The best candidate for regulatory T (Treg) cell lineage-determining factor is currently the Forkhead box transcription factor FOXP3. FOXP3 up-regulation has been linked to TCR-mediated signals, and in mice the abrogation of TCR expression or signals also prevents FoxP3 expression. In contrast, the TCR dependence of human FOXP3 is assumed but not established. In this study we show on a single cell level that 1.4% (range 0.1–3.8%) of CD4−CD8− thymocytes in healthy humans express FOXP3, two thirds of them without any detectable αβ TCR. These TCR−FOXP3+ cells were mostly CD25− and did not express γδ TCR or B cell, NK cell, or monocyte-associated markers. Like mature Treg cells, they were mostly CD2+CD127low and expressed cytoplasmic CTLA-4. Our results suggest that in immature human thymocytes the expression of FOXP3 precedes surface TCR, in which case TCR-mediated signals cannot be responsible for the thymic up-regulation of FOXP3.
The natural regulatory T (Treg)3 cells, originally described as CD4+ T cells also expressing CD25 (1, 2), have recently emerged as key controllers of the immune system. A definitive step in their thymic development is the up-regulation of the Forkhead box transcription factor FoxP3, in murine models shown to be both necessary and sufficient for conferring a full regulatory phenotype (3, 4, 5). A large body of literature suggests that Treg cell commitment in the thymus is driven by self-Ags and the resultant population is autoreactive, indicating that TCR-mediated signals are involved in the up-regulation of FoxP3 (3, 4, 5, 6, 7). In support of the indispensable role of TCR, FoxP3 is found exclusively in αβ TCR+ cells and abrogation of TCR expression or TCR-mediated signals also prevents FoxP3 expression (8).
In humans, the data on the development of FOXP3-expressing natural Treg cells are much more limited, but several factors have led to the implicit acceptance of dependence on TCR-mediated signals. These factors include extrapolation from the murine studies, a well-proven link between TCR stimulation and FOXP3 induction in mature T cells, and the fact that FOXP3 expression has only been reported in TCR+ populations (3, 4, 5). However, there are no data directly addressing the timing of FOXP3 up-regulation in the human thymus. In this study we show that a subset of CD4−CD8− thymocytes expresses FOXP3, mostly without a surface TCR. Our results suggest that FOXP3 up-regulation in the human thymus precedes TCR expression.
Materials and Methods
Patients and samples
Thymic tissue was obtained from otherwise healthy children undergoing corrective cardiac surgery (n = 8, four females, mean age 4.4 mo, range 4 days to 10.5 mo) and PBMC were from healthy adult volunteers (n = 6, four females, mean age 26 years). Informed consent was obtained from the volunteers and the parents of the children, and the study was approved by the ethics committee of Helsinki University Hospital, Helsinki, Finland. Thymocytes were released from the tissue samples by mechanical homogenization and freshly analyzed. PBMC were isolated by using Ficoll-Hypaque gradient centrifugation (Amersham Biosciences). Double-negative thymocytes were negatively selected and γδ TCR+ PBL were positively selected either by immunomagnetic separation using mAbs and Dynabeads (Dynal), or by flow cytometric sorting using the FACSAria instrument (BD Biosciences) as described (9). The purity of the isolated cells was typically >90%.
Anti-human FOXP3 mAb conjugated to PE (clone 236A/E7) was purchased from eBioscience. The anti-FOXP3 clone 150D/E4 (10) was a gift from Dr. A. Banham (University of Oxford, Oxford, U.K.), and the anti-pre-Tα mAb (11) was from Dr. M. Toribio (Universidad Autónoma de Madrid, Madrid, Spain). They were used with PE- or FITC-labeled anti-mouse Ig (BD Biosciences), respectively. A mix containing anti-human CD14, CD19, and CD56 mAb was purchased from Dynal and used with the FITC-labeled anti-mouse Ig reagent. All other mAb, and an isotype-matched control mAb, were direct conjugates and were purchased from BD Biosciences. Intracellular detection of FOXP3 and CTLA-4 was performed on cells permeabilized with the FOXP3 permeabilization kit (eBioscience) according to the manufacturer’s instructions. The samples were analyzed with the FACSAria instrument.
Cells were lysed using TriPure isolation reagent (Roche) and total RNA was isolated with RNeasy mini kit columns (Qiagen). First-strand cDNA was synthesized by using an oligo(dT) primer (Sigma-Aldrich) and avian myeloblastosis virus reverse transcriptase (Finnzymes). Quantitative PCR was done using TaqMan Universal PCR master mix and intron-spanning primer-probe assays for FOXP3 and β-actin, commercially available from Applied Biosystems. The reactions were analyzed using an iCycler-IQ instrument (Bio-Rad) and the values taken from the exponential phase of the reaction. FOXP3 levels were normalized against β-actin.
