SEARCH

SEARCH BY CITATION

Keywords:

  • FOXP3;
  • Human T-cell development;
  • Treg cells;
  • Thymus

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Treg cells, best identified by the expression of the transcription factor FOXP3, play a crucial role in maintaining self-tolerance. Natural Treg cells constitute an independent thymus-derived T-cell lineage whose developmental program in humans is still ill-defined. Here, we provide evidence of a Treg-cell differentiation pathway at the double positive (DP) stage, prior to commitment to the CD4+ or CD8+ lineage, in pediatric thymuses. FOXP3+ DP cells displayed a functional IL-7 receptor and increased Bcl-2 levels that may protect them from cell death/negative selection, and an activated/suppressive phenotype that was lost as CD4 single positive (SP) cells matured and acquired egress markers. A subpopulation of FOXP3+ DP thymocytes expressing CD103 likely represents the precursor of FOXP3+ CD8SP cells, which homogeneously expressed this mucosal-homing molecule. Finally, co-cultures of DP thymocytes with primary thymic epithelial cells and multiple linear regression analyses support that FOXP3+ SP cells are largely derived from FOXP3+ DP thymocytes. Overall, our data suggest that human Treg-cell lineage commitment significantly occurs at the DP stage with possible implications for the diversity and autoreactivity of the natural Treg-cell repertoire.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

The thymus is essential for the establishment and renewal of the peripheral T-cell compartment with a diverse repertoire, able to efficiently respond to novel pathogens, yet tolerant to self-antigens. Tolerance is at least partly achieved through the thymic production of a subset of T cells termed “natural” Treg cells. Treg cells modulate the activation and function of other T cells, as well as of other cell populations of the immune system, playing important roles not only in the prevention of autoimmune diseases, but also in other clinical settings, such as tumor immunity and persistent infections 1, 2. Currently, the forkhead/winged-helix transcription factor FOXP3 is considered as the best available marker to identify Treg cells. The essential role of FOXP3 in human Treg-cell development and function is attested by the association of FOXP3 defects with the fatal condition IPEX (immune dysregulation, polyendocrinopathy, enteropathy, X-linked) syndrome 3. Although understanding Treg-cell development in the human thymus is critical for their manipulation 4, the developmental program that Treg-cell precursors undergo to give rise to mature FOXP3+ cells is still unclear.

T-cell development proceeds through a series of stages that can be tracked by the expression of CD3, CD4 and CD8. In humans, CD4CD8CD3 triple negative cells first acquire CD4 (CD4+ immature single positive) and then CD8 to become CD4+CD8+ (double positive, DP) cells. DP cells bearing TCR-αβ complexes able to engage self-MHC molecules are signaled to survive (positive selection) and to differentiate into functionally mature CD4+ single positive (CD4SP; CD4+CD8CD3high) or CD8+ single positive (CD8SP; CD4CD8+CD3high) T cells. Data from mouse models suggest that the strength and duration of TCR signaling drive the CD4+ versus CD8+ fate-decision during thymic differentiation, and the negative selection of cells with potentially auto-reactive TCRs 5. TCR signaling has also been shown to be essential for Treg-cell differentiation in the thymus 6, 7. Treg-cell lineage commitment is thought to occur within a very narrow window of affinity of TCR–ligand interactions between positive selection of conventional T cells and negative selection of high-affinity self-reactive T cells 8, 9. Indeed, increased self-reactivity is considered a hallmark of Treg cells 10.

Initial studies in human fetal and pediatric thymuses, relying on IL-2 receptor α chain (IL-2Rα, CD25) as a Treg-cell marker, identified CD25 expression in CD4SP, as well as in DP and CD8SP, thymocytes 11–17. Suppressive function was ascribed to CD25+ CD4SP thymocytes, but also to the CD25+ DP and CD25+ CD8SP populations 11, 12, 14, 16–18. We and others have assessed FOXP3 expression in the human thymus and reported its expression at early stages of T-cell development, likely impacting on the diversity and autoreactivity of the Treg-cell repertoire 19–21.

We investigated here the development of Treg cells in the human thymus and provide evidence that a considerable fraction of FOXP3+ SP cells derives from FOXP3+ DP thymocytes. Our data support a model whereby FOXP3 expression in human DP thymocytes is associated with the expression of a functional IL-7 receptor, as well as of markers of Treg-cell activation, which decline as thymocytes mature. On the other hand, the mucosal homing molecule CD103 is expressed by a subpopulation of FOXP3+ DP thymocytes and by the large majority of FOXP3+ CD8SP cells, but not by FOXP3+ CD4SP thymocytes, which suggests that FOXP3+ CD8SP cells may arise from FOXP3+ DP precursors and exit the thymus with a preferential mucosal homing phenotype.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Human FOXP3equation image DP thymocytes comprise cells uncommitted to the CD4equation image or CD8equation image lineage

FOXP3 was shown to be expressed throughout human T-cell development 19, 21 with DP thymocytes representing the most immature population where considerable amounts of FOXP3 mRNA and protein are detected 19. Accordingly, in all 35 thymuses analyzed a population of FOXP3+ DP thymocytes, corresponding approximately to a third of the total FOXP3+ CD4SP thymocytes and twice the total of FOXP3+ CD8SP cells, was identified after careful exclusion of cell aggregates (Fig. 1A and B). The latter technical detail is relevant, as overestimation of FOXP3+ DP cells in the murine thymus, due to doublet formation, has been described 22.

