Selective depletion of FOXP3high cells by Fas–Fas-L–induced apoptosis occurs in CD4+CD25+-enriched populations during repeated expansion

Cytotherapy, 2013; 15: 1286e1296

Selective depletion of FOXP3high cells by FaseFas-Leinduced apoptosis occurs in CD4DCD25D-enriched populations during repeated expansion

WEI ZHANG1,2, SINDU NAIR1,2, ROBERT DANBY1,2,3, ANDY PENIKET3 & DAVID J. ROBERTS1,2,3 1

Division of Clinical Laboratory Sciences, Radcliffe Department of Medicine, John Radcliffe Hospital, Oxford, United Kingdom, 2NHS Blood and Transplant Oxford Centre, John Radcliffe Hospital, Oxford, United Kingdom, and 3Cancer and Haematology Centre, Churchill Hospital, Oxford, United Kingdom Abstract Background aims. Expansion of anti-CD25 bead-isolated human Tregs culture has paradoxically resulted in reduced suppressive activity, but the mechanism(s) responsible for these observations are poorly defined. Methods. Magnetic-bead isolated human CD25þ cells were expanded with anti-CD3/CD28 beads and high doses of rhIL-2. Detection of Fas and Fas ligand (Fas-L) expression, activation of Caspase 8, cell proliferation and cytokine production was evaluated by multi-color fluorescence-activated cell sorting analysis. The role of FaseFas-Lemediated cell death was dissected through the use of agonist or antagonist monoclonal antibodies directed at Fas and Fas-L. Results. Repeated expansion of bead-enriched CD4þCD25þ cells generated a cellular product with markedly reduced suppressive activity and with significantly increased CD8þ T cells and CD4þ T cells producing interferon-g and/or interleukin-2. We showed that FaseFas-Lemediated apoptosis of CD4þFOXP3high cells and rapid cell-cycling of CD8þ T cells were collectively responsible for the reduced proportion of CD4þFOXP3high cells in expanded cultures. The depletion of CD4þFOXP3high cells and activation of Caspase 8 in CD4þFOXP3high cells was attenuated by Fas antagonist antibody, ZB4, in short-term culture. However, the loss of CD4þFOXP3high cells during expansion was not prevented by either Fas or Fas-L antagonist antibodies. Conclusions. Taken together, the data show that FaseFas-Lemediated apoptosis may limit the expansion of anti-CD25 bead-isolated cells in vitro. Key Words: apoptosis, cell therapy, expansion, Fas, FOXP3, Treg

Introduction The seminal work of Sakaguchi et al. (1) established the role of regulatory T cells (Tregs) in maintaining immunological self-tolerance. The ability of these cells to modulate immune responses has suggested the possibility of the use of isolated and/or expanded Tregs for the prevention or therapy of autoimmune or alloimmune diseases (2e4). In murine models, Tregs effectively suppressed undesirable immune responses in autoimmune diseases, abrogated stem cell transplant related graft-versus-host disease and prevented transplant arteriosclerosis in a humanized mouse system (5e8). By introducing additional specific markers, such as CD45RA, or CD127, a highly purified population of human Tregs have been isolated by fluorescence-activated cell sorting (FACS) and reliably expanded in vitro by means of

polyclonal TCR stimulation with exogenous interleukin (IL)-2 (3,9). These cells can maintain their suppressive function in vitro, and when transfused to patients can ameliorate steroid-resistant severe acute graft-versus-host disease. Alternatively, less homogeneous populations of Tregs have been isolated through the use of current, widely available good manufacturing practiceecompliant bead technology by first depleting non-CD4 T cells, then CD25-positive selection, followed by expansion in the presence of rapamycin (10). The cost of these in vitro manipulations is one limiting factor for Tregs to be generated and tested in more clinical trials. If one-step CD25-bead isolated Treg populations are to be manipulated in vitro for clinical use, it is crucial to understand the mechanisms underlying

Correspondence: Wei Zhang, MD, PhD, Division of Clinical Laboratory Sciences, Radcliffe Department of Medicine, Level 4, Academic Block, John Radcliffe Hospital, Headley Way, Oxford, OX3 9DU, UK. E-mail: [email protected] (Received 14 March 2013; accepted 28 May 2013) ISSN 1465-3249 Copyright Ó 2013, International Society for Cellular Therapy. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcyt.2013.05.021

Fas/Fas-L and human regulatory T-cell apoptosis the failure to sustain suppressor function during their expansion. Fas-Fas ligand (Fas-L) interactions mediate apoptosis and play a fundamental role in immune homeostasis in mammals (11). Genetic defects in Fas or Fas-L expression or downstream signalling lead to autoimmune lymphoproliferative syndromes (12,13). Fas is readily detected on 50% of T cells from ex vivo human peripheral blood. By contrast the expression of Fas-L is tightly regulated in the immune system through multiple mechanisms including transcriptional regulation, intracellular storage and metalloproteinase cleavage. Consequently, human T lymphocytes only transiently express surface Fas-L during TCR stimulation, when it becomes the main mediator of activation-induced cell death (14e17). The activation state of T cells can influence their survival by modulating the levels of Fas expression. For example, activated human memory CD4þ T cells are more sensitive to Fas-Fas-Leinduced apoptosis than are naive CD4þ T cells (14). Furthermore, human CD8þ and CD4þ T cells show substantial differences in the potential consequences of Fas/ Fas-L signaling. Fas-L can deliver “reverse” costimulatory signals to CD8þ T cells, thereby enhancing their proliferation (18). Human CD4þFOXP3þ Tregs from peripheral blood can be further divided into CD45RA FOXP3higheactivated Tregs and CD45RAþFOXP3low naive/resting Tregs. It has been observed that activated Tregs die rapidly, whereas resting Tregs proliferated and converted into activated Tregs in vitro and in vivo (19). Fritzsching et al. (20) have previously reported that ex vivo human CD25þ Tregs are sensitive to FasLemediated apoptosis; their sensitivity correlates with the levels of CD25. Furthermore, Tregs undergoing homeostatic expansion showed increased susceptibility to Fas-mediated apoptosis (21). We hypothesised that Fas/Fas-L interaction during expansion of antie CD25ebead-isolated cells, which are enriched for Tregs, causes the selective loss of FOXP3high activated Tregs, which may explain, at least in part, the decline in Tregs function during in vitro expansion. To test our hypothesis, that cells resistant to Fas/ Fas-Leinduced cell death may have survival advantage over mature/activated Tregs during in vitro expansion, we isolated and expanded Tregs from PBMCs and showed that a subpopulation of CD4þFOXP3high cells expressing the highest levels FOXP3, also expressed high levels of Fas and is highly sensitive to Fas-Fas-Lemediated killing. Selective loss of these CD4þFOXP3high cells, concomitant outgrowth of CD8þ T cells and persistent and/ or development of cytokine-producing helper T cells are collectively responsible for the decline in suppressive function of CD25 beadeisolated cells after

