Unlike graft-resident donor-derived hemopoietic APCs, which decrease in number over time after transplantation, vascular endothelial cells are lifelong residents of a vascularized allograft. Endothelial cells are potent APCs for allogeneic CD8+ T lymphocytes but are unable to induce proliferation of allogeneic CD4+ T lymphocytes. Although the reason for this differential response has been poorly understood, here we report that alloantigen presentation by vascular endothelium to CD4+ T lymphocytes activates and induces CD4+25+Foxp3+ regulatory T cells, which can inhibit proliferation of alloreactive T cells both in vitro and in vivo. This process occurs independently of B7.1 costimulation but is dependent on programmed death ligand 1 (B7-H1). This finding may have important implications for tolerance induction in transplantation.
Several investigators, including our group, have described that vascular endothelium isolated from various sources, such as the murine aorta and pulmonary parenchyma, constitutively express MHC class I as well as costimulatory molecules (1). We have shown that endothelial cells can act as APCs to allogeneic CD8+ T lymphocytes and can trigger allograft rejection via CD8+ direct allorecognition (2, 3). However, despite expression of both inducible MHC class II and costimulatory molecules, vascular endothelium is unable to induce cell division in allogeneic CD4+ T cells in vitro or in vivo (4, 5). More importantly, endothelial cells are unable to initiate acute graft rejection via CD4+ direct allorecognition (6). Based on these observations, the interaction between vascular endothelium and allogeneic CD4+ T cells has been generally viewed as a neutral encounter (4, 5, 6). In this study, we have further analyzed the interaction between vascular endothelium and allogeneic CD4+ T lymphocytes and show that activated vascular endothelium generates alloantigen-specific CD4+25+Foxp3+ regulatory T cells, which inhibit T cell proliferation both in vitro and in vivo. As vascular endothelium persists for the life of the allograft, the ability to induce regulatory CD4+ T cells may be important for tolerance induction.
Materials and Methods
Animals and cell isolation
Primary and secondary cultures
CD4+ T lymphocytes were resuspended in RPMI 1640 medium and cultured with endothelium activated with 500 U/ml murine IFN-γ (R&D Systems) for 72°C before coculture (2, 4). For some experiments, CTLA4Ig fusion protein (15 μg/ml), hamster Ig (15 μg/ml), rat IgG2a (1 μg/ml) (BD Pharmingen), anti-programmed death ligand 1 (PDL-1)4 (1 μg/ml), or anti-PDL-2 (1 μg/ml) blocking Abs (eBioscience) were added during the primary CD4+T cell/endothelial coculture. After 5 days in culture, the CD4+ T cells were reselected by CD4+ magnetic beads and added to a secondary culture as “regulators” of syngeneic CFSE-labeled CD8+ responder cells. CD4+ T cell survival and recovery within these primary cultures ranged between 25 and 40%. Stimulator cells consisted of irradiated (30 Gy) B6 or BALB/c CD90-depleted (midiMACS system; Miltenyi Biotec) splenocytes. In vivo adoptive transfer experiments were performed as described previously (2, 4).
Flow cytometry, RT-PCR, and ELISA
All Abs and reagents for flow cytometric analysis were purchased from BD Pharmingen or eBioscience and were primarily conjugated with either PE, allophycocyanin, or peridinin chlorophyll protein-cyanin 5.5. Intracellular Foxp3 staining was performed using a commercially available kit (eBioscience). Flow cytometric sorting of CD4+ T cells was performed on a MoFlow (DakoCytomation) high-speed sorter based on CD25 expression. Purity of the CD4+25− cell fraction was consistently >99.7%, and the purity of the CD4+25+ fraction ranged between 85 and 95%.
Cytokine analysis of culture supernatant was performed using commercially available ELISA kits (BioSource International). Quantitative and semiquantitative RT-PCR were performed using previously published primer sequences (7). Quantitative analysis was performed after normalization of the data to 18 sRNA, and relative Foxp3 expression was calculated by setting expression in CD4+25− cells to 1. All statistical analysis was performed by Student’s t test.
