of antiviral IgG antibodies (Fig. 3 F). These results emphasize the biological significance of the TRAF3-mediated T_reg cell regulation.

A hallmark of T cell–dependent antibody responses is the formation of GCs, which are required for various events of the humoral immune response, including antibody class switching, somatic hypermutation, and affinity maturation (Ramiscal and Vinuesa, 2013). Because the T_reg cell–specific TRAF3 deficiency preferentially promoted the production of high-affinity antibodies of a major secondary isotype, IgG, we examined the role of TRAF3 in regulating GC formation by flow cytometry, based on the typical surface markers of GC B cells. As expected, immunization of the WT animals with SRBCs induced production of GC B cells, which peaked around 1 wk and declined around 2 wk after the immunization (Fig. 4 A). Importantly, the Traf3^Treg-KO mice produced a markedly higher frequency of GC B cells than the WT mice (Fig. 4, A and B). Parallel immunofluorescence assays also revealed the formation of enlarged GCs in the spleen of the Traf3^Treg-KO mice after SRBC immunization (Fig. 4 C).

Formation of GCs is associated with antibody somatic hypermutations, a mechanism which generates high-affinity antibodies (Allen et al., 1987; Weiss et al., 1992). Given the enhanced GC formation in Traf3^Treg-KO mice, we analyzed the antibody somatic hypermutation frequency in the immunized WT and Traf3^Treg-KO mice based on the tryptophan (W) 33 to leucine (L) hotspot mutation in the Ig heavy chain variable gene V_H186.2 (Allen et al., 1988). Consistent with the elevated GC formation in the Traf3^Treg-KO mice, these mutant animals had a profoundly higher frequency of W33 to L mutations (Fig. 4 D).

GC formation and antibody class switching are dependent on T_FH subtype of CD4⁺ helper T cells. We found that in response to SRBC immunization, the WT and Traf3^Treg-KO mice produced a comparable frequency of T_FH cells (Fig. 4, E and F). Interestingly, however, the T_FH cells derived from the Traf3^Treg-KO mice displayed hyperexpression of genes encoding several cytokine genes, including IL-4, IL-10, IL-17, and IFN-γ (Fig. 4 G). Furthermore, the expression of several transcription factors, such as Bcl-6, Gata-3, and RORγt, were also elevated in the T_FH cells derived from the immunized Traf3^Treg-KO mice (Fig. 4 G). Of note, T_FH cells are known to produce different effector cytokines depending on the nature of antigens, which appears to influence the antibody class switching in a humoral immune response (Bauquet et al., 2009; Linterman et al., 2012). Several of these cytokines, IL-4, IL-17, and IFN-γ, have been shown to mediate B cell help and induction of antibody class switching toward the generation of different IgG subtypes (Bauquet et al., 2009; Mitsdoerffer et al., 2010; Linterman et al., 2012). Thus, the perturbed T_FH cell activation provided explanations for the aberrant formation of GCs and hyper production of IgG isotypes in Traf3^Treg-KO mice.

Recent studies identified a subset of T_reg cells, the T_FR cells, which controls the function of T_FH cells and GC reactions (Chung et al., 2011; Linterman et al., 2011; Wollenberg et al., 2011). T_FR cells are characterized by expression of the chemokine receptor CXCR5, which mediates their migration into the lymphoid follicles (Linterman et al., 2012). In addition, the T_FR cells also express high levels of the co-inhibitory molecule PD-1 and the transcription factor Bcl-6, respectively exerting negative and positive roles in T_FR regulation (Chung et al., 2011; Linterman et al., 2011; Wollenberg et al., 2011; Sage et al., 2013).

