Naturally occurring CD4+CD25+ regulatory T cells (Treg) are crucial in immunoregulation and have great therapeutic potential for immunotherapy in the prevention of transplant rejection, allergy, and autoimmune diseases. The efficacy of Treg-based immunotherapy critically depends on the Ag specificity of the regulatory T cells. Moreover, the use of Ag-specific Treg as opposed to polyclonal expanded Treg will reduce the total number of Treg necessary for therapy. Hence, it is crucial to develop ex vivo selection procedures that allow selection and expansion of highly potent, Ag-specific Treg. In this study we describe an ex vivo CFSE cell sorter-based isolation method for human alloantigen-specific Treg. To this end, freshly isolated CD4+CD25+ Treg were labeled with CFSE and stimulated with (target) alloantigen and IL-2 plus IL-15 in short-term cultures. The alloantigen-reactive dividing Treg were characterized by low CFSE content and could be subdivided by virtue of CD27 expression. CD27/CFSE cell sorter-based selection of CD27+ and CD27− cells resulted in two highly suppressive Ag-specific Treg subsets. Each subset suppressed naive and Ag-experienced memory T cells, and importantly, CD27+ Treg also suppressed ongoing T cell responses. Summarizing, the described procedure enables induction, expansion, and especially selection of highly suppressive, Ag-specific Treg subsets, which are crucial in Ag-specific, Treg-based immunotherapy.
Naturally occurring CD4+CD25+ regulatory T cells (Treg)3 are important in the regulation of T cell homeostasis and pathogenesis (1, 2). Recently, it was demonstrated that CD4+CD25+ Treg are critical regulators of the human immune response in autoimmunity (3, 4, 5), allergy (6), and infectious diseases (7). Obviously, the use of CD4+CD25+ as a tool in Treg-based immunotherapy is currently intensively studied. In fact, in a variety of animal models, CD4+CD25+ Treg, either freshly isolated or ex vivo expanded, were successfully used to prevent transplant rejection and autoimmunity (8, 9, 10, 11, 12, 13, 14).
Human naturally occurring CD4+CD25+ Treg comprise 5–10% of all CD4+ peripheral T cells (15, 16, 17), and like their murine counterpart, they are anergic and suppressive when stimulated via the TCR (15, 16, 17, 18, 19). The human CD4+CD25+ Treg population displays a polyclonal TCR-Vβ repertoire (20, 21), indicating that a broad variety of Ags might be recognized. No unique phenotypical selection marker is available for human naturally occurring CD4+CD25+ Treg; the cells are best characterized by dose-dependent suppressor function in in vitro suppression assays. The effector mechanism(s) by which Treg exert their suppressive function is largely unknown. In most studies, cell contact between Treg and effector T cells appeared required for the suppressive effect (22); in some cases, a role for IL-10, TGF-β (23, 24, 25, 26), and/or CTLA-4 (25, 27, 28) was demonstrated.
Of great interest are recent studies demonstrating that the success of Treg-based immunotherapy critically depends on the Ag specificity of the Treg. In a transgenic mouse model of multiple sclerosis, myelin basic protein-specific Treg, compared with Treg-specific for other Ags, were superior in disease protection (29). Similarly, in mouse autoimmune diabetes models, small numbers of ex vivo expanded Ag-specific Treg reversed diabetes after disease onset (13, 14). This was due to the fact that Ag-specifically expanded CD4+CD25+ Treg were far more effective in preventing autoimmune diabetes than polyclonally (using anti-CD3 and anti-CD28 mAb) expanded Treg (13). Also, ex vivo expansion of freshly isolated mouse CD4+CD25+ Treg with host-type stimulator cells resulted in better inhibition of graft-vs-host disease compared with Treg-activated with third-party stimulator cells (9, 11). Together, these findings indicate that ex vivo expanded, Ag-specific Treg, rather than polyclonally expanded populations, are superior in suppressing graft rejection and autoimmunity. Therefore, the availability of a protocol to induce and select potent immunosuppressive human Treg with defined Ag specificity seems crucial for the success of clinical Treg therapy. Clinical implementation of Treg therapy, next to the ex vivo selection of human Ag-specific Treg, will require the availability of sufficient Treg numbers. At present the ex vivo expansion of CD4+CD25+ Treg is well established (15, 16, 21, 30, 31). Notably, the use of highly suppressive Ag-specific Treg will reduce the total number of Treg needed for Treg-based immunotherapy; quality reduces the need for quantity.
