Smad2 Positively Regulates the Generation of Th17 Cells*

Development of Foxp3+ regulatory T cells and pro-inflammatory Th17 cells from naive CD4+ T cells requires transforming growth factor-β (TGF-β) signaling. Although Smad4 and Smad3 have been previously shown to regulate Treg cell induction by TGF-β, they are not required in the development of Th17 cells. Thus, how TGF-β regulates Th17 cell differentiation remains unclear. In this study, we found that TGF-β-induced Foxp3 expression was significantly reduced in the absence of Smad2. More importantly, Smad2 deficiency led to reduced Th17 differentiation in vitro and in vivo. In the experimental autoimmune encephalomyelitis model, Smad2 deficiency in T cells significantly ameliorated disease severity and reduced generation of Th17 cells. Furthermore, we found that Smad2 associated with retinoid acid receptor-related orphan receptor-γt (RORγt) and enhanced RORγt-induced Th17 cell generation. These results demonstrate that Smad2 positively regulates the generation of inflammatory Th17 cells.


Development of Foxp3 ؉ regulatory T cells and pro-inflammatory Th17 cells from naive CD4 ؉ T cells requires transforming growth factor-␤ (TGF-␤) signaling. Although Smad4 and
Smad3 have been previously shown to regulate Treg cell induction by TGF-␤, they are not required in the development of Th17 cells. Thus, how TGF-␤ regulates Th17 cell differentiation remains unclear. In this study, we found that TGF-␤-induced Foxp3 expression was significantly reduced in the absence of Smad2. More importantly, Smad2 deficiency led to reduced Th17 differentiation in vitro and in vivo. In the experimental autoimmune encephalomyelitis model, Smad2 deficiency in T cells significantly ameliorated disease severity and reduced generation of Th17 cells. Furthermore, we found that Smad2 associated with retinoid acid receptor-related orphan receptor-␥t (ROR␥t) and enhanced ROR␥t-induced Th17 cell generation. These results demonstrate that Smad2 positively regulates the generation of inflammatory Th17 cells.
TGF-␤, signaling through a heteromeric TGF␤RII and TGF␤RI complex, activates the phosphorylation of Smad2 and Smad3, which associate with the common partner Smad4, and then translocate to the nucleus (4). We have previously shown that TGF-␤ signaling through TGF␤RI is required for generation of both Th17 and iTreg cells (5). Both Smad4 and Smad3 play an important role in the induction of Foxp3 expression upon TGF-␤ stimulation of naive T cells. Although Smad4 is dispensable for Th17 cell generation (5), Smad3 deficiency leads to enhanced Th17 cell development in vitro and in vivo (6). Thus, TGF-␤-signaling mechanisms for Th17 cell differentiation still remain unclear.
In the present study, we have determined the role of Smad2 and found that Smad2 is required for induction of both iTreg cells as well as Th17 cells in vitro. In the experimental autoimmune encephalomyelitis (EAE) model, mice with a deficiency of Smad2 in T cells exhibited reduced disease severity and defective Th17 cell generation. Thus, Smad2 is crucial for the generation of Th17 cells.

EXPERIMENTAL PROCEDURES
Mice-C57BL/6 and OT-II T cell receptor transgenic mice were purchased from The Jackson Laboratory Smad2 floxed mice (7) were bred with CD4-Cre mice (8). Mice were housed in the specific pathogen-free animal facility at the MD Anderson Cancer Center, and the animal experiments were performed at the age of 6 -10 weeks using protocols approved by Institutional Animal Care and Use Committee.
In Vitro Regulatory T Cell Suppression Assay-FACS-sorted naive CD4 ϩ CD25 Ϫ CD62L hi CD44 lo T cells were stimulated in the presence or absence of FACS-sorted CD4 ϩ CD25 hi natural regulatory T cells (nTregs) at different ratios with 3000 Rads irradiated T cell-depleted splenocytes and soluble anti-CD3 (1 g/ml) for 3 days. Proliferation was determined by the addition of [ 3 H]thymidine to the culture for the last 8 h.
KLH Immunization and EAE-Mice were immunized with keyhole limpet hemocyanin (KLH) as described previously (9). EAE induction was performed and analyzed as indicated previously (10). Disease scores from three independent experiments were combined, and p values were calculated using Student's t test by comparing the disease scores.

