SEARCH

SEARCH BY CITATION

Keywords:

  • CD26;
  • regulatory T cells;
  • effector T cells;
  • flow cytometry;
  • magnetic sorting

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED
  8. Supporting Information

A major obstacle hampering the therapeutic application of regulatory T (Treg) cells is the lack of suitable extracellular markers, which complicates their identification/isolation. Treg cells are normally isolated via CD25 (IL-2Rα) targeting, but this protein is also expressed by activated CD4+ effector T (Teff) lymphocytes. Other extracellular (positive or negative) Treg selection markers (e.g., HLA-DR, CD127) are also nonspecific. CD26 is an extracellular peptidase whose high expression has been traditionally used as an indicator of immune activation and effector functions in T cells. Now, we provide flow cytometry data showing high levels of CD26 within CD4+CD25 or CD4+FoxP3−/low effector T (Teff) lymphocytes, but negative or low levels (CD26−/low) in Treg cells selected according to the CD4+CD25high or the CD4+FoxP3high phenotype. Unlike the negative marker CD127 (IL-7Rα), which is down modulated in CD4+ Teff lymphocytes after TCR triggering, most of these cells upregulate CD26 and take a CD4+CD25+/highCD26+ phenotype upon activation. In contrast, there is only a slight upregulation within Treg cells (CD4+CD25highCD26−/low). Thus, differences in CD26 levels between Treg and Teff subsets are stable, and assessment of this marker, in combination with others like CD25, FoxP3, or CD127, may be useful during the quantitative evaluation or the isolation of Treg cells in samples containing activated Teff lymphocytes (e.g., from patients with autoimmune/inflammatory diseases). © 2012 International Society for Advancement of Cytometry

Regulatory cells are included in different leukocyte populations, like CD8+ (1) or CD4+ (2, 3) T subsets. CD4+CD25high natural Treg (nTreg) cells constitute a small CD4+ subpopulation (<5% of CD4+ T lymphocytes) with a thymic origin. Numerical or functional deficit of nTreg cells is linked to autoimmune diseases such as multiple sclerosis (MS), Type 1 diabetes or rheumatoid arthritis (RA) (4). In addition, they are also important to prevent (or delay) grafts rejection or disproportionate responses to bacterial/viral antigens (5, 6). Alternative subpopulations of regulatory T cells (e.g., TH3 and Tr1), called “adaptive” or “induced” regulatory CD4+ T cells (iTreg), are generated in the periphery (7). Both nTreg and iTreg lymphocytes (altogether, Tregs) limit the biological activities (e.g., proliferation, cytokine production) of adaptive “effector” cells, a miscellaneous group consisting of CD4+ helper T cells (TH1, TH2, TH17, TFH, TH9, TH22), CD8+ cytotoxic T lymphocytes and B cells. To finely tune the strength of effector responses, Treg cells employ various suppressor mechanisms, like inhibitory soluble molecules (adenosine, TGFβ, IL-10) or cell contact-mediated pathways (e.g., membrane cytokines like TGFβ1 or surface molecules like CTLA-4) (8, 9).

Any research aimed at controlling Treg function, either enhancing (e.g., in autoimmune diseases) or blocking it (e.g., in cancer), will hold a great interest (5, 6, 10, 11). However, this kind of research faces several challenges; for example, how to distinguish regulatory (Treg) from effector (Teff) CD4+ T populations. Human Treg cells constitutively express surface proteins like CD25, CD45RO, CTLA-4, HLA-DR, or GITR (see Table 1 in Supporting Information) (2, 12), but these markers are neither present in 100% of Tregs or exclusive of this cell lineage. Three examples can illustrate this point: CD25 (IL-2Rα), FoxP3, and CD127 (IL-7Rα).

The majority of human Treg cells strongly and constitutively express CD25 (CD25high). However, conventional/effector T cells (2, 12, 13) and a portion of CCR7+ central memory T lymphocytes start expressing CD25 upon TCR-mediated activation (14). Therefore, even highly pure CD4+CD25high Treg populations may contain a significant fraction of proinflammatory Teff cells (15). On the other hand, FoxP3 is the most specific Treg marker (2). Despite this fact, FoxP3 still shows a transitory and activation-dependent expression in CD4+CD25 Teff cells, which together with its intracellular nature disqualifies this marker for Treg identification and, especially, isolation purposes (16). CD127 is another protein whose levels inversely correlate with FoxP3 expression in Treg lymphocytes (17, 18). Nevertheless, the mere existence of an underlying disease (e.g., HIV infection) (19) or the in vitro activation (20) cause an intense CD127 down modulation on formerly CD127+ Teff cells. Other alternative markers have arisen more recently. Markus Kleinewietfeld et al. (21) have reported CD49d (α chain of VLA-4 integrin) as expressed in most of IFN-γ or IL-17-producing proinflammatory T cells but reduced on Tregs, even though data from the same group (22) as well as other researchers (23) still reflect some degree of CD49d expression in some subsets of Treg cells. Mandapathil et al. (24) have focused on CD39 as a positive selection marker. This ectoenzyme catalyzes the generation of AMP from ATP, which is necessary to produce adenosine, an important mediator of active suppression (22). Fifty to 90% of CD4+CD39+ T lymphocytes are FoxP3+ and express low levels of CD127 (24). However, activated T cells upregulate CD39 (25), and a novel population of human CD4+CD39+FoxP3 T cells that produce IFN-γ and IL-17 has been found (26). Thus, it seems that more phenotypic studies on Treg cells are still necessary (27).

