Stability of human rapamycin-expanded CD4+CD25+ T regulatory cells

Abstract

Background The clinical use of ex vivo-expanded T-regulatory cells for the treatment of T-cell-mediated diseases has gained increasing momentum. However, the recent demonstration that FOXP3⁺ T-regulatory cells may contain interleukin-17–producing cells and that they can convert into effector cells once transferred in vivo raises significant doubts about their safety. We previously showed that rapamycin permits the ex vivo expansion of FOXP3⁺ T-regulatory cells while impairing the proliferation of non-T-regulatory cells. Here we investigated the Th17-cell content and the in vivo stability of rapamycin-expanded T-regulatory cells as pertinent aspects of cell-based therapy.Design and Methods T-regulatory-enriched cells were isolated from healthy volunteers and were expanded ex vivo with rapamycin with a pre-clinical applicable protocol. T-regulatory cells cultured with and without rapamycin were compared for their regulatory activity, content of pro-inflammatory cells and stability.Results We found that CD4⁺CCR6⁺CD161⁺ T cells (i.e., precursor/committed Th17 cells) contaminate the T-regulatory cells cultured ex vivo in the absence of rapamycin. In addition, Th17 cells do not expand when rapamycin-treated T-regulatory cells are exposed to a “Th17-favorable” environment. Rapamycin-expanded T-regulatory cells maintain their in vitro regulatory phenotype even after in vivo transfer into immunodeficient NOD-SCID mice despite being exposed to the irradiation-induced pro-inflammatory environment. Importantly, no additional rapamycin treatment, either in vitro or in vivo, is required to keep their phenotype fixed.Conclusions These data demonstrate that rapamycin secures ex vivo-expanded human T-regulatory cells and provide additional justification for their clinical use in future cell therapy-based trials.

Introduction

In recent years, there has been growing recognition of the ability of human FOXP3 T-regulatory (Treg) cells to suppress adverse immune responses. Much research effort has been devoted to implementing Treg-cell therapy in the clinic1,2 and results of the first clinical trials have been recently reported.3,4 Given their reduced circulating frequency (i.e., 5–10% of the CD4 T cells) FOXP3 Treg cells, to be of any therapeutic use in cell-therapy trials, need to be expanded ex vivo. The intrinsic reduced ability of Treg cells to proliferate in vitro can be reversed by potent stimulation through the T-cell receptor (TCR) in the presence of high doses of exogenous interleukin (IL)-2.5 Protocols for the expansion of human FOXP3 Treg cells performed under conditions of good manufacturing practice (GMP) have been shown to be feasible.3,6–8 Importantly, these culture conditions are also highly advantageous for the expansion of effector CD25 T cells, which may contaminate the starting peripheral FOXP3 Treg cells. This obstacle can be bypassed by: (i) the use of cord blood Treg cells,4 (ii) isolation of very pure peripheral Treg cells through flow-based cell sorters3,9,10 or (iii) the addition of rapamycin to the culture, which preferentially inhibits proliferation of effector T cells while sparing Treg cells.11,12

IL-17 is a pro-inflammatory cytokine with non-redundant functions in the clearance of extracellular pathogens.13 In humans, IL-17 is associated with many inflammatory disorders such as rheumatoid arthritis, asthma, multiple sclerosis, Crohn’s disease and allograft rejection.14 The majority of IL-17 derives from a population of CD4 T cells known as Th17.13 The combination of cytokines which stimulates the development and stabilizes human Th17 cells is a subject of much debate.15 It appears that transforming growth factor (TGF)-β, IL-21, IL-23, IL-6, and IL-1β at various dosages and combinations, depending on the experimental conditions and the starting T-cell subset, are essential for Th17-cell generation/expansion.15

It has recently been shown that a memory fraction of CD25FOXP3 T cells contains a significant number of IL-17-producing cells which lack regulatory activity.16,17 In the context of concerted efforts to use expanded populations of Treg cells for adoptive therapy in human T-cell-mediated diseases, the risk of Th17-cell contamination is a key, as transfer of contaminant pro-inflammatory cells may exacerbate rather than ameliorate disease. Thus, developing methods of securing ex vivo-expanded Treg cells intended for clinical use and inhibiting their pro-inflammatory potential in vivo have become all-important aspects of Treg cell-mediated immunotherapy.18

We set up a clinically applicable protocol for the expansion of human peripheral Treg cells using rapamycin to set the basis for an investigator-initiated clinical trial. Th17-cell content and the in vivo stability of the rapamycin-expanded human FOXP3 Treg cells were investigated as critical pertinent aspects of cell-based therapy.

