Nature Immunology
- 7, 1157 - 1165 (2006)
Published online: 1 October 2006; | doi:10.1038/ni1398
A pathway regulated by cell cycle inhibitor p27Kip1 and checkpoint inhibitor Smad3 is involved in the induction of T cell toleranceLequn Li1, Yoshiko Iwamoto1, Alla Berezovskaya2 & Vassiliki A Boussiotis1, 31 Transplantation Biology Research Center, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts 02129, USA. 2 Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts 02129, USA. 3 Department of Medicine Division of Hematology and Oncology, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts 02129, USA.
Correspondence should be addressed to vboussiotis@partners.org Peripheral tolerance is essential for immunological homeostasis. Tolerant T cells are thought to arise after T cell receptor ligation in conditions that are nonpermissive for replication. Here we have investigated the function of the cell cycle inhibitor p27Kip1 in tolerance induction in vivo using naive T cell receptor–transgenic cells lacking the cyclin-dependent kinase (Cdk)–binding domain of p27Kip1(p27 ). Wild-type but not p27 cells underwent tolerization. Tolerized wild-type cells had impaired Cdk2 and Cdc2 kinase activity and failed to phosphorylate the checkpoint inhibitor Smad3, leading to enhanced expression of the Cdk inhibitor p15. In contrast, p27 cells proliferated in tolerizing conditions because of Cdk kinase activation and phosphorylation of Smad3, which resulted in no upregulation of p15. Smad3 'knockdown' prevented tolerance induction, whereas expression of a Smad3 mutant resistant to Cdk-mediated phosphorylation recapitulated molecular and functional events of tolerance. Thus, p27Kip1 is required during induction of tolerance and Smad3 regulates T cell responses 'downstream' of p27Kip1.Maintenance of peripheral tolerance is essential for homeostasis of the immune system. Tolerance in vivo and its in vitro counterpart anergy are defined as the inability of T cells to produce interleukin 2 (IL-2) and expand their populations after stimulation with competent antigen-presenting cells (APCs). Anergy of helper T cells is induced by ligation of the T cell receptor (TCR) by antigen alone and is prevented by costimulation or IL-2 (refs. 1,2). Given the dual properties of costimulation and IL-2 to induce T cell proliferation and to prevent the induction of anergy, it has been hypothesized that IL-2 and the costimulatory molecule CD28 may prevent anergy because they induce T cell proliferation. Consequently, it has been proposed that anergy may be the result of TCR engagement in the absence of proliferation1. Subsequent studies have supported that view by showing that blockade of cell cycle progression by rapamycin or by forced upregulation of the cell cycle inhibitor p27Kip1 induces T cell anergy even in the presence of costimulation2,
3.
The cyclin-dependent kinase (Cdk) inhibitor p27Kip1 is a member of the 'Cip-Kip' family of Cdk inhibitors and has been reported to interact with Cdk2, the adapter Grb2, and Rho family GTPases4,
5,
6,
7. Cdk2 is activated and associates with cyclin E at the late G1 phase and with cyclin A at the S phase of the cell cycle. Through its interaction with Cdk2, p27Kip1 can inhibit the activation of cyclin E–Cdk2 and cyclin A–Cdk2 complexes. Although p27Kip1 also associates with cyclin D–Cdk4 (or cyclin D–Cdk6), this interaction does not seem to induce inhibition of cyclin D–associated enzymatic activity but instead functions to sequester p27Kip1, thereby facilitating cyclin E–Cdk2 activation8. The inhibitor p27Kip1 also associates with Cdc2 (also called Cdk1) and regulates cyclin E–Cdc2 activity independently of Cdk2 (ref. 9). Silencing or inhibiting Cdc2 affects S-phase entry specifically in the absence of Cdk2, suggesting that Cdk2 may have a dominant function in this9. Cdk2 promotes cell cycle progression in part by phosphorylating the transcription factor Rb and related 'pocket proteins', thereby reversing their ability to sequester E2F transcription factors, leading to expression of E2F-regulated genes5. Cdk2 also phosphorylates the signal transducers Smad2 and Smad3 (ref. 10). Smad3 inhibits cell cycle progression from the G1 to the S phase, and impaired phosphorylation of its Cdk-specific sites renders it more effective in executing this function. In contrast, Cdk-mediated phosphorylation of Smad3 reduces Smad3 transcriptional activity and antiproliferative function10.
In T lymphocytes, p27Kip1 is a critical inhibitor of cell cycle progression. Deficiency of p27Kip1 in mice leads to multiorgan hyperplasia with disproportional enlargement of lymphoid organs11,
12,
13. Downregulation of p27Kip1 is essential for normal thymocyte development14 and for population expansion of mature T lymphocytes in response to mitogenic stimuli15,
16,
17. Stimulation with IL-2 or through the IL-2 receptor leads to downregulation of p27Kip1 protein expression18, thereby contributing to cell cycle progression and clonal expansion. CD28 costimulation also leads to downregulation of p27Kip1 by mediation of ubiquitination and subsequent degradation of p27Kip1 and by induction of IL-2 production and downregulation of p27Kip1 mediated by the IL-2 receptor17,
18. Upregulation of p27Kip1 correlates with induction of anergy in vitro and tolerance in vivo3,
19,
20,
21,
22. Moreover, forced expression of p27Kip1 in T lymphocytes leads to the induction of anergy3. In contrast, p27Kip1-deficient primary T lymphocytes are resistant to anergy induction in vitro, suggesting that expression of p27Kip1 is indispensable in anergy induction22. Consistent with that, p27Kip1-deficient recipients reject cardiac allografts in conditions that induce long-term allograft survival in wild-type recipients23. Despite those extensive studies, it remains unclear whether p27Kip1 itself causes the induction of tolerance of naive T cells in vivo. Moreover, the mechanism by which p27Kip1 may mediate its effects during the induction of tolerance has not been determined.