thumbnail image

Figure 1. FOXP3 expression in DP thymocytes. (A) Flow cytometry analysis of FOXP3 expression in a representative thymus (3 year old boy). Total thymocytes after exclusion of cell aggregates illustrating the definition of DP, CD4SP and CD8SP subsets is shown (top). FOXP3 expression in CD3high thymocytes within DP, CD4SP and CD8SP subsets and the counterpart FOXP3-population are shown (bottom). (B) Frequency of FOXP3+ DP CD3high, FOXP3+ CD4SP and FOXP3+ CD8SP cells within total thymocytes. (C) Sorted DP cells were treated with pronase and cultured overnight at 37°C. Controls included pronase-treated DP cells left at 4°C. A representative example of the re-expression of pronase-cleaved surface molecules and controls within FOXP3+ and FOXP3 cells after exclusion of aggregates and dead cells is shown (left). The frequency of thymocytes (mean of replicates) re-expressing CD4 (CD4+CD8), CD8 (CD8+CD4) or both (DP) co-receptors within FOXP3+ cells after overnight culture at 37°C of pronase-treated purified DP cells is shown (right). Each circle represents a thymus. (D) Histogram overlays and graphs show the expression, as analyzed by flow cytometry, of CD69 (left) and CD27 (right), in FOXP3+ (open histograms and bars) or FOXP3 (filled histograms and bars) CD3high thymocytes at the DP, CD4SP and CD8SP stages. Graphs show the relative MFI (median fluorescence intensity), defined as the ratio between the MFI of a marker in a given population and the MFI of the same marker within FOXP3+CD4SP cells, of CD69 (n=11; except for FOXP3+DP CD3high, n=10, and FOXP3+CD8SP, n=6) and the frequency of CD27+ cells (n=12; except for FOXP3+DP CD3high, n=10, FOXP3+CD8SP, n=7). Data are presented as mean + SEM. p-values were generated using the Wilcoxon matched pairs test: *p<0.05, **p< 0.01, ***p<0.001, ns: not significant. (E) TCR Vβ family distribution at the DP CD3high stage within FOXP3+ (open circles) or FOXP3 (black circles) cells as assessed by flow cytometry. Each line connects the frequency of a given family within the FOXP3+ DP CD3high and FOXP3 DP CD3high populations of the same thymus.

Download figure to PowerPoint

It is known that DP cells may already be committed to either the CD4SP or CD8SP lineage, and have thus shut-down the synthesis of either CD4 or CD8, but still express both co-receptors at their surface 5, 23. Therefore, our first aim was to experimentally assess whether the FOXP3+ DP population we detected in the human thymus were “true” DP cells. For this purpose, we employed a re-expression assay, which consisted of stripping surface molecules with pronase and analyzing their re-expression after short-term culture 23, 24. This technique allowed us to identify DP cells and CD4SP- or CD8SP-committed DP-like cells. Pronase treatment of sorted DP cells completely abolished surface CD4 and CD8 expression, and no re-expression of these molecules was observed when thymocytes were kept at 4°C (Fig. 1C). In contrast, pronase-treated DP thymocytes cultured at 37°C clearly re-expressed CD4 and/or CD8. Importantly, we found that a proportion of FOXP3+ DP cells re-expressed both CD4 and CD8 (Fig. 1C), thus confirming the active synthesis of both co-receptors.

Overall, these results show that bona fide, CD4+ and CD8+ lineage-uncommitted, FOXP3+ DP cells are present in the human thymus.

FOXP3equation image DP CD3high cells display an early mature phenotype and a diverse TCR Vβ repertoire

In spite of the evidence of non-commitment to the CD4+ or CD8+ lineage of FOXP3+ DP cells, the high levels of CD3 expressed suggest some degree of maturity (Fig. 1A) 19. Therefore, we compared the phenotype of FOXP3+ cells at the DP CD3high stage with that of FOXP3 DP cells expressing similar levels of CD3 (Fig. 1A). FOXP3+ DP CD3high cells expressed significantly higher levels of the early activation marker CD69, commonly associated with positive selection 25, than FOXP3 DP CD3high cells (Fig. 1D). Moreover, we found that more than 90% of FOXP3+ DP CD3high cells already expressed the maturation marker CD27 25, in contrast to approximately 20% of FOXP3 cells at a similar developmental stage (Fig. 1D).

FOXP3+ DP CD3high cells exhibited a diverse TCR repertoire as assessed by their surface TCR Vβ family distribution, and similar TCR Vβ family profiles were found between FOXP3+ and FOXP3 DP CD3high cells (Fig. 1E). Together, these results suggest that FOXP3+ DP CD3high cells constitute a pool of mature, positively selected thymocytes that display a diverse repertoire.