The particularly prominent phenotype of the Traf3^Treg-KO mice in T_FH cell activation and humoral immune responses prompted us to examine the induction and function of T_FR cells. We analyzed the T_FR cells based on their surface expression of CXCR5 and PD-1 within the population of GFP⁺ (expressed from the Foxp3-GFP-hCre transgene) T_reg cells. As expected, the spleens of the WT mice contained a small population of CXCR5⁺PD-1⁺GFP⁺ T_FR cells, and this population was significantly increased in response to SRBC immunization (Fig. 5, A and B). Under unimmunized conditions, the T_FR frequency of the Traf3^Treg-KO mice was only moderately lower than that of the WT mice (Fig. 5, A and B). However, these mutant animals were attenuated in antigen-stimulated production of T_FR cells after SRBC immunization (Fig. 5, A and B). To further confirm this remarkable result, we stained the T_FR cells based on their surface expression of CXCR5 and intracellular expression of Bcl-6 and Foxp3. This method once again revealed a defect of the Traf3^Treg-KO mice in T_FR cell production in response to immunization with SRBCs (Fig. 5, C and D). Thus, although TRAF3 is dispensable for the homeostasis of T_reg cells, it is important for the induction of a specific subset of T_reg cells, the T_FR cells, involved in the control of GC formation and humoral immune responses. This finding explains the aberrant induction of GCs and high-affinity antibodies in the Traf3^Treg-KO mice.

To assess the molecular mechanism by which TRAF3 regulates T_reg cell function, we examined the expression of signature molecules on WT and TRAF3-deficient T_reg cells. Although the TRAF3 deficiency did not affect the expression of most of the T_reg cell surface markers analyzed, it caused an increase in the frequency of CD103^hi T_reg cell populations (Fig. 6 A). The increase in CD103^hi T_reg cells might be caused by their elevated activation, as expression of this molecule is known to be associated with effector/memory type T_reg cells (Huehn et al., 2004). Interestingly, the TRAF3-deficient T_reg cells had a profound reduction in the expression level of ICOS (Fig. 6 A), a co-stimulatory molecule thought to mediate specific T_reg cell functions, such as IL-10 gene induction and T_FR cell development (Ito et al., 2008; Busse et al., 2012; Redpath et al., 2013; Sage et al., 2013). QPCR analysis further confirmed the defect in Icos gene expression (Fig. 6 B). Moreover, the TRAF3-deficient T_reg cells were also impaired in the induction of IL-10 expression (Fig. 6, B–D). These findings further suggest the involvement of TRAF3 in the regulation of specific aspects of T_reg cell functions.

Because Traf3^Treg-KO mice had a defect in T_FR cell induction, we tested whether TRAF3 was required for antigen-stimulated ICOS expression on T_FR cells. As seen with the total T_reg cells, ICOS expression was reduced on T_FR cells under unimmunized conditions (Fig. 6 E). Importantly, the defect of the TRAF3-deficient T_FR cells in ICOS expression became much more drastic after immunization of the mice with the protein antigen SRBCs (Fig. 6 E). Furthermore, this defect appeared to be cell intrinsic because it was not detected in the Foxp3⁻ non-T_reg cells (Fig. 6 F). A mixed bone marrow adoptive transfer experiment revealed that the TRAF3-deficient T_reg cells had reduced ICOS expression even when they were developed in the presence of WT T_reg cells in the same recipient mice (Fig. 6 G), thus further suggesting a cell-intrinsic role for TRAF3 in the regulation of ICOS expression in T_reg cells.

The signaling function of TRAF3 is complex and may vary among cell types (Hildebrand et al., 2011). TRAF3 is known to bind to NIK and mediate NIK degradation, thereby negatively regulating the noncanonical NF-κB signaling (Sun, 2012). Loss of TRAF3 causes accumulation of NIK and constitutive activation of noncanonical NF-κB in B cells and T cells (Liao et al., 2004; He et al., 2006; Xie et al., 2007; Gardam et al., 2008). To examine how TRAF3 regulates ICOS expression, we examined the potential involvement of noncanonical NF-κB activation. We used a transgenic mouse expressing a stabilized form of NIK lacking its TRAF3-binding motif (NIKΔT3) under the control of a loxP-flanked stop cassette (Sasaki et al., 2008). By crossing these mice with the Foxp3-GFP-hCre mice, we generated T_reg-conditional NIKΔT3 transgenic (NIKΔT3^Treg-Tg) mice. As seen with the Traf3^Treg-KO mice, the NIKΔT3^Treg-Tg mice displayed a moderate increase in memory-like effector T cells (Fig. 7 A). This result suggested that deregulated noncanonical NF-κB activation might partially impair the function of T_reg cells. However, the NIKΔT3^Treg-Tg mice did not show obvious abnormalities in antibody responses (Fig. 7 B) or reduced expression of ICOS and IL-10 in T_reg cells (Fig. 7, C and D). In fact, the level of ICOS was even slightly enhanced in these mutant T_reg cells (Fig. 7 C), probably reflecting their abnormal activation. Consistent with this idea, another activation marker, CD103, was also enhanced on the NIKΔT3 T_reg cells (Fig. 7 E).

