Blood endothelial cells (ECs) act as gatekeepers to coordinate the extravasation of different T cell subpopulations. ECs express defined panels of adhesion molecules, facilitating interaction with blood circulating T cells. In addition to the mere adhesion, this cellular interaction between ECs and transmigrating T cells may also provide signals that affect the phenotype and function of the T cells. To test the effects of ECs on regulatory T cells (Treg) we set up cocultures of freshly isolated murine Treg and primary ECs and assessed the phenotype and function of the Treg. We show that Treg upregulate programmed death-1 (PD-1) receptor expression, as well IL-10 and TGF-β secretion after contact to ECs. These changes in phenotype were accompanied by an increased suppressive capacity of the Treg. Blockade of the PD-1 and/or the IL-10 secretion in the in vitro suppression assays abrogated the enhanced suppressive capacity, indicating relevance of these molecules for the enhanced suppressive activity of Treg. In aggregate, our data show, that ECs increase the immunosuppressive potential of activated Treg by upregulation of PD-1 and stimulation of the production of high levels of IL-10 and TGF-β. Therefore, one can speculate that Treg during transendothelial transmigration become “armed” for their suppressive function(s) to be carried out in peripheral tissues sites.
Naturally occurring CD4+CD25+Foxp3+ regulatory T cells (Treg) are well characterized by their capacity to suppress proliferation of conventional T cells in vitro and protect humans and animals from severe autoimmunity (1, 2). Thus, depletion of Treg in healthy animals induces development of inflammatory bowel disease and diabetes and exacerbates experimental autoimmune encephalomyelitis (EAE) in immunized mice (3–6). Studies with scurfy mice lacking the forkhead transcription factor Foxp3, which plays a crucial role for Treg development, suffer from multiorgan inflammatory disease that can be overcome by reconstitution of the mice with Treg (7, 8). Overall, these data demonstrate that Treg are required for the induction and maintenance of peripheral tolerance.
Treg suppress activation of conventional CD4+ T cells (Tconv) via different mechanisms. For instance, it is known that most Treg subpopulations suppress immune responses by inhibitory cytokines, such as IL-10 and TGF-β. Despite their ability to suppress immune responses by soluble factors it is well documented that Treg suppress by cell–cell contact via, namely, CTLA-4 and LAG-3 (9). Another molecule that is proposed to be involved in the cell-contact dependent immunosuppression is the programmed death-1 (PD-1) receptor. PD-1 is expressed on TCR ligation on activated Treg, but also on conventional T cells albeit in lower amounts (10). Humans and animals with a deficiency in PD-1 expression are susceptible to autoimmune diseases (11, 12). Notably, PD-1−/− T cells have been found to be incapable to control autoimmunity. Moreover, blockade of PD-1/B7-H1 interactions have been shown to accelerate the outcome of several autoimmune diseases, such as EAE, diabetes, and colitis (13, 14). Likewise, PD-1/B7-H1 interactions are critically involved in the suppression of antiviral T cell activity (15). Overall, the PD-1/B7-H1 pathway is expected to be a key regulatory mechanism for immunosuppression in the periphery.
Treg meet their respective MHC–Ag complexes presented by APC in secondary lymphoid organs resulting in Treg activation and de novo expression of tissue-specific adhesion molecules and chemokine receptors (16). These molecules enable Treg to enter inflammatory sites in peripheral tissues. Recently, it has been shown that reconstitution of scurfy mice with FuT7−/− Treg that inefficiently migrate into the skin, also failed to suppress cutanous autoimmune-inflammation (17). Moreover, accumulation of Treg was found at sites of chronic Leishmania major infection and in tumors, where these cells suppress parasite- and tumor-specific immune responses (18, 19). Thus, these data show that tissue-homing of Treg is critically involved in the progression of immune responses.
