Obtaining regulatory T cells from uraemic patients awaiting kidney transplantation for use in clinical trials

Authors


Correspondence: D. Berglund, Department of Surgical Sciences, Section of Transplantation Surgery, Uppsala University, Akademiska sjukhuset, 751 85 Uppsala, Sweden.

E-mail: david.berglund@surgsci.uu.se, david.berglund@igp.uu.se

Summary

Adoptive transfer of regulatory T cells (Tregs) has been proposed for use as a cellular therapy to induce transplantation tolerance. Preclinical data are encouraging, and clinical trials with Treg therapy are anticipated. In this study, we investigate different strategies for the isolation and expansion of CD4+CD25highCD127low Tregs from uraemic patients. We use allogeneic dendritic cells (DCs) as feeder cells for the expansion and compare Treg preparations isolated by either fluorescence activated cell sorting (FACS) or magnetic activated cell sorting (MACS) that have been expanded subsequently with either mature or tolerogenic DCs. Expanded Treg preparations have been characterized by their purity, cytokine production and in-vitro suppressive ability. The results show that Treg preparations can be isolated from uraemic patients by both FACS and MACS. Also, the type of feeder cells used in the expansion affects both the purity and the functional properties of the Treg preparations. In particular, FACS-sorted Treg preparations expanded with mature DCs secrete more interleukin (IL)-10 and granzyme B than FACS-sorted Treg preparations expanded with tolerogenic DCs. This is a direct comparison between different isolation techniques and expansion protocols with Tregs from uraemic patients that may guide future efforts to produce clinical-grade Tregs for use in kidney transplantation.

Introduction

Today, solid organ transplantation necessitates the use of lifelong treatment with immunosuppressive drugs. As a consequence, transplanted patients experience increased risks of infections [1, 2] and malignancies [3, 4], as well as drug-specific side effects [5]. At the same time many allografts are lost due to chronic graft dysfunction [6-8]. Only rarely can the immunosuppressive treatment be discontinued in kidney recipients with maintenance of good allograft function, defined as operational tolerance (OT) [9-13]. Therefore, the induction and detection of OT after organ transplantation are areas of intense investigation, and substantial progress has been made recently in these fields [14-16]. Tolerance has also been described after simultaneous renal and haematopoietic stem cell transplantation [17-19]; however, this requires a conditioning regimen and is not without risk. Furthermore, OT alone has been shown to prevent chronic graft dysfunction [20].

Regulatory T cells (Tregs) have been proposed for use as a cellular therapy to induce OT after organ transplantation [21-25], the main approach being isolation of Tregs from prospective transplant recipients followed by ex-vivo expansion and subsequent reintroduction into the patient. Preclinical data are encouraging [26-31], and although many questions remain regarding human Treg therapy they are likely to be answered only by well-designed clinical trials. Recent trials have demonstrated a therapeutic effect of Tregs for the treatment/prevention of human graft-versus-host disease (GVHD) in recipients of bone marrow transplants [32-34].

Treg preparations can be isolated by either fluorescence activated cell sorting (FACS) or magnetic activated cell sorting (MACS) and most protocols use the phenotypic combination of CD4+CD25highCD127low. This phenotype yields high frequencies of CD4+CD25highforkhead box protein 3 (FoxP3)+ T cells with immunoregulatory properties [35]. Although FACS remains a challenging technique it allows for high purities of Tregs, and in addition good manufacturing practice (GMP)-compliant cell sorters are entering the market. Once sorted there are many strategies for the expansion of Tregs. For instance, the use of allogeneic feeder cells makes antigen-specific expansion possible [36, 37].

End-stage renal failure and dialysis affect both innate and adaptive immunity [38]. For the successful completion of clinical trials with Tregs in renal transplantation more data are needed on the functional properties of Tregs isolated from patients with renal failure. In this study, we investigate different strategies for the isolation and expansion of CD4+CD25highCD127low Tregs from uraemic patients. In our protocol, allogeneic dendritic cells (DCs) were used as feeder cells for Treg expansion and we compared the functional properties of Treg preparations isolated by either FACS or MACS and expanded subsequently with either mature DCs (mDC) or tolerogenic DCs (DC-10). Furthermore, we include the rapid (7 h or overnight) assessment of Treg-mediated inhibition of CD69 expression on responding T cells that may be used as a test of suppression in lieu of the gold standard carboxyfluorescein diacetate succinimidyl ester (CFSE) assay.