Results and Discussion
Coculture of CD4+ T lymphocytes with activated vascular endothelium results in the expansion of activated CD4+25+Foxp3+ regulatory T cells, which can inhibit T cell proliferation in an alloantigen-specific fashion
To further evaluate the interaction between T cells and endothelium, we analyzed the supernatants collected from cocultures of activated B6 vascular endothelium and CBA CD8+ and CD4+ T cells, respectively (Fig. 1 A). CD8+ T cell/endothelial cell coculture supernatants had high levels of IFN-γ but no IL-2. Lack of IL-2 is consistent with previous reports describing production of IL-2 by CD8+ T cells for predominantly autocrine use (8). In the supernatants of CD4+ T/endothelial cell cocultures, we observed low but detectable levels of IL-2 and remarkably high levels of IL-10 and TGF-β compared with CD8+ T/endothelial cell cocultures. Because a similar cytokine pattern has been described previously during the generation of regulatory T lymphocytes, we decided to evaluate whether coculturing vascular endothelium with allogeneic CD4+ T cells results in the generation of regulatory T cells (9, 10).
We noticed an increase in the percentage of CD25+ cells after 5 days of coculturing CBA CD4+ T cells with activated B6 endothelium (Fig. 1,B). As surface expression of CD25 in CD4+ T cells is characteristic of both regulatory and activated alloreactive responder cells, we documented increased expression of Foxp3 mRNA by quantitative RT-PCR in CBA CD4+ T lymphocytes after coculture with activated B6 vascular endothelium suggesting the expansion of regulatory T cells (Fig. 1,C). Flow cytometrically we were able to determine that the vast majority of CD4+25+ T cells express Foxp3 after coculture with allogeneic vascular endothelium, while only a small fraction of the CD25+ population did not express Foxp3 (Fig. 1 D). The CD4+25+Foxp3+ T cells also demonstrated up-regulation of surface glucocorticoid-induced TNFR (GITR) and intracellular CTLA4 when compared with freshly isolated, resting CD4+25+Foxp3+ naturally occurring T cells, suggesting activation of this cell population by activated allogeneic vascular endothelium.
We next used a two-stage culture system to test the inhibitory properties of our cells. First, CBA CD4+ T cells were cocultured with activated B6 vascular endothelium for 5 days (primary culture), and subsequently, these CBA CD4+ T cells were mixed as “regulators” with responder CBA CD8+ T cells and irradiated T cell-depleted B6 splenocyte stimulators (secondary culture). CBA CD8+ T cells proliferated vigorously when stimulated by irradiated, T cell-depleted B6 splenocytes (Fig. 1,E). The addition of CBA CD4+ T cells that had been cocultured with activated B6 endothelium during the primary culture resulted in inhibition of CBA CD8+ T cell proliferation (Fig. 1,E). Near complete inhibition of CD8+ T cell proliferation could be achieved with a 3:1 “regulator”:responder cell ratio in secondary cultures. The importance of CD4+ T cell TCR engagement by inducible MHC class II on vascular endothelium during primary culture was demonstrated by abrogation of suppression when CBA CD4+ T cells were cocultured with MHC class II-deficient B6 endothelium before transfer to secondary cultures as “regulators” (Fig. 1,E). Suppression was alloantigen specific as CBA CD4+ T cells, stimulated primarily by B6 endothelium, did not inhibit CBA CD8+ T cell proliferation induced by stimulators derived from a third party (BALB/c splenocytes) during the secondary culture (Fig. 1E). Of note, CBA CD4+ T cells stimulated primarily by B6 endothelium did not undergo proliferation and remained anergic upon secondary culture with B6 splenocytes as well (data not shown).
We next set out to evaluate whether CD4+ T cells, cocultured with activated vascular endothelium, are able to inhibit proliferative responses in vivo. We have described previously that CBA CD8+ T cells proliferate vigorously after adoptive transfer into a supralethally irradiated B6 mouse (2). However, when CBA CD4+ T cells that had been cocultured previously with activated B6 vascular endothelium in primary culture were cotransferred along with CBA CD8+ T cells, the proliferative responses of the CBA CD8+ T cells were markedly inhibited. Analogous to the in vitro observation, this inhibition was alloantigen specific as CBA CD4+ T cells cocultured primarily with activated B6 vascular endothelium were unable to affect CBA CD8+ T cell proliferation after adoptive transfer into a supralethally irradiated BALB/c mouse (data not shown).
