Correspondence to: B. R. Ludviksson, Department of Immunology, Landspitali University Hospital, Building #14 at Eiriksgata, 101 Reykjavik, Iceland. E-mail: firstname.lastname@example.org
The expression of the integrin αE (CD103), may enhance the retention of regulatory T cells to peripheral inflammatory sites and possibly contribute to their suppressive potential. The aim of this study was to define the regulatory role of IL-2 and TGF-β1 on the CD103 expression and the optimal in vitro conditions for the induction/expansion of human CD4+ and CD8+ Tregs. Cord blood mononuclear cells (CBMC) were stimulated under various culture conditions, including anti-CD3, anti-CD28, IL-2 and TGF-β1. TGF-β1 and IL-2 were both required for optimal expression of CD103. In addition, TGF-β1 and IL-2 synergistically induced CD103 expression on CD8+ T cells, whereas, only additive induced expression was noted on CD4+ T cells. Surprisingly, CD103 expression was not dependent upon CD28 costimulation. IL-2 also played a central role in CD103 expression by CD25hi Foxp3+ Tregs. IL-2, TGF-β1 and anti-CD3 defined the optimal stimulatory conditions favouring the induction/expansion of both CD4+ and CD8+ human Tregs from naive CBMC. Thus, this study provides new insights into the regulatory role of IL-2 upon CD103 expression by human cord blood CD4+ and CD8+ T cells. Furthermore, it identifies the in vitro culture conditions driving the differentiation of the novel phenotype CD4+ and CD8+ CD103+ CD25hi Foxp3+ Tregs from human CBMC.
Several different T cell subsets that possess regulatory properties have been described (Reviewed in [1, 2]). However, currently only the CD4+ CD25hi lineage expressing the transcription factor Foxp3 has been shown to be essential for the preservation of self-tolerance and immune homoeostasis [3, 4]. These CD4+ CD25hi Foxp3+ T cells are at present classified into thymic-derived natural Tregs (nTregs), peripherally induced Tregs (iTregs), and finally, the in vitro–induced iTregs [5, 6]. These three populations have been thoroughly studied in mice and humans over two decades, but due to discordant results, the exact nature of their stimulatory requirements and suppressive function is under intense investigation. The general consensus is that the differentiation of nTregs depends on strong TCR- and CD28-mediated activation [7-11], which is independent of TGF-β1 and IL-2 [12, 13]. However, more recent studies in mice show that when neither TGF-β1 signalling nor IL-2 is present, no nTregs are generated . In contrast, most studies suggest that iTregs require weak TCR-driven activation [15, 16] and are TGF-β1 and IL-2 dependent [5, 6]. Additionally, CD8+ CD25hi Foxp3+ T cells have been classified using similar criteria because they can be induced in vitro and have been detected in the thymus and within peripheral tissues [17-19]. Although, these CD8+ Tregs have been reported to possess similar regulatory properties as their CD4+ counterparts , the significance of this CD8+ Treg subset with regard to self-tolerance and balanced immune responses remains to be clarified. The therapeutic use of Tregs for the resolution of inappropriate immune responses is the optimal goal of most Treg studies, but several important questions remain to be addressed regarding their behaviour. One of the key elements in Treg function is their migration and retention at inflammatory sites. This has been addressed in murine studies showing that although the integrin CD103 is not necessary for the migration of T cells to sites of inflammation, it favours prolonged retention of Tregs at these sites [20, 21]. Therefore, these studies suggest that CD103 could be important to enable Tregs to maintain at local inflammatory sites.
