SMAD2 Is Essential for TGFβ-mediated Th17 Cell Generation*

TGFβ is the quintessential cytokine of T cell homeostasis. TGFβ signaling is required for the efficient differentiation and maintenance of CD4+FOXP3+ T cells that inhibit immune responses. Conversely, in conjunction with the inflammatory cytokine IL-6, TGFβ promotes Th17 cell differentiation. The mechanism by which TGFβ signals synergize with IL-6 to generate inflammatory versus immunosuppressive T cell subsets is unclear. TGFβ signaling activates receptor SMADs, SMAD2 and SMAD3, which associate with a variety of nuclear factors to regulate gene transcription. Defining relative contributions of distinct SMAD molecules for CD4 T cell differentiation is critical for mapping the versatile intracellular TGFβ-signaling pathways that tailor TGFβ activities to the state of host interaction with pathogens. We show here that SMAD2 is essential for Th17 cell differentiation and that it acts in part by modulating the expression of IL-6R on T cells. Although mice lacking SMAD2 specifically in T cells do not develop spontaneous lymphoproliferative autoimmunity, Smad2-deficient T cells are impaired in their response to TGFβ in vitro and in vivo, and they are more pathogenic than controls when transferred into lymphopenic mice. These results demonstrate that SMAD2 is uniquely essential for TGFβ signaling in CD4+ T effector cell differentiation.

Transforming growth factor ␤ (TGF␤) is an immune modulatory cytokine that is critical for the development and homeostasis of the immune system (1). The absence of TGF␤ activity in mice leads to deregulated, hyperactive T cell activations resulting in early onset fatal multiorgan autoimmunity. Recent studies have highlighted the dominant function of TGF␤ in the differentiation of T cell subsets. TGF␤ can initiate transcription of Foxp3 in naive conventional CD4 ϩ T (Tconv) 2 cells to gen-erate induced regulatory T (iTreg) cells. Moreover, in conjunction with the pro-inflammatory cytokine IL-6, TGF␤ promotes differentiation of Tconv cells to Th17 cells (2)(3)(4) that contribute to the pathogenesis of autoimmune disorders. However, Th17 cells are also necessary for combating infectious diseases such as oral candidiasis and enteritis, caused by Candida albicans and Citrobacter rodentium, respectively (5). Hence, for CD4 ϩ T cells, TGF␤ controls the balance between inflammation and quiescence.
Although extensive progress has been made in understanding the effect of TGF␤ on T cell differentiation and function, detailed characterization of the biochemical basis of TGF␤ signaling in T cells has lagged significantly. In common with other cell types, TGF␤ binds to TGF␤ receptor II on T cells, which leads to the recruitment of TGF␤ receptor I to the complex and activation of cytoplasmic transcription factors SMAD2 and SMAD3. In most cells, activated SMAD2 and SMAD3 translocate to the nucleus in a complex with SMAD4 to regulate target gene transcription (6). Some genes in the TGF␤ pathway are redundantly regulated by SMAD2 and SMAD3, but unique target genes of specific SMADs also exist, suggesting that each possesses distinct functions. Hence, differential activation of SMADs may biochemically account for context-dependent pleiotropic effects of TGF␤.
Using mice deficient in individual SMADs, it has been shown that Smad3 and Smad4 are necessary for efficient TGF␤-mediated generation of FOXP3 ϩ T cells but not for the differentiation of naive CD4 ϩ T cells to the Th17 lineage (7)(8)(9). However, results from a recent study led to the contrasting conclusion that receptor SMADs are not necessary for TGF␤ signaling in T cells (10). Here, analysis of T cellspecific Smad2-deficient mice indicates that in contrast to SMAD3 and SMAD4, SMAD2 plays a non-redundant role in the generation of Th17 cells in vitro and in vivo. The diminution in IL-17 production by CD4 ϩ T cells correlates with accelerated loss of Il6ra expression and a corresponding decrease in STAT3 activation in Smad2-deficient T cells, suggesting that SMAD2 specifically modulates the cross-talk between TGF␤ and IL-6 in Th17 cell differentiation.

EXPERIMENTAL PROCEDURES
Mice-Smad2 fl/fl mice (11) generated in the Robertson laboratory were provided by Dr. Richard Flavell. These mice were backcrossed six times to the C57BL/6 background before analysis. Smad2 conditional KO (CKO) mice were generated using hCD2 Cre Tgϩ mice (C57BL/6 background). Rag1 Ϫ/Ϫ and C57BL/6 mice were purchased from The Jackson Laboratory. For C. rodentium infection, 10 10 cfu of C. rodentium (DBS 100) in 10% sodium bicarbonate was administered by oral gavage. All experiments used mice of 6 -12 weeks of age and were approved by Institutional Animal Care and Use Committee.