In conventional T cells, TRAF3 has a role in mediating TCR-stimulated activation of the MAP kinase ERK (Xie et al., 2011). We found that loss of TRAF3 in T_reg cells also inhibited the activation of ERK by the TCR and CD28 signals (Fig. 8 A). To test whether attenuated ERK activation contributed to the reduced expression of ICOS in T_reg cells, we used a selective ERK inhibitor, U0126. Indeed, the ERK inhibitor severely inhibited the induction of ICOS expression (Fig. 8 B). ICOS promoter contains an AP1-binding site that is important for promoter activation by the TCR/CD28 signals (Watanabe et al., 2012). Consistently, we found that cross-linking the TCR and CD28 in WT T cells led to rapid activation of AP1 that bound to the ICOS AP1 probe (Fig. 8 C). Moreover, the TRAF3 deficiency attenuated the activation of AP1 (Fig. 8 C). These findings suggest that TRAF3 regulates ICOS gene induction via an ERK–AP1 signaling pathway.

A recent study suggests that Icos-KO mice have a defect in T_FR cell induction, although it is unclear whether this function is cell intrinsic (Sage et al., 2013). We used a T_reg cell adoptive transfer model to examine whether the expression of ICOS in T_reg cells is required for the development and function of T_FR cells. We adoptively transferred the T cell–deficient Tcrb/Tcrd double KO (Tcrb/Tcrd-dKO) mice with WT naive CD4⁺ T cells together with T_reg cells purified from WT, Traf3^Treg-KO, or Icos-KO mice and then challenged the mice with SRBC antigen. The different T_reg cells were highly similar in homeostasis after the adoptive transfer (Fig. 9 A). In contrast, although the WT T_reg cells efficiently suppressed the induction of GCs in the immunized recipient mice, the TRAF3-deficient T_reg cells were largely defective in this function (Fig. 9 B). The similar deficiency was observed with the ICOS-deficient T_reg cells (Fig. 9 B).

We next examined the production of high-affinity IgG antibodies in the immunized recipient mice. Consistent with the deficiency of the Traf3-KO and Icos-KO T_reg cells in suppressing the GC formation, they also displayed attenuated function in the control of IgG production compared with the WT T_reg cells (Fig. 9 C). Moreover, the recipient mice of Traf3-KO and Icos-KO T_reg cells contained substantially higher levels of plasma cells (Fig. 9 D) and T_FH cells (Fig. 9 E), coupled with reduced frequency of T_FR cells (Fig. 9 F). Collectively, these data suggest that TRAF3 and ICOS have a T_reg cell–intrinsic role in mediating the generation and function of T_FR cells.

The Foxp3⁺ T_reg cells represent a functionally diverse population composed of subsets that regulate different immune functions (Campbell and Koch, 2011; Josefowicz et al., 2012). The signaling mechanism that regulates the functional diversity of T_reg cells is still poorly understood. In the present study, we have identified TRAF3 as a signaling factor that mediates the effector function of T_reg cells, particularly their control of GC formation and antibody responses. Our data suggest that TRAF3 is involved in the induction of the recently defined T_FR cells, a subset of T_reg cells specialized in the control of GC reactions (Chung et al., 2011; Linterman et al., 2011; Wollenberg et al., 2011). Consequently, the mice with T_reg cell–specific TRAF3 deficiency display heightened responses to antigens in the formation of GCs and production of high-affinity antibodies.

