Isolation and expansion of thymus‐derived regulatory T cells for use in pediatric heart transplant patients

Regulatory T‐cells (Tregs) are a subset of T cells generated in the thymus with intrinsic immunosuppressive properties. Phase I clinical trials have shown safety and feasibility of Treg infusion to promote immune tolerance and new studies are ongoing to evaluate their efficacy. During heart transplantation, thymic tissue is routinely discarded providing an attractive source of Tregs. In this study, we developed a GMP‐compatible protocol for expanding sorted thymus‐derived CD3+CD4+CD25+CD127– (Tregs) as well as CD3+CD4+CD25+CD127–CD45RA+ (RA+Tregs) cells. We aimed to understand whether thymic RA+Tregs can be isolated and expanded offering an advantage in terms of stability as it has been previously shown for circulating adult CD45RA+ Tregs. We show that both Tregs and RA+Tregs could be expanded in large numbers and the presence of rapamycin is essential to inhibit the growth of IFN‐γ producing cells. High levels of FOXP3, CTLA4, and CD25 expression, demethylation of the FOXP3 promoter, and high suppressive ability were found with no differences between Tregs and RA+Tregs. After freezing and thawing, all Treg preparations maintained their suppressive ability, stability, as well as CD25 and FOXP3 expression. The number of thymic Tregs that could be isolated with our protocol, their fold expansion, and functional characteristics allow the clinical application of this cell population to promote tolerance in pediatric heart transplant patients.


Introduction
Heart transplantation is a lifesaving procedure for pediatric patients with end-stage heart disease. Long-term success is Correspondence: Giovanna Lombardi e-mail: giovanna.lombardi@kcl.ac.uk limited by chronic inflammation and the use of immunosuppressive drugs [1]; therefore, new strategies are needed to induce indefinite transplant survival. So far, phase I clinical trials have shown safety and feasibility of regulatory T-cells (Tregs) therapy in the setting of transplantation and autoimmunity [2]. Similarly to conventional CD4 + T cells, Tregs arise from thymus following maturation. This organ is routinely discarded during heart transplantation and is considered to be an attractive source of autologous Tregs, especially for pediatric patients where it is not possible to collect large amounts of blood. In 2015, Dijke et al. showed for the first time the feasibility of isolating and expanding thymic resident Tregs [3]. These cells showed higher suppressive ability both in vitro and in vivo in comparison to circulating Tregs. Furthermore, the same group evaluated different expansion and freezing protocols for translating this population into the clinic [4].
One of the main challenges when isolating Tregs for clinical use is the purity of the resultant cell product. Following immunomagnetic isolation, thymic Tregs showed a greater purity compared to peripheral blood Tregs; however, 15-20% of the cells were still negative for FOXP3 [3,4]. This is due to the protocols enrolled for cell isolation and the presence of thymic cells expressing CD25 while they lack FOXP3 [5], a population that might be converted in effectors cells [6].
Good manufacturing practice (GMP) compliant cell sorters are now available in the clinic allowing the isolation of pure CD4 + CD25 + CD127cells as well as Treg subpopulation like CD4 + CD25 + CD127 -CD45RA + . We have previously shown that for some specific diseases, such as Crohn, blood-derived expanded CD45RA + Tregs were functionally superior and more stable compared to total CD4 + CD25 + CD127 -Tregs [7]. Furthermore, blood-derived CD45RA + Tregs appear to be the ideal candidate population in the evolving landscape of CAR-Treg technology as they are more stable during expansion [8]. In thymic tissue, CD45RA + Tregs represent the cells that will move into the bloodstream following upregulation of CD31 [9]. As the thymus has the potential to become an alternative Treg source for autologous and also an HLA-matched cell therapies (in the prevention of graft vs. host disease), we aimed to develop an expansion protocol that can be used in future clinical trials with thymic Tregs and compare CD3 + CD4 + CD25 + CD127 -(Tregs) with CD3 + CD4 + CD25 + CD127 -CD45RA + (RA + Tregs) to understand whether the latter offered an advantage.