1287

repeated expansion. These data underline the importance of the use of unmanipulated beadisolated CD4þCD25þ cells for therapy or expanded highly purified CD4þCD25þ cells in appropriate clinical settings. Methods Ethics statement Buffy coats were used from whole blood donations taken from UK blood donors who gave informed consent and with ethical permission from National Health Service Blood and Transplant (NHSBT, Bristol, UK; Research Ethics Committee reference No. 08/H0607/41). CD25-positive cell and other cell isolation PBMCs were isolated from 24-to 48-h-old buffy coats by centrifugation on Ficoll-Paque (GE Healthcare, Hatfield, UK). CD25þ cells were isolated directly, without CD4þ T-cell enrichment, by incubation of PBMC with 5 mL of antieCD25coated microbeads per 107 cells for 15 min at 4 C (Miltenyi Biotech, Surrey, UK). After one wash in incubation buffer, the cell suspension (1  108/mL) was passed through a 30-mm sieve before loading onto the LS column attached to the VarioMACs magnet (Miltenyi Biotech). CD25þ cells remained attached to the column, which was washed three times with 3 mL of buffer to remove non-bound cells. After the final wash, the column was removed from the magnetic field and the positively selected cells were flushed out. CD25þ cells were then passed two or three more times over fresh columns as described above to enrich the cells with the highest levels of CD25. Positively selected cells from the first, second and third columns were termed CD25Sel-1, CD25Sel-2 and CD25Sel-3 cells, respectively. In vitro expansions were performed with the use of the CD25Sel-2 cells as described below. The yield of enriched population was quantified as [(number of positively selected cells)  (% cells in the lymphocyte gate)  (% of CD4þ cells) (% of CD4þ cells positive for FOXP3)]/[(total starting cell number)  (% cells in the lymphocyte gate)  (% of CD4þ cells)  (% of CD4þ cells positive for FOXP3)]  100%. CD8þ T cells and naive CD4þ (nCD4þ) T cells were isolated with reagents from Miltenyi, following the manufacturer’s instructions. Cell culture Cells were cultured as previously described (6). All reagents were obtained from Sigma, Poole, UK,

1288

W. Zhang et al.

unless otherwise stated. In brief, cells were cultured in Roswell Park Memorial Institute 1640 media supplemented with 2 mmol/L L-glutamine, 100 units/mL of penicillin, 100 mg/mL of streptomycin and 1 mmol/L sodium pyruvate; 10% heat-inactivated pooled human AB serum was obtained from NHSBT, Bristol, UK. For expansion, CD25Sel-2 cells were stimulated with anti-CD3/anti-CD28 beads (Invitrogen, Paisley, UK) at a one bead per cell plus recombinant human (rh)IL-2 (1000 U/mL) (Chiron/Novatis, Surrey, UK). Anti-CD3/anti-CD28 beads were removed with a magnet (Invitrogen) on day 5, and the cultures were maintained in medium supplemented with rhIL-2 (1000 U/mL) for 2 more days. On day 7, aliquots of cells were taken for suppression assays (and incubated in medium plus 200 U/mL of rhIL-2 for 1 more day), intracellular cytokine staining and further rounds of expansion as described above. In vitro suppression assays The suppressive activity of the CD25Sel-2 population was tested ex vivo after each round of expansion over a range of responder:regulatory cell ratios by measuring their ability to prevent proliferative responses of cryopreserved autologous PBMC (5  104) stimulated with irradiated allogeneic PBMC (5  104), which were from the same donor for a given responder. Proliferation was measured after 5 days by addition of 3H-thymidine for the last 16 hours of culture, before harvesting and counting on a flat-bed liquid scintillation counter. Assays were performed in triplicate.

a 1% overlap with the levels in these activated CD8þ T cells in the same cultures; those overlapping with the remaining 99% were termed CD4þFOXP3þ cells. In the CD8-spiked cultures, cells were stained for CD3, CD4, CD8, Fas and FOXP3 or CD3, CD4, CD8, Fas-L and FOXP3 at the time points indicated. CD3þCD4þ were compared with CD3þCD8þ cells for expression of Fas and Fas-L on the cell surface. For cytokine stimulation, CD25Sel2 was stimulated for 4 hours with 50 ng/mL phorbol myristate acetate, 1 mmol/L ionomycin and 5 mg/mL Brefeldin A. Brefeldin A was omitted for negative controls. Samples were stained for surface markers before they were fixed and permeabilized for intracellular staining of FOXP3 and cytokines IL-2 and interferon (IFN)g (Becton Dickinson), as indicated. The FOXP3 fix/perm buffer kit from Miltenyi was used for all intracellular staining.

Detecting apoptotic cells Staining for fluorescent labeled inhibitor of caspases (FLICA) was performed with the use of the FAM Caspase 8 assay kit (Invitrogen). In brief, cells were incubated with the FLICA staining reagent for the last hour of the 24-h incubation at a 1:30 final concentration in the dark at 37 C, washed once with FLICA washing buffer and once with PBS. Dead cells were excluded by counter-staining with a Live/ Dead Fixable Staining Kit (Invitrogen), which uses an amine-reactive fluorescent dye to evaluate mammalian cell viability by flow cytometry (22).

