Prof. Herman Waldmann, Sir William Dunn School of Pathology, Oxford University, South Parks Road, Oxford OX1 3RE, United Kingdom
Regulatory T (Treg) cells are critically important for the maintenance of immunological tolerance. Both centrally arising natural nTreg cells and those emerging in the periphery in response to TGF-β, iTreg cells, play a role in the control of unwanted immune responses. Treg cells adopt multiple mechanisms to inhibit effector T cells, yet it is unclear whether these mechanisms are shared by nTreg cells and iTreg cells alike. Here, we show that iTreg cells, like nTreg cells, are able to out-compete naïve T cells in clustering around dendritic cells (DCs). However, using both a tamoxifen-responsive inducible Foxp3 retroviral construct and TGF-β-induced iTreg cells from hCD2-Foxp3 knock in reporter mice, we show that it is prior antigen-induced activation rather than Foxp3 expression per se that determines the ability of iTreg cells to competitively cluster around DCs. We found no difference in the capacity of iTreg cells to displace naïve T cells around DCs to that of Tr1, Th1, Th2, or Th9 cells. An important difference was, however, that clustering of iTreg cells around DCs, just as for naïve T cells, did not effectively activate DCs.
Regulatory T (Treg) cells are a specialized subset of CD4+ T cells that can mediate dominant tolerance to self and nonself by suppressing the activation and proliferation of lymphocytes []. Natural Treg (nTreg) cells generated in the thymus are characterized by the expression of Foxp3, a member of the forkhead/winged-helix family of transcription factors. T cells with “regulatory” properties can also be induced in the periphery under a variety of conditions. Such T cells include TCRαβ+CD4−8−NK1.1− double negative Treg cells, CD4+CD25+ Treg cells in animals that had been rendered tolerant by coreceptor blockade [], Foxp3+-induced Treg (iTreg) cells, Tr1 cells [], and a newly discovered subset of IL-35-secreting T (iTr35) cells []. Graca et al. [] also showed a new subset of iNKT Foxp3+ cells induced in vivo or in vitro with TGF-β. Chen et al., Fantini et al., Fu et al. and we [[6-10]] showed that Treg cells can be induced in vitro by TCR activation in the presence of TGF-β. A proportion of these TGF-β-conditioned T cells express Foxp3.
A recent report, based on the notion that Treg cells dominate sites of antigen presentation, compared the in vitro aggregation of nTreg cells and effector/responder T cells around dendritic cells (DCs) by dye-labeling the T-cell populations []. nTreg cells were found to out-compete responder T cells in forming aggregates around DCs, thus preventing responder T cells from accessing the DCs. This process was antigen dependent. Functionally, nTreg cells incubated with responder T cells and DCs led to lower expression of surface activation markers CD80 and CD86 than in DCs incubated with responder cells alone [].
If competitive aggregation was an important part of nTreg-cell function, we hypothesized that peripherally induced Treg (iTreg) cells might perform similarly, and consequently impact antigen presentation by DCs. To establish the validity of the aggregation assay as a measure of Treg cells’ function, we investigated whether competitive aggregation is a sole property of regulatory cells or is shared by other functional T-cell types. Using confocal microscopy and quantitative software tools, we examined the capacity for in vitro aggregation of TGF-β-conditioned T cells (as representative of iTreg cells), comparing them with naïve and activated cells and also polarized T-cell clones. In addition, to determine whether just like nTreg cells, iTreg cells would not activate DCs [] we compared the effect of TGF-β-conditioned T cells on the upregulation of the costimulatory receptors CD86 and CD40 on DCs with that of proinflammatory polarized cell clones. Indeed, they proved inefficient at such activation.
nTreg cells out-compete naïve CD4+T cells in the formation of DC aggregates
It has previously been demonstrated in vitro that CD4+CD25+ nTreg cells can out-compete naïve T cells in aggregating around splenic DCs in an assay using the OVA-specific DO11.10 TCR transgenic system (H-2d) []. This competitive ability of nTreg cells was considered a potential mechanism of suppression of naïve T-cell activation. Here, we asked whether TGF-β-induced Treg cells could function in a similar manner and, in addition, whether these Treg cells could out-compete polarized “proinflammatory” T-cell clones as well as naïve T cells. First, using the HY antigen-specific A1 system (H-2k), we asked whether nTreg cells from female A1 TCR transgenic mice [] would also behave competitively in aggregation around DCs. nTreg cells and naïve T cells were sorted from (A1.RAG−/− × CBA) F1 mice based on CD4 and CD25 expression. The T-cell populations were pretreated with either a green (PKH-67) or red (PKH-26) membrane dye prior to coincubation and incubated together at 1:1 ratio with bone marrow-derived DCs (BMDCs) and HYEk peptide (100 nM) for 48 h. IL-2, important for the function and maintenance of nTreg cells [], was also added to the cultures at 5 ng/mL. Naïve T cells were labeled with a green-membrane dye while nTreg cells were treated with the red dye (Fig. 1A top panel). After 24 h of coincubation, the cultures were then examined using confocal microscopy. Three random clusters were photographed in each culture condition, and a tally of the number of red and green cells in each cluster was made using ImageJ®.
