Regulatory T-Cell Therapy in Transplantation: Moving to the Clinic

Abstract

Regulatory T cells (Tregs) are essential to transplantation tolerance and their therapeutic efficacy is well documented in animal models. Moreover, human Tregs can be identified, isolated, and expanded in short-term ex vivo cultures so that a therapeutic product can be manufactured at relevant doses. Treg therapy is being planned at multiple transplant centers around the world. In this article, we review topics critical to effective implementation of Treg therapy in transplantation. We will address issues such as Treg dose, antigen specificity, and adjunct therapies required for transplant tolerance induction. We will summarize technical advances in Treg manufacturing and provide guidelines for identity and purity assurance of Treg products. Clinical trial designs and Treg manufacturing plans that incorporate the most up-to-date scientific understanding in Treg biology will be essential for harnessing the tolerogenic potential of Treg therapy in transplantation.

Regulatory T-cell (Treg) therapy is planned at multiple transplant centers around the world. Factors critical to the efficacy of Treg therapy are dose, specificity, and adjunct immunosuppression.

One of the major challenges facing the field of transplantation is the management of immunosuppression. Although immunosuppression is necessary to prevent immune attacks of the transplanted organ, it also imposes substantial morbidity and mortality risks for transplant recipients. Chronic global immunosuppression impairs immune responses to microbial pathogens and hinders tumor immunosurveillance. Often, infections and posttransplant cancers, rather than allograft rejection, are the major contributors of transplant-related mortality, especially beyond the first year after transplant (Penn 1990; Euvrard et al. 2003; Soltys et al. 2007). In addition to these immunological complications, immunosuppressive drugs are often causes of morbidity owing to their off-target effects such as nephrotoxicity, diabetes, hyperlipidemia, hypertension, cardiovascular diseases, and obesity (Berenson et al. 1992; Textor et al. 2000; Nair et al. 2002; Ojo et al. 2003). All these complications may necessitate reduction or even withdrawal of immunosuppression that leads to graft rejection and graft loss. Therefore, the key to improving immunosuppression after transplantation is to selectively block the immune responses against the graft without impeding other protective immune functions or causing nonspecific toxicities.

In this article, we propose that such goal can be accomplished by harnessing the natural immune regulatory mechanisms using cell-based therapies. Various types of T cells have been shown to contribute to transplant tolerance. These include the CD4⁺CD25⁺ regulatory T cells (Tregs) that express the transcription factor FOXP3, IL-10-producing Tr1 cells, CD8⁺28⁻ T cells, and anergic T cells. In this article, we will focus on the FOXP3-expressing Tregs. We summarize parameters that are important for effective application of Treg therapy to prevent graft rejection in experimental models and review advances in translating these preclinical experiences to the clinic.

ADVANTAGES OF Treg THERAPY IN TRANSPLANTATION

None of the current immunosuppressive drugs can suppress immune responses to transplant antigens without potentially altering immune surveillance toward tumor antigens and microbial pathogens. This is because immunosuppressive drugs target common pathways of immune activation, such as calcium signaling (cyclosporin A, FK506), purine biosynthesis (Mycophenolate), and T-cell costimulation (CTLA4Ig). Other immunosuppressive drugs such as antithymocyte globulin, Campath-1, and anti-CD20, massively and nonspecifically delete immune cells. In contrast, T cells have an extraordinary ability to distinguish minute differences among different antigens. Aside from the specificity conferred by the T-cell receptors, specificity of therapeutic T cells is also amplified by their ability to seek their targets throughout the body and deliver effector functions locally where they are most effective and specific. These properties underlie the remarkable efficacy of T-cell therapy in treating drug-resistant recurrent cancers (Restifo et al. 2012; Scholler et al. 2012). For example, using autologous cytotoxic T cells engineered to express a chimeric receptor that recognizes CD19, 19 out 20 patients with end-stage therapy-refractory B-cell lymphoma have been successfully treated (Porter et al. 2011; Scholler et al. 2012). Thus, T cells can be used to deliver highly specific and targeted therapies.

