Multiple sources of human Tregs have been explored clinically. Peripheral blood collected by apheresis or leukapheresis is the most accessible and often the only option for autologous therapy and is therefore the most used source of starting material to date. Human Tregs are relatively rare in PBMCs, comprising only 5-10% of peripheral CD4⁺ T cells in healthy adults, and this can be further reduced in patients with autoimmune disease (2, 9, 18, 19). This inherent variability in the percentage of circulating Tregs can be compounded in the variability in the amount of Tregs post isolation and increases the risk of manufacturing failures. Allogeneic Treg therapies manufactured from partially HLA-matched healthy donor-derived peripheral blood or umbilical cord blood (UCB) have been successfully (in Phase I studies) aimed at treating GvHD (12, 20). UCB is enriched in naive Tregs that possess inherent expansion potential and broad TCR repertoire. However, a single UCB unit contains significantly less Tregs than PBMCs (~5-7.5x10⁶), and Tregs derived from UCB often require multiple rounds of expansion (the implications of which are discussed later in this review) (12, 21).

Alternative sources of naive Tregs currently being explored for allogeneic products include pediatric thymuses, which are routinely removed during pediatric cardiac surgeries, and differentiation of iPSC or conventional T cells into Tregs. iPSC derived CD4⁺ Tregs are an emerging technology with the potential to address the current limitations of low yield related to ex vivo expansion, manufacturing variability and specificity. These cells have been shown to be as functional and suppressive as natural (CD4⁺CD25⁺CD127^-) Tregs both ex vivo and in animal models of GvHD and can be concomitantly engineered during expansion to express CAR (22). iPSC derived Tregs offer a potentially renewable and scalable source of Tregs, however, as this class of cells are only being assessed at the preclinical stage, they will not be further discussed in this review. Using the thymus as the source of Tregs is advantageous due to the abundance of available Tregs. On average, 1-2% of thymocytes are CD4⁺CD25⁺ which translates to ~5x10⁸ cells. This is roughly 100x more than what is obtained from a UCB unit, and more than the number in the peripheral blood of an adult. The lack of activated conventional T cells in the thymus means that CD25⁺ exclusively marks CD4⁺ Tregs, which can facilitate isolation of a pure population of Tregs. The feasibility of manufacturing clinical material using this approach is currently being explored in a Phase I/II clinical trial assessing the ability of Tregs to prevent rejection after pediatric heart transplant (NCT04924491) (23) (24) (25). In vitro re-programming of conventional T cells into Tregs has been achieved through ectopic expression of Foxp3 (26, 27) as well as through culture with IL-2, rapamycin and TGF-B and was shown to be safe (NCT01634217) (13). Although feasible for clinical manufacturing, re-programming approaches require additional reagents as well as testing to prove the stability of the product, increasing the complexity of both manufacturing and release testing.

Human cells or tissues intended for implantation, transplantation, infusion or transfer into a human recipient are regulated as human cells, tissues and cellular and tissue-based products (HCT/Ps) under 21 CFR parts 1270 and 1271. These regulations require establishment registration, screening and testing of donors to reduce transmission of infectious diseases and also establish current good tissue practices (cGTP). Sponsors should think carefully about both the manufacturing and regulatory implications when choosing their starting material. While the manufacturing of allogeneic therapies is typically more reproducible than that of autologous therapies, the regulatory requirements for allogeneic products are more extensive than those for autologous products in order to assure patient safety. If multiple donors are being used, special attention must be paid to possible cell interactions that could result in unexpected immune responses or alter the performance of the cells.

Like conventional T cells, Tregs are functionally and phenotypically diverse and can be divided into subsets based on cell surface marker expression. Identifying whether particular subsets are more attractive therapeutically or whether the appropriateness of different subsets varies by application (i.e., GvHD vs, autoimmune disease) is essential to the field but at this point remains unknown. A detailed discussion and comparison of the different Treg subpopulations is beyond the scope of this review and is discussed in detail elsewhere (28, 29).

Due to their low frequency in all starting materials, Tregs must be purified prior to expansion by either magnetic-activated cell sorting (MACS) or fluorescent-activated cell sorting (FACS). As Tregs do not express any unique cell surface markers, there is considerable variation in the methods used to isolate Tregs for clinical use, with the only commonality between all methods being the use of CD25 as a positive selection parameter. Both MACS and FACS require specialized equipment and reagents, highly trained personnel and are labor intensive process steps that take several hours per sample (~6 hrs for MACS, >12 for FACS).