In this study we developed a robust, ex vivo, CFSE dye-based technique to induce and purify highly potent human alloantigen-specific Treg subsets from freshly isolated polyclonal CD4+CD25+ Treg. Fresh CD4+CD25+ Treg were CFSE-labeled and stimulated with target alloantigen and IL-2 plus IL-15. Dividing alloantigen-reactive Treg subsets were characterized on the basis of low CFSE content and the presence or the absence of CD27. This CD27/CFSE cell sorter-based isolation resulted in a CD27+CD25+CD70−CD62L+CTLA4high (CD27+Treg) and a CD27−CD25−CD70+CD62L−CTLA4int (CD27−Treg) CD4+ Treg subset. After purification, an increase in the suppressive potential of both Treg populations was observed. Both alloantigen-specific Treg subsets showed high effector-suppressor potential; 50% inhibitory ratios of Treg vs responder T cells (Tresp) were 1:500 and 1:50 for CD27+Treg and CD27−Treg, respectively. Suppressor function was cell contact dependent and could not be neutralized by mAb against CTLA-4 or CD27. The Treg subsets were further distinguished by their migration receptor pattern and cytokine production profile. The Treg subsets showed suppression at multiple levels; they suppressed naive and memory T cell division (CD27+Treg and CD27−Treg), ongoing T cell division (CD27+Treg), and T cell effector function, which are all implicated in transplant rejection.
This CFSE-based method for ex vivo induction and selection of alloantigen-specific Treg subsets offers a potent tool for donor-specific, Treg-based immunotherapy.
Materials and Methods
Isolation of cells
PBMC were isolated by density gradient centrifugation (Lymphoprep; Nycomed Pharma) of buffy coats obtained from normal healthy donors. CD4+ T cells were purified from PBMC by negative selection using mAbs directed against CD8 (RPA-T8), CD14 (M5E2), CD16 (3G8), CD19 (4G7), CD33 (P67.6), and CD56 (B159; BD Biosciences) combined with sheep anti-mouse Ig-coated magnetic beads (Dynal Biotech). This resulted in a CD4+ T cell enrichment of >95% and the absence of CD8+ T cells. From purified CD4+ T cells, naturally occurring CD4+CD25+ T cells were isolated using the MACS CD25+ magnetic microbead method (Miltenyi Biotec), using half the amount of beads recommended by the manufacturer. In some experiments CD4+CD25high were isolated from purified CD4+ T cells by high purity flow cytometric cell sorting (Beckman Coulter). CD4+CD25+ regulatory T cells were immediately used after isolation, whereas all other cell types were either used fresh or after thawing of liquid nitrogen-stored cell stocks. HLA typing was conducted as described previously (32). High purity CFSE- and/or CD27-based cell isolation was performed by flow cytometric cell sorting (Beckman Coulter).
Primary mixed lymphocyte cultures (MLC)
Primary MLC were performed by culturing 5 × 104 isolated CD4+CD25− T cells with 0.5–1 × 105 fully HLA mismatched gamma-irradiated (30 Gy) stimulator PBMC (target alloantigen) in 200 μl of culture medium (RPMI 1640 with Glutamax supplemented with pyruvate (0.02 mM), 100 U/ml penicillin, 100 μg/ml streptomycin, (all from Invitrogen Life Technologies), and 10% human pooled serum (HPS)) at 37°C, 95% humidity, and 5% CO2 in 96-well, round-bottom plates (Greiner). Proliferation was analyzed by [3H]thymidine incorporation using a gas scintillation counter (Canberra Packard; Matrix 96 beta counter). To this end, 0.037MBq (1 μCi) [3H]thymidine (ICN Pharmaceuticals) was added to each well, cells were harvested after 8 h of culture, and [3H]thymidine incorporation was measured. The 3H incorporation was expressed as the mean ± SD counts per 5 min of at least triplicate measurements.
Allogeneic restimulation assays to study T cell anergy
T cell anergy was examined in restimulation assays (32). CD4+CD25+ or CD4+CD25− cells were primed with (target) alloantigen in the presence of recombinant human IL-2 (12.5 U/ml; Proleukine) and IL-15 (10 ng/ml; BioSource International). After expansion, the cells were harvested, washed, and allowed to rest for 2 days in culture medium containing 2–5% HPS. Next, the viable cells were isolated by density gradient centrifugation and restimulated (1–2.5 × 104 cells/well) with gamma-irradiated (30 Gy) stimulator PBMC (1 × 105 cell/well) derived from either the target donor stimulator PBMC pool or from completely HLA mismatched (third-party Ag) blood donors. Proliferation was examined at the indicated time points.
MLC coculture assay to study T cell effector-suppressor function
The effector-suppressor capacity of cells was studied in an MLC coculture suppression assay. CD4+CD25+ or CD4+CD25− T cells were primed with (target) alloantigen in the presence of recombinant human IL-2 and IL-15. After expansion, the cells were harvested, washed, and allowed to rest for 2 days in medium containing 2–5% HPS, and the viable cells were collected by density gradient centrifugation. The cells of interest were added to a newly set-up primary MLC (consisting of freshly thawed original responder PBMC and gamma-irradiated (30 Gy) stimulator PBMC) or to a restimulation assay (set up with allogenetically primed responder CD4+ T cells and gamma-irradiated (30 Gy) stimulator PBMC). Ag specificity was examined in cocultures that were performed with third-party stimulator PBMC that were fully class I and class II HLA mismatched with the (target) allogeneic stimulator PBMC. In some experiments, round-bottom plates coated with 5 μg/ml anti-CD3 mAb (UCHT1; BD Biosciences) were used to activate T cells.
