Smad3 Is Essential for TGF-β1 to Suppress IL-2 Production and TCR-Induced Proliferation, but Not IL-2-Induced Proliferation¹

The lethal multiorgan inflammation phenotype in TGF-β1 knockout mice demonstrates that the presence of TGF-β1 is essential for the maintenance of immune homeostasis ( 1, 2). Studies using TGF-β1 receptor type II (TβRII)³ dominant-negative transgenic mice show that TGF-β1 signaling deficiency within cells of hemopoietic origin is sufficient to cause a lethal inflammatory disorder ( 3). CD4⁺ and CD8⁺ T cell-specific TβRII dominant-negative transgenic mice produce a similar phenotype, although less severe than TGF-β1 knockout mice ( 4). CD2 promoter-driven TβRII transgenic mice show predominantly spontaneous activation and hyperproliferation of peripheral CD8⁺ T cells ( 5). These observations illustrate that TGF-β1 signaling of T cells is required for T cell homeostasis.

TGF-β1 initiates signaling by binding to and bringing together type II (TβRII) and type I (TβRI) receptor serine/threonine kinases on the cell surface. Ligand binding enables TβRII to phosphorylate TβRI, which, in turn, propagates signaling through phosphorylation of the effector proteins, including Smads. TGF-β1 signals though four structurally similar Smad proteins, which constitute three functionally distinct classes: the receptor-regulated Smads (R-Smads) (Smad2 and Smad3), the comediator Smad (Smad4), and the inhibitory Smad (Smad7). Smad2 and Smad3 are directly phosphorylated and activated by TβRI and heterodimerize with Smad4. The activated Smad complex then translocates into the nucleus, and, in a cooperative manner with other nuclear cofactors, regulates the transcription of target genes. Activated TβRI is required to maintain continual nucleocytoplasmic shuttling of the Smad proteins essential for TGF-β1 signaling ( 6). Smad7 negatively regulates TGF-β1 signaling by inhibiting TβRI-induced R-Smad phosphorylation and mediating E3 ubiquitination of the TβRI ( 7). The Smad3/Smad4 heterocomplex can bind directly to CAGA-containing Smad-binding elements to regulate gene expression ( 8). Smad2 has been shown to bind to CG-rich regions, but not Smad-binding elements ( 9, 10). Cooperative interaction between Smad3 and transcription factors (e.g., CREB, AP-1, NF-κB) as well as transcriptional coactivators (e.g., CBP/p300 and P/CAF) and corepressors (e.g., TG-interacting factor, c-Ski, and Sno) has been substantiated in numerous TGF-β1-responsive genes ( 11, 12, 13). Loss of Smad3 was found to result in impaired mucosal immunity and a loss of T cell growth inhibition by TGF-β1 ( 14, 15). Embryonic lethality of Smad2 and Smad4 knockout mice exemplifies nonredundant signaling pathways among Smad protein family members. TGF-β1 also signals through Smad-independent pathways, such as those involving the mitogen-activated protein kinases and phosphoinositol 3-kinase ( 16, 17).

Activation of T cells through the TCR results in activation of a cascade of signaling pathways leading to the up-regulation of IL-2 production and IL-2 responsiveness. IL-2 promotes cell growth by activating genes such as c-myc, cyclin D2, cyclin D3, and cyclin E ( 18) and inactivating expression of cyclin-dependent kinase (Cdk) inhibitors, in particular, p27^Kip ( 19, 20). Conversely, the profound negative regulatory effects of IL-2 are illustrated by lymphomegaly associated with polyclonal T and B cell expansion in IL-2-, IL-2Rα-, and IL-2Rβ-deficient mice ( 21). The T cell-mediated autoimmune-like phenotype of these IL-2 signaling-deficient mice is associated with profound defects in the number and function of CD4⁺ CD25⁺ regulatory T cells ( 22, 23, 24, 25).