Immunophenotyping and cytokine staining

Induction and prevention of apoptosis in CD4þFOXP3high T cells

Two monoclonal antibody (mAb) panels were used in a six-color FACS analysis. Panel I included mAbs against CD3, CD4, CD8, CD25 (M-A251), CD127 (Becton Dickinson, Oxford, UK) and FOXP3 (3G3) (Miltenyi Biotech). Panel II consisted of mAbs against CD3, CD4, CD8, CD16, CD20 and CD56 (Becton Dickinson). For PBMC, at least 5  105 events were acquired for each sample and 2e5  104 for CD25-selected cells. Data were analyzed with FACSDiva 5 or 6 software (Becton Dickinson). Samples stained with fluorochrome-conjugated mAb minus 1 for CD25 and CD127 were used to set positive and negative gates. The CD3 population was used as the internal negative control for intracellular FOXP3 staining, and its threshold for positivity was set at the upper 1% of staining on the CD3 cells. A proportion of CD8þ T cells also upregulated FOXP3 after CD25Sel-2 expansion. The FOXP3high cut-off in the CD4þ cells was set as

T cells and CD3/CD28 beads stick strongly to each other, and depleting CD3/CD28 beads during early days of stimulation often resulted in cell loss. For these experiments, CD25Sel-2 cells from the first expansion were stimulated with 1 mg/mL of plate-bound anti-CD3 and 0.5 mg/mL of soluble anti-CD28 (Biolengend/Cambridge Biosciences, Cambridge, UK) in complete culture media and 1000 U/mL of rhIL-2, as indicated. To induce apoptosis, cells were incubated for 2 h with the agonist anti-Fas immunoglobulin (Ig)M mAb CH-11 (MBL/CaltagMedSystems Ltd, Claydon, UK) or control mouse IgM (Serotec, Kidlington, UK) over a range of concentrations between 0.125 to 1 mg/mL. To prevent activation-induced cell death, cells were incubated for 24 h with the blocking IgG1 mAb ZB4 (MBL) or control mouse IgG1. Cells were harvested and stained for CD3/CD4/CD8/FOXP3/FLICA and with the Live/Dead Fixable Staining Kit as described above.

1289

All analyses were performed with the use of GraphPad Prism 4 (Heame Scientific Software, Kilkenny, Ireland). The two-way analysis of variance test was used to compare three experimental groups. Results Isolation of Tregs We enriched Tregs by capturing them on antieCD25coated magnetic beads and derived populations of CD25-enriched T cells after one, two or three rounds of sequential selection through the use of LS and MS columns, named CD25Sel-1, CD25Sel-2 and CD25Sel-3, respectively. Successive CD25-column selections increased the frequency of CD4þCD25þ FOXP3þ cells with concomitant decreases of both CD8þ cells and CD4þCD127þ cells (Supplementary Table IA). Over three rounds of selection, this method enriched the proportion of CD4þFOXP3þ cells of CD3þCD4þ T cells of PBMC from 3.6%  0.6% to 66.5%  4.1%, with 7%  0.5% Treg recovery in the CD25Sel-2 population. The purity increased to 79.4%  1.1%, but the yield fell to 4.5%  1.3% after the third round of selection (Supplementary Table IA). We used CD25Sel-2 for further experiments as a compromise between purity and yield. Some CD3 cell subsets persisted, notably CD20þ B cells and the CD16þCD56þ natural killer cells and an unidentified CD3CD16CD20CD56 population (Supplementary Table IA). We analyzed expression of CD127 and FOXP3 on the CD3þCD4þ cells and noted progressive enrichment of CD127FOXP3þ cells and concomitant depletion of CD127þFOXP3 cells during successive rounds of selections, as expected (Figure 1). The median fluorescent intensity of CD25 was similar CD127FOXP3 and on CD127þFOXP3þ,  þ þ CD127 FOXP3 CD4 T cells (1145  212.7, 924  86.2 and 1349  196.1). It was significantly lower on the CD127þFOXP3cells (491.6  77.91), consistent with the depletion of this population during successive rounds of selection. Expansion of enriched Tregs and their subsequent phenotype(s) We used the CD25Sel-2 population for further expansion with anti-CD3/CD28 beads. Treg numbers were expanded 200-fold after two rounds of expansion: starting from 5  105 isolated CD25Sel-2 to 5.7  106  2.1  106 (mean  standard error of the mean) after one round of expansion and from 5  105 to 10.7  106  0.9  106 (mean  standard error of the mean) after the second expansion.

100

0

80 50

60 40

100

20 0

150 PB M

C

CD 25 Se l-1

CD 25 Se l-2

Target cell recovery (%)

Statistical analyses

% of cells in the CD4+ gate

Fas/Fas-L and human regulatory T-cell apoptosis

CD 25 Se l-3

Figure 1. Effect of consecutive column isolation on Treg purity and Treg cell recovery. Percentages are given of the four subsets of cells defined by CD127 and FOXP3 staining in the CD4 gate of PBMCs and their changes after sequential selection with the use of anti-CD25 beads and their respective target cell recovery. The four subsets of cells are CD127þFOXP3 (-), CD127þFOXP3þ (:), CD127eFOXP3 (A), CD127eFOXP3þ (C), cell recovery of putative Tregs (V). PBMC (n ¼ 21), CD25Sel-1 (n ¼ 14), CD25Sel-2 (n ¼ 12), CD25Sel-3 (n ¼ 3), recovery of putative Tregs (n ¼ 3e12). Data are shown as mean  standard error of the mean.

The proportion of CD4þ T cells increased after one round of expansion but then declined significantly (percentage of CD3þCD4þ at ex vivo and after first and second expansions were 89.8  3, 93.5  1.8 and 78.7  3.9; first expansion versus second expansion, P < 0.005) (Figure 2A and Supplementary Table IB). The CD3þCD8þ lymphocytes remained unchanged after one round of expansion but increased significantly after two rounds to greater than that in the original unexpanded population (Figure 2A) (percentage of CD3þCD8þ at ex vivo and after first and second expansions were 3.5  1.2, 2.6  0.7 and 10.2  2.2) (P < 0.05) (Figure 2A). Over two rounds of expansion, the proportion of CD3CD20 null cells rose, whereas the proportion of both B cells and especially natural killer cells declined (Figure 2B and Supplementary Table IB). Conventional T cells are co-purified and expanded in Treg-expanded cultures Some IL-2- and/or IFNg-producing cells were co-purified with the putative FOXP3þIL-2 or FOXP3þIFNg Tregs (Figure 3A,B). The frequency of IL-2eproducing and/or IFNg-producing cells, as well as the IL-2 and/or IFNg and FOXP3 co-expressing cells, rose steadily after each round of expansion. In contrast, the frequency of FOXP3þIL-2 declined significantly by the second round of expansion (from 78.1%  2.1% at ex vivo to 68.3%  1.8% after first and to 46.2%  1.3% after second expansion) (ex vivo versus second, P < 0.001; first versus second, P < 0.05), and FOXP3þIFNg cells declined significantly (from 77.4%  2.5% at