Blocking studies were performed in the presence of 20 μg/ml mAbs directed against CTLA-4 mAb (Innogenetics) and CD27 (DakoCytomation (M-T271) or Beckman Coulter (1a4CD27)), IL-10 (MAB217), and TGFβ (MAB1835; R&D Systems Europe).
The contact dependency of Treg was studied in Transwell experiments using 24-well plates (Greiner). Briefly, 1 × 106 CD4+CD25− autologous Tresp and 1 × 106 gamma-irradiated stimulator PBMC were cultured in the lower compartment of the well. Titrated numbers of Treg were cultured in the presence 1 × 106 of gamma-irradiated target allogeneic stimulator PBMC in the Transwell insert (0.4 μm pore size; Millicell; Millipore). On day 5 of the cultures, equivalent culture volumes were transferred from the lower compartment of the 24-well plate to a 96-well, round-bottom plate and analyzed for proliferation.
Flow cytometry, intracellular cytokine staining, and Abs
Cells were phenotypically analyzed by four- or five-color flow cytometry as described previously (32). The following conjugated mAb were used: CD3 (UCHT1), CD4 (MT310), CD8 (DK25), CD27 (M-T271), CD45RA (4KB5), CD45RO (UCHL1) FITC- or PE-labeled (DakoCytomation); CD25 (M-A251) PE, CD70 (Ki-24) PE, CD134 (ACT35) PE, integrin β7 (FIB504) PE, and CCR7 (3D12) Pe-Cy7 (BD Biosciences); CD122 (CF1) PE and CD152 (BNI13) PE (Beckman Coulter; eBioscience); CD4 (T4) ECD, CD4 (T4) PC5 and CD25 (B1.49.9) PC5, and CD62L (DREG54) ECD (Beckman Coulter); and GITR (110416) PE (R&D Systems). Isotype-matched Abs were used to define marker settings. Intracellular analysis of CTLA-4 was performed after fixation and permeabilization, using Fix and Perm reagent (AnDerGrub). Intracellular cytokines were studied by flow cytometry after stimulation with PMA and ionomycin, as described previously (32). The following mAb were used; anti-human IL-2, IL-4, and IL-10 (PE-labeled) and IFN-γ and TNF-α (FITC-labeled; BD Biosciences).
Western blot analysis of Forkhead/winged helix transcription factor gene (FoxP3)
Cells (1 × 107 cells/ml) of each T cell subset were lysed in buffer containing PBS, 1% Igepal CA-630, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM EDTA (chemicals from Sigma-Aldrich), and Protease Inhibitor Cocktail (BD Biosciences). The lysates were cleared by centrifugation, diluted 1/1 in Laemmli sample buffer (0.188 M Tris-HCl (pH 6.8), 6% SDS, 30% glycerol, 15% 2-ME, and 0.03% Bromphenol Blue) and boiled for 5 min before separation on a 10% SDS-PAGE gel. The samples were electrotransferred to polyvinylidene difluoride membranes (Millipore; Immobilon-P) after electrophoresis. Filters were blocked in TBS with 0.1% Tween 20 and 5% skim milk (BD Biosciences) and subsequently incubated with polyclonal purified mouse anti-human FoxP3 (eBioscience) overnight at 4°C, The primary Ab was detected by HRP-conjugated rabbit anti-mouse Ab (DakoCytomation). Blots were developed with ECL lighting (Amersham Biosciences). Equal loading was confirmed after stripping and reprobing with anti-β-actin mAb (Sigma-Aldrich).
T cells (0.5–2 × 106) were labeled with 0.2–1 μM carboxyfluorescein diacetate succinimidyl ester (CFDA-SE; Molecular Probes) just before stimulation. In the cell, esterases cleave the acetyl group, leading to the fluorescent diacytylated CFSE. Cell division accompanied by CFSE dilution was analyzed by flow cytometry.
ELISA for cytokine detections
Cell-free culture supernatants were stored at −80°C, thawed, and subsequently analyzed for the presence of IFN-γ and TNF-α by ELISA (Sanquin), according to the manufacturer’s instructions.
Chromium release assay to examine the induction of CTLs
The cytotoxic potential of alloantigen-primed T cells was examined in a 51Cr release assay as reported previously (32). Briefly, allogeneic 51Cr-labeled PHA blasts were used as target cells and tested in triplicate at different E:T cell ratios. Culture supernatants were examined for released 51Cr on a gamma irradiation counter (Wallac; 1470 gamma counter). Control experiments were performed using autologous PHA blasts. The cytotoxic capacity is shown as the percent lysis of the indicated target, calculated according to the following equation: % lysis = 100% × ((cpm sample release − cpm spontaneous release)/(cpm total release − cpm spontaneous release)).