Materials and methods

Patients

Uraemic patients awaiting living donor kidney transplantation were enrolled into the study. A total of nine patients were included, all with stage V chronic kidney disease at the time of inclusion (Table 1). All patients had given written informed consent prior to leukapheresis and ethical approval of the study was obtained from the Uppsala regional ethical review board (Dnr 2010-069). As a reference, five healthy blood donors were used as controls for assessing the purity of expanded Treg preparations.

Table 1. Demographics of the uraemic patients included in the present study.
AgeGenderCause of uraemiaType of dialysis
24MaleNephrotic syndromeHaemodialysis
40MalePolycystic kidney diseaseHaemodialysis
55FemaleImmune complex glomerulonephritisHaemodialysis
58FemaleUnknownPeritoneal dialysis
61MaleImmunoglobulin A nephritisPeritoneal dialysis
65FemaleMesangioproliferative glomerulonephritis with vasculitisHaemodialysis
66FemaleRecurrent pyelonephritisHaemodialysis
66MaleMesangiocapillary glomerulonephritisCombined peritoneal and haemodialysis
68MaleHypertensionPredialytic (about to start haemodialysis)

All reagents used in the study were of non-clinical grade, and equivalent GMP-compliant reagents is advocated when producing Treg preparations for the use in clinical trials.

Isolation of CD4+ cells

Peripheral blood mononuclear cells (PBMCs) were obtained from the uraemic patients by leukapheresis for 60 min and subsequent centrifugation at 210 g for 30 min over a Ficoll-Paque gradient (GE Healthcare, Uppsala, Sweden). Adherent cells were removed by incubation in T175 flasks for 2 h at 37°C in complete media (CM) consisting of RPMI-1640 (Gibco, Invitrogen, Carlsbad, CA, USA) with 1% penicillin–streptomycin, 1% 4-(2-hydroxyethyl)-1-piperazine-ethanesulphonic acid (HEPES), 0·5% L-glutamine, 0·04% β-mercaptoethanol and supplemented with 2% pooled human AB serum (pooled, sterile-filtered and heat-inactivated AB serum from 15 to 20 healthy blood donors, tested for pathogenic contamination according to hospital standards). The non-adherent cells were separated further into CD4+ cells by negative MACS selection (Miltenyi Biotec, Bergisch Gladbach, Germany), the reagents were titrated and separation was performed according to the manufacturer's instructions. At least 30 × 106 CD4+ T cells were cryopreserved for later use in functional assays using 45% CM, 45% human AB serum and 10% dimethyl sulphoxide (DMSO; Sigma, St Louis, MO, USA).

FACS for CD4+CD25highCD127low T cells

The pre-enriched CD4+ cells were cultured overnight in CM with 10% AB serum and low-dose interleukin (IL)-2 (30 U/ml). The cells were stained subsequently with the following antibodies: CD4-fluorescein isothiocyanate (FITC), CD25-phycoerythrin (PE) and CD127-allophycocyanin (APC) (all from BD Biosciences, San Jose, CA, USA). Staining was performed in CM for optimal cell viability. A sample of the cells was also stained with 7- aminoactinomycin (7-AAD) (Via-Probe; BD Biosciences) to assess cell viability. After staining, the cells were filtered through a cell strainer cap with a 35-μm nylon mesh and resuspended in CM at 30 × 106 cells/ml. Next, the cells were sorted for CD4+CD25highCD127low using a FACSAria III (BD Biosciences). Post-sort analysis was performed to confirm purity. Sorted cells were collected in CM with 10% AB serum.

MACS for CD4+CD25+CD127dim/– T cells

To compare phenotypical and functional differences between FACS- and MACS-isolated Tregs from uraemic patients some CD4+ cells were separated further by MACS into CD4+CD25+CD127dim/– T cells, according to the manufacturer's instructions (Miltenyi Biotec).

Differentiation of dendritic cells

Plastic adherent cells were obtained from healthy blood donors or uraemic patients (as described above) and differentiated subsequently into either mature (mDC) or tolerogenic dendritic cells (DC-10), as described previously [39].