A balance between stimulatory and inhibitory signals is essential to maintain immune homeostasis. A growing body of evidence has demonstrated that the proliferative effects of IL-2 and TCR stimulation are opposed by TGF-β1. In the present study, we used Smad3^−/− mice to determine that Smad3 signaling plays a role in IL-2 production and T cell expansion. We show that TGF-β1 inhibits anti-CD3 plus anti-CD28-induced steady state IL-2 mRNA and IL-2 protein in a Smad3-dependent manner. In agreement with previous reports, we also show that Smad3 is critical for TGF-β1-mediated inhibition of anti-CD3 plus anti-CD28-induced T cell proliferation. By contrast, Smad3 was not essential for TGF-β1 to inhibit IL-2-induced T cell proliferation. These results provide the first evidence that TGF-β1 signals through Smad3-independent pathways to suppress T cell proliferation.

Smad3 knockout (Smad3^−/−) and wild-type (WT) mice were obtained from crosses of Smad3 heterozygotes and genotyped using PCR, as described ( 15). Mice were evaluated at 6–12 wk of age. WT littermates on a C57BL/SV129 background were used as controls.

All reagents were purchased from Sigma-Aldrich (St. Louis, MO), unless otherwise noted. Human rTGF-β1 was purchased from R&D Systems (Minneapolis, MN). Anti-CD3 (145-2C11), anti-CD28 (37.51), and ELISA protein standards were purchased from BD PharMingen (San Diego, CA). Mouse antiserum recognizing Smad3 was a generous gift from P. ten Dijke (The Netherlands Cancer Institute, Amsterdam, The Netherlands). Rabbit polyclonal Ab recognizing phosphorylated Smad2 was purchased from Upstate Biotechnology (Lake Placid, NY). HRP-conjugated secondary Abs for Western immunoblotting and ECL detection reagents were purchased from Amersham Pharmacia Biotech (Piscataway, NJ). Human rIL-2 was purchased from BioSource International (Camarillo CA). CFSE was purchased from Molecular Probes (Eugene, OR).

Mice were sacrificed via cervical dislocation. Single cell suspensions of whole splenocytes or thymocytes were prepared and activated with 2 μg/ml plate-bound anti-CD3 (145-2C11) and 1 μg/ml soluble anti-CD28 (37.51) in RPMI 1640 supplemented with 1% bovine calf serum (HyClone Laboratories, Logan, UT), 5 × 10⁻⁵ M 2-ME, 100 U/ml penicillin, and 100 μg/ml streptomycin. Cells were harvested at 2 h for mRNA analyses. Supernatants were recovered for ELISA analysis at 24 h. For proliferation, cells (2.5 × 10⁶ cells/ml; 200 μl/well) were cultured in quadruplicate in 96-well flat-bottom tissue culture plates in the presence or absence of TGF-β1 for 72 h. Cells were pulsed with [³H]TdR (1.0 μCi/well; NEN, Boston, MA) for the last 16 h of the culture. Tritium incorporation was quantified using a beta scintillation counter.

Splenocytes or thymocytes (2 ml at 5 × 10⁶ cells/ml) were lysed in a hypotonic buffer (10 mM HEPES, 1.5 mM MgCl₂), and nuclei were pelleted by centrifugation at 6700 × g for 5 min. Nuclei were lysed in a hypertonic buffer (30 mM HEPES, 1.5 mM MgCl₂, 450 mM NaCl, 0.3 mM EDTA, and 10% glycerol), supplemented with 1 mM DTT, 1 mM PMSF, 1 μg/ml aprotinin, leupeptin, and sodium fluoride, and 1 mM sodium orthovanadate for 30 min on ice. Samples were centrifuged at 17,500 × g for 20 min, and the supernatant was retained. Protein was quantified using the bicinchoninic acid assay (Sigma-Aldrich). Aliquoted samples were stored at −80°C. Nuclear extracts (25 μg/lane) were resolved by 10% SDS-PAGE, transferred to nitrocellulose membranes (Amersham Pharmacia Biotech), and immunostained with a mouse antiserum recognizing Smad3 or a rabbit polyclonal Ab recognizing phoshorylated Smad2. Bound Abs were visualized by incubation with HRP-conjugated goat anti-rabbit IgG (Promega, Madison, WI), followed by ECL detection (Pierce, Rockford, IL). Membranes were stripped and reprobed with mouse anti-β-actin (loading control).