CD103 is the alpha chain of the integrin αEβ7 and provides tissue retention at sites enriched in E cadherin, particularly at the epithelial lining of gut, lungs and skin but also at sites of inflammation . Besides participating in normal immune responses, CD103+ T cells have been associated with pathogenic inflammatory responses such as graft-versus-host-disease (GVHD)  and also in several autoimmune disorders such as inflammatory bowel disease, psoriasis and Sjögrens syndrome [23-25]. Furthermore, the expression of CD103 by T cells has been associated with a regulatory function of CD4+ and CD8+ T cells [26, 27]. As CD103 may provide retention at sites of inflammation, its expression by regulatory T cells could be beneficial for the sustained localisation of regulatory cells where their suppressive function is required. However, the microenvironment governing their simultaneous differentiation and CD103 expression remains to be elucidated. TGF-β1 is to date the best documented inducer of CD103 expression in T cells [28-30]. TGF-β1 may induce CD103 expression through Foxp3 , or through other mechanisms, which are currently unidentified for extrathymic T cells. Moreover, although IL-2 and stimulatory conditions have been shown to play an essential role in the TGF-β1-induced expression of Foxp3, there have been remarkably few studies on the regulatory role of these parameters upon TGF-β1-mediated expression of CD103. We and others have shown that the inhibitory effect of TGF-β1 upon IL-2 secretion is antagonized by CD28-mediated costimulation [32, 33], However, it is currently unknown whether the strength of the TCR/pMHC interaction and/or its costimulatory milieu is crucial in the induction of Tregs. In addition, it is important to define the stimulatory conditions that favour the induction of human CD8+ Foxp3+ Tregs, given their possible role in immune homoeostasis and autoimmunity. Finally, as recently pointed out by Rötzschke et al. murine and human Tregs differ significantly with regard to CD103 expression, and therefore, research on human subjects are needed. The aim of the present study was therefore to delineate the microenvironment promoting the induction of human iTregs from naive human cord blood T cells.
Material and methods
Isolation of mononuclear cells
The highest fraction found for naive undifferentiated human T cells is found amongst Cord blood mononuclear cells (CBMC). Therefore, cord blood samples were obtained from normal deliveries after receiving informed consents from their mothers. Mothers with history of autoimmune disorders were excluded. The study was approved by the Landspitali bioethics committee. Mononuclear cells were isolated from heparinised blood diluted in half by PBS and centrifuged over Ficoll (Sigma Aldrich, St Louis, MO, USA) at 1372 g at RT for 30 min.
2 × 105 CBMC were cultured per well in 96-well U-bottom plates for 3 days, unless otherwise stated, in the presence or absence of TGF-β1 (10 ng/ml) and/or IL-2 (100, 1000 or 10.000 pg/ml) in the serum-free medium AimV (Invitrogen Gibco, Grand Island, NY, USA). The cells were stimulated with 0.1 μg/well of immobilized anti-human CD3ε antibody (Clone UCHT1), with or without 1 μg/ml soluble anti-human CD28 antibody (Clone 37407). Cytokines and antibodies were purchased from R&D systems, Minneapolis, MN, USA.
Freshly isolated CBMC were stained with Glycophorin A-FITC ((JC159; DAKO, Glostrup, Denmark), CD3-FITC (SK7; BD Biosciences, San Jose, CA, USA), CD4-PE/-APC (11830, R&D systems) and CD8-PE/-PerCP, CD103-FITC (Ber-ACT8; BD Biosciences) CD49d-FITC (7.2; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), Integrin β7-PE (FIB504; BD Biosciences) and matching isotype controls. The cells were placed on ice for 20 min, and washed 1× with PBS, fixed with 0.5% formalin in PBS and stored at 4 °C until analysed. Cultured cells were washed with PBS, stained with CD103-FITC, CD49d-FITC, Integrin b7-PE, CD8-PE/PerCP, CD4-PE/APC and matching isotype controls as described before.
Cells were washed twice with 2-ml staining buffer (PBS with 2 mm EDTA, 0.5% BSA and 0.1% sodium azide) and stained for CD103-FITC, TβRII-PE (polyclonal Goat Ab; R&D systems), CD8-PerCP, CD4-APC and matching isotype controls at room temperature for 30 min and then washed twice again with staining buffer, fixed and stored as described before.