Abs, Flow Cytometry, and Cell Sorting-Cells were stained for surface markers, intracellular cytokines, and transcription factors using monoclonal antibodies (mAbs) and intracellular kits purchased from BD Biosciences and eBioscience. Anti-activin RII Ab was purchased from Santa Cruz Biotechnology. Samples were acquired on a BD LSRII cytometer, and data were analyzed using FlowJo software (Treestar). Naive CD4 T cells and natural FOXP3 ϩ Treg (nTreg) cells were sorted to Ͼ95% purity (MoFlow cytometer, Dako Cytomation).

RESULTS AND DISCUSSION
Smad2 Is Not Required for Peripheral T Cell Homeostasis-Mice deficient in TGF␤ or TGF␤ receptor II on T cells succumb to lethal, early onset lymphoproliferation and autoimmunity (1). Smad2 Ϫ/Ϫ mice are embryonic lethal. To determine the function of Smad2 in the maintenance of T cell homeostasis downstream of TGF␤, we generated mice deficient in Smad2 specifically in T cells by crossing CD2 promoter-Cre transgenic (Tg) mice to Smad2 fl/fl mice (referred to as Smad2 CKO mice). In these mice, Smad2 is deleted from the genome of developing T cells at the CD25 ϩ CD4 Ϫ CD8 Ϫ stage of thymic maturation, and Smad2 mRNA is virtually undetectable in peripheral T cells (supplemental Fig. 1A). Smad2 CKO mice are healthy. We observed normal cellularity, T cell subset composition, and T cell phenotypic marker expression in the thymus and peripheral lymphoid organs in young Smad2 CKO mice as compared with Cre Tg controls. A subtle but consistent increase in the frequency and number of nTreg cells was observed in the thymus and spleen of Smad2 CKO animals (supplemental Fig. 1, B and C), although the rate of proliferation, as measured by Ki67 staining, among Treg cells was not altered (sup-plemental Fig. 1B). Smad2 CKO nTreg cells function normally as they can control the colitogenic T cells in lymphopenic Rag1 Ϫ/Ϫ mice (supplemental Fig. 1D) and are able to inhibit Tconv cell proliferation in vitro (data not shown).
Smad2 Regulates CD4 ϩ T Cell Differentiation into iTreg and Th17 Cells-To determine whether TGF␤-activated SMAD2 mediates unique function in CD4 ϩ T cell proliferation and differentiation, we first assayed for TGF␤-mediated suppression of CD4 ϩ Tconv cell division and differentiation to iTreg and Th17 subsets. TGF␤ suppresses Tconv cell proliferation by inducing cell cycle arrest. To assay TGF␤ sensitivity of Smad2 CKO CD4 ϩ Tconv cells, naive CD4 ϩ T cells were labeled with the cell cycle dye CFSE and activated with or without TGF␤. As expected, WT CD4 ϩ T cells showed diminished proliferation in the presence of TGF␤, with the proportion of divided cells (CFSE lo ) decreased by ϳ50% as compared with cultures without TGF␤. In contrast, Smad2 CKO T cells were relatively insensitive to TGF␤, as indicated by the limited difference in the proportion of divided cells with TGF␤ (Fig. 1A). However, when the concentration of TGF␤ in the cultures was increased, Smad2 CKO T cells responded to TGF␤, and proliferation was reduced. These results suggest that there is a dose-dependent impairment in TGF␤ signaling in Smad2 CKO T cells.
CD4 ϩ T cells stimulated with TGF␤ in vitro convert to FOXP3-expressing CD4 ϩ T cells that resemble nTreg cells. We observed that in Smad2 CKO T cells, there is a partial, but significant, decrease in the TGF␤-induced differentiation to FOXP3 ϩ CD4 ϩ T cells (Fig. 1B). Smad3 Ϫ/Ϫ T cells were reported to also have impairment in the FOXP3 induction in vitro (8,9). The extent of impairment in those studies is similar to that observed here with Smad2 CKO T cells, suggesting that each regulatory SMADs contribute to FOXP3 induction in T cells.