CD26 is a serine protease with dipeptidyl peptidase IV (DPPIV) activity (28). Activated/memory T cells display a CD26high phenotype, and TH1 cytokines like IL-12 raise the number of CD26 molecules on T lymphocytes (28, 29). Thus, high surface levels of this protease are an indication of, at least, TH1 effector responses (28, 29). Ligation of CD26 leads to activation of signal transduction proteins (e.g., ERK, p56lck, CD3ζ, ZAP-70, CARMA-1/NF-κB), cell proliferation, and cytokine (IL-2, IFN-γ) production (28, 30). Both cis-interaction of CD26 with CD45 (31, 32) and trans-association of CD26 (T cell) with caveolin-1 (APCs) (33) seem important for CD26 functions. However, despite the large amount of data supporting the costimulatory role of CD26 in T cells, an extensive research on CD26 levels in Treg cells has not been undertaken. The present work shows that CD26 is present on FoxP3-expressing activated CD4+ Teff cells, but reduced or absent from Tregs. This CD26−/low phenotype is stable, and therefore useful to differentiate these two antagonistic CD4+ T cell subsets.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED
  8. Supporting Information

Materials

Phytohemagglutinin (PHA; catalog no. L1668), paraformaldehyde (PFA; catalog no. P-6148), penicillin-streptomycin solution (catalog no. P0781), 5(6)-Carboxyfluorescein diacetate N-succinimidyl ester (CFSE; catalog no. 21888-25MG-F), and RPMI-1640 culture medium (catalog no. R4130) were obtained from Sigma-Aldrich (Spain), fetal bovine serum (FBS; catalog no. DE14-801F) from Biowhittaker (Lonza Iberica, Spain), and Ficoll-Paque PLUS (catalog no. 17-1440-03) from GE Healthcare (Spain). Mouse antibodies against human molecules CD127 (CD127-PE, catalog no. 557938, and CD127 Alexa Fluor 647, catalog no. 558598; both IgG1 clone hIL-7R-M21), CD25 (CD25-FITC, catalog no. 555431, and CD25-APC, catalog no. 555434; both IgG1, clone M-A251), CD39 (CD39-FITC, clone TU66, catalog no. 561444), CD49d (CD49d-APC, IgG1, clone 9F10, catalog no. 559881), FoxP3 (anti-FoxP3-PE, IgG1, clone 259D/C7, catalog no. 560046), and IFN-γ (anti-IFN-γ-PE, IgG1, clone B27, catalog no. 554701), as well as the mouse isotypic control IgG1-APC (clone MOPC-21, catalog no. 555751) and the Alexa Fluor 647 mouse IgG1 isotype control (clone MOPC-21, catalog no. 557714), were from BD Pharmingen (BD Biosciences, Spain). Mouse antibodies against human CD4 (CD4-PerCP, IgG1, clone SK3, catalog no. 345770) and CD26 (CD26-FITC, IgG2a, clone L272, catalog no. 340426) were purchased from BD Immunocytometry Systems (BD Biosciences). Mouse mAb against human CD26 (TP1/16) was purified from a hybridome supernatant and used pure (in combination or not with APC goat anti-mouse Ig; catalog no. 550826, BD Pharmingen) or FITC-labeled (Fluorotag FITC Conjugation Kit, catalog no. FITC1-1KT; Sigma). Murine mAb against human CD26 clone TP1/19 (CD26-FITC or APC, IgG2b, catalog no. 26F-100T or 26A-100T) was provided by Immunostep (Salamanca, Spain). Isotypic controls IgG1-FITC (clone MOPC-21; catalog no. F6397) and IgG1-PE (clone MOPC-21; catalog no. P4685) were purchased to Sigma. Buffers used for intracellular staining of FoxP3 and IFN-γ were BD Pharmingen stain buffer (catalog no. 554656), BD FACS™ lysing buffer (catalog no. 349202), Human FoxP3 buffer set (catalog no. 560098), BD Cytofix/Cytoperm buffer™ Plus (catalog no. 555028), and BD Perm/Wash™ (catalog no. 554723), all from BD Biosciences.

Methods

PBMCs purification and culture conditions

Human buffy coats from healthy donors were kindly provided by Centro de Transfusiones de Galicia (Santiago de Compostela, Spain). Once informed consent for the donation was obtained, blood samples (healthy subjects) were drawn into EDTA/K2E or LH/170 IU tubes (BD vacutainer; BD Biosciences; catalog no. 367525 and 367526, respectively) via sterile venipuncture (Medical Service at the University of Santiago) according to the ethics committee guidelines. Peripheral blood mononuclear cells (PBMCs) were isolated from either buffy coats or anticoagulant treated blood samples using Ficoll® density gradient centrifugation, as previously described (29, 32). PBMCs were either used directly or in vitro cultured (37°C, 5% CO2 humidified atmosphere). RPMI-1640 growth medium was supplemented with 10% FBS, 100 μg/ml streptomycin and 100 UI/ml penicillin (complete medium). Stimulation of PBMCs was performed by using 1 to 2 μg/ml PHA (phytohemagglutinin from Phaseolus vulgaris) for 5 days in either 6/24-well plates or 75/150 cm2 flasks. Alternatively, PBMCs were activated with beads coated with monoclonal antibodies against CD2/CD3/CD28 (Treg suppression inspector human, Miltenyi Biotec, Auburn, CA; 0.5–1 bead per lymphocyte; catalog no. 130-092-909). In IFN-γ assays, culture medium was supplemented with GolgiPlug™ (BD Biosciences; 1/1,000 dilution) during the last 4 h of incubation before immunostaining.

Human Treg purification

Viable (>90%; trypan blue exclusion assay) Treg cells were isolated from either resting or activated PBMCs by two different magnetic methods. The first one (Dynabeads® Regulatory CD4+CD25+ T cell Kit; Life-Technologies, Spain; catalog no. 113.63D) was a “classical” procedure. During the first step, CD4+ T cells were enriched by negative selection by means of a cocktail of monoclonal antibodies (Antibody Mix Human CD4; mouse IgG antibodies against CD14, CD16a, CD16b, CD56, CDw123, CD36, CD8, CD19 and glycophorin A) and anti-mouse IgG Abs linked to super paramagnetic polystyrene beads (Depletion MyOne™ Dynabeads). Afterwards, both CD4+CD25 Teff and CD25high Treg cells were purified from CD4+ T cells by means of beads coated with antibodies against CD25 (Dynabeads CD25). Finally, magnetic beads were removed from Treg cells (DETACHaBEAD® buffer). Cell recovery was calculated as follows: (number of cells within the Treg enriched fraction × Treg purity)/(starting cell number × starting Treg percentage).