Design and Methods

Blood samples

Peripheral blood was obtained from healthy donors’ buffy coats obtained after written informed consent in accordance with the Declaration of Helsinki under the protocol approved by the San Raffaele Scientific Institute Ethics Committee (protocol TIGET PERIBLOOD).

T-regulatory cell purification

Peripheral blood mononuclear cells (PBMC) were separated by density-gradient centrifugation over Lymphoprep (Axis-shield, Oslo, Norway). Some PBMC were cryopreserved for later use. The remaining PBMC were processed using CD8 MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s instructions. Treg cells were then isolated from the negative cell fractions (CD8 cells) using CD25 MicroBeads (Miltenyi Biotec) following the manufacturer’s instructions.

Flow cytometry

The cells were stained for surface antigens with the following monoclonal antibodies: anti-CD4 PerCP (clone SK3, BD, Franklin Lakes, NJ, USA), anti-CD25 APC (clone 2A3, BD), CD161 PE (clone HP3G10, eBioscience), and CCR6-biotin (clone 11A9, BD) and streptavidin APC7 (BD). Intracellular staining for human FOXP3 and Helios was performed using the anti FOXP3-Alexa 488 and anti-Helios-PE monoclonal antibodies (clone 259D and clone 22F6, respectively, BioLegend, San Diego, CA, USA) according to the manufacturer’s instructions. Samples were acquired on a FACSCanto (BD) and analyzed with FCS Express V3 software (DE Novo Software, Los Angeles, CA, USA).

Cell cultures

CD8CD25 T cells isolated from PBMC of healthy subjects were plated at 0.1×10/mL in X-vivo 15 medium (Lonza, Verviers, Belgium) supplemented with 5% pooled AB human serum (Lonza) and 100 U/mL penicillin/streptomycin (Bristol-Myers Squibb, Sermoneta, Italy) in the presence or absence of 100 nM rapamycin (i.e. 100 ng/mL) (Rapamune, Whyet, Maidenhead, Berkshire, UK).12 The cells were activated with GMP grade beads coated with anti-CD3 and anti-CD28 antibodies (T Cell Expander from Dynal Biotech ASA, Oslo, Norway) at a 1:3 ratio (cells:beads). At day 2 post-activation, IL-2 (500 U/mL) (Proleukin, Chiron, Amsterdam, the Netherlands) was added to the culture and replenished as shown in Figure 1. Beads were removed after 14 days of culture by magnetic adherence and the expanded cells were used for further analyses.

Suppression assay with TMED and TRPM cells

Previously frozen PBMC were thawed and stained with CFSE (Molecular Probes, Eugene, OR, USA)11 and a suppression assay was performed as described elsewhere by our group.12 The percentage of CFSE divided cells was calculated using the formula: [(n. of precursors that proliferated_1-6/n. of total precursors_0-6)×100]. The percentage of CFSE cells divided in the presence of expanded cells was compared to the percentage of CFSE divided cells in the absence of any added cells.12

Cytokine detection in TMED and TRPM cells

The 14-day cultured T cells were plated at a density of 0.5×10 cells in 500 μL in 48-well plates and stimulated for 6 h with latex-beads (Molecular Probes) pre-coated with anti-CD3 (OKT3 clone) and anti-CD28 (clone CD28.2, BD Pharmingen) monoclonal antibodies at 500 ng and 50 ng for 10 beads, respectively at a 1:2 ratio (cells:beads) in the presence of 10 μg/mL brefeldin A (Sigma-Aldrich). At the end of the 6-h activation, the cell surface was stained with anti-CD4 PerCp monoclonal antibody (clone SK3, BD). Subsequently, intracellular staining was performed with the following monoclonal antibodies: anti-IL-2 PE (clone MQ1-17H12), anti-IL-4 PE (clone MP4-25D2, anti-IL-10 PE (clone JES3-9D7), anti-IFNγ APC (clone B27), and anti-TNFα PE (clone MaB11) (all from BD Pharmingen), and anti-IL-17 PE (clone eBio64CAP17, from eBioscence). Each stain was performed in the presence of anti-FOXP3 Alexa488 monoclonal antibody (clone 259D, Biolegend) according to the manufacturer’s protocol.