W. Zhang et al.

A

B

% of cells in the CD3+ gate

100

P<0.005

75 50 25

P<0.05

0

% of cells in the CD3- gate

1290

P<0.001

100 75

P<0.001 P<0.001

50 25 0

CD20+

CD3+CD8+

CD3+CD4+

P<0.001

CD20-

Figure 2. Frequencies of CD3þCD4þ, CD3þCD8þ, CD3CD20þ and CD3CD20 cells in the CD25Sel-2 before and after each expansion. (A) CD4þ and CD8þ T cells within the CD3þ gate; ex vivo, open bar, n ¼ 10; first expansion (gray), n ¼ 9; second expansion (black), n ¼ 8, mean  standard error of the mean. (B) CD20þ and CD20 cells within the CD3 gate; ex vivo (open), n ¼ 3; first expansion (gray), n ¼ 4; second expansion (black), n ¼ 5. Mean  standard error of the mean values are shown.

ex vivo to 77.2%  1.8% after first and 46.2%  1.9% after second expansion) (ex vivo versus second, P < 0.01; first versus second, P < 0.01) (Figure 3B). Suppressive function of Tregs peaked after one round of expansion We tested the ability of CD25Sel-2 cells to suppress the proliferation of mixed lymphocyte reaction (MLR) (Figure 4A). We used the allogeneic PBMC from the same donor as the allogeneic stimulator for each CD25Sel-2 in the MLR to minimize variation

A

Ex vivo

1st expansion

caused by human leukocyte antigen disparity between different stimulators and so were able to compare the suppressive function of CD25sel-2 ex vivo and post each round of expansion. The CD25Sel-2 cells from expansion-1 showed much greater suppressive ability than from expansion-2 or ex vivo (P < 0.01) (Figure 4B). The ratio of CD25sel-2 and autologous PBMC ex vivo, after one and two rounds of expansion required to achieve 50% suppression, were 1:1e1/4:1, 1/16:1e1/32:1 and 1:1e1/2:1, respectively. In addition, the “background” responses of the Treg cells alone to

B

2nd expansion

IL-2&FoxP3

3

3

4

6

11

21

12

82

7

83

27

41

FoxP3 IFNγ 1

20

3

76

1

12

4

84

6

59

4

31

FoxP3

% of cells in the CD4+ gate

IL-2

90 80 70 60 50 40 30 20 10 0

p<0.001 P<0.05

P<0.05

IL-2+ FoxP3-

IL-2+ FoxP3+

IL-2FoxP3-

IL-2FoxP3+

P<0.01 P<0.01

IFNγ&FoxP3 90 80 70 60 50 40 30 20 10 0

P<0.05

IFNγ+ FoxP3-

IFNγ+ FoxP3+

IFNγFoxP3-

IFNγFoxP3+

Figure 3. Cytokine production in CD3þCD4þ T cells from CD25sel-2 cells ex vivo and after first and second rounds of expansions. (A) Representative FACS plots showing staining for IL-2 versus FOXP3, IFNg versus FOXP3 in CD4þ T cells from CD25Sel-2 cells ex vivo, after expansion-1 and expansion-2. (B) Summaries of IL-2 and IFNg production versus FOXP3 by CD4þ cells ex vivo (open), after the first (gray) or second expansions (black). IL-2 and FOXP3, n ¼ 6; IFNg and FOXP3, n ¼ 5.

Fas/Fas-L and human regulatory T-cell apoptosis

A

B P<0.01

P<0.01

6.4×10 3.2×10

3.0×10

Cell number

CPM

4.0×10

1291

2.0×10 1.0×10

1.6×10 8.0×10 4.0×10 2.0×10

0 1:1

1/2:1 1/4:1 1/8:1 1/16:1 1/32:1 1/64:1 0:1

Expanded CD25Sel-2:Responder

1.0×10

Ex vivo

1st exp.

2nd exp.

Figure 4. Suppressive activity of CD25Sel-2 cells in MLR assays ex vivo and after first or second rounds of expansion. (A) Representative data from one experiment (representative of five) with the use of CD25Sel-2 cells from expansion-1. (B) Comparison of the suppressive ability of the CD25Sel-2 cells before and after each expansion, expressed as the absolute (or the closest) numbers of ex vivo or expanded CD25Sel-2 cells required to suppress MLRs by 50% (ex vivo, n ¼ 3; expansion-1, n ¼ 4; expansion-2, n ¼ 5).

allogeneic PBMC were consistently lower after the first expansion compared with the background responses after expansion-2 (1209  283 cpm versus 9235  2584 cpm, respectively; P < 0.01). Expression of Fas and Fas-L on different populations To identify the causes of the enrichment of CD8þ T cells after repeated stimulation, we compared Fas and Fas-L expression on nCD4þ T cells, CD8þ T cells and CD25Sel-2 from the same donors at the end of first expansion. CD8þ T cells and nCD4þ T cells were isolated with reagents from Miltenyi to >97% purity (data not shown). Cell surface

expression of Fas-L on all three populations was low (Figure 5A). The level of Fas was similar on nCD4þ T cells, CD8þ T cells and the FOXP3 T cells of CD25Sel-2. However, the intensity of Fas was much higher on the FOXP3þ, particularly on FOXP3high T cells of CD25Sel-2 (Figure 5A). Cell proliferation in different T-cell populations We measured T-cell growth in subpopulations of T cells by spiking CD25Sel-2 or nCD4þ T cells with autologous CD8 cells at a 4:1 ratio. The ratio and the differential expression of Fas on CD4þ and CD8þ T cells was confirmed on the spiked samples before

Figure 5. Expression and the relation of Fas and Fas-L to FOXP3, on CD3þCD4þ of naive, CD25Sel-2 and CD3þCD8þ T cells at the times indicated. (A) Cell surface expression of Fas, Fas-L and FOXP3 on nCD4þ T cells, CD4þ T cells in CD25Sel-2 and CD8þ T cells at the end of the first expansion. (B) Staining of CD8-spiked samples with CD3, CD4, CD8 and Fas before the second expansion; nCD4þ T cells and CD25Sel-2 were spiked with autologous CD8þ T cells at a ratio of 4:1. Expression of Fas on CD8þ T cells (red), nCD4þ T cells (green) and CD4þ T cells in the CD25Sel-2 (green). (C) Staining of CFSE-pulsed, spiked samples with CD3, CD4 and CD8 on day 5 of the second stimulation with plate coated 1 mg/mL CD3 and 0.5 mg/mL of soluble CD28. Dilution of CFSE in CD8þ T cells (green), nCD4þ T cells (red) and CD4þ T cells in the CD25Sel-2 (red) in CD8þ T cells spiked samples. Data are representative of three experiments.