Here we have examined the function of p27Kip1 during induction of tolerance of naive T cells in vivo using DO11.10 TCR-transgenic T cells deficient in recombination-activating gene 2 (Rag2) and lacking the cyclin-Cdk–binding domain of p27Kip1 (called 'p27 ' cells here). This approach disrupts the interaction of p27Kip1 with cyclin-Cdk complexes13, whereas other p27Kip1 interactions remain unaffected. We adoptively transferred Rag2-/- DO11.10 TCR-transgenic CD4+ T cells (called 'DO11.10' cells here) or Rag2-/- DO11.10 TCR-transgenic p27 CD4+ T cells (called 'DO11.10p27' cells here) into syngeneic wild-type recipient mice and compared the development of immune responses to an immunogenic or tolerizing stimulus in vivo. Expression of p27Kip1 in naive T cells was required for induction of tolerance in vivo. Moreover, our studies show that Smad3 is a critical component of a pathway 'downstream' of p27Kip1 regulating T cell responses.
Results
Similar primary responses in naive control and p27 cells
Mice deficient in p27Kip1 have more CD4+ memory T cells and a larger percentage of T lymphocytes in the S phase of the cell cycle than do wild-type mice11,
24. Those issues create potential caveats and make it difficult to use mice with homozygous deficiency of the gene encoding p27Kip1 for study of the function of p27Kip1 in the activation of naive T cells. Here we used DO11.10 mice, which express a major histocompatibility complex class II–restricted transgenic TCR specific for an ovalbumin peptide consisting of amino acids 323–339 (OVA(323–339)) that can be detected with a clonotypic antibody25. To generate p27Kip1-deficient mice with naive T cells, we used DO11.10 TCR-transgenic mice crossed onto the Rag2-/- background, thereby eliminating endogenous TCRs, and subsequently crossed the Rag2-/- DO11.10 strain onto the p27 background. Naive T cells from DO11.10 and DO11.10p27 mice had similar expression of surface CD44, CD25 and CD62L (data not shown), allowing comparison of naive cells differing only in expression of the p27Kip1 Cdk-binding domain.
Initially we compared the responses of DO11.10 and DO11.10p27 T cells to antigen-specific stimulation in vitro. During primary stimulation, naive DO11.10 and DO11.10p27 had no difference in the kinetics of their responses, although the peak proliferation of DO11.10p27 cells was slightly greater than the proliferation of DO11.10 cells (Fig. 1a). We obtained the same pattern of responses using various antigen concentrations (Supplementary Fig. 1 online). We detected no differences in IL-2 production (Fig. 1b). During secondary responses, the proliferation of DO11.10p27 cells was 1.8- to 3-fold higher than the proliferation of DO11.10 cells (Fig. 1c), and DO11.10p27 cells produced more IL-2 (Fig. 1d). Thus, naive DO11.10 cells lacking the Cdk-binding domain of p27Kip1 had enhanced proliferation that was greatest after antigenic rechallenge.
p27 cells are resistant to tolerance induction
Next we evaluated the function of p27Kip1 in regulating the responses of naive T cells to immunizing (priming) or tolerizing stimulus in vivo using the well established DO11.10 adoptive transfer model20,
25. To induce tolerance in vivo, we blocked costimulation by using a combination of a cytotoxic T lymphocyte antigen fusion protein (CTLA-4–Ig) plus monoclonal antibody (mAb) to CD40 ligand (CD40L) to inhibit signaling through both the B7-CD28 and the CD40-CD40L pathways26. We purified CD4+ T cells from DO11.10 and DO11.10p27 mice and adoptively transferred the cells into syngeneic wild-type recipient mice. Then, 1 d later, we left recipient mice untreated (naive), treated them with PBS followed by immunization with OVA(323–339) (primed) or treated them with CTLA-4–Ig plus mAb to CD40L followed by immunization with OVA(323–339) (tolerized).
To determine whether tolerance or priming had been induced during in vivo stimulation in the recipients of adoptively transferred cells, we assessed proliferation and IL-2 production after restimulation of DO11.10+ T cells with antigen in vitro. DO11.10 and DO11.10p27 cells from the naive mice had similar proliferation and IL-2 production (Fig. 2a,b). DO11.10 cells from primed mice were capable of proliferating and producing copious IL-2 (Fig. 2c,d). In contrast, DO11.10 cells that encountered a tolerizing stimulus in vivo proliferated poorly and had almost undetectable IL-2 production (Fig. 2c,d). DO11.10p27 cells that encountered a tolerizing stimulus in vivo had proliferation and IL-2 production indistinguishable from that of DO11.10p27 cells from primed mice (Fig. 2e,f). Tolerizing treatment reduced the number of IL-2-producing DO11.10 cells to one third that of priming treatment. In contrast, we detected similar numbers of IL-2-producing DO11.10p27 cells after tolerizing or priming treatment (Supplementary Fig. 2 online). Thus, T cells lacking the Cdk-binding domain of p27Kip1 are resistant to tolerance induction.