Briefly, adherent cells were differentiated into mDC by culturing in CM with 10% AB serum supplemented with recombinant human granulocyte–macrophage colony-stimulating factor (rhGM-CSF) (100 ng/ml) and rhIL-4 (10 ng/ml) for 5 days. On days 3 and 5 half the media was replaced and rhGM-CSF and rhIL-4 was replenished in the original concentrations. On day 6 lipopolysaccharide (LPS) was added (1 μg/ml) and the cells were harvested on day 7 by trypsin digestion and gentle scraping.

Adherent cells were differentiated into DC-10 by culturing in CM with 10% AB serum supplemented with rhGM-CSF (100 ng/ml), rhIL-4 (10 ng/ml) and rhIL-10 (10 ng/ml) for 7 days. On days 3 and 5 half the media was replaced and rhGM-CSF, rhIL-4 and rhIL-10 was replenished in the original concentrations. As described by Roncarolo et al., DC-10 are non-adherent and only the non-adherent fraction was collected at day 7 [39].

To ascertain that the protocols for differentiation of mDC and DC-10 result in DCs with the desired phenotypical characteristics, the protocols were tested on the adherent cells from three different healthy blood donors.

Expansion of isolated CD4+CD25high/+CD127low/– T cells

CD4+CD25highCD127low T cells isolated by FACS or CD4+CD25+CD127dim/– T cells isolated by MACS were expanded using a combination of DCs, IL-2 and anti-CD3 antibody. The rationale for this is to provide Tregs with antigen-specific signals through the addition of DCs, while IL-2 and anti-CD3 provides additional proliferation signals to increase the expansion further. Tregs were co-cultured with irradiated (25 Gy) allogeneic dendritic cells (mDC or DC-10) at a 1:1–1:5 DC : Treg ratio. Initially, T cells and DCs were cultured overnight in 6- or 12-well plates in CM with 10% AB serum and 30 U/ml IL-2. The following day anti-CD3 antibody (okt3, 30 ng/ml, Ortho Biotech, NJ, USA) and IL-2 (500 U/ml, Proleukin, Novartis, Basel, Switzerland) were added. Every second day, approximately half the media was replaced and new IL-2 was added. The expansions were ongoing for 12–20 days. When needed, as assessed by visual inspection cell expansions were split and/or transferred to larger wells or culture flasks.

Phenotypic analysis

The phenotype of isolated and ex-vivo expanded Tregs was evaluated by flow cytometry using the following conjugated monoclonal antibodies: CD4-FITC (BD Biosciences), CD25-PE (BD Biosciences) and FoxP3-APC (eBioscience, San Diego, CA, USA; clone 236A/E7). After surface staining with CD4 and CD25 cells were fixed and permeabilized for 30 min using a FoxP3 staining buffer kit (eBioscience), according to the manufacturer's instructions.

The phenotypes of mDC and DC-10 were assessed by surface staining with monoclonal antibodies directed against CD1a, CD14, CD40 and CD86 (all from BD Biosciences). Cells were analysed using a FACSCanto II cytometer (BD Biosciences).

Treg suppression assay

Autologous CD4+ T cells, cryopreserved previously from the future kidney recipient, were thawed and used as responding cells (Tresp) in the Treg suppression assay. The Tresp (1 × 105 cells) were stimulated with CD3/CD28 beads (Dynal; Invitrogen) at a 4:1 bead : cell ratio (titrated previously to give maximal stimulation) in a 96-well round-bottomed plate. Tregs rested previously overnight in 10 U/ml IL-2 were added at a 1:1 Treg: Tresp ratio to a final volume of 300 μL in CM with 10% human AB serum. In some experiments the Treg: Tresp ratio was titrated 1:1, 1:2, 1:4, 1:8, 1:16 and 1:32 to confirm a Treg dose–response. Appropriate controls were included, such as stimulated and unstimulated Tresp, non-Treg controls that mimic Tregs (to exclude Tresp suppression due to overgrowth) and Treg alone (to assess Treg viability).

After 7 h, or overnight, the CD69 expression was assessed on Tresp by flow cytometry by gating only on CD4+CD25 events, as described previously [40]. As Tregs also express CD69 upon activation, this gating strategy was used to exclude Tregs from the analysis.

