High TCR diversity ensures optimal function andhomeostasis of Foxp3+ regulatory Tcells



Dominant tolerance to self-antigen requires the presence of sufficient numbers of CD4+Foxp3+ Treg cells with matching antigen specificity. However, the size and role of TCR repertoire diversity for antigen-specific immuno-regulation through Treg cells is not clear. Here, we developed and applied a novel high-throughput (HT) TCR sequencing approach to analyze the TCR repertoire of Treg cells and revealed the importance of high diversity for Treg-cell homeostasis and function. We found that highly polyclonal Treg cells from WT mice vigorously expanded after adoptive transfer into non-lymphopenic TCR-transgenic recipients with low Treg-cell diversity. In that system, we identified specific Treg-cell TCR preferences in distinct anatomic locations such as the mesenteric LN indicating that Treg cells continuously compete for MHC class-II-presented self-, food-, or flora-antigen. Functionally, we showed that high TCR diversity was required for optimal suppressive function of Treg cells in experimental acute graft versus host disease (GvHD). In conclusion, we suggest that efficient immuno-regulation by Treg cells requires high TCR diversity. Thereby, continuous competition of peripheral Treg cells for limited self-antigen shapes an organ-optimized, yet highly diverse, local TCR repertoire.


Polyclonal Treg cells establish and maintain unresponsiveness to self-antigen, regulate tolerance to food and flora antigen, and control T-cell-mediated inflammatory responses 1, 2. It is believed that the repertoire of natural (thymic) Treg cells is selected by recognition of self-antigen in the thymus 3–9 and further shaped by self-antigen recognition in the periphery 10–13. This involves TCR-MHC class II:peptide interactions with higher avidity than during positive selection of naïve CD4+ T cells 3, 14. However, due to intraclonal competition, only a limited number of thymocytes expressing the same TCR specificity are selected to develop into natural Treg cells, which ensures the generation of a highly diverse Treg-cell TCR repertoire 15, 16. In addition to the well-established essential regulation of Treg-cell homeostasis by IL-2 17–20, previous studies suggested that organ-specific self-antigen preferentially drives the survival and/or expansion of peripheral Treg-cell clones 11, 13, 21. Further studies showed that transferred antigen-specific Treg cells were proliferating in target-organ draining lymph nodes 22 and, along the same line, that transferred Treg cells from target-organ draining lymph nodes were more efficient in suppressing autoimmune disease than those of non-draining lymph nodes 23–25. Nishio et al. reported that unspecific TCR stimulation with anti-CD3 mAb surmounted intraclonal competition of Treg cells in lymph nodes and spleen and provided additional support for the notion that Treg-cell expansion in the periphery depends on self-antigen recognition 26. However, new data showed that the Treg-cell pool can remain self-sustained over months 27. Recently, comprehensive high throughput (HT) sequencing studies revealed a very high TCR diversity in human Treg cells, comparable to other T-cell subsets including naïve T cells 28. This led us to the hypothesis that broad TCR diversity may be important for Treg-cell homeostasis and immuno-regulatory function. To address this, we compared highly diverse Treg cells from WT mice with less diverse Treg cells derived from Rag-sufficient TCR-transgenic (TCR-Tg) mice. In the latter, endogenous TCR rearrangements permit the generation of natural Treg cells with a polyclonal, albeit narrower, TCR repertoire compared with WT mice. Therefore, TCR-Tg mice turned out to be a valuable tool for analyzing the physiological impact of TCR diversity on Treg-cell function. In this system, we performed adoptive transfer experiments and revealed a robust homeostatic advantage of WT Treg cells in TCR-Tg recipients with a less complex Treg-cell repertoire. Such sustained survival and expansion of transferred Treg cells allowed us to recover sufficient numbers of WT Treg cells to correlate their TCR sequences and organ-specific distribution. Furthermore, we analyzed the influence of TCR repertoire size on in vitro suppressive capacity of Treg cells and compared these results with their ability to suppress allogeneic T-cell responses in an in vivo model of lethal acute GvHD. We conclude that, within the limitations of an IL-2-dependent homeostatic niche, TCR diversity is required for optimal Treg-cell homeostasis and suppressive function.


Treg cells in TCR-Tg mice use non-clonotypic TCR rearrangements and have a polyclonal TCR-repertoire

In this study, we used Rag-sufficient OT-II TCR transgenic mice in which the TCR repertoire of Treg cells is limited to non-clonotypic ‘escapees’ that are selected on endogenous Tcrb and/or Tcra rearrangements. To monitor and sort Foxp3+ Treg cells, we crossed male homozygous TCR-Tg and female Foxp3-eGFP reporter mice. Male F1-offspring are hemizygous for Foxp3-eGFP and carry the pre-rearranged TCR. GFP+ Treg cells in TCR-Tg mice expressed no or only low levels of the clonotypic TCR and are selected for endogenous TCR rearrangements (Supporting Information Fig. 1) 29, 30. These observations and previous studies of Treg cells with restricted TCR rearrangement options 7, 12, 31 supported the hypothesis that Treg-cell repertoires of TCR-Tg mice are diverse but narrower than those of congenic WT mice.