1292

W. Zhang et al.

restimulation with plate-coated CD3, soluble CD28 and IL-2 (Figure 5B). The spiked samples were then pulsed with carboxyfluorescein succinimidyl ester (CFSE) and stimulated as described above. The rate of CFSE dilution was similar among nCD4þ T cells and spiked CD8þ T cells (Figure 5C). The ratio of nCD4þ to CD8þ T cells remained at 4:1 at the end of culture. However, when the growth of a mixture of CD25Sel-2 and CD8þ T cells was analyzed, CFSE diluted further in the CD8þ T cells than in the CD4þ T cells of the CD25Sel-2, which suggests a faster rate of proliferation of the CD8þ T cells compared with CD4þCD25þ T cells. Indeed, the ratio of CD4 and CD8 T cells reversed at the end of culture (Figure 5C). The remaining CD4þ T cells in the CD25Sel-2 had a CFSE intensity similar to nCD4þ T cells, which indicates that CD4þ T cells in the CD25Sel-2 and nCD4þ T cells have undergone a similar degree of proliferation in the presence of rhIL-2. Fas-Fas-Lemediated apoptosis We assessed expression of FOXP3 in CD4þ cells of CD25Sel-2 before and after each round of expansion. We found that there was a selective loss of FOXP3high cells during the second round of expansion. The positive correlation between FOXP3 and Fas led us to assess the sensitivity of CD4þ T cells in the CD25Sel-2 cells to Fas-mediated apoptosis induced by the agonist anti-Fas mAb CH-11 (Figure 6A). We observed that this Fas agonist caused dose-dependent reductions in numbers of CD4þFasþFOXP3þ cells (30.5%  3.7% in controls to 14.5%  0.5% with CH-11 at 1 mg/mL, P < 0.05) and especially in CD4þFasþFOXP3high cells (from 35.2%  3.6% to 19.1%  4.5% at 0.125 mg/mL, P < 0.01; to 5.4%  0.5% and 5.2%  0.4% at 0.5 and 1 mg/mL,

respectively, P < 0.001), with concomitant enrichment of cells that lacked expression of FOXP3 (Figure 6B). We noted that the effect of CH-11 reached saturation at 0.5 mg/mL. Fas antagonist antibody prevented CD4þFOXP3high cell apoptosis by reducing Caspase 8 activation If FaseFas-L interactions cause apoptosis of CD4þFOXP3high cells, then apoptosis of these cells may be reduced by blocking FaseFas-L interactions. We therefore tested whether expansion-related apoptosis can be prevented by the Fas-blocking antibody ZB4. We found a significant dose-dependent increase in viable CD4þFOXP3high cells after 24 h. The percentage of CD4þFOXP3high cells in the experimental groups increased from 13.8%  2.8% to 33.6%  2.8% at 0.5 mg/mL of ZB4 (P < 0.05) and to 37.2%  3% at 1 mg/mL of ZB4 (P < 0.01). There was a concomitant reduction of the CD4þFOXP3þ cells (from 60.3%  3.5% to 45%  2.4% and 45%  2.8% at 0.5 and 1 mg/mL of ZB4, respectively) (P < 0.05), whereas the percentage of CD4þFOXP3 cells remained unchanged (Figure 7A). Staining for FLICA, a marker of Caspase 8 activation, was most evident in cells expressing high levels of FOXP3 in both the control and experimental groups; within them, it was reduced from 20.5%  0.2% in controls to 12.6%  1.4% at 1 mg/mL ZB4 (P < 0.05) (Figure 7B). Taken together, these data suggest that dose-dependent protection from Fas/Caspase 8emediated apoptosis can be achieved by Fas-blocking antibodies. Neither Fas nor Fas-L antagonist protected FOXP3high cells in long-term culture If FaseFas-L interactions reduce the number and function of CD4þFOXP3high cells in the cultures

Figure 6. Effect of Fas agonist CH-11 on cells expressing different levels of FOXP3. (A) CD25Sel-2 at the end of first expansion was incubated with or without CH-11 at the indicated concentration. Profiles of CD4þ T cells in the CD25Sel-2 after 2-h incubation with (a) medium, (b) 0.125, (c) 0.5 or (d) 1 mg/mL of CH-11 mAb. Results of a representative experiment of three experiments are shown. (B) CH-11einduced cell depletion/enrichment in cells expressing FOXP3 at the indicated levels. Control IgM at 1 mg/mL (open), CH-11 at 0.125 (light gray), 0.5 (dark gray) or 1 (black) mg/mL, n ¼ 3.

Fas/Fas-L and human regulatory T-cell apoptosis

1293

Figure 7. Short- and long-term effects of Fas or Fas-L antagonist ZB4 or NOK-1 on cells expressing different levels of FOXP3. (A) Dosedependent protection from loss of CD4þFOXP3high cells by the Fas-antagonistic ZB4 mAb at the end of 24-h incubation, n ¼ 4. (B) At the end of experiments, cells were stained for Live/Dead fixable stain to exclude dead cells before being further stained for CD3, CD4, FOXP3 and FLICA staining for detecting activated Caspase-8. Percentage of FLICAþ events in cells expressing FOXP3 at the indicated levels from cultures with/without ZB4 at doses are indicated. Control IgG at 1 mg/mL is shown as (open bar), mAb ZB4 at 0.25 (light gray), 0.5 (dark gray) or 1 (black) mg/mL. n ¼ 3. (C) Effect of Fas or Fas-L antagonist on cells expressing FOXP3 at the indicated levels at the end of second expansion; Control antibody (open bar) or NOK-1, a Fas-L antagonist mAb (thin line bar) or ZB4 (black bar) both at 1 mg/mL were added to the culture throughout the second expansion, n ¼ 3. Only P < 0.05 is shown.

