Transforming growth factor-β (TGF-β) maintains self-tolerance through a constitutive inhibitory effect on T-cell reactivity. In most physiological situations, the tolerogenic effects of TGF-β depend on the canonical signaling molecule Smad3. To characterize how TGF-β/Smad3 signaling contributes to maintenance of T-cell tolerance, we characterized the transcriptional landscape downstream of TGF-β/Smad3 signaling in resting or activated CD4 T cells. We report that in the presence of TGF-β, Smad3 modulates the expression of >400 transcripts. Notably, we identified 40 transcripts whose expression showed Smad3 dependence in both resting and activated cells. This ‘signature’ confirmed the non-redundant role of Smad3 in TGF-β biology and identified both known and putative immunoregulatory genes. Moreover, we provide genomic and functional evidence that the TGF-β/Smad3 pathway regulates T-cell activation and metabolism. In particular, we show that TGF-β/Smad3 signaling dampens the effect of CD28 stimulation on T-cell growth and proliferation. The impact of TGF-β/Smad3 signals on T-cell activation was similar to that of the mTOR inhibitor Rapamycin. Considering the importance of co-stimulation on the outcome of T-cell activation, we propose that TGF-β-Smad3 signaling may maintain T-cell tolerance by suppressing co-stimulation-dependent mobilization of anabolic pathways.
Transforming growth factor-beta (TGF-β) is a highly conserved cytokine that has evolved as a pleiotropic mediator that governs broad aspects of cell and tissue development, differentiation and homeostasis.1, 2 In the vertebrate immune system, TGF-β is essential for maintenance of immune tolerance.3 Indeed, TGF-β-deficient mice (or mice lacking a functional TGF-β receptor) suffer from fatal auto-immunity caused by spontaneous activation of self-reactive CD4 T cells.4, 5, 6, 7 Moreover, recent evidence shows that loss of TGF-β responsiveness in mature T cells decreases the threshold for T-cell activation.8 Importantly, TGF-β is abundantly expressed in secondary lymphoid organs and constitutive TGF-β signaling is present in resting CD4 T cells.9 Through tonic inhibitory signaling in CD4 T cells, TGF-β enforces tolerance by limiting T-cell activation signals and likely through the modulation of yet undetermined tolerance enforcing genes.
TGF-β is synthesized as a pro-peptide and secreted in the extracellular space. Once activated by proteases, it binds to its hetero-tetrameric receptor on the surface of the cells, which is composed of two type I and two type II chains. Both chains are endowed with serine/threonine kinase activity capable of activating several signaling pathways. However, the TGF-β receptor complex (TGF-βR) operates mainly through a so-called canonical pathway by phosphorylating the receptor-associated Smads (R-Smad): Smad2 and/or Smad3. Activated R-Smads form hetero/homodimers and then bind a co-Smad (Smad4) before moving as trimeric complexes in the nucleus. The Smad complexes can bind DNA and interact with a large variety of co-activators and co-repressors that are cell and context specific and modulate the transcription of hundreds of genes.2, 10
Current evidence supports that Smad3 is the main mediator downstream of TGF-β in T cells.11, 12 Although not recapitulating the severe auto-immune manifestations of TGF-β deficiency, Smad3−/− mice are subject to inflammatory diseases.13 In a murine hematopoietic cell transplantation model, we have previously showed that healthy Smad3-deficient allogeneic donor cells could induce lethal graft vs. host disease , while their WT littermates did not. Consistent with the pleiotropic effects of TGF-β, Smad3 maintained tolerance to the host by restraining CD4 activation and Th1 skewing in a cell-intrinsic manner, as well as by limiting the expansion and cytotoxic activity of myeloid effectors. We also found that Smad3 mediates the TGF-β-dependent inhibition of CD4 T-cell proliferation.14 A likely possibility is that the TGF-β/Smad3 signaling axis limits CD4 T-cell proliferation by interfering with converging activation pathways leading to CD4 T-cell activation and division. This would be consistent with our previous finding that T-cell receptor (TCR) triggering swiftly decreases Smad3 phosphorylation,15 suggesting that the interruption of the TGF-β/Smad3 axis is functionally relevant for productive CD4 T-cell activation. While several transcriptional targets of Smad3 in T cells have been identified (for example, Foxp3 (ref. 16)), we lack a systems level characterization of the transcriptional landscape controlled by Smad3 in resting and activated T cells.