When the clusters were examined, nTreg cells showed a strong preferential aggregation around BMDCs compared to the naïve effector cells such that approximately 70% of cells in the clusters were nTreg cells (Fig. 1A bottom panel and B), thus confirming the data of Onishi et al. and Sarris et al. [, ].
TGF-β-conditioned T cells and activated T cells both out-compete naïve T cells for DC clustering
Having confirmed that nTreg cells can out-compete naïve T cells, we next asked whether TGF-β-conditioned CD4+ T cells could also out-compete naïve CD4+ T cells. Splenocytes were conditioned in TGF-β for 7 days in the presence of BMDCs and their cognate peptide as described in the Materials and methods. CD4+ T cells were then enriched from the cultures using Histopaque density gradients. On FACS analysis, the TGF-β-conditioned cells were 72% positive for Foxp3 expression (Fig. 2A). Activated T cells (peptide activated but without TGF-β conditioning and 0% Foxp3 expression) were also cultured with naïve T cells to assess clustering. For the aggregation assay, the different T-cell populations were incubated for 24 h with BMDCs and peptide. As seen in Fig. 2B top panel, TGF-β-conditioned cells (TTGF-β-con) were found to out-compete naïve T cells (Tn) over a range of different Tn:TTGF-β-con cell ratios. However, activated T cells (Ta) also out-competed naïve T cells at all cell ratios (Fig. 2B lower panel). There was no statistically significant difference in the ability of TGF-β-conditioned T cells or peptide-activated T cells to out-compete naïve T cells in DC clustering (Fig. 2C).
Competitive aggregation of TGF-β-conditioned T cells and cTreg cells is not dependent on Foxp3 expression
Because we found that nTreg cells and TGF-β-conditioned T cells can out-compete naïve T cells around DCs, and activated T cells also display this ability, we next asked what role, if any, expression of Foxp3 had in endowing these cells with competitive clustering ability. To this end, we used retrovirally transduced conditional Treg (cTreg) cells (cFoxp3-transduced T cells treated with 4-hydroxytamoxifen in vitro). cTreg cells were derived by transducing CD4+ T cells isolated from CBA or Marilyn.RAG1−/−.Foxp3hCD2 (see Materials and methods). T cells transduced with cFoxp3 retrovirus expressed abundant GFP-tagged Foxp3 protein predominantly in the cytoplasm (Fig. 3A middle panel). Upon addition of 4-hydroxytamoxifen (4HT) the fluorescence translocated to the nucleus indicating the Foxp3 protein had entered the nucleus (Fig. 3A lower panel). Expression of Foxp3 increased following transduction reaching a peak after 64 h of culture (Fig. 3B). Transduced cells cultured in the presence of 4HT seemed to express approximately threefold more Foxp3 than those cultured without 4HT. (This may be a consequence of nuclear sequestering of Foxp3 reducing its exposure to cytoplasmic proteases.) We then tested the ability of 4HT to switch the cFoxp3-transduced cells to a regulatory phenotype. cFoxp3 transduced Marilyn.RAG−/− CD4+ T cells or empty vector control cells were mitomycin C treated to inhibit their proliferation prior to coincubation with untransduced Marilyn.RAG−/− CD4+ T cells and female mitomycin-treated C57BL/6 BMDCs with cognate peptide. cFoxp3-transduced T cells cultured with 4HT significantly inhibited the proliferation of naïve Marilyn.RAG−/− CD4+ T cells in cocultures (Fig. 3C). In the absence of either 4HT or translocated cFoxp3, effector cell proliferation was not significantly inhibited by these cells.