The polyclonal Treg-cell TCR repertoire diversity is constricted in TCR-Tg mice

HT sequencing has recently become available to comprehensively characterize TCR repertoires on the level of nucleotide sequences. We chose primers spanning the variable region between the constant Cα and 12 V-elements of the Vα8 (also TRAV12) family. A similar sequencing method was recently applied to describe and compare the TCR repertoire of human naïve and memory CD8+ T cells 32. After sequence analysis of several thousands of individual Tcra rearrangements, we used this information pars pro toto to characterize and compare TCR diversity in Treg cells sorted from Foxp3-eGFP (here used as WT) and Foxp3-eGFP×OT-II TCR-Tg. Figure 1A depicts 23 718 individual rearranged Tcra sequences from each WT and TCR-Tg Treg cells by size distribution. Both of these ‘virtual Vα8-Cα spectratyping’ plots showed similar strong bias for multiples of three nucleotides, reflecting a preference for in-frame VJ rearrangements. Among the 23 718 Tcra sequences of both Treg-cell populations, we found high numbers of unique sequences, namely 10 746 clones with one single copy (and 2139 clones with two copies) in WT Treg cells and 6377 clones with one single copy (and 1341 clones with two copies) in Treg cells from OT-II TCR-Tg mice (Fig. 1B). Of note, the most abundant sequence in WT Treg cells had 71 copies, whereas 15 sequences from the TCR-Tg Treg cells had more than 100 and up to 1254 copies (Fig. 1B). Total numbers of all individual sequences added up to 14 622 different sequences for Treg cells from WT and only 9275 for TCR-Tg Treg cells. Thus, Treg-cell diversity in the TCR-Tg mice was reduced to 63% of the WT (Fig. 1C). Subsequently, we compared all productive VJ rearrangements according to the international ImMunoGeneTics information system IMGT® 33. Among the 23 718 sequences of each pool, 10 353 individual productive VJ rearrangements on the nucleotide level were found in WT and 5657 in TCR-Tg Treg cells (Fig. 1C). These encoded 6123 and 3459 distinct CDR3α respectively (Fig. 1C). These data suggested that on the amino acid level, the diversity of TCR antigen recognition in OT-II TCR-Tg Treg cells was reduced at least to 56% of WT. Qualitative comparison showed that 1295 of the CDR3α sequences from the TCR-Tg were identical to those from WT Treg cells (Fig. 1D). Collectively, our HT sequencing data showed that TCR-Tg Treg cells were essentially normal on a single cell basis but that their TCR repertoire was less diverse than that of WT Treg cells.

Figure 1.

Treg cells in TCR-Tg mice have a narrow TCR-repertoire. 1×106 Foxp3+ Treg cells were sorted from WT and TCR-Tg mice to a purity of >96% and Vα8-Cα amplicons were amplified from cDNA before high-throughput sequencing. (A) Length distribution of 23 718 CDR3α-spanning Vα-Cα sequences derived from WT (blue) and TCR-Tg (red) Treg cells. (B) Frequency of individual Vα-Cα sequences observed in WT (blue) and TCR-Tg (red) Treg cells. The copy number of Vα-Cα sequences is plotted against the number of individual nucleotide sequences. (C) Bars show summary of rearrangement analysis of 23 718 Vα-Cα sequences of WT (blue) and TCR-Tg (red) Treg cells using IMGT/HighV-QUEST. (D) Proportional Euler diagram shows amount of CDR3 regions with identical amino acid sequences from WT (blue) and TCR-Tg (red) Treg cells. Analysis of one of two independent sorting and sequencing experiments that gave similar results is shown.

Treg cells with a broad TCR repertoire expand in hosts with a narrow TCR repertoire