Two hours after activation, total RNA was isolated using Tri Reagent (Molecular Research Center, Cincinnati, OH). A rIL-2 internal standard (IS) was used to quantify IL-2 mRNA expression by quantitative/competitive RT-PCR, as previously described ( 26). Briefly, total RNA and IS rcRNA of known amounts were reverse transcribed into cDNA using oligo(dT)15 as primers. The IL-2 primers produced a 391-bp amplified product from the cellular RNA and a 474-bp product from the IS rcRNA. IL-2 mRNA was quantified using a Gel Doc 1000 imaging system (Bio-Rad, Hercules, CA), as described previously ( 26). The number of transcripts was calculated from a standard curve generated from the density ratio between the IL-2 PCR product and varying spiked amounts of IS. The primer sequences used were (5′ to 3′): forward primer (TGCTCCTTGTCAACAGCC) and reverse primer (TCATCATCGAATTGGCACTC).

ELISA was used to detect levels of secreted IL-2. All Abs used in the ELISA were purchased from BD PharMingen. Cytokine detection and analyses were performed with an ELISA plate reader and the DeltaSoft 3-computer analysis program (BioMetallics, Princeton, NJ).

For CFSE labeling, 5 × 10⁶ unsorted splenocytes or thymic T cells were washed twice in staining buffer (PBS supplemented with 0.1% BSA) and resuspended in 250 μl of staining buffer; 250 μl of freshly diluted 2 μM CFSE (Molecular Probes) was added and cells were incubated at room temperature for 10 min with frequent mixing. An equal volume of FCS was added to stop labeling and cells were washed three times with culture medium containing 10% FCS. Cells were then activated with plate-bound α-CD3 plus soluble α-CD28, as described above. The cell division status of cells was determined by measuring CFSE fluorescence of 7-amino actinomycin D (7-AAD) negative-gated cells at 72 h after activation. Briefly, 10⁶ cells were incubated for 10 min in FACS buffer (PBS plus 1% BSA plus 0.01% sodium azide) with 10 μg/ml 2.4G2 Ab to block FcRs before labeling with TCR-α β-chain (H57) (BD Biosciences, San Jose, CA). After two washes, the cells were analyzed on a FACSCalibur (BD Biosciences). CFSE fluorescence was detected with the FL1 detector (530). Data were analyzed using CellQuest (BD Biosciences). Dead cells were excluded by 7-AAD staining, according to the manufacturer’s protocol (BD Biosciences). In some experiments, anti-CD69 (H1.2F3) (BD Biosciences) and mean cell size (forward side scatter) were acquired using CellQuest to monitor cell activation status.

Quantitation of CSFE dilution was determined, as previously described ( 27). Briefly, the TCR-β⁺ (clone H57-597; BD Biosciences) 7-AAD⁻ cells within each division peak (i) were enumerated by calculating the area under the curve (CellQuest software; BD Biosciences). Each of these values was divided by 2ⁱ and normalized to 100% to determine the number of input cells contributing to each peak. These input values were then plotted against division number to generate a normal population distribution. The mean division number for the population was determined from a best fit of these data to a Gaussian curve using Prism4 (GraphPad, San Diego, CA). Unstimulated, CFSE-labeled cells were used to verify the peak corresponding to the undivided population.