To exert their immunosuppressive functions in the periphery, Treg home into inflamed tissues. This process is mediated by endothelial cells (ECs) via adhesion molecules. Th1/Th2 cells secrete cytokines enhancing expression of adhesion molecules on ECs and as a consequence, the cells are recruited to the site of inflammation (20, 21). In addition to this well-characterized crosstalk between EC and T cells, activated ECs also express MHC molecules and several T cell costimulatory molecules and therefore are able to stimulate proliferation and cytokine production of murine CD8+ T cells (22). Proliferation of murine CD4+ T cells is not induced by ECs (23). Instead, ECs convert conventional CD4+ T cells into Treg (24). These data show that activated ECs not just facilitate T cell transmigration but also affect the polarization of immune responses in the tissue. In contrast, the effects of ECs on Treg function remains elusive. In this report, we investigated the role of ECs on the regulation of Treg phenotype and suppressor capacity. In this study, we show that Treg upregulate PD-1 expression and produce significantly increased levels of IL-10 and TGF-β after contact to ECs. These changes in phenotype were accompanied by a significant increase of the suppressive capacity of Treg that is primarily mediated by upregulation of the suppressive molecule PD-1. Because Treg interact with ECs during extravasation from the blood into the inflamed tissue, the ability of ECs to increase Treg function might be an important mechanism to augment the immunosuppressive function of Treg in peripheral organs.
Materials and Methods
BALB/c mice were purchased from Charles River Laboratories (Sulzfeld, Germany) and housed at the animal facility of the University of Heidelberg. Sex-matched mice between 6 and 10 wk of age were used in all experiments. All experiments were performed according to institutional guidelines.
Reagents and mAb
Cell culture media and supplements were purchased from PAA Laboratories (Cölbe, Germany), except indicated reagents. The following Abs were used in this study: anti-CD4 FITC, anti-Foxp3 FITC staining kit, anti–PD-1 FITC, biotinylated anti–PD-1, rat IgG2b FITC, anti-CD105 (unlabeled), blocking anti–B7-H1, neutralizing anti–IL-10, and respective isotype controls rat IgG1, IgG2a, and IgG2b4 was obtained from Vector Laboratories (Burlingame, CA).
Purification of CD4+CD2− T cells and CD4+CD25+ Treg
T cell subpopulations were isolated from lymph nodes and spleen of untreated BALB/c mice by magnetic microbeads (Miltenyi Biotech, Bergisch Gladbach, Germany). First, CD4+ T cells were negatively enriched using the Miltenyi untouched T cell kit according to manufacturers instructions. Purity of CD4+ T cell population was judged by FACS staining (>95%). In a second step, CD4+ T cells were further separated into CD4+CD25− and CD4+CD25+ subpopulations. CD4+ T cells were incubated with PE-labeled anti-mouse CD25 and anti-PE microbeads. CD4+CD25+ Treg were retained in MIDI-MACS columns. Conventional CD4+CD25− T cells were concentrated in the flow-through. Purity of Treg and Tconv populations was judged by FACS staining. Treg and Tconv were cultured in RPMI 1640 supplemented with 10% (v/v) FCS, 1% penicillin/streptomycin (v/v), 1% l-glutamine (v/v), HEPES (v/v), and 5 nM β-mercaptoethanol (v/v) (Sigma-Aldrich, Deisenhofen, Germany). For all experiments, the regulatory function of CD4+CD25+ T cells was analyzed by their ability to suppress proliferation of CD4+CD25− T cells in the presence of 0.5 μg/ml soluble anti-mouse CD3 and CD28 mAbs.
Purification of ECs from lung
The ECs were isolated from lungs of BALB/c mice as described previously (25). Briefly, lungs were harvested, minced into small pieces, and digested with collagenase A (CellSystems, St. Katharinen, Germany) for 1 h at 37°C. To obtain a single-cell suspension, an additional digestion step with trypsin/EDTA for 10 min at room temperature (RT) was added. After thoroughly washing, 1 × 107 cells/ml were incubated with 10 μg/ml mouse immunglobulin (Dianova) for 30 min at 4°C. For positive enrichment of ECs, the lung cells were incubated with saturating amounts of anti-mouse CD31, anti-mouse CD105 mAb, and biotinylated isolectin B4
Ab-staining and flow cytometry of Treg and EC