To investigate how TCR diversity would affect their homeostasis, we performed adoptive cell transfers. In former studies, Treg cells adoptively transferred into WT mice have been followed for up to several wks, although recovery rates were generally very low 34, 35. Here, purified Foxp3+ WT Treg cells with a broad TCR repertoire showed a robust and continuous expansion when transferred into TCR-Tg hosts with restricted Treg-cell TCR diversity (Fig. 2A and B). After 2 months, donor Treg cells constituted approximately 20% of all Treg cells in the recipient blood and peripheral lymph nodes (pLNs). Conversely, this phenomenon was not observed when TCR-Tg Treg cells with a narrow TCR repertoire were transferred into WT hosts (Fig. 2B, left panel). Moreover, transfer of CD4+Foxp3 WT cells led neither to a considerable expansion of CD45.1 donor cells nor to a significant conversion into Foxp3+ WT Treg cells (Fig. 2B, right panel). These results could be confirmed in a different experimental system by employing luciferase-expressing Treg cells 36 with a WT TCR repertoire. Monitoring Foxp3-specific light emission clearly showed similar WT Treg-cell expansion in OT-II TCR-Tg hosts over time (Fig. 2C). In addition, the same effect of efficient competition with the less diverse endogenous host Treg cells was observed after transfer into a different strain of TCR-Tg hosts, namely OT-I (Fig. 2C). Furthermore, WT donor Treg-cell frequency in TCR-Tg hosts correlated to the input dose (Fig. 2D) and their expansion was associated with higher proliferation rates and an activated phenotype (Fig. 2E and F). Re-analysis of recovered donor Treg cells 2 months after transfer revealed a reduced but still highly diverse TCR repertoire when compared with similar numbers of the control group (Supporting Information Fig. 2). Therefore, sustained survival and expansion was not restricted to a small number of clones but included a broad set of donor Treg cells. This suggests that TCR diversity and continuous self-antigen recognition control the total size of the peripheral Treg-cell pool independently of homeostatic factors such as IL-2. Of note, exogenous administration of recombinant IL-2 increased Treg-cell proliferation and absolute numbers in both WT and TCR-Tg mice (Supporting Information Fig. 3). Taken together, these adoptive transfer experiments revealed a hitherto unappreciated role for TCR diversity in Treg-cell homeostasis and imply that it is probably a combination of TCR specificity and TCR-independent factors that determine on the one hand the competitive/homeostatic fitness of Treg cells and on the other hand the total pool/niche size.

Figure 2.

Treg cells with a broad TCR repertoire expand in hosts with a narrow TCR repertoire. Adoptive transfers of FACS-sorted WT Treg cells or CD4+Foxp3 into WT or TCR-Tg recipients and vice versa. (A) Analysis of WT Treg cells in pLNs (inguinal, brachial) of TCR-Tg recipients 9 wk after transfer of 4×106 WT Treg cells. Numbers next to CD45.1GFP+ gate show mean frequency ±SD, n=6, of CD45.1 WT Treg cells (blue) of all Foxp3+ cells. (B) Lines indicate proportion of the indicated donor T cells of all CD4+Foxp3+ (left panel) and CD4+Foxp3 (right panel) in blood samples of recipients (n=1–2). (C) 1.1×107 enriched CD4+ cells from Foxp3.LuciDTR-4 donors were injected into WT, OT-I, and OT-II recipients (n=2). Mice were repeatedly imaged 5 min after injection of D-luciferin at the indicated time points. (D) Bar graphs show mean frequency + SEM of WT Treg cells in TCR-Tg recipients gated as in (A) 9 wk after transfer of either 2×106 (hatched blue) or 4×106 (blue) FACS-sorted WT Treg cells. P-value calculated by two-tailed unpaired t-test. (E) Dot plot showing at 20 h after BrdU injection the mean frequency±SD, n=3, of BrdU+ endogenous Treg cells (red) and transferred WT Treg cells (blue) in TCR-Tg recipients 9 wk after transfer of 4×106 FACS-sorted WT Treg cells gated as in (A). (F) Representative dot plot showing expression of CD62L and CD44 of endogenous Treg cells (red) and transferred WT Treg cells (blue) in TCR-Tg recipients gated as in (A) 9 wk after transfer. Data shown are representative of three independent experiments.

Treg cells from WT and TCR-Tg mice are not intrinsically different

In principle, endogenous rearrangements in TCR-Tg mice were able to produce any potential TCR chain combination and thus there were no distinct gaps in their Treg-cell repertoire. However, we still observed a few qualitative differences on the Treg-cell population level. Jα-usage of the analyzed Vα8 family sequences in Treg cells from TCR-Tg mice showed an increased proportion of the elements TRAJ5*01 and TRAJ34*02 (Supporting Information Fig. 4), while Jα-element usage was consistent in independent experiments for both types of Treg cells (Supporting Information Fig. 4). It is likely that this biased Jα-usage reflects preferential selection of Tcra rearrangements that can efficiently pair with the clonotypic Tcrb chain. Furthermore, productive VJ rearrangements in TCR-Tg mice included on average more non-templated N-nucleotides compared with WT Treg cells (6.679±0.079 versus 5.89±0.050 N-nucleotides; p<0.0001). Also, we found lower isoelectric point (pI) values of pH 9.289±0.029 for the Treg cells from TCR-Tg mice versus pH 9.473±0.021 (p<0.0001) in WT. This is in line with a report showing that CDR3α regions of Treg cells have a relatively high pI 12. Because of these significant, albeit subtle, differences, we wondered whether individual Treg cells derived from TCR-Tg mice were intrinsically less competitive than WT Treg cells. For that reason, we generated mixed BM chimeras of WT and TCR-Tg mice and compared thymic and peripheral Treg-cell levels. When a 1:1 ratio of both donors was used to reconstitute lethally irradiated recipients, we found only a marginal contribution of TCR-Tg precursors to the generation of the thymic and peripheral Treg-cell pool (Fig. 3). This is consistent with the assumption that only a few T-cell precursors in TCR-Tg mice are able to rearrange proper endogenous TCR chains prior to positive selection by the transgenic TCR. However, in chimeras derived from 20 parts TCR-Tg to 1 part WT BM, approximately 15% of thymic Treg cells were from the TCR-Tg donor as defined by the congenic markers Thy.1.1 and Thy1.2 (Fig. 3). This frequency did not decrease in the periphery, indicating that TCR-Tg donor-derived Treg cells showed similar fitness to compete for peripheral Treg-cell niches once successfully developed in competition with WT Treg cells. We cannot rule out that the repertoire of TCR-Tg donor-derived Treg cells may be skewed in a competitive environment. However, we can conclude that rearrangement of endogenous TCR chains in OT-II TCR-Tg mice generates Treg cells that individually are as fit as Treg cells in WT mice.