FOXP3equation image DP CD3high cells include a unique activated subpopulation

We assessed the expression of other Treg-cell markers within the DP CD3high thymocyte population and found a clear expression of CD25 and CTLA-4 in association with FOXP3, in agreement with their clear suppressive function in vitro (Fig. 2A and B and Supporting Information Fig. 1A and B). Notably, within FOXP3+ cells the levels of CD25 and CTLA-4 were frequently higher in DP CD3high than in CD4SP cells, despite the high levels of those markers observed in the latter population (Fig. 2B and Supporting Information Fig. 1A). Conversely, CD8SP cells expressed low levels of CD25 and CTLA-4 (Fig. 2A and B and Supporting Information Fig. 1A).

thumbnail image

Figure 2. FOXP3+ DP CD3high cells in the human thymus present an activated phenotype and express functional IL-7 receptor. (A) Frequency of Treg-cell-associated markers CD25, CTLA-4 and CD39 within FOXP3+ (open circles) or FOXP3 (grey circles) cells at the CD3high DP, CD4SP and CD8SP stages. Each circle represents a thymus. (B) Relative MFI (ratio between the MFI of a marker in a given population and the MFI of the same marker within FOXP3+ CD4SP cells) of CD25, CTLA-4 and CD39 within FOXP3+ cells expressing the marker at each stage. (C) Representative flow cytometric analysis of CD39 expression in relation to CD45RA within FOXP3+DP and FOXP3+CD4SP thymocytes. CD25high cells are highlighted in black over the contour plots representing the CD25neg/low population. Data are representative of four experiments each using a different thymus. (D) Relative MFI of HLA-DR within FOXP3+ (open bars) or FOXP3 (grey bars) cells at the CD3high DP, CD4SP and CD8SP stages, as related to the HLA-DR MFI within FOXP3+ CD4SP cells (n=8). (E) Histogram overlays and graph show the expression of CD127 (IL-7Rα), as analyzed by flow cytometry, within FOXP3+ (open histograms and bars) and FOXP3 (filled histograms and bars) CD3high thymocytes at the DP, CD4SP and CD8SP stages (n=8). (F) Representative dot plots showing phospho-STAT5 (P-STAT5) expression in relation to FOXP3 within DP thymocytes after exposure of total thymocytes to IL-7 (50 ng/mL) or PBS (unstimulated control); one of three experiments each using a different thymus. Data are presented as mean + SEM. p-values were generated using the Wilcoxon matched pairs test: *p<0.05, **p< 0.01, ***p<0.001, ns: not significant.

Download figure to PowerPoint

CD39 (ectonucleoside triphosphate diphosphohydrolase 1) is an ectoenzyme that has been associated with Treg-cell activity 26. Within the FOXP3+ population the expression of CD39 was also significantly higher in DP CD3high than in CD4SP thymocytes (Fig. 2A and B and Supporting Information Fig. 1). Of note, CD39high cells defined a population that expressed very high levels of FOXP3, CD25 and CTLA-4 (Supporting Information Fig. 2A and B). Interestingly, CD45RA expression at the CD4SP stage was associated with negative/low levels of CD39 and CD25 (Fig. 2C), suggesting that egress from the thymus with a CD45RA+ phenotype may be preceded by the down-regulation of those markers. CD8SP cells did not express high levels of CD39, in line with the fact that they lack a CD25high population (Fig. 2A and B and Supporting Information Fig. 1). Of note, CD73 (ecto-5′-ectonucleotidase), which dephosphorylates AMP, a CD39 product, to generate the immunosuppressive nucleoside adenosine 26, was not concomitantly expressed with CD39, being mainly expressed on CD8SP cells irrespective of concomitant FOXP3 expression (Supporting Information Fig. 2C and D).

HLA-DR expression in peripheral cells has been reported to identify a mature, functionally distinct Treg-cell subset that expresses higher levels of FOXP3 and mediates contact-dependent suppression 27. HLA-DR expression was highest among FOXP3+ DP CD3high cells, being particularly marked within the CD25high population (Fig. 2D and Supporting Information Fig. 2B).

In summary, our results show that FOXP3+ DP CD3high cells express other Treg-cell-associated markers, are suppressive, and include an activated subpopulation that expresses high levels of CD25, CTLA-4, HLA-DR and CD39. The decline of the expression levels of these markers at the SP stages suggests that this activation is transitory and that Treg cells exit the thymus with a resting phenotype, similar to that of peripheral naïve-like Treg cells 28.

FOXP3equation image DP CD3high cells express functional IL-7 receptor

In humans, low levels of surface IL-7 receptor alpha chain (IL-7Rα, CD127) have been associated with a regulatory phenotype 29, 30. In agreement, we found lower expression of CD127 on FOXP3+ as compared with FOXP3 cells at the SP stages (Fig. 2E). In contrast, we found significantly higher levels of CD127 in the FOXP3+ as compared with the FOXP3 population within DP CD3high cells (Fig. 2E).