A recent study suggests that TRAF3 ablation in total T cells, by crossing the Traf3-floxed mice with CD4-Cre mice, causes an increase in the percentage of Foxp3⁺ T_reg cells (Xie et al., 2011). Because TRAF3 is knocked out during the double-positive stage of thymocyte development in this approach, it is thought that TRAF3 may regulate the development of T_reg cells (Hildebrand et al., 2011). Our present study revealed that ablation of TRAF3 in committed T_reg cells, using the Foxp3-Cre system, also caused a moderate, but significant, increase in the T_reg cell frequency. Our finding is consistent with the previous observation and further suggests that TRAF3 may negatively regulate both the early development and the late-stage maturation or survival of T_reg cells. Interestingly, despite their increased T_reg cell frequency, the Traf3^Treg-KO mice had impaired conventional T cell homeostasis, characterized by an increase in the percentage of CD4⁺ T cells with memory/effector surface markers. This result suggests a positive role for TRAF3 in the regulation of T_reg cell function. In contrast, the phenotype of the Traf3^Treg-KO mice in T cell homeostasis is considerably weaker than that observed in mice lacking Foxp3 or other factors required for the development and core functions of T_reg cells. The Traf3^Treg-KO mice also had only moderate immune cell infiltrations in the nonlymphoid tissues, particularly the lung. Consistently, adoptive transfer experiments revealed that the TRAF3-deficient T_reg cells had weakly reduced ability to suppress the expansion and pathogenesis of CD4⁺ T cells. We observed reduced expression of IL-10 in the TRAF3-deficient T_reg cells, which may contribute to their partially impaired function in maintaining immune homeostasis.

Despite the relatively weak phenotype in immune homeostasis, the Traf3^Treg-KO mice had a prominent abnormality in antigen-stimulated GC reactions. In response to T-dependent antigens, these mutant animals formed enlarged GCs in the spleen and produced markedly more GC B cells, coupled with hyper production of high-affinity IgG subtypes. Gene expression analyses suggest the aberrant activation of T_FH cells. Of particular interest is the elevated production of various effector cytokines, including IL-4, IL-10, IL-17, and IFN-γ, by the T_FH cells derived from the immunized Traf3^Treg-KO mice. Production of these cytokines by normal T_FH cells has been observed previously and thought to regulate B cell proliferation and antibody class switching (Rousset et al., 1992; King et al., 2008; Bauquet et al., 2009; Mitsdoerffer et al., 2010). Thus, deregulated production of these effector cytokines by T_FH cells may contribute to the heightened GC formation and antibody responses in the Traf3^Treg-KO mice. The aberrant T_FH cell activation was apparently caused by the defect of these mutant animals in producing T_FR cells. Upon immunization with a T-dependent antigen, the WT mice had a substantial increase in the T_FR cells, whereas the Traf3^Treg-KO mice were severely attenuated in this response. In an adoptive transfer model, the TRAF3-deficient T_reg cells also displayed a severe defect in generating T_FR cells and suppressing GC responses. Under the transfer conditions, the TRAF3-deficient T_reg cells also had significantly reduced ability to suppress the production of T_FH cells. These results suggest that TRAF3 may regulate a signaling pathway in T_reg cells that is particularly important for their conversion into T_FR cells during an immune response.

The generation of T_FR cells from Foxp3⁺ T_reg cells is critically dependent on the transcription factor Bcl-6 (Chung et al., 2011; Linterman et al., 2011; Wollenberg et al., 2011). Our data suggested that TRAF3 is not required for expression of Bcl-6 or the various T_reg cell signature molecules. Instead, TRAF3 appeared to be required for maintaining the high-level expression of ICOS in T_reg cells because the TRAF3 deficiency greatly reduced, although did not completely block, the expression of ICOS at both RNA and protein levels. By using an adoptive transfer approach, we obtained in vivo evidence that ICOS expression in T_reg cells is crucial for T_reg cell–mediated suppression of GC formation and antibody responses. Like the TRAF3-deficient T_reg cells, the ICOS-deficient T_reg cells were severely attenuated in the generation of T_FR cells and suppression of GC induction in response to immunization. These findings provide a mechanistic insight into the function of TRAF3 in regulating T_FR cell production and antibody responses. We believe that ICOS induction is an important, but probably not the only, mechanism by which TRAF3 regulates T_FR cell induction and antibody responses.