Figure 3.

Frequency and fitness of WT and TCR-Tg donor-derived Treg cells in mixed BM chimeras. Lethally irradiated C57BL/6 WT mice received BM transplantation of either a 1:1 or 1:20 mixture of Thy1.2Foxp3-eGFP WT to Thy-1.2+Foxp3-eGFP-OT-II, respectively. Bar graphs show mean proportion–SEM of Treg cells that developed from Thy1.2 (WT=white) and Thy1.2+ (TCR-Tg=black) precursors among all CD4+Foxp3+ Treg cells in different organs as analyzed by FACS 8 wk after reconstitution. Data are representative of two independent experiments with n=4 mice for each condition.

Organ-specific maintenance and expansion of transferred Treg cells

A recent study suggested that the Treg-cell repertoire varies by anatomical location 13. However, it was so far difficult to address the influence of TCR specificity on Treg-cell homing in adoptive transfer experiments because recovery rates were not sufficient. Here, 9 wk after adoptive transfer, the distribution of WT Treg cells into TCR-Tg hosts showed a clear preference for pLN and spleen over mesenteric lymph nodes (mLNs) (Fig. 4A). Input Treg cells were pooled from spleens and all lymph nodes, comprising approximately 15–20% mLN-derived Treg cells. In contrast, one would likely need to perform a very high number of experiments in order to decide whether significant organ-specific homing might occur after transfer into WT mice because recovery rates were approximately 100-fold lower (Fig. 4B). It is possible that dissimilar expression of gut-associated lymphoid tissue (GALT) homing receptors of the donor Treg cells additionally influenced their migration in the host. When comparing Treg cells from spleen, pLN, and mLN of WT and OT-II TCR-Tg mice, we found that the frequency of double-positive cells for the GALT homing markers CCR9 37 and of the homing/activation marker CD103 38 was increased in mLNs compared with that in pLNs (Fig. 4C). However, we largely observed only minor differences in the expression of CCR9 and CD103 (Fig. 4C). Hence, it is plausible that antigen specificity and expression of a distinct homing signature correlate with each other and are not mutually exclusive determinants of the observed organ-specific Treg-cell distribution. In mLNs, available MHC class II presented antigen may also comprise considerable proportions of intestinal antigen derived from food and bacterial flora. Therefore, we investigated the TCR sequence overlap of re-isolated donor Treg cells from spleen, pLN, mLN, and LPL (lamina propria lymphocytes) 9 wk after adoptive transfer of WT Treg cells as described for Fig. 2. We were able to analyze several thousands of recovered Treg cells and revealed strikingly overlapping Tcra rearrangements in mLN and intestinal LPL (Fig. 5A). Comparing the 25 most abundant CDR3 sequences from each tissue, we found that mLN and LPL samples shared 14 out of 25 identical AA sequences, whereas only one was similar between pLN and mLN or pLN and LPL (Fig. 5B and Table 1). Next, we asked to what extent such organ-specific expansion would be specific for Treg cells as compared with Foxp3 T cells. Therefore, we performed adoptive transfers of either pLN or mLN whole lymphocyte suspensions from CD45.1 WT mice into CD45.1+ TCR-Tg recipients (Fig. 6A). The percentage of input Foxp3+ Treg cells among all CD4+-gated T cells was similar in both cell suspensions. Nine wks after transfer of pLN cells, the frequency of Treg cells among all CD45.1CD4+ input T cells was assessed. It had increased in spleen, pLN, and mLN (Fig. 6A and B), which is in line with the Treg-cell expansion after transfer of purified Treg cells shown above. A decreased proportion among LPL may reflect antigen-specific expansion of Foxp3CD4+ T cells. At the same time, transfer of mLN cells resulted in stable proportions of Treg cells in LPL and elevated frequencies in both mLN and pLN (Fig. 6A and B). Interestingly, expansion of mLN-derived Treg cells was similar in pLN and mLN, although lower than the expansion after transfer of pLN suspensions (Fig. 6B). In conclusion, these results suggested that, besides homing receptor cues, organ-specific TCR shaping created distinct, highly diverse but still overlapping TCR repertoires in pLNs and mLNs. After transfer, such locally optimized TCR repertoires supported the maintenance of donor Treg cells in their respective organs of origin.

Figure 4.