CD27/CFSE cell sorter-based isolation of alloantigen-specific regulatory T cells
Freshly isolated purified CD4+CD25+ regulatory T cells were labeled with CFSE, and subsequently a total of 2.5 × 104 labeled cells was cultured with 1 × 105 fully HLA-mismatched stimulator cells (target) in the presence of IL-2 (10 U/ml) and IL-15 (10 ng/ml), using 96-well, round-bottom plates. On days 9–11 of the cultures, cells were collected, washed, and labeled with anti-CD27-PE. Next, the dividing cells characterized by low CFSE content were sorted based on high CD27 (CD27+Treg) expression and the absence of CD27 (CD27−Treg; Fig. 1). The sorted cells were rested 1–2 days in medium containing 2–5% HPS and 5 ng/ml IL-15. Next, 2.5 × 104 rested cells were restimulated with 1 × 105 target allogeneic stimulator PBMC in the presence of IL-2 and IL-15 for 9–11 days. During this culture period, the content of the wells was split, and fresh culture medium containing IL-2 plus IL-15 was added. Next, the cells were collected, allowed to rest for 2 days, and examined for function and phenotype (Fig. 1).
Ex vivo induction of alloantigen-specific Treg from freshly isolated naturally occurring CD4+CD25+ Treg
Previously, we showed that costimulation blockade induced alloantigen-specific regulatory T cells that can be successfully expanded using Ag and T cell growth factors (32). We further developed this expansion protocol to determine whether we could successfully induce alloantigen-specific Treg starting with naturally occurring CD4+CD25+ Treg. CD4+CD25+ and CD4+CD25− T cells were isolated from freshly isolated PBMC (Fig. 2,A) and examined for their proliferative potential upon alloantigen stimulation in the presence of exogenously added human rIL-15 and/or IL-2 (Fig. 2,B). Compared with CD4+CD25− T cells, CD4+CD25+ Treg cells appeared hyporesponsive upon alloantigen stimulation alone. Addition of IL-2, IL-15, or a combination of the two reversed this hyporesponsive state. Especially the combined use of IL-2 and IL-15 resulted in profound proliferation of allogenetically stimulated CD4−CD25+ T cells. Cytokine stimulation alone, in the absence of Ag, was not sufficient to induce proliferation of CD4+CD25+ T cells (Fig. 2 B).
Next, the effector-suppressor function and Ag specificity of the (target) alloantigen-expanded CD4+CD25+ Treg was addressed in MLC coculture suppression assays. The expanded CD4+CD25+ Treg showed dose-dependent suppression (50% inhibition at a Treg:T responder ratio of 1:20; Fig. 2,C). The expanded Treg appeared Ag-specific, because the suppressor function was only activated upon stimulation with target alloantigen stimulator PBMC and not upon stimulation with third-party HLA-mismatched PBMC (Fig. 2,C). Restimulation assays revealed that the expanded CD4+CD25+ T cells remained anergic. Ag specificity of the alloantigen expanded CD4+CD25+ Treg was confirmed in restimulation assays, because only stimulation with target alloantigen, but not third-party Ag, and exogenous IL-2 could reverse the hyporesponsive state of the Treg (Fig. 2 D).
In short-term cultures of 9–11 days, freshly isolated CD4+CD25+ and CD4−CD25− cells stimulated with alloantigen and IL-2 plus IL-15 showed a 6.6 ± 4.6-fold (n = 7; range, 3.2–14.4) and 20.9 ± 9.4-fold (n = 7; range, 10.5–36) expansion, respectively. Similar expansion levels were found in subsequent stimulation cycles without loss of Treg function (data not shown). Thus, starting with freshly isolated CD4+CD25+ Treg in short-term ex vivo cultures, we demonstrated the induction and expansion of Treg populations with defined Ag specificity.
CD27/CFSE-based selection and purification of ex vivo induced Ag-specific Treg subsets with high effector-suppressor potential
For the purpose of Ag-specific Treg therapy, we further focused on the development of a method that allows both generation and selection of alloantigen-specific Treg from naturally occurring Treg. Therefore, we exploited the feature that Ag reactive Treg proliferate well upon stimulation with alloantigen and IL-2 plus IL-15 (Fig. 2). Thus, we labeled freshly isolated CD4+CD25+ Treg with CFSE and analyzed cell division upon allogeneic stimulation in the presence of exogenously added IL-2 plus IL-15. As anticipated, CFSE analysis of cell division on day 4 of the cultures revealed that a minor fraction of the CD4+CD25+ Treg pool was dividing in response to allogeneic stimulation and IL-2 plus IL-15 (Fig. 3,A). This was in contrast to polyclonal stimulation with anti-CD3 mAb and IL-2 plus IL-15, which drove virtually all CD4+CD25+ Treg into cell division (Fig. 3 B).
Then we studied differentiation markers on the dividing cell population. Strikingly, on day 6 of culture (but also earlier), two dividing Treg subsets marked by differential CD27 expression could be distinguished: a CD27+ (CD27+Treg) and a CD27− (CD27−Treg) population (Fig. 3,A). The CD27+Treg showed high level expression of CD27 (CD27high; Fig. 3, A and C). On day 6 of cultures, the majority of both CD27+Treg and CD27−Treg were CD45RO+/CD45RA− and CD25+ (data not shown). In contrast to freshly isolated CD4+CD25+ cells, the majority of dividing fresh CD4+CD25− cells lost CD27 expression, whereas only a residual fraction was CD27+ (these cells will be referred to as CD27+CD4 control and CD27−CD4 control, respectively). Of note, the use of highly purified flow cytometric cell-sorted CD4+CD25high cells, instead of magnetic bead-isolated CD4+CD25+ (our standard isolation procedure), also resulted in CD27+ and CD27− Treg subsets after stimulation with alloantigen and IL-2 plus IL-15 (data not shown).
