Mesenteric lymph node
Regulatory T cells
There is an increasing amount of knowledge on the functional properties of regulatory T cells (Treg) in the adult immune system, but data on the generation and function of these cells during human embryonic development are scarce. In this study, we show that in the fetal thymus, double-positive cells initiate expression of CD25, GITR, CTLA4 and CD122 during their transition from the CD27– to the CD27+ stage. Moreover, CD4+CD25+ fetal thymocytes already have the potential to suppress proliferation of CD25– cells. After leaving the thymus, FoxP3+CD4+CD25+ Treg enter the fetal lymph nodes and spleen, where they acquire a primed/memory phenotype. A model is proposed for the development of human fetal Treg that encompasses two sequential maturation steps: initiation of a regulatory phenotype and suppressive activity in the thymus; and subsequent activation within the peripheral lymphoid organs. Upon activation, FoxP3+CD4+CD25+ Treg suppress potentially deleterious responses by autoreactive lymphocytes and maintain homeostasis within the developing fetus.
See accompanying Commentary: http://dx.doi.org/10.1002/eji.200525996
CD4+CD25+ regulatory T cells (Treg) survey all regions of the body and act as immunological sentinels by preventing or terminating immune responses by autoreactive lymphocytes 1–3. The fundamental importance of this Treg autoimmune surveillance has become apparent in both mice and men, with the observation that the absence of these cells, as a result of gene targeting or spontaneous mutations, invariantly leads to widespread autoimmunity and often premature death 4–7. Moreover, deregulation of CD4+CD25+ Treg is increasingly linked to various human diseases 8–10. Naturally occurring CD4+CD25+ Treg are thymus derived and are selected by self-antigen recognition 11–14. However, conclusive data on the developmental pathway of these cells in the human thymus and on the relation to the development of conventional CD4+CD25– T cells are scarce 11, 12.
Within the thymus, T cell development progresses through several clearly defined stages, which can be separated based on surface expression of CD1a and CD27 15. The earliest precursors that are committed to the T cell lineage express CD1a in the absence of CD27. This phenotype is maintained throughout the CD4/CD8 double-positive (DP) stages, as well as in the early immature CD4 single-positive (SP) stage (I-DP, II-DP and II-CD4SP). These cells subsequently acquire expression of CD27 during the transition to III-DP (followed by III-CD8SP) and III-CD4SP. Finally, human thymocytes down-regulate CD1a during development into mature CD4 or CD8 SP T cells (IV-CD4SP or IV-CD8SP; reviewed in 15).
There are no conclusive data on whether Treg branch off from this developmental pathway as a separate lineage or whether they are generated by differences in signal strength during TCR selection. Within the mature pool of T cells present in the thymus, Treg are present. From a developmental point of view, the earliest human intra-thymic Treg were reported in the human postnatal thymus 11, 12, 16. At this time, mature CD4+CD25+ T cells exist with many of the phenotypic characteristics of adult Treg and the capacity to suppress proliferation of naive T cells in in vitro assays 11, 12. Our current knowledge on Treg during embryonic development has been generated by making use of pre-term cord blood-derived cells 17. Pre-term cord blood is the only available source of fetal blood, and it has to be assumed that it is an accurate reflection of the cellular composition of the fetal circulation. Upon maturation in the fetal thymus, these CD4+CD25+ cells enter the periphery and are present in the circulation in elevated numbers when compared to adult peripheral blood. In contrast to the adult situation, the majority of cord blood Treg have a naive phenotype, are GITR negative, yet express FoxP3 and are able to suppress proliferation of polyclonally activated CD25– T cells in in vitro assays 17, 18. Only a small population of cells does present the phenotype normally associated with human Treg: CD4+CD25+CD45RO+GITR+. However, in contrast to CD4+CD25+CD45RO+ cells from adult peripheral blood 19, CD4+CD25+CD45RO+ T cells from cord blood lack the capacity to react, or suppress responses, against self-antigens 18. These data suggest that fetal Treg undergo additional, previously unappreciated, maturation steps during which they become responsive to self-antigens.