To assess the ability to respond to IL-2, unsorted splenocytes or thymocytes (2 × 10⁶/ml) were stimulated with 5 μg/ml Con A for 12 h, washed, stripped of cell surface-bound IL-2 by brief glycine buffer (pH 4) incubation ( 28), purified over Histopaque (Sigma-Aldrich), washed twice in complete medium, and restimulated in complete medium containing 1% FCS, with or without exogenous IL-2, in the presence or absence of TGF-β1 (5 ng/ml) for 72 h. Cells were pulsed with [³H]thymidine (1.0 μCi/well; NEN) for the last 16 h of the culture. For CFSE profiles, cells were stimulated with Con A for 12 h in the presence of 10% FCS (Gemini Bio-Products, Woodland, CA), washed with 10 μg/ml α-methyl mannoside (Sigma-Aldrich) ( 29), stripped of surface-bound IL-2, purified, washed twice with complete medium, and labeled with CFSE, as described above. Twenty-four, 48, and 72 h after addition of exogenous IL-2, cells were washed and stained with anti-TCR-β. Aliquots were then stained with PE anti-CD25, PE anti-CD122, or PE anti-CD132 (BD Biosciences). Annexin V staining was performed in HBSS containing calcium and magnesium for 10 min at 37°C, according to the manufacturer’s protocol (BD Biosciences). TCR-β⁺ 7-AAD⁻ cells were gated for analyses.

Data were evaluated by a parametric ANOVA, and Dunnett’s two-tailed t test was used to compare treatment groups with the naive control. Significant differences were determined at the level of p < 0.05.

In these studies, we used mice homogeneous for a null mutation of exon 8 of the Smad3 gene (Smad3^−/−) ( 15) and WT C57BL/SV129 littermate controls. We confirmed deletion of Smad3 protein in the spleen and thymus of these mice by Western immunoblot analyses (Fig. 1 A). Further characterization of the Smad3^−/− mice indicated, as previously reported ( 15), comparable numbers of splenic CD4⁺ and CD8⁺ T cell numbers between Smad3^−/− and WT mice (data not shown). T cell development was not dysregulated in 6- to 12-wk-old Smad3^−/− mice, as measured by normal total cell yields and percentages of thymic CD4⁺ and CD8⁺ subpopulations (data not shown).

ELISA was used to determine whether Smad3 plays a regulatory role in IL-2 production. WT or Smad3^−/− splenocytes were stimulated with anti-CD3 plus anti-CD28 for 24 h in the presence or absence of increasing concentrations of TGF-β1. As illustrated (Fig. 1 B), TGF-β1 inhibited IL-2 protein secretion by WT splenocytes in a concentration-dependent manner, and this inhibition was absent in Smad3^−/− splenic T cells, except at the highest concentration of TGF-β1 tested. A small amount (8–12 U/ml) of secreted IL-2 protein was detected in unstimulated Smad3^−/− splenocytes, but not WT splenocytes following 24 h in culture (data not shown).

Because IL-2 expression is highly regulated at the level of transcription and Smad3 predominantly functions to regulate gene transcription, we used quantitative RT-PCR to investigate whether Smad3 regulates IL-2 expression at the mRNA level. WT or Smad3^−/− splenocytes were stimulated with anti-CD3 plus anti-CD28 for 2 h in the presence or absence of TGF-β1. Anti-CD3 plus anti-CD28 induced steady state IL-2 mRNA expression in both WT and Smad3^−/− splenocytes, and TGF-β1 markedly attenuated this induction in WT, but not Smad3^−/− splenocytes (Fig. 1,C). We substantiated these findings by demonstrating that Smad3 was also essential for TGF-β1-mediated attenuation of anti-CD3 plus anti-CD28-induced steady state IL-2 mRNA in thymocytes (Fig. 1 D). Collectively, these results demonstrate that suppression of IL-2 production by TGF-β1 is mediated predominantly through a Smad3-coupled signaling pathway and is manifested, at least in part, at the level of steady state mRNA.