Using gene expression microarrays, we characterized the transcriptional landscape downstream of TGF-β/Smad3 signaling in naive CD4 T cells that were either quiescent or in early phases of activation. We found that Smad3 had little non-redundant effects on transcript levels in the absence of TGF-β. However, the addition of physiological concentrations of TGF-β to T-cell cultures revealed that Smad3 regulates the expression of >400 transcripts. The superimposition of CD3ɛ and CD28 agonistic stimuli mitigated the differences between WT and Smad3−/− CD4 T cells. This suggests that in the early phases of T-cell activation, the imprint of the TGF-β/Smad3 pathway on transcription remains present, albeit at a decreased degree. Smad3 targets belonged to several distinct pathways and encompassed known and putative regulators of T-cell function and tolerance. Furthermore, we provide evidence that, in CD4 T cells, Smad3 regulates a number of biological processes, and of particular relevance, cell growth and metabolism. Functional assays showed that the TGF-β/Smad3 pathway restricts CD4 T-cell growth and proliferation by mitigating the effects of CD28 co-stimulation. These effects were analogous to those of the mTOR inhibitor rapamycin. We thus reveal an unsuspected link between the TGF-β/Smad3 signaling axis and the control of cell growth pathways involved in early CD4 T-cell activation and proliferation.
The TGF-β/SMAD3 signaling pathway modulates >400 transcripts in quiescent and early activated naive CD4 T cells
Sorted naive CD62L+, CD44− CD4 T cells from WT or Smad3−/− littermates were cultured either in the absence or presence of exogenous TGF-β. In a third condition, naive CD4 T cells were subjected to both TGF-β and agonistic anti-CD3ɛ and anti-CD28 (αCD3ɛ/CD28) stimulation to reproduce early T-cell activation (Figure 1a). We cultured CD4 T cells for 6 h, which is sufficient to reach a plateau in SMAD-mediated transcription15 and, in the case of anti-CD3ɛ/CD28 stimulation, sufficient to initiate the transcription of IL-2, an early and determining step in T-cell activation (Iwashima et al.17; Supplementary Figure 1). After 6 h of culture, the number of differentially expressed transcripts (defined as 1.5-fold difference in expression and a P-value of <0.05) between WT and Smad3−/− cells was 59 in the absence of exogenous TGF-β. This number increased to 418 in the presence of TGF-β, in accordance with the notion that the key role of Smad3 is to transduce TGF-β signals (Figure 1b; Supplementary Table 1 for the list of all differentially expressed transcripts across conditions). The number of differentially expressed transcripts decreased to 142 in the presence of both TGF-β and anti-CD3ɛ/CD28 (Figure 1b). The latter observation is consistent with the notion that T-cell activation attenuates TGF-β-dependent Smad3 signaling.15 Accordingly, unsupervised hierarchical clustering performed on selected transcripts showing expression variation across conditions in WT cells showed maximal disparities between WT and Smad3−/− cells in the presence of TGF-β alone (Figure 1c). Moreover, while some transcripts were overexpressed in WT relative to Smad3−/− cells, others were repressed, suggesting that Smad3 is both a transcriptional activator and repressor in naive CD4 T cells. Of note, some transcripts overexpressed in WT cells showed high fold change (up to 13-fold) and low P-value (Figure 1b) in both the TGF-β and TGF-β+αCD3ɛ/CD28 conditions. In contrast, transcripts overexpressed in Smad3−/− cells relative to WT cells tended to show more modest differential expression and little consistency across experimental conditions (Figure 2a). Since the tolerogenic effects of the TGF-β/Smad3 pathway operate in both quiescent and TCR-stimulated cells, we examined the transcripts that were differentially expressed in both the TGF-β and TGF-β+αCD3ɛ/CD28 conditions.
The gene expression signature of the TGF-β/Smad3 pathway in resting and early activated naive CD4 T cells
In the two conditions where TGF-β stimulation was applied, we found that 36 and 4 transcripts were consistently overexpressed in WT and Smad3−/− cells, respectively (Figure 2a and b). In other words, 40 transcripts were modulated in a TGF-β/Smad3-dependent manner in both resting and TCR/CD28-stimulated CD4 T cells. These 40 gene transcripts showed great heterogeneity in terms of molecular function or contribution to biological processes (Figure 2c). Importantly, nearly half of these genes have previously been described as modulators of immune cell function (Table 1).