RESULTS AND DISCUSSION
Smad2 Deficiency Leads to a Partial Reduction in TGF-␤induced Foxp3 Expression-To address the role of Smad2 in T cell differentiation, Smad2 floxed mice (7) were crossed with CD4-Cre transgenic mice (8) to generate mice lacking Smad2 in T cells (Smad2 tKO), which results in the absence of Smad2 but not Smad3 protein in CD4 ϩ and CD8 ϩ T cells but not in B cells (supplemental Fig. 1). Interestingly, T and B cells primarily express Smad2 that is of higher molecular weight than that of Smad3, indicating that it contains exon 3 and thus has defective DNA binding capacity, unlike Smad3 (16,17).
Smad2 tKO mice showed normal numbers of CD4 ϩ and CD8 ϩ T cells in thymus, as well as similar maturation markers as compared with wild-type littermates (supplemental Figs. 2 and 3). Furthermore, similar frequencies and total cell numbers . p values were calculated using Student's t test. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. LN, peripheral lymph nodes; MLN, mesenteric lymph nodes. B, CD4 ϩ CD25 Ϫ CD62L hi CD44 lo naive T cells from Smad2 WT mice were cultured in the presence or absence of different ratios of Smad2 KO or WT CD4 ϩ CD25 ϩ natural regulatory T cells in triplicate wells with irradiated T cell-depleted splenocytes and stimulated with 1 g/ml anti-CD3. Proliferation was assayed 72 h after treatment by adding [ 3 H]thymidine to the culture for the last 8 h. A representative of three independent experiments is shown. C and D, FACS-sorted naive CD4 ϩ CD25 Ϫ CD62L hi CD44 lo T cells from Smad2 fl/fl CD4CreϪ (Smad2 WT) or Smad2 fl/fl CD4Creϩ (Smad2 tKO) mice were stimulated with plate-bound anti-CD3 and anti-CD28 under iTreg conditions (TGF-␤, IL-2, anti-IFN-␥, and anti-IL-4) for 4 days. Foxp3 expression was analyzed by intracellular staining (C). After differentiation, cells were restimulated with anti-CD3 for 4 h, and cDNA was prepared. Gene expression profile was analyzed by real-time RT-PCR (D). Data were normalized to a reference gene Actb. The lower expression level for each gene was referred as 1. A representative of three independent experiments is shown. Error bars in B and D indicate mean Ϯ S.D. of CD4 ϩ and CD8 ϩ T cells were observed in periphery, with a slight decrease in CD8 ϩ T cell frequencies in spleens but not their total numbers (supplemental Fig. 2 and data not shown). Moreover, CD4 ϩ T cells found in peripheral tissues contained more activated/memory cells as compared with wild-type counterparts (supplemental Fig. 4).
Unlike Smad4 tKO or Smad3 KO mutant mice, mice lacking Smad2 in T cells exhibit enhanced CD4 ϩ CD25 ϩ Foxp3 ϩ nTreg cells in all the lymphoid tissues analyzed (Fig. 1A). This phenotype resembles the TGF␤RI conditional KO mice, which also showed enhanced nTreg levels in thymus after 1 week of birth (18). Peripheral Smad2-deficient nTreg cells showed a similar suppressive phenotype as wild-type (WT) nTreg cells in vitro (Fig. 1B).
To analyze the role of Smad2 in Treg induction, FACS-sorted naive CD4 ϩ CD25 Ϫ CD62L hi CD44 lo T cells were activated with plate-bound anti-CD3 and anti-CD28 in the presence of TGF-␤, IL-2, and neutralizing antibodies against IFN-␥ and IL-4. Under this polarizing condition, we found a reduction in Foxp3 expression in cells lacking Smad2 (Fig. 1C). Reduction in Foxp3 expression was detected also at the mRNA level (Fig.  1D). These results suggest that Smad2, as well as Smad3 and Smad4 (5, 6), regulates de novo Foxp3 transcription.
Absence of Smad2 Reduces Th17 Cell Generation in Vitro-We next determined the requirement of Smad2 in the genera-tion of Th17 cells. Naive T cells lacking Smad2 showed reduced IL-17-producing T cells as compared with WT counterparts when increasing TGF-␤ concentrations were utilized to drive Th17 cell differentiation ( Fig. 2A). By real-time RT-PCR, we found that IL-17, IL-17F, and CCL20 mRNA expression was reduced in Smad2-deficient T cells (Fig. 2B). Interestingly, IL-21 or IL-22 levels were found to be similar or even slightly enhanced in Smad2-deficient T cells (Fig. 2B and data not shown), supporting that TGF-␤ does not regulate the expression of these cytokines. Furthermore, no significant differences were observed in expression of Th17-specific transcription factors ROR␥t, ROR␣, AHR (aryl hydrocarbon receptor), IRF4 (interferon regulatory factor 4), BATF (basic leucine zipper transcription factor, ATF-like), and Ib ( Fig. 2B and data not shown). Expression of Th1 and Th2 transcription factors T-bet and GATA3, respectively, was also similar between Smad2 KO and WT Th17 cells (Fig. 2B and data not shown), suggesting that the defect in Th17 cell generation in Smad2deficient cells is not due to enhanced differentiation toward alternative Th cell lineages. Reduced IL-17-producing cells were observed in the absence of Smad2 (supplemental Fig. 5) when Th17 cells were generated using a combination of IL-1, IL-6, and IL-23 cytokines (9).