Tresp used in the 4-day proliferation assay were labelled with carboxyfluorescein diacetate succinimidyl ester (CFSE; Sigma-Aldrich, St Louis, MO, USA), according to the manufacturer's instructions. Briefly, 1 × 106 cells/ml were stained with a working solution of 10 μM CFSE in phosphate-buffered saline (PBS) with 0·1% human serum albumin (HSA) for 10 min at 37°C. The staining was quenched by adding 5 volumes of ice-cold CM and put on ice for 5 min. The cells were finally washed three times in CM. Suppression in the CFSE assay was determined after 4 days using the proliferation platform in FlowJo (OS version 7·6·5; Tree Star Inc., Ashland, OR, USA). The division index (DI, average number of times each cell has divided) was calculated by FlowJo and the percentage of suppression (S) was computed as:

display math

where a is the DI in the presence of Tregs and b the DI in the absence of Tregs.

Cytokine profile

The cytokine profile of expanded Tregs was assessed by co-culturing Tregs with allogeneic mDC from the same third-party donor used in the expansion, at a 1:1 Treg : mDC ratio. Cells were cultured in round-bottomed 96-well plates with CM and 10% human AB serum. Treg alone and mDC alone were included as controls. Supernatants were collected after 72 h and analysed for the following cytokines: IL-2, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12, IL-13, IL-15, 1L-17, IL-21, IL-22, IL-31, IL-33, IL-35, chemokine (C–C) ligand 20 (CCL20), interferon (IFN)-γ, tumour necrosis factor (TNF)-α and transforming growth factor (TGF)-β. All soluble mediators were analysed by Luminex xMAP technology (Merck Millipore, Darmstadt, Germany), except for TGF-β, IL-35 and granzyme B, which were analysed separately by enzyme-linked immunosorbent assay (ELISA) (from eBioscience), according to the manufacturer's instructions. The ability of Tregs to secrete a specific cytokine upon stimulation was computed as:

display math

where Tregstim and Tregnon-stim denote the amount (pg/ml) of cytokine secretion from mDC-stimulated and non-stimulated Tregs, respectively, while mDC denotes the amount (pg/ml) of cytokine secretion from mDC alone.

Sterility testing

Expanded Tregs were analysed for contamination by endotoxin testing and microbiological sampling, as used routinely in the Department of Clinical Microbiology (accredited by the Swedish Board for Accreditation and Conformity Assessment) at Uppsala University Hospital.

Karyotyping

To rule out a potential for malignant transformation due to any large chromosomal abnormalities, karyotyping was performed on expanded Tregs. The analyses were performed according to standard procedures at the Cytogenetic Laboratory (accredited by the Swedish Board for Accreditation and Conformity Assessment) of the Sahlgrenska University Hospital (Gothenburg, Sweden).

Statistical analysis

Paired Student's t-test or Wilcoxon's matched-pairs signed-rank test were used to compare cells sorted by FACS or MACS and expanded subsequently by mDC or DC-10. Wilcoxon's matched-pairs signed-rank test was used when the Kolmogorov–Smirnov test of normality failed. The Mann–Whitney U-test was used to compare the purities of Treg preparations from uraemic patients and healthy blood donors. A P-value <0·05 was considered statistically significant; * is used to denote P < 0·05, ** denotes P < 0·01, and *** denotes P < 0·001. Data are presented as mean ± standard error of the mean (s.e.m.).

GraphPad Prism, version 5·0c, was used for all statistical computations. The FlowJo software (OS version 7·6·5; Tree Star Inc.) was used for processing and visualizing flow cytometric data.

Results

Isolation and expansion of CD4+CD25highCD127low and CD4+CD25+CD127dim/– T cells

In this study, nine patients waiting for living donor kidney transplantation underwent leukapheresis without adverse events during any of the procedures. An overview of the process used in this study to obtain Treg preparations is shown in Fig. 1. The leukapheresis products were purified further by MACS for CD4+ T cells and by FACS for CD4+CD25highCD127low T cells. One part of the leukapheresis product was used for the FACS-sorting into CD4+CD25highCD127low T cells (Fig. 2a) and another part was used for isolating CD4+CD25+CD127dim/– T cells by MACS. Post-sort analysis after FACS showed that the sorted products occasionally displayed an increase in small-sized cell debris with a concomitant loss in cell number, which indicates that cell viability was affected. However, in general the sorted cells were of good purity, with only small amounts of debris (Fig. 2b). The sorted CD4+CD25highCD127low cells constituted between 2·2 and 6·7% of the total number of CD4+ cells (Supporting information, Fig. S1).

Figure 1.