We next performed intra-genotype (that is, between WT cells or between Smad3−/− cells) comparisons to assess the level of expression of these 40 transcripts after culture in the presence vs. absence of TGF-β (Table 1; Figure 2d). Our objective was twofold: to confirm that these transcripts were truly regulated by TGF-β and to evaluate whether TGF-β might affect their expression in the absence of Smad3. Strikingly, 38 out of 40 gene transcripts were found to be differentially expressed between the medium only and TGF-β conditions in either WT or Smad3−/− cells. In WT cells, TGF-β exposure increased by at least 1.5-fold (P-value<0.05) all but one gene transcripts (Ubash3b), previously found to be overexpressed in WT compared with Smad3−/− cells in the TGF-β condition. The same analysis revealed that out the 36 transcripts overexpressed in WT relative to Smad3−/− cells (Figure 2a), 18 were upregulated by TGF-β in Smad3−/− cells (Figure 2d). This suggests that TGF-β/Smad3 targets fall into two main categories: those where transduction of TGF-β signals is entirely Smad3 dependent (they are not induced by TGF-β in Smad3−/− cells) or others where Smad3 enhances effects of TGF-β signals that can be transduced by alternative pathways downstream of the TGF-βR (probably by Smad2). Conversely, Olfr1437 and Tpm2 which are overexpressed in Smad3−/− cells relative to WT cells are induced by TGF-β only in Smad3−/−, but not in WT cells suggesting that Smad3 impedes the TGF-β-dependent transcription of these genes. Interestingly, Cpt1c that is overexpressed in Smad3−/− compared with WT cells has a lower expression in TGF-β-treated WT cells compared with WT cells incubated in medium only, suggesting that TGF-β, through Smad3, might repress the expression of Cpt1c. Globally, our data demonstrate that the TGF-β/Smad3 pathway leaves a consistent imprint on the transcriptome of CD4 T cells involving in most cases an upregulation of transcript levels.
Since the TGF-β/Smad3 pathway exerts an important tolerogenic effect, we explored whether the above described gene signature could unveil potential TGF-β/Smad3-dependent tolerogenic mechanisms. At the individual gene level, the TGF-β/Smad3 signature across conditions of quiescence and early T-cell activation revealed several regulators of immunity, some previously known to be TGF-β pathway targets (such as Maf18). Interestingly, several differentially expressed transcripts were related to regulatory T cell (TReg) biology (Cxcr4,19 Ctla2a20 and Nrp1 (ref. 21)). Of note, the TReg master transcription factor FoxP3 was not differentially expressed between WT and Smad3−/− cells (data not shown). The most differentially expressed gene transcript between WT and Smad3−/− cells, in all three experimental conditions was Cathepsin W (Ctsw), which is known to be upregulated in activated CD8 and NK cells,22 but has no defined role in CD4 T cells. Moreover, by extending our search to transcripts differentially expressed in either the TGF-β or TGF-β+αCD3ɛ/CD28 condition, we identified several gene transcripts, which are candidate immunoregulatory genes downstream of the TGF-β/Smad3 pathway (Table 2). Although the role of each of these genes as ‘tolerance-enforcing’ or ‘T-cell activating’ may differ in various contexts, we noted that the transcripts of the IL-10 receptor, a master cytokine of T-cell tolerance, as well as that of the TReg-associated transcription factor Foxo3a23 were induced in a Smad3-dependent manner. On the contrary, Smad3−/− cells overexpressed the transcripts of calcineurin (Ppp3ca), a central mediator in calcium-dependent activation of T cells, and of two effector granzymes. In summary, the TGF-β/Smad3 axis transcriptionally regulates several key modulators of immune cell biology, but also many other candidates whose function in CD4 T-cell physiology is unknown.