TH17-cell differentiation/expansion

The T_MED and T_RPM cells were plated at a density of 0.5×10/mL and cultured with 5 ng/mL TGF-β (R&D System, Minneapolis, MN, USA), 25 ng/mL IL-21 (Cell Sciences, Canton, MA, USA) and 25 ng/mL IL-23 (R&D System) as previously described.19 On day 3, IL-23 was replenished. On day 7 the cells (i.e., T_MED-17, and T_RPM-17) were collected and used for further analysis.

Suppression assay with TMED-17 and TRPM-17 cells

Previously frozen PBMC were thawed and stained with CFSE (Molecular Probes) as previously described.11 T_MED-17 and T_RPM-17 cells were added at the ratio of 0.3:1 (T cells:responder cells) and stimulated with Treg Suppression Inspector beads (Miltenyi Biotec) at a ratio of 1:0.3 (cells:beads). Seven days later the cells were collected and stained with anti-CD4 and anti-CD8 monoclonal antibodies and analyzed by flow activated cell sorting (FACS).

Cytokine detection in TMED-17, TRPM-17 cells

T_MED-17, T_RPM-17 cells were stimulated for 6 h with TPA (CalBiochem, Lucerne, Switzerland) and ionomycin (Sigma-Aldrich) at 10 ng/mL and 500 ng/mL, respectively, in the presence of 10 μg/mL brefeldin A (Sigma-Aldrich). At the end of the 6-h activation, the cell surface was stained with anti-CD4 and anti-CD25 monoclonal antibodies, while intracellular staining was performed with anti-IL-17 or anti-IL-2, and anti-IFNγ APC monoclonal antibodies.

Non-obese diabetic/severe combined immunodeficiency mouse model

Six- to 8-week old female NOD/scid mice were obtained from Charles-River Italia (Calco, Italy). The experimental protocol was approved by the San Raffaele Scientific Animal Care and Use Committee (IACUC#387). The day before human cell transfer, mice were treated with 1 mg of anti-CD122 (TMβ-1) monoclonal antibody (Santa Cruz, Tebubio, Le Perray-en-Yuelines, France) i.p. as previously described.20 On day 0, the mice received total body irradiation (TBI) with a single dose of 350 cGy (γ irradiation from a linear accelerator) and 8×10 cells/mouse were transferred i.p.. Three times a week the general appearance, body weight and mobility of the mice were evaluated and once a week peripheral blood was collected from the tail vein and chimerism analysis was performed. Human chimerism was determined weekly by flow cytometry, after bleeding from the tail vein, using the monoclonal antibodies anti-human-CD45 APC (clone HI30) and anti-mouse-CD45 PerCp (clone 30F11) (both from BD-Pharmingen). The percentage of chimerism was calculated using the formula: [h-CD45/(h-CD45+m-CD45)x100] as described elsewhere.21 To improve human chimerism, some of the mice were treated with 2×10 IU/mouse/day i.p. of recombinant human IL2 (Proleukin, Chiron) for 7 days after cell transfer.21

Phenotype, cytokine production profile and suppressive activity of T cells recovered from non-obese diabetic/severe combined immunodeficiency mice

Thirteen or 19 days after human cell transfer into NOD/scid mice, the animals were sacrificed and peripheral blood, thymus, spleen and lymph nodes were collected and pooled after smashing (T_MED-mouse and T_RPM-mouse). The total cells recovered from each mouse varied accordingly to the cell-type injected. Total cells recovered from mice injected with CD3 T cells: 4.5±1.2×10, with T_MED: 0.2±0.5×10, and with T_RPM: 0.5±0.1×10 (average ± SD, n=7). Total cells from mice injected with CD25T_RPM cells were analyzed by flow cytometry or were stimulated for 6 h with TPA (CalBiochem) and ionomycin (Sigma-Aldrich) in the presence of brefeldin A (Sigma-Aldrich). At the end of the 6-h activation, the cell surface was stained with anti-human CD4 PerCp (clone SK3, BD) and anti-human-CD45 APC (clone HI30, BD-Pharmingen) monoclonal antibodies. Subsequently, the intracellular staining was performed with anti-IL-17 or anti-IL-2, and anti-IFNγ monoclonal antibodies.