In order to assess the functional activity of CD127 on FOXP3+ DP cells we measured the degree of STAT5 phosphorylation (P-STAT5) upon exposure to IL-7. The clear up-regulation of P-STAT5 observed confirmed that a subpopulation of FOXP3+ DP thymocytes expressed functional IL-7 receptor and was responsive to IL-7 (Fig. 2F and Supporting Information Fig. 3A and B). Importantly, whereas the FOXP3+ SP subsets also showed a comparable up-regulation of P-STAT5, these levels did not differ from their FOXP3 SP counterparts, in contrast with the clear difference observed at the DP stage (Fig. 2F and Supporting Information Fig. 3A and B). Since CD127 expression within FOXP3+ DP cells was associated with high levels of CD25 (Supporting Information Fig. 4A) we also compared the expression of P-STAT5 in CD25neg, CD25low and CD25high DP cells in response to different concentrations of IL-7. We found higher levels of P-STAT5 in DP cells expressing high levels of CD25, while CD25low and CD25neg cells were less susceptible to IL-7 stimulation (Supporting Information Fig. 4B).

Given the expression of IL-7Rα and IL-2Rα on FOXP3+ DP thymocytes we asked whether IL-7 or IL-2 were involved in the modulation of FOXP3 and/or CD25 expression on these cells. Total thymocytes cultured in the presence of IL-2, IL-7 or TCR stimulation (anti-CD3 plus anti-CD28 mAbs) were assayed after 6 and 22 h of culture in order to minimize the contribution of cell proliferation. We found that IL-2 was able to increase both FOXP3 and CD25 expression levels within FOXP3+ DP thymocytes (Supporting Information Fig. 4C and D). Interestingly, although IL-7 did not directly impact on FOXP3 expression, it did increase the levels of CD25 within FOXP3+ DP cells at the later time point (Supporting Information Fig. 4C and D). These results confirm that FOXP3+ DP cells are susceptible to IL-7-mediated signals, which consequently lead to an up-regulation of CD25 levels on FOXP3+ DP cells and/or a preferential survival of CD25highFOXP3+ cells in culture. Given the activated phenotype of FOXP3+ DP CD3high cells, we further assessed whether IL-2, IL-7 or TCR stimulation impacted on their CD39 and/or HLA-DR expression. Surprisingly, the expression of both molecules was severely reduced upon culture, irrespective of the stimuli provided (Supporting Information Fig. 4E and F), suggesting that other cues, such as interaction with thymic stroma, may be required for their induction/maintenance.

Overall, these data show that FOXP3+ DP human thymocytes express functional IL-7Rα, suggesting that IL-7 may play a role in human Treg-cell development.

CD103 is already present in FOXP3equation image DP CD3high cells and is expressed by most FOXP3equation image CD8SP cells

CD103 (integrin alpha E, αE) is expressed by a subset of peripheral Treg cells 31, 32 and is thought to be important for T-cell homing to mucosal sites 32, 33. At the DP CD3high stage, we found significantly higher levels of CD103 on FOXP3+ than FOXP3 cells (Fig. 3). Of note, CD103 expression at the DP CD3high stage was associated with low levels of CD25 (Fig. 3A). At the SP stages, CD103 was significantly more expressed on CD8SP than CD4SP cells within the FOXP3+ subset (Fig. 3B and C). Interestingly, the large majority of FOXP3+ CD8SP cells expressed CD103, in contrast to only approximately 10% of FOXP3 CD8SP cells (Fig. 3B). The observation of high levels of expression of a mucosal homing molecule within FOXP3+ CD8SP is particularly relevant, since FOXP3+ CD8 T cells are rarely found in human peripheral blood and have been described in the colon mucosa 34, 35.

thumbnail image

Figure 3. CD103 is expressed by FOXP3+ DP CD3high and by the majority of FOXP3+ CD8SP human thymocytes. (A) Representative contour plots showing the expression of CD103 in relation to CD25 within FOXP3+ (top) and FOXP3 (bottom) thymocytes at the CD3high DP, CD4SP and CD8SP stages. (B) Frequency of CD103+ cells within FOXP3+ (open bars) and FOXP3 (grey bars) thymocytes at the CD3high DP, CD4SP and CD8SP stages (n=9; except for FOXP3+DP CD3high and FOXP3+CD8SP, n=8). (C) Relative MFI of CD103 within CD103+ FOXP3+ (open bars) or FOXP3 (grey bars) cells at the CD3high DP, CD4SP and CD8SP stages, as related to the CD103 MFI within FOXP3+CD4SP cells (n=8). Data are presented as mean + SEM. p-values were generated using the Wilcoxon matched pairs test: *p<0.05, **p<0.01, ns: not significant.

Download figure to PowerPoint

FOXP3+ CD8SP CD103+ cells showed a diverse TCR Vβ distribution similar to FOXP3CD8SP CD103+ and CD103 cells (Supporting Information Fig. 5).

Although the CD4/CD8 ratio was much higher in FOXP3+ as compared with FOXP3 SP subsets in the human thymus (9.9±0.9 versus 3.1±0.3 for CD4SP/CD8SP ratio in FOXP3+ and FOXP3 cells, respectively; n=19), a distinct FOXP3+ CD8SP population was identified in all the thymuses evaluated (Fig. 1A and B). Notably, 11.6±1.9% of the CD103+ CD8SP population expressed FOXP3.

Our data show that FOXP3+ CD8SP cells generated in the human thymus already express the mucosal homing molecule CD103 and support the possibility that this commitment may, at least in part, occur at the DP stage.