Similarly, immune tolerance mediated by Tregs is also highly antigen specific. Immunosuppressive functions of Tregs are activated by the engagement of their T-cell receptors locally at the site of antigen deposition. In a mouse model of autoimmune diabetes and autoimmune pancreatitis, we have observed complete protection against islet destruction using islet antigen-specific Tregs, whereas autoimmune attack of the surrounding exocrine pancreas progressed without hindrance (Meagher et al. 2008). In the transplant setting, tolerance maintained by Tregs is specific to the graft donor, whereas unrelated grafts are rejected (Joffre et al. 2008). Tolerance can spread to a new transplant antigen, and this “linked suppression” is mediated by Tregs in the graft. The establishment of linked suppression requires that the new antigen and the antigen the host is already tolerant to are present in the same graft. Thus, immune tolerance to transplant antigens mediated by Tregs is highly specific and highly localized. Despite their high antigen specificity, Tregs are highly versatile and can control responses of various immune cells including conventional CD4⁺ T cells, CD8⁺ T cells, natural killer (NK) cells, natural killer T (NKT) cells, B cells, and various antigen-presenting cells. Moreover, Tregs have a collection of more than a dozen different immunosuppressive mechanisms and can deploy different strategies depending on the tissue microenvironment (Tang and Bluestone 2008; Yamaguchi et al. 2011). In comparison to standard immunosuppressive drugs used in transplantation today, Tregs are “smart” therapeutic agents that are highly antigen specific and highly adaptable, capable of selectively targeting graft tissues by tuning their activities in response to the tissue microenvironment.

Besides the problems with long-term global immunosuppression and off-target toxicities, most of the current immunosuppressive drugs prevent tolerance induction and create a dependence on continuous immunosuppression. Immunosuppressive drugs used in the clinic today are selected based on their ability to prevent immune activation. Research from the past three decades shows that acquisition of immune tolerance to self-antigens and transplantation antigens is an active process that requires antigen exposure. A tolerogenic antigen exposure leads to inactivation of antigen-reactive T cells through apoptosis, anergy, and induction and expansion of immune regulatory mechanisms that maintain tolerance. The current approach of immunosuppressing transplant recipients blindfolds the immune system to prevent rejection, but also impedes tolerance induction. This may explain why spontaneous transplant tolerance is rare and the best predictor of tolerance is time after transplant (Sanchez-Fueyo 2011), likely through the cumulative effects of a low level of donor antigen exposure over a long period of time. To induce tolerance and freedom from chronic immunosuppression, transplant recipients should receive immunoregulatory regimens that block rejection while permitting donor antigen recognition. In this regard, Tregs can induce other T cells to acquire regulatory functions in vitro (Dieckmann et al. 2002; Jonuleit et al. 2002; Oliveira et al. 2011). In a mouse model of transplantation, Treg therapy not only prevents rejection, but also allows the induction of new Tregs with broader specificities through a process of infectious tolerance (Waldmann 2008). Therefore, Treg therapy can potentially turn the graft tissue from a target of immune attack into a tolerogenic organ that promotes its own long-term survival.

CRITICAL PARAMETERS FOR EFFECTIVE Treg THERAPY IN TRANSPLANTATION

Dosing

Tregs ensure normal immune homeostasis by providing a counterbalance for the effector arm of the immune system. In an immunologically quiescent state, Tregs represent ∼5%–10% of CD4⁺ T cells in lymphoid tissues. At this ratio, Tregs prevents unwanted immune activation by reducing expression of costimulatory molecules CD80 and CD86 via CTLA-4-mediated trogocytosis (Muthukumar et al. 2005; Dijke et al. 2007) and by sopping up IL-2 and other common γ-chain-binding cytokines (Pandiyan et al. 2007; O’Gorman et al. 2009). During an active immune response, Tregs proliferate, traffic, and accumulate at the site of inflammation, particularly at the later phase of the response, to restore normal immune homeostasis using a wider array of effector mechanisms including immunosuppressive cytokines IL-10, IL-35, TGF-β, cell-surface ATPases, granzyme-dependent killing of antigen-presenting cells, IL-9-mediated recruitment of mast cell, etc. (Tang and Bluestone 2008; Yamaguchi et al. 2011). Consequently, the number of Tregs often increases with inflammation and graft rejection (Muthukumar et al. 2005; Dijke et al. 2007). In the transplant setting, such increases are usually not sufficient and too late to prevent graft damage; therefore, effective Treg therapy should be given at a dose sufficient to tip the balance in favor of Tregs before the rise of effector responses.