Figure 1.Protocol used for the expansion of enriched human CD25⁺ T cells. (A) The diagram outlines the time points (D: day) and the procedures followed for the ex vivo expansion of CD25⁺-enriched Treg cells. T_MED are the cells expanded in the absence of rapamycin, T_RPM are the cells expanded in the presence of rapamycin. The protocol is described in detailed in the Design and Methods section. (B) The total cell fold expansion after 14 days of culture was determined (left panel). CD25⁺ T_MED (empty symbols) and CD25⁺ T_RPM cells (filled symbols) were collected and counted at the end of the culture. Each dot represents one experiment. The average fold expansion is shown (bar and number). One representative time-course expansion of CD25⁺ T_MED (empty symbols) and CD25⁺ T_RPM cells (filled symbols) is presented (right panel).

Alternatively, T cells were purified by using human-CD3 MicroBeads (Miltenyi Biotec) according to the manufacturer’s instructions and used as suppressor T cells in a suppression assay. Previously frozen PBMC (responder cells) were thawed and stained with CFSE (Molecular Probes) as described elsewhere.11 T_MED-mouse and T_RPM-mouse cells were added at the ratio of 0.3:1 and 1:1 (T cells:responder cells) and stimulated with Treg Suppression Inspector beads (Miltenyi Biotec) at a ratio of 1:0.3 (cells:beads). Seven days later the cells were collected and stained with anti-CD4 and anti-CD8 monoclonal antibodies and analyzed by FACS.

Statistical analysis

All statistical analyses were performed using a two-tailed Student’s t-test. P values less than 0.05 were deemed statistically significant.

Results

Human Treg cells were enriched from PBMC using GMP-grade available magnetic beads, which are the best alternative to clinical grade flow-based cell sorters.8 We first depleted CD8 T cells and then positively selected CD25 cells. The purified cell subset (termed CD25) contained a median of 80±10% CD4CD25 cells and 61±5% of CD25FOXP3 cells (data not shown). CD25 cells were cultured ex vivo to obtain the best Treg-cell expansion in the shortest time possible. Increasing doses of GMP-grade anti-CD3/CD28–coated beads (from 1:1 to 1:5 cells:beads) and IL-2 (from 100 to 1000 U/mL), and different periods of culture (from 7 to 21 days) in the presence or absence of rapamycin (T_RPM and T_MED cells, respectively) were tested. The optimal expansion protocol was 14 days of culture in the presence of three beads per cell and 500 U/mL of exogenous IL-2 (scheme depicted in Figure 1A).

T_RPM cells expanded 1 log less than T_MED cells with a 7-day time-lag (Figure 1B), as we previously observed upon expansion of unfractionated human CD4 T cells.12 T_RPM cells contained an average of 51% FOXP3-expressing cells (range: 32–80%). Helios, a member of the Ikaros transcription factor family, is thought to be a specific marker of thymic-derived Treg cells.22 We found an average of 73% (range, 60–85%) of Helios cells within the FOXP3-expressing T_RPM cells, suggesting a selective expansion of thymic-derived Treg cells in the presence of rapamycin. In addition, T_RPM cells displayed a reproducible in vitro suppressive capacity towards both autologous and allogeneic responder T cells in a dose-dependent manner. In contrast, CD25 T cells expanded in the absence of rapamycin (i.e., T_MED cells) almost completely lost FOXP3-expressing cells (average: 14%; range, 3–30%) and the few cells remaining FOXP3 expressed low levels of Helios (average: 28%; range, 20–35%) and retained no in vitro suppressive function in any condition tested (Figure 2 and Online Supplementary Figure S1). The TCR Vβ repertoire remained polyclonal in both T_MED and T_RPM cells demonstrating the absence of a specific T-cell clone expansion in the 14-day culture (Online Supplementary Table S1). We previously showed that rapamycin can be considered as a “selective Treg cell growth-survival” factor since it inhibits proliferation of non-Treg cells,12 which follow different molecular signaling pathways than Treg cells.23,24 Analysis of the cytokine production profile of T_MED and T_RPM cells confirmed that the presence of rapamycin in the culture reduced the frequency of T cells releasing pro-inflammatory cytokines (Figure 3).