Cells were washed once with 1-ml PBS and stained with surface markers as described before, and then the cells were permeabilized and fixed with the Foxp3 staining buffer set (eBioscience, San Diego, CA, USA) and stained with Foxp3-APC (236A/E7-APC; eBioscience) or appropriate isotype control as described by the producer. The cells were analysed immediately after staining.
Definition of CD25hi
CD25hi were determined on the basis of the double-positive CD25 and Foxp3 expression as shown in Fig. 1.
Analysis of flow cytometric data
Ten to 20,000 cells were collected and analysed by flow cytometry (FACScan; BD Biosciences) using the programs CellQuest (BD Bioscience) and Flow Jo (Tree Star, Inc. Ashland, OR, USA).
Data are presented as mean values with SEM. The program GraphPad Prism (Version 5.03; GraphPad Sofware Inc, La Jolla, CA, USA) was used to analyse the data and to draw graphs. Data that were not normally distributed or with unequal variance were transformed when analysed with parametric tests. One-way analysis of variance (anova) was used to compare the means of the treatment groups. The Dunnett post-test was used to further analyse the statistical difference between the control and other groups. The two-way anova was used to determine whether two factors had a synergistic effect on a measured variable. The formal definition of synergy was used as presented by Slinker  (‘The biological response to both treatments A and B together will be greater than the sum of the response to treatment A alone and the response to treatment B alone’). Consequently, the null hypothesis can be formulated as: the combined effect of factor A and B is the same as the sum of the individual effect of A and B. The F statistic is computed for the interaction effect (non-additive effects) between two factors in a two-way anova and can therefore be used to test this hypothesis. The Bonferroni multiple comparison test was used to further analyse the statistical difference between selected groups. P < 0.05 was considered significant. When only two groups were compared, the Student′s t-test or the Mann–Whitney test was applied.
Phenotype of the CBMC
The phenotype of the isolated CBMC (n = 11) was as follows (mean% (SEM)): Glycophorin A (red blood cells) 6.52% (1.52), CD103+/CD4+ = 0.96% (0.29), CD103+/CD8+ = 8.55% (2.01), CD4+ CD49d+ = 9.31% (3.78), CD8+ CD49d+ = 13.43% (4.45).
IL-2 and TGF-β1 regulate the induction/expansion of CD103+ T cells
To determine the roles of TGF-β1, IL-2 and stimulatory conditions in the regulation of the induction/expansion of naive human CD103+ T cells, CBMC were isolated from cord blood and stimulated with plate-bound anti-CD3 with or without anti-CD28 in a serum-free medium. TGF-β1, IL-2 or both were added into selected cultures. The maximum dose of IL-2 was 10,000 pg/ml, which is comparable to the levels secreted by naïve human T cells stimulated with anti-CD3 and anti-CD28 in the presence of TGF-β1 under serum-free conditions . As shown in Fig. 2A, the addition of TGF-β1 or IL-2 alone had limited effects, whereas the addition of TGF-β1 and IL-2 combined demonstrated a significant effect on the induction of CD103+ cells within the CD4+ (P < 0.0001, Bonferroni's Multiple Comparison Test) and CD8+ (P < 0.0001, Bonferroni's Multiple Comparison Test) T cell populations. Furthermore, as shown in Fig. 2B, the induction of CD103+ CD4+ and CD103+ CD8+ T cells was evident through IL-2 in a dose-dependent manner. In order to determine whether the cytokines acted independently or synergistically, we analysed the data with a two-way anova. Interestingly, there was no synergistic interaction between IL-2 and TGF-β1 for CD4+ T cells stimulated with anti-CD3 alone (F = 1.59, P = 0.25), whereas a highly significant synergistic interaction was demonstrated for CD8+ T cells (F = 9.43, P < 0.0001). Notably, the increase in CD103 expression was more pronounced amongst CD8+ T cells compared with the CD4+ subpopulation (fold increase in CD103 expression: CD4+ = 26.9 ± 64.3; versus CD8+ = 80.5 ± 158.0; P = 0.03, Paired t-test). Interestingly, it was observed that costimulation with anti-CD28 reduced the effect of IL-2 and TGF-β1 on the CD103 expression for both phenotypes and abolishing the synergistic interaction between IL-2 and TGF-β1 amongst CD8+ T cells (F = 0.52; P = 0.48). It should be noted that the level of CD103 expression (MFI) did not increase following IL-2 and/or TGF-β1 treatment (Fig. 2B).