We examined the competitive clustering ability of these cTreg cells by coincubation with naïve CD4+ T cells together with B6 BMDCs and HYAb peptide. 4HT was also added in vitro to ensure functional “activation” of cFoxp3 (50 nM). As controls, empty vector-transduced CD4+ cells with 4HT, and also cFoxp3-transduced CD4+ cells without 4HT, were used. The various cell populations were incubated with naïve CD4+ cells at 2:1, 1:1, and 1:2 ratios (Fig. 4A). While cTreg cells could out-compete naïve CD4+ cells, both empty vector-transduced cells and cFoxp3-transduced CD4+ cells without 4HT treatment could also do the same. There was no evidence that the aggregation potency of cTreg cells was vastly superior to that of control-transduced cells at various cell ratios (Fig. 4A and B). Hence, functional Foxp3 may not be essential for the competitive aggregation of cTreg cells.
To further investigate whether Foxp3 was important for competitive aggregation of iTreg cells, Foxp3+, and Foxp3- populations of TGF-β-conditioned T cells from Marilyn.RAG1−/−.Foxp3hCD2 mice [] were purified based on their expression of hCD2 and CD4 (Fig. 5A and B). hCD2+ (Foxp3+) and hCD2− (Foxp3−) cells were then coincubated with naïve T cells. At both 10:1 and 1:1 ratios, there was no statistically significant difference in the aggregation potency of hCD2+ and hCD2− cells compared to naïve T cells. This suggests that Foxp3+ TGF-β-conditioned T cells are no better at out-competing naïve T cells than Foxp3- TGF-β-conditioned T cells (Fig. 5C).
TGF-β-conditioned T cells and cTreg cells do not out-compete other activated T cells in aggregation around DCs
Having established that both regulatory cells and activated Foxp3− CD4+ T cells could out-compete naïve T cells, we hypothesized that previous antigen activation could be more important for the ability of T cells to competitively cluster around DCs. Onishi et al.  attributed the competitive aggregation of nTreg cells to their LFA-1 expression, as LFA-1-deficient nTreg cells fail to out-compete naïve T cells in aggregation. As LFA-1 is upregulated on T-cell activation [], this would be consistent with the observation that antigen-activated CD4+ T cells show superior competitive aggregation compared to naïve CD4+ T cells. Moreover, nTreg cells may also be more efficient than naïve CD4+ T cells, as they may have constant exposure to the repertoire of self-antigens [].
We tested whether there was a Treg cell-intrinsic superiority in competitive aggregation by comparing the competitive clustering ability of activated CD4+ T cells with CD4+ T cells transduced with cFoxp3 empty vector or cFoxp3 in the presence and absence of 4HT. Sorted cTreg cells (cFoxp3-transduced CD4+ T-cell, 4HT treated) were coincubated with untransduced but anti-CD3 activated CD4+ T cells for 24 h. Empty vector-transduced cells and non-4HT-treated cFoxp3-transduced cells were also coincubated with untransduced-activated cells for comparison. cTreg cells did not out-compete activated CD4+ T cells more so than control-transduced cells at the cell ratios that were examined — 2:1, 1:1, and 1:2 (Fig. 6).
Polarized T-cell clones also out-compete naïve T cells
While nTreg cells, iTreg cells, and cTreg cells all express Foxp3, they also share the common characteristic of being antigen experienced. nTreg cells are thought to constantly survey antigen in the periphery and cTreg cells and iTreg cells are induced during activation with cognate antigen. Prior antigen experience may explain the result that empty vector-transduced CD4+ T cells and Foxp3- TGF-β-conditioned cells can out-compete naïve T cells in aggregation equally well compared to cTreg cells and iTreg cells. To confirm if antigen experience is sufficient for competitive aggregation of CD4+ T cells, polarized cell clones (Th1, Th2, Tr1, Th9) were incubated with naïve CD4+ T cells (see Materials and methods on the generation of polarized cell clones). Both naïve CD4+ T cells and polarized cell clones were established from A1.RAG−/− mice. In contrast to naïve T cells, polarized T-cell clones had been activated for 14 days with the male spleen cells prior to the aggregation assay.
As seen in Fig. 7A, all polarized cell clones were found to out-compete naïve CD4+ T cells (Tn). This was statistically significant at all cell ratios for Th9 and Th2 cell clones, and at 2:1 and 1:1 cell ratios for Th1 and Tr1 cell clones. At the 1:1 ratio, Th9, Tr1, and Th1 appeared to be the most competitive at aggregating, with around 70% cells in the T cell‒DC clusters being red (polarized cell clone). Hence, antigen-experienced proinflammatory and regulatory cell types all appear to out-compete naïve antigen-inexperienced cells.