Precursor–progeny relationship between FOXP3equation image DP CD3equation image and FOXP3equation image SP thymocytes

In order to exclude the possibility that FOXP3+ DP thymocytes are dead-end cells and to assess the precursor–progeny relationship of human FOXP3+ DP and SP thymocytes we purified CD25+ DP CD3high cells, which are largely composed of FOXP3+ cells, and co-cultured them with allogeneic primary thymic epithelial cells (TECs; Fig. 4A). This system allowed us to determine that some CD25+ DP CD3high thymocytes remain DP, in agreement with the results of our re-expression assay (first section of Results), and that a considerable proportion differentiated into FOXP3+ CD4SP and FOXP3+ CD8SP cells (Fig. 4A). In contrast, CD25 DP CD3high cells differentiated poorly into FOXP3+ DP or SP cells.

thumbnail image

Figure 4. Analysis of the precursor–progeny relationship between DP CD3high and SP stages of FOXP3+ and FOXP3 human thymocytes. (A) Differentiation of sorted CD25+ DP (top) and CD25 DP (bottom) thymocytes in co-culture with allogeneic primary TECs for 5 days. Data are representative of five experiments each using a different thymus. (B) Relationship of FOXP3+ CD4SP (left) and FOXP3+ CD8SP (right) with FOXP3+ DP CD3high thymocytes. Multiple regression modeling was performed to assess which thymocyte populations have an effect on FOXP3+ SP populations (statistical results are depicted in Table 1) using cell numbers, estimated as described in Figure1A. Observed (filled circles) and fitted (open circles) values are shown. Each circle represents one thymus. (C) Representative flow cytometry analysis of the expression of Ki67 (top) or Bcl-2 (bottom) within FOXP3+ (open histograms) or FOXP3 (filled histograms) thymocytes at the CD3high DP, CD4SP and CD8SP stages. Graphs show the frequency of Ki67+ cells (n=11; except for FOXP3+ DP CD3high, n=9, and FOXP3+ CD8SP, n=6) and the relative MFI (ratio between the MFI in a given population and the MFI within FOXP3+ CD4SP cells) of Bcl-2 (n=10; except for FOXP3+ CD8SP, n=5). Data are presented as mean + SEM. p-values were generated using the Wilcoxon matched pairs test: *p<0.05, **p<0.01, ***p<0.001, ns: not significant.

Download figure to PowerPoint

To further investigate the precursor–product relationship between FOXP3+ DP CD3high and FOXP3+ SP thymocytes suggested by our results, we undertook a multiple linear regression analysis in which a set of explanatory variables is related to a given output variable by means of a linear combination. In this analysis, we considered FOXP3+ SP thymocyte cell counts as the output variables and the remaining thymocyte cell counts as putative explanatory variables (shown in Table 1). This analysis led to quite high R2 adjusted estimates (>0.68), which suggests that the FOXP3+ SP thymocyte cell counts are mainly explained by the explanatory variables considered in the models, supporting a strong positive effect of FOXP3+ DP CD3high cells on FOXP3+ CD4SP and FOXP3+ CD8SP but not on FOXP3 CD4SP and FOXP3 CD8SP cells, irrespective of the total thymocyte frequency (Table 1 and Fig. 4B). Similar qualitative results were obtained using the log 10 of cell frequencies (Supporting Information Table 1). These results suggest that the variation observed on FOXP3+ SP cell counts is intimately related to their FOXP3+ DP counterparts. Thus, FOXP3+ DP pool seems to be the main source of the FOXP3+ SP cells. Quantitatively, the fitted models predict that one FOXP3+ DP CD3high thymocyte would give rise to 3.3 FOXP3+ CD4SP thymocytes. Such a result is in agreement with the 3-fold difference in frequencies shown in Fig. 1B, and shows that the fitted models are capturing the main features of the data. In the case of FOXP3+ CD8SP pool, one FOXP3+ DP CD3high thymocyte seems to originate 0.5 FOXP3+ CD8SP thymocytes.

Table 1. Multiple regression analysisa)
CoefficientFOXP3+CD4SP (R2 adjust=0.89)FOXP3+CD8SP (R2 adjust=0.69)
 Estimatep-ValueEstimatep-Value
  • a)

    a) Using the number of live thymocytes of each subset (analyzed as shown in Fig. 1A).

  • b)

    b) FOXP3CD4SP or FOXP3CD8SP for the FOXP3+CD4SP or FOXP3+CD8SP populations, respectively.

Intercept−14800.37−930.79
FOXP3+DP CD3high3311<0.001509.90.01
FOXP3DP CD3high−13.490.55−0.920.86
FOXP3 counterpartb)58.31<0.0019.400.21
Total thymocytes−3.690.36−0.400.70

It is worth noting that FOXP3 CD4SP cells also have a non-negligible effect on their FOXP3+ CD4SP counterparts (Table 1). In this case, the model predicts that one FOXP3 CD4SP thymocyte would lead to 0.06 FOXP3+ CD4SP thymocyte on average. For FOXP3+ CD8SP cells, this number is slightly lower (0.01 FOXP3+ CD8SP cells per FOXP3 CD8SP cell). This rather small effect might be an explanation for the lack of statistical significance of FOXP3 CD8SP in the respective model (Table 1 and Supporting Information Table 1).