Overview of the regulatory T cell (Treg) production process. Step 1: the starting material is comprised of peripheral blood mononuclear cells (PBMCs) collected from both the intended organ donor and recipient. We prefer the use of leukapheresis, because this greatly increases the yield. Step 2: the PBMCs are purified further over a Ficoll density gradient. Step 3: the PBMCs are separated with regard to their adherent properties by which monocytes (attaching to plastic) and lymphocytes (non-plastic adherent) are separated. Step 4: Tregs with a phenotype of CD4+CD25highCD127low are sorted from the non-plastic adherent cells of the potential organ recipient using fluorescence activated cell sorting (FACS), with the possibility of performing a pre-enrichment with magnetic activated cell sorting (MACS) for CD4+ T cells prior to FACS. Monocytes from the potential organ donor are stimulated with granulocyte–macrophage colony-stimulating factor (GM-CSF), interleukin (IL)-4 and lipopolysaccharide (LPS) by which the cells differentiate into mature dendritic cells (mDC). Some of these cells are cryopreserved for use in possible restimulations and functional analyses. Step 5: Tregs from the potential organ recipient are mixed with irradiated mDC from the potential organ donor. The culture is supplemented with anti-CD3 antibody and IL-2. Steps 6–8: the Treg product is assessed for sterility, function, etc. If the cells meet the set release criteria for the particular study they can be administered to the patient.

Figure 2.

Sorting CD4+CD25highCD127low cells from a representative uraemic patient. (a) First, the lymphocyte population was selected using their characteristics in forward- (FSC) and side-scatter (SSC). Next, the CD25high expressing cells were selected. As evident in the dot-plot, these cells expressed on average slightly lower levels of CD4 and the inferior boundary of the gate was set to include this putative population. Finally, cells expressing low levels of CD127 were selected. Only cells present in all three gates were sorted. (b) Post-sort analysis was performed after the sorting. In general, post-sort purity was satisfying, but we occasionally noted decreased cell viability, as evident by an increased amount of cell debris.

Sorted T cells were mixed with either allogeneic mDC or DC-10 overnight, under resting conditions, with low-dose IL-2 (10 U/ml). The phenotypes of mDC and DC-10 corresponded with that described previously [39] with regard to their expression of CD1a, CD14, CD40 and CD86 (Supporting information, Fig. S2). The next day, anti-CD3 antibody (30 ng/ml) and IL-2 (500 U/ml) were added. Typically, expanding clusters of T cells could be observed by light microscopy after 2–3 days (Supporting information, Fig. S3a) and upon closer visualization T cells were clustering around individual DCs (Supporting information, Fig. S3b). The expansion protocol with mDC increased the number of cells up to ∼200-fold (Fig. 3) and the expansion was ongoing for 12–20 days (median 14 days), after which the cells were harvested and rested overnight in low-dose IL-2 (10 U/ml). No difference in the expansion was observed between any of the Treg preparations. The next day, phenotypic analysis was performed on the rested Tregs (Fig. 4a,b). Treg preparations sorted by FACS and expanded with mDC were, on average, 72% CD4+CD25highCD127low cells, but were not significantly higher than Treg preparations sorted by MACS and expanded with mDC. The contaminating cell population consisted of CD4+ lymphocytes not positive for both CD25 and FoxP3 (Fig. 4a).

Figure 3.

Expansion of CD4+CD25highCD127low T cells using dendritic cells (DCs). The number of fluorescence activated cell sorting (FACS)-sorted CD4+CD25highCD127low cells and the corresponding number of cells after expansion with mature DCs (mDC). There is no apparent correlation between the number of sorted regulatory T cells (Tregs) and the total yield after expansion.

Figure 4.

Phenotype of expanded regulatory T cells (Tregs). (a) The expanded cells constituted a coherent population. Here, a representative example of Tregs expanded with mature dendritic cells (mDC) is shown. (b) Treg preparations sorted by fluorescence activated cell sorting (FACS) and expanded with mDC were purer than Treg preparations expanded with DC-10 (P < 0·05). The Treg preparations sorted by magnetic activated cell sorting (MACS) dropped occasionally to purities <20%, whereas Treg preparations sorted by FACS were, on average, 72% pure.

If the original sort is performed using flow cytometry, a post-expansion purity of 72% is somewhat low. We therefore compared the purities of flow-sorted mDC-expanded Treg preparations from uraemic patients and healthy blood donors. The Treg preparations from healthy blood donors displayed higher purities compared to Treg preparations from uraemic patients, 92·6 versus 72·0% (P = 0·0051, Supporting information, Fig. S4).