 | | Figure 2. DO11.10p27 T cells are resistant to tolerance induction in vivo. | | |  | DO11.10 and DO11.10p27 cells were adoptively transferred into syngeneic recipient mice that were treated with no antigen (a,b; Naive) or were treated subcutaneously with OVA(323–339) in adjuvant alone (c–f; Primed) or together with mAb to CD40L plus CTLA-4–Ig (c–f; Tolerized). CD4+ T cells were purified from lymph nodes collected on day 15 after treatment and were cultured with irradiated BALB/c splenocytes and OVA(323–339) (concentrations, horizontal axes (a,c,e)). Proliferation was measured at day 3 of culture (a,c,e) and IL-2 in supernatants of cultures at 24 and 48 h was measured by ELISA (b,d,f). The percentage of KJ1-26+ cells from each individual mouse in each treatment group was determined by flow cytometry and the number of KJ1-26+ cells per culture was calculated by multiplication of the frequency of KJ1-26+ cells in each sample by the number of total CD4+ cells per well (1.5  105); mean c.p.m. values at each peptide dose were divided by the number of KJ1-26+ cells (a,c,e). Results represent data from two (a,b) or four (c–f) independent experiments.
 Full Figure and legend (59K) |
| |
p27 cells progress to G1-S despite tolerizing treatment
DO11.10 cells tolerized in vivo are incapable of progressing to the S phase of the cell cycle after rechallenge20. To determine the effect of the p27Kip1 Cdk-binding domain on G1-S blockade after tolerance induction, we assessed the activation of cyclin E that occurs at the G1- to S-phase transition. We primed or tolerized DO11.10 and DO11.10p27 cells in vivo, then collected CD4+ T cells from recipients of adoptive transfer and restimulated equal numbers of cells positive for mAb KJ1-26 (a clonotypic antibody that detects the DO11.10 TCR) with OVA(323–339) and wild-type APCs for various times in vitro. We detected cyclin E activation after restimulation of in vivo–primed DO11.10 cells (Fig. 3a). In contrast, activation of cyclin E was impaired in DO11.10 cells that had received a tolerizing stimulus in vivo (Fig. 3a). Notably, DO11.10p27 cells that had encountered either a priming or tolerizing stimulus in vivo were capable of activating cyclin E after rechallenge (Fig. 3b).
Next we examined the function of p27Kip1 using a different method to induce tolerance. In this established experimental approach using the same DO11.10 TCR-transgenic model20,
25, tolerance or priming of the adoptively transferred TCR-transgenic T cells in syngeneic recipient mice can be induced by means of different routes of antigen administration in vivo. Subcutaneous administration of OVA(323–339) induces immunogenic stimulation, whereas intravenous administration of OVA(323–339) induces tolerance. As with the costimulation blockade experiments, DO11.10 but not DO11.10p27 cells from mice given the tolerizing treatment had impaired proliferation and IL-2 production after antigenic rechallenge (data not shown). These results confirmed that p27Kip1 was critical in regulating tolerance induction of naive T cells in vivo.
p27 cells are suppressed by regulatory T cells
Induction of tolerance by blockade of costimulation may require regulatory T cells (Treg cells)21,
27. Because in our experimental model DO11.10 and DO11.10p27 cells were transferred into syngeneic wild-type recipients, we determined whether p27 cells might be resistant to tolerance induction because they were resistant to suppression by syngeneic, wild-type Treg cells in the host. We cultured DO11.10 or DO11.10p27 cells with various numbers of syngeneic Treg cells. Treg cells suppressed the proliferation (Fig. 4a) and IL-2 production (Fig. 4b) of DO11.10p27 cells to the same as or an even greater extent than that of wild-type DO11.10 cells. Thus, resistance to tolerance induction in DO11.10p27 cells is not due an intrinsic resistance of p27 cells to the suppressive function of Treg cells.
Equal number of divisions for control and p27 cells
It has been suggested that cell division may prevent induction of tolerance1. We determined whether the resistance of DO11.10p27 cells to undergo tolerance induction was related to a distinct cell division profile during antigen-specific stimulation in vivo. After priming stimulation, DO11.10 and DO11.10p27 cells underwent the same number of divisions in vivo (Fig. 5a). Similarly, both cell types underwent the same number of divisions after tolerizing stimulation (Fig. 5b), although both had fewer divisions than with priming stimulation (Fig. 5c,d). During priming treatment for both cell types, most cells (68–73%) underwent five or six rounds of divisions. However, 7.4% of DO11.10 cells and 13.8% of DO11.10p27 cells underwent seven rounds of divisions, thereby resulting in a modestly higher (1.8-fold) in vivo population expansion of DO11.10p27 than of DO11.10 cells (Fig. 5e,f). During tolerizing stimulation, most cells (54–61%) underwent four or five rounds of division and only a minor fraction (3–5%) had completed six divisions. In these conditions, there was an even smaller difference in the in vivo population expansion (1.4-fold) of DO11.10p27 versus DO11.10 cells (Fig. 5g,h).
| | |
Rb is phosphorylated in TCR-transgenic control and p27 cells
Phosphorylation of Rb is associated with cell division5. In both DO11.10 and DO11.10p27 cells, the phosphorylation and retarded electrophoretic mobility of endogenous Rb was induced by either priming or tolerizing stimulation in vivo (Fig. 6a). In both types of cells, the heavily phosphorylated Rb isoform with slower electrophoretic mobility predominated and the ratio of heavily phosphorylated Rb to phosphorylated Rb was higher after priming stimulation (Fig. 6a). Thus, despite the finding that only DO11.10 became tolerant, both DO11.10 and DO11.10p27 cells demonstrated phosphorylation of endogenous Rb and underwent the same number of divisions. These results collectively suggested that mechanisms other than cell division probably account for the distinct susceptibility of DO11.10 and DO11.10p27 cells to tolerance induction.