undergoing expansion, we hypothesized that Fas and Fas-L antagonist may prevent the decline in CD4þFOXP3high cells. We therefore added either Fas or Fas-L antagonist antibody ZB4 or NOK-1, both at 1 mg/mL, throughout the second round expansions to see if we could selectively prevent the loss of FOXP3high cells. There was no significant difference in the proportion of CD4þFOXP3high cells between the controls and experimental groups (Figure 7C). In the presence of either antibody, the cells formed larger clumps, and cell counts were higher in the experimental groups compared with the controls, which suggests that these antagonist antibodies were able to protect cell death. However, the effect was not specific to CD4þFOXP3high cells (Figure 7C). Discussion In the present study, we show that in bead-isolated CD4þCD25Sel-2 cells enriched for CD4þFOXP3þ T cells, the level of FOXP3 expression is upregulated by initial TCR stimulation but declined after repeated TCR stimulation. We noticed that the majority of CD4þFOXP3þ T cells expressed higher levels of Fas than CD4FOXP3 cells in the CD25Sel2, as well as CD8þ and nCD4þ T cells from the same donor, which had been stimulated in the similar manner. Furthermore, there is a positive correlation in the levels of FOXP3 and Fas on

individual cells expressing both molecules in the CD4þCD25Sel-2. We have shown that the loss of CD4þFOXP3high cells in the CD25Sel-2 was accelerated by Fas agonist and partially prevented by Fas antagonist in the presence of TCR stimulation. Demonstration of Fas-L expression by FACS can be difficult because of the low level of basal expression and dynamic upregulation in special circumstances. We did not observe a clear difference in Fas-L expression between CD4þ from CD25Sel-2, CD8þ and nCD4þ T cells when staining of Fas-L was performed at the end of the first round of expansion without further TCR stimulation. With the use of an amplification technique, Weiss et al. (23) showed that conventional T cells (Tcons) expressed higher levels of Fas-L than Tregs when examined 20 h past TCR stimulation. TCR stimulation and the presence of Tcons in the CD25Sel-2 therefore could have contributed to CD4þFOXP3high cell death by inducing Fas-L expression on Tcons and by further upregulating Fas expression on CD4þFOXP3high cells. When labelled human Tregs were transfused back, it was found that CD4þCD45ROþCD25high FOXP3þ cells have a fast turnover rate in vivo in steady state and have critically short telomeres and low telomerase activity (24). Consistent with the above observation, mature murine CD4þCD25þFOXP3þcells are highly sensitive to activation-induced cell death (25). Pathways, other than FaseFas-L, could have contributed to the decline of CD4þFOXP3high cells

1294

W. Zhang et al.

after repeated TCR stimulation. Even in the presence of ZB4, CD4þFOXP3high cells still had the highest percentage of FLICAþ events compared with that of CD4þFOXP3 cells and CD4þFOXP3þ cells. This is consistent with the failure to rescue CD4þFOXP3high cells by Fas or Fas-Leblocking antibodies, ZB4 or NOK-1, in the second round of expansion. Furthermore, Tregs express many other receptors of the tumour necrosis factor (TNF) superfamily (26). The net effect of the tumour necrosis factor receptor superfamily signalling may also differentially regulate the homeostasis of Tregs and other T-cell compartments (27,28). The molecular mechanisms that predispose CD4þFOXP3high cells to increased sensitivity to cell death warrant future study. Significant increase in the frequency of CD8þ T cells after the second round of expansion only occurs in CD25Sel-2 spiked but not the nCD4þ T cell spiked cultures, this suggests that rapid proliferation of CD8þ T cells and their survival advantage over CD4þFOXP3high cells could be responsible for this outcome. Depletion of CD8þ T cells is therefore mandatory before Treg expansion. Tregs have been shown to be reprogrammed to downregulate FOXP3 and gain the ability to produce cytokines such as IFNg or IL17, especially in an inflammatory or lymphopenic environment (29,30). We could not rule out the possibility that CD4þ T cells expressing low levels of FOXP3 acquired cytokine-producing ability after strong TCR stimulation and CD28 costimulation. This could be partially responsible for enrichment of CD4þFOXP3, CD4þIL-2þFOXP3þ and CD4þIFNgþFOXP3þ cells in the expanded cultures. Instead of showing Treg function as a percentage of suppression in the MLR, we have shown the number of CD25Sel-2 required to achieve approximately 50% of suppression. We reasoned that knowing the cell numbers help when the time comes for deciding the dosing strategy for Treg therapy. We have shown a significant decline in suppressive function in CD25Sel-2 after the second expansion, which contains a significant higher number of CD8þ T cells. It is possible that proliferation of the CD8þ T cells contributed to the higher counts and therefore masked the suppression by CD25Sel-2. CFSE-based suppression assays or the use of CD25Sel-2 after CD8þ T-cell depletion will be more informative. However, we have provided evidence that the remaining CD4þ T cells express FOXP3þ at lower levels than those cells from the first expansion and many of them also produced IL2 and IFNg. Given that the level of FOXP3 expression has been shown to be important for maintaining Treg phenotype and function (31), it is likely that CD4þFOXP3þ T cells