Figure 2.Phenotype and suppressive function of 14-day cultured CD25⁺ T cells. (A) The frequency of CD4⁺FOXP3⁺ T cells within CD25⁺ T_MED and CD25⁺ T_RPM cells was established by FACS by gating on alive cells. One representative dot plot is shown on the left and all the results are included in the right vertical scatter plot in which each dot represents one experiment. The average CD4⁺FOXP3⁺ T-cell content is shown (bar and number). (B) The frequency of Helios⁺ cells within CD25⁺ T_MED and CD25⁺ T_RPM cells was established by FACS by gating on CD4⁺FOXP3⁺ T cells (as shown in panel A). One representative dot plot from three experiments is shown. (C) The suppressive ability of CD25⁺ T_MED (empty symbols) and CD25⁺ T_RPM cells (filled symbols) was tested by co-culture experiments with CSFE-labeled autologous (left scatter plot) and allogeneic (right scatter plot) PBMC (responder cells) at a ratio of 0.4:1 (expanded T cells: responder cells). The amount of CD4⁺ and CD8⁺ CFSE⁺ T cells proliferating in the absence or presence of cultured T cells was calculated as described in the Design and Methods section. Percentage of suppressive activity of each tested expanded T-cell population and the average suppressive function an shown (bar and number). Differences that are statistically significant are highlighted by an asterisk (*).

Recent findings demonstrate that FOXP3 T cells contain some IL-17-producing non-Treg cells16,17 and that naïve FOXP3 Treg cells can differentiate into Th17 cells in the presence of IL-2 and lineage-specific polarizing factors.25 As a consequence, we wondered whether the CD25 T-cell starting population contained significant numbers of Th17-cells or Th17-cell precursors, recently found to be highly enriched within the CCR6CD161 cell fraction,26,27 which might expand upon the 14-day culture. The starting CD25 T-cell subset was slightly enriched in CCR6CD161 T cells, as compared to the unfractionated PBMC (Figure 4A). We, therefore, tested the IL-17-producing cell content in T_MED and T_RPM cells. Low amounts of IL-17IFN-γ cells were present in the expanded CD25 T cells irrespectively of the presence of rapamycin (Figure 4B). Nevertheless, T_MED cells contained a higher frequency of CCR6CD161 cells as compared to T_RPM cells (Figure 4C). Rapamycin, therefore, inhibits the differentiation and growth of CCR6CD161 T cells during ex vivo expansion of CD25-enriched T cells. In spite of this, we could not rule out that a subsequent exposure of T_RPM cells to a “Th17-favorable environment” in the absence of exogenous rapamycin would lead to Th17-cell differentiation and/or expansion. To resolve this issue, T_MED and T_RPM cells were cultured in the presence of IL-23, IL-21 and TGF-β (Figure 4D) (termed T_MED-17 and T_RPM-17), a cytokine-cocktail known to expand and stabilize Th17 cells.13,19 Seven days of in vitro culture in this “Th17-favorable environment” and in the absence of rapamycin led to the expansion of IL-17-producing cells (at high or low frequency depending on the donor) in T_MED but not in T_RPM cells (Figure 4E). These data were also confirmed by a high frequency of CCR6CD161 cells in T_MED-17 but not in T_RPM-17 cells (data not shown). In addition, the FOXP3 cell-content (Figure 5A) and regulatory activity of T_RPM-17 cells were as good as those prior to the Th17-cell condition exposure (Figure 5B).

One major concern regarding the use of ex vivo-expanded Treg cells in immunotherapy trials is the risk of their in vivo conversion into effector cells and loss of suppressive ability. We, therefore, decided to test the stability of T_RPM cells in vivo upon transfer into NOD/scid mice (protocol depicted in Figure 6A). T_MED, T_RPM, or freshly isolated CD3 T cells were injected into total body irradiated (TBI)-NOD/scid mice in the absence of any additional rapamycin treatment. T cells previously cultured ex vivo (i.e., T_MED and T_RPM cells) did not expand in vivo as efficiently as freshly isolated T cells (Figure 6B), despite the fact that the frequency of circulating human CD3CD25 cells was almost identical in the mice receiving either freshly isolated T cells or T_MED and T_RPM cells 13 days following in vivo transfer (Figure 6C). This is likely due to T-cell exhaustion given by the powerful ex vivo TCR-mediated expansion, as previously demonstrated by others.28 Additional treatment of recipient mice with recombinant human IL-2 did not improve human T-cell chimerism in any of the treated mice (data not shown). Despite the lack of evident circulation of human T cells in the peripheral blood of NOD/scid mice injected with T_MED and T_RPM cells, human cells were found in their spleens and lymph nodes. We, therefore, isolated human T cells from secondary lymphoid organs of NOD/scid mice 19 days after adoptive cell transfer. Interestingly, the recovered T_RPM (i.e., T_RPM-MOUSE) cells remained FOXP3 and almost all the FOXP3 cells were Helios–expressing cells and contained no CCR6CD161 precursor/committed Th17 cells (Figure 6D). The cytokine production profile of T_RPM-MOUSE cells changed slightly towards more IL-2 and IFN-γ producing cells (average 11% and 8%, respectively). Conversely, IL-17 cells were almost undetectable in all recovered T_RPM-MOUSE cells (Figure 6E). Unfortunately, the number of T_MED cells recovered from the injected mice was too low to perform the above-mentioned tests. Importantly, T_RPM cells retained their suppressive activity even 19 days after in vivo “parking” in TBI-treated host mice (Figure 6F) while neither unexpanded T cells nor T_MED cells recovered from injected mice displayed any suppressive activity (data not shown).