In this study, we provide a detailed description of human Treg present within the fetal thymus, spleen and lymph nodes (LN). The appearance of CD4+CD25+ Treg during thymic development is discussed, and we describe the presence of antigen-experienced FoxP3+ Treg in the fetal secondary lymphoid organs.
Emergence of fetal CD25+ thymocytes
To determine the kinetics of Treg generation during fetal development, human fetal thymi, ranging in age from 13 to 17 weeks of gestation, were analyzed for the appearance of CD25+ cells (Fig. 1). Thymocytes were gated on CD25+ (Fig. 1B) and CD25– (Fig. 1A) cells and subsequently subdivided into three major sequential developmental stages: CD1a+CD27–, containing the most immature DP cells; CD1a+CD27+, in which most DP cells are present; and finally CD1a–CD27+ cells, which are mainly made up of mature SP thymocytes. In the fetal thymus, the earliest cells that express CD25 are DP cells found within the CD1a+CD27– population. Surprisingly, a low percentage of these cells still show dim expression of surface CD3, indicating their early immature state. Additionally, the CD27– population contains very few CD25+CD3highCD4SP cells. The CD25– cells at this stage are also predominantly DP, expressing dim to high levels of surface CD3. The CD25–CD4SP cells are divided in CD3low and CD3high cells, unlike their CD25+ counterparts, which are all CD3high. The first CD8+CD25+ thymocytes are found in the subsequent CD1a+CD27+ stage. Similar to the CD25–CD8+ cells at this stage, CD3 expression varies from dim to high. In the most mature stage, CD1a–CD27+, all thymocytes express high levels of CD3, and the vast majority of cells are either CD4SP or CD8SP. Finally, when cells in this stage were assessed for levels of CD25 expression (Fig. 1C), only DP and CD4SP cells expressed high levels of CD25, while the CD8SP cells showed lower CD25 expression. These data indicate that expression of CD25 is acquired at the CD1a+ stage during initiation of CD27 expression by DP cells.
Phenotype of CD4+CD25+ and CD8+CD25+ fetal thymocytes
To determine whether the CD25+ cells found in the fetal thymus correlate with the described CD4+CD25+ regulatory cells in the postnatal thymus 11, 12, the expression of several surface markers characteristic for Treg was analyzed (Fig. 2). DP as well as SP CD4 or CD8 cells were gated on high expression of CD3, to select for the most mature cells. We found CD4+CD25+ T cells from the fetal thymus to have a surface profile that is in complete agreement with the reported phenotype of human Treg: expression of GITR, CD122 (IL-2R β chain) and intracellular CTLA4 (CTLA4ic), in combination with low expression of CD127 (IL-7R α chain). In contrast, the CD4+CD25– T cells have almost the reversed phenotype, with an absence of GITR, CD122 and CTLA4ic as well as higher levels of CD127. The CD8+CD25+ T cells that develop in the fetal thymus have similar characteristics as their CD4+ counterparts, which is in line with the described presence of CD8+CD25+ Treg in the postnatal thymus 16.
We noted two differences between the CD8+CD25+ and CD4+CD25+ cells: one was that the level of CD25 on CD8+CD25+ was considerably lower (Fig. 1C), a feature previously observed by Cosmi et al. 16, and second that the CD8+CD25+ population contained a considerable CTLA4ic-negative fraction (Fig. 2C). Surprisingly, the DP CD25+ thymocytes had already acquired many of the Treg characteristics, with the exception of CD127. On the DP population, CD127 was expressed at low levels, yet slightly higher than in the SP populations. In contrast, the DP CD25– cells showed only very weak CD127 expression. This indicates that CD127 is up-regulated in CD25+ cells with faster kinetics than in CD25– cells, yet fails to reach the level of expression typically seen in SP CD25– cells. The early up-regulation of CD127 on CD25+ thymocytes does not lead to the preferential formation of a thymic stromal lymphopoietin (TSLP) receptor complex 20 on these cells, since the majority of CD25+ thymocytes lack expression of the TSLP receptor (data not shown).