Purified CD27+Treg and CD27−Treg subsets were analyzed in more detail for effector-suppressor function and T cell anergy. To this end, divided CFSElow/CD27+ and CFSElow/CD27− cells were cell sorted by flow cytometry as indicated in Fig. 3,C. Immediately after sorting, CD27+Treg and CD27−Treg showed suppressive capacity upon stimulation with target alloantigen (Fig. 4 A).
Because only minute cell numbers were obtained after cell sorting, the cells were further expanded by stimulation with (target) alloantigen in the presence of IL-2 and IL-15 before additional study. In short-term cultures of 9–11 days, 10.2 ± 7.7 (range, 1.2–20.0) and 27.5 ± 3.4 (range, 22.6–30.0) mean fold expansions were observed for the CD27+Treg and CD27−Treg subsets, respectively (n = 5). For comparison, CD27+CD4 and CD27−CD4 control cells showed mean expansion levels of 35.1 ± 13.0 (range, 16.8–50.7) and 41.5 ± 18.6 (range, 22.4–72.0), respectively.
Both CD27+Treg and CD27−Treg subsets showed potent effector-suppressor potential in MLC coculture suppression assays conducted with naive autologous CD4+CD25− responder T cells and target alloantigen stimulator PBMC (50% inhibition at Treg:Tresp ratios of 1:500 and 1:50, respectively; Fig. 4,B). CD27+Treg consistently appeared as ∼10 times more potent suppressors than CD27−Treg (Fig. 4,B). It is unlikely that suppression by the CD27−Treg population was due to contamination with CD27+Treg, because the suppressive CD27+Treg, which showed the characteristic CD27high expression, were absent from the CD27−Treg population (Fig. 6 A). Furthermore, we excluded that a toxic effect of the highly purified Treg populations is part of the suppressive effect that we observed, because suppressed cells survived, and the Treg did not affect baseline proliferation of epithelial cells (data not shown).
The suppressor function of both CD27+Treg and CD27−Treg was Ag-specific, because target alloantigen induced clear suppression, whereas MLC coculture assays performed with third-party, HLA-mismatched stimulator cells were hardly affected (Fig. 4,B). Only at relatively high cell doses of CD27+Treg, suppression of third-party responses was observed (Fig. 4 B).
The Ag specificity of the of the CD27+Treg and CD27−Treg CD4+CD25+ Treg was confirmed in restimulation assays, because only stimulation with target alloantigen, but not third-party alloantigens, and exogenous IL-2 could reverse the hyporesponsive state of the Treg (Fig. 4,C). Notably, there was a difference in response to target alloantigen and IL-2 between the two Treg subsets in restimulation assays. Although addition of exogenous IL-2 to CD27+Treg resulted in limited reversal of proliferation (i.e., deep anergy), addition of IL-2 to CD27−Treg resulted in complete recovery of the proliferative response (i.e., classical anergy; Fig. 4 C).
Thus, starting with a freshly isolated CD4+CD25+ polyclonal Treg population, we demonstrated the induction, expansion, and selection, of two highly suppressive, alloantigen-specific Treg subsets.
CD27+Treg and CD27−Treg subsets suppress T cell effector function
After studying the inhibition of proliferation, we analyzed the potential of CD27+Treg and CD27−Treg subsets to suppress effector functions. Both Treg subsets suppressed accumulation of IFN-γ and TNF-α (Fig. 5,A) in primary mixed lymphocyte cultures. Again, CD27+Treg were more potent in their suppressor function compared with CD27−Treg. Also, the induction of cytotoxic effector cells after allogeneic priming of T cells was suppressed by the Treg subsets (Fig. 5 B).
CD27+Treg and CD27−Treg subsets are distinguished by surface marker expression and cytokine-producing potential
We noted that CD27+Treg constitutively expressed CD25 and CD27, whereas CD27−Treg, like CD27+/−CD4 control cells, became CD25 negative at rest and turned CD25 positive upon activation (Fig. 6 A). This phenomenon was observed after CD27/CFSE-based selection (data not shown) as well as during the subsequent culture steps. Apparently, the dynamics of CD25 expression of the CD27− Treg subset are fundamentally different from those of the CD27+ Treg subset. Moreover, in contrast to control CD27+CD4 cells, CD27+Treg consistently showed high CD27 expression levels (CD27high).