The second approach was the Human CD4+ CD127lowCD49d Regulatory T Cell Enrichment Kit (StemCell Technologies, Grenoble, France; catalog no. 19232). This fully negative selection protocol uses CD127 and CD49d to isolate the Treg population from a PBMCs suspension. We have followed strictly the manufacturer's manual (EasySep™ protocol) in combination with the EasySep magnet (for standard 5 ml polystyrene tubes; catalog no. 18000). Despite this fact, a low Treg viability was always observed.

Flow cytometry assays

All the incubations during immunofluorescence protocols were performed in the dark, and unstained cells, isotype controls, and single fluorochrome stained cells were used to set-up the flow cytometer. In all cases, recommended concentrations were employed for all monoclonal antibodies. In both proliferation and immunofluorescence experiments, sample acquisition (typically 20,000 to 200,000 events) was performed on a FACScalibur Flow Cytometry System (BD Biosciences), a 2-laser, 4-color flow cytometer used for clinical samples at the USC University Hospital Complex (CHUS). This instrument has the following laser-fluorochrome combinations: (1) 488 nm blue laser—fluorescein isothiocyanate (FITC), 5(6)-carboxyfluorescein diacetate N-succinimidyl ester (CFSE), phycoerythrin (PE), and peridinin-chlorophyll-protein complex (PerCP); (2) 635-nm red-diode laser—allophycocyanin (APC) and AlexaFluor 647. We used WinMDI software (Dr. Joe Trotter, The Scripps Institute, Flow Cytometry Core Facility) for data analysis.

For CFSE-based proliferation studies, PBMCs (RPMI medium, 10 × 106 cells/ml) were firstly incubated with 5 mM CFSE for 8 min (RT) in the dark. To block the reaction, an equal volume of FBS was added and cells were thoroughly washed before counting. Initial cell density was 50,000 cells/well (0.25 × 106 PBMC/ml; 96-round bottom well microplates), and 2 μg/ml PHA was used as mitogen (34). Each experimental condition was tested nine times. At the end of the incubation period (5 days), wells were pooled to generate single triplicates. Cells were washed with PBS pH 7.4, and indirectly labeled with either isotype antibody or a purified mAb against CD26 (TP1/16) plus GAM-APC. Unlabeled (CFSE) cells served as negative controls, while unstimulated CFSE-labeled PBMCs allowed us to identify those lymphoblasts that did not divide.

For the multicolor extracellular staining of unfractionated leukocytes, either isotype (isotype-FITC, isotype-PE, isotype-APC) or specific (CD26-TP1/19-FITC, CD127-PE, CD4-PerCP, or CD25-APC) monoclonal antibodies were added to test tubes. Hundred microliters of EDTA anticoagulated whole blood was mixed with antibodies by gentle vortexing and left for 30 min (RT). Afterwards, erythrocytes were lysated (2 ml 1× BD FACS lysing solution/test tube; 15 min, RT), and samples centrifuged (200g, 5 min, RT), washed with 2 ml of PBS pH 7.4, 1% FBS, 0.1% sodium azide (washing solution), and fixed with 2 ml ice-cold 2% PFA in PBS pH 7.4 (30 min, RT). Finally, cells were washed again and resuspended in 1 ml of washing solution.

During the simultaneous detection of extracellular (CD4, CD25, CD26, CD49d, CD39) and intracellular (FoxP3) proteins in either resting or PHA-activated PBMCs, samples were washed with BD Pharmingen Stain Buffer, counted (hemocytometer) and adjusted to 10 × 106 cells/ml. For extracellular labeling, adequate volumes of isotype (isotype-FITC and isotype-APC) or specific antibodies (CD26-FITC TP1/16, CD4-PerCP, and either CD25-APC or CD49d-APC) were placed inside flow cytometry tubes and 1 × 106 cells (100 μl) added. In some experiments (Fig. 5), samples were simultaneously stained with CD49d-APC and CD127 Alexa Fluor 647 (same fluorescence channel). After incubation (20 min, RT), cells were washed (2 ml of BD Pharmingen Stain Buffer per tube) and centrifuged at 250g 10 min RT. Then, cell pellet was fixed with 1× Human FoxP3 buffer A (2 ml/tube; 10 min at RT), centrifuged again (500g, 5 min, RT) and the wash step repeated. Fixed and extracellularly labeled cells were permeabilized with 1× Human FoxP3 buffer C (0.5 ml/tube) for 30 min at RT. After washing, permeabilized cells were incubated (30 min, RT) with isotype-PE or anti-FoxP3-PE. Cells were washed again, centrifuged (500g, 5 min, RT) and fixed (1 ml, 2% PFA) during 30 min at RT. Finally, leukocytes were resuspended in 1 ml of BD Pharmingen Stain Buffer per tube.

We used a similar protocol to evaluate the percentage of IFN-γ-producing T lymphoblasts. In brief, GolgiPlug™ treated lymphoblasts were also extracellularly labeled with diverse combinations of isotypic (FITC- and APC-labeled) or specific (CD26-FITC TP1/16, CD25-FITC, CD4-PerCP, and CD49d-APC or CD25-APC) mAbs. Once washed, cells were fixed and permeabilized (20 min, 4°C) with 250 μl of BD Cytofix/Cytoperm buffer, washed two times with 1 ml of 1× BD Perm/Wash™ and incubated with anti-IFN-γ-PE or isotype-PE in 50 μl of 1X BD Perm/Wash (30 min, 4°C). Finally, leukocytes were washed twice (1 ml/each, 1X BD Perm/Wash) and resuspended in BD Pharmingen Stain Buffer (1 ml).