We have previously demonstrated that TGF-β1 suppresses anti-CD3 plus anti-CD28-induced T cell proliferation in a time- and concentration-dependent manner ( 30). Having established a role for Smad3 in IL-2 production, we next investigated whether Smad3 was essential for TGF-β1-mediated inhibition of anti-CD3 plus anti-CD28-induced T cell proliferation. Splenocytes or thymocytes were stimulated with anti-CD3 plus anti-CD28 for 72 h in the presence or absence of TGF-β1. TGF-β1 significantly impaired [³H]TdR incorporation in WT splenocytes (Fig. 2,A) and thymocytes (Fig. 2 B), and this suppression was greatly diminished in the absence of Smad3. Interpretation of these proliferative results is complex because [³H]TdR incorporation does not discriminate between proliferation and changes in cell survival. TGF-β1 can independently regulate both of these processes.

To more definitively characterize the role of TGF-β1 and Smad3 on cell proliferation, we directly evaluated early cell activation, division, and death as determined by cell size and CD69 surface expression, CFSE dilution, and 7-AAD and annexin V staining, respectively. TGF-β1 did not markedly impair T cell blastogenesis (forward light scatter) in either WT or Smad3^−/− splenocytes following anti-CD3 plus anti-CD28 stimulation (Fig. 3, A and B, respectively). Likewise, TGF-β1 had little or no effect on either the incidence (percentage) or magnitude (mean fluorescence intensity (MFI)) of CD69 expression in either WT or Smad3^−/− splenocytes at 24 h following anti-CD3 plus anti-CD28 stimulation (Fig. 3, C and D). At 48 h, CD69 MFI was slightly lower in the presence of TGF-β1, but there was no difference between WT and Smad3^−/− T cells (Fig. 3, E and F). Collectively, these results suggest that TGF-β1-mediated inhibition of IL-2 production and proliferation is not a consequence of impaired early T cell activation.

To investigate directly the effect of Smad3 on cell division, splenic T cells and thymocytes were labeled with CFSE and stimulated with anti-CD3 plus anti-CD28 for 72 h in the presence or absence of TGF-β1. Percentages of cells contained within each CFSE peak were calculated from histograms of 7-AAD⁻ TCR-β⁺-gated, CFSE-labeled splenocytes (Fig. 4,A), and these percentages are plotted against the number of cell divisions (Fig. 4, B and C). In the absence of exogenous TGF-β1, Smad3 deletion did not affect the proportion of cells that entered into the first division during the first 72 h after anti-CD3 plus anti-CD28 stimulation (i.e., WT, 75%; Smad3^−/−, 78%). TGF-β1, in a concentration-dependent manner, suppressed proliferation of WT splenic T cells as demonstrated by an increased percentage of cells in the undivided cohort (shown as 0 division) and a concomitant decrease in the percentage of cells that had undergone four or more divisions (Fig. 4,B). The suppressive effect by TGF-β1 was markedly diminished in Smad3^−/− splenic T cells (Fig. 4 C). These results clearly establish a role for Smad3 in the inhibition of TCR-induced cell proliferation.

Once T cells have been stimulated to up-regulate the IL-2R α-chain (CD25), IL-2 can bind to the high affinity IL-2R complex and induce cell proliferation independent of sustained TCR signaling ( 31). Because IL-2-induced proliferation is susceptible to down-regulation by TGF-β1 ( 32), we next investigated whether Smad3 contributes to the suppression of IL-2-induced proliferation by TGF-β1. For these experiments, splenocytes or thymocytes were first stimulated with Con A for 16 h to up-regulate surface expression of CD25. Con A-induced activation was blocked by washing the cells with α-methyl mannoside, a specific ligand for Con A ( 29). Cells were then washed in a low pH buffer to remove receptor-bound IL-2, and incubated for 72 h in fresh medium containing exogenous IL-2 in the presence or absence of TGF-β1. Flow cytometric analyses verified that Con A up-regulated CD25 expression in both WT and Smad3^−/− T cells (Fig. 5,A), and Smad3^−/− T cells were slightly more responsive. In comparison, Con A had little to no effect on expression of the IL-2R β-chain (CD122) or the common γ-chain (CD132) (Fig. 5, B and C, respectively).