For FACS analysis 1 × 105 T cells or ECs were resuspended in PBS and 1% FCS (v/v) and incubated with combinations of the following mAb (dilution): anti-CD4 FITC (1:200), anti–PD-1 FITC (1:200), anti–PD-1 bio (1:500), rat IgG2b (1:200), anti-CD4 APC (1:400), anti-CD25 PE (1:400), anti-CD31 PE (1:100), anti–ICAM-1 FITC (1:200), and anti–VCAM-1 FITC (1:200). For analysis of surface markers, cells were incubated with the indicated mAb for 30 min at 4°C. After washing, the cells were resuspended in PBS and 1% FCS (v/v). For detection of PD-1 by biotinylated mAb, cells were incubated with SA-APC (1:1000) after binding of PD-1 for additional 30 min at 4°C prior analysis. For indirect staining, the cells were incubated for 30 min at 4°C with blocking rat serum after binding of the unlabeled first mAb. Thereafter, the cells were incubated for 30 min at 4°C with a FITC- or PE-labeled secondary mAb (both 1:400). For discrimination between live and dead cells, 1 × 105 cells were incubated with 3 μl 7-aminoactinomycin D (7-AAD; Beckman Coulter) for 30 min at RT prior analysis. Intracellular staining of Foxp3 was performed using the Foxp3 staining kit according to the manufacturer’s instructions. For intracellular IL-10 staining Treg and ECs were treated with golgi-plug (final dilution 1:1000, BD Pharmingen, Heidelberg, Germany) during overnight coculture. Treg were harvested and incubated with FITC-labeled mAb directed to cell surface PD-1. Afterward the cells were fixed with 4% (w/v) paraformaldehyde (Sigma-Aldrich) for 15 min at RT. Fixed cells were incubated for 30 min with biotinylated anti–IL-10 mAb or isotype control mAb, both diluted 1:200, in PBS supplemented with 1% FCS (v/v), 0.2% saponin (w/v) (Sigma-Aldrich), and 1 μg/ml goat-serum (Dianova). After washing, the cells were incubated with SA-APC in a final dilution at 1:1000 in PBS and 1% FCS (v/v) and 0.2% saponin (w/v) for 30 min. After washing, the cells were resuspended in PBS and 1% FCS (v/v). In all experiments cells were analyzed using FACS-Canto flow cytometer and DIVA software.
ECs were detached from culture flasks by adding a PBS-based enzyme-free dissociation buffer (Life Technologies). The cells were incubated with the buffer at 37°C. Every 5 min, the degree of dissociation of the cells was controlled microscopically. Complete detachment of the cells was achieved after 10–15 min. Thereafter, the reaction was stopped by adding EC culture medium and the cells were washed thoroughly. A suspension of 1 × 105reg and Tconv were isolated as described previously. Immediately after isolation, 1 × 106/ml cells were activated with 10 μg/ml plate-bound anti-mouse CD3 mAb and 0.5 μg/ml soluble anti-mouse CD28 mAb. After 24 h, the cells were harvested and washed thoroughly. The 5 × 105 activated Treg or Tconv were cocultured with 1 × 105 previously activated ECs overnight in RPMI 1640-based culture medium. After overnight coculture, T cells were harvested and culture supernatants were collected for further analysis. For contact-independents cocultures, 5 × 105 Treg were cultured in 100 μl medium in 0.4 μm transwell inserts (Corning, Amsterdam, The Netherlands) and 1 × 105 ECs were cultured in 400 μl medium in the well. For neutralization experiments, cocultures were performed in the presence of 10 μg/ml anti-human cross-reactive TGF-β mAb or 5 μg/ml anti-mouse IL-10 mAb. In parallel, isotype-matched control mAb were used to assess specificity of the neutralizing mAb. Complete neutralization of cytokines was assessed by ELISA of culture supernatants. For blocking of PD-1/B7-H1 interactions during cocultures, ECs were incubated with 10 μg/ml anti-B7-H1 or isotype control mAb prior coculture with 5 × 105 Treg.
T cell proliferation assay
EC-exposed Treg were cultured with 1 × 105 freshly isolated CD4+CD25− T cells at the indicated amounts in the presence of anti-CD3/CD28 (both 0.5 μg/ml, soluble). After 3 d, the cocultures were pulsed with 1 μCi 3H-thymidine (Amersham, Germany). After additional 16 h, the cultures were harvested and incorporation of radioactivity was determined by liquid scintillation using a Trilux Betaplate reader (Wallace, Turku, Finland). For neutralization experiments, cocultures were performed in the presence of 10 μg/ml anti-human cross-reactive TGF-β mAb or 5 mg/ml anti-mouse IL-10 mAb. In parallel, isotype-matched control mAb were used to assess specificity of the neutralizing mAb. For blocking of PD-1/B7-H1 interactions during suppression assays, 1 × 107/ml CD4+CD25− T cells were preincubated with 10 μg/ml anti–B7-H1 or isotype control mAb. After thoroughly washing, 1 × 105 preincubated Tconv were cultured with indicated amounts of EC-exposed Treg.