Preferential expansion of transferred Treg cells in peripheral LN and spleen as opposed to GALT. Comparison of donor Treg cell frequency in different lymphoid organs. (A, B) Bar graphs show mean frequency + SD of WT donor Treg cells 9 wk after transfer in (A) TCR-Tg recipients or (B) WT recipients gated as described in Fig. 2A in spleen, pLN, and mLN. P-values calculated by two-tailed unpaired t-test. (C) Dot plots show expression of CCR9 and CD103 on Treg cells from WT (upper panel) or TCR-Tg mice (lower panel) in spleen, pLN, mLN and LPL. Numbers in plots show mean frequency±SD of the corresponding gate; FACS plots are representative of three independent experiments.

Figure 5.

Biased yet diverse organ-specific TCR repertoires reveal overlapping antigen-specificities of Treg cells in mLN and intestinal lamina propria. (A, B) Statistical analysis of TCR repertoire similarity of transferred WT Treg cells between organs 9 wk after transfer into TCR-Tg recipients. (A) Morisita-Horn indices are calculated to compare similarity of CDR3 amino acid sequences of recovered Treg cells between indicated organs. (B) Proportional Euler diagram shows overlap of the 25 most abundant CDR3 regions in Treg cell samples recovered from pLN (green), mLN (red), and LPL (blue) 9 wk after transfer into TCR-Tg recipients. Data were pooled from n=3 mice, comprising 5979, 4798, and 1938 functionally rearranged clones from pLN, mLN, and LPL respectively. Detailed sequence information is shown in Table 1.

Figure 6.

Expansion after transfer is Treg-cell specific and depends on the origin of donor cells. 1×107 lymphocytes from either pLN (A, left side; B, black) or mLN (A, right side; B, white) of WT donors were adoptively transferred into TCR-Tg recipients. (A) Upper dot plots show the input frequency of GFP+ Treg cells among all CD4+ T cells and lower dot plots show the frequency of GFP+ Treg cells among all transferred CD4+ T cells in pLN and LPL of recipients 9 wk after transfer. (B) Input frequency of GFP+ Treg cells was defined as 1, bar graphs show fold change of GFP+ Treg cells 9 wk after transfer. Data were combined from three independent experiments with two mice per group each. p-Values calculated by two-tailed unpaired t-test.

Table 1. Top CDR3 sequences of transferred WT Treg cells
  1. a

    The 25 most abundant CDR3 sequences among all recovered transferred WT Treg cells were chosen from all productive clones of pLN (5979), mLN (4798), and LPL (1938) and highlighted in green, red, and blue respectively.

  2. b

    a) Numbers in brackets indicate for each tissue the total amount of clones with productive Tcra rearrangements (IMGT) of donor Treg cells recovered from TCR-Tg mice 9 wk after transfer of WT Treg cells.

  3. c

    b) Number of clones that code for the corresponding CDR3 AA sequence in the left column.

original image

A broad TCR-repertoire is required for Treg-cell mediated suppression of GvHD

Next, we investigated the impact of Treg-cell repertoire diversity on their genuine function, i.e. their capacity to suppress T-cell activation. In an in vitro system based on T-cell activation with anti-CD3 mAb, Treg cells from TCR-Tg mice were equally efficient as Treg cells from WT mice in suppressing the proliferation of CD8+ and of CD4+ T cells (Fig. 7A and B). In contrast, in an experimental model of acute GvHD 35 less diverse Treg cells were less efficient than WT Treg cells in preventing the lethal disease (Fig. 7C and D). Co-transfer of allogeneic Treg cells derived from OT-II TCR-Tg mice showed only alleviation of the disease but not protection from GvHD (Fig. 7C and D). Taken together, these results suggest that the impact of TCR diversity on Treg-cell function is context dependent. Limited diversity is not per se affecting the capability of Treg cells to suppress unspecific T-cell proliferation in vitro, but it is an essential prerequisite for effective suppression of allogeneic T-cell activation in vivo in a MHC-mismatched GvHD model.

Figure 7.

TCR-repertoire size compromises Treg-cell function in vivo in experimental GvHD but not in vitro. (A, B) In vitro suppression by WT and TCR-Tg Treg cells. Treg cells were co-cultured with CFSE-labeled Thy1.2+ T cells in the presence of anti-CD3/anti-CD28 beads. Frequency of proliferated CFSElow (A) CD8+ and (B) CD4+ Thy1.2+ T cells are shown for the indicated ratios of Treg cells to effector cells. Data are representative of three independent experiments with four replicates each. (C, D) In vivo suppression of experimental GvHD. (C) Score and (D) survival of lethally irradiated BALB/c mice after transplantation of allogeneic BM and effector T cells alone or together with equal numbers of WT or TCR-Tg Treg cells as indicated. Data are pooled from three independent experiments.


Here, we present evidence that TCR diversity is an essential aspect of Foxp3+ Treg-cell homeostasis and function. Treg cells with a broader TCR repertoire exhibited sustained survival and expansion in hosts with less diverse Treg cells, which likely reflected their advantage in competition for self peptides and other peptides presented by MHC class II. Adoptive transfer experiments revealed that the TCR repertoire of Treg-cell populations varied by anatomical location. Functionally, our data strongly suggest that TCR diversity is a critical factor for efficient Treg-cell mediated suppression of experimental acute GvHD.