The effects of TGF-β1 on IL-2-induced [³H]TdR incorporation are plotted as a function of IL-2 concentration (Fig. 6). These results illustrate that TGF-β1 inhibited IL-2-induced proliferation of WT splenocytes and thymocytes (Fig. 6, A and C). Surprisingly, in contrast to TCR-induced proliferation, TGF-β1-mediated inhibition of IL-2-induced proliferation was not disrupted in the Smad3^−/− T cells (Fig. 6, B and D). Exogenous IL-2 did not augment proliferation of thymocytes or splenocytes that had not been previously stimulated with Con A, indicating that up-regulation of the IL-2R α-chain was essential for IL-2-induced proliferation (Fig. 6).

We next used CFSE dilution to track the history of IL-2-induced cell division and to corroborate our [³H]TdR incorporation results. For these studies, Smad3^−/− and WT splenic T cells and thymocytes were first stimulated overnight with Con A to up-regulate the IL-2R α-chain, labeled with CFSE, and then stimulated with either 0, 5, or 10 U/ml IL-2 in the presence or absence of 5 ng/ml TGF-β1. At 72 h after addition of exogenous IL-2, both WT and Smad3^−/− splenic T cells (Fig. 7,A) and thymocytes (Fig. 7,B) displayed multiple peaks of CFSE intensity, indicating that each of these populations divided in an asynchronous manner in response to IL-2. The percentages of undivided cells were calculated from histograms of viable splenic T cells or thymocytes and these percentages were plotted (Fig. 8, A and B, respectively). Compared with WT T cells, a slightly greater percentage of Smad3^−/− T cells divided in response to IL-2. Nonetheless, TGF-β1 inhibited both Smad3^−/− and WT T cells from entering into the first division, as indicated by an increase in the percentage of undivided cells (Fig. 8, A and B).

To quantify the effect of Smad3 on progression of the divided population through subsequent rounds of division, population mean division numbers were calculated, as described in Materials and Methods. In response to IL-2, the mean division number was greater in Smad3^−/− T cells compared with WT T cells. Again, TGF-β1 inhibited mean division number independent of Smad3 (Fig. 8, C and D). Overall, these results demonstrate for the first time that TGF-β1 can inhibit T cell proliferation independent of Smad3 and show that the mechanism by which TGF-β1 inhibits T cell proliferation is dependent upon the mode of T cell activation.

TβRI synchronously phosphorylates Smad2 and Smad3 following TGF-β1 engagement, yet these two signaling molecules are functionally distinct as evidenced by embryonic lethality vs adult phenotype following their targeted deletion, respectively. Because of the distinct nature of these two Smad proteins, it is possible that TGF-β1 mediates some of its growth-inhibitory effects through a Smad2 pathway. Therefore, we investigated whether TGF-β1-induced phosphorylation of Smad2 remained intact in the absence of Smad3. WT or Smad3^−/− splenocytes and thymocytes were cultured for 30 min in the presence or absence of TGF-β1, and whole cell lysates were collected and analyzed via Western blotting. As illustrated in Fig. 9, for both splenocytes and thymocytes, Smad2 protein was present and its phosphorylation enhanced by TGF-β1. Thus, signaling through Smad2 provides a plausible alternative pathway whereby TGF-β1 may inhibit T cell proliferation independent of Smad3. Other pathways are also possible, as presented in Discussion.