CD4+CD25+ thymocytes suppress proliferation of CD4+CD25– cells
The reported ability of CD4+CD25+ thymocytes from postnatal human thymi to suppress proliferation of CD4+CD25– cells in vitro11, 12 prompted us to test the suppressive capacity of fetal CD4+CD25+ cells. CD25+CD4+CD1– cells as well as CD25–CD4+CD1– cells from fetal thymi were sorted to purity, mixed in a 1:1 ratio and cultured with irradiated PBMC and soluble anti-CD3 antibody (Fig. 3). In this assay, fetal CD4+CD25+ thymocytes were indeed able to potently suppress the proliferative response of CD25– thymocytes, showing that already within the fetal thymus functional Treg are present.
CD4+CD25+ T cells in fetal secondary lymphoid organs
In line with reports on pre-term cord blood Treg 17, the mesenteric LN (MLN) (Fig. 4), as well as the spleen (data not shown), contained a large population of CD4+CD25+ T cells, ranging from 13% to 22% of total CD4 cells. This percentage remained constant during development from week 14 to week 17, as shown in Fig. 4. CD8+CD25low cells were also present in the LN, although, in contrast to the elevated percentage of CD25+ cells within the CD4+ population, the percentage of CD8+CD25low cells was similar to that of the fetal thymus.
While CD4+CD25+ Treg in pre-term and normal-term cord blood have mainly a naive phenotype, CD45RA+RO– with an absence of GITR (17, 18, 21 and our unpublished observations), Treg in adult peripheral blood mostly have a primed/memory phenotype 19, 22–24. This discrepancy directly implies that CD45RA+, naive fetal Treg undergo antigen-mediated activation during fetal or postnatal development. Since such antigen encounter events are likely to take place in the secondary lymphoid organs, fetal MLN and spleens were analyzed for the presence of CD4+CD25+ Treg populations (Fig. 5). Surprisingly however, and in contrast to (pre-term) cord blood Treg 17, 18, 21, flow cytometric analysis revealed that cells within the secondary lymphoid organs displayed the classical CD45RO+ antigen-experienced phenotype associated with adult CD4+CD25+ Treg (Fig. 5; CD25–, filled histograms; CD25+, black line; all gated on CD4+CD3+). All CD4+CD25+ T cells expressed the αβ TCR, and a large portion of these cells expressed GITR. Moreover, all were positive for CD122, CD95 and CTLA4ic, as well as CD127– and CD62Llow. In addition, fetal CD4+C25+ T cells, found in the non-inflamed fetus, expressed the early activation marker CD69, suggestive of TCR signaling.
None of these cells expressed the integrin αE (CD103), which is expressed on a subset of CD4+CD25+ T cells in adult mice 25. Since these data suggest that CD4+CD25+ Treg get activated within the LN, we analyzed the expression of GITR on the CD4+CD25+CD45RO– versus CD4+CD25+CD45RO+ cells (Fig. 5B). This indeed showed that upon activation and up-regulation of CD45RO, Treg initiate expression of GITR. Finally, to exclude clonal expansion in situ, a flow cytometric analysis of the Vβ repertoire of CD4+CD25+ and CD4+CD25– T cells in the LN was performed, which showed diverse Vβ usage indistinguishable between CD25+ and CD25– cells (data not shown) 26.
The CD4+CD25+ T cell population in fetal spleen had a similar surface phenotype and was found in the same percentages as in the LN (data not shown).