If not crossed to a Rag-deficient background, TCR-Tg mice contain functional Treg cells that develop through thymic selection of endogenous, non-clonotypic TCR rearrangements 14, 39, 40. Only in rare exceptions, e.g. in AND- or HA- TCR-Tg mice 41, 42, a limited number of clonotypic thymocytes was shown to develop into Foxp3+ Treg cells 15, 16, 43. Here, the use of broadly available OT-II TCR-Tg as Treg-cell recipients allowed efficient in vivo expansion of adoptively transferred WT Treg cells with a broader TCR repertoire. Moreover, congenic markers in combination with the eGFP-reporter in the Foxp3 locus assured unambiguous detection of Treg cells after adoptive transfer. To the best of our knowledge, such a robust expansion of adoptively transferred Treg cells as described here is unprecedented in non-lymphopenic mice.

Several studies in humans and mice have implied that TCR diversity is an important feature of Treg cells. A comprehensive study on one single human T-cell repertoire recently concluded that Treg cells were the most diverse T cells 28. The authors predicted 89 920 TCRα CDR3 sequences in Treg cells (defined as CD4+CD25+) compared with 58 325 in all other naive and transitional CD45RA+ non-Treg cells. This is in line with former data obtained by spectratyping of human Treg-cell CDR3 regions 44, 45. Furthermore, earlier studies using classical sequencing approaches also found at least similar diversity in mouse Treg cells 6, 7. Our study demonstrated that the TCR repertoire of WT mouse Treg cells was indeed very broad, however, at least TCR-Vα8 CDR3 diversity was found to be even higher in WT Foxp3CD4+ T cells than in Treg cells (Supporting Information Fig. 2). Recent studies suggested that thymic intra- and interclonal competition for limited antigen presented on MHC class II may be an important mechanism to generate Treg cells with a broad TCR spectrum 15, 16, 46. This was specific for natural Treg cells but not for Foxp3CD4+ T cells and thus led to the conclusion that TCRs from Treg cells may on average have higher affinity for self-peptide-MHC. Here, our findings suggest that Treg cells continue to compete in a similar way for limited matching self-peptide-MHC complexes after egress from the thymus, which in turn ensures an optimal organ-specific but diverse repertoire of relevant self-reactive TCR in the periphery.

The impact of TCR repertoire diversity on Treg-cell function is controversial. Regarding the prevention of autoimmune disease, previous studies on the effective suppression of EAE through Treg cells with limited TCR repertoires came to divergent conclusions 47, 48. A recent study by Adeegbe et al. found that limited TCR diversity of transferred Treg cells was a risk factor for autoimmune disease in IL-2Rbeta−/− mice 49. Intriguingly, non-obese diabetic mice were recently shown to select a low diversity Treg-cell TCR repertoire 50.

Understanding the parameters that govern Treg-cell homeostasis will be critical for the design of future Treg-cell-based intervention strategies. Sufficient availability of organ-specific antigen must be considered in translational attempts to manipulate organ-specific autoimmunity with engineered Treg cells of known self-peptide specificity. Otherwise, exogenous therapeutic Treg cells may be lost quickly after transfer. Previous studies suggested that organ-specific self-antigen preferentially drives the survival and/or expansion of organ-specific Treg-cell clones 11, 13, 21, 22. Our results also support the view that the antigen specificity of Treg cells changes by anatomical location, although TCR sequences of recovered Treg cells from pLNs and mLNs were largely overlapping. This may be the result of two possible scenarios. Either Treg cells recirculate less than naïve T cells or differences are due to selective local survival.

Importantly, our study infers that Treg-cell diversity is connected to diversity and availability of specific self- and foreign-antigen and thus the amount of DCs presenting it on MHC class II. In accord, it was recently shown that DC ablation reduced Treg-cell frequencies 51, 52, whereas an increase of DC numbers by FLT3L treatment led to expansion of peripheral naturally occurring Treg cells 52, 53. However, in the latter report, it was concluded that Treg-cell proliferation was mainly IL-2 dependent. In our study, we also recognized IL-2 as a master regulator that controls the absolute size of the Treg-cell pool. We propose that an optimal and maximally broad organ-specific Treg-cell TCR repertoire is continuously shaped by inter- and intraclonal competition for diverse antigen. Within a peripheral Treg-cell niche, sufficient population diversity seems to be crucial for proper Treg-cell function. Hence, in future studies, HT-sequencing analysis of Treg-cell diversity may be suitable to predict the relative risk of T-cell-mediated diseases.