The TGF-β/SMAD3 transcriptional axis modulates the expression of transcripts related to T-cell activation networks and metabolism
In an attempt to interrelate the 40 members of the gene signature, we performed exploratory searches using Ingenuity Pathway Analysis tool. We found that the TGF-β/Smad3 signature could be associated with several networks. One network contained 18/40 of the signature gene transcripts and had the highest score (42). The likelihood of finding another network within the database that would contain the same number of signature genes by chance was 10−42 (Figure 3). This top network, named Cardiovascular system development and function/Organismal development/Cancer, included at its core several central pathways relevant to T-cell activation such as Akt, NFκb, Nfat, Map kinases and Stat5. Although this does not imply a functional relationship, such association between our gene signature and pathways having a determining role in T-cell activation was largely unsuspected. We extended our search to determine whether TGF-β/Smad3 signaling and transcription could be impacting other yet undetermined pathways. The study of differentially expressed transcripts according to pre-defined criteria might have overlooked relevant contribution of the TGF-β/Smad3 axis to the transcriptional regulation of specific biological processes or pathways. We used the Gene Set Enrichment Analysis resource, a non-threshold based algorithm, to assess the representation of our data sets into pre-defined gene sets from curated databases.24 Using the KEGG database, we found that when cells were cultured in unsupplemented media, only four gene sets were differentially represented in WT vs. Smad3−/− cells (data not shown), in accordance with the minimal differences noted between WT and Smad3−/− cells in this condition (Figure 1b). However, in the TGF-β and TGF-β+αCD3ɛ/CD28 conditions, respectively 18 and 14 gene sets were overrepresented in WT cells. Despite a high number of differentially expressed transcripts, the number of gene sets overrepresented in Smad3−/− cells was small (two in the TGF-β condition and none in the TGF-β+αCD3ɛ/CD28 condition). Our main findings are summarized in Table 3 and suggest a contribution of the TGF-β/Smad3 pathway in three broad categories: (1) nucleic acid and protein synthesis and degradation, (2) transcription and (3) metabolism. The possible regulation of oxidative phosphorylation and protein catabolism, two hallmarks of quiescent T cells,25 evoke the possibility that the TGF-β/Smad3 pathway might regulate T-cell tolerance by modulating metabolic pathways.25, 26, 27, 28, 29 Given our findings relating signature genes with key players of cell metabolism and growth such as Akt (Figure 3) and GO categories such as Metabolic process (Figure 2c), the Gene Set Enrichment Analysis findings further suggest that the TGF-β/Smad3 pathway might modulate T-cell biology through the control of energy and cell growth pathways. T-cell tolerance and activation are inextricably linked to their metabolic state30, 31 and as such, cell growth pathways could be functionally targeted by tolerogenic TGF-β and Smad3 signaling and transcription.
TGF-β antagonizes CD28-dependent co-stimulation in a Smad3-dependent manner
During T-cell activation, T-cell metabolism switches from catabolic to anabolic and profoundly alters the fate of naive T cells.25, 32 After antigen recognition by the TCR, the engagement of co-stimulatory molecules (signal 2) has a major role in coupling productive T-cell activation with the reassignments of metabolic fluxes.33, 34 We examined whether the Smad3-dependent suppressive effects of TGF-β on sorted naive CD4 T-cell proliferation were similar in the presence or absence of CD28 co-stimulation. Specifically, we analyzed both the proliferation kinetics (number of daughter cells generated per proliferating cell) and the fraction of cells that underwent at least one cell division.35 We knew from previous work that TGF-β/Smad3 signaling blunts anti-CD3ɛ/CD28-mediated CD4 T-cell proliferation.14, 36 We confirmed and extended these data by showing that while Smad3 is necessary to mediate the anti-proliferative effects of TGF-β on T cells stimulated by anti-CD3ɛ and anti-CD28, Smad3 is dispensable when anti-CD3ɛ is used alone (Figures 4a and b). In other words, TGF-β-mediated suppression of T-cell proliferation is either Smad3 dependent or independent depending on whether activation involves CD28 co-stimulation or not. Moreover, a conspicuous feature of cells activated with anti-CD3ɛ alone compared with those stimulated with anti-CD3ɛ/CD28 was a reduced cell size as assessed by forward scatter (FSC) (Figure 4a). This was to be expected since CD28 co-stimulation is known to increase cell size.37 The salient finding was that the growth suppressing effect of TGF-β (at both 2.5 and 5 ng ml−1) on co-stimulated CD4-cells was significantly less important in Smad3−/− relative to WT cells (Figures 4a–c; Supplementary Figure 2). WT cells were smaller than Smad3−/− cells at each division (Figure 4d). Moreover, this difference in cell size between WT and Smad3−/− cells was apparent at 36 h after stimulation, that is, before the first cell division (Figures 4e and f). Increase in cell growth precedes the first division after T-cell activation. This suggests that in the presence of TGF-β Smad3−/− CD4 T cells mobilize the cell growth machinery more readily after TCR and CD28 triggering than their WT counterparts. Together, our data show that the TGF-β/Smad3 signaling axis impedes the effects of CD28 co-stimulation on naive CD4 T-cell growth and proliferation.