At rest, the majority of CD27+Treg were positive for CD62L (84%), CD152 (88%), and β1 (CD29; 71%) and β7 integrin (74%) family members; a fraction of the cells expressed GITR (50%) and CD122 (32%); a minor population expressed OX-40 (CD134, 14%) and PD-1 (8%); and the majority of the population was negative for CD70 (Fig. 6,B). Virtually all CD27−Treg were positive for CD70 (87%), GITR (84%), β1 (97%) and β7 integrin (99%; both integrin members were expressed at a high level; Fig. 6 B), a fraction of the cells expressed CD152 (55%) and CD122 (34%), a small proportion of cells expressed CD134 (13%) and PD-1 (8%), and the majority were negative for CD62L (3%). The TNFR family members 4-1BB and CD30 were absent on both CD27+Treg and CD27−Treg (data not shown).
Cytokine production by either Treg subset was analyzed by intracellular cytokine staining after stimulation by PMA and ionomycin. The CD27+Treg population was characterized by relatively high numbers of IL-10 and low numbers of IL-2-, IFN-γ-, and TNF-α-producing cells (Fig. 6 C). In contrast, the CD27−Treg population was especially comprised of IL-2, IFN-γ-, and TNF-α-producing cells.
Finally, we showed that FoxP3, a transcription factor that is strongly associated with Treg in mice (33, 34), is expressed in both expanded 27+Treg and 27−Treg subsets as well as in the expanded 27+CD4 and 27−CD4 control cell populations (Fig. 6 D). Of note, in concert with our findings, it was recently demonstrated that human CD4+CD25− T cells express FoxP3 upon activation (35).
To summarize, Ag-specific CD27+Treg and CD27− Treg subsets were distinguished on the basis of distinct expression levels of CD25, CD27, CD70, CTLA-4, the migratory markers CD62L, β1 and β7 integrins, and a characteristic cytokine profile.
Effector suppressor function of CD27+Treg and CD27−Treg is contact dependent and does not rely on CD27 signaling
Currently, little is known about how Treg mediate their suppressive effect. They might perform their immunosuppressive function in either a contact-dependent or -independent manner (22) via CTLA-4 interactions (25, 27, 28) or by cytokines such as IL-10 and TGF-β (23, 24, 25, 26). We thus investigated whether these mechanisms play a role in the suppressor function of CD27+Treg or CD27−Treg. Additionally, the requirement for CD27 signaling in effector-suppressor function was examined.
In Transwell MLC coculture suppression assays, we demonstrated that for both subsets, contact between Treg and responder T cells is required for effector-suppressor function (Fig. 7,A). This suppression could not be reversed by mAb directed against CD27 or CTLA-4 interactions (Fig. 7, B and C). Preliminary experiments suggested that anti-IL-10 or anti-TGF- mAb were unable to restore the proliferative response of the suppressed T cells (Fig. 7,C). As anticipated, anti-CTLA-4 mAb enhanced the proliferative response of the responder T cells in the cocultures regardless of the absence or presence of Treg (Fig. 7 C). In none of these experiments did anti-CTLA4 mAb restore proliferation to the levels observed in the absence of Treg.
Both CD27+Treg and CD27−Treg subsets suppress memory CD4+ T cells, whereas only CD27+Treg suppress ongoing T cell responses
Both CD27+Treg and CD27−Treg subsets clearly showed suppression of naive alloreactive CD4+CD25− T cells (Figs. 4, 5, and 7). This suggests that both Treg subsets are of potential interest for (donor) Ag-specific immunotherapy especially in nonprimed patients. However, for use in primed patients and in patients suffering from ongoing T cell-mediated autoimmune disease, it is crucial to establish whether these Treg can prevent ongoing and memory T cell responses. To study the potential of the Treg subsets to suppress ongoing T cell responses, Treg were titrated and added to an MLC coculture suppression assay on day 0, 2, or 3 of the culture. In contrast to CD27−Treg, only CD27+Treg suppressed ongoing T cell responses (Fig. 8,A). Likewise, to study the capacity of Treg subsets to suppress memory T cell responses, MLC coculture suppression assays were conducted with allogenetically primed CD4+ memory responder T cells and target stimulator PBMC, and titrated numbers of Treg (Fig. 8,B). Both subsets were able to suppress memory T cell proliferation, but, compared with CD27−Treg, CD27+Treg were more potent suppressors (Fig. 8 B).
Immunosuppressive Treg are of great interest for immunotherapy in the prevention of transplant rejection and treatment of autoimmunity. Currently it is recognized that Treg are particularly effective when they have specificity for their target Ag(s) (9, 11, 12, 13, 14). Moreover, the availability of target Ag-specific Treg will reduce the number of cells needed for immunotherapy. Peripheral blood contains limited numbers of naturally occurring CD4+CD25+ Treg, this population shows broad Ag reactivity. Hence, for Ag-specific immunotherapy with CD4+CD25+ Treg, the ex vivo induction, expansion, and selection of target Ag-specific Treg are necessary.