In Vitro Suppression Assays

Human PBMCs were activated with 2 μg/ml PHA for 5 days to generate T lymphoblasts. To favor Treg cell proliferation (2- to 10-fold), culture medium was also supplemented (day 3) with 200 IU/ml IL-2 (Peprotech). CD4+CD25high and CD4+CD25highCD26−/low Treg subsets were purified from T lymphoblasts (Dynabeads® Regulatory CD4+CD25+ T cell Kit; Life-Technologies, Spain) and tested for suppression in co-cultures with autologous CFSE-labeled PBMCs as responder cells (1 × 105/well; 250 μl/well). The suppression assay was performed in round-bottom 96-well plates with a dilution series ranging from 1:1 to 4:1 of responder cells:Treg cells. To induce proliferation, responder cells were stimulated with beads coated with mAbs against CD2/CD3/CD28 (Treg suppression inspector human, Miltenyi Biotec, Auburn, CA; 1 bead per cell) in the presence of 200 IU/mL IL-2 for 4 to 5 days. Additionally, responder cells and Treg cells were cultured separately, either with or without beads/IL-2. Measurements were always carried out in triplicate. Potency of suppression was calculated at 1:1 ratio as [1-(proliferation ofTreg:responders coculture/proliferation of responders alone)].

Statistics

P values were calculated by the nonparametric Mann–Whitney U-test using IBM SPSS statistics 19 software. P values =0.05 were considered to be significant.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED
  8. Supporting Information

As previously reported (29), resting lymphocytes express low levels of CD26. However, CD26 is upregulated during cell activation (plateau phase at 4–6 days; Fig. 1A), and CD26 levels are positively correlated with cell size (forward scatter/FSC measurements) (Fig. 1A). CD26 up regulation takes place especially amongst CD4 T-cells (Fig. 1B), and actively dividing T lymphocytes display higher numbers of CD26 molecules as they progress through new division rounds (Fig. 1C, 5-days PHA blasts, >90% enriched in CD3+ T cells; see CD26 geomean values plotted against the number of cell divisions in line graph). Moreover, as the percentage of IFN-γ+ secreting CD4+ T cells that also stained for CD26 after 5 days of stimulation with PHA is very high (81.5 ± 15.7%; n = 5 donors), this extracellular peptidase allows tracing IFN-γ-producing TH1 cells. Therefore, CD26 can be used, at least, as an activation/TH1 marker.

thumbnail image

Figure 1. CD26 is an activation marker. (A) Human PBMCs, cultured and activated with 2 μg/ml PHA for 5 days in 24-well plates (0.25 × 106 cells/ml), were harvested and immunostained with CD4(SK3)-PerCP and CD26(TP1/16)-FITC. During flow cytometry data acquisition (20,000 events) and analysis, lymphocytes were gated (R1; not shown) according to their size (FSC) and complexity (SSC), three subsets differentiated based on cell size (small, medium, big), and CD26 levels detected in each one of these subpopulations. The discontinuous line indicates the CD26 expression in resting lymphocytes (day 0). (B) CD26 levels in CD4+ and CD4 populations from resting (day 0) and PHA-activated lymphocytes (day 5). (C) CFSE-labeled PBMCs, seeded at 0.25 × 106 cells/ml in 96-round well microplates, were either kept unstimulated (upper left dot plot) or activated during 5 days with 2 μg/ml PHA (upper right dot plot). Then, cells were indirectly labeled with isotype control antibody or mAb against CD26 (TP1/16) plus GAM-APC, and CD26 expression (percentage and geomean) plotted against the number of cell divisions of gated lymphocytes (FSC vs. SSC plot; R1 gate, not shown). Results in (A), (B) and (C) belong to representative experiments. Data in (A) and (C) are shown as mean ± standard deviation (SD) of triplicate measurements.

Download figure to PowerPoint

Treg cells display an “activated-like” (CD4+CD25high CD45RO+) phenotype. Consequently, we set out to determine whether Treg cells at the steady-state expressed CD26. About 5.3 ± 2.2% (n = 10; data not shown) of resting CD4+ T lymphocytes are CD25+ (2, 12), but only those displaying the highest expression of CD25 (CD25high, ∼1–2% CD4+ cells) are actual Treg cells (Fig. 2) (2). In contrast, CD4+ populations with intermediate levels of CD25 (CD25low) contain a mixture of Teff and immature Treg lymphocytes (2). This fact, together with the continuous expression of CD25, makes the boundary for CD25high Treg populations not exempt of arbitrariness (10). However, we observed that CD4+CD25high Treg lymphocytes (R2 gate) were the subset with the lowest percentage of positive cells and the most reduced molecular density for both CD26 and CD127 antigens (Fig. 2). In contrast, high levels of CD127 and CD26 were found amongst CD4+CD25 Teff cells (R4 gate), whereas CD4+ lymphocytes expressing intermediate densities of CD25 (R3 gate) consisted of regulatory (CD26−/lowCD127−/low) and effector (CD26+CD127+) populations (more clearly observed for CD127). We confirmed these results with three different CD26 mAbs clones (TP1/19, TP1/16, and L272), pointing out that this result does not depend on a particular epitope. Thus, when CD26 levels were measured with the TP1/16 mAb (∼20% stronger staining than the TP1/19 mAb in Fig. 2), the following results were obtained: 68.3 ± 12.2% of CD4+CD25high T cells, 76.9 ± 11% of CD4+CD25+, and 91.4 ± 5.3% of CD4+CD25 T cells were CD26+ (n = 10). In addition, intracellular staining of CD4+CD25high Treg cells also revealed low levels of CD26; i.e., there is not internal pool (not shown).

thumbnail image

Figure 2. Resting CD4+CD25high Treg cells display a low or null expression of CD26. Human whole blood was directly labeled with the following fluorescent monoclonal antibodies: CD26(TP1/19)-FITC, CD127(hIL-7R-M21)-PE, CD4(SK3)-PerCP, and CD25(M-A251)-APC. Subsequently, erythrocytes were lysated and samples washed and fixed before acquisition (200,000 events). During data analysis, lymphocytes were gated (R1) according to their forward (FSC) and side (SSC) scatter properties, and three subsets (R2, R3, and R4) selected based on CD4 and CD25 levels. In each one, extracellular expression of both CD26 and CD127 markers was measured as percentage of positive cells (%) and mean fluorescence intensity (geometric mean, Gm). CV, coefficient of variation. Md, median. One representative experiment out of 10 performed is shown.