The TGF-β/Smad3 axis inhibits CD4 T-cell activation through a rapamycin-like effect
The cell growth and proliferation that follows productive TCR and co-stimulatory receptor engagement has been shown to be highly dependent on the mTOR kinase27 (through the mTORC1 complex) and its two main substrates S6K and 4eBP1, which have a positive and a negative effect on translation, respectively. Phosphorylation activates S6K and inhibits 4eBP1. According to current evidence, mammalian cell growth is mostly controlled by S6K, whereas proliferation and protein synthesis rely on both substrates.38 We performed western blot analyses on WT and Smad3−/− CD4 T cells after 36 and 72 h of culture to assess the quantity and phosphorylation status of the two main mediators downstream of the mTOR kinase; S6K and 4eBP1 (Supplementary Figure 3; Figure 5a). In concordance with our carboxyfluorescein succinimidyl ester (CFSE) dilution assays and FSC measurements, there were notable differences between WT and Smad3−/− cells in the presence of TGF-β. After TCR stimulation, all substrates, phosphorylated S6K and 4eBP1 as well as actin increased in T cells (Supplementary Figure 3). However, we noted that both phosphorylated and native S6K and 4eBP-1 proteins were decreased (relative to actin) by TGF-β in naive CD4 T cells WT, but not in Smad3−/− cells after in vitro stimulation (Figure 5a). In these conditions, the ratio of phospho-S6K and phospho-4eBP-1 relative to total S6K or 4eBP-1 remained unchanged (data not shown). These findings demonstrate that TGF-β and Smad3 affect the quantity of the two most important downstream effectors of the mTOR pathway. A decrease in the amount of active phospho-S6K could explain the inhibitory effect of TGF-β/Smad3 on cell growth and proliferation. Prolonged exposure to rapamycin has been similarly associated with a decrease expression of phospho-4eBP-1 and unphosphorylated 4eBP-1 in other models but its biological significance is unclear.39, 40
We next asked whether rapamycin would reproduce the effects of TGF-β. The addition of rapamycin at a dose of 5 ng ml−1 in the absence of exogenous TGF-β, significantly blunted the hypertrophic and proliferative responses of CD28 co-stimulated cells. The effect on cell size occurred irrespective of the Smad3 genotype (Figure 5b). Likewise, the effect of rapamycin on proliferation kinetics was similar between WT and Smad3−/− cells (Figure 5c). However, the combination of TGF-β and rapamycin, nearly abolished the proliferative response of WT cells as the number of cells reaching the first division was only 5% and proliferation kinetics decreased. In Smad3−/− cells, the effect of the combination was much less stricking as 24% of cells divided with no significant change in the calculated number of daughter cells per dividing cells (Figure 5c). Importantly, TGF-β and/or rapamycin did not induce significant cell death in our system that could have impacted on cell size, proliferation as well as protein measurements (Supplementary Figure 4).
In conclusion, rapamycin recapitulated most effects of TGF-β on CD4 T-cell size and proliferation. However, the effect of rapamycin on cell size and proliferation kinetics was Smad3 independent, which might indicate that rapamycin could palliate for lack of TGF-β/Smad3-dependent inhibition of the mTOR pathway. Moreover, our data show a functional synergy between TGF-β and rapamycin, which is most evident in WT cells. The presence of additive effects between TGF-β and rapamycin could reflect a convergence between the inhibitory mechanisms underlying the action of the TGF-β/Smad3 pathway and rapamycin on T-cell activation. Globally, these data suggest a negative functional relationship between the TGF-β/Smad3 and the mTOR pathways in CD4 T cells.
The tolerogenic mechanisms orchestrated by TGF-β are diverse and operate in several cell types. However, a crucial aspect is to enforce peripheral tolerance by limiting CD4 T-cell response to low affinity self-antigens and to limit ongoing immune responses. Smad3, a canonical mediator of TGF-β signaling, is instrumental in these processes.11, 14, 41 This work was undertaken to discover how TGF-β/Smad3 signals maintain tolerance in CD4 T cells. By performing our analyses in both quiescent and early activated cells, we were able to investigate the genetic impact of the TGF-β/Smad3 pathway in resting CD4 T cell as well as during the transition from quiescence to activation. Our results show that following TGF-β stimulation, Smad3 has a partially or totally non-redundant effect on >400 transcripts. This confirms that the other R-Smad, Smad2 as well as the other pathways downstream of the TGF-β receptor cannot fully palliate for the lack of Smad3. We identified a group of 40 gene transcripts that were consistently differentially expressed between WT and Smad3−/− cells after exposure to TGF-β, whether αCD3ɛ/CD28 stimulation was applied or not. This signature included several putative immunomodulatory genes, but also genes whose role