Treg suppression assay

In the Treg suppression assay CD4+ T cells from the patients awaiting living donor kidney transplantation were used as responding cells, and were stimulated by beads coated with CD3/CD28 antibodies. Autologous Tregs were added to assess the ability to suppress activation and proliferation of CD4+ T cells. The assay was harvested at two time-points, at 7 h or overnight (short-term) and after 4 days (long-term). In the short-term assay, suppression was determined by CD69 expression on CD4+CD25 T cells (Fig. 5a). Tregs sorted by FACS exhibited a superior ability to decrease CD69 expression compared to Tregs sorted by MACS. Furthermore, Tregs sorted by FACS and expanded with mDC were superior to Tregs sorted by MACS and expanded with DC-10 (Fig. 5b).

Figure 5.

Short-term functional assessment of regulatory T cells (Tregs) from uraemic patients. (a) Gating for CD69-expressing responder cells. Unstained responder cells (top row) and stained, non-stimulated responder cells (second row) were included as controls. In the third column (CD69+ gate) only the CD25 cells are shown. Suppression was calculated as the ratio of CD69 expression on stimulated responder cells without (third row) and with (fourth row) the addition of autologous Tregs. (b) Treg preparations sorted by fluorescence activated cell sorting (FACS) and expanded with mDC were superior to Treg preparations sorted by magnetic activated cell sorting (MACS) and expanded with either mature dendritic cells (mDC) or DC-10 (P < 0·05 and P < 0·001, respectively). Furthermore, suppression of CD69 expression in FACS-sorted and mDC-expanded Treg preparations was consistently more than 20%.

In the long-term assay, suppression was determined by the decrease in proliferation of CFSE-labelled CD4+ T cells. Again, the suppressive ability of autologous Tregs sorted by FACS and expanded with mDC was higher compared to those sorted by MACS and expanded by DC-10 (Fig. 6a).

Figure 6.

Long-term functional assessment of regulatory T cells (Tregs) from uraemic patients. (a) In the long-term suppression assay, assessed by inhibition of carboxyfluorescein diacetate succinimidyl ester (CFSE)-labelled responder cells, Treg preparations sorted by fluorescence activated cell sorting (FACS) and expanded with mature dendritic cells (mDC) were superior to Treg preparations sorted by magnetic activated cell sorting (MACS) and expanded with DC-10. (b) Titration of the Treg : Tresp ratio. A representative titration of the ratio between Tregs, sorted by FACS and expanded with mDC, and autologous CD4+ cells. About 50% suppression was observed even at Treg : Tresp ratios of 1:32. Titrations beyond 1:32 were not performed.

Upon titration of the Treg : CD4 ratio, almost 50% suppression in the CFSE assay was noted at a ratio of 1:32 (Fig. 6b). Further titration below a ratio of 1:32 was not performed.

Cytokine profile

Cytokine concentrations were assessed in the supernatant after co-culturing Tregs with allogeneic mDC. The following cytokines did not show any measurable levels: IL-4, IL-6, IL-9, IL-12p70, IL-21, IL-22, IL-31, IL-33, IL-35 and TGF-β.

Upon stimulation, Tregs sorted by FACS and expanded with mDC secreted only low amounts of IL-5, IL-13, IL-17, IFN-γ and TNF-α, whereas IL-10 secretion was increased substantially (P < 0·05). Notably, FACS-sorted Tregs expanded with mDC secreted more IL-10 than FACS-sorted Tregs expanded with DC-10. In contrast, Tregs sorted by MACS secreted higher levels of IFN-γ (P < 0·05) upon stimulation and there was a trend (P < 0·1) towards higher secretion of IL-5, IL-13 and TNF-α (Fig. 7a–f).

Figure 7.

Cytokine secretion of expanded regulatory T cells (Tregs). (a) Treg preparations sorted by fluorescence activated cell sorting (FACS) and expanded with mature dendritic cells (mDC) secreted low levels of interferon (IFN)-γ, less than cells sorted by magnetic activated cell sorting (MACS) (P < 0·05). (b) Treg preparations sorted by FACS and expanded with mDC secreted more interleukin (IL)-10 than Treg preparations sorted by FACS and expanded with DC-10 (P < 0·05). (c–e) There was a trend (P < 0·1) towards higher secretion of tumour necrosis factor (TNF)-α, IL-5 and IL-13 in Treg preparations sorted by MACS compared to Tregs sorted by FACS. (f) The secretion of IL-17 was low in general and did not differ between Treg preparations. (g) Treg preparations sorted by FACS and expanded with mDC secreted more granzyme B than Treg preparations expanded with DC-10 (P < 0·05).