To examine the aggregation competitiveness of TGF-β-conditioned T cells in comparison with other antigen-experienced cell types, the former were coincubated with activated but non-polarized T cells and polarized cell clones at varying ratios. There was no statistically significant difference in the competitive aggregation of TGF-β-conditioned T cells with activated T cells compared with dye-control TGF-β-conditioned T-cell incubations (Fig. 7B). Furthermore, TGF-β-conditioned cells competed poorly in aggregation against the polarized cell clones Th1 and Tr1 (Fig. 7B). Hence, while cTreg cells and TGF-β-conditioned T cells compete favorably against naïve T cells, they do not appear to have a competitive advantage in relation to other antigen-experienced cell populations.
DCs are important for both the induction of primary immune responses and the promotion of immunological tolerance [[18, 19]]. Upon encounter with CD4+ T cells, DCs are activated via CD40-CD40L interaction, leading to the upregulation of CD80 and CD86 costimulatory molecules and release of cytokines. CD86 is one of the key molecules for the amplification of the T-cell response []. CD40 is also upregulated upon BMDC maturation and activation. To examine whether iTreg cells can modulate the expression of activation markers such as CD40 and CD86 on DCs, BMDCs were examined after 24 or 48 h culture with TCR-transgenic TGF-β-conditioned T cells in the presence of cognate peptide (100 nM). For comparison, naïve T cells and polarized T-cell lines were also coincubated with BMDCs (Fig. 7C).
BMDCs coincubated with Th1 and Th9 polarized CD4+ cells for 24 h were found to express high levels of CD86 (MFI: 584 and 326, respectively), with much lower levels found on those coincubated with Th2, Tr1, naive, and TGF-β-conditioned T cells (MFI: 71, 63, 59, and 87) at 24-h postincubation (Fig. 7C). The same is also true at 48 h. Similarly, at 24 h, CD40 was significantly upregulated by BMDCs coincubated with Th1 and Th9 polarized cell clones (MFI: 609 and 561), but lower levels were found on those coincubated with TGF-β-conditioned cells, Th2, Tr1, and naive cells (MFI: 15, 188, 268, 20). A similar picture also emerged for the expression of class II MHC on BMDCs, with Th1 and Th9 polarized cells producing the highest levels of expression and lower levels produced by TGF-β-conditioned cells, Tr1 cells, and naïve T cells.
Hence, antigen-experienced cells such as TGF-β-conditioned T cells and polarized CD4+ clones, while effective in competitive aggregation compared with naïve T cells, are not equal in their ability to induce upregulation of activation markers CD86, CD40, and class II MHC on BMDCs. TGF-β-conditioned T cells appear less effective at activating BMDCs than polarized inflammatory cell types such as Th1 and Th9 cells. Taken together, TGF-β-conditioned T cells can not only physically displace naïve T cells from contact with DCs but also only modestly activate the DCs themselves. Like “civil servants,” these physically competitive but functionally inert cells may contribute to the dampening of the immune response in an inflammatory context by limiting the capacity of naïve CD4+ T cells to bind to DCs [].
The mechanisms which Treg cells adopt to inhibit activation of naïve effector cells is an area of intense scrutiny. Multiple ways which Treg cells may inhibit T-cell activation have been proposed. These include secretion of inhibitory cytokines such as IL-10, TGF-β, and IL-35 [[22, 23]], competition for growth factors [], production of antiinflammatory mediators such as adenosine [], and expression of membrane bound inhibitory receptors such as Fas, Granzyme B, CTLA4, and TGF-β. In addition, Treg cells can condition DCs to upregulate catabolic enzymes involved in breaking down essential amino acids thus producing a tolerogenic microenvironment []. Treg cells have also been shown to competitively cluster around antigen presenting cells [[11, 14]].
We have previously hypothesized that the latter mechanism might occur and proposed the “civil service” hypothesis for Treg cells’ immune regulation []. This hypothesis compares Treg cells to “civil servants”: competitive but functionally inert cells that diminish the capacity of naïve CD4+ T cells to bind to DCs. Recent studies have reported the ability of natural Treg cells to out-compete naïve effector T cells for clustering around DCs [[11, 14]]. In this report, we showed that both natural and TGF-β induced iTreg cells can out-compete naïve T cells from clustering around DCs. The capacity of iTreg cells to out-compete naïve T cells was dependent on prior antigen activation and independent of Foxp3 expression. Indeed polarized Th1, Th2, Tr1, and Th9 cells were also capable of out-clustering naïve T cells on DCs but they differed from Treg cells in that they led to upregulation of activation markers on DCs whereas Treg cells failed to activate DCs.