in immunobiology is unknown. Our microarray data suggest that the ancestral canonical TGF-β pathway mediates its potent effects by impacting several distinct pathways, in line with the characteristic pleiotropy of this cytokine in different organ systems. Likewise, the use of gene set enrichment algorithms revealed that the TGF-β/Smad3 axis impinged on several distinct biological processes, including pathways relevant to T-cell activation. Of particular interest, our data suggest that catabolic/anabolic pathways and oxidative phosphorylation are regulated by TGF-β and Smad3 in naive CD4 cells. We therefore explored the role of TGF-β/Smad3 in the modulation of growth and energy pathways in CD4 T cells. We found that the relationship between TGF-β, metabolism and cell growth was striking since TGF-β is an ancestral switch coupling energy sources with cellular fates. In C. elegans, where the interaction between R-Smads and Co-Smad orthologs is antagonistic rather than synergistic,42 the TGF-β ortholog Daf7 is secreted in the presence of abundant energy sources and prevents a pause in the larval development (Dauer stage). In mammals, TGF-β modulates growth and glucose metabolism in fibroblasts43 and epithelial cells by positively regulating the mTOR pathway.44 However, TGF-β promotes muscle atrophy by counterbalancing Akt-mTOR signaling.45 In hematopoietic stem cells, TGF-β is at the nexus of the quiescence/cycling decision by opposing stromal-derived factor-1 activation of the mTOR pathway.46 Moreover, a functional antagonistic relationship between the mTOR and TGF-β pathway is present as mTOR-deficient T cells show hyperactive TGF-β signaling.47 This is in accordance with data obtained with other cell systems that have also demonstrated a similar inhibitory function of the mTOR pathway on TGF-β-Smad signaling.48, 49, 50 However, evidence of a reciprocal inhibitory effect of the TGF-β/Smad3 axis on the mTOR pathway activity in T lymphocytes had been heretofore lacking. Our findings suggest such a relationship as the cellular effects of TGF-β/Smad3 signaling resemble those of pharmacologic mTOR inhibition. Our data based on functional observations support the idea that the TGF-β/Smad3 pathway is a strong inhibitor of growth in co-stimulated naive CD4 T cells. Whether the TGF-β pathway interacts directly with the Akt-mTOR pathway or indirectly through upstream inhibition of CD28 signaling remains an open question. This negative interaction could also be transcriptionally regulated through the TGF-β/Smad3-dependent modulation of several genes. Our microarray data revealed that the TGF-β/Smad3 axis controls the expression of genes related to metabolic and growth pathways and also key immunity-related genes. As such, the overexpression of the IL-10ra gene in WT cells could contribute to the mitigation of CD28 signaling since IL-10 is known to prevent CD28 tyrosine phosphorylation and subsequent phosphatidyl-inositol 3-kinase recruitment and activation of the Akt-mTOR pathway.51, 52 Likewise, the overexpression of Forkhead box O3a (Foxo3a) in WT cells might also contribute to a cross-talk between the TGF-β and Akt-mTOR pathways as Foxo transcription factors are regulated by the Akt-mTOR pathway. However, it should be noted that the role of Foxo3a in T-cell tolerance and physiology is complex and remains controversial.23, 53 Interestingly, 4 out 40 TGF-β/Smad3 signature genes are related to lipid biology and metabolism (Apol9a, Apol9b, Acsbg1 and Cpt1c). Although conceptually related to the possibility that TGF-β is a regulator of metabolism in T cells, the modulation of lipid pathways is likely to contribute to several other aspects of T-cell biology such as mediator synthesis. Thus far, Cpt1c has been associated with IL-15-driven memory CD8 T-cell generation,54 but whether one could speculate about its role in naive CD4 T cells in relation with our finding that the transcript for this gene is negatively regulated by the TGF-β/Smad3 pathway is unclear at the moment.
Our finding that TGF-β impedes the effects of CD28 co-stimulation in a Smad3-dependent manner opens a new avenue to explain the tolerogenic effects of TGF-β in CD4 T cells. Indeed, the mitigation of co-stimulatory signals can explain how TGF-β increases the threshold for productive T-cell activation. It is known for example that TGF-β exposure can interfere with early steps in T-cell activation, notably by impeding calcium fluxes through an inhibitory effect on Itk phosphorylation, or by attenuating signaling downstream of the TCR.55, 56 Our results add to these previous studies by demonstrating a functional antagonism between the TGF-β/Smad3 pathway and CD28 signaling in T-cell activation. Lack of appropriate co-stimulation and the inability to mobilize the anabolic/glycolytic pathways can hamper T-cell activation and induce T-cell anergy. Accordingly, we propose that TGF-β-Smad3 signaling may maintain T-cell tolerance by suppressing co-stimulation-dependent mobilization of anabolic pathways.