Although a large body of literature supports the role of CD4+25+ T cells in regulating immune responses, some investigators have suggested that the CD4+25− T cell fraction can also inhibit T cell proliferation and mediate transplantation tolerance (11). Based on the Foxp3 expression in CD4+25+ T cells after coculture with endothelium (Fig. 1,D), we hypothesized that this population would have suppressive function. To this end, CBA CD4+ T cells were flow cytometrically sorted based on CD25 expression after 5 days of coculture with activated B6 vascular endothelium. Suppression was limited to the CD4+25+ T cell fraction, and the CD4+25− T cells were unable to suppress proliferation of CD8+ T cell responders in secondary cultures (Fig. 1 F).
To evaluate the role of naturally occurring CD4+25+ regulatory T cells in inhibition of CD4+ T cell proliferation in primary T cell/endothelial cell cultures, we depleted the CD4+25+ T cell fraction from freshly isolated CBA CD4+ T cells by flow cytometric sorting. Unlike the case for unfractionated CD4+ T cells, CD4+25− T cells did proliferate when cultured with activated B6 vascular endothelium (Fig. 1 G). Taken together with the up-regulation of CTLA4 and GITR in the CD4+25+ T cell fraction, these findings support the notion that the coculture of CD4+ T cells with allogeneic vascular endothelium results in the activation of naturally occurring CD4+25+Foxp3+ regulatory T cells as well as expansion of the CD4+25+Foxp3+ regulatory T cell pool.
Expansion of regulatory T cells results from the conversion of CD4+25− to CD4+25+Foxp3+ cells
Although we described activation of naturally occurring CD4+25+Foxp3+ regulatory T cells by vascular endothelium, consistent with previous reports (12), no proliferation of this cell population was evident in our cultures by CFSE staining. This raised the possibility that the increase in the CD4+25+Foxp3+ cells after coculture with cytokine-activated endothelium was due to the conversion of CD4+25− T cells into CD4+25+ regulatory cells.
We further tested this possibility by culturing sorted CBA CD4+25− T cells with activated B6 vascular endothelium. After 5 days of coculture, as many as 10% of the initially >99.7% pure CD4+25− T cells expressed surface CD25. The newly generated CBA CD4+25+ cells had surface GITR and intracellular CTLA4 similar to naturally occurring CD4+25+ regulatory cells (Figs. 2,A). Expression of Foxp3, which was undetectable by semiquantitative RT-PCR in freshly isolated CBA CD4+25− T cells, was documented after 5 days of coculture with activated B6 vascular endothelium indicating the generation of regulatory Foxp3+ T cells (Fig. 2 B).
Consistent with previous reports, we were able to detect a population of CD4+CD25lowFoxp3+ T cells within freshly isolated CD4+ T cells (Fig. 1,D) (13). In fact, a recent report suggests that CD4+CD25+Foxp3+ regulatory T cells arise from CD25lowFoxp3+ cells (13). For our sorting experiments described above, we discarded CD25low as well as CD25+ cells and isolated CD25− cells only. These cells do not express Foxp3 before coculture with vascular endothelium but after coculture Foxp3 was induced (Fig. 2 B). Therefore, we conclude that the coculture of CD4+ T cells with allogeneic vascular endothelium results in the generation of CD4+25+Foxp3+ regulatory T cells from CD4+25− cells. Although our data indicate that our original CD25− T cells do not express Foxp3, we cannot rule out the possibility that a small population of CD25lowFoxp3+ cells with Foxp3 present below the levels of detection by PCR contribute to the generation of CD25+Foxp3+ regulatory cells.
The generation of CD4+25+ regulatory T cells depends on PDL-1 costimulation
Although some reports have shown B7.1 to play a critical role in the generation and homeostasis of regulatory T cells, others have suggested that these processes can occur independently of B7.1 (14, 15, 16). Based on the expression of B7.1 on endothelial cells (1, 2), we hypothesized that B7.1 is important to the activation and/or generation of regulatory T cells in our system. Interestingly, CBA CD4+ T cells cocultured in the absence of B7.1 stimulation through the addition of soluble CTLA4Ig or use of vascular endothelium isolated from B7-deficient mice during primary culture retained the same phenotype with expansion of the CD4+25+ fraction as well as up-regulation of CTLA4 and GITR similar to CD4+25+ T cells generated in the presence of B7.1 costimulation (data not shown).