Evaluation of thymic Tregs: Phenotype and numbers
Thymocytes were obtained from discarded thymi (age range 0.5-149 months, Table 1) following mechanical and enzymatic digestion and the median yield was 0.65 × 10 9 cells per gram of tissue (range 0.35-1.08 × 10 9 thymocytes per gram). This wide range was related to the quality of the tissue collected during the surgery as no correlation between the age of the patients and the number of cells recovered was found (data not shown). Thymocytes were then phenotypically characterized to determine their cellular composition (Supporting Information Fig. S1A and Table S1). The mean percentage of single positive (SP)-CD4 was 35.36 ± 6.5% of the whole thymocytes with 27.21 ± 6.93% of the cells being positive for CD45RA + (Supporting Information Table S1). We then evaluated the proportion of CD25 + CD127cells in the SP-CD4 compartment as the expression of these two markers are routinely used for Treg sorting. CD3 + CD4 + CD8 -CD25 + CD127cells were 11.8 ± 3.63% of the whole SP-CD4 with 84.57 ± 5.68% of them expressing FOXP3 (Supporting Information Fig. S1A and Table  S1). The mean expression of CD3 + CD4 + CD25 + CD127 -FOXP3 + in all the samples was 9.86 ± 2.52% of the total SP-CD4. The difference between the percentage of CD3 + CD4 + CD8 -CD25 + CD127and CD3 + CD4 + CD8 -CD25 + CD127 -FOXP3 + was in line with reports showing the presence of thymic CD25 + cells negative for FOXP3 in both mouse [10] and humans [5]. RA + Tregs represented the 38.7 ± 5.13% of the total Tregs, and no correlation was found between the age of the patients and the expression of CD45RA (data not shown). Next, the expression of functional and activation markers together with chemokine receptors was evaluated. No statistically significant differences were found in the expression of CD39, CTLA-4, TIGIT, HLA-DR, CCR6, CCR7, CD62L, CXCR3, CXCR4, CD38, CD69, and HELIOS between Tregs and RA + Tregs (Supporting Information Fig. S1B and C). Additionally, all CD3 + CD4 + CD8 -CD25 + CD127cells did not  produce IL-2 following stimulation with PMA/Ionomycin (data not shown).
Highly suppressive and stable Tregs are expanded following cell sorting CD3 + CD4 + CD8 -CD25 + lymphocytes were isolated by CD8 + depletion followed by CD25 enrichment ( Table 1). The yield was 2.02 × 10 6 Tregs/g (range 0.59-6.65) and more than 80% of the cells were CD3 + CD4 + CD8 -CD25 + CD127 -(range 72-90.8%) (Supporting Information Fig. S2A and B). To compare the expansion of Tregs and RA + Tregs, the two populations were sorted according to the gating strategy shown in Supporting Information Fig. S2A; while Tregs included 14.3 ± 2.6 CD4 + CD25 + FOXP3cells, this population was almost absent in RA + Tregs (Supporting Information Fig. S2C). Sorted cells were expanded polyclonally (anti-CD3/CD28 beads and IL-2) in the presence or absence of rapamycin. After 36 days, the mean fold expansion rate was greater than 900 in all the Treg preparations (Fig. 1A). The viability of the Tregs was ≥90% in all the preparations (Fig. 1B) and no statistically significant differences were found between the four culture conditions. Of note, a higher fold expansion was obtained after 36 days of culture, compared to previously published protocols where nonsorted cells have been expanded for 15 days [3,4]. Furthermore, all the cells were CD4 + CD25 + CD127and CD4 + CD25 + Foxp3 + (Fig. 1C and D), while the expansion of nonsorted cells lead to a lower FOXP3 expression [4]. The phenotype at the end of the culture (day 36) was comparable between Tregs and RA + Tregs and the presence of rapamycin in culture did not modify the expression of the main Treg markers (Fig. 1E and F).
In detail, expanded Tregs were all positive for CTLA-4, HELIOS, CD38, and homing receptors such as CD62L and CXCR4. Conversely, activation and functional markers such as CD69 and CD39 and homing receptors such as CCR6 and CXCR3 were poorly or not expressed ( Fig. 1E and F). Of note, when expanded for 15 days by using a GMP-compatible method, nonsorted Tregs partially lose the expression of CTLA-4 [4]. Following expansion, Tregs upregulated the expression of TIGIT and CXCR4, whereas the expression of CCR7 and CD69 was lost (Supporting Information Fig. S3B and C). The majority of RA + Tregs downregulated the expression of CD45RA after expansion indicating an acquired memory phenotype. However, they lacked the expression of markers routinely expressed by expanded Tregs from blood such as CXCR3 and CD39 [11,12].
In line with both MacDonald et al. and Dijke et al. [3,4], expanded Tregs showed a high suppressive ability in vitro in a dose-dependent manner ( Fig. 2A).
To evaluate the stability of the expanded cells, Tregs and RA + Tregs were cultured in the presence of two pro-inflammatory cocktails of cytokines known to convert them into Thelper-17 [13]. As control, low dose of IL-2 only was used. No IL-17producing Tregs were detected in all the conditions, while IFN-γproducing cells were detected only in Tregs expanded in absence of rapamycin ( Fig. 2B and C). These data confirmed the impor-tance of rapamycin in culture to obtain a stable Treg product suitable for cell therapy. To further define cell stability at the end of the culture, we tested the methylation status of 11 CpG sites within the TSDR region. The mean percentage of methylation of the whole region was less than 10% in all the Treg preparations (Fig. 2D).
In our experience, cell freezing at the end of the Treg expansion has been a very helpful step for the application of Treg therapy. This step will be applied in the proposed clinical trial with thymic Tregs. Compatible with this idea, we have evaluated the phenotype and function of our Treg products soon after thawing. All the cells showed a viability greater than 70% over a period of 2 years (Supporting Information Fig. S3C), the Tregs retained their phenotype (Fig. 2E and Supporting Information Fig. S3D) and suppressive ability (Fig. 2F). Furthermore, almost no IL-2-producing cells were detected (% of IL-2 + cells <1% in all the preparations), while the proportion of TNF-α-producing cells was lower than 10% ( Fig. 2G and Supporting Information Fig. S3E).