CD4+CD25+ Treg from fetal LN express FoxP3 and suppress proliferation of CD4+CD25– T cells
Expression of FoxP3 was determined on cDNA from CD4+CD25+ and CD4+CD25– cells sorted to purity from fetal LN (Fig. 6A). Using decreasing concentrations of cDNA, CD4+CD25– cells did not express detectable levels of FoxP3, while CD4+CD25+ fetal T cells show clear expression of FoxP3 at all three concentrations. To test whether the fetal primed/memory CD4+CD25+ Treg within the LN are functional, their ability to suppress proliferation of fetal CD4+CD25– T cells was determined in vitro (Fig. 6B). Cells were sorted from fetal MLN, mixed in a 1:1 ratio and cultured in the presence of irradiated PBMC and soluble anti-CD3 antibody for 96 h. Proliferation was determined using [3H]thymidine incorporation. Fig. 6B shows two representative examples of such an in vitro suppression assay. CD4+CD25+ Treg from fetal MLN inhibited proliferation of CD4+CD25– T cells, showing that functional Treg are present in the fetal LN.
During embryonic development, a daunting challenge has to be met when organ formation and tissue remodeling coincide with the emergence of an adaptive immune system that has the potential to mount possibly fatal immune responses. Our data are the first to show that within the human fetal thymus, Treg branch off from the developmental pathway of conventional T cells during the transition of DP cells from CD27– to CD27+. At this point, the exact molecular signals that initiate this lineage decision remain to be determined, but they could possibly be linked to the strength of TCR signaling or the repertoire of self-antigens presented 27–29. Upon exit from the thymus, CD4+CD25+ Treg enter the circulation as naive, CD45RA+ cells. It has been clearly shown that in this naive state, Treg have yet to acquire the ability to respond to defined autoantigens 18. The capacity to do so seems to be linked to the transition from this naive phenotype to a more primed/memory-like phenotype. These results led us to propose that this transition occurs within the fetal secondary lymphoid organs, mediated by the recognition of (self)-antigens (Fig. 7). However, we cannot exclude the possibility that a population of conventional CD4+CD25– T cells is converted to CD4+CD25+ Treg in the periphery 30, 31. The model we propose is supported by the described presence of two distinct Treg populations in the murine system 32, a dichotomy that becomes apparent upon antigen recognition in the steady state by the self-specific TCR. While part of the Treg remain in a quiescent state, Treg activated by self-antigens up-regulate activation markers and enter into cell cycle. Together with the fact that CD4+CD25+ Treg need to be activated via their TCR to acquire suppressive activity, this indicates that in the human fetus, Treg seem to undergo this activation step within the fetal secondary lymphoid organs.
Following maturation in the thymus (Fig. 7), both conventional CD4+CD25– T cells as well as regulatory CD4+CD25+ T cells start circulating through the body in a naive state 17, 18, 21. Like all T cells, the most likely site for these thymic emigrants to first encounter high doses of antigen will be in the LN and the spleen, sites to which antigens are actively transported by dendritic cells, as well as via the lymphatics. Within the fetal environment, the antigen load will have a unique composition, being a mixture of maternally derived antigens that have crossed the placental barrier and self-antigens, which are likely to be present in high amounts due to continuous tissue remodeling and massive cellular apoptosis. Conventional T cells that recognize self-antigens outside of the LN or spleen, in the absence of professional antigen-presenting cells, will be triggered to apoptosis or anergy 33, 34. However, a part of these potentially autoreactive T cells will first encounter their cognate antigen within the lymphoid organs, where it is likely to be presented by professional antigen-presenting cells, and as a result get activated. We observed that the CD25+CD69+ population in the fetal LN is larger than the CD25+GITR+ population (Fig. 5). There are two possible explanations: First, following activation of naive Treg, CD69 appears earlier than GITR, or the CD69+GITR–CD25+ cells are derived from conventional CD25–CD4+ cells as a consequence of activation by autoantigens. Second, T cells reacting to maternally derived antigens will also be activated within the LN and spleen and acquire the aptitude to mount a potentially lethal immune response. We propose that the developing immune system of the human fetus has devised an intricate way of preventing such immune responses. The CD4+CD25+ Treg that leave the thymus as naive cells likewise encounter their cognate antigen in the peripheral lymphoid organs. These antigens are likely to be autoantigens, which is in line with the previously reported, developmentally acquired response of Treg to myelin proteins 18. As a result, Treg get activated, start to express CD69, GITR and CD45RO and are then able to suppress a broad range of responses. In this way, conventional T cells that escape both central and peripheral tolerance and get activated within the LN or spleen will subsequently come under control of these activated CD4+CD25+ Treg that reside within the secondary lymphoid organs (Fig. 7).