Smad2 Is Required for Th17 Cell Generation in Vivo-To analyze the requirement for Smad2 in vivo, Smad2 tKO or WT littermate mice were first immunized with KLH emulsified in complete Freund's adjuvant. Seven days after the immunization, spleen and draining lymph nodes were harvested, and cells were restimulated with KLH protein ex vivo to evaluate cytokine production. We found reduced frequencies and total cell numbers of IL-17-producing T cells from Smad2 tKO mice as compared with WT counterparts (supplemental Fig. 6A). Moreover, Smad2-deficient T cells produced significantly lower IL-17 and IL-17F cytokines upon ex vivo restimulation with increasing concentrations of KLH in splenocytes and draining lymph node cells, as measured by ELISA in the culture supernatants (supplemental Fig. 6B and data not shown). On the other hand, no significant differences in IL-22 or IFN-␥ expression were observed between the two groups of mice, suggesting that Smad2 deficiency affects primarily IL-17 and IL-17F cytokines.
Th17 cells have been shown to be important for mediating inflammatory responses and autoimmune diseases (1). Thus, to further understand the function of Smad2 in vivo, we utilized . p values were calculated using Student's t test comparing the disease score within each day. *, p Ͻ 0.05; **, p Ͻ 0.005. B and C, mononuclear cells infiltrating the central nervous system from the EAE mice were isolated on day 14 after the second immunization and restimulated with PMA, ionomycin, and GolgiStop for 6 h. CD4 and CD11b expression was determined (B), and IL-17-or IFN-␥-expressing cells were measured by intracellular staining on a CD4 ϩ T cell gate (C). In B and C, a representative dot plot is shown, and the combined results (frequency or total cell number) for each group are indicated. p values were calculated using Student's t test. *, p Ͻ 0.05, **, p Ͻ 0.005. D, splenocytes from the above mice were stimulated with MOG peptide, and cytokine expression levels were measured by ELISA. Data shown are a representative example of three independent experiments with consistent results. p values were calculated using Student's t test. *, p Ͻ 0.05. the EAE model. Both WT and Smad2 tKO mice showed similar disease onset and disease severity at an early stage (Fig. 3A). However, although Smad2 tKO mice started to recover by day 10, WT littermates showed sustained disease severity (Fig. 3A). At day 14 after the second immunization, we analyzed central nervous system infiltration and found significantly lower frequency of CD4 ϩ T cells in inflamed tissue from Smad2 tKO mice as compared with WT counterparts (Fig. 3B). Moreover, a reduction in the total CD4 ϩ T cell number was also observed (Fig. 3B). We further investigated the cytokine production of those CD4 ϩ T cells infiltrating the central nervous system and found a significant reduction in frequencies and total cell numbers of both IL-17 ϩ and IL-17 ϩ /IFN-␥ ϩ cells (Fig. 3C). However, the frequencies of IFN-␥-producing CD4 ϩ T cells were not affected between the two groups (Fig. 3C), further demonstrating a specific role of Smad2 in Th17 cell generation in vivo. The diminished disease severity in Smad2 tKO mice was not due to enhanced regulatory T cells infiltrating the CNS (data not shown). Moreover, when restimulated ex vivo with MOG peptide, splenocytes showed decreased Th17 cytokine production, albeit similar proliferation (Fig. 3D). Thus, these results showed that Smad2 is required for appropriate Th17 immune responses in vitro and in vivo.