Materials and methods
The 129-Smad3tm1Par/J strain was obtained from The Jackson Laboratory (Bar Harbour, ME, USA) and mice were housed under specific pathogen-free conditions. WT (Smad3+/+) and Smad3−/− mice were littermates obtained through breeding of heterozygous parents. All work involving mice was conducted under protocols approved by the Comité de Déontologie de l’Expérimentation sur les Animaux from the University of Montréal.
Cell sorting and RNA extraction and cDNA microarray
Cells were obtained from WT or Smad3−/− mice spleen and cervical lymph nodes. Single-cell suspensions were prepared by mechanical separation of the tissues and red blood cell lysis. CD4 T cells were then enriched using anti-CD19 and anti-CD8 beads (Milteny Biotec, Auburn, CA, USA) according to manufacturer’s instructions. Cells were stained with antibodies against TCR-β, CD19, CD4, CD8, CD44 and CD62L (BD Biosciences, San Jose, CA, USA). Naive cells were sorted on a FACS Aria (BD Biosciences) according to the following phenotype: TCR-β, CD4 and CD62L positive and CD19, CD8 and CD44 negative. After sorting, cells were left on ice for 2 h in serum-free conditions. Naive CD4 T cells were then plated in round-bottom 96-well plates at a concentration of 500 000 per ml (100 000 cells in 200 μl of serum-free X-Vivo-15 medium (Lonza, Basel, Switzerland) at 37 °C. Where indicated, TGF-β was added at a dose of 2.5 ng ml−1 and anti-CD3ɛ/anti-CD28 stimulation was done using plate bound anti-CD3 (10 μg ml−1) (BD Biosciences) and soluble anti-CD28 (1 μg ml−1) (eBiosciences, San Diego, CA, USA). After 6 h of incubation, cells were harvested, pooled according to genotype and condition and centrifuged for 5 min at 300 g. Supernatant was removed and RNA extraction was done using the Qiagen (Hilden, Germany) RNAeasy column mini kit according to manufacturer’s instructions. To achieve higher RNA concentration and purity, the RNA obtained from two to three different mice per genotype and condition was pooled and processed using the Qiagen micro kit, yielding one experimental replicate. A total of three experimental replicates per genotype and conditions were obtained. The integrity and concentration of RNA was then assessed using the Agilent 2100 bioanalyser (Agilent Technologies, Santa Clara, CA, USA). An RNA integrity number of 9.5 or more was deemed acceptable for reverse transcription. Five hundred nanograms of total RNA was reverse transcribed using oligo dT primers and SuperScript II (Invitrogen, Carlsbad, CA, USA) at 42 °C. Following purification (using Qiagen PCR purification column as per the manufacturer’s protocol), up to 1 μg of purified cDNA in a total volume of 40 μl was mixed with 40 μl of 5′ Cy3-labeled random nonamers for the test sample or 5′ Cy5-labeled random nonamers for the control sample (Trilink Biotechnology San Diego, CA, USA) and heated at 95 °C for 10 min After a quick chill in ice/water slurry for 10 min, the samples were mixed with 10 μl of 50 × dNTP mix (10 mM dATP, dTTP, dGTP, dCTP each in 1 × Tris-EDTA buffer), 8 μl of water and 2 μl of Klenow fragment (3′–5′ exo-) (NEB) and incubated at 37 °C for at least 2 h. The labeling reaction was then stopped using 10 μl 0.5 M EDTA, purified with 11 μl NaCl and 110 μl isopropanol, washed with 80% Ethanol and resuspended in 40 μl water. Following quantification by NanoDrop (NanoDrop products, Wilmington, DE, USA), 6 μg of Cy3-labeled material and 6 μg of Cy5-labeled material were resuspended in a total of 6 μl of water. The labeled material was mixed with 9 μl NimbleGen 2 × hyb buffer (Roche NimbleGen Inc., Madison, WI, USA), 3.6 μl NimbleGen Hyb component and 0.4 μl NimbleGen alignment oligo, for a total volume of 19 μl. The samples were heated at 95 °C for 5 min and transferred to 42 °C for at least 5 min before loading. Samples were hybridized overnight unto NimbleGen Mouse Gene Expression 385K Arrays at 42 °C in a Maui hybridization station with mixing. Arrays were scanned at 5 μm resolution on an Axon 4000B scanner using GenePix version 6.1 (Molecular Devices, Sunnyvale, CA, USA). Features were extracted using the Pair Reports function of NimbleScan version 2.5 (Roche NimbleGen Inc.).