We took advantage of two well-characterized experimental systems to explore the function of Foxp3 in DC clustering. The tamoxifen responsive conditional Foxp3 system was originally used to show that experimental collagen-induced arthritis could be ameliorated by transfer of cFoxp3-transduced T cells followed by tamoxifen administration []. In the absence of tamoxifen the cells homed to secondary lymph nodes and participated in the autoimmune reaction. Administration of tamoxifen rapidly and efficiently converted the cells to a regulatory phenotype both in vitro and in vivo. The human CD2-Foxp3 knockin mouse has a membrane reporter (hCD2/hCD52 fusion protein) gene construct “knocked in” to the Foxp3 locus []. All cells in these animals that express Foxp3 also express this cell surface tag. This system allowed us to purify 99% pure Foxp3-expressing cells as well as to examine the relative importance of “TGF-β conditioning” versus Foxp3 expression.
Using the combination of these two systems we were able to conclude that neither Foxp3 expression nor TGF-β conditioning alone were necessary for acquisition of competitive clustering ability. Two questions arise from these observations: first, what molecular mechanisms are responsible for the enhanced clustering ability and second, why does iTreg cells clustering not result in enhanced DC activation? As all the T-cell populations used in this study had the same TCR specificity, differential MHC:peptide binding cannot be invoked as an explanation for the increased prevalence of Treg cells around DCs. Increased Treg cells clustering around DCs compared to naïve T cells may either be a consequence of increased binding via adhesion molecules, longer contact time at the immune synapse or a decrease in relative cell movement of Treg cells compared to naïve T cells. The three explanations are not mutually exclusive.
Previous studies addressing the interaction of natural Treg cells with DCs have highlighted two molecules, LFA1 and neuropilin-1 as being necessary for enhancing the interaction time of the nTreg cells with the DCs. Sarris et al.  showed that neuropilin-1, expressed by nTreg cells but not by naïve T cells was necessary for prolonged interaction and increased the nTreg cells' sensitivity to limiting concentrations of cognate antigen presented by immature DCs. Neuropilin-1 is a receptor for class III semaphorins and is a coreceptor for vascular endothelial growth factor and latent and active TGF-β1 []. LFA-1 was shown to be necessary for nTreg cells/DC clustering in addition to down modulation of CD80/CD86 on the DC cell surface []. The two molecules differ in distribution with neuropilin-1 being expressed by CD4+CD25+ nTreg cells but downregulated on activation of naïve T cells. LFA1 on the other hand is expressed by most T cells, playing an important role in the formation of the immune synapse []. Thus Neuropilin-1 may play a more specific role in increasing contact time of nTreg cells with DCs whereas it is likely that multiple T-cell types use LFA1.
In relation to the issue of DC activation, we asked whether the different effects of the distinct T-cell lines were related to expression of key DC-interacting molecules. We measured expression of cell surface and intracellular LFA1, CTLA4, and CD40L on polarized T-cell clones; Th1, Th2, Tr1, Th9, and iTreg cells (Supporting Information Fig. 2) and found no correlation between the level of expression of these receptors and the ability of these T cells to induce activation of DCs, indicating that T-cell to DC costimulation or coinhibition were not the direct causes of differential activation.
We recently demonstrated that iTreg cells under the influence of TGF-β upregulate CD73 []. CD73 is an ectonucleotidase which in conjunction with CD39 on the cell surface converts ATP, a relatively inflammatory molecule to adenosine that has potent anti inflammatory effects. Thus although it is possible that adenosine or other soluble mediators might, in principle, be able to inhibit DC activation in the presence of AMP substrate, such substrate was not added in the studies described.
Finally, our results may go some way to explaining the well-documented phenomenon of antigen competition. Antigen competition describes the inhibited immune response to a protein antigen if it is coadministered at the same time or within a short-time interval and at the same anatomical site as another antigen []. This effect is particularly important for the design of vaccines and is seen following administration of some combined vaccines such as diptheria, pertussis, and tetanus combinations among others. Antigen competition may be influenced by multiple factors including antigen processing events, affinity of the TCR for MHC/peptide, and epitope concentration. An immunodominant peptide may induce more rapid activation of its cognate CD4+ T-cell thus leading to competitive aggregation of these cells on DCs at the expense of other CD4+ T cells with different specificities. Treg cells may be acting as “gate keepers” using this mechanism to limit de novo responses while allowing ongoing responses to continue.
Materials and methods
CBA/Ca (CBA), A1.RAG−/− [], Marilyn.RAG1−/−, (A1.RAG−/− × CBA)F1, and Marilyn.RAG1−/−.Foxp3hCD2 [] mice have been described previously. All mice were bred and maintained under specific pathogen-free conditions at the Sir William Dunn School of Pathology (University of Oxford, Oxford, UK). All experimental procedures received local ethics committee approval and were conducted in accordance with the Home Office Animals (Scientific Procedures) Act of 1986.