The signaling function of TRAF3 is complex and differs in different cell types and in the context of different receptors (Hildebrand et al., 2011). One important function of TRAF3 is to mediate degradation of NIK and, thereby, negatively regulate the noncanonical NF-κB signaling (Liao et al., 2004). This negative signaling function of TRAF3 is particularly important for maintaining normal homeostasis and function of B cells, although its role in other cell types remains unclear (He et al., 2006; Xie et al., 2007; Gardam et al., 2008). Our data suggest that the activation of noncanonical NF-κB might contribute to the moderate defect of the TRAF3-deficient T_reg cells in maintaining CD4⁺ T cell homeostasis because perturbed CD4⁺ T cell homeostasis was also detected in mice with conditional expression of stable form of NIK (NIKΔT3^Treg-Tg mice). However, the noncanonical NF-κB activation is unlikely responsible for the functional defect of the TRAF3-deficient T_reg cells in the regulation of GC reactions. Unlike the TRAF3 deficiency, conditional expression of NIKΔT3 in T_reg cells did not cause hyperproduction of IgG or other antibody isotypes. The NIKΔT3-expressing T_reg cells did not show any defect in ICOS expression either. Instead, our data suggest a positive role for TRAF3 in mediating the TCR signaling in T_reg cells. As seen in conventional T cells (Xie et al., 2011), the loss of TRAF3 in T_reg cells resulted in attenuated activation of the MAP kinase ERK. We obtained evidence that ERK is important for TCR-stimulated ICOS expression in T_reg cells.

In summary, we have identified TRAF3 as a signaling factor that mediates the effector function of T_reg cells, particularly in the control of humoral immune responses. Although TRAF3 is dispensable for the homeostasis of T_reg cells, it is crucial for antigen-stimulated production of T_FR cells. Our data suggest that the T_reg cell regulatory function of TRAF3 involves induction of the co-stimulatory molecule ICOS, which is the result of impaired ERK activation. Our study has also revealed the requirement of T_reg cell–specific ICOS in the induction of T_FR cells and the control GC responses. These findings not only establish a novel role of TRAF3 in the regulation of T_reg cell function and humoral immune responses, but also shed new light onto the signaling mechanism regulating the specific functions of T_reg cells.

Foxp3-GFP-hCre BAC transgenic mice (Zhou et al., 2008; The Jackson Laboratory) were backcrossed for nine generations to the C57BL/6 background and then crossed with Traf3-floxed mice (C57BL/6 background; Gardam et al., 2008) to produce age-matched Traf3^+/+Foxp3^GFP-hCre (termed WT) and Traf3^fl/flFoxp3^GFP-hCre (termed Traf3^Treg-KO) mice. In some experiments, these mice were further crossed with the Rosa26-YFP Cre reporter (termed R26^YFP; C57BL/6J background; The Jackson Laboratory) to generate WT and Traf3^Treg-KO mice that express YFP in their T_reg cells (termed WT-R26^YFP and Traf3^Treg-KOR26^YFP mice, respectively). The Traf3-floxed mice were also crossed with CD4-Cre (C57BL/6 background; The Jackson Laboratory) mice to produce mice harboring TRAF3 deletion in total T cells (called Traf3^TKO mice). NIKΔT3 mice (C57BL/6 background; The Jackson Laboratory) express a stabilized form of NIK lacking the TRAF3-binding motif (NIKΔT3) from the Rosa26 locus under the control of a loxP-flanked stop cassette (Sasaki et al., 2008). These mice were crossed with the Foxp3-GFP-hCre mice (backcrossed to C57BL/6 background) to generate T_reg-conditional NIKΔT3 transgenic (NIKΔT3^Treg-Tg) and internal WT control mice. B6.SJL (expressing the CD45.1 congenic marker), Tcrb/Tcrd-dKO, Icos-KO, and Rag1-KO mice (all in C57BL/6 background) were obtained from the Jackson Laboratory. For all experiments, WT littermates were used as controls. Mice were maintained in a specific pathogen–free facility of the University of Texas MD Anderson Cancer Center, and all animal experiments were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of the University of Texas MD Anderson Cancer Center.