Overall, these results support that: (i) at least part of the developmental pathway of FOXP3+ cells involves a sequential progression from the DP CD3high to the SP stages; and (ii) the induction of FOXP3 expression at the SP stage is likely a less frequent event than previously anticipated from mouse studies.

Notably, the distribution of TCR Vβ families of FOXP3+ cells at the DP CD3high, CD4SP and CD8SP stages exhibited comparable profiles (Supporting Information Fig. 6).

The strong positive effect of FOXP3+ DP CD3high cells on FOXP3+ SP populations also supports a distinct contribution of cell death and/or proliferation during FOXP3+ as compared with FOXP3 T-cell development. Indeed, we found a significantly lower proportion of Ki67-expressing cells in FOXP3+ as compared with FOXP3 cells at the DP CD3high and CD4SP stages (Fig. 4C). Furthermore, the FOXP3+ DP CD3high cell subset expressed significantly higher levels of Bcl-2 than FOXP3 cells, indicating that FOXP3+ cells may be more resistant to apoptosis (Fig. 4C).

Together, our results support that a significant proportion of FOXP3+ cells at the CD4SP and CD8SP stage originate from FOXP3+ DP CD3high cells, likely through a balance of cell death and proliferation distinct from that involved in the development of FOXP3 thymocytes.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

We provide here evidence of significant commitment to the Treg-cell lineage at the DP stage in the human thymus. We showed that FOXP3 expression in DP thymocytes was associated with considerable levels of functional IL-7Rα and with the activation of a transcriptional program similar to that observed in fully differentiated peripheral “activated” Treg cells. Moreover, we found that the mucosal homing molecule CD103 was expressed by a subpopulation of FOXP3+ DP thymocytes, as well as by the large majority of human FOXP3+ CD8SP cells. Finally, both the differentiation of CD25+ DP cells in a co-culture system and the statistical analyses of progenitor–product relationships in these pediatric thymuses support that a significant proportion of FOXP3+ SP cells might indeed derive from FOXP3+ DP cells with likely implications for the autoreactivity and diversity of the natural Treg-cell repertoire.

We confirmed that FOXP3 expression in human DP thymocytes occurs prior to CD4+ or CD8+ lineage commitment using a re-expression assay, consistently showing active synthesis of both CD4 and CD8 and lack of commitment to either T-cell subset in a subpopulation of FOXP3+ DP thymocytes. A previous study has reported the expression of recombination-activating gene 2 (RAG2) in CD25+ DP thymocytes, which has been interpreted as a possible sign of immaturity 21. Nevertheless, we showed that the majority of FOXP3+ DP thymocytes displayed a phenotype associated with positive selection and maturity, in agreement with the phenotype previously described in CD25+ DP thymocytes in human fetal thymuses 15, 16.

We also found that FOXP3+ DP thymocytes expressed IL-7Rα, which contrasted with the low levels found in FOXP3+ SP cells, a feature associated with the peripheral Treg-cell phenotype 29, 30. Although not in line with a recent report regarding FOXP3 expression in DP cells in postnatal thymuses 21, it is in agreement with the previous data that reported CD25 expression on DP cells of human thymic fetuses in conjunction with significant levels of IL-7Rα 16. Importantly, we further demonstrated that the IL-7Rα expressed in FOXP3+ DP thymocytes was functional, as indicated by the downstream phosphorylation of STAT5 in a significant fraction of cells upon IL-7 stimulation and the associated up-regulation of CD25 on these cells. The presence of functional IL-7Rα on FOXP3+ DP thymocytes suggests an involvement of IL-7 in human Treg-cell development. This hypothesis is further supported by our observation of significantly higher levels of Bcl-2, an anti-apoptotic family member induced by IL-7, in the FOXP3+ DP than the FOXP3 DP subset. The high levels of Bcl-2 and the activated/suppressor phenotype reported in the FOXP3+ populations may contribute to some protection from apoptosis/negative selection, as previously suggested for mouse thymic Treg cells 36.

Our results show that a genetic program reminiscent of that found in effector-suppressive Treg cells in the periphery is likely to be induced in FOXP3+ DP thymocytes, including high levels of CD25, CTLA-4, HLA-DR and CD39. In agreement with a previous study 17, we confirmed the suppressive properties of CD25+DP CD3high thymocytes. Those molecules were also expressed by a subpopulation of FOXP3+ CD4SP cells, though at lower levels, supporting the idea that the activated FOXP3+ DP cells preferentially differentiate into the CD4+ lineage. Activated FOXP3+ CD4SP cells expressed low levels of CD45RA, a marker associated with T-cell egress from the thymus and with peripheral naïve T cells, suggesting that the activation markers are progressively down-regulated in these cells before leaving the thymus. Consequently, T cells exiting the thymus as CD45RA+ are likely to display low levels of CD25, CTLA-4 and FOXP3 and to be CD39 and IL-7Rα−/low. In agreement, the expression of CD45RA in CD4SP human thymocytes 37 and in FOXP3+ CD4+ T cells 28 was associated with low levels of CD25. In addition, the fraction of FOXP3+ CD45RA+ CD4+ T cells expressing CD39 in the periphery was shown to be negligible 38.