Impaired Cdk activity in tolerized control but not p27 cells
Because the Cdk-binding domain of p27Kip1 that negatively regulates activation of cyclin E–Cdk2 and cyclin A–Cdk2 is deleted in DO11.10p27 cells that are resistant to tolerizing treatment, we hypothesized that Cdk2 might be a critical target of p27Kip1 for tolerance induction. Cdk2 kinase activity was higher in DO11.10 cells from primed mice and was lower in cells from tolerized mice (Fig. 6b). In contrast, DO11.10p27 cells given either a tolerizing or a priming stimulus in vivo had similar Cdk2 activity (Fig. 6b). Because p27Kip1 also associates with and can inhibit the activity of Cdc2, we also assessed Cdc2 kinase activity. Cdc2 activation had a pattern similar to that of Cdk2 in all experimental populations (Fig. 6b).
Distinct Smad3 transactivation in control and p27 cells
Cdk proteins phosphorylate Smad2 and Smad3, negatively regulating the Smad3 transcriptional activity and cell growth–inhibitory function10. Maximum Cdk-mediated Smad3 phosphorylation occurs at the G1-phase–S-phase junction and requires activation of Cdk2 (ref. 10). Like Cdk2, Cdc2 can also phosphorylate Smad3 (F. Liu, personal communication). Because tolerizing treatment resulted in impaired Cdk2 and Cdc2 activation, we determined whether Smad3 might be differentially phosphorylated in DO11.10 and DO11.10p27 cells during priming or tolerizing stimulation. We examined Cdk-mediated phosphorylation of endogenous Smad3 using an antibody specific for Smad3 phosphorylated at serine 212 (S212), a site specifically phosphorylated by Cdk proteins10. Phosphorylation of Smad3 at S212 was undetectable in naive cells (data not shown). Priming stimulation induced phosphorylation of Smad3 at S212 in both DO11.10 and DO11.10p27 cells (Fig. 6c). Tolerizing treatment resulted in impaired phosphorylation of Smad3 at S212 in DO11.10 cells but not in DO11.10p27 cells (Fig. 6c).
Based on those findings, we determined whether differential Cdk-mediated phosphorylation of Smad3 differentially regulated Smad3 transcriptional activity in primed versus tolerized cells after rechallenge with antigen. We used a Smad-dependent reporter28 to assess Smad3 transcriptional activity by luciferase assay. DO11.10 cells from tolerized mice had abundant Smad3 transcriptional activity, whereas DO11.10 cells from primed mice had much less Smad3 transcriptional activity (Fig. 6d). In contrast, DO11.10p27 cells from mice of both treatment groups had low Smad3 transcriptional activity (Fig. 6d). To analyze the functional consequences of those findings, we examined expression of p15, a known downstream transcriptional target of Smad3. After restimulation, expression of p15 was upregulated in DO11.10 but not in DO11.10p27 cells that had encountered a tolerizing stimulus (Fig. 6e).
Those results collectively indicated that tolerance induction results in impaired Cdk-mediated phosphorylation of Smad3 in vivo, leading to abundant transcriptional activation of Smad3 and upregulation of the Cdk4-Cdk6 inhibitor p15 after rechallenge. Those events might account for the inability of tolerized cells to progress past the G1 phase of the cell cycle after restimulation. Furthermore, those events required the Cdk-binding domain of p27Kip1, as they were not detected in DO11.10p27 T cells that had encountered a tolerizing stimulus.
Smad3 is a critical component in tolerance induction
Next we addressed whether Smad3 had a 'causative' function downstream of p27Kip1 in the induction of T cell tolerance. First we 'knocked down' endogenous Smad3 in cells using short hairpin RNA specific for Smad3 (Smad3 shRNA) and examined the susceptibility of these 'Smad3-knockdown' cells to tolerance induction. Smad3 shRNA induced a 90% reduction in Smad3 protein compared with the amount in cells treated with vector expressing shRNA specific for luciferase control ('control-knockdown' cells), as determined by immunoblot (Supplementary Fig. 3 online). We adoptively transferred Smad3-knockdown and control-knockdown DO11.10 and DO11.10p27 cells into syngeneic recipient mice that we subsequently primed or tolerized in vivo. We recovered TCR-transgenic CD4+ T cells from draining lymph nodes and cultured the cells in vitro with wild-type APCs and OVA(323–339). Naive Smad3-knockdown DO11.10 and DO11.10p27 cells showed similar proliferation and IL-2 production (data not shown). Those responses were higher than those of control-knockdown DO11.10 and DO11.10p27 cells, consistent with published observations with T lymphocytes from Smad3-deficient mice29,
30. Control-knockdown DO11.10 cells from primed mice were capable of proliferating and producing IL-2, whereas control-knockdown DO11.10 cells from tolerized mice proliferated poorly and had almost undetectable IL-2 production (Fig. 7a,b). Smad3-knockdown DO11.10 cells that encountered either priming or tolerizing stimulus showed similar proliferation and IL-2 production after rechallenge, indicating that knockdown of endogenous Smad3 rendered them resistant to the induction of tolerance (Fig. 7a,b).