Anti-CD3e (145-2C11), anti-CD4 (H129.19 or RM4-5, FITC, PerCP or allophycocyanin conjugated), anti-CD25 (PC61, PE conjugated), anti-CD86 (GL1, allophycocyanin conjugated), and anti-MHC class II I-A/I-E (2G9) antibodies were purchased from BD Pharmingen. Anti-FoxP3 (FJK-16S, FITC, PE, or allophycocyanin conjugated) was purchased from eBiosciences. Anti-CD40 (1C10, PE conjugated) was purchased from Southern Biotech. Anti-human CD2 domain-2 (YTH655) was produced in house. DRAQ5 was purchased from Biostatus.
Cytokines for cell culture and synthetic peptides
IL-2, IL-7, and lyophilized recombinant human TGF-β-1 were purchased from Peprotech (UK) IL-2 and IL-7 were reconstituted in sterile water TGF-β1 was reconstituted in 4 mM HCl with 1 mg/mL BSA. Supernatants enriched for GM-CSF were harvested from a cultured X63 cell line transfected with cDNA constructs encoding murine GM-CSF (D. Gray, Edinburgh). GM-CSF-containing supernatant was used at 2% v/v, which was equivalent to 5 ng/mL of GM-CSF. Synthetic peptides, HYEk (REEALHQFRSGRKPI) [] and HYAb (NAGFNSNRANSSRSS) [], were reconstituted in PBS from lyophilized powder and both used at a maximum working concentration of 100 nM.
Bone marrow DC preparation
BMDCs were prepared using a protocol adapted from Inaba et al. []. Briefly, femurs and tibias were flushed with R10 medium (RPMI-1640 medium (Lonza) containing 10% v/v FCS (Gibco), 50 μg/mL of Penicilin-Streptomycin, (1% v/v) Sodium Pyruvate solution, 0.1 M HEPES buffer solution (all from PAA laboratories GmbH) and 50 μM 2-mercaptoethanol) to extrude the bone marrow. The cells were then passed through a 70-μm cell strainer, pelleted by centrifugation, and resuspended in 45 mL R10 medium supplemented with 2% GM-CSF-containing supernatant. The cells were cultured for 7 days, with removal of nonadherent cells and the addition of fresh medium on day 3. On day 6, BMDCs were replated and resuspended in fresh R10 medium containing GM-CSF. On day 7 of culture, BMDCs were harvested. Depending on experimental needs, BMDCs were treated before use with 25 μg/mL Mitomycin C (MMC; Sigma-Aldrich, UK) for 30 min at 37°C to inhibit cell division.
Splenic CBA or C57BL/6 CD4+ cells were isolated by negative selection using the AutoMACS separator as per the manufacturer's instructions (Miltenyi Biotec, UK).
In vitro generation of induced Treg cells
Splenocytes from A1.RAG−/− or Marilyn.RAG−/− mice were red blood cell depleted with tris buffered ammonium chloride for 30 s before washing in complete RPMI medium. A total of 5 × 105 splenocytes were then resuspended per well in a 24-well plate (Corning, US) along with 1 × 105 mitomycin-treated CBA female or B6 female BMDCs. HYEk or HYAb peptide was added at 100 nM. For TGF-β-conditioned iTreg cell cultures, recombinant human TGF-β was also added at 2 ng/mL. Control cell cultures received peptide only. On day 7, lymphocytes from the cultures were harvested by Histopaque (1.083 g/mL; Sigma-Aldrich) separation. The percentage of Foxp3 expression was assessed using flow cytometry and was typically 20–40% for A1.RAG−/− TGF-β-conditioned cultures, and 50–90% for Marilyn.RAG−/−.
Preparation of polarized CD4+T-cell clones
Th1 and Th2 clones (R2.2F1 and R2.4E9, respectively) were established as described previously []. Briefly, spleen cells were isolated from female A1.RAG−/− mice that had rejected two male skin grafts. The cells (5 × 105/well) were polarized by three cycles of weekly stimulation with male 5 × 106 spleen cells in the presence of IL-2 (50 U/mL) for Th1 and IL-4 (200 U/mL) for Th2. Viable cells were then cloned at limiting dilution in 96-well plates with APCs (2 × 105 male CBA MMC-treated spleen cells), and the appropriate cytokine. These were then expanded into 24-well plates and maintained by fortnightly stimulation of 2 × 105 clone cells with 5 × 106 male MMC-treated spleen cells and the appropriate cytokine.