from the second expansion are weaker in their ability to suppress than the ones from the first expansion. A small fraction of CD3, CD20 and CD16CD56-null cells were also identified in the ex vivo CD25Sel-2 and at moderate frequencies after expansion. The CD3, CD20 and CD16CD56null cells in the ex vivo CD25Sel-2 are mostly large cells, which suggests a myeloid origin. While the majority of these null cells in the expanded CD25Sel-2 were cells in the lymphocyte gate. T cells can downregulate CD3 expression especially when a strong TCR stimulation is given, and may account for some of the CD3, CD20 and CD16CD56null cells after expansion. The identity and influence of these cells could not be assessed and would require further investigation. The most promising expansion data for the clinical use of Tregs in allogeneic hematopoietic stem cell transplantation uses cord bloodederived Tregs or FACS sorted CD45RAþCD25þ cells (32,33). Faithful epigenetic modulation at the FOXP3 loci was thought to be responsible. It could also be that naive Tregs are resistant to FaseFas-Lemediated cell death (33,34). Obtaining highly purified Tregs by FACS ensures low contamination of Tcons and consequently less Fas-L generation in response to TCR stimulation could have also accounted for success in Treg expansion in this setting. Rapamycin has been shown to maintain the functionality of expanded Tregs by selectively suppressing the proliferation of contaminating Tcons (35). Rapamycin alone, however, is less potent than rapamycin and retinoid acid together in preventing the loss of demethylation in the Treg-specific demethylated region (TSDR) (36). Differences namely the strength, duration and frequency of TCR stimulation in different expansion protocols could also account for variations in Treg expansion. It is important to increase Treg expansion efficiency and quality. The expansion-related changes in Treg homing ability warrant further study. Thus far, limited data indicates that expanded human Tregs from either cord or adult donors are short-lived (3,9,33). In contrast, ex vivoeisolated CD25þ T cells have shown promising benefit in a phase I-II clinical trial in haplo-identical hematopoietic stem cell transplantation (37). Deciding on the appropriate cell isolation method tailored to each clinical setting is therefore important. Our findings has raised the possibility that Tregs in vivo are sensitive to FaseFas-Leinduced apoptosis, and this is part of the homeostatic mechanisms involved in initiating and maintaining immune responses. Further investigation of Treg expansion and control of expansion by FaseFas-L interactions may give some novel insights into the development of immune responses.

Fas/Fas-L and human regulatory T-cell apoptosis FOXP3 TSDR demethylation has been found to be responsible for stable constitutive FOXP3 expression in Tregs. A significant reduction in the percentage of FOXP3þ cells was found in highly purified FACS-sorted CD4þCD25þCD127 Tregs after 2 weeks of expansion and a more significant reduction after 3 weeks of expansion (38). Hoffmann et al. (38) have further shown that whereas changes in FOXP3 TSDR demethylation are minimal during a 2-week expansion culture, they were observed after a 3-week expansion culture. In fact, Hoffman et al. (38) also demonstrated that loss of FOXP3 expression in Treg clones derived from highly purified CD4þCD25þCD45RAþ Tregs required these cells to be expanded for a minimum of 6 weeks, with six to eight times stimulation. It is conceivable that changes in FOXP3 TSDR demethylation in cells expressing FOXP3 require longer expansion than what is described in our study. We have shown that a preferential loss of CD4þFOXP3high cells occurred during the second round of CD25Sel-2 expansion. CD4þFOXP3high cells express high levels of Fas and are highly sensitive to FaseFas-Lemediated apoptosis. At the end of the second expansion, CD25Sel-2 contains a significantly increased number in CD8þ T cells and CD4þ T cells producing IFNg and/or IL-2, which shows a reduced suppressive function. Acknowledgments WZ was funded by NHSBT and Department of Haematology Trust Fund. The research benefited from funding by the United Kingdom National Health Service R&D Directorate through the National Institute of Health Research (NIHR) (RP-PG0310-1003) and the National Institute for Health Biomedical Research Centre Program. RD is supported by the Wellcome Trust Academic Clinical Fellowship Scheme. The authors wish to thank Dr Abigail Lamikanra and Professor Nick Wilcox for critical reading of the manuscript. Disclosure of interests: The authors have no commercial, proprietary, or financial interest in the products or companies described in this article. References 1. Sakaguchi S, Sakaguchi N, Asano M, Itoh M, Toda M. Immunologic self-tolerance maintained by activated T cells expressing IL-2 receptor alpha-chains (CD25): breakdown of a single mechanism of self-tolerance causes various autoimmune diseases. J Immunol. 1995;155:1151e64. 2. Negrin RS. Role of regulatory T cell populations in controlling graft vs host disease. Best Pract Res Clin Haematol. 2011;24: 453e7.

1295

3. Edinger M, Hoffmann P. Regulatory T cells in stem cell transplantation: strategies and first clinical experiences. Curr Opin Immunol. 2011;23:679e84. 4. Roncarolo MG, Battaglia M. Regulatory T-cell immunotherapy for tolerance to self antigens and alloantigens in humans. Nat Rev Immunol. 2007;7:585e98. 5. Trenado A, Sudres M, Tang Q, Maury S, Charlotte F, Gregoire S, et al. Ex vivo-expanded CD4þCD25þ immunoregulatory T cells prevent graft-versus-host-disease by inhibiting activation/differentiation of pathogenic T cells. J Immunol. 2006;176:1266e73. 6. Nadig SN, Wieckiewicz J, Wu DC, Warnecke G, Zhang W, Luo S, et al. In vivo prevention of transplant arteriosclerosis by ex vivo-expanded human regulatory T cells. Nat Med. 2010; 16:809e13. 7. Edinger M, Hoffmann P, Ermann J, Drago K, Fathman CG, Strober S, et al. CD4þCD25þ regulatory T cells preserve graft-versus-tumor activity while inhibiting graft-versus-host disease after bone marrow transplantation. Nat Med. 2003;9: 1144e50. 8. Takahashi T, Kuniyasu Y, Toda M, Sakaguchi N, Itoh M, Iwata M, et al. Immunologic self-tolerance maintained by CD25þCD4þ naturally anergic and suppressive T cells: induction of autoimmune disease by breaking their anergic/ suppressive state. Int Immunol. 1998;10:1969e80. 9. Trzonkowski P, Bieniaszewska M, Juscinska J, Dobyszuk A, Krzystyniak A, Marek N, et al. First-in-man clinical results of the treatment of patients with graft versus host disease with human ex vivo expanded CD4þCD25þCD127- T regulatory cells. Clin Immunol. 2009;133:22e6. 10. Hippen KL, Merkel SC, Schirm DK, Sieben CM, Sumstad D, Kadidlo DM, et al. Massive ex vivo expansion of human natural regulatory T cells (T(regs)) with minimal loss of in vivo functional activity. Sci Transl Med. 2011;3:83ra41. 11. Nagata S. Apoptosis by death factor. Cell. 1997;88:355e65. 12. Nagata S. Human autoimmune lymphoproliferative syndrome, a defect in the apoptosis-inducing Fas receptor: a lesson from the mouse model. J Hum Genet. 1998;43:2e8. 13. Wu J, Wilson J, He J, Xiang L, Schur PH, Mountz JD. Fas ligand mutation in a patient with systemic lupus erythematosus and lymphoproliferative disease. J Clin Invest. 1996;98: 1107e13. 14. Lenardo M, Chan KM, Hornung F, McFarland H, Siegel R, Wang J, et al. Mature T lymphocyte apoptosiseimmune regulation in a dynamic and unpredictable antigenic environment. Annu Rev Immunol. 1999;17:221e53. 15. Bossi G, Griffiths GM. Degranulation plays an essential part in regulating cell surface expression of Fas ligand in T cells and natural killer cells. Nat Med. 1999;5:90e6. 16. Suda T, Okazaki T, Naito Y, Yokota T, Arai N, Ozaki S, et al. Expression of the Fas ligand in cells of T cell lineage. J Immunol. 1995;154:3806e13. 17. Kayagaki N, Kawasaki A, Ebata T, Ohmoto H, Ikeda S, Inoue S, et al. Metalloproteinase-mediated release of human Fas ligand. J Exp Med. 1995;182:1777e83. 18. Peter ME, Budd RC, Desbarats J, Hedrick SM, Hueber AO, Newell MK, et al. The CD95 receptor: apoptosis revisited. Cell. 2007;129:447e50. 19. Miyara M, Yoshioka Y, Kitoh A, Shima T, Wing K, Niwa A, et al. Functional delineation and differentiation dynamics of human CD4þ T cells expressing the FoxP3 transcription factor. Immunity. 2009;30:899e911. 20. Fritzsching B, Oberle N, Eberhardt N, Quick S, Haas J, Wildemann B, et al. In contrast to effector T cells, CD4þCD25þFoxP3þ regulatory T cells are highly susceptible to CD95 ligand- but not to TCR-mediated cell death. J Immunol. 2005;175:32e6.