Download figure to PowerPoint

FoxP3 is still the preferable marker to accurately identify the Treg subset, even though the existence of nonregulatory FoxP3low T cells in normal individuals also precludes the use of FoxP3 as a sole marker for Treg cells (2). For that reason, cell size (FSC), CD25, CD127, and especially, CD26 levels, were evaluated in resting CD4+FoxP3high (R2), CD4+FoxP3low (R3), and CD4+FoxP3 (R4) T cells from healthy individuals (Fig. 3A). Our results clearly showed that, compared with CD4+FoxP3−/low lymphocytes (R3 and R4), CD4+FoxP3high Treg cells (R2) were mainly CD25high leukocytes with an intermediate size and a dim expression of CD4, CD127, and CD26 (82.2–85.2% vs. 46.5% for CD127; 82.9–85.2% vs. 39.4% for CD26). Thus, our findings suggest that CD26−/low levels are linked to Treg cells in resting lymphocyte populations, while high CD26 levels can be found in CD4+ subsets with effector functions, which is in line with data from Mandapathil et al. (24).

thumbnail image

Figure 3. Phenotypic stability of CD4+CD25highCD26−/low Treg cells upon activation. PBMCs were isolated by Ficoll density gradient separation, and either directly labeled (resting, day 0) (A) or activated in complete medium supplemented with 1 μg/ml PHA before immunofluorescent labeling with antibodies (activated, 5 days) (B). In both cases, cells were initially incubated with the following mAbs: CD26-FITC (clone TP1/16), CD4-PerCP (clone SK3), and either CD25-APC (clone M-A251) or CD127-Alexa Fluor 647 (clone hIL-7R-M21). Then, cells were fixed/permeabilized and stained with anti-FoxP3-PE (clone 259D/C7). Background values were set with the corresponding isotype mAbs. During data analysis (WinMDI), the lymphocyte population was selected according to its cell size (FSC) and complexity (SSC) (R1). Three subpopulations of CD4+ T cells were identified as a function of its FoxP3 expression (R2, CD4+FoxP3high Treg; R3, CD4+FoxP3low; R4, CD4+FoxP3), and the level of CD25, CD127, and CD26 as well as the cell size (FSC) quantified afterwards. Numbers on top of histograms represent the percentage of positive cells for each marker (%) as well as the corresponding intensity of fluorescence (geomean; in brackets). Data belong to one representative experiment out of four performed.

Download figure to PowerPoint

However, both CD25 and CD26 antigens also reflect the cell activation status. Consequently, we wanted to exclude the possibility of blurry limits between different T helper subsets for CD26 levels because of cell activation. Thus, data in resting PBMCs were compared with those obtained after subjecting the same cells to 5 days of in vitro polyclonal activation (Fig 3B). As expected, the nonspecific mitogenic stimuli (PHA) caused an intense FoxP3 up regulation in both CD4+ and CD8+ T cells (16), as well as augmented levels of CD25 within the CD4+FoxP3−/low (R3 and R4) and, especially, the CD4+FoxP3high Treg subset (R2 gate) (14). On the contrary, the percentage of CD127+ cells (but not the fluorescence intensity value) was diminished (Fig. 3B) in all CD4+ T subsets (R2-R4) (19, 20). Therefore, our data support the notion that there is not a clear boundary between Treg and Teff lymphocytes under activating conditions when selection is exclusively based on extracellular markers such as CD25 or CD127.

In clear contrast, steady state levels of CD26 were only slightly upregulated (see fluorescence intensity data) by in vitro stimulation in all CD4+ T subsets (Fig 3B), which likely reflects that augmented levels of CD26 are mainly confined to the CD4 cell subset (Fig. 1B). More important, different levels of CD26 expression between FoxP3−/low (Teff) and FoxP3high (Treg) subsets were kept or even magnified after in vitro activation. Similar data were obtained with CD2/CD3/CD28 beads, or when the TH1-polarizing cytokine IL-12 was used as costimulus (not shown). Thus, our results collectively demonstrate that CD26 could be used as additional criterion, in conjunction with CD25 (or FoxP3) and CD127, to distinguish Teff from Treg in both resting and activated populations. Moreover, Supporting Information Figure S1 provides an example where either CD4, CD25, CD26, or cell size (FSC) were used with FoxP3 to segregate both T subsets. As shown, the boundary between both FoxP3−/low and FoxP3high cells was blurred in resting CD4+ T lymphocytes, when just CD4 and FoxP3 levels were considered, whereas the most polarized FoxP3 levels in activated CD4+ T cells allowed a better discrimination. However, the use of complementary Treg features (e.g., CD25high, CD26−/low, or FSCintermediate), in conjunction with CD4 and FoxP3, significantly improved the selection of this subset.

Other currently used Treg selection markers are CD49d (α chain of the VLA-4 integrin) (21) and CD39 (22). Expression of CD25 and CD26 antigens was compared with CD49d levels on Treg cells (CD4+FoxP3high; R2) and Teff lymphocytes (FoxP3−/low; R3) under resting (not shown) and activating conditions (Fig. 4A). We found marked differences for CD25 and CD26 antigens, but not for CD49d. Consequently, our results fit with previously published data reporting the presence of CD49dhigh subsets among Treg cells (22, 23). Of note, CD26 also was better than CD49d as negative selection criterion to discriminate IFN-γ-producing cells within the CD4+CD25high compartment in PHA-activated PBMCs (Supporting Information Fig. S2).

thumbnail image

Figure 4. Levels of CD26, CD49d and CD39 antigens in Treg (FoxP3high) and Teff (FoxP3low) lymphocytes. PBMCs were purified and expanded with PHA during 5 days before inmunofluorescent labeling. Cells were extracellularly stained with CD4-PerCP (cloneSK3) and different combinations of CD26-FITC (clone TP1/16), CD39-FITC (clone TU66), CD25-APC (clone M-A251), CD26-APC (clone TP1/19), and CD49d-APC (clone 9F10) mAbs. After that, cells were fixed/permeabilized and stained with anti-FoxP3-PE (clone 259D/C7). During analysis, lymphocytes were gated (size vs. complexity plots; R1) and two subpopulations (R2 and R3) distinguished based on CD4 and FoxP3 expression: CD4+FoxP3+/high Treg cells (R2; depicted as grey color in A) and CD4+FoxP3low/− Teff lymphocytes (R3; depicted as black color in A). Finally, either plasma membrane levels of CD25, CD26, and CD49d, each one in combination with FoxP3 (A), or both CD39 and CD26 antigens (B), were evaluated. The figure shows one representative experiment out of three performed.