Magnetic sorting is performed with the CliniMACS Plus (Miltenyi Biotec), and Tregs can be isolated using a one- or two-step procedure. In one-step isolation protocols, CD25⁺ T cells are positively selected, whereas in two-step protocols, CD8⁺ and/or CD19⁺ are first positively depleted, then CD25⁺ cells are positively selected, yielding a CD4⁺CD25⁺ isolated product. The use of a one- vs. two-step protocol is largely determined based on the starting material. In allogeneic trials using UCB, isolation is typically performed with a one-step procedure as UCB cells express high levels of CD25 (CD25^high), and the risk of CD25⁺ contaminating effector cells in this starting material is low. Conversely, two-step processes are common when peripheral blood is being used as the starting material. Post-isolation purity of the one-step process is typically low (<70%, median), whereas the two-step process yields significantly greater purity (>90%, median) (30). Isolation is a balance between selection of a sufficiently pure population and avoidance of relevant cell exclusion. Purity therefore comes at the cost of post isolation yield, regardless of the method or type of product.

Magnetic sorting is limited by the number of surface markers that can be used and is only capable of binary sorting based on the presence or absence of the selected markers. FACS is not only capable of using multiple markers but also allows for precise isolation of specific cell populations such as CD4⁺CD25^highCD127^lo/- which have been shown to have a higher frequency of Foxp3 expression and greater potency in vitro (1). CD45RA expression identifies naïve Tregs and is another marker that is being included in FACS based isolation protocols. CD4⁺CD25⁺CD127^lo CD45RA⁺ Tregs isolated have been shown to have an epigenetically stable Foxp3 locus (TSDR demethylation), enhanced suppressive ability and reduced Th17 plasticity in vitro compared to CD4⁺CD25⁺CD127^lo CD45RA- Tregs (31–33). This approach was successfully used to isolate and expand Tregs in the TRIBUTE study (NCT03185000) for the treatment of inflammatory bowel disease (33). The post-isolation purity of FACS isolated Tregs can be as high as 98%, however cell recovery is often modest (<80% of input) due to loss of cells during the pre-enrichment and staining steps and the long processing time. Based on the indication and route of administration, a balance must be achieved between yield and purity. The lengthy processing time of FACS also limits the number of cells that can be processed via cell sorting, hampering the number of Tregs that can be isolated for expansion. In an effort to increase the number of cells that can be sorted using FACS, some groups perform magnetic sorting of CD4⁺ cells followed by FACS. This manufacturing process was found to yield a product with >95% Foxp3⁺ cells (34, 35). There are currently two 21 CFR part 11 compliant cell sorters used for cell sorting of adoptive cell therapies that can be qualified for use in a GMP environment; Miltenyi Biotec’s Tyto and Sony Biotechnology’s CGX10. The Tyto is a closed, cartridge style benchtop cell sorter equipped with three lasers, allowing for up to 10-parameter cell sorting. Sony Biotechnology’s fully closed CGX10 cell isolation system is equipped with four lasers and replaces the FX500, which utilized exchangeable fluidics and was housed in a Class II BSC to enable qualification for GMP use. The creation of flow cytometers that can be qualified for cell sorting has enabled the use of FACS in the manufacture of clinical products. This highlights the need for dedicated and expert personnel to maintain and operate these specialized instruments.

Regardless of the isolation approach used to purify Tregs, antibodies are the critical raw material in this phase of manufacturing. cGMP compliant MACS antibodies are available from Miltenyi Biotec for all of the routinely used isolation markers (CD4, CD8, CD19, and CD25). As these antibodies are produced by a single supplier, all cell therapy manufactures requiring these antibodies acquire them from the same source. This can result in high costs, long lead times and potential manufacturing delays if supply chain issues arise. While there are numerous suppliers of FACS antibodies, cGMP-compliant antibody options (clone and fluorochrome combinations) are limited. Care should be taken early in the planning of clinical manufacturing, as preclinical non-GMP reagents may never translate to the GMP environment, resulting in different processes. Should sponsors need to use an antibody that is not available in GMP format, the cost to create a controlled, cGMP-compliant antibody is > $1 million USD. As the antibodies are critical to the manufacturing process, lot to lot variability must be controlled and in later phases of clinical development, these critical raw materials will need to be tested and released to ensure control is maintained.