Large-scale ex vivo expansion of human CD4+CD25+ Treg by stimulation with anti-CD3 and anti-CD28 mAb and high dose IL-2 (100 U/ml or more) have successfully been established (21, 31). This polyclonal, non-Ag-specific, expansion increases the total number of Treg, but leaves the frequency of putative Ag-reactive Treg unchanged (21). This is also clear from our CFSE experiments showing that polyclonal stimulation with anti-CD3 mAb and IL-2 plus IL-15 provoked cell division in almost all CD4+CD25+ T cells, whereas alloantigen stimulation and IL-2 plus IL-15 only drove cell division of the alloantigen-specific CD4+CD25+ T cells. In this study we used this feature combined with a CD27/CFSE-based purification protocol to isolate alloantigen-specific Treg from the naturally occurring CD4+CD25+ Treg pool. To our knowledge this is the first study that demonstrates the ex vivo induction, selection, and expansion of Ag-specific Treg subsets obtained from freshly isolated CD4+CD25+ Treg.
Recently, it was demonstrated that a cell division-based selection approach by CFSE-cell sorting proved successful in the isolation of alloantigen-reactive and nonalloreactive T cells when using conventional responder T cells (36). The alloantigen-experienced primed T cells exclusively resided in the CFSElow cell population. In this study we demonstrate that a comparable approach permits isolation of potent alloantigen-specific Treg. Moreover, cell sorter isolation based on CFSE intensity combined with the expression of CD27 enabled selection of Ag-specific CD27+ and CD27− Treg subsets, which are both of interest for Ag-specific immunotherapy.
The Ag-specific Treg subsets described in this study showed highly potent suppressor-effector function, with 50% inhibition achieved at Treg:Tresp ratios of 1:500 to 1:50 for CD27+Treg and CD27−Treg subsets, respectively. These effector-suppressor effects are far more potent than what has previously been reported for freshly isolated human CD4+CD25+ (15, 16, 17, 37), CD4+CD25high Treg (38) (50% inhibition at Treg:Tresp ratios of 1:1 up to 1:10), or large-scale, polyclonally expanded Treg (50% inhibition at Treg:Tresp ratios of 1:1 up to 1:32) (21, 31).
Our Ag-specific purification procedure includes multiple Ag-specific expansion cycles. After each expansion cycle, enhancement of effector-suppressor potential occurred, which is best exemplified by a reduction in 50% inhibitory E:T cell ratios of 1:5 for freshly isolated CD4+CD25+ Treg (data not shown), 1:20 after one allogeneic expansion cycle, up to 1:50–1:500 after CD27/CFSE-based cell sorting and Ag-specific expansion. Although an increase in suppressor potential might be the result of preactivation (18, 31), in this study we show that the enrichment of target Ag-specific Treg strongly contributes to an increase suppressor potential.
Recently, Godfrey et al. (31) and Hoffmann et al. (21) demonstrated large-scale polyclonal expansion (100- and 2800-fold expansion, respectively) of freshly isolated human CD4+CD25+ Treg. After expansion, both groups found a CD62L+CD27+ CD4+CD25+ Treg population that expressed CTLA intracytoplasmatically. Interestingly, using our alloantigen-induced protocol and a CD27/CFSE-based cell sorting approach, we obtained two phenotypically distinct suppressive CD4+Treg subsets, characterized as either CD27+CD25+CD70−CD62L+CTLA4high or CD27−CD25−CD70+CD62L−CTLA4int. The CD4+CD25+ Treg isolation procedures used by Godfrey and Hoffman (21, 31) were highly stringent. Therefore, to exclude that the isolation procedure might induce the observed differences, we simultaneously performed experiments with highly purified flow cytometry-sorted CD4+CD25high Treg. This protocol resulted in similar CD27+Treg and CD27−Treg populations (data not shown).
Treg are not defined by CD25 expression per se. In this study we show that upon in vitro activation of freshly isolated CD4+CD25+ Treg, next to a CD25+ population, also a CD25− suppressive subset emerges. Notably, this subset distribution only becomes apparent after rest. Also, in experimental transplant (39) and autoimmune (40, 41) models, both CD4+CD25+ as well as CD4+CD25− suppressive Treg populations have been detected. In concert with our human in vitro data, in vivo mouse CD4+CD25+ Treg appeared ∼5- to 10-fold more potent than the CD4+CD25− Treg (39, 41). The fact that CD25− Treg proved suppressive in vivo supports the use of ex vivo generated alloantigen-specific CD25− CD27−Treg next to or in combination with CD25+CD27+Treg for immunotherapy.
Control of the normal immune responses by Treg has been proposed to take place in different phases (1): a steady state, a response phase in which aggressive T cells and Treg are stimulated and proliferate, and a regulatory phase where Treg effector-suppressor function dominates. The question arises of why CD27+ and CD27− Treg subset heterogeneity emerges. CD27+Treg reveal high suppressor activity and poor growth potential, whereas CD27−Treg, although suppressive by nature, proliferate strongly upon encounter with Ag and T cell growth factor. In vivo, both a long-lived, nondividing and a proliferating Treg subset were described (42). The long-lived cells are believed to reside in the lymph node, whereas the rapidly dividing Treg recirculate. The CD27+Treg subset that we describe might be committed to local effector-suppressor function in the draining lymph node. The dividing CD27−Treg subset may be destined to enter the peripheral Treg pool, which will require high cell numbers, thereby increasing the likelihood of encountering the target Ag, and facilitate a balance shift toward regulation at the expense of aggressive T cells.