Download figure to PowerPoint

Unlike CD49d, CD39 is a marker with a clearly divergent expression pattern for Treg (R2; ∼72% CD39+) and Teff (R3; ∼3% CD39+) cells cultured under activating conditions (Fig. 4B), but not at the steady state (data not shown; Treg, ∼3% CD39+; Teff ∼0.1% CD39+). Moreover, CD39 is linked to adenosine production (22), an immunosuppressive molecule whose accumulation is also favored by the CD26−/low phenotype of activated CD39+ Treg cells (Fig 4B), since the CD26 protease is the main adenosine deaminase (ADA) binding protein (28).

According to Kleinewietfeld et al. (21), antibodies against CD49d and CD127 alone are sufficient to isolate “untouched” FoxP3+ Treg cells free of contaminating CD25+ Teff cells. To determine whether the combined use of these two markers is enough to accurately discriminate Treg cells from activated Teff cells, “untouched” CD4+CD127lowCD49d Treg lymphocytes were purified from resting or in vitro expanded PBMCs using the EasySep™ Human CD4+CD127lowCD49d Regulatory T Cell Enrichment Kit (StemCell Technologies). Treg cells isolated from resting PBMCs were a homogeneous population (see CD4 and FoxP3 levels), mostly negative/low for CD127 and CD49d (as expected) and with a homogeneous CD26−/low phenotype (see R2 and R3 gates in Fig. 5A). On the contrary, the CD4+CD127lowCD49d cell fraction purified from activated PBMCs was heterogeneous, whether regarding the level of CD4 and FoxP3 molecules or according to CD26 levels (see R2 and R3 gates in Fig. 5B). This heterogeneity was caused by a preferential isolation of a FoxP3low subset (R3), which displays a higher CD26 expression than the FoxP3high subset (R2) (Fig. 5B). Due to the activation-dependent reduction of CD127+ cells within the Teff subset (Fig. 3), we speculate that this FoxP3lowCD26+ phenotype (R3) may be the signature of T cells with unstable FoxP3 expression, activated-memory phenotype, and the capability of producing IL-2, IFN-γ or IL-17 (35, 36).

thumbnail image

Figure 5. The combined use of CD127 and CD49d does not avoid the presence of FoxP3lowCD26high T cells within Treg populations isolated from activated PBMCs. Regulatory T cells were purified from resting (A) or activated PBMCs (1 μg/ml PHA, 5 days) using the EasySep™ Human CD4+CD127lowCD49d Regulatory T Cell Enrichment Kit (StemCell Technologies). Treg cells were stained with CD26-FITC (clone TP1/16), anti-FoxP3-PE (clone 259D/C7), CD4-PerCP (clone SK3), and CD49d-APC/CD127-AlexaFluor 647 (Clone 9F10/hIL-7R-M21) mAbs. During analysis (A and B), Treg lymphocytes were gated (forward vs. side scatter; R1) and two subpopulations (R2 and R3) distinguished based on the expression of CD4 and FoxP3: CD4+FoxP3high (R2) and CD4+FoxP3low (R3). Finally, both CD26 and CD127/CD49d levels were evaluated for R2 and R3 in new density plots, reflecting that Treg lymphocytes purified from PBMCs (A) are a more homogeneous population as compared with the Treg population isolated from activated PBMCs (B). A high mortality rate has been constantly observed for Treg cells purified with this protocol (see forward vs. side scatter plot is in A and B). One representative experiment out of three performed is shown.

Download figure to PowerPoint

According to our findings, inclusion of CD26-specific mAbs in Treg purification protocols may help to deplete contaminating FoxP3low T cells and allow the isolation of highly enriched FoxP3high cells, especially from previously expanded “bulk” T-cell populations or samples from inflammatory/autoimmune diseases. Most of commercially available protocols are based on the CD4+CD25high phenotype. According to our data in resting T lymphocytes, 81.9 ± 11.2% (n = 10) of CD4+CD25+ T cells were CD26. Thus, unlike the use of CD45RA to isolate “naive” CD4+CD25high Treg cells (24–61% of resting Treg cells) (37), the CD26 strategy should yield higher Treg numbers (38). As a first approach, we determined by means of a pilot flow cytometry experiment whether the strategy of selecting the CD4+CD25highCD26−/low yielded more homogeneous Treg populations (according to cell size and FoxP3; Supporting Information Fig. S3) than the usual CD4+CD25high phenotype. We used a broader gating strategy and a lower than usual CD25 boundary. CD4+CD25 T cells were also gated for comparison, revealing that ∼95% of them (R4 gate) were FoxP3−/low cells. However, FoxP3low cells (40–60%) were also detected amongst the CD4+CD25high Treg lymphocytes (R2 + R3 gates). In contrast, the use of a CD4+CD25highCD26 phenotype (R2 gate) to select the Treg subset yielded a more homogeneous population, as the percentage of FoxP3high cells reveals (in Supporting Information Fig. S3, 43%[RIGHTWARDS ARROW]61% for resting PBMCs, and 59[RIGHTWARDS ARROW]75% for activated PBMCs).