The secretion of granzyme B by Tregs was measured in the tissue culture supernatant after stimulation with mDC. Tregs expanded with mDC secreted more granzyme B than Tregs expanded with DC-10 (Fig. 7g).

Sterility and karyotyping

No cell-batches were positive for endotoxin or microbes after up to 3 weeks of expansion and functional testing. In addition, four batches underwent karyotypic analysis with no signs of aberrant chromosomes.

Discussion

Chronic kidney disease is known to affect the immune system and uraemic patients on dialysis suffer from a stage of chronic inflammation [38]. It was shown recently that functional Tregs can still be obtained from uraemic patients [41]. However, there are insufficient data on different modes of sorting and expanding Tregs from patients with renal failure. The present study demonstrates that Tregs can be produced from uraemic patients in numbers that are likely to be sufficient for adoptive cell transfer in the setting of clinical kidney transplantation. The fact that functional Tregs can be obtained from uraemic patients indicates that chronic inflammation in vivo does not abrogate their immunoregulatory potential indefinitely.

CD4+CD25high/+CD127low/– Tregs can be isolated from uraemic patients by both FACS and MACS, and expanded subsequently with either mDC or DC-10. However, Treg preparations sorted by FACS and expanded with mDC have a higher purity of CD4+CD25highFoxP3+ cells compared to Treg preparations expanded with DC-10. In addition, MACS-sorted Treg preparations expanded with mDC occasionally displayed purities of less than 20%. No definitive recommendation can be given regarding the mode of Treg-sorting and -expansion from uraemic patients. We do, however, note that Treg preparations sorted by FACS and expanded with mDC together have some advantages over the other Treg preparations with regard to purity, cytokine secretion and/or suppressive ability.

The expansion protocol for FACS-sorted Tregs, with allogeneic mDC as feeder cells, should be further scalable for use in clinical trials. If not dividing each leukapheresis into four parts, as we have performed in this study, and instead one large expansion is performed, the number of cells could potentially approach a magnitude of 109. Also, there is a possibility of performing additional rounds of stimulation, and preliminary data show that this can further increase the expansion substantially while preserving the phenotype.

The Treg preparations expanded with mDC displayed a purity of 72%, which is lower than expected when sorting with flow cytometry. We therefore made a comparison with mDC-expanded preparations from healthy blood donors to assess whether this reflects the state of uraemia or is a result of our methodological approach. We observed that the purity of expanded Treg preparations from healthy blood donors was higher at 92·6%, compared to the Treg preparations from uraemic patients. This underlines the importance of assessing Treg preparations in the context of a specific disease state. Furthermore, we speculate that the optimal sorting algorithm for Tregs may differ between patients and healthy blood donors. This has potentially important implications for clinical Treg sorting and deserves further attention.

DC-10, described originally by Roncarolo et al. [39], are known to differentiate naive CD4+ T cells into IL-10-secreting Tr1 cells. We hypothesized that expansion of CD4+CD25highCD127low T cells with DC-10 would yield an increased ability of Tregs to secrete IL-10 and at the same time differentiate any contaminating CD4+ effector cells into Tr1 cells. Somewhat unexpectedly, we noted that FACS-sorted Treg preparations expanded with mDC were superior in secreting IL-10 upon stimulation compared to FACS-sorted Treg preparations expanded with DC-10. This may suggest that Tregs are not affected by DC-10 in the same way as are naive CD4+ T cells. Furthermore, FACS-sorted Treg preparations expanded with mDC secreted more granzyme B than those expanded with DC-10. We therefore suggest that DCs of varying maturity provide Tregs with different functional properties, e.g. with regard to cytokine secretion. This has direct implications when choosing what feeder cells to use in Treg-expansion protocols.