The CD27+Treg population described in this study appeared extremely suppressive. These cells expressed very high levels of CD27. A link between CD27 expression and suppressor function appeared obvious, but we showed that this is not the case. Because interactions between CD27 and its ligand CD70 play a role in controlling expansion, survival, and/or effector cell differentiation in T cell and B cell biology (43, 44, 45), it may well be that the expression of CD27 is implicated in the homeostasis of CD27+Treg.
From a total number of 250–300 × 106 PBMC, 6.4 ± 7.2 × 106 allo-specific CD27+Treg and 28.9 ± 23.7 × 106 allo-specific CD27−Treg were obtained (Table I). It is likely that immunotherapy will require higher cell numbers to be infused. Comparable to the report by Guinan et al. (46), we aimed at a yield of 0.5 × 107/kg dedicated Treg for immunotherapy. After CD27/CFSE cell sorting of CD27+Treg or CD27−Treg, we obtained 0.7 × 106 and 0.8 × 106 Treg (Table I), with a mean expansion of 10.2- and 27.5-fold, respectively. This implies that to obtain the required CD27+Treg and CD27−Treg, three and two expansion cycles are needed, respectively. Alternatively, Treg yield will be boosted by increasing the number of CD4+ cells at the start of the procedure using aphaeresis products (e.g., 9 l aphaeresis will yield 5–12 × 109 T cells, instead of 8–9 × 107 CD4+ cells recovered in this study) and by optimizing the culture conditions and the selection procedure. In fact, for clinical application of the Treg subsets, a magnetic bead selection might be warranted. Such an alternative selection procedure is feasible, because the CD27+ and CD27− Treg subsets show reciprocal expression of CD27 and CD70, respectively. Cursory examination of a magnetic bead-based isolation approach combined with anti-CD27 and anti-CD70 mAb enabled purification of functional CD27+Treg and CD27−Treg. This procedure was further optimized.
|.||Start PBMC .||CD4− .||CD4+CD25+ .||First Alloexpansion (n = 7) .||CD27/CFSE Sort Using Flow Cytometry .||.||Second Alloexpansion (n = 5) .||.|
|.||.||.||.||.||CD27+ .||CD27− .||CD27+ .||CD27− .|
|Mean cell numbers (×106)||250–300||60–80||1.58 ± 0.4||9.7 ± 5.81||0.7 ± 0.4||0.8 ± 0.9||6.4 ± 7.2||28.9 ± 23.7|
|.||Start PBMC .||CD4− .||CD4+CD25+ .||First Alloexpansion (n = 7) .||CD27/CFSE Sort Using Flow Cytometry .||.||Second Alloexpansion (n = 5) .||.|
|.||.||.||.||.||CD27+ .||CD27− .||CD27+ .||CD27− .|
|Mean cell numbers (×106)||250–300||60–80||1.58 ± 0.4||9.7 ± 5.81||0.7 ± 0.4||0.8 ± 0.9||6.4 ± 7.2||28.9 ± 23.7|
In transplantation, control of the immune response by Treg primarily takes place in the lymphoid organs (47) and at the graft site (48). Hence, for successful Treg therapy, it is imperative that the ex vivo generated Treg migrate to these sites. Migration of T cells is controlled by adhesion molecules such as CD62L integrin β1 and β7 family members and chemokine receptors, such as CCR7 (49, 50, 51). CD27+Treg expressed CD62L, the β1 (CD29) and β7 integrin family members, whereas the chemokine receptor CCR7 (data not shown) was only marginally expressed. CD27−Treg were devoid of CD62L and CCR7, but did express high levels of β1 and β7 integrins. Although little is known about the in vivo trafficking characteristic of ex vivo generated Treg, the differences in migration receptor pattern between CD27+Treg and CD27−Treg suggests that the cells might function at distinct sites.
To summarize, the efficacy of Treg-based immune therapy critically depends on the Ag specificity of the Treg (13, 14). Therefore, we developed a CD27/CFSE cell sorter-based ex vivo culture procedure to induce, select, and expand alloantigen-specific Treg starting of with freshly isolated CD4+CD25+ Treg. This approach enabled selection and expansion of two highly suppressive subsets of multipotent alloantigen-specific Treg. The availability of Ag-specific Treg subsets with high suppressor potential will probably reduce the total number cells needed for immunotherapy. This will facilitate the implementation of Ag-specific Treg therapy in living donor transplants, thereby guiding permanent graft survival and even tolerance without the need for life-long, nonspecific immunosuppressive drugs.
We are very grateful to A. Pennings and G. Vierwinden (Central Hematology Laboratory, Radboud University Nijmegen Medical Center) for expert cell sorting. We thank Ronald van Beek for skilled Western blot analysis, J. Coenen for critical reading of the manuscript, and C. van den Brink for technical assistance.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by Research Grant C02-2004 from the Dutch Kidney Foundation.
Abbreviations used in this paper: Treg, regulatory T cell; FoxP3, Forkhead/winged helix transcription factor gene; HPS, human pooled serum; MLC, mixed lymphocyte culture; Tresp, responder T cell.