After overnight culture of ECs with Treg or Tconv supernatants were collected. Secretion of IL-10 (BD Bioscience) and TGF-β11, the cytokine was acid activated by 1 N HCL prior assay.
Paired Student t tests were used for statistical analysis.
Characterization of lung-derived primary murine ECs
Primary ECs were isolated from lungs of BALB/c mice using magnetic beads. EC cultures showed the typical cobblestone morphology and were able to take up LDL (Fig. 1A). FACS analysis (Fig. 1B) revealed that these cells express the EC marker CD105, isolectin B4, and as previously described, low levels of CD31. Furthermore, isolated EC expressed adhesion molecules ICAM-1 and VCAM-1 and failed to express the common leukocyte Ag CD45 (data not shown). As the expression of surface markers remained stable during passages 5–12, for subsequent analysis of Treg–EC interactions ECs from passages 7–11 (Fig. 1B) were used.
ECs affect function and cytokine release of Treg
To identify effects of ECs on Treg function, we activated EC with TNF-α and IFN-γ (100 and 500 ng/ml) and Treg with anti-CD3 and anti-CD28 (10 and 0.5 μg/ml) Abs overnight. Activated Treg and ECs were cocultured for 12 h. Thereafter, Treg were separated from EC cocultures and their capacity to suppress proliferation of conventional CD4+CD25− T cells (Tconv) was assessed. As expected, activated Treg strongly suppressed proliferation of syngenic Tconv in a dose-dependent manner (Fig. 2A). Interestingly, Treg that had been cultured with activated ECs, suppressed proliferation of CD4+CD25− T cells even more potently as compared with activated Treg that were not in contact with ECs (n = 6; ***p < 0.001). To further investigate whether activation of the ECs is mandatory to increase the suppressive function of Treg, we cocultured Treg with nonactivated (Fig. 2A) or partially activated ECs using either TNF-α or IFN-γ alone (data not shown). In this study, we recorded an increased suppressive capacity as compared with untreated ECs. However, this increase was minor as compared with results obtained with fully (TNF-α and IFN-γ) activated ECs. Therefore, in further experiments, we always used fully activated ECs.
To exclude that the observed differences in the suppressive capacity of the Treg are due to cell death, the number of dead cells was assessed by FACS analysis of 7-AAD+ cells. No substantial differences in the number of 7-AAD+ Treg were detectable after culture of the Treg cells with ECs or medium alone (Fig 2B). Thus, these data indicate that ECs augment the suppressive capacity of activated Treg.
Activated Treg are known to produce the immune modulatory cytokines IL-10 and TGF-β that are involved in suppression of CD4+ T cell responses (9). To analyze whether ECs affect the production of these cytokines by Treg, we cultured activated Treg together with activated ECs or medium alone overnight and measured IL-10 and TGF-β in the tissue culture supernatants by ELISA. After coculture of Treg, together with ECs, significantly higher concentrations of IL-10 (137% increase) were detectable in the tissue culture supernatants (n = 4; **p = 0.007), as compared with the supernatants of activated Treg alone (Fig. 2C, left). Moreover, this increased IL-10 production was contact dependent on cellular contact between Treg and ECs, as IL-10 production of Treg separated from ECs by semipermeable membranes, was not significantly increased as compared with control Treg (n = 4) (Fig. 2C, right). To further test whether engagement of the TCR by a cognate MHC/peptide complex is necessary to trigger increased IL-10 release, we cocultured Treg purified from mice harboring an OVA-specific transgenic TCR (DO11.10), together with MHC class II-matched, EC (BALB/c), and tested the tissue culture supernatants for IL-10. Similar to results obtained with regular BALB/c Treg, we measured increased levels of IL-10 (Supplemental Fig. 1). Thus, as the Ag recognized by the DO11.10 T cells, namely, OVA, was not present in the cultures, these data indicate that IL-10 release by Treg is induced independently from cognate MHC/TCR interactions.
For TGF-β (Fig. 2D), we also show increased levels in cocultures of Treg and ECs (n = 4; **p = 0.012). But in contrast to IL-10, the increased TGF-β production was independent from cell-to-cell contact, as high levels of this cytokine were also detected in tissue culture supernatants of Treg-EC cocultures, which were setup in transwell assays. Therefore, TGF-β production, as opposed to IL-10 release, is likely induced by soluble substances.