The integrins CD49d (α4) and CD103 (αE) can form functional heterodimers with the integrin β7. However, the addition of IL-2 and TGF-β1 alone or together did not affect the expression of CD49d, suggesting that CD49d and CD103 are regulated through different pathways (data not shown). Similarly, the expression of CD49d by CD25hi Foxp3+ Tregs was not induced on either CD4+ or CD8+ T cell subset in the presence of these cytokines (data not shown).
IL-2 and TGF-β1 induce/expand CD25hi Foxp3+ T cells
As IL-2 and TGF-β1 are critical for the induction and proliferation of CD4+ CD25hi Foxp3+ Tregs [5, 6], we next evaluated the induction of CD25hi Foxp3+ Tregs in response to the stimulatory conditions described above. As shown in Fig. 3, the combined addition of TGF-β1 and IL-2 during TCR-directed stimulation alone induced an eight-fold increase in CD25hi Foxp3+ CD4+ Tregs (P < 0.05, Dunnett's Multiple Comparison Test) and an eighteen-fold increase in CD25hi Foxp3+ CD8+ Tregs (P < 0.05, Dunnett's Multiple Comparison Test). However, IL-2 or TGF-β1 alone did not induce the differentiation of CD25hi Foxp3+ Tregs. Moreover, a two-way anova demonstrates that the observed increase in CD25hi Foxp3+ Tregs following the addition of both cytokines combined during TCR-induced stimulation alone, was not driven through a synergistic interaction (Interaction: CD4; F = 0.57; P = 0.46 and CD8; F = 0.74; P = 0.82). Finally, as shown in Fig. 3B, increased CD103 expression following IL-2/TGF-β treatment was only found on the CD4+/FoxP3+ supbopulation, but not on the CD4+/FoxP3−.
The expression of CD103 is induced by IL-2 and TGF-β1 on iTregs
It has been suggested that CD103 expression characterises subgroups of CD4+ and CD8+ Tregs [26, 27]. Therefore, we evaluated the regulation of CD103 expression amongst CD4+ and CD8+ CD25hiFoxp3+ Tregs. The addition of TGF-β1 or IL-2 alone did not significantly affect the expression of CD103 on CD25hi Foxp3+ Tregs. However, when added together, IL-2 and TGF-β1 significantly enhanced its expression when stimulated with anti-CD3 alone. For CD4+ T cells, the induction in CD103+CD25hi Foxp3+ was 26-fold, while it was 93-fold for CD8+ T cells (CD4+ versus CD8+, P = 0.05, Mann–Whitney test) (Fig. 4). The two-way anova demonstrated that IL-2 and TGF-β1 tended to have a synergistic effect upon the percentage of CD103+ CD25hi Foxp3+ for both phenotypes if stimulated through TCR alone (interaction: CD4 F = 3.44, P = 0.08 and CD8, F = 3.73, P = 0.07).