Similar to previously published reports, murine vascular endothelium in our cultures expresses surface PDL-1 (17) but does not express PDL-2 (data not shown). We next sought to investigate the role of PDL-1 in the generation of regulatory T cells. Interestingly, we found that PDL-1 blockade during the primary CD4+ T cell/endothelial coculture resulted in inhibition of CD4+25+ T cell expansion and a decrease in markers of CD4+25+ T cell activation with levels of CTLA4 and GITR comparable to those found in freshly isolated CD4+25+ regulatory T cells (Fig. 3,A). No up-regulation of Foxp3 mRNA was seen in CD4+ T cells in the presence of PDL-1 blocking Ab by quantitative RT-PCR (Fig. 3,B). Intracellular staining with Foxp3 extended our RT-PCR data as we found that the addition of anti-PDL-1 Ab prevented the expansion of the CD4+25+Foxp3+ T cell population seen in cocultures without PDL-1 blockade (Figs. 1,D and 3 C). Interestingly, unlike the case for cocultures of endothelium with allogeneic CD4+ T cells without PDL-1 blockade, CD4+CD25+Foxp3− T cells were also present when PDL-1-blocking Ab was added. Moreover, CBA CD4+ T cells primarily cultured with activated B6 vascular endothelium in the presence of PDL-1 blockade were unable to inhibit CBA CD8+ T cell proliferation in secondary cultures (data not shown). Thus, our data indicates that PDL-1 blockade prevents both the activation and expansion of CD4+25+Foxp3+ regulatory T cells. This observation is supported by other reports that have also shown PDL-1 to be critical to the induction of regulatory T cells (18, 19, 20).
Because of progressive diminution of donor-derived hemopoietic APCs, it is generally believed that indirect, self-restricted Ag presentation is the predominant allorecognition pathway with increasing duration after solid organ engraftment. Therefore, it has been postulated that indirect allorecognition is the predominant mode of late Ag presentation to both regulatory as well as CTL (21, 22). Our recent observation that nonhemopoietic cells, which are present for the life of the allograft, can mediate rejection via CD8+ direct allorecognition has opened the possibility of an alternative pathway mediating late allorecognition (3). To our knowledge, this report is the first demonstration of CD4+25+Foxp3+ regulatory T cell generation via CD4+ direct alloantigen presentation by nonhemopoietic cells. Our findings are consistent with and extend recent reports describing the peripheral conversion of CD4+25− nonregulatory T cells into CD4+25+Foxp3+ regulatory T cells in other models (23, 24).
Several observations in transplantation biology may be explained by our finding. Direct CD4+ alloantigen presentation by graft-resident hemopoietic APCs is an important mechanism of alloreactivity and initiation of allograft rejection early after organ transplantation. As these cells progressively decrease in number over time, this pathway of allorecognition is blunted by regimens such as temporary recipient lymphocyte depletion or transient costimulatory blockade. Interestingly, such treatments can also induce regulatory T cells (25, 26). The underlying mechanism for the induction of regulatory T cells by these regimens is currently unknown, but based on our data, we can speculate that it may result from delayed allorecognition. This may allow donor-derived hemopoietic APCs to decrease in number, thus increasing the importance of direct alloantigen presentation to CD4+ T cells by vascular endothelium. Because transplantation of a larger mass of allogeneic tissue has similarly been shown to lead to allograft acceptance, it is possible that delivering a larger mass of endothelium or other parenchymal cells may favor tolerogenic Ag presentation over alloreactivity induced by professional hemopoietic APCs (27). Thus, our finding that alloantigen presentation by vascular endothelium to CD4+ T cells leads to the induction and activation of alloantigen-specific CD4+25+Foxp3+ regulatory T cells could have important implications for the design of tolerance induction strategies without global recipient immunosuppression.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was partially supported by the American Society of Transplant Surgeons (to A.S.K.) and National Institute of Health Grant R01HL041281-10A2 (to G.A.P.).
Abbreviations used in paper: PDL-1, programmed death ligand1, GITR, glucocorticoid-induced TNFR.