The increase of both, IL-10 and TGF-β production, was specific for Treg-EC cocultures, because in the tissue culture supernatants of activated ECs cultured together with activated Tconv no increased levels of IL-10 and TGF-β were detectable (Fig. 2C, middle; Fig. 2D, right). Thus, activated EC affect the production of the inhibitory cytokines IL-10 and TGF-β by activated Treg, but not by Tconv.
ECs trigger upregulation of PD-1 expression by Treg
We next investigated the phenotype of activated EC-exposed Treg. To this end, we cultured activated Treg together with activated ECs for 12 h and analyzed the cells by FACS. Treg were identified by being Foxp3+CD25+ and were further characterized by the expression of activation markers, adhesion molecules, and costimulatory molecules.
In this study, we were not able to detect increased expression of CD69, CD62L, LFA-1, and ICAM-1 (data not shown). In contrast, changes in surface expression of the inhibitory molecule PD-1 on Treg were detectable (Fig. 3A, 3B, left panel). We observed only low expression levels of PD-1 on freshly isolated resting Treg that were mildly upregulated on activation with anti-CD3/CD28 mAb. Notably, during exposure to ECs, activated Treg increased expression of PD-1 by 81% as shown for mean fluorescence intensity (MFI) (n = 5; **p = 0.006). No upregulation was detectable when activated Treg were cultured overnight in the absence of ECs.
CD4+CD25− conventional T cells also upregulated expression of PD-1 on activation with anti-CD3/CD28 mAb, but the overall PD-1 was less pronounced when compared with Treg (Fig. 3B, right panel) and in contrast to Treg, Tconv failed to increase PD-1 expression on coculture with ECs. These data show that ECs selectively upregulate PD-1 expression on activated Treg.
Further analysis of the regulation of PD-1 expression by Treg during coculture with ECs revealed that PD-1 was upregulated in a contact-independent fashion, as determined in semipermeable transwell systems (n = 4; **p = 0.008) (Fig. 3C, left). Furthermore, PD-1 was significantly upregulated after culture of Treg with tissue culture supernatants of activated ECs. This effect was comparable to upregulation of PD-1 observed during Treg-EC cocultures (n = 4; *p = 0.03) (Fig. 3C, right). Thus, these observations demonstrate that ECs release soluble factor(s) capable of upregulating PD-1 expression. To exclude that traces of TNF-α and IFN-γ, which used for EC activation, were responsible for the increased PD-1 expression, we cultured Treg with these cytokines overnight. No upregulation of PD-1 was detectable after these cultures. Thus, substances other than TNF-α and IFN-γ, produced by the activated ECs trigger upregulation of PD-1 on Treg.
Several reports demonstrated a role for PD-1 in regulating the susceptibility of cells for apoptosis (26). Thus, to exclude that the observed increased suppressor capacity of the EC-exposed Treg might be due to the release of soluble factors during apoptosis, we compared the number of annexin-positive cells within the PD-1low and PD-1high expressing Treg population. Comparable amounts of annexin-positive cells within both Treg subpopulations (∼8%) were detectable by FACS analysis (Fig. 3D). Furthermore, the overall amount of apoptotic cells was very low compared with “heat-shocked” controls that had been incubated at 42°C for 30 min. Altogether these findings suggest that activated ECs modulate the phenotype and function of activated Treg, without affecting the viability of the cells.
The suppressor capacity of EC-exposed Treg is increased by means of IL-10 secretion and PD-1 expression
To analyze the potential role of the immunosuppressive molecules IL-10, TGF-β, and PD-1 in the increased suppressor capacity of EC-exposed Treg, we measured the suppressive capacity of EC-exposed Treg in conventional suppression assays in the presence of neutralizing anti–IL-10, anti–TGF-β, and blocking anti–B7-H1 Abs.