Given these preliminary data, both CD4+CD25high Treg and CD4+CD25 Teff lymphocytes were isolated from either freshly isolated or PHA-expanded PBMCs by means of the Dynabeads® Regulatory CD4+CD25+ T cell Kit (Life Technologies/Invitrogen). As expected, the starting percentage of Treg cells amongst resting PBMCs was low (0.84 ± 0.37%; n = 6), recovering ∼86% (n = 3) of these Treg cells as CD4+CD25high fractions with a cell purity of 87% (assessed by CD25 expression; n = 3). In contrast, we have detected a slightly higher starting percentage of CD4+CD25high cells amongst PHA-activated PBMCs (3.34 ± 2.08%, n = 6). Only 68% (n = 7) of these lymphocytes was collected within CD4+CD25high fractions. Moreover, the percentage of positive cells for CD25 and FoxP3 was, respectively, 83.6 ± 5.14% and 63.63 ± 9.12% (n = 3), which can be explained because of the presence of activated CD25+ Teff cells.

To improve the purity of Treg cells obtained from PHA-expanded PBMCs, we partially removed CD26+ cells from CD4+ T lymphocytes by means of the addition of TP1/16 mAb to the cocktail of mouse IgG mAbs during the first step. Consequently, a significant decrease in the number of CD4+ T cells was noted: 27.39 ± 8.93 [RIGHTWARDS ARROW]18.67 ± 5.53 (× 106) CD4+ T cells, 28.84 ± 5.47% reduction (n = 7). Once purification was completed, flow cytometric analysis revealed a low expression of CD26 in CD4+CD25 Teff cells and a bimodal pattern for the CD4+CD25high Treg subset, which comprises CD26−/low and CD26high subtypes (Fig. 6A). The use of mAbs against CD26 caused an almost full depletion of CD26+ cells in CD4+CD25 Teffs, but only a partial decline amongst CD4+CD25high Tregs (∼40%; Figs. 6B and 6C). However, despite this incomplete CD26 depletion, the percentage of positive cells for CD25 (83.6 ± 5.14% [RIGHTWARDS ARROW] 91.2 ± 5.05%; n = 3) and FoxP3 (63.63 ± 9.12% [RIGHTWARDS ARROW] 75.33 ± 10.91%; n = 3) (Fig. 6C) was raised, as well as the percentage of CD25highFoxP3high cells (e.g., from 66.8 to 84.5% in Figs. 6A and 6B). Moreover, there was a significant rise in the number of CD127+ Treg cells after CD26-depletion (8.9 ± 9.2% CD127+ Treg [RIGHTWARDS ARROW] 29.7 ± 0.8% CD127+ Treg, n = 2 donors), which fits with a higher suppressive activity (Fig. 6D; 8.29 ± 2.17% [RIGHTWARDS ARROW] 17.98 ± 1.93%).

thumbnail image

Figure 6. The use of CD26 as a negative selection marker enhances the levels of FoxP3 and CD25 in CD4+CD25high Treg populations isolated from activated PBMCs. CD4+CD25high Treg cells were purified from activated PBMCs (1 μg/ml PHA, 5 days) using the Dynabeads® Regulatory CD4+CD25+ T Cell Kit (Life Technologies/Invitrogen). During the depletion of non-CD4+ cells, the cocktail of monoclonal antibodies was supplemented (B) or not (A) with CD26 (TP1/16) mAb in order to deplete CD26+ cells. Isolated CD4+CD25high Treg and CD4+CD25 Teff cells were stained with CD26-FITC (TP1/16), anti-FoxP3-PE (259D/C7), CD4-PerCP (SK3), and CD25-APC (M-A251) mAbs. During analysis, viable CD4+ lymphocytes were selected (R1 and R2) and FoxP3/CD25 levels and CD26 expression evaluated. According to CD26, two different subsets (CD26−/low and CD26high) were observed within CD4+CD25high Treg cells (A), which points out to the presence of contaminating activated Teff cells (CD26high). Reduction of this CD26high subset within the CD4+CD25high Treg lymphocytes raised the percentage of CD25/FoxP3 double positive cells (B). (C) Columns plot summarizes the percentage of reduction/increase in the fluorescence intensity of CD26, FoxP3, and CD25 in Tregs from three different healthy donors after the partial depletion of CD26+ cells (mean ± standard deviation). (D) Suppression of autologous (CFSE-labeled) responder PBMCs proliferation (y-axis; % of cells that divided at least once) by coculture with increasing numbers of either CD4+CD25high (Treg) or CD4+CD25highCD26−/low (CD26-depleted Treg) lymphocytes isolated from preactivated (PHA + IL-2; 5 days) PBMCs (x-axis). Measurements were carried out in triplicate. Significant results are indicated by single asterisks (P value is =0.05; Mann-Whitney U-test).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED
  8. Supporting Information

CD26 is an activation/memory marker capable of transmitting costimulatory signals (28, 30). Thus, interaction between CD26 (T cell) and caveolin-1 (APCs) leads to NF-κB activation and recruitment of CARMA1/Bcl10/IκB (28, 33), a complex implied in thymic development of Treg cells (39), while blockade of this costimulatory interaction causes T cell cycle arrest (40, 41). Nevertheless, CD26 may also have inhibitory functions. For example, disease severity is increased in CD26−/− mice in certain autoimmune models (42, 43), with augmented T-cell proliferation rates (43) and production of TH1 cytokines (IFN-γ, TNF-α) (42, 43) but diminished TGF-β1 synthesis (43). Indeed, several inhibitory peptides or naturally occurring ligands of CD26/DPPIV promote the secretion of TGF-β1 (43-45). Therefore, all these findings point to some kind of connection between CD26 and the Treg phenotype that deserves to be ascertained, especially because Treg lymphocytes, with an “activated-like” (e.g., CD25high, HLA-DR+) and “memory” (CD45RO+) phenotype, were expected to be CD26+ (46, 47) but not CD26−/low cells. However, our results are in agreement with the CD26−/low profile of Treg-like leukocytes in classical Hodgkin's lymphoma (48). Moreover, they fit with recent studies in human Treg cells (24), the latest in April this year (49), where authors describe variable CD26 levels in different TH subsets: TH17 > TH1 > TH2 > Treg cells.