Figure 3.Cytokine production profile of 14-day cultured CD25⁺ T cells. (A) The frequency of T_MED and T_RPM FOXP3^+/− cells producing the indicated cytokines was established by FACS. The percentages of FOXP3⁺ (upper right) and FOXP3⁻(lower right) cells producing the given cytokines upon 6-h activation in one representative experiment out of three are shown. FOXP3 expression prior and after 6-hour activation is shown. (B) The average cytokine production profile, irrespective of FOXP3 expression, upon 6-h activation of CD25⁺ T_MED (empty bars) and CD25⁺ T_RPM cells (filled bars) determined by FACS (n=3, mean ± SEM).

Overall, these in vitro and in vivo data strongly suggest that rapamycin not only allows the ex vivo expansion of Treg cells voided of inflammatory-producing cells but also fixes their phenotype.

Discussion

Clinical trials with ex vivo-expanded human FOXP3 Treg cells isolated from peripheral blood are currently under question mainly because purified CD25 T cells may contain contaminant non-Treg cells.16,17 The current study shows that a pre-clinical grade protocol for the expansion of functional peripheral Treg cells with rapamycin is feasible. In addition, we demonstrate that rapamycin-expanded human Treg cells are not contaminated by Th17 cells and that they maintain a stable phenotype even upon in vivo transfer with no need for additional rapamycin treatment.

Figure 4.Th17 cell-content in CD25⁺ T cells and in 14-day cultured cells. (A) The frequency of CCR6⁺CD161⁺ cells (gated on CD4⁺ T cells) in the unfractionated PBMC and in the purified CD25⁺ starting T cell subset was established by FACS. One representative dot plot is shown on the left and all the results generated in three separate experiments are included in the right histogram (mean ± SEM). (B) The frequency of IL-17⁺IFN-γ⁻ cells in CD25⁺ T_MED and in CD25⁺ T_RPM cells was established by FACS. One representative dot plot is shown on the left and all the results generated in three separate experiments are included in the right histogram (mean ± SEM). (C) The frequency of CCR6⁺CD161⁺ cells (gated on CD4⁺ T cells ) in T_MED and T_RPM cells was established by FACS. One representative dot plot is shown on the left and all the results generated in three separate experiments are included in the right histogram (mean ± SEM). Data sets were compared by using the t-test. (D) The diagram outlines the time points (D: day) and the procedure followed for the expansion of Th17 cells starting from CD25⁺ T_MED and CD25⁺ T_RPM cells. (E) The frequency of IL-17⁺IFN-γ-cells in CD25⁺ T_MED-17 and CD25⁺ T_RPM-17 culture was established by FACS. One representative dot plot is shown on the left and all the results are included in the right vertical scatter plot in which each dot represents one experiment. The data are divided into high (left scatter plot) and low IL-17–producers (right scatter plot) to better appreciate the differences between CD25⁺ T_MED-17 and CD25⁺ T_RPM-17 cells. All data sets were compared by using the paired t-test and the differences were statistically significant (P=0.036).