For blocking of PD-1/B7-H1 interactions, we preincubated Tconv with an anti-mouse B7-H1 mAb (10 μg/ml) prior coculture with EC-exposed Treg (Fig. 4A). In this study, we show that blockade of PD-1/B7-H1 interaction resulted in a reduction of the suppressor capacity of the Treg (n = 4; **p = 0.006). In the presence of blocking B7-H1 mAb, EC-conditioned Treg were no longer able to suppress T cell proliferation in a Treg/Tconv ratio of 1:256. This effect was specific for blocking of PD-1/B7-H1 interactions, because pretreatment of the cells with an IgG2b isotype control mAb did not alter the suppression. However, although significant effects of anti–B7-H1 Abs on Treg mediated suppression were only obtained at a Treg/Tconv ratio of 1:256, in all experiments performed we also saw a reduced suppression at higher (1:128) Treg/Tconv ratios, supporting our hypothesis that B7-H1 expression contributes to the increased suppressive capacity of Treg. When neutralizing IL-10, we also recorded a reduced suppressive capacity of Treg (43 ± 14% suppression in presence of anti–IL-10 versus 75% suppression in controls. At a Treg/Tconv ratio of 1:256; n = 4). However, this reduction was not significant (n = 4) and less pronounced when compared with blockage of the PD-1/B7-H1 interactions during the suppression assays (n = 4) (Fig. 4B). Finally, neutralization of TGF-β had no effect on the suppressor capacity of EC-exposed Treg (n = 4) (Fig. 4C). Therefore, we conclude that the augmented expression of the inhibitory molecule PD-1 on EC-exposed Treg attributes to the increased suppressive capacity of the Treg. In contrast, IL-10 plays a minor role in affecting the suppressive capacity of the EC-exposed Treg.
To further evaluate a possible interrelation between PD-1 and IL-10 during the induction of the increased suppressive capacity of EC-exposed Treg, we analyzed EC-exposed Treg for coexpression of PD-1 and intracellular IL-10 by flow cytometry. Costaining of EC-exposed Treg for PD-1 and IL-10 revealed that Treg producing substantial amounts of intracellular IL-10 also express high amounts of PD-1 (Fig. 4D). Thus, EC promote development of activated Treg that coexpress high levels of PD-1 and IL-10. Therefore, these molecules might have a synergistic effect on the increased suppressor capacity of EC-exposed Treg. To test this hypothesis, we performed suppression assays with EC-exposed Treg in the presence of B7-H1 blocking mAb together with neutralizing anti–IL-10 mAb. For blocking both, the PD-1/B7-H1 interaction as well as the effects of IL-10, Tconv were preincubated with anti–B7-H1 and cocultured together with EC-exposed Treg in the presence of neutralizing IL-10 mAb as described previously. This combined blockage resulted in an augmented reduction of the suppressor capacity of EC-exposed Treg compared with individual blockade of either PD-1/B7-H1 or IL-10 alone (Fig. 4E). In this study, the suppressor capacity of EC-exposed Treg was significantly reduced at a Treg to Tconv ratio of 1:64 (33 ± 11% inhibition; **p < 0.001). In contrast, blocking of either PD-1/B7-H1 or IL-10 alone (at this Treg to Tconv ratio) had no effect on the suppressive capacity of the EC-conditioned Treg. Thus, these data show that EC-exposed Treg suppress the proliferation of Tconv by both IL-10 as well as PD-1 and thereby PD-1 provides a dominant regulatory signal.
ECs stimulate PD-1 upregulation, IL-10, and TGF-β production of Treg by distinct mechanisms
Our observation, that EC-exposed Treg upregulate PD-1 expression and produce high amounts of IL-10 and TGF-β, prompted us to analyze whether expression of these molecules by Treg is regulated via a positive feedback loop. Because PD-1 is upregulated on EC-exposed Treg in a contact-independent manner, we asked whether the soluble factors IL-10 and TGF-β produced by EC-exposed Treg are involved in upregulation of PD-1 expression. Therefore, we neutralized these cytokines during overnight Treg-EC cocultures as described previously and analyzed PD-1 expression on Treg by FACS. Blockade of IL-10 as well as of TGF-β did not affect upregulation of PD-1 expression on EC-exposed Treg (Fig. 5A, 5B).
Because ECs upregulate B7-H1 cell surface expression on stimulation with IFN-γ we next asked whether interactions between PD-1 expressed on Treg and B7-H1 expressed on ECs might be involved in the increased production of IL-10 by the Treg. To this end, ECs were preincubated with 10 μg/ml anti–B7-H1 or the respective isotype control mAb prior overnight coculture with activated Treg. Thereafter, the culture supernatants were analyzed for IL-10 by ELISA (Fig. 5C). These experiments show that IL-10 production of Treg was not affected by PD-1/B7-H1 interaction between Treg and ECs during coculture.
Taken together, these data demonstrate that upregulation of PD-1 as well as production of IL-10 and TGF-β by Treg are independent of one another. Thus, we suggest that EC stimulate PD-1 expression as well as IL-10 and TGF-β by Treg via distinct mechanisms.