The expression of Foxp3 is predominantly expressed amongst the CD25hi CD103+ phenotype
CD25 is one of the key phenotypic markers for CD4+Tregs [37-39]. However, CD103+ has also been associated with Foxp3 expression in human CD4+ T cells regardless of their CD25 expression . Therefore, we next addressed the question which phenotype was most strongly associated with Foxp3; CD103+ CD25lo or CD103+ CD25hi. As shown in Fig. 5, the Foxp3 expression was predominantly associated with the CD103+ CD25hi phenotype, which was consistently more positive for Foxp3 than the corresponding CD103+ CD25lo phenotype. This difference was observed for both CD4+ and CD8+ T cells and was most pronounced when stimulated in the presence of both IL-2 and TGF-β1, regardless of costimulation. Moreover, the CD103+ CD25 hi phenotype also expressed significantly higher levels of Foxp3 (MFI) when stimulated with anti-CD3 in the presence of IL-2 and TGF-β1 (CD4+P = 0.006 and CD8+P = 0.007, Paired t-test, n = 5–6) and when stimulated with anti-CD3/CD28 (CD4+P = 0.001 and CD8+P = 0.03, Paired t-test, n = 6). In contrast, when CD103 expression was evaluated independently of CD25, CD103+ T cells expressed equal levels of FoxP3 as their CD103- counterparts (data not shown).
The effect of IL-2 and TGF-β1 upon TβRII expression
TGF-β1 signals through two different receptor units: TGF-β1 receptor type I (TβRI) and type II (TβRII) . Of these two, the TβRII has been shown to control the TGF-β1 responsiveness of T cells . Thus, we next evaluated the effect of IL-2 and TGF-β1 on the expression of TβRII. Although, TGF-β1 tended to suppress the expression at all time points after the first hour, the difference did not reach a statistical significance at any of the time points tested in three independent experiments (Fig. 6). Similarly, the addition of TGF-β1, IL-2 or both into cultures did not significantly affect the MFI of the expressed TβRII (data not shown). Therefore, the synergistic effect of TGF-β1 and IL-2 could not be attributed to changes in TβRII expression. However, due to beta error, it is conceivable that a larger sample size would alter our findings regarding the effect of IL-2 and TGF-β1 on the expression of TβRII. In contrast, the increased responsiveness of CD8+ T cells to IL-2 and TGF-β1 in terms of CD103 expression, was reflected by a significantly greater expression of TβRII by CD8+ T cells [TβRII MFI (SEM) CD8+ = 21.53 (±1.69) versus CD4+ = 19.75 (±1.58); P = 0.02, paired t-test, n = 6].
The use of in vitro–induced Tregs to resolve inflammation and to induce a long-term tolerance is the ultimate therapeutic goal for various inflammatory disorders. In this study, we demonstrate a novel regulatory role for IL-2 on TGF-β1-mediated ex vivo induction of CD103+ human T cells. In addition, it demonstrates a clear synergistic effect of IL-2 and TGF-β1 upon CD103 expression for CD8+ T cells and an additive effect for CD4+ T cells.
Furthermore, in this study, we define the optimal stimulatory conditions for the ex vivo generation of recently described and novel phenotype of iTregs involving both the CD4+ and CD8+ population characterised by CD103+/CD25hi/Foxp3+ expression from human CBMC. The maximal frequency of both lineages was obtained after TCR/CD3-induced activation in the presence of IL-2 and TGF-β1 without CD28-mediated costimulation.
In contrast to previous studies [25, 29, 42, 43], we observe that TGF-β1 alone is not sufficient for the maximal expression of CD103 due to its dependence on IL-2, although, TGF-β1 tends to induce its expression. This is more likely reflected by the fact that in all of these studies, IL-2 was included in the culture medium [25, 29, 42, 43]. Our findings comply with findings originating from murine T cells demonstrating that IL-2 has a critical role in CD103 expression . However, in contrast to our findings, the CD103 inducing effect of IL-2 on murine T cells was specific for CD4+, while CD8+ T cells were unaffected .