We previously reported that rapamycin permits the preferential enrichment and subsequent expansion of mouse11 and human12 FOXP3 Treg cells from a population of unfractionated CD4 T cells. In these previous studies we showed that three rounds of powerful in vitro stimulation were required to expand an average of 30% FOXP3 cells with sustained suppressive activity. In this work we aimed to start with a purer Treg-cell fraction using GMP-grade available magnetic beads and to shorten the cell culture. Interestingly, the purity of our starting cell population was very similar to that recently reported by Brunstein et al. who have just performed a clinical trial in humans with ex vivo purified and expanded cord-blood derived CD25 T cells.4 This further supports the validity of our approach for expanding clinical-grade peripheral CD25 T cells. In addition, it is clear that defining Treg cell dosing strategies in new clinical trials requires large-scale Treg-cell expansion capacity. The protocol described here led to an average 45-fold expansion of our starting Treg-cell fraction. This increase is limited as compared to that of other protocols performed in the absence of rapamycin2 but is absolutely sufficient to obtain, for instance, 3×10/kg of expanded cells (for an average patient of 60 kg), which is the highest dose shown to be effective and safe in cord blood-transplanted patients.4

We believe that the novelty of our work lies in two major aspects. First, Th17 cells do not differentiate and expand upon ex vivo culture of magnetically-isolated CD25 Treg cells only in the presence of rapamycin. The recent findings demonstrating that memory FOXP3 T cells contain some IL-17-producing non-Treg cells16,17 and that naïve FOXP3 Treg cells differentiate into Th17 cells in the presence of IL-2 and lineage-specific polarizing factors25 raise significant doubts about the safety of the protocols commonly used for Treg-cell expansion. The experimental procedure must in fact prevent the growth and development of IL-17-producing or IL-17-precursor cells, which could contaminate the starting T-cell fraction. In our experimental conditions, the starting CD25 T-cell subset did indeed contain a certain number of already committed Th17 cells or their precursors26,27 which expanded efficiently over the 14-day culture period only in the absence of rapamycin. Importantly, the subsequent exposure of T_RPM cells to a Th17-favorable environment in the absence of new addition of rapamycin to the culture did not lead to Th17-cell differentiation or expansion. These data demonstrate that rapamycin imprints a fixed phenotype, not dependent upon the chronic presence of rapamycin, to the expanded Treg cells. The molecular mechanism underlying such a phenomenon is currently under active investigation.

It has been shown that rapamycin reduces de novo, in vitro generation of IL-17-producing cells by TGF-β+IL-6 while it permits TGF-β-induced development of FOXP3 cells starting from murine CD4 T cells.29 This could be ascribed to blockade of the biological activity of IL-6 by rapamycin.29 Our new findings demonstrate that rapamcyin has an inhibitory effect on human Th17-cell growth and development starting from a Treg-cell–enriched fraction and, importantly, it enables the expansion of Treg cells rich in Helios cells and void of Th17-cell contaminants.

The second novel aspect of our work is the demonstration that T_RPM cells preserve their regulatory function even upon in vivo transfer into TBI-treated immunodeficient mice in the absence of any additional rapamycin treatment, thus further supporting their fixed phenotype. The in vivo suppressive potential of FOXP3 Treg cells, freshly isolated30 or previously expanded in the presence31 or absence32,33 of rapamycin, has already been shown in xeno-graft-versus-host disease (GvHD) murine models. However, to our knowledge, to date no experiments have been reported in which only ex vivo-expanded Treg cells have been transferred into TBI-treated NOD/scid mice and subsequently recovered. It is well known that pre-transplant irradiation of immunodeficient mice prompts a massive cytokine storm, which promotes human T-cell engraftment, activation and development of lethal xeno-GvHD.34 The TBI-induced release of TNF-α, IL-1, and IL-6 promotes the activation of host antigen-presenting cells and up-regulation of MHC antigens,35 which are crucial for the induction of lethal xeno-GvHD by human T cells.36 We found that expanded Treg cells have a reduced in vivo proliferative capacity as compared to the same number of freshly isolated T cells, confirming data from the past performed with total T cells.28 However, the transferred T cells did survive for as long as 19 days after in vivo transfer (latest time point analyzed) and despite being exposed to the critical TBI-induced cytokine storm their phenotype and suppressive activity remained stable.

Figure 5.FOXP3 cell content and suppressive ability of CD25⁺T-17 cells. (A) The frequency of FOXP3⁺ T cells within CD25⁺T_MED and CD25⁺T_RPM prior and after Th17 cell expansion was established by FACS. One representative dot plot out of four independent experiments is shown. (B) The suppressive ability of CD25⁺ T_MED, CD25⁺ T_MED-17, CD25⁺ T_RPM cells and CD25⁺ T_RPM-17 cells was tested by co-culture experiments with CSFE-labeled PBMC (responder cells) at a ratio of 0.3:1 (expanded T cells : responder cells). The amount of CD4⁺ and CD8⁺ CFSE⁺ T cells proliferating in the absence or presence of cultured T cells was calculated as described in the Design and Methods section. Percentage of suppressive activity of each tested expanded T-cell populations and the average suppressive function are shown (bars and numbers). Differences that are statistically significant are highlighted by an asterisk (*).