Tregs are similar to conventional T cells in that they both requiring TCR and CD28 co-stimulatory signals for maximum expansion but are also highly dependent on IL-2 for survival, expansion and stability (9). Tregs are expanded using protocols similar to those used for conventional T cells; in culture bags or G-Rex bioreactors for 14–36 days, with CD3/CD28 stimulation and supplemented with IL-2 (9, 36). Anti-CD3/anti-CD28 coated microbeads and artificial APCs (aAPC) are the most widely used approaches for stimulating Treg expansion. However, cGMP compliant alternatives such as soluble CD3/CD28 antibody complexes have been developed (ImmunoCult Human CD3/CD28 T cell activator, STEMCELL and TransAct, Miltenyi) and are gaining popularity as they provide similar levels of activation, and are easily removed from culture media while minimizing cell loss. To reduce the risk of contaminating effector T cells, some groups add rapamycin, an approved mTOR inhibitor, to the expansion media to exploit the differential role of mTORC1 and mTORC2 in Tregs and effector T cells (Teff), resulting in the selective suppression of Teff growth due to the Foxp3 induced expression of Pim2, a serine/threonine kinase that confers rapamycin resistance (37, 38) (37–39). Although the addition of rapamycin improves purity, it reduces Treg expansion ~ 10 fold, necessitating longer ex vivo culturing times (11, 40). The addition of rapamycin is predominantly used when Tregs are isolated by CD4⁺CD25⁺ MACS to improve the purity of the final product. Expansion methods vary greatly between protocols with regards to culture media, additives, duration of culture and the type, timing and frequency of antigen stimulation, making it difficult to compare results across different products (11).

The use of autologous cells poses multiple challenges when it comes to expansion. First and foremost, in the reproducibility of the manufacturing process and the ability to produce an adequate number of cells. Tregs used in autologous therapies are isolated from the patient, so not only can the cells be dysfunctional, but there can be substantial variability in the extent of dysfunction between patients. This leads to heterogeneity during the expansion phase, both in the Tregs ability to expand and the phenotype of the final product (increased interferon production compared to Tregs from healthy donors) (36). Multiple trials have documented the failure of patient samples to meet the minimum cell number required for infusion (12, 20, 41). Due to the manufacturing challenges of autologous products, it is not uncommon for manufacturing feasibility to be included as a clinical endpoint of the trial. These issues underscore the need for a better understanding of starting material and their impact on Treg manufacturing. Through a deeper understanding, potential failures could be identified and manufacturing criteria adjusted to reduce the risk of failures.

Manufacturing of allogeneic Tregs from healthy donors is substantially more reproducible, although variability does exist. UCB is a common starting material for allogeneic therapies but due to the smaller number of Tregs in UCB, multiple rounds of expansion are necessary to achieve dose (12, 13), extending the cGMP manufacturing time and increasing the amount of reagents needed to manufacture, resulting in drastically higher manufacturing costs, diversion to a. While longer clinical expansion protocols have successfully manufactured Tregs with >75% CD4⁺CD25⁺Foxp3⁺ (38), prolonged expansion can have negative functional impact, such as diversion to a Th2 (IL-4 producing) Treg functional state (42). Thymic Tregs although naive, do not proliferate as effectively as UCB or blood Tregs during ex vivo expansion and are not widely implemented (19). However, recent advances in clinical grade expansion protocols for thymic Tregs have demonstrated the ability to produce highly pure, functional T-regs in 10–23 days (43). Moreover, the ability to cryopreserve and recover cells with >95% viability and 80% Foxp3+ post freeze-thaw further highlights the potential of this material for use in clinical manufacturing, making it likely to become a more prominent starting material in the future.

Ex vivo expansion protocols often require a large number of single use reagents and consumables, such as specialized media, cytokines (IL-2), antigen stimulation reagents (CD3/CD28 beads, artificial APCs), drugs (rapamycin), culture bags or bioreactors and serum. These raw materials and consumables must be released by the manufacturer for use. Often, for early phase raw material release can rely on the certificate of analysis presented by the manufacturer of the raw material. However, in situations where cGMP-compliant material may not be available, a risk assessment should be undertaken as to the impact of using research grade material in the manufacturing process of clinical material. This risk assessment, inclusive of an evaluation of controls that may be in place (such as in-process and/or release testing), may necessitate developing and conducting tests of the raw material itself. The test(s) should be conducted on the incoming raw material, upon receipt, for the release of the raw material for use in manufacturing by the product developer. Additionally, single-source reagents often pose a risk to the supply chain. In these situations, it is advisable to understand and monitor raw material and consumables lead time to receiving the material once an order is placed. A thorough understanding of material supply chain will allow for the ordering of safety stocks to mitigate supply chain associated risks to manufacturing campaigns.