In the presence of IL-6 and TGF␤, CD4 ϩ T cells differentiate into Th17 cells. Smad2 CKO T cells were significantly impaired in Th17 lineage differentiation (Fig. 1, C and D), as indicated by decreases in IL-17A ϩ T cells in various culture conditions. Critically, the differentiation of Smad2-deficient CD4 T cells to Th17 lineage was dictated by the concentration of both TGF␤ and IL-6. Smad2 CKO T cells have a more substantial impairment in converting to the Th17 lineage at lower concentrations of TGF␤ and/or IL-6 than at higher concentrations of both of these cytokines (Fig. 1, C and D). This suggests that alterations in both TGF␤ and IL-6 signaling pathways in Smad2 CKO T cells are responsible for the reduced efficiency of Th17 cell generation. Differentiation of Smad2 CKO T cells in the presence of IL-1␤ and IL-6 to Th17 lineage cells (14) was also less efficient than with control T cells (Fig. 1E). IL-2 signaling is inhibitory for IL-17 production by CD4 T cells, and blocking IL-2 enhances Th17 differentiation in vitro (15). However, the defect in Th17 differentiation of Smad2 CKO T cells was not restored by the blockade of IL-2 signaling in Th17-inducing cultures (Fig. 1F).
We next determined whether other cytokines that activate SMAD2 are also impaired in function in T cells from Smad2 CKO mice. Activins are members of the TGF␤ family of cytokines, and activin A has been shown to exhibit a marked pref-erence for SMAD2 activation over SMAD3 in CD4 ϩ T cells (16). Activin receptor II (ActRII) is not expressed on naive CD4 ϩ T cells, but it is up-regulated upon activation (Fig. 1G). Activin A with IL-6 can promote the generation of IL-17-secreting CD4 ϩ T cells without TGF␤ addition, although only ϳ15% of activated Tconv cells detectably express ActRII. Activin-mediated Th17 generation from naive Smad2 CKO CD4 ϩ T cells was also significantly impaired as compared with control CD4 ϩ T cells (Fig. 1G).
Finally, to determine whether IL-17 production is dependent on SMAD2 exclusively downstream of TGF␤ and IL-6, we compared IL-17 production in innate ␥␦T cells that constitute the major early source of IL-17 in vivo. In contrast to ␣␤T cells, IL-17 production in ␥␦T cells does not require TCR signaling or IL-6 but is dependent on TGF␤ and SMAD3 (17). Ex vivo ␥␦T cells from lymph nodes of Smad2 CKO mice were not different from control ␥␦T cells in IL-17A secretion (Fig. 1H), indicating that SMAD2 is dispensable for innate IL-17A production. In sum, SMAD2 is uniquely required to efficiently induce IL-17 in adaptive ␣␤T cells downstream of the TGF␤ family of cytokines and IL-6.
Smad2 Regulates IL-6R␣ Expression and STAT3 Phosphorylation-The finding that Th17 differentiation by IL-6 and TGF␤ is SMAD2-dependent led us to examine the alterations in the IL-6-signaling cascade in Smad2-deficient T cells. It has been reported that TGF␤ up-regulates IL-6R␣ expression in activated CD4 ϩ T cells (18). Further, increased and sustained STAT3 phosphorylation in activated CD4 ϩ T cells was observed in the presence of both TGF␤ and IL-6, as compared with IL-6 alone (19). These results suggest that one function of TGF␤ in promoting Th17 cell generation is to enhance and/or prolong IL-6 signaling in T cells. To investigate whether the IL-6-signaling pathway is altered in Smad2 CKO T cells, we measured the amounts of Il6ra transcripts in stimulated T cells. In our hands, available Abs to IL-6R␣ are not sufficiently sensitive to be useful for flow cytometry, so other methods were used to quantify Il6ra expression. By quantitative PCR analysis, we observed a dramatic down-modulation of Il6ra mRNA expression in activated Smad2 CKO T cells (Fig. 2, A and B), as well as in Th17 culture conditions (data not shown). When Smad2 CKO CD4 ϩ T cells were cultured with IL-6 alone, there was a significant decrease in intracellular phosphorylated STAT3 at early time points (15 and 30 min) as compared with control CD4 ϩ T cells (Fig. 2, C and D; supporting Western blots are in supplemental Fig. 2) in an IL-6 concentration-dependent manner. In contrast to the alteration in IL-6 signaling, no changes in expression of two essential factors of Th17 differentiation, ROR␥t (Fig. 2B) and RUNX1 (data not shown), were observed in Smad2 CKO T cells. These results suggest that Smad2 CKO CD4 ϩ T cells have a decreased capacity to respond to IL-6 and that the synergy between TGF␤ and IL-6 in promoting Th17 differentiation is likely to involve SMAD2 regulation of IL-6R expression. Smad2-deficient Tconv Cells Cause More Severe Colitis-It has been shown that IL-17A is protective during colitis induction, and CD4 ϩ T cells that cannot produce IL-17A cause more aggressive colitis in Rag1 Ϫ/Ϫ recipients (20). To determine the in vivo relevance of the in vitro defects in IL-17 production by Smad2 CKO CD4 ϩ T cells, we assayed whether naive Smad2 CKO CD4 ϩ T cells produce IL-17 when transferred to lymphopenic Rag1 Ϫ/Ϫ recipients. Three weeks after T cell transfer, there was a significant decrease in IL-17A ϩ CD4 ϩ  Mice infected with C. rodentium were sacrificed at 2 weeks after infection. Representative flow cytometric profiles of CD4 ϩ T cells expressing the activation markers in the spleens are shown (two experiments, minimum three mice/ genotype/experiment). E, CD4 ϩ T cells from lymphoid tissues (splenocytes are shown here, but a similar pattern was observed in all tissues) of mock-infected and infected animals were analyzed for the expression of IL-17A and/or IFN␥ by ics. Left, representative flow cytometric profiles; right, averages (n ϭ 5/genotype) with statistics (as in Fig. 1) of the frequencies of CD4 ϩ IL-17A ϩ . UI, uninfected; CB, Citrobacter-infected.
T cells isolated from the peripheral lymphoid organs and colonic lamina propria of Rag1 Ϫ/Ϫ recipients that had been reconstituted with naive Smad2 CKO CD4 ϩ T cells as compared with control CD4 ϩ T cells (Fig. 3, A and B). Interestingly, the decrease in IL-17 production was not consistently observed in Smad2 CKO CD4 ϩ T cells secreting both IL-17A and IFN␥. The decrease in overall IL-17 production by activated Smad2 CKO CD4 ϩ T cells in lymphopenic Rag1 Ϫ/Ϫ mice was correlated with the more severe colitis induced by the transferred Smad2 CKO T cells as revealed by more rapid and severe weight loss in the recipients (Fig. 3C), as well as more severe intestinal damages (data not shown). These results demonstrate that SMAD2 is necessary for normal production of IL-17 by CD4 ϩ T cells in a lymphopenic environment.
C. rodentium Infection in Smad2 CKO Mice Elicits Diminished Th17 Cell Induction-The IL-17 family of cytokines is required for efficient clearance of the gut pathogen C. rodentium. To determine whether pathogen-driven IL-17 production by CD4 ϩ T cells also requires SMAD2, we infected Smad2 CKO mice with C. rodentium. In C57BL/6 mice, the infection reaches maximal pathogen load at 7-9 days, and by 14 days, it is resolved. Ten days after infection, Smad2 CKO mice had comparable numbers of activated T cells in the mesenteric lymph nodes and spleen as WT infected mice ( Fig. 3D and data not shown). However, there was a significant reduction in the frequency of Th17 cells in the lymphoid organs of infected Smad2 CKO mice (spleen shown in Fig. 3E and other tissues showed a similar trend, data not shown). In Smad2 CKO mice, IL-17 ϩ CD4 ϩ T cells accumulated on average to ϳ50% of the numbers seen in control infected mice (Fig. 3E). These results demonstrate that during infection, optimal Th17 cell generation requires SMAD2.
Our results show that SMAD2 is specifically utilized for optimal IL-17 production from CD4 ϩ T cells downstream of the TGF␤ family of cytokines. A similar finding using independently derived T cell-specific Smad2-deficient mice was just published (21). In comparison, Smad3 Ϫ/Ϫ or Smad4 CKO mice do not exhibit significant impairments in Th17 generation (7,9). Smad4-independent Th17 differentiation suggests the possibility that the unique TRIM33 interaction with SMAD2 is central to the process (22). The reduction in Th17 cell differentiation of Smad2 CKO T cells in vitro is most obvious at lower concentrations of TGF␤ and/or IL-6. That TGF␤ dose is a critical determinant in regulatory versus inflammatory T cell subset generation has been proposed (23), and it is likely that under saturating cytokine culture conditions, normal biochemical circuits can be subverted and masked. An essential function of SMAD2 for Th17 cell differentiation leaves open the possibility that selective regulatory SMAD activation may underpin the varied fate of TGF␤-signaled T cells. Further work is required to dissect how and in what in vivo conditions TGF␤-activated SMAD2 intersects with TCR-and IL-6-signaling pathways to dictate Th17 subset differentiation.