Accumulating data demonstrate that Treg not only suppress T cell responses in secondary lymphoid organs but also in inflamed tissues. To do so, Treg colocalize with effector T cells (Teff) and dendritic cells (DCs) at the site of inflammation and inhibit Teff function and DC migration into the draining lymph nodes (27, 28). In addition, we and others have shown that Treg suppress transendothelial migration of CD8+ Teff cells in a model of contact hypersensitivity in vivo and suppress the transendothelial migration of PBMC from gastric carcinoma patients in vitro (29, 30). Treg by themselves migrate more efficiently through autologous endothelium when compared with Tconv, which has been shown in vitro for human pancreatic carcinoma patients (31). Altogether, these studies strongly suggest that ECs in addition to professional APC influence Treg function(s) in the periphery. In our current study, we demonstrate that primary ECs contribute to the immunoregulatory capacity of Treg by increasing their suppressive potential. Thereby, EC stimulate the production of high amounts of IL-10 and TGF-β by activated Treg and trigger upregulation of PD-1 cell surface expression.
The suppressive cytokine TGF-β has been implicated in the suppressor capacity of Treg in several in vivo models. But blocking of TGF-β did not affect Treg suppressor capacity in vitro. Thus, the role of this cytokine for in vitro suppression of T cell proliferation is discussed controversially (32). Nevertheless, one well-characterized function of TGF-β is its ability to convert Tconv into Treg (9, 19). Recent observations from Zhang et al. propose that TGF-β produced by Treg in combination with IL-10 reduces the migration of DCs from inflamed tissues into peripheral lymph nodes (27). In accordance with our observation that Treg produced elevated amounts of TGF-β and IL-10 during coculture with ECs, one can speculate that ECs, by inducing production of these inhibitory cytokines in Treg, contribute to this inhibition of DCs efflux out of inflamed tissues. Therefore, despite the fact that TGF-β produced by Treg during exposure to ECs plays only a minor role in directly affecting the Treg suppressor capacity in vitro, it might play an important role for the generation/maintenance of an immunosuppressive environment during inflammation in vivo by suppressing the efflux of activated DCs.
In addition to the increased release of soluble factors, we also observed an upregulation of PD-1 on Treg after exposure to ECs. In several animal models, the PD-1/B7-H1 pathway has been identified to play a key role for the suppression of Teff functions in the periphery. For example, in animal models of autoimmune diabetes, blocking of interactions between PD-1 expressed by self-reactive T cells and B7-H1 expressed by islet cells inhibited the onset of disease (33–35). In addition, disease progression of EAE was increased after blocking of B7-H1 expressed by inflammatory cells, astrocytes, vascular ECs, and microglia within the CNS (36, 37). Despite these studies on the role of PD-1 on Teff, functional analysis of PD-1 signaling in Treg has been shown that the suppressive capacity of these cells is at least in part dependent on their PD-1 expression (38, 39). That is, Treg isolated from PD-1−/− mice possess a reduced suppressor capacity (40). Furthermore, involvement of the PD-1/B7-H1 pathway has been shown for Treg-mediated immunosuppression after allogenic skin transplantation (41) and also for the reduction of tumor immunity in vivo (42, 43). In accordance with our observation that ECs possess the capacity to mediate upregulation of PD-1 on Treg and more importantly that PD-1/B7-H1 interactions were primarily responsible for the increased suppressor capacity of the Treg, we conclude that ECs augment the Treg suppressor capacity by mediating upregulation of PD-1 on Treg during transendothelial migration into inflamed tissues in vivo.
Recently, it has been shown that simultaneous blockade of PD-1/B7-H1 and IL-10R increases the frequency, proliferation rate, and IFN-γ production of lymphocytic choriomeningitis virus-specific CD8+ T cells (44). Accordingly, combinatory blockade of PD-1/B7-H1 interactions and IL-10 inhibited T cell proliferation by EC-exposed Treg more effectively as compared with PD-1/B7-H1 blockade alone. In contrast, neutralizing of IL-10 alone had only minor effects on the suppressor capacity of the EC-exposed Treg in vitro. Thus, EC-exposed Treg inhibited proliferation of Tconv via both PD-1/B7-H1 interactions and IL-10, with PD-1/B7-H1 interactions being the major negative regulatory pathway. IL-10 had a minor effect on the suppression of T cell proliferation in the in vitro assays but it might play a role in fine tuning the suppression of immune responses by other mechanisms. For example, IL-10 produced by Treg in inflamed tissues suppress production of proinflammatory cytokines by Teff (28). Furthermore, we have previously shown that IL-10 is involved in the inhibition of transendothelial migration of Teff into the inflamed tissues. Thus, we further suggest that IL-10 produced by EC-exposed Treg not only contributes to the capacity of Treg to suppress proliferation of Tconv but also might influence T cell responses in inflamed tissues by inhibition of Teff migration and production of proinflammatory cytokines.