Conclusion
In conclusion, CD3 + CD4 + CD25 + CD127 -Tregs showed the same phenotype, stability, and methylation level as the CD3 + CD4 + CD25 + CD127 -CD45RA + Tregs, suggesting that the additional step of isolating the CD45RA + subpopulation is not essential. Most importantly, thymic Tregs can be isolated and expanded to provide sufficient numbers (>3×10 7 cells/kg) for clinical application in a tolerance protocol. In addition, we have shown that the expanded thymic Tregs can go through a process of freezing and thawing without compromising their efficacy or stability. Overall, our data provide additional information for the translation of thymus-derived Tregs to the clinic.

Tissue processing
Discarded thymi from pediatric heart surgery were received from "Great Ormond Street Hospital" at University College London (UCL). Informed consent was obtained from all the donors prior to enrolment into the study (ethical approval: 06-MI-13(B)). Thymocytes were obtained after mechanical and enzymatic digestion by using dissociation medium consisting of X-Vivo 15 (Lonza) supplemented with 5% of human serum AB male (BioWest), amphotericin B (0.5 μg/mL; Thermo Fisher Scientific), collagenase (0.2 mg/mL; Merk), and deoxyribonuclease I (5 mg/mL; Merk). Thymic fragments were transferred into GentleMacs C tubes (Miltenyi Biotech) for mechanical digestion.

Flow cytometry analysis
Freshly isolated thymocytes and expanded Tregs were stained with two different panels. Intracellular staining was performed using Foxp3/Transcription Factor Staining Buffer Set (eBioscience) according to the manufacturer's instruction. Stained cells were acquired on 5-lasers LSRFortessa X20 (BD) and analyzed using Flowjo. Guidelines for the use of flow cytometry in immunological studies were adhered to Cossarizza et al. [14].

Tregs isolation, expansion, and cryopreservation
Tregs were isolated in two steps: CD8 + T cell depletion followed by enrichment of CD25 + cells using immunomagnetic beads (Miltenyi Biotech). After isolation, cells were labeled with the following dyes for cell sorting (FACSAria, BD): LIVE/DEAD® Fixable Near-IR Dead Cell Stain Kit, CD4, CD3, CD8, CD25, CD127, CD45RA (all from Biolegend). After sorting, total and CD45A + Tregs were counted and cultured in X-Vivo 15 (Lonza, UK) supplemented with 5% of human AB serum (BioWest), IL-2 (1000 IU/mL; Proleukin, Novartis), and in the presence/absence of rapamycin (100 nM; LC-Laboratories, USA). All the cells were activated with MACS GMP ExpAct Treg Kit (1:1 bead to cell ratio; Miltenyi). IL-2 and rapamycin were replenished every 2 days following activation. Cells were re-stimulated twice every 10-12 days. At the end of the expansion, cells have been aliquoted into cryotubes following resuspension in serum free freezing media (Biological Industries) placed at −80°C overnight and then stored in liquid nitrogen for up to 2 years.

Suppression assay
CD4 + CD25-T cells (Teffs) were labeled with 2.5 μM CFSE and activated with anti-CD3/CD28 beads (Invitrogen) at 40:1 (cell/bead) ratio. Teffs were then cultured alone (1 × 10 5 ) or cocultured with HLA-A2-mismatched Treg at different ratios. After 5 days, cells were stained with anti-HLA-A2 and data were acquired. The percentage of suppression was calculated based on the proliferation of Teffs alone compared to the percentage of proliferation observed in the presence of iTreg cells.

Methylation assay
CpG methylation analysis has been executed on 250 000 cells from each preparation. CpG methylation analysis was determined by pyrosequencing of bisulfite-modified genomic DNA and was conducted by EpigenDx, as previously described by us [15].