| | Figure 7. Smad3 is a critical component of a pathway downstream of p27Kip1 in regulating induction of productive immunity or tolerance. | | |  | Proliferation (a,c) and IL-2 production (c,d) of DO11.10 (a,b) or DO11.10p27 (c,d) T cells transfected with control shRNA (control-KD) or Smad3 shRNA (Smad3-KD) and adoptively transferred into syngeneic recipient mice that then received priming or tolerizing treatment; lymph nodes were collected on day 5 after treatment and transfected DO11.10 cells were isolated and cultured with irradiated BALB/c splenocytes and OVA(323–339) (5  g/ml). (a,c) Proliferation, measured at day 3 of rechallenge culture and expressed as 'stimulation index' (SI)44. (b,d) IL-2 in supernatants at 48 h of culture. (e–h) Proliferation (e,g) and IL-2 production (f,h) of DO11.10 (e,f) or DO11.10p27 (g,h) T cells transfected with vector control or Smad3-Tm, then adoptively transferred into syngeneic recipient mice that then received priming or tolerizing treatment, followed by culture and analysis of responses as described in a–d. Results are representative of two separate experiments with three mice per type of transfection and treatment group (error bars, s.d. of triplicates of each sample).
Full Figure and legend (49K) |
| |
Control-knockdown DO11.10p27 cells were resistant to tolerance induction and showed similar responses after priming or tolerizing treatment (Fig. 7c,d). Smad3-knockdown DO11.10p27 cells remained resistant to tolerance induction. Moreover, those cells had greater proliferation and IL-2 production than did control-knockdown DO11.10p27 cells (Fig. 7c,d). These data are consistent with published observations indicating a synergistic effect between p27 deficiency and Smad3 deficiency in enhancing T cell responses31.
As a second approach to determine whether Smad3 caused tolerance induction, we used a Smad3 'triple mutant' that is resistant to Cdk-mediated phosphorylation ('Smad3-Tm'). We transfected DO11.10 and DO11.10p27 cells with Smad3-Tm and vector control and adoptively transferred the cells into syngeneic recipient mice, which we then primed or tolerized in vivo. We recovered TCR-transgenic CD4+ T cells from draining lymph nodes and cultured them with syngeneic APCs and OVA(323–339). DO11.10 cells transfected with vector control had enhanced in vitro responses after encountering the priming stimulus and reduced responses after encountering the tolerizing stimulus in vivo (Fig. 7e,f). In contrast, DO11.10 cells transfected with Smad3-Tm that had encountered either a priming or tolerizing stimulus in vivo had reduced proliferation and IL-2 production (Fig. 7e,f). Thus, expression of a Smad3 mutant resistant to Cdk-mediated phosphorylation did not alter the susceptibility of DO11.10 cells to tolerizing treatment but rendered them resistant to priming treatment.
DO11.10p27 cells transfected with vector control that then encountered either a priming or tolerizing stimulus in vivo were capable of proliferating and producing IL-2 after rechallenge (Fig. 7g,h), indicating that they remained resistant to tolerance induction. In contrast, for DO11.10p27 cells transfected with Smad3-Tm, proliferative capacity after priming or tolerizing treatment was reduced and IL-2 production was impaired (Fig. 7g,h). These results collectively provided evidence that nonphosphorylated Smad3 is a negative regulator of T cell activation and that Cdk-mediated phosphorylation of Smad3 is at least one critical mechanism by which p27Kip1 deficiency regulates enhanced responses after priming and resistance to tolerance induction in vivo.
Smad3-Tm recapitulates the molecular findings of tolerance
Only two of the transfection conditions described above altered cell responses to tolerizing treatment: Smad3 knockdown rendered DO11.10 cells resistant to tolerance, whereas Smad3-Tm rendered DO11.10p27 cells susceptible to tolerizing treatment. To determine the mechanisms involved in those functional outcomes, we determined whether expression of p15 and cell cycle progression were altered. Expression of p15 was not upregulated in Smad3-knockdown DO11.10 cells, in contrast to control-knockdown DO11.10 cells that encountered a tolerizing stimulus (Fig. 8a). Upregulation of p15 in control-knockdown cells was associated with blockade of progression to the S phase, as determined by lack of cyclin A expression (Fig. 8a). In contrast, Smad3-knockdown DO11.10 cells did not show upregulation of p15 and were capable of progressing to the S phase, as determined by expression of cyclin A (Fig. 8a).
| | |
DO11.10p27 cells transfected with Smad3-Tm had upregulated expression of p15 compared with that of similarly treated cells transfected with vector control (Fig. 8b). Upregulation of p15 was more prominent in cells that had encountered tolerizing stimulus (Fig. 8b) and was inversely correlated with expression of cyclin A (Fig. 8b). Thus, downregulation of endogenous Smad3 signaling in DO11.10 cells recapitulated the molecular pattern of cell cycle progression in DO11.10p27 cells despite tolerizing treatment, whereas expression of the Smad3 mutant resistant to Cdk-mediated phosphorylation in DO11.10p27 cells recapitulated the molecular pattern of cell cycle blockade in tolerized DO11.10 cells.
Discussion
In this study we have investigated the function of p27Kip1 in the development of peripheral T cell tolerance in vivo using the DO11.10 adoptive transfer model. Our adoptive transfer approach eliminated the concerns of non–TCR-mediated stimulation of T cells in an activated environment in p27Kip1-deficient mice11,
24 and provided the means to study the intrinsic function of p27Kip1 in the activation of naive T cells. Our studies have demonstrated that after antigen-specific rechallenge, DO11.10p27 T cells given a tolerogenic stimulus in vivo showed a vigorous immune response, characterized by considerable proliferation and IL-2 production; in contrast, similarly treated DO11.10 T cells proliferated poorly and did not secrete IL-2. Our in vivo studies have provided evidence that p27Kip1 has an active function required for tolerance induction.