In those cases in which fetal Treg development is perturbed, as is the case with immune dysregulation, polyendocrinopathy, enteropathy, X-linked (IPEX)-syndrome, the clinical appearance of autoimmunity takes place very early after birth 5, 6, 35–38. This indicates that activation of autoreactive T cells by self-antigens had already initiated within the developing fetus.
Summarizing, we show that human Treg specification occurs in DP thymocytes during transition to the CD27+ stage, and propose a model in which CD4+CD25+ Treg within the fetal secondary lymphoid organs get activated, most likely by their cognate antigen, and subsequently act to educate the developing immune system and prevent unwanted immune responses in the fetus. In this way, T cells that escape central and peripheral tolerance are kept in check by this previously unappreciated third way of peripheral tolerance induction occurring within the secondary lymphoid organs.
Materials and methods
Fetal and postnatal tissues
The use of human fetal and postnatal tissues was approved by the Medical Ethical Committee of the Academic Medical Center and was contingent on informed consent. Fetal tissues were obtained from elective abortions. Gestational age was determined by ultrasonic measurement of the diameter of the skull and ranged from 13 to 17 weeks. MLN were dissected from the mesentery using dissecting microscopes, and cell suspensions were made by digestion with 0.5 mg/ml collagenase type IV (Sigma, St. Louis, MO) in PBS for 30 min at 37°C, while stirring continuously, and subsequently filtered through a nylon mesh. Spleens and thymi were digested by mincing and gently pressing the tissue through a fine nylon mesh, after which splenic erythrocytes were removed by alkaline lysis.
The following antibodies were used: anti-TCRαβ-FITC; -CD4-peridinin chlorophyll a protein (PerCP)-Cy5.5; -CD8-allophycocyanin (APC)-Cy7; -CD25-PE; -CD25-PE-Cy7; -CD27-PE; -CD45RA-APC; -CD45RO-APC; -CD62L-FITC; -CD62L-PE; -CD69-FITC; -CD69-PE; -CD95-APC; -CD103-FITC; -CD122-PE; -CD127-PE; -CTLA4-biotin (all purchased from Becton Dickinson, San Jose, CA); anti-GITR-PE (R&D Systems, Minneapolis, MN) and anti-CD1a-FITC (Serotec, Oxford, UK). Flow cytometry was performed on a BD LSRII (Becton Dickinson).
In vitro suppression
In vitro proliferation assays were performed as described 19. CD4+CD25+ and CD4+CD25– cells were sorted to purity and co-cultured in a 1:1 ratio (2×104 cells each) in the presence of 2.5×104 irradiated (900 rad) allogeneic adult PBMC and 0.4 µg/ml anti-CD3 antibody (Pelicluster-CD3; Sanquin, Amsterdam, The Netherlands) in RPMI/20% human serum for 4 days. During the last 16 h, 0.7 mCi [3H]thymidine was added. Proliferative responses are expressed as the mean [3H]thymidine incorporation (cpm) of triplicate wells ± SD.
mRNA was isolated and reverse transcribed using standard protocols. Primers used were: FoxP3 Fw: 5′-TCCCAGAGTTCCTCCACAAC, Rev: 5′-TGAGTGTCCGCTGCTTCTC; and HPRT Fw: 5′-TATGGACAGGACTGAACGTCTTG, Rev: 5′-GACACAAACATGATTCAAATCCC.
The Bloemenhove Clinic in Heemstede, The Netherlands, is thanked for providing fetal tissues. Berend Hooijbrink is acknowledged for help with FACS sorting and maintenance of the FACS facility.