Results are expressed as mean ± SD. p values were calculated with Student’s two-tailed t test.
Results and Discussion
In the murine thymus, FoxP3 is currently thought to be up-regulated at the CD4+CD8+ double-positive (DP) stage, following TCR α-chain rearrangement and surface expression of αβ TCR (12, 13). Flow cytometric analysis showed that in the human thymus (n = 8) 69.9 ± 12.8% of all FOXP3+ cells belonged to the CD4+CD8− single-positive (SP) subset, and 61.2% of these cells also expressed CD25 (Table I). Previous studies have shown that these cells are functionally mature Treg cells, indistinguishable from circulating CD25high Treg cells (14). Smaller fractions of the FOXP3+ thymocytes belonged to the DP or CD8+CD4− SP populations, (17.9 ± 7.7% and 11.0 ± 7.2%, respectively). However, in every thymus examined, a small subset of FOXP3+ cells was also found in the CD4−CD8− double-negative (DN) subset (Fig. 1,A). The FOXP3+ cells accounted for 1.4 ± 1.1% of all DN cells, expressed FOXP3 at the same intensity as the FOXP3+ CD4 SP cells, were CD34−, and only 22.3% of them were CD25+ (Table I).
To exclude the possibility that the 236A/E7 mAb we used to detect FOXP3 might cross-react with an unknown structure expressed in the DN cells, we repeated the FACS analysis with a mAb (150D/E4; Ref. 10) specific to a different epitope with similar results (data not shown). Control staining with the second-step reagent alone showed no significant background staining. To further ascertain that the DN thymocyte population expressed FOXP3, we performed an immunomagnetic depletion of CD4+ and CD8+ cells and then used quantitative PCR to measure the amount of FOXP3 mRNA in the isolated DN thymocytes. There was no significant difference between the DN cells and thymocytes in general (Fig. 1,B), a result in good agreement with the FACS data (Table I). As a point of comparison, the level of FOXP3 mRNA in DN thymocytes was 30 times higher than that in peripheral blood γδ T cells, not reported to express FOXP3 (Fig. 1 B). Together, these data show that some DN thymocytes express both FOXP3 protein and mRNA.
The DN stage contains the earliest T cell precursors, which have not yet completed TCR gene rearrangement and therefore cannot express surface αβ TCR (15). In contrast, TCRhigh DN Treg cells have also been reported, although only from peripheral blood (16). We therefore performed a four-color FACS analysis to determine TCR expression in the FOXP3+ thymocyte subsets. The great majority of all FOXP3+ thymocytes expressed αβ TCR (Table I). In the FOXP3+ CD4 SP population this fraction was even bigger, and almost all of these cells were TCRhigh (Fig. 1, C and D). In contrast, less than one-third of the FOXP3+ DN cells expressed αβ TCR, and even in the TCR+ cells the expression level was significantly lower than in the CD4 SP cells (mean fluorescence intensity of TCR staining, 129 ± 36 vs 256 ± 48; p < 0.01). There was no correlation between TCR and CD25 expression in the FOXP3+ DN cells. Staining with an isotype- and fluorochrome-matched control mAb revealed no significant binding (<0.1% in DN cells) or differences between the subsets (Fig. 1 C). The TCR−FOXP3+ subset also did not express intracellular TCR and was therefore not an artifact caused by TCR internalization. Expression of pre-Tα was not observed, either. These results show that the normal human thymus contains a subset of FOXP3+ DN cells lacking TCR expression. Immunofluorescence staining of frozen tissue sections showed that these TCR−FOXP3+ cells localized to cortical areas of the thymus (data not shown).
In addition to T cell precursors, the DN thymocytes contain a heterogeneous population of non-T cell leukocytes, so it was important to exclude the possibility of ectopic FOXP3 expression in non-T cells. Staining with mAb specific to CD14, CD19, CD56, or γδ TCR showed that, although the DN cells contained a substantial positive fraction, the FOXP3+ DN thymocytes were not costained and thus were not monocytes, B cells, NK cells, or γδ T cells (Fig. 2,A). Epithelial cells expressing epithelial cell adhesion molecule (EpCAM) were detected at a low frequency (0.2%; range 0.0–0.7%) in some of the samples, but none of the EpCAM+ cells expressed any FOXP3. That the FOXP3+ DN thymocytes belonged to the T cell lineage was further supported by the fact that most of them expressed CD2, an early T/NK lineage marker (Fig. 2 B) (17).