Materials and methods


C57BL/6-Foxp3eGFP (here: WT) 54, C57BL/6-Foxp3.LuciDTR-4 36, and C57BL/6-Tg(TcraTcrb)425Cbn/J (here: OT-II/TCR-Tg) 55 mice have been described. The Thy1.1 allele of the C57BL/6-Foxp3eGFP mice was derived from the original Bruce 4 ES cells 35. C57BL/6J, BALB/cJ, C57BL/6-Tg(TcraTcrb)1100Mjb/J (here: OT-I), and C57BL/6.SJL-Ptprca (CD45.1) mice were obtained from Charles River (Germany). Mice were bred and housed under specific pathogen free (SPF) conditions in the central animal facility of Hannover Medical School (Germany) and used at 6–12 wk of age. All experiments were approved by the Local Institutional Animal Care and Research Advisory committee and authorized by the local government. This study was conducted in accordance with the German Animal Welfare Law and with the European Communities Council Directive 86/609/EEC for the protection of animals used for experimental purposes.

Antibodies and reagents

Anti-CD4-PacificOrange (RmCD4-2), anti-CD4-PacificBlue (GK1.5), anti-CD4-Cy5 (RmCD4-2), anti-CD8β-PacificOrange, anti-CD8β-biotin (RmCD8), and anti-CD62L-PacificOrange (MEL-14) were purified from hybridoma supernatants and conjugated in house. Anti-CD44-PacificBlue (IM7), anti-TCRβ-allophycocyanin-Alexa750 (H57-597), anti-Thy1.2-PE (MMT1), and anti-CD62L-allophycocyanin-AlexaFluor780 (MEL-14) were obtained from eBioscience. Anti-CD25-PerCP-Cy5.5 (PC61), anti-BrdU-Alexa647 (mglG1k), anti-Thy1.1-biotin (HIS51), anti-CD45.1-Alexa405 (A20), anti-CD103-PE (M290), anti CD8α-allophycocyanin-Cy7 (53-6.7), anti-Vα2-PE (B20.1), anti-Vβ3-PE (KJ25), anti-Vβ4-PE (KT4), anti-Vβ5-biotin (MR9-4), anti-Vβ6-PE (RR4-7), anti-Vβ7-PE (TR310), anti-Vβ8-PE (F23.1), anti-Vβ11-PE (RR3-15), and Streptavidin coupled to PE-Cy7 or PerCP were purchased from BD Bioscience. CCR9 staining with rat anti-mouse CCR9 (7E7-1-1) was performed as described 56. Human rIL-2 (Roche) was obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH.

Cell isolation

Lymph nodes and spleens were mashed through a 100-μM nylon gauze and washed with PBS/3% FCS (PAA). Spleen and blood samples were treated with erythrocyte-lysis buffer. For isolation of LPL, gut content and Peyer's patches were removed before intestines were opened longitudinally, washed twice in cold PBS/3% FCS, and incubated 3×15 min in HBSS (Gibco) with 10% FCS and 2 nM EDTA at 37°C. After each incubation step, tubes were shaken for 10 s and the supernatant was discarded. Intestines were washed once in PBS, incubated 2×45 min in RPMI 1640 (Gibco) containing 10% FCS, 0.24 mg/mL collagenase A (Roche), and 40 U/mL DNase I (Roche) at 37°C, then tubes were shaken for 10 s, and cell suspensions pooled, resuspended in 40% Percoll (Amersham) in RPMI 1640/PBS, overlaid onto 70% Percoll in RPMI 1640/PBS, and centrifuged at 2000 rpm for 20 min at room temperature. LPL were recovered from the interphase and washed with PBS/3% FCS.

Flow cytometry and BrdU labeling

To assess BrdU incorporation, mice received 2 mg BrdU in PBS i.p. and were sacrificed after 20 h. Before staining, cell suspensions were incubated at 4°C for 5 min with Fc block (mAb 2.4G2). For multicolor staining, cells were incubated with antibodies for 15 min at 4°C, washed with PBS/3% FCS, and, if required, incubated with secondary reagent SAv-PerCP or SAv-PE-Cy7 for 10 min at 4°C and washed with PBS/3% FCS. BrdU staining was performed with the APC BrdU Flow Kit (BD Biosciences) according to manufacturer's protocol. Flow cytometric analysis was performed on an LSR II cytometer (BD Bioscience) equipped with the BD FACSDiva software. Post acquisition analyses were conducted using the FlowJo software (Treestar).