Conversely, we showed that the expression of CD73, an ectoenzyme that, in conjunction with CD39, generates the suppressive molecule adenosine, was mainly restricted to CD8SP cells, and this was observed irrespectively of FOXP3 expression. Accordingly, a recent study reported the lack of CD73 expression by circulating human CD4+ Treg cells 38. Furthermore, our results allocate the acquisition of CD73 to CD8SP cells in the thymus, which, on egress, will most likely enrich the pool of CD73+ naïve CD8+ T cells described in humans 39.

We identified, in all the 35 thymuses evaluated, a considerable proportion of FOXP3+ cells among the CD8SP population, as previously reported 14, 21. Although FOXP3+ CD8SP did not share the activated phenotype observed in FOXP3+ DP and CD4SP cells, previous studies have shown that CD8SP cells purified on the basis of CD25 expression have suppressive properties 14. FOXP3+ CD8+ T cells are rarely found in the peripheral blood 34, 35, but CD25+FOXP3+ CD8+ T cells have been identified in normal colonic tissue and were found to be increased in colorectal cancer tissue 34. Interestingly, we found that the vast majority of FOXP3+ CD8SP cells in the human thymus express CD103, suggesting that they have the capacity to preferentially home to the mucosa. A population of CD103+CXCR3+ CD8SP cells was previously described in the human postnatal thymus, and featured an effector phenotype, given their expression of perforin and granzyme A and their ability to produce IFN-γ and IL-2 33. We also documented a population of CD103+ CD8SP cells that do not express FOXP3, suggesting that the human thymus concomitantly produces two populations of CD8SP cells expressing CD103, able to migrate to mucosal tissues, where the regulatory FOXP3+ subset may serve to counteract the activity of their effector FOXP3 counterparts. Furthermore, our data support a scenario whereby FOXP3+ DP cells that express CD103 may be diverted into the CD8+ lineage. We found very low levels of CD103 within CD4SP cells, regardless of FOXP3 expression, despite the association of this molecule with peripheral CD4+ Treg cells, particularly in mice 31, 32. This in agreement with what has been described in cord blood, where CD103+ CD4+ T cells are virtually absent 31.

The results of the co-culture system of sorted thymocyte subpopulations with TECs supported that a considerable proportion of mature FOXP3+ cells arises from FOXP3+ DP thymocytes. In agreement, it was recently suggested that human Treg-cell progenitors may selectively reside within mature DP thymocytes expressing high levels of CD69 and TCR-αβ, since these cells develop into CD4SP Treg cells in response to activated autologous plasmacytoid and myeloid DCs 40, although not excluding the possibility of FOXP3 induction at the SP stages 41, 42. Additionally, our multiple regression analyses show a significant statistical dependency of the FOXP3+ DP and SP populations, further supporting their direct precursor–product association. The numerical difference in those populations may be explained by a significant commitment to the Treg-cell lineage at the DP stage together with an accumulation of FOXP3+ SP cells prior to their exit, due to the time-lag incurred by maturation of cells in the thymic medulla 43. Recent murine studies have also demonstrated the possibility of Treg-cell generation in the cortex, followed by rapid migration into the medulla, where they mature 44, 45.

In conclusion, our data support a model in which significant induction of FOXP3 occurs at the DP stage, prior to CD4+ or CD8+ lineage commitment, in the human thymus. FOXP3+ DP thymocytes were shown to express an activated phenotype that is reminiscent of that found in peripheral activated Treg cells but is associated with the expression of a functional IL-7 receptor, which may serve to protect them from cell death/negative selection. A subpopulation of FOXP3+ DP expressing CD103 is likely to give rise to a population of FOXP3+ CD8 T cells with preferential homing to the mucosa. The early commitment to the Treg-cell lineage during human T-cell development is anticipated to impact on their repertoire and function.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Samples

Thymic specimens were obtained from routine thymectomy performed during pediatric corrective cardiac surgery at the Hospital de Santa Cruz, Carnaxide, Portugal, after parent's written informed consent. Study was approved by the Ethical Board of the Faculty of Medicine of Lisbon. Cells from pediatric thymuses (newborns to 13-year-old children) were recovered through tissue dispersion and separated on a Ficoll-Paque PLUS (GE Healthcare) density gradient.

Flow cytometry

The following anti-human monoclonal antibodies (BD Biosciences, eBioscience, DakoCytomation, Miltenyi Biotec or R&D Systems) were used: CD3 (UCHT1 or OKT3), CD4 (RPA-T4), CD8α (SK1 or RPA-T8), CD25 (2A3), CD27 (O323), CD39 (eBioA1), CD69 (FN50), CD73 (AD2), CD103 (B-Ly7), CD127 (40131), HLA-DR (L243), Bcl-2 (124), CTLA-4 (BNI3), FOXP3 (PCH101), Ki67 (B56). After surface staining cells were fixed, permeabilized and stained for FOXP3 as well as other intracellular molecules using a FOXP3 staining kit from eBioscience, according to manufacturer's protocol. 8- to 10-Parameter analysis was performed on the FACSCanto or FACSAria (BD Biosciences). Data were analyzed using the FlowJo (TreeStar). Cell aggregates were stringently excluded based on area and width parameters of both forward and side scatter.