TGF-β1 is produced by nearly all mammalian cell types and functions in an autocrine and paracrine manner. Transgenic mouse models have confirmed that TGF-β1 directly targets T cells to play a pivotal role in maintaining immune homeostasis ( 4). TGF-β1 responsiveness is influenced by the activation and differentiation status of the T cell, the strength of the T cell activation signal, and the microenvironment ( 30, 33, 34, 35). Details of the precise nature by which TGF-β1 inhibits IL-2 production and T cell proliferation remain undefined. TGF-β1 signals through a network of intracellular Smad-dependent and -independent cascades. Integration of these signaling cascades and cross talk with signal transducers from non-TGF-β1 pathways provide a mechanism to establish flexibility and fine tuning of TGF-β1 signaling. Several signal transducers are common among TCR, IL-2, and TGF-β1 signaling (e.g., NF-κB, extracellular signal-regulated kinase, and p38) and suggest that regulatory integration among these pathways is likely. Smad3 is highly expressed in lymphocytes, becomes phosphorylated in response to TβRII ligation, and transduces signals from the cell membrane into the nucleus. We used Smad3 knockout mice to investigate the role of intracellular Smad3 signaling on IL-2 production and proliferation in T cells ex vivo. We demonstrate that Smad3 is essential for TGF-β1-mediated inhibition of IL-2 production and TCR-induced proliferation, but is not required for TGF-β1 to inhibit IL-2-induced proliferation. Thus, we establish that TGF-β1 signals through at least two intracellular signaling pathways to inhibit T cell expansion.

We have determined that anti-CD3 plus anti-CD28-induced IL-2 production is a target of Smad3/TGF-β1 signaling and have identified the site of TGF-β1-induced inhibition to be either at the level of transcription or mRNA stability. Our results demonstrating a requirement for Smad3 in regulating anti-CD3 plus anti-CD28-induced steady state IL-2 mRNA are consistent with an earlier observation showing that TGF-β1 did not inhibit soluble anti-CD3 plus APC-stimulated steady state IL-2 mRNA in Smad3-deficient T cells ( 14). Our studies; however, relied strictly on plate-bound anti-CD3 for stimulation; therefore, we can conclude that it is the Smad3 in T cells that is necessary for TGF-β1-induced inhibition of IL-2 production.

The transcription factors AP-1, NF-AT, and NF-κB bind at distinct sites within the IL-2 promoter and interact to regulate activation of the iI2 gene in T cells ( 36, 37). Repression of il2 expression associated with T cell anergy ( 38) has been linked with binding of CREB/cAMP-responsive element modulator to the −180 site of the iI2 promoter ( 39). Multiple CAGA motifs lie juxtaposed to or overlap these critical transcription factor-regulatory response elements (i.e., AP-1, NF-AT, NF-κB, and CREB) in the proximal promoter/enhancer region (+1 to −300) of the iI2 promoter (GenBank accession number NM_008366). Similar Smad-binding CAGA motifs have been identified as essential TGF-β1-response elements in numerous other TGF-β1-responsive genes ( 8, 11, 40). The mechanism(s) underlying the role for Smad3 in regulating iI2 expression remains undefined. However, the proximity of the CAGA motifs to AP-1, NF-AT, NF-κB, and cAMP-responsive element modulator/CREB cis-regulatory response elements provides a mechanism for cooperativity between Smad3 and TCR-regulated transcription factors in regulating iI2 expression. Smad3, however, can do more than just bind DNA; it also plays an important role in chromatin remodeling. The coactivator, CBP/p300, which has intrinsic histone acetyltransferase activity, and the corepressor, TG-interacting factor, which binds to histone deacetylases, can both bind to Smad3 in a mutually exclusive manner. Therefore, Smad3 may regulate il2 expression through chromatin remodeling of the il2 locus ( 41, 42, 43).