Expression microarray data analysis
Robust Multi-Array analysis normalized data was log2-transformed and filtered. Features having an intensity above the background intensity in <3 samples (out of a total of 18 microarrays) were excluded. The data were then analyzed using Bioconductor packages (http://www.bioconductor.org/) and R statistical language (www.r-project.org). Using the R package limma, we fitted linear models to identify genes regulated between the different conditions.57 Features having a fold change greater than 1.5 or smaller than −1.5 and a P-value smaller than 0.05 were considered as significantly regulated based on previous studies using primary T cells.9, 58 Heat maps and hierarchal clustering were done using the MEV (MultiExperiment Viewer) package.59 The Panther tool (www.pantherdb.org) was used to categorize gene transcripts according to molecular function and biological process (Gene Ontology analysis). Gene Set Enrichment Analysis (www.broadinstitute.org/gsea) and Ingenuity Pathway analysis (Ingenuity Systems, www.ingenuity.com) were performed according to the providers instructions. Differentially expressed transcript lists included as Tables or Supplementary Material were manually curated to remove doublets (features corresponding to the same gene transcripts) and exclude annotations that were withdrawn or discontinued from the National Center for Biotechnology Information (NCBI) (last accessed in April 2012). The entire data sets have been deposited in NCBI’s Gene Expression Omnibus (GEO)60 and are accessible under the accession number GSE40494.
In vitro T-cell assays
Naive CD4 T cell sorting was performed with a BD Aria (BD Bioscience) according to the following phenotype: TCRβPosCD4PosCD8NegCD62LHighCD44Neg. T-cell stimulation was achieved with plate bound anti-CD3ɛ (10 μg ml−1) and soluble anti-CD28 (1 μg ml−1), in the presence or absence of TGF-β (2.5 or 5 ng ml−1) in complete RPMI medium. Rapamycin was purchased from Sigma-Aldrich (St Louis, MO, USA). CFSE labeling was performed according to manufacturer instruction (Invitrogen). At the end of the incubation period, cells were acquired on a FACS Canto (BD Biosciences). Cell death assessment was done using propidium iodine staining (BD Biosciences). The data were analyzed with the BD FACSDiva (V.6.1.3, BD Biosciences) and FlowJo (V7.2.5, TreeStar Inc., Ashland, OR, USA) softwares. FSC change was calculated as the percentage of cells in each division that reached a twofold increase in its FSC size when compared with non-divided cells. Cell proliferation was assessed by CFSE dilution and related calculations were done as previously described.35
CD4 T cells were lysed in SDS lysis buffer (50 mM Tris, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100, 5 mM NaF and 2 mM NaVO4, plus a cocktail of protease inhibitors and phosphatase inhibitors (Roche, Basel, Switzerland). Protein solutions were loaded on 12% SDS–PAGE, with each well of the same gel loaded with total cell lysate from an identical number of cells (500 000 or 750 000). After migration, transfer onto PVDF membranes (GE Healthcare, Piscataway, NJ, USA) was done. After 30 min blocking in 2% BSA in TBST (25 mM Tris–HCl pH 8.0, 150 mM NaCl, 0.05% Tween-20), Western blotting was done by overnight incubation at 4 °C in the presence of indicated primary antibodies. Protein staining of membrane was revealed by incubation with specific secondary antibodies for 2 h at room temperature, and then with ECL Immun-Star Chemiluminescent Detection kit (Bio-Rad, Hercules, CA, USA). Acquisition of .gel files was done with a Fujifilm LAS-4000 (Fujifilm Corporation, Tokyo, Japan). The relative level of each protein was quantified (GelQuant.NET software provided by biochemlabsolutions.com) by normalizing the intensity of each protein band to an internal calibrator sample (protein expression in the αCD3ɛ/CD28 treated Smad3−/−). Then, a ratio between normalized expression of the different proteins of interest over actin within the same sample was performed.
Gene Expression Omnibus
This work was supported through grants held by JSD (Leukemia/lymphoma society of Canada (LLSC) and the Maisonneuve-Rosemont hospital foundation) as well as by an LLSC grant to CP. JSD holds a career award from the Fonds de recherche du Québec-Santé (FRQS) and Claude Perreault holds a Canada research chair in immunobiology. We would like to thank the animal care facility personnel both at the IRIC and CR-HMR as well as Danièle Gagné, Gaël Dulude and Martine Dupuis for cell sorting.
About this article
Supplementary Information accompanies this paper on Genes and Immunity website (http://www.nature.com/gene)
Nature Communications (2017)