The Tr1 clone (Tr1D1) was generated from splenocytes of female A1.RAG−/− mice, stimulated in recombinant IL-10 (R&D Systems, 10 ng/mL) for 3 weeks before cloning [[3, 34]]. The clone was maintained by culturing 2 × 105 cloned Tr1D1 cells with 5 × 106 male splenocytes in 24-well plates in medium containing IL-2 and IL-4, each at 20 U/mL. The Th9 clone (3E7) was generated by culturing CD4+ cells from female A1.RAG−/− spleens with female CBA spleen cells in 100 nM HYEk peptide, 1 ng/mL rTGF-β, and 10 ng of mIL-4 for 1 week. Viable cells were recovered and restimulated as previously. After three cycles of weekly activation, cells were cloned by limiting dilution in 96-well plates with 2 × 105 female CBA/Ca splenocytes, 100 nM of HYEk peptide, 20 U/mL IL-2, 10 ng/mL IL-4, and 2 ng/mL TGF-β. Cells that secreted IL-9 were expanded.
Fluorescence-activated cell sorting (FACS)
Cells were washed twice in PBS 0.5% BSA and then incubated for 10 min at room temperature with 10 μg/mL anti-FcR block (2.4G2). Cells were then incubated with primary antibody for 30 min in the dark at 4°C. Following washing and fixation with 2% paraformaldehyde, analysis was performed on a FACSCalibur (Becton Dickinson, NJ) with dual laser excitation (488 and 633 nm). Data acquisition was performed with inter-channel compensation using CellQuest version 3.1 software (Becton Dickinson, NJ) and data analysis with FlowJo 7.2.4 software (Treestar Inc., Oregon).
For staining of Foxp3, the Foxp3-staining kit (eBioscience, CA) was used according to the manufacturer's directions. Briefly, surface staining was first performed followed by washing in FACS wash buffer. Cells were then resuspended in fixation and permeabilization buffer for 30 min at 4°C. Following this, cells were washed in permeabilization buffer and incubated with PE or allophycocyanin-conjugated anti-Foxp3 for 30 min at 4°C. Then, cells were washed further in permeabilization buffer and finally resuspended in PBS 0.5% BSA.
Mo-Flo FACS sorting of cells
Flow cytometry sorting was performed by Nigel Rust at the Sorting Facility at the Sir William Dunn School of Pathology (University of Oxford) using a MoFlo sorter (Dako Cytomation, Glostrup). This instrument has three lasers emitting at the wavelengths of 488, 633, and 405 nm.
Isolation of nTreg cells
nTreg cells were isolated using flow cytometry from A1.RAG−/− × CBA F1 mice based on CD4 and CD25 expression. CD4+ splenocytes isolated as described above were stained with PE-conjugated anti-hCD2 and MoFlow FACS sorting was performed to obtain 98% pure hCD2+Foxp3+ nTreg cells.
Production of conditional Foxp3 retroviral supernatant
Retroviral vectors except pCL-ECO were as described []. The conditional Foxp3 construct (cFoxp3) was made by fusing Foxp3 cDNA with ERT2, a modified oestrogen-binding domain, at its C-terminal. A green-fluorescent protein (GFP) gene was cloned in-frame with Foxp3 after the first five codons from the 5′-end in order to produce an ERT2-F(gfp)oxp3 fusion protein that can be visualized in the cell. These were then cloned into a Moloney Murine Leukemia Virus (MMLV) backbone, a circular plasmid that also contained a GPI-linked and IRES-driven rat CD8a (rCD8) gene. The empty control vector contained only the GFP and rCD8 segments, but no Foxp3. The packaging vector, pCL-ECO, was purchased from Addgene (Cambridge, MA). The day before transfection, a confluent plate of HEK 293eT cells were split at one in three and plated in 6-well plates, at approximately 40% confluency with 2 mL per well. Approximately 16 h later transfection mix was prepared. In preparing the transfection mix, water was first added, followed by 12 μL CaCl2, then the plasmid DNA to a total volume of 100 μL. A cotransfection was performed with 2 μg of pCL-ECO packaging plasmid (Imgenex, San Diego, CA) and 2 μg of the retroviral vector (cFoxp3). While vortexing the transfection mix on a low setting, 100 μL 2 × HBS was added drop-wise to achieve a total volume of 200 μL. After waiting 30 s, the transfection mix (200 μL) was added drop-wise to a single well in the 6-well plates. Six hours following transfection, the medium in the well was replaced with fresh medium. A further 24 h later, the viral supernatant was harvested and filtered through a 0.45-μm filter. Protamine sulphate (600 μg/mL) was added to the viral supernatant.