TCR amplification and HT sequencing

Total RNA was isolated using the RNeasy Mini Kit (Qiagen) from 6×105 FACS-sorted GFP+ Treg cells, purity >95%, from either 2 WT or 2 OT-II donors per experiment. cDNA templates were synthesized using SuperScript® II reverse transcriptase (Invitrogen) according to manufacturer's recommendation. To generate template libraries of rearranged TCR CDR3 regions from Treg-cell cDNA for the Genome Sequencer FLX System (454 sequencing, Roche), we used primers spanning the variable region between constant Cα and V elements of the Vα8 family (comprising TRAV12-1*01, TRAV12-1*03, TRAV12-1*04, TRAV12-1*05, TRAV12D-2*01, TRAV12D-2*02, TRAV12D-2*03, TRAV12D-2*04, TRAV12D-2*05, TRAV12D-3*01, TRAV12D-3*02, and TRAV12D-3*03). (For primers and PCR conditions please see Supporting Information Table 1.) Forward and reverse primers contained at their 5′ ends the universal adapter sequences and a multiplex identifier (MID) respectively. Amplicons were purified by agarose gel electrophoresis and QIAquick Gel Extraction Kit (Qiagen), and quantified by Quant-iT™ dsDNA HS Assay Kit (Invitrogen). Single PCR amplicon molecules were immobilized onto DNA Capture Beads within an oil–water emulsion to enable clonal amplification in a second PCR process with universal primers. The emulsion was then disrupted and isolated beads were loaded onto PicoTiterPlates. Sequencing reactions were performed by ultra-deep 454 pyrosequencing on the Genome Sequencer FLX system (Roche Applied Sciences). Productive rearrangements and CDR3α regions were defined by comparing nucleotide sequences to the reference sequences from IMGT®, the international ImMunoGeneTics information system® (http://www.imgt.org) 33. Rearrangements were analyzed and CDR3α regions were defined using IMGT/HighV-QUEST 57.

Adoptive cell transfer

For transfers of purified cell populations, suspensions from pooled spleens and lymph nodes (inguinal, brachial, axillary, submandibular, and mesenteric) were enriched by magnetic beads (CD4+ T Cell Isolation Kit, MiltenyiBiotec) and subsequently sorted into Foxp3+ and Foxp3 cells by FACS. 4×106 or 2×106 of either Foxp3+ or Foxp3 sorted cells, 1×107 unpurified pLN or mLN cell suspensions, or 1.1×107 enriched CD4+ cells from Foxp3.LuciDTR-4 donors were resuspended in 150 μL PBS and injected into the lateral tail vein of indicated recipient mice. After 9 wk, mice were sacrificed and pLN, mLN, spleen, and the small intestine were taken to recover and analyze transferred Treg cells identified by congenic markers and GFP reporter fluorescence.

Bioluminescence imaging

Mice were imaged 5 min after i.p. injection of 4.5 mg D-luciferin (SynChem, Elk Grove Village, IL, USA) using the IVIS 100 imaging system (Xenogen, Hopkinton, MA, USA) as described 36. Analysis was performed with the Living Image software (v2.50, Xenogen).

Generation of BM chimeras

Lethally irradiated (9 Gy) C57BL/6 WT recipients received adoptive transfer of a total number of 1×107 BM cells that were either a 1:1 or a 1:20 mixture of Thy1.2Foxp3-eGFP WT to Thy1.2+Foxp3-eGFP OT-II, respectively. Chimeric mice were analyzed at 8–10 wk after transfer.

Experimental acute GvHD model

The C57BL/6 (H-2b) into BALB/c (H-2d) acute GvHD model was performed as described elsewhere 35. In addition, some groups received 0.5×106 sorted Treg cells from WT or OT-II mice with a purity of >95%.

In vitro suppression assay

Thy1.1+ Treg cells from either WT or OT-II donors were co-cultivated in round-bottom 96-well plates with MACS-enriched CFSE-labeled Thy1.2+ T cells at the indicated ratios under stimulatory conditions applying RPMI 1640 supplemented with 10% FCS, 2 mM glutamine and antibiotics, 100 IU/mL rh-IL2, and 1.5 μL T-cell expander beads (anti-CD3/anti-CD28, Dynal). After 4 days of co-culture, proliferation was assessed by flow cytometry determining CFSE dilution on live Thy1.2+ T cells. Dead cells were identified by counterstaining with 4′,6-diamidino-2-phenylindol.

Statistical analysis

Averages and SD or SEM were calculated with Graphpad Prism®. Group data were compared with the two-tailed unpaired t-test. Similarity between two sequenced TCR repertoires was statistically measured by the Morisita-Horn index 58. This index ranges between 0 (complete dissimilar) and 1 (identical) and is comparatively resistant to sample size. Proportional Euler diagrams were generated using the program VennMaster, which is available at http://www.informatik.uni-ulm.de/ni/staff/HKestler/vennm/doc.html.


The authors thank Andreas Krueger and Oliver Pabst for discussions and carefully reading the MS. The authors thank Mathias Herberg, Georgios-Leandros Moschovakis, Sebastian Seth, Henrike Fleige, Sabrina Woltemate, Kerstin Püllmann, Monika Bischoff, Anna Smoczek, Frano Malinarich, Manuel Winter, and Vijaykumar Chennupati for help. They also acknowledge the assistance of the Cell Sorting Core Facility of the Hannover Medical School. The authors are grateful to Véronique Guidicelli and the IMGT® team for their helpful collaboration and the analysis of nucleotide sequences on the IMGT/HighV-QUEST web portal, prior to its public availability. This work was supported by grants from the Deutsche Forschungsgemeinschaft SFB621-A14 (IP), and Deutsch-Französische Hochschule/Université franco-allemande DFH-UFA G2RFA 104-07-II. L. F. is supported by Hannover Biomedical Research School.

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