A characteristic marker of Treg cells is the intracellular expression of CTLA-4, a molecule that plays an important role in the Treg cell function (3). Most FOXP3+ thymocytes expressed CTLA-4 (Table I), and this was true of the FOXP3+ DN cells as well (Fig. 3,A). The FOXP3+ DN cells were also CD127low, a feature recently reported to be typical of human peripheral blood Treg cells (18, 19, 20). Comparison of the FOXP3+ DN and FOXP3+ CD4 SP cells showed no significant differences in the expression of either CTLA-4 or CD127, whereas in the FOXP3− CD4 SP and DN thymocytes the expression pattern was clearly different (Fig. 3,A). Other Treg cell-associated molecules, GITR and CD39, were expressed by a subset of FOXP3+ CD4 SP cells, but no expression was found in the FOXP3+ DN cells (data not shown). Importantly, the two FOXP3+ subsets differed in the expression of CD69, a marker up-regulated in response to TCR engagement (21). Of the FOXP3+ CD4 SP cells, 80.8 ± 4.7% were CD69+, and of the FOXP3+ DN cells only 13.7 ± 6.3% were CD69+ (Fig. 3 B), indicating that the latter cells have not received TCR-mediated signals. The level of CD5 expression, which in the thymus has been shown to correlate with the strength of TCR-signaling (22), was also different. In FOXP3+ CD4 SP cells the mean fluorescence intensity of CD5 was 32 ± 20, whereas in the FOXP3+ DN cells it was only 3 ± 1 (n = 6).
Taken together, our data show that a subset of DN thymocytes in healthy children expresses FOXP3 in the absence of a TCR while already sharing some of the characteristics of mature Treg cells (Table I). In the absence of suitable surface markers, the isolation and functional analysis of the FOXP3+ DN thymocytes is problematic. Our attempts to isolate them indirectly by depleting cells expressing CD4, CD8, CD14, CD19, CD56, and CD127 were only partially successful. The average frequency of FOXP3+ DN cells after depletion was 38.6% (range 10.3–80.2%); in coculture experiments with autologous CD4 SP thymocytes or PBMC the sorted cells displayed no suppressive effect or IL-10 production (not shown). On a theoretical basis, it is highly unlikely that these TCR− cells would already be functional Treg cells, as TCR signaling is needed for Treg cell activation and suppressive function (3, 4). The inability to separate viable FOXP3+ cells also precluded cell cultures and differentiation assays, so the developmental role of the FOXP3+ DN thymocytes cannot be proven. However, given that the TCR− DN thymocytes represent the pool of earliest T cell precursors (15), it seems reasonable to suggest that the TCR−FOXP3+ DN thymocytes contain the precursors of FOXP3+ Treg cells. Assuming a broadly similar developmental pathway for Treg cells and other T cells, the FOXP3+ DN cells would then proceed through a TCR+ DP stage to give rise to the mature CD4+ and CD8+ Treg cells.
Although our results are at variance with murine data, recent work suggests that in mice also the induction of a Treg cell genetic program may be an early event (23) and precedes the expression of FoxP3 (24, 25). It is also important to note that our results do not dispute the crucial role of TCR in human Treg cell commitment. Analysis of the FOXP3 promoter area has identified several TCR-responsive sites (26) and TCR ligation can lead to up-regulation of FOXP3 (27), although this is not necessarily accompanied by suppressive function (28, 29, 30). In all likelihood the posited Treg cell maturation pathway involves TCR-dependent checkpoints, perhaps selecting for autoreactive clones among a stochastically generated FOXP3+ precursor pool (3, 4, 5, 6, 7). However, our data show that TCR expression is not a prerequisite for FOXP3 up-regulation and strongly suggest that, contrary to the current view, TCR-mediated signals are not responsible for the induction of FOXP3 in human thymocytes.
We thank Alison Banham for the gift of 150D/E4 anti-FOXP3 mAb, Maria Toribio for pre-Tα mAb, Marja-Liisa Karjalainen-Lindsberg for expert advice on thymus histology, and Aaro Miettinen for reviewing the manuscript.
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 supported by grants from the Academy of Finland, the Finnish Cultural Foundation, and the Helsinki University Science Foundation and by research funds from the Helsinki University Hospital.
Abbreviations used in this paper: Treg, regulatory T cell; DN, CD4−CD8− double negative; DP, CD4+CD8+ double positive; SP, single positive.