Studies of anergy induction in vitro and tolerance induction in vivo have shown that tolerized T cells contain substantial p27Kip1 and are blocked at the G1 phase of the cell cycle2,
3,
19,
20,
32. That event is not simply correlative but seems to be 'causative' in tolerance induction. Experiments using antigen-specific T cell clones have provided compelling evidence that arrest of T cells at the G1 phase of the cell cycle by rapamycin results in the induction of anergy even in the presence of costimulation or IL-2 (ref. 2). In contrast, blockade of the cell cycle at the S phase by hydoxyurea does not result in the induction of anergy2. Those data indicate that it is neither costimulation nor proliferation itself that prevents anergy induction but instead the biochemical events that occur after progression from G1 to the S phase of the cell cycle2. Our data here have demonstrated how biochemical events that occur during the G1- to S-phase transition but do not directly involve cell division may determine the fate of T cells toward induction of tolerance or productive immunity, as predicted2. After encountering a tolerizing stimulus, DO11.10 and DO11.10p27 T cells underwent the same number of cell divisions in vivo, but only DO11.10 cells became tolerant, whereas DO11.10p27 cells were resistant to tolerance induction. A critical biochemical event at the G1- to S-phase junction that occurred only in the DO11.10p27 cells during tolerizing treatment was sustained activation of Cdk2 and Cdc2.
An unexpected finding of our studies was the induction of Cdk-mediated phosphorylation of Smad3 at S212 and suppression of Smad3 transcriptional activity after priming. In contrast, tolerizing stimulation resulted in impaired phosphorylation of Smad3 at S212 and abundant Smad3 transcriptional activity in DO11.10 T cells. The Cdk interacting-inhibitory domain of p27Kip1 seemed to be required for those tolerance-related events, as DO11.10p27 cells had more Cdk2 and Cdc2 activity and phosphorylation of Smad3 at S212 and less Smad3 transcriptional activity after tolerizing stimulation. As a consequence, there was upregulation of the Cdk4- and Cdk6-specific inhibitor p15, a transcriptional target of Smad3, in DO11.10 but not DO11.10p27 cells that had encountered tolerizing stimulus. Our results are consistent with observations of other cell types indicating that Cdk-mediated phosphorylation reduces Smad3 transcriptional activity and antiproliferative function. Conversely, alteration of the Cdk phosphorylation sites increases Smad3 transcriptional activity, leading to more expression of p15 (ref. 10). Thus, by regulating Smad3 phosphorylation, p27Kip1 may affect T cell responses because it regulates the transcriptional machinery independently of its direct function on the cell cycle. Consistent with that idea, studies have shown that p27Kip1 can negatively regulate tumor growth independently of its involvement in cell proliferation33.
Cdk-mediated phosphorylation of Smad3 reduces both basal and transforming growth factor- (TGF-)–mediated Smad3 transcriptional activity, thereby opposing TGF--mediated function. In contrast, Smad3 with alterations in its Cdk-phosphorylation sites shows enhanced responses to TGF--mediated transcription10. TGF- is produced during antigenic stimulation of T lymphocytes and suppresses their responses34. Genetic approaches have indicated that endogenous TGF- is important in the regulation of tolerance induction and maintenance of T cell quiescence35,
36. Our studies here have indicated that priming stimuli oppose TGF-–Smad3 signaling and the functional outcome because of Cdk-mediated phosphorylation of Smad3, leading to suppression of Smad3 transcriptional activation. In contrast, tolerizing stimuli facilitate TGF-–Smad3 signaling because of impaired Cdk-mediated phosphorylation of Smad3, leading to abundant Smad3 transcriptional activity.
There is compelling evidence that Smad3 negatively regulates IL-2 transcription. Forced expression of Smad3 in Jurkat T cells inhibits IL-2 transcription37, whereas Smad3-deficient T cells show enhanced IL-2 transcription30. Smad proteins function as 'nodes' for the integration of various signal-transduction pathways, including MAP kinases, Ras, Jak-STAT and Wnt38. Thus, Smad3-mediated signals may impair IL-2 production by interfering with pathways involved in IL-2 transcription39. An alternative hypothesis is that Smad3 may directly bind to Smad-binding sites in the promoter of the gene encoding IL-2, thereby inhibiting IL-2 transcription30,
37. Dependent on the context of promoters, activated Smad proteins interact with different DNA-binding cofactors, recruiting transcriptional coactivators or corepressors to distinct target gene promoters38. By that mechanism, Smad3 may positively regulate the transcription of p15 and negatively regulate the transcription of IL-2. Cdk-mediated phosphorylation of Smad3 reduces its DNA-binding activity and suppresses its ability to regulate both the induction and repression of gene transcription in various cell lines10. Consistent with that, our studies have shown that priming stimulation of T cells, which induced Cdk-mediated phosphorylation of Smad3, resulted in low expression of p15 and high expression of IL-2. In contrast, tolerizing stimulation, which induced impaired Cdk-mediated phosphorylation of Smad3, resulted in more expression of p15 and less expression of IL-2.