Nevertheless, our data go beyond, as they show that this CD26−/low phenotype is stable upon in vitro T-cell expansion and can be used during the isolation/identification of Treg cells. In addition, this CD26−/low phenotype may explain why the TH1 cytokine IL-12 causes a strong upregulation of CD26 on “bulk” T-cell cultures, whereas IL-2 (CD25-ligand) induces only a slight increase (29). Still, several questions remain unanswered: for example, whether this CD26−/low phenotype is based on differential transcription/translation, shedding, or exosomes/microvesicles releasing processes, whether CD26 level on Tregs depends on their tissue localization (50), or how the various Treg/Teff cells ratios found in different diseases correlate with levels of circulating sCD26 (51).

Researchers can only get high numbers of Treg cells by means of tedious two-step magnetic procedures based on CD25 (35-38). However, CD25 is also expressed by FoxP3low T cells, a group of lymphocytes that might fit with naïve or uncommitted Tregs but also activated Teff cells (2, 37, 52, 53). For that reason, more efficient technologies have been placed on the research market. Nevertheless, these new protocols may not be entirely effective yet, as our article points out. Thus, negative selection antigens, like CD127 (17, 18) or CD49d (21), certainly provide an additional way of ascertaining the right phenotype. Nevertheless, this does not seem to be sufficient because of the intense reduction in CD127+ cells upon TH-cell activation (19,20,54, our data), the presence of CD127high subsets within activated Treg cells (50, our data), or the small differences for CD49d between FoxP3high and FoxP3−/low cells (our data).

The limited amount of biological samples or the low abundance of Treg cells in peripheral blood are additional problems. On this regard, several protocols have been developed for the in vitro expansion of Treg cells (35-38), but many of them are associated with potential risks: e.g., the in vitro outgrowth of contaminating Teff cells or the possibility of “plastic” Treg cells reverting to FoxP3low Teff cells (2, 35, 37). To circumvent these problems, CD45RA have been promoted as a discriminating marker to identify a subset of naive CD4+CD25high Treg cells that maintain FoxP3 expression during in vitro culture (37). However, it remains to be determined if there is a sufficient number of such “naive” Treg cells in blood samples of patients for in vitro expansion and autologous adoptive transfer (37), especially considering the age-dependent differentiation from naive to effector memory Treg cells (55). Moreover, it is known that even CD45RAhigh Treg cells can give rise to nonregulatory T cells when cultured in a milieu containing high-dose IL-2 (2, 35, 37). In clear contrast, our study suggests using a different strategy. This new approach consist in the in vitro expansion of bulk lymphocytes before isolating the Treg subset on the basis of a more reliable Treg/Teff differentiation marker: CD26. Thus, antibodies against this peptidase could be included during the first negative selection step of any magnetic protocol to get rid of unwanted CD26+/high Teff (TH1, TH2, TH17) lymphocytes and achieve higher numbers of either CD4+CD25high or CD4+CD39high Treg populations (our data, 24,49). Moreover, the phenotype CD4+CD127lowCD49dCD26 may be also enough to select Treg cells through a fast and activation-degree independent “single-step” purification protocol.

Which functional explanation there might be for this CD26−/low phenotype in human Treg cells? Several ligands have been found for CD26, as the tyrosine phosphatase CD45RO (31, 32) or adenosine deaminase (ADA). The last one is an ectoenzyme involved in degradation of adenosine, which is an immunosuppressive molecule (22, 56). Consequently, a low level of CD26 could alter the membrane location of CD45RO (32) or reduce the amount of surface ADA on Treg cells (57), which may explain either their anergic phenotype in vitro or their net production of adenosine (22). Alternatively, interaction of Caveolin-1 (APC) with CD26 (T-cell) leads to the up regulation of CD86 on APCs (58). Interestingly, Treg lymphocytes down-regulate CD80 and CD86 on APCs through a CTLA-4-dependent (9) mechanism. CD80/CD86 molecules are differentially regulated (59) and seem to have divergent functions (60). For example, CD86 is highly responsive to proinflammatory stimuli (LPS, TNFα, IFN-γ, CD40-CD40L ligation) and shows a preference for CD28, whereas CD80 seems rather specific for the inhibitory receptor CTLA-4 (60, 61). Therefore, if we assume a co-stimulatory role for the CD26/caveolin-1 interaction and a subsequent CD86 (but not CD80) upregulation on APCs (58), a CD26−/low phenotype in Treg cells may contribute to keep low levels of proinflammatory CD86 molecules on APCs, leading indirectly to a lack of response (anergy) or apoptosis in antigen-primed naïve T cells.

In conclusion, the present data substantiate a link between null/low levels of CD26 and immunosuppressive functions within either CD4+CD25high, CD4+CD39high, or CD4+CD127−/lowCD49d subsets. In addition, they support further investigation to determine whether the CD26−/low phenotype is a reliable alternative for the delineation or purification of human Treg cells by means of extracellular markers, which may stimulate the arising of new protocols and the progress in Treg field. Finally, a patent application for the data described in this paper was done on January 26, 2010, which has recently entered the Patent Cooperative Treaty (PCT) phase.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED
  8. Supporting Information

The authors are grateful to the Medical Service at the University of Santiago, and to Dr. Juan E. Viñuela, Immunology Service, Clinic University Hospital of Santiago (CHUS), for technical assistance. In addition, the authors want to thank Ana M. Carballido and Alejandro González for their collaboration.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
CYTO_22117_sm_SuppFig1.tif5222KSupporting Information Figure 1.
CYTO_22117_sm_SuppFig2.tif2913KSupporting Information Figure 2.
CYTO_22117_sm_SuppFig3.tif4497KSupporting Information Figure 3.
CYTO_22117_sm_SuppTab1.tif106KSupporting Information Table 1.
CYTO_22117_sm_SuppTab2.tif112KSupporting Information Table 2.
MIFlowCyt-Item-Location.doc106KSupporting Information: MIFlowCyt

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