Mononuclear cells were isolated from the spleen and intestinal lamina propria as previously described (Reiley et al., 2006; Chang et al., 2012). T_reg cells were purified by flow cytometric cell sorting based on their expression of CD25 and the YFP marker from WT-R26^YFP or Traf3^Treg-KO-R26^YFP mice. Flow cytometric analyses and cell sorting were performed (Reiley et al., 2006) using a FACSCalibur (BD) and FACSAria (BD), respectively. For intracellular cytokine staining (ICS), cells were stimulated with 50 ng/ml PMA plus 750 ng/ml ionomycin for 4–6 h in the presence of 10 µg/ml monensin, and the fixed cells were incubated with the indicated antibodies and subjected to flow cytometry. FITC-conjugated anti-CD95 (BD); PE-conjugated anti–PD-1, anti-CD138, anti–Bcl-6, anti-CTLA4, and anti-CXCR3 (BD); allophycocyanin (APC)-conjugated anti-GL7 and anti-CXCR5 (BD Bioscience); and PE-Cy7–conjugated anti-CD8 and anti-CD4 (eBioscience) were used. Other antibodies were described previously (Chang et al., 2012).

Age-matched WT and Traf3^Treg-KO mice (6–8 wk old) were immunized i.p. with 2 × 10⁹ SRBCs (Cocalico Biologicals), 100 µl NP-KLH (1 mg/ml), or 100 µl NP-Ficoll (1 mg/ml) and sacrificed at the indicated times for analyzing serum antibody concentration and GC formation.

WT and Traf3^Treg-KO mice were subjected to intranasal infection with a nonlethal dose (25 tissue culture ID₅₀ units in 25 µl PBS per mouse) of influenza virus A/PR8 (H1N1) under isoflurane anesthesia. Body weight change was monitored every another day for 13 d. On day 13, influenza-specific antibodies in the serum were detected by ELSIA as described previously (Elsner et al., 2012). In brief, ELISA plates were coated with influenza virus overnight at 4°C, blocked with 1% BSA in PBS, and then loaded with serum samples prediluted in 1% BSA. After 3 h of incubation, the plates were washed three times with PBST buffer, and the bound antibody was detected by with biotinylated goat anti–mouse antibody and streptavidin-HRP using TMB as substrate.

Organs were removed from sacrificed mice, fixed in 10% neutral buffered formalin, embedded in paraffin, and sectioned for hematoxylin-eosin staining. For GC formation analysis, SRBC-immunized mice were sacrificed after 10 d of immunization. Frozen spleen sections were fixed in methanol/acetone (2:1) for 10 min, followed by blocking for 30 min in a biotin- and serum-free SuperBlock buffer (Thermo Fisher Scientific). The GCs were detected using biotinylated peanut agglutinin (Vector Laboratories) followed by Alexa Fluor 594–conjugated streptavidin (Molecular Probes). The sections were mounted using an antiquenching agent (Vectashield; Vector Laboratories) and analyzed under a fluorescent microscope (DP70; Olympus). Panoramic pictures were taken and the GCs counted. The area and perimeter of each GC from each mouse (n = 3) were obtained using ImageJ software (National Institutes of Health). The area and perimeter are expressed in micrometers. GC B cells were analyzed by flow cytometry after staining with FITC-conjugated anti-CD95, PE-conjugated anti-B220, and APC-conjugated anti-GL7, and plasma B cells were analyzed similarly by staining with FITC-conjugated anti-B220 and PE-conjugated anti-CD138. For detecting T_FH cells, the cells were stained with FITC-conjugated anti-CD4, PE-conjugated PD-1, and APC-conjugated CXCR5, followed by permeabilization with a Foxp3 staining kit (eBioscience) and intracellular staining with PE-conjugated anti–Bcl-6. T_FH cells were detected similarly except the inclusion of intracellular staining with FITC-conjugated anti-Foxp3 or the GFP marker.

For SRBC or NP-specific antibody detection, Nunc Immuno Plate MaxiSorp plates (Thermo Fisher Scientific) were coated with SRBC lysate, NP9-BSA, or NP26-BSA and incubated overnight at 4°C. After blocking with 1% BSA in PBS, the plates were incubated with serially diluted standard or samples and then with HRP-conjugated detection antibodies. After colorimetric reaction (tetramethylbenzidine; Moss), the absorbance at 450 nm was measured using an ELISA plate reader. For IL-10 detection, Nunc plates were coated with 1 µg/ml anti–IL-10 capture antibody in 50 mM sodium bicarbonate and incubated overnight at 4°C. The coated plates were incubated with supernatant of the T_reg cell culture and subjected to colorimetric detection as described above.