Retroviral transduction of CD4+T cells
AutoMACS sorted CD4+ cells were cultured at 2 × 105 per well in a 96-well plate containing plate-bound anti-CD3 (KT3, 4 μg/mL). The cells were activated for 24 h and then resuspended in a 1:2 mixture of viral supernatant and complete medium (IMDM + 10% v/v FCS + 10 μM β-mercaptoethanol) supplemented with 5 ng/mL of recombinant mIL-2 and 5 ng/mL of recombinant mIL-7. This was followed by centrifugation at 500 g for 2 h at 32°C. Cells were analyzed 40 h later.
DC and T cell coclustering assay
The clustering assay was based on that described by Onishi et al. []. Naïve T cells (Tn), iTreg cells, nTreg cells, and T-cell clones were labeled with membrane linker dyes PKH-67 (green dye) or PKH-26 (red dye) (Sigma-Aldrich, UK) according to the manufacturer's instructions. Labeling was performed in 15 mL polypropylene tubes. Briefly, cells were washed and resuspended in serum-free medium prior to resuspension in Diluent C, supplied by the manufacturer (Sigma-Aldrich). Immediately, an equal volume of membrane linker dye diluted in Diluent C was added to obtain a final cell concentration of 5 × 106 cells per mL of total reaction volume and dye concentration of 1 μM. The cells were incubated for 2–3 min. An equal volume of 1% PBS/BSA was added to stop the reaction. This was followed by washing the cells three times in complete medium. Cells were counted and competing T-cell populations were cocultured along with unlabeled BMDCs in 96-well round-bottomed plates. A total number of 60,000 T cells consisting of various ratios of competing T-cell populations of 10:1, 5:1 2:1, 1:1, 1:2, 1:5, or 1:10 were cultured with 6000 BMDCs in a total volume of 200 μL in each well. Cognate peptide was also added at a concentration of 10 nM. Twenty-four hours later, the contents of each well were gently transferred into a well of an 8-well chamber slide (Lab-Tek, NY) and cell clusters were allowed to settle to the bottom of the well before examining under a confocal microscope.
Confocal microscopy was performed on a Zeiss LSM 510 META laser scanning microscope (Carl Zeiss MicroImaging GmbH, Göttingen) using a ×20 objective. The chamber slide was housed in a heated chamber adjusted to 4% CO2. Quantitative analysis of cell clusters were carried out using ImageJ software (Rasband, W.S., U. S. National Institutes of Health, Bethesda, Maryland, USA) and LSM Image Browser software version 4.2 (Carl Zeiss MicroImaging GmbH, Göttingen). The presence of a DC within T cell:DC clusters was confirmed by focusing throughout the plane of the cluster. DCs were identified in these clusters by size, morphology, and lack of PKH-26 or PKH- 67 dye staining. Additional control cultures lacking peptide, where clustering did not occur, confirmed the requirement for antigen for cluster formation.
Proliferation and suppression assays
Effector cells were Marilyn.RAG−/− CD4+ cells enriched by negative selection using AutoMACS. “Suppressor” cells were Marilyn.RAG−/− CD4+ cells that had been transduced with retroviral vectors for 40 h and then sorted on CD4 and GFP expression. BMDCs were obtained from the bone marrow of female B6 mice. “Suppressor” cells and DCs were incubated with mitomycin C to inhibit their proliferation (25 μg/mL) for 30 min at 37°C and washed three times before being added to the assay. Each well of a 96-well plate contained 104 DCs, 2 × 104 “suppressors” and 2.5 × 104 effectors. HYAb peptide was added as a dilution series (100 nm, 10 nm, 1 nm, 0). Cultures were incubated for 72 h. To measure proliferation, 3H-thymidine (GE Healthcare, UK) was added (0.5 μCi/well) for the last 18 h. The count of the incorporated isotope was then measured by a scintillation counter.
Statistical analysis was performed using Graphpad Prism software version 4.0. For cell clustering studies involving multiple test groups, ANOVA analysis was used followed by Bonferroni or Dunnett's post-test. Student's t-test with two-tailed analysis was used for binary data sets. p-values less than 0.05 were considered significant.
The authors thank the Pathology Support Building staff for their excellent contribution. This work was supported by UK Medical Research Council grants G7904009 and G1000215 (D.H. and H.W.). F. S. R. received a PhD scholarship from Fundaçãopara a Ciência e Tecnologia, Portugal.
Conflict of interest
The authors declare no financial or commercial conflict of interest.