In our CBMC human system, CD8+ T cells preferably express CD103 and show more responsiveness towards the CD103 inducing effect of IL-2 and TGF-β1 than CD4+ T cells. Our findings are in agreement with previous studies showing that CD8+ T cells are more responsive towards the CD103 inducing effect of TGF-β1 than CD4+ T cells [25, 45]. This could be due to the observed difference in CD8+ versus CD4+ T cell TβRII expression. However, additional mechanisms could be involved such as the difference in TGF-β1-mediated signalling .
The regulation of CD103 expression is highly complex and, although, TGF-β1 has been shown to induce its expression, no TGF-β1 responsive elements have been found in the promoter region of the CD103 gene . We evaluated whether the IL-2 and TGF-β1-mediated induction of CD103 expression was associated with the TβRII. However, this was not found to be true for either phenotype (CD4+ versus CD8+). Therefore, the exact mechanism behind this effect remains to be found.
Our previous studies have shown that the suppressive effect of TGF-β1 upon IL-2 secretion is prevented by CD28-mediated costimulation . However, CD28-mediated costimulation in this study did not have any effect upon CD103 expression, and similar results have been found by others . Unlike murine CD4+ CD25- T cells that require both TGF-β1 and IL-2 for the induction of Foxp3[5, 48], human CD4+CD25-T cells have been shown to express Foxp3 in response to TCR-directed stimulation alone [49-55]. Moreover, the Foxp3 expression in murine T cells is consistently associated with a regulatory function, whereas it is not necessarily associated with a suppressive function in human CD4+ T cells . Two recent studies have reported that human CD8+ T cells express Foxp3 in response to anti-CD3 stimulation alone [49, 57]. However, this expression was only associated with a regulatory function in one of these studies. The characteristic CD25hi Foxp3+ iTreg phenotype of the CD103+ CD8+ population induced in our study may suggest that an iTreg population with a regulatory function was induced.
Our findings strongly support the concept that IL-2 and TGF-β1 are not only necessary for the in vitro differentiation of murine iTregs, but also for naïve human iTregs. More importantly, this was observed not only for the CD4+ but also naïve human CB CD8+T cells. In accordance with the findings of Tran et al., we observed that the greatest increase in Foxp3 expression by both phenotypes was achieved by the addition of exogenous TGF-β1 and IL-2.
In this study, we demonstrate that a significant proportion of cord blood CD4+ and CD8+ T can be induced to become CD103+ CD25hi Foxp3+. However, this was most apparent amongst CD8+ iTregs (~25%) compared with CD4+ iTregs (~5%). The functional relevance of CD103 expression by human Tregs is uncertain. CD103 is rarely expressed by human Tregs in CB or adult PB and its expression has not been associated with an effector/memory-like activation status as in mice . Murine Tregs expressing CD103 have been shown to be more potent suppressors than their CD103-counterparts [59, 60]. However, less is known regarding the relevance of CD103 expression by human T cells. As the primary effect of CD103 expression is to provide enhanced retention to peripheral sites, its expression on iTregs is likely to promote their role at inflammatory sites in peripheral tissues. As CD103 has been shown to contribute directly to the cytotoxic function of CD8+ effector T cells, it is possible that CD103 also has a functional relevance for the polarisation of molecules involved in the regulatory function of human CD8+ CD25hi Foxp3+ Tregs through a cell–cell contact mechanism as described by others [17, 62].
This study demonstrates a novel role of IL-2 for the TGF-β1-induced generation of CD103 expression by human CB CD4+ and CD8+ T cells. Furthermore, it defines the optimal in vitro culture conditions driving the induction of CD103+ CD25hi Foxp3+ CD4+ and CD8+ Tregs from human CBMC. Thus, our study provides further and novel insights into the generation of CD103+ iTregs that could be utilised for future therapeutic modalities targeted at various inflammatory diseases.
This work was supported by the Landspitali University Hospital Research Fund and the University of Iceland Research Fund. The authors thank Dr. Maren Henneken and Dr. Ingileif Jonsdottir for critical reading of the manuscript.