It is evident that many issues need to be resolved prior to using ex vivo-expanded human peripheral FOXP3 Treg cells in clinical trials.2 A common point of discussion among scientists with a strong interest in Treg-cell therapies is the limit of cytokine-producing expanded Treg cells that can be safely infused into a patient. Our data show that, upon in vivo transfer, Treg cells skew their cytokine production profile a bit more towards INF-γ/IL-2-producing cells but, importantly, this profile does not lead to a reduced regulatory function and the mere transfer of these cells in vivo does not cause xeno-GvHD. We, therefore, believe that rather than setting a “cytokine-producing Treg cell threshold” one should set up appropriate tests to monitor the regulatory stability of the transferred cells.

Figure 6 A-C.In vivo plasticity of 14-day cultured CD25⁺ T cells. (A) The diagram outlines the time points (D: day) and the procedure followed for the in vivo transfer into NOD/scid mice of freshly isolated human CD3⁺ T cells, or 14-day cultured CD25⁺ T_MED and CD25⁺ T_RPM. (B) Human chimerism detected in the peripheral blood of NOD/scid mice upon day 13 and day 19 of in vivo transfer of the indicated cells. (C) The frequency of circulating human CD3⁺CD25⁺ T cells was established by FACS by gating on human-CD45⁺ cells 13 days following in vivo transfer. One dot plot representative of one mouse out of three mice per group is shown.

Figure 6 D-F.(D) The phenotype of CD25⁺T_RPM-MOUSE cells recovered 19 days after transfer into NOD/scid mice was tested. The frequency of CD25⁺FOXP3⁺T cells was established by FACS by gating on hu-CD45⁺CD4⁺ cells (left dot plot); the frequency of Helios⁺ cells was established by FACS by gating on hu-CD45⁺CD4⁺FOXP3⁺CD25⁺ cells (middle dot plot); and the frequency of CD161⁺CCR6⁺ T cells was established by FACS by gating on hu-CD45⁺CD4⁺ cells (right dot plot). One dot plot representative of one mouse out of three mice analyzed is shown. (E) The cytokine production profile of CD25⁺T_RPM-MOUSE cells recovered 19 days after transfer into NOD/scid mice was tested. The frequency of IL-17⁺ and IFN-γ⁺ T cells (left dot plot) and the frequency of IFN-γ⁺/IL-2⁺ T cells (right dot plot) was established by FACS by gating on human CD45⁺CD4⁺ cells. One dot plot representative of one mouse out of three mice tested is shown. (F) The suppressive ability of CD25⁺T_RPM-MOUSE cells recovered 19 days after transfer into NOD/scid mice was tested. CSFE-labeled allogeneic PBMC (responder cells) were cultured with CD25⁺T_RPM-MOUSE cells at a ratio of 0.3:1 (upper panel) and 1:1 (lower panel) (T_RPM-MOUSE cells: responder cells). The amount of CFSE⁺ T cells proliferating in the absence (dotted lines) or presence (solid lines) of the given cells was calculated as described in the Design and Methods section. Percentages of suppressive activity of each tested T-cell population are shown. One representative experiment is shown.

In summary, the demonstration that Treg-enriched cells expanded in the presence of rapamycin lack contaminant Th17 cells and are stable upon in vivo transfer in a murine model of TBI-induced inflammation in the absence of rapamycin administration provides additional justification for the clinical use of these cells in future cell therapy-based trials.

Acknowledgments

We thank all Manuela Battaglia’s and Maria-Grazia Roncarolo’s laboratory staff (San Raffaele Scientific Institute, Milano, Italy) for helpful and continuous discussion. We also thank Prof. Toshiyuki Tanaka (Osaka University, Osaka, Japan) for kindly providing the TMβ-1 hybridoma.

Footnotes

• Authorship and Disclosures The information provided by the authors about contributions from persons listed as authors and in acknowledgments is available with the full text of this paper at www.haematologica.org.
• Financial and other disclosures provided by the authors using the ICMJE (www.icmje.org) Uniform Format for Disclosure of Competing Interests are also available at www.haematologica.org.
• The online version of this article has a Supplementary Appendix.

• Received January 28, 2011.
• Revision received April 15, 2011.
• Accepted May 5, 2011.