In addition, the reagents used for Treg manufacturing can often introduce additional complexity. For example, Treg culture media is typically supplemented with 5-10% human AB serum to provide critical growth factors to cells. Not only is the serum costly, but it necessitates additional testing (both prior to use and during release testing) creating additional quality and compliance burden. Due to the variability of serum, serum lots must be qualified before release for use to confirm they support cell growth. Whenever a new lot of serum is to be used, an assessment of old and new lots using a reference material must be performed to confirm that expansion is similar. Serum-free media such as OpTmizer (ThermoFisher) and ImmunoCult-XF (StemCell) that contain serum replacement supplements are being used in the manufacture of other T cell therapies and may become more common in Treg manufacturing if it can be demonstrated that they enable similar levels of expansion as serum containing media.

The widespread implementation of cell-based therapies is limited by the complexity and scalability of the manufacturing process and the lack of a reliable manufacturing platform that ensures consistent production of clinical grade material at the required therapeutic doses. A variety of culture systems are currently used for the manufacture of cell-based therapies, and these are reviewed in detail elsewhere (44). Cell culture bags, G-Rex flasks and stirred flasks that are manufactured with gas permeable polymers or membranes are the simplest, least expensive and most widely used systems to date for manufacturing Treg therapies. These bags and flasks can have multiple aseptic ports for media input/output, sampling and harvest, but require additional equipment and manual processing at all steps. G-Rex flasks are among the most popular of these expansion systems due to their affordability and scalability.

Bioreactors (rocking motion (WAVE), stirred tank and hollow fiber (Quantum) are more mechanized culture systems that allow certain steps to be completed in a closed aseptic model, lowering the risk of operator-based errors during manufacturing. They can be equipped with pH and dissolved oxygen probes to provide real-time control of these variables. Currently, bioreactors are not commonly used in the manufacture of Tregs. Semi-automated and automated systems like the CliniMACS Prodigy and Lonza Cocoon require minimal human intervention. The CliniMACS Prodigy is a single closed system instrument capable of magnetic separation, cell expansion with automated media exchange and formulation. While not known to be capable of cell separation or formulation, the Cocoon supports cell transfection, transduction and expansion and can link several systems together in order to perform parallel large-scale cell expansion with full electronic control over the process in a single network, features especially beneficial for the commercial manufacturing of off-the-shelf allogeneic cell products (44). These cGMP-compliant systems reduce infrastructure requirements at the manufacturing facility however, they are costly and at the present moment, not more efficient in terms of time and yield than less advanced methods.

Upon completion of ex vivo expansion, cells must be formulated and filled for administration to patients. The final formulation of a drug product is dictated by indication, route of administration, storage requirements and shelf-life stability need. A key decision at this stage of Treg manufacturing is whether to cryopreserve the cells or administer them fresh. Cryopreservation is typically achieved through the use of cryopreservation solutions containing dimethyl sulfoxide (DMSO) and USP grade components, such as CryoStor CS10 (StemCell). Cell concentration is based on the intended clinical dose (cells/kg) and as such there is significant variability between trials as doses range from 1–100 x10⁶. Once formulated, cells are cryopreserved using controlled-rate freezers with validated freezing protocols, followed by long term storage in liquid nitrogen. This facilitates large-scale production at centralized manufacturing facilities, enables long-term storage and reduces logistical hurdles by allowing for flexibility around the time of infusion to accommodate changes in patient health, while also affording more time for release testing. Due to these benefits, cryopreservation has been widely implemented in the manufacture of both autologous and allogeneic cell-based therapies such as TIL and CAR-T. However, DMSO itself is a known inhibitor of viability and cryopreservation of PBMCs and purified Tregs cells has been associated with loss in cell yield, decreased viability as well as impaired cytokine production and suppressive capacity and reduced surface marker expression (FoxP3) that essential for proper Treg function (45–48). These effects can all dramatically affect the clinical safety and efficacy of this therapy (49, 50). Conversely, others have shown Tregs can be successfully frozen and thawed without compromising their phenotype (51–54). These discrepancies are likely due to the variations in the manufacturing process (starting material, method of expansion, purity of the final product) as well as differences in cryopreservation protocols (cryopreservation agent and % DMSO, cell concentration and freezing protocol), as there is currently no standard protocol. These mixed reports have led to uncertainty around the feasibility of cryopreserving Tregs, and as a result, clinical trials have been a mix of cryopreserved and freshly administered Tregs, with the majority using administering fresh products. Novel cryopreservation agents with potentially less deleterious effects on cell viability are currently being explored preclinically (55, 56). Development of cryopreservation media and standardized methods for Treg therapy should leverage the findings and progress being made in conventional T cell-based therapy, which is more clinically advanced.