Treg preparations sorted by MACS also secrete high levels of IL-10 upon stimulation. However, this should be interpreted in view of a simultaneously high secretion of IL-5, IL-13, IFN-γ and TNF-α, whereas FACS-sorted Treg preparations secrete low levels of these cytokines. We also note that the secretion of IL-17 is low both for Tregs sorted by FACS and MACS. Thus, FACS-sorted Tregs expanded with mDC secrete the tolerogenic cytokine IL-10 selectively without the addition of non-tolerogenic cytokines. This suggests that FACS- and MACS-sorted Treg preparations have different cytokine profiles. In part, this may be explained by the contamination of effector T cells, which is likely to be greater with MACS. However, it is also possible that the difference in cytokine secretion is not due simply to a small contaminating subpopulation but rather to smaller contributions from most cells. Indeed, we observed a different cytokine profile in expanded Tregs of comparable purities that had been sorted by different techniques (i.e. MACS or FACS). This may suggest that the sorting criteria can affect the functional characteristics of Tregs beyond what can be detected by the CD4+CD25highFoxP3+ phenotype. Indeed, setting the gates for sorting Tregs with FACS is performed by visual inspection, making possible versatile and fine-tuned adjustments. Future studies may hopefully add additional markers to be used in the sorting process and can perhaps also standardize, for example, the optimal CD25high and CD127low gates for FACS-sorting Tregs.

The in-vitro functional assessment of Treg preparations is often based on proliferation assays, e.g. the widely used CFSE assay, where responder cells are labelled with CFSE. However, these assays have the shortcoming of requiring approximately 4 days for completion, at which time the original Tregs may have changed their properties. Assays utilizing humanized mouse models require even more time, and are therefore not an option for clinical trials. In the present study we show that the suppressive ability of Treg preparations from uraemic patients can be assessed rapidly using a flow cytometry-based assay, where the expression of the activation marker CD69 is measured on responder cells. This provides an opportunity for the rapid completion of a Treg suppression assay and may be more appropriate for the use in clinical trials. This has already been suggested by Lord et al. [40], who studied a similar short-term assay in healthy subjects.

Adoptive transfer of Tregs has been tested rigorously in preclinical models, and Lombardi et al. [37] showed recently that alloantigen-specific Tregs from healthy subjects have improved efficacy in preventing graft rejection compared to polyclonal Tregs. We therefore attempted to include DCs from the intended organ donor as alloantigen-specific stimulators in the Treg suppression assays to simulate the in-vivo alloresponse; however, this did not yield a reliable and reproducible activation of the responder cells (results not shown), perhaps because the number of alloreactive cells that is induced to proliferate and divide is relatively low. Albeit not antigen-specific, stimulation with CD3/CD28 beads has proved to be highly reproducible, and for the present we believe that this is an appropriate way to stimulate responder cells in Treg suppression assays. We have also used phytohaemagglutinin (PHA) to stimulate responder cells (results not shown), but CD3/CD28 beads have yielded more consistent stimulations. Nevertheless, a convenient in-vitro assay that can estimate the level of antigen-specificity for Tregs is highly desirable, and should be investigated further in future studies.

The risks associated with the intravenous administration of autologous Tregs expanded ex vivo seem to be small, and no infusion-related adverse events have been noted in any of the clinical trials with Tregs in the treatment of GVHD [32-34] or type 1 diabetes [42]. Also, the risk of malignant transformation is probably minimal; indeed, we observed no abnormalities when karyotyping Tregs that had undergone prolonged expansion for at least 3 weeks. Furthermore, there are extensive data from the field of T cell therapy in cancer, where donor lymphocyte infusions (DLIs) and tumour-infiltrating lymphocytes (TILs) have been administered intravenously without an increased risk for development of lymphomas/leukaemias [43-48].

Many transplant centres are now aiming to initiate Phases I–II clinical trials with adoptive Treg therapy in solid organ transplantation. In this study we present a direct comparison between different isolation techniques and expansion protocols with Tregs from uraemic patients that may guide future efforts to produce clinical grade Tregs for use in renal transplantation.

Acknowledgements

The authors thank Jan Grawé and Dirk Pacholsky at the Science for Life Laboratory BioVis Technology Platform in Uppsala for assistance on the flow cytometric cell-sorting on the FACSAria III. This study was supported by grants from the Tommy and Gösta Andersson Memorial Foundation, the Professor Lars-Erik Gelin Memorial Foundation, the Swedish Medical Research Council (16X-12219, K2011-65X-12219-15-6 and 2008-2205), the Swedish ALF fund and the Juvenile Diabetes Research Foundation International.

Disclosure

The authors of this manuscript have no conflicts of interest to disclose.

Ancillary