Flow cytometry analysis of the TCR Vβ repertoire

TCR Vβ repertoire was determined using the IOTest Beta Mark kit (Beckman Coulter). Cells were incubated for 10 min with the reagent mixtures provided by the kit and subsequently stained for surface molecules and intracellular FOXP3 as described in Flow cytometry.

Flow cytometry analysis of phospho-STAT5

Thymocytes were stained for surface molecules, washed and incubated with IL-7 (0.1, 1, 10 or 50 ng/mL; R&D Systems) or PBS for 15 min at 37°C. For analysis based on CD25 expression, cells were immediately fixed in 2% formaldehyde for 10 min at 37°C and subsequently permeabilized with ice-cold 90% methanol for 30 min on ice. Thymocytes were then washed and incubated with anti-phospho-STAT5 (Y694; BD Biosciences) for 1 h at room temperature 46. For FOXP3 analysis, an additional step of intracellular staining (see Flow cytometry) was performed after IL-7 incubation and before fixation in 2% formaldehyde.

Co-receptor re-expression assay

DP thymocytes were sorted to a high degree of purity using FACSAria (BD Biosciences). Sorted cells were treated with 0.5 mg/mL pronase (Calbiochem) for 30 min at 37°C and subsequently cultured at 37°C or 4°C overnight. Analysis of the re-expression of CD4 and CD8 was performed by sequentially staining for surface molecules and intracellular FOXP3, as described above. Dead cells were excluded from the analysis using LiveDead Fixable Viability Dye (Molecular Probes).

Thymocyte cultures

Thymocytes were cultured in complete medium alone or supplemented with IL-2 (10I U/mL), IL-7 (10 ng/mL) or soluble anti-CD3 plus anti-CD28 mAbs (both 1 μg/mL; eBioscience) for 6 or 22 h. Cells were then stained for surface molecules and FOXP3 as described above. The following reagent was obtained through the AIDS Research and Reference Program, Division of AIDS, NIAID, NIH: Human rIL-2 from Dr. Maurice Gately, Hoffmann – La Roche. For thymocyte co-cultures with primary TECs, CD25 DP CD3high and CD25+ DP CD3high cells (2.5×105 cells) were purified using FACSAria and co-cultured separately with allogeneic 2.5×104 TECs for 5 days. Purities were always higher than 97% for DP within the CD25 DP CD3high population and 95% within CD25+ DP CD3high cells. The frequency of CD25 was lower than 3% within the former and higher than 70% in the latter. Human TECs were isolated as previously described 47. To evaluate the suppressor function of a given population, the population under test was plated at 2.5×104/well in U-shaped 96-well plates together with an equal number of CD25 CD4SP thymocytes (target cells). Stimulation was provided by CD3/CD28 Dynabeads (bead:cell ratio 1:4, Invitrogen). Proliferations were monitored at day 3 by the addition of 1 μCi/well of tritiated thymidine for the last 16 h of culture.

Statistical analysis

Statistical analysis was performed using GraphPad Prism v5.01 (GraphPad Software) and R software (www.r-project.org). The results were presented as arithmetic mean±SEM. Two-sample data were compared using Wilcoxon matched pairs test. Assuming that data come from a random process in steady state, multiple linear regression analyses were performed in order to disentangle the relationship between cell counts of different thymocyte populations. In general, multiple linear regression aims to quantify the effect of a set of explanatory (or input) variables on a dependent (or output) variable through means of an appropriate linear combination, which seems to hold on present data. This statistical approach has the advantage of not only detecting the most important input variables for the output, but also predicting how the output changes with the input variables. In the present work, the cell counts of either FOXP3+ SP populations were considered as the output variables while the cell counts of the remaining thymocyte populations were defined as putative explanatory variables. Cell counts were estimated based on the number of live thymocytes of each subset (analyzed as shown in Fig. 1A). Parameter estimation was done through maximum likelihood estimation while the statistical significance of the estimates was assessed through Z-score tests. For the sake of interpretability, the parameter estimates were multiplied by 1000 in order to quantify the expected changes on the output variable when there is an increase of 1000 cells in the corresponding input thymocyte population. To validate purely statistical assumptions of the models, an appropriate residual analysis was performed. In all tests, significance level was setup at 5%.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

This work was supported by grant (PTDC/SAU-MI/66248/2006) from “Fundação para a Ciência e a Tecnologia” (FCT) and by “Programa Operacional Ciência e Inovação 2010” (POCI2010) to AES. H. N. C. received a scholarship from FCT co-financed by POCI 2010. The authors thank M. Abecasis and R. Anjos for human thymus sample collection; B. Silva-Santos, J. Barata, R. B. Foxall and R. M. M. Victorino for critical review of the manuscript.

Conflict of interest: The authors declare no financial or commercial conflict of interest.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Detailed facts of importance to specialist readers are published as ”Supporting Information”. Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors.

FilenameFormatSizeDescription
eji_201141614_sm_SupplInfo.zip2803KSupplInfo

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.