Smad2 Binds to and Synergizes with ROR␥t in Th17 Cell Induction-Because Smad2-deficient T cells showed impaired Th17 cytokine expression but maintained normal levels of both ROR␣ and ROR␥ transcription factors, we considered the possibility that Smad2 may not be required for ROR expression but is important for their function. Given that Smad3 can bind to ROR␥t (6), we next examined whether Smad2 can also associate with ROR␥t. Similar to Smad3, Smad2 but not Smad4 was able to bind ROR␥t when co-expressed in HEK 293T cells, and this binding was increased upon co-expression of a constitutively active form of rat TGF␤RI (TGF␤RI T202D) (Fig. 4A) (6,19). Because both Smad2 and Smad3 showed similar binding to ROR␥t, we next evaluated their affinity to ROR␥t by doing a competitive co-immunoprecipitation experiment. We found that Smad3 competed with Smad2 for binding to ROR␥t and actually inhibited the binding of Smad2 to ROR␥t, whereas Smad2 did not affect Smad3 binding. Thus, these results suggest that Smad3 might indirectly inhibit ROR␥t function by blocking Smad2 binding (supplemental Fig. 7). It has been recently suggested that PP2A differentially regulates Smad2 and Smad3 phosphorylation by directly dephosphorylating Smad3 under hypoxia conditions (20) which suggests that although TGF␤R signaling might induce Smad2 and Smad3 phosphorylation similarly, other proteins might also regulate their phosphorylation status independently. Thus, this differential dephosphorylation of Smad2/3 might be the mechanism utilized in Th17 cells to favor Smad2 over Smad3 phosphorylation.
Next, we investigated whether Smad2 regulates ROR␥t-dependent generation of Th17 cells. For that purpose, ROR␥t and/or constitutively activated Smad2 were overexpressed in T cells by retroviral transduction. Overexpression of ROR␥t but not Smad2 led to the generation of IL-17-producing cells and up-regulation of genes associated with Th17 phenotype (Fig. 4,  B and C). Interestingly, overexpression of constitutively active Smad2 together with ROR␥t greatly enhanced the induction of IL-17-producing cells as compared with ROR␥t expression alone (Fig. 4B). In addition, Smad2 enhanced ROR␥t-dependent IL-17, IL-17F, and CCL20 mRNA expression (Fig. 4C). This effect appears to be unique to Smad2 as a constitutive form FIGURE 4. Smad2 directly binds to ROR␥t and synergizes in the generation of Th17 cells. A, HEK 293 T cells were transiently transfected with 6ϫMyc-tagged Smad3, 6ϫMyc-tagged Smad2, 2ϫMyc-tagged Smad4, FLAG-tagged ROR␥t, and/or His-tagged TGF-␤RI T202D. After 48 h, lysates were prepared and immunoprecipitated with an anti-FLAG mAb (IP) followed by immunoblotting (IB) with anti-FLAG or anti-Myc (bottom two panels). The top three panels indicate Western blot of whole cell lysates (WCL). B and C, FACS-sorted naive OT-II CD4 ϩ T cells were activated with OVA peptide-pulsed splenic APCs under neutral (anti-IL-4, anti-IFN-␥, and anti-TGF-␤) conditions and co-infected with two bicistronic retroviruses (IRES-GFP or IRES-hCD2) expressing ROR␥t-GFP, Smad2-2SD-hCD2, and/or GFP or hCD2 vector controls. After 4 days, GFP ϩ hCD2 ϩ -infected cells were FACS-sorted. B, GFP ϩ hCD2 ϩ cells were stimulated with PMA, ionomycin, and GolgiStop for 4 h, and IL-17-producing cells were determined by intracellular staining. C, GFP ϩ hCD2 ϩ cells were restimulated for 4 h with anti-CD3, and mRNA expression of the indicated genes was analyzed by real-time RT-PCR. The data shown were normalized to expression of a reference gene Actb. The lowest expression of each gene was referred to as 1 and corresponds to naive T cells. The data represent at least three independent experiments with consistent results.
Error bars indicate mean Ϯ S.D. Rel. mRNA levels, relative mRNA levels.
of Smad3 did not regulate IL-17 expression in the presence or absence of ROR␥t in the same system (data not shown). Furthermore, we observed an increase in endogenous ROR␥ expression itself when ROR␥t was overexpressed, and such induction was further enhanced by co-expression of Smad2 (Fig. 4C). However, no differences were observed in other Th17-specific transcription factors such as ROR␣, BATF, and IB ( Fig. 4C and data not shown). Thus, these results suggest that Smad2 might act as a coactivator for ROR␥t, leading to enhanced Th17 cell generation.
In summary, in the present study, we studied mice with deficiency of Smad2 in T cells. We found that Smad2, like Smad4 and Smad3, was partially required for induction of iTreg cells. However, unlike Smad3 or Smad4, Smad2 not only binds to but also synergizes with ROR␥t in the generation of Th17 cells. More importantly, Smad2-deficient T cells had reduced capability to differentiate into Th17 cells, and mice with deficiency of Smad2 in T cells showed reduced Th17 cell responses in vivo and amelioration of EAE disease symptoms. Our results provide a basis for understanding the reciprocal regulation of Th17 and regulatory T cells.