The ability of Smad3 ablation to selectively blunt suppression of TCR-induced proliferation by TGF-β1, but not IL-2-induced proliferation, demonstrates that TGF-β1 can suppress T cell expansion through Smad3-independent pathways. These results add a new dimension to previous reports suggesting a complete nonresponsiveness of Smad3^−/− T cells to growth inhibition by TGF-β1 ( 14, 15, 44), and are consistent with the observation that TGF-β1-induced inhibition of B cell growth is also not dependent on Smad3 ( 14, 15). Moreover, the ability of TGF-β1 to effectively suppress IL-2-induced proliferation of Smad3^−/− splenic T cells and thymocytes, but its failure to inhibit TCR-induced proliferation of Smad3^−/− T cells suggests that IL-2 is not the sole or even the major mitogenic stimulus for anti-CD3 plus anti-CD28-activated T cells. This observation is consistent with results obtained with IL-2^−/− and IL-2R^−/− mice, whose T cells proliferate vigorously in vivo in the absence of IL-2 ( 23, 45, 46, 47). TGF-β1 rapidly turns off c-myc and prevents the turning off of Cdk inhibitors (i.e., p15^INK4B, p21^WAF1, and/or p27^KIP1) to suppress G₁-phase Cdk activity and induce G₁-phase cell cycle arrest. These effects have been shown to rely on direct physical interaction of Smad3 with sequence-specific cis-regulatory elements ( 48, 49, 50). Therefore, it is conceivable that TGF-β1 suppresses TCR-induced proliferation by directly targeting transcription of these Smad3-dependent genes.

In contrast to TCR-induced proliferation, Smad3 was dispensable for TGF-β1-mediated inhibition of IL-2-induced proliferation. One possible explanation for this difference may be the fact that STAT5a and STAT5b are uniquely activated in response to IL-2, but not through the TCR complex ( 51). These STAT proteins are critical for IL-2-induced proliferation ( 52). Reports concerning the effect of TGF-β1 on Jak/STAT signaling are somewhat controversial, but generally suggest that the inhibition is downstream of both the STAT5 and Shc pathways ( 53, 54, 55). Although recent studies by Nelson et al. ( 53) raise the possibility of a context-dependent role for Smad3 in IL-2-induced proliferation, one cannot exclude an alternative explanation that the dominant-negative Smad3 protein used in these studies impairs endogenous Smad2 as well as Smad3 ( 56). We show, in Fig. 9, that TGF-β1 phosphorylates Smad2 in splenocytes and thymocytes from Smad3 knockout mice, and propose that Smad2 signaling may contribute to the observed Smad3-independent inhibition of IL-2-induced proliferation. Although we have not yet defined the mechanism, Smad2 may indirectly stabilize corepressor complexes through direct binding to histone deacetylases to inhibit IL-2-induced proliferation through STAT5-coupled signaling ( 55). Alternatively, TGF-β1 may induce G₁/S-phase cell cycle arrest by signaling through Smad2 to turn off the Cdk tyrosine phosphatase, Cdc25. Cdc25 is up-regulated by IL-2 and subsequently activates Cdk4 and Cdk6; TGF-β1 inhibits Cdc25 expression by modulating histone acetylation at the E2F response element ( 57).

It is also possible that TGF-β1 inhibits IL-2-induced proliferation through Smad-independent mechanisms. For example, IL-2 activates phosphatase 2A (PP2A), which, in turn, inactivates p70^s6 kinase to permit transition of cells from G₁ into S phase. TGF-β1 has been shown to directly inhibit PP2A and prevent subsequent cyclin production ( 58). One study demonstrated that Smads are not required for TGF-β1-mediated inactivation of PP2A/p70^s6 kinase in epithelial cells ( 59). It remains to be determined whether TGF-β1 mediates a similar regulatory response in T cells.

In summary, this study reveals that TGF-β1 signals through Smad3-dependent and -independent pathways to suppress T cell expansion. We show that Smad3 is essential for TGF-β1 to suppress TCR-induced IL-2 production and cell proliferation, but not IL-2-induced proliferation in T cells ex vivo. The ability of TGF-β1 to use several intracellular pathways provides a mechanism to selectively facilitate or inhibit responses to different stimuli that a T cell receives during an immunological response.

We sincerely thank Anita Roberts, John Letterio, Peter ten Dijke, Carl-Henrik Heldin, Lalage Wakefield, and Sharon Wahl for contributing mice, reagents, and valuable discussions.