Interestingly, ECs control the generation of a PD-1high IL-10–producing Treg subset via distinct mechanisms. PD-1 upregulation occurred contact-independent via soluble factor(s) produced by activated ECs, whereas an increase in IL-10 production was exclusively detectable in contact-dependent cocultures. Thus, this supports the notion that ECs contribute to the control of Treg function in the periphery at different time points as well as different sites. For instance, upregulation of PD-1 on Treg by soluble factor(s) derived from ECs might occur already in the bloodstream. Because interactions between PD-1 and B7-H1 on activated endothelium are proposed to be involved in transendothelial migration (45), this process might enhance Treg migration into the inflamed tissue by a feedback loop. However, an increase in IL-10 production was exclusively observed after physical contact of Treg and ECs. This might be a mechanism to avoid systemic release of IL-10 into the bloodstream, but increases specific suppression of tissue resident Tconv and DCs, once the Treg have migrated through the endothelium.
A detailed mechanism as to how EC facilitate the upregulation of PD-1 expression and IL-10 release by Treg is not clear yet. We tested whether release of vascular endothelial growth factor (VEGF) or expression of IDO may account for the Treg activating effects of the EC. We found (data not shown) that our ECs only secreted minor amounts of VEGF and expressed only negligible amounts of IDO. Moreover, the expression was not influenced by activation of the ECs with TNF-α/IFN-γ and therefore does not correlate with the increased Treg-activating capacity of activated ECs. Thus, these data indicate that neither IDO nor VEGF are the sole factors responsible for the induction of increased suppressor capacity of the Treg. In addition, the IL-10 release was also stimulated in Treg harboring an Ag-specific (transgenic) TCR in absence of the cognate Ag and therefore is likely to be Ag and TCR independent. This is in line with recent results published by Vignali et al. (46), showing that Treg can become activated independently from TCR engagement.
Nevertheless, we recently found that free ATP in tissues, which may be released in response to inflammatory stimuli, such as contact sensitizers, activate Treg. Thus, it is tempting to speculate that ATP, produced by ECs in response to inflammation and injury (47), may account for the activation and the increased IL-10 production of the Treg. This hypothesis is further supported by findings, showing that PD-1 is upregulated by engagement of the adenosine A2A receptor (48). Thus, Treg-derived adenosine, which is a degradation product of ATP, may be involved in the upregulation of PD-1. Although the involvement of adenine nucleotides in the EC-guided modulation of Treg functions is likely, it needs further proof.
In summary, our data show that activated ECs increase the immunosuppressive potential of activated Treg via triggering upregulation of PD-1 and stimulation of IL-10 and TGF-β release. Therefore, one can speculate that during inflammatory conditions, which render ECs activated, ECs enhance the ability of Treg to 1) augment the capacity of Treg to suppress proliferation of Tconv, mainly triggered by the enhanced PD-1 expression and 2) generate an immunosuppressive environment by IL-10 and TGF-β in inflamed tissues. Thus, it would be of interest to identify the factors involved in the EC-mediated augmentation of Treg function for therapeutic use.
Disclosures The authors have no financial conflicts of interest.
This work was supported by grants to K.M. and A.H.E. (SFB 405 B15, B16; German Research Foundation Grant KM 1924/2-2; the Wilhelm Sander Foundation, European Union Grant LSHC-CT-2005-518178, and the Helmholtz Association: Alliance on Immunotherapy of Cancer).
The online version of this article contains supplemental material.
Abbreviations used in this paper:
- 7-aminoactinomycin D
- experimental autoimmune encephalomyelitis
- endothelial cell
- low-density lipoprotein
- mean fluorescence intensity
- programmed death-1
- room temperature
- conventional CD4+ T cell
- thymidine deoxyribose
- effector T cell
- regulatory T cell
- vascular endothelial growth factor.
- Received July 29, 2009.
- Accepted March 5, 2010.
- Copyright © 2010 by The American Association of Immunologists, Inc.