Purified T_reg and CD4⁺ T cells were lysed in RIPA buffer supplemented with inhibitors of phosphatases and the proteases. For analyzing ERK phosphorylation, the cells were stimulated with agonistic anti-CD3 and anti-CD28 antibodies using a cross-linking method (Reiley et al., 2007) before lysis. The cell lysates were fractionated by SDS-PAGE and subjected to immunoblot assays with the indicated antibodies. For analyzing RNA expression, total cellular RNA was isolated and subjected to real-time quantitative RT-PCR assays as previously described (Chang et al., 2011). The gene-specific primer sets are listed in Table S1.

Nuclear extract preparation and EMSA were performed as described previously (Ganchi et al., 1992) using ³²P-radiolabeled oligonucleotide probes. The probes used were mouse ICOS AP1, 5′-TTCATCCATCTAGTCATTCATTTACGCATC-3′; and NF-Y, 5′-AAGAGATTAACCAATCACGTACGGTCT-3′.

The in vitro T_reg cell suppression assay was performed as previously described (Chang et al., 2012). In vivo T_reg cell assays were performed essentially as described previously (Chang et al., 2012). For examining the in vivo roles of T_reg cells in the regulation of GC reactions, sorted CD4⁺CD62L^hiCD44^loCD25⁻ naive T cells from B6.SJL mice (5–6 wk old) and sorted CD4⁺GFP⁺CD25⁺ T_reg cells from WT, Traf3^Treg-KO, and Icos-KO mice (5–6 wk old) were intravenously transferred into Tcrb/Tcrd-dKO mice. Recipient mice were immunized with SRBC 2 d after adoptive transfer and sacrificed 14 d later for analyzing serum antibodies and lymphocyte populations.

WT and Traf3^Treg-KO mice were immunized with SRBCs and sacrificed after 10 d for isolating GC B cells by flow cytometric cell sorting based on the B220⁺CD95⁺GL7⁺ surface markers. Somatic hypermutation was analyzed based on the W33 to L hotspot mutation, essentially as previously described (Muramatsu et al., 2000). In brief, V_H186.2 transcripts were amplified by PCR using the V_H186.2 (5′-GCTGTATCATGCTCTCTTG-3′) and CHγ-1 (5′-GGTGCTATCCGGTCCCTGAGT-3′) primer pairs, and the PCR products were inserted into the pGEM-T vector by TA cloning. Randomly picked cDNA clones were sequenced for identification of the W33 to L mutations.

Prism software (GraphPad Software) was used for two-tailed unpaired Student’s t tests. P-values of <0.05 or <0.01 were considered significant and very significant, respectively.

Table S1 lists the gene-specific primers used for real-time RT-PCR analyses.

We thank the Centenary Institute of Cancer Medicine and Cell Biology for TRAF3-floxed mice. We also thank the personnel from the Flow Cytometry, DNA Analysis, Histology, and Genetically Engineered Mouse core facilities at the University of Texas MD Anderson Cancer Center for their technical assistance.

This study was supported by grants from National Institutes of Health (AI057555, AI064639, GM84459, and AI104519 to S.-C. Sun and CA131207 to S.E. Ullrich) and partially supported by a Sister Institution Network Fund and a seed fund from the Center for Inflammation and Cancer at the University of Texas MD Anderson Cancer Center.

The authors have no conflicting financial interests.

Author contributions: J.-H. Chang and H. Hu designed and performed most of the experiments, prepared the figures, and wrote part of the manuscript; J. Jin performed the histology experiment; N. Puebla-Osorio performed the GC immunofluorescence staining experiment; Y. Xiao provided technical help; R. Brink contributed the Traf3-floxed mice; B.E. Gilbert contributed the influenza virus; S.E. Ullrich was involved in supervision of N. Puebla-Osorio; and S.-C. Sun supervised the work and wrote the manuscript.