1296

W. Zhang et al.

21. Matsuoka K, Kim HT, McDonough S, Bascug G, Warshauer B, Koreth J, et al. Altered regulatory T cell homeostasis in patients with CD4þ lymphopenia following allogeneic hematopoietic stem cell transplantation. J Clin Invest. 2010;120:1479e93. 22. Pozarowski P, Huang X, Halicka DH, Lee B, Johnson G, Darzynkiewicz Z. Interactions of fluorochrome-labeled caspase inhibitors with apoptotic cells: a caution in data interpretation. Cytometry A. 2003;55:50e60. 23. Weiss EM, Schmidt A, Vobis D, Garbi N, Lahl K, Mayer CT, et al. Foxp3-mediated suppression of CD95L expression confers resistance to activation-induced cell death in regulatory T cells. J Immunol. 2011;187:1684e91. 24. Vukmanovic-Stejic M, Zhang Y, Cook JE, Fletcher JM, McQuaid A, Masters JE, et al. Human CD4þ CD25hi Foxp3þ regulatory T cells are derived by rapid turnover of memory populations in vivo. J Clin Invest. 2006;116:2423e33. 25. Miyao T, Floess S, Setoguchi R, Luche H, Fehling HJ, Waldmann H, et al. Plasticity of Foxp3(þ) T cells reflects promiscuous Foxp3 expression in conventional T cells but not reprogramming of regulatory T cells. Immunity. 2012;36:262e75. 26. Croft M. The role of TNF superfamily members in T-cell function and diseases. Nat Rev Immunol. 2009;9:271e85. 27. Yolcu ES, Ash S, Kaminitz A, Sagiv Y, Askenasy N, Yarkoni S. Apoptosis as a mechanism of T-regulatory cell homeostasis and suppression. Immunol Cell Biol. 2008;86:650e8. 28. Nagar M, Jacob-Hirsch J, Vernitsky H, Berkun Y, BenHorin S, Amariglio N, et al. TNF activates a NF-kappaBregulated cellular program in human CD45RA- regulatory T cells that modulates their suppressive function. J Immunol. 2010;184:3570e81. 29. Bailey-Bucktrout SL, Bluestone JA. Regulatory T cells: stability revisited. Trends Immunol. 2011;32:301e6. 30. Hori S. Regulatory T cell plasticity: beyond the controversies. Trends Immunol. 2011;32:295e300. 31. Rudra D, deRoos P, Chaudhry A, Niec RE, Arvey A, Samstein RM, et al. Transcription factor Foxp3 and its

32.

33.

34.

35.

36.

37.

38.

protein partners form a complex regulatory network. Nat Immunol. 2012;13:1010e9. Hoffmann P, Eder R, Boeld TJ, Doser K, Piseshka B, Andreesen R, et al. Only the CD45RAþ subpopulation of CD4þCD25high T cells gives rise to homogeneous regulatory T-cell lines upon in vitro expansion. Blood. 2006;108:4260e7. Brunstein CG, Miller JS, Cao Q, McKenna DH, Hippen KL, Curtsinger J, et al. Infusion of ex vivo expanded T regulatory cells in adults transplanted with umbilical cord blood: safety profile and detection kinetics. Blood. 2011;117:1061e70. Fritzsching B, Oberle N, Pauly E, Geffers R, Buer J, Poschl J, et al. Naive regulatory T cells: a novel subpopulation defined by resistance toward CD95L-mediated cell death. Blood. 2006;108:3371e8. Battaglia M, Stabilini A, Migliavacca B, Horejs-Hoeck J, Kaupper T, Roncarolo MG. Rapamycin promotes expansion of functional CD4þCD25þFOXP3þ regulatory T cells of both healthy subjects and type 1 diabetic patients. J Immunol. 2006;177:8338e47. Golovina TN, Mikheeva T, Brusko TM, Blazar BR, Bluestone JA, Riley JL. Retinoic acid and rapamycin differentially affect and synergistically promote the ex vivo expansion of natural human T regulatory cells. PLoS ONE. 2011;6: e15868. Di Ianni M, Falzetti F, Carotti A, Terenzi A, Castellino F, Bonifacio E, et al. Tregs prevent GVHD and promote immune reconstitution in HLA-haploidentical transplantation. Blood. 2011;117:3921e8. Hoffmann P, Boeld TJ, Eder R, Huehn J, Floess S, Wieczorek G, et al. Loss of FOXP3 expression in natural human CD4þCD25þ regulatory T cells upon repetitive in vitro stimulation. Eur J Immunol. 2009;39:1088e97.

Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jcyt.2013.05.021