Our observations have provided a link between p27Kip1 and Smad3 in the regulation of tolerance induction in naive T cells in vivo. A link between p27Kip1 and TGF-–Smad3 has been supported by published studies. Originally, p27Kip1 was identified as an inhibitor of cyclin-Cdk complexes in cells arrested by TGF-40. In addition, T cells lacking p27Kip1 or p21Cip1 or, more notably, lacking both p27Kip1 and p21Cip1 are less susceptible to the inhibitory effects of TGF-41,
42. Moreover, studies of the regulation of 'leukemogenesis' have provided genetic evidence for a functional link between p27Kip1 and Smad3 (ref. 31). Those studies have indicated that leukemia cells from pediatric patients with T cell acute lymphocytic leukemia have less p27Kip1 protein than do normal T lymphocytes and have no detectable Smad3 protein expression. Smad3 deficiency in Smad3-/- mice is not associated with T cell leukemia29, indicating that loss of Smad3 signaling alone is insufficient to initiate 'leukemogenesis'. Similarly, although p27Kip1-deficient mice have more T cells, they do not develop leukemia11,
12,
13. Loss of one Smad3 allele by p27Kip1-deficient mice results in massive lymphocytic infiltrates in multiple organs, nephropathy and cardiomyopathy31. T cell leukemia also develops in 10% of these p27-knockout Smad3+/- mice31. Those results indicate that p27Kip1 deficiency reduces the sensitivity of T lymphocytes to the inhibitory effects of Smad3 and, therefore, that Smad3 is 'haploinsufficient' in preventing T cell hyperactivation, autoimmunity and 'leukemogenesis'. Further studies are needed to determine the importance of the interaction between p27Kip1 and Smad3 in the regulation of T cell responses and to potentially exploit their properties for therapeutic benefit in autoimmunity and leukemia.
Methods
Mice.
BALB/c mice 6–8 weeks old were purchased from Charles River Laboratory and were used as syngeneic recipients. Mice lacking the cyclin-Cdk–binding domain of p27Kip1 (p27) were provided by A. Koff (Memorial Sloan-Kettering Cancer Center, New York, New York) and were bred onto the BALB/c background for 14 generations as described for Ctla4-/- mice20. DO11.10 TCR-transgenic Rag2-deficient mice were purchased from Taconic and were bred 'onto' p27 on the BALB/c background. Mice were maintained in our breeding colony and were cared for in accordance with US National Institutes of Health and institutional guidelines (Massachusetts General Hospital Subcommittee on Research Animal Care-OLAW Assurance A3596-01).
In vitro T cell cultures.
For analysis of primary antigen-specific T cell responses of naive cells, CD4+ cells (1 106 cells/ml) from DO11.10 and DO11.10p27 mice were purified and were cultured for 1–4 d in 96-well plates with various concentrations of OVA(323–339) in the presence of APC samples that were depleted of T cells and treated with mitomycin C (10 106 cells/ml). Proliferation was assessed by [3H]thymidine incorporation for the last 16–18 h of culture. For analysis of secondary responses, CD4+ cells and APCs at the density described above were cultured for 4 d in 24-well plates with OVA(323–339) (1 g/ml). T cells were then collected and were allowed to 'rest' overnight, and then T cells (1 105 cells/ml) were stimulated for 3 d in 96-well plates with various concentrations of OVA(323–339) in the presence of APCs (5 106 cells/ml). Proliferation was assessed as described above. Supernatants were collected from primary and secondary cultures at various times and IL-2 concentrations were measured by enzyme-linked immunosorbent assay (ELISA).
Adoptive transfer and immunization.
For CD4+ T cell transfer, lymphocytes were collected from pooled peripheral lymph nodes and spleens of DO11.10 and DO11.10p27 mice. Then, CD4+ T cells were isolated by positive selection with microbeads coated with antibody to CD4 (Miltenyi Biotec). T cells (3 106 cells per mouse) were adoptively transferred intravenously into nonirradiated, syngeneic BALB/c wild-type recipient mice. At 1 d after cell transfer, recipients were left untreated (naive) or were treated intraperitoneally with PBS (primed) or with antibody to CD40L (250 g/mouse) plus CTLA-4–Ig (250 g/mouse; both from Bioexpress Cell culture Services; tolerized). Then, 24 h later, primed and tolerized mice were immunized subcutaneously with 100 g of OVA(323–339) in incomplete Freund's adjuvant. At various times, cells were collected from draining lymph nodes and spleen for use in experiments.
Analysis of cell divisions in vivo.
Purified T cells from DO11.10 and DO11.10p27 mice (10 106 cells/ml) were labeled for 30 min at 37 °C with the intracellular fluorescent dye CFSE (5 M 5(and 6)-carboxyfluorescein succunimidyl ester; Molecular Probes). Then, cells were washed twice with cold RPMI 1640 medium containing 10% FCS, were resuspended in PBS and were transferred intravenously into BALB/c mice (5 106 cells per mouse). Syngeneic hosts were left untreated (naive) or were treated with PBS followed by immunization with OVA(323–339) (primed) or with CTLA-4–Ig plus mAb to CD40L followed by immunization with OVA(323–339) as described above (tolerized). Then, 3 d later, lymphocytes were isolated from the draining lymph nodes of the BALB/c hosts. The number of cell divisions on CFSE-stained cells and the percentage of cells that had undergone a specific number of divisions were determined as described43. Cells were also stained with mAb KJ1-26 and CFSE analysis of KJ1-26+ T cells was done by flow cytometry.
Treg cell suppression assay.
For analysis of antigen-specific T cell responses, CD4+ cells (1 106 cells/ml) from DO11.10 and DO11.10p27 mice were purified and were cultured for 1–4 d in vitro with various concentrations of OVA(323–339) in the presence of APC samples depleted of T cells and treated with mitomycin C (10 106 cells/ml). Proliferation was assessed as described above. For analysis of suppressive responses to CD4+CD25+ T cells, CD4+CD25- wild-type or DO11.10p27 T cells (5 105 cells per well) were cultured for 96 h in 96-well flat-bottomed plates containing APC samples depleted of T cells (5 105 APCs), 1 g/ml of mAb to CD3 (2C11) and various numbers of CD4+CD25+ T cells isolated from wild-type BALB/c mice. The proliferation of responder T cells was assessed as described above.
Restimulation assay after in vivo immunization.
For analysis of T cell priming in vivo, CD4+ T cells were collected from naive, primed or tolerized recipient mice on day 15 after immunization. Proliferative responses were measured by culture for 72 h of CD4+ T cells (3 106 cells/ml) with irradiated (3,000 rads) APCs (10 106 cells/ml) and OVA(323–339). The number of KJ1-26+ cells for each group of recipient mice was determined by flow cytometry and proliferation was normalized to the number of input KJ1-26+ cells. Supernatants were collected from plates and cytokine concentrations were measured by ELISA.
Flow cytometry.
For analysis of surface antigen expression, mAb to CD4 (JK1.5; eBioscience) and mAb KJ1-26 (KJ-126; Caltag) were used. For intracellular IL-2 staining, T cells were restimulated for 24 h in vitro with OVA(323–339) in the presence of APCs as described above. Brefeldin A (eBioscience) was added for the last 6 h of the culture. Cells were collected and were stained with allophycocyanin-conjugated mAb to CD4 and fluorescein isothiocyanate–conjugated mAb KJ1-26. Then, cells were fixed, were made permeable and were stained with antibody to IL-2 (clone JES6-5H4; eBioscience) according to the manufacturer's instructions.
Immunoblot, immunoprecipitation and in vitro kinase reactions.
For analysis of cyclin E, Cdk2 and Cdc2 activation, antiserum specific for cyclin E (sc-481; Santa Cruz Biotechnology), Cdk2 (sc-163; Santa Cruz Biotechnology;) or Cdc2 (9114; Upstate Biotechnology) was used for immunoprecipitation, followed by an in vitro kinase reaction with histone H1 as the exogenous substrate, as described3. Reaction products were separated by 10% SDS-PAGE and were transferred to a polyvinyldifluoride membrane that was then exposed to X-ray film. For analysis of Cdk-mediated phosphorylation of Smad3, antiserum specific for Smad2 and Smad3 (sc-8332; Santa Cruz Biotechnology) was used for immunoprecipitation, followed by immunoblot with an antibody that detects phosphorylation of Smad3 at S212 (the Cdk-specific phosphorylation site). For immunoblot analysis, antiserum specific for Smad2 and Smad3 (sc-6032; Santa Cruz Biotechnology) and antiserum to Smad3 (ab28379; Abcam; for immunoblot of Smad3-knockdown and control-knockdown cells) were used. The mAb to Rb was from BD Pharmingen (clone JI12-906), the antiserum to p15 was from Cell Signaling (4822) and the antiserum to -actin was from Santa Cruz Biotechnology (sc-1615). Densitometry was done as described20.
T cell transfection and reporter assays.
DO11.10 and DO11.10p27 T cells were adoptively transferred intravenously into syngeneic recipient mice and additional treatments were done as described above. Then, 15 d later, CD4+ T cells were collected from naive, primed or tolerized recipient mice and were transfected (5 106 cells per condition) with plasmids by nucleofection with the Amaxa nucleofection apparatus. For analysis of Smad3 transcriptional activity, cells were transfected with the Smad-dependent 4xSBE-luciferase reporter (20 g/sample) and Smad3 plus Smad4 expression plasmids (2.5 g each/sample)28. Transfection efficiency was normalized by cotransfection of cDNA encoding green fluorescent protein. Expression was assessed by flow cytometry and was consistently similar for all transfection conditions (28–30%). Cells were stimulated for 48 h with irradiated (3,000 rads) syngeneic APCs pulsed with OVA(323–339), and luciferase activity was measured according to an established protocol37. The number of KJ1-26+ cells for each group of transfected cells was determined by flow cytometry, and luciferase activity was normalized to the number of input KJ1-26+ cells. DO11.10 and DO11.10p27 cells isolated from four host mice in each treatment group (primed and tolerant) were transfected individually and were analyzed for luciferase activity.
Mammalian expression plasmids and transfection.
For generation of the plasmid expressing Smad3 shRNA, the following specific oligonucleotides were used: upper, 5'-GATCCACCTGAGTGAAGATGGAGATTCAAGAGATCTCCATCTTCACTCAGGTTTTTTTACGCGTG-3'; lower, 3'-AATTCACGCGTAAAAAAACCTGAGTGAAGATGGAGATCTCTTGAATCTCCATCTTCACTCAGGTG-5'. These were cloned under control of the U6 promoter into the pSIREN-DNR-DsRed expression vector (Clontech, BD). Vector expressing shRNA specific for luciferase served as a control. Smad3-Tm was subcloned into the pIRES2-EGFP vector (Clontech, BD); empty vector served as a control. Purified DO11.10 or DO11.10p27 T cells were transfected with plasmids by nucleofection with the Amaxa nucleofection apparatus, according to the manufacturer's instructions (Mouse T Cell Nucleofector Kit Amaxa Biosytems). Purified T cells were suspended in nucleofector solution (3 106 cells/100 l) and were mixed with 3 g of plasmid. Samples were transferred into cuvettes, were transfected with nucleofector program X-01 and were then immediately transferred into 12-well plates and were cultured in nucleofector medium for 3 h. Then, cells were collected and counted and were immediately transferred into syngeneic recipient mice (3 106 cells per mouse). At 3 h after adoptive transfer, mice were given priming or tolerizing treatment in vivo according t |