Two approaches can be applied to estimate the effective dose of Tregs in humans to prevent graft rejection: one based on achieving a percentage of Tregs required to tip the balance to tolerance and the other based on allometric scaling from mouse transplant models. Early proof-of-principle experiments in mouse models relied on the use of adoptive transfer of a mixture of effector T cells and Tregs to lymphopenic hosts to precisely control and balance between the two populations (Hara et al. 2001; Graca et al. 2002). In these settings, a high ratio of at least one Treg per two effector T cells, and sometimes as high as five Tregs per one effector cell, is needed to prevent rejection (Nishimura et al. 2004; Golshayan et al. 2007). This suggests that a minimum of 33% Tregs is required to prevent rejection. Experiments in lymphoreplete mice also find that 30% Tregs in the grafts is associated with prevention of graft rejection (Fan et al. 2010). Interestingly, Tregs are also found to accumulate to 30% in tumor tissues and are thought to contribute to the suppression of antitumor immunity. Thus, 30% Tregs is likely the tipping point between productive immunity and tolerance. We have estimated that an average adult human has 166 × 10⁹ CD4⁺ T cells in the body, including lymphoid and nonlymphoid organs, and among which 13 × 10⁹ are Tregs (Tang and Lee 2012). To increase the percentage of Tregs to 30% by using Treg therapy alone, 53 × 10⁹ Treg would be needed. Aside from the technical challenge of producing such a high dose of Tregs (see below), it is not clear if Tregs can engraft when infused at such a high bolus dose. However, 10 times less (5 × 10⁹) Treg would be sufficient if Treg therapy is combined with prior depletion of 90% of the T cells using an immune T-cell-depleting agent such as antithymocyte immunoglobulin.

Scaling for immune cell therapy may be less complicated and unpredictable than that for small molecule drugs because pharmacokinetic and pharmacodynamic properties of immune cells (i.e., rules for cell trafficking, turnover, and dose response) are relatively insensitive to the differences in body mass and metabolic rate of different species (Wiegel and Perelson 2004; Perelson and Wiegel 2009). For example, clinical data obtained from hematopoietic stem cell transplant shows that the minimal effective dose for neutrophil and platelet reconstitution in humans is consistent with that predicted by allometric scaling of data from animal models. Thus, the number of Tregs needed to achieve efficacy in humans may be extrapolated using numbers from mouse experiments multiplied by the ratio between the sizes of the human and the mouse lymphoid compartments. One way to approximate the relative size of the mouse and human lymphoid compartments is by comparing the lymphocyte cellularity of the spleens. Adult human spleens contain 70 × 10⁹ lymphocytes (Westermann and Pabst 1992; Ganusov and De Boer 2007; Nylen et al. 2007), 1000 times that in adult mouse spleens. Therefore effective Treg dose in humans is likely 1000 times that found in mouse models. There is very little experimental evidence for effective Treg dose in lymphoreplete mouse models. In a mouse islet transplant model, 1 × 10⁶ unmanipulated polyclonal Tregs were able to prolong graft survival by 2 wk (Zhang et al. 2009). Others used 2 to 5 × 10⁶ Tregs enriched for donor antigen reactivity and also observed limited prolongation of graft survival unless combined with other treatments (Golshayan et al. 2007; Joffre et al. 2008; Tsang et al. 2008). We have found that 30 × 10⁶ polyclonal Tregs was able to induce long-term graft survival only when combined with substantial deletion of donor-reactive T cells from the hosts (K Lee and Q Tang, unpubl.). Thus, using allomeric scaling, effective dose of polyclonal Tregs for preventing rejection in humans is estimated to be >30 × 10⁹.

The total number of Tregs in an adult human is estimated to be 13 × 10⁹ and most of the Tregs reside in lymphoid organs (Tang and Lee 2012). The number of Tregs circulating in the blood is 0.25 × 10⁹, which is the most that can be isolated from an autologous source for therapeutic use. It is important to point out that there is limited information about the percentage of Tregs in the body that actually circulates. Tregs in the skin and gut have been suggested to be largely resident with little ability to circulate. This is especially true of memory Tregs (Rosenblum et al. 2011). In a recent clinical trial in type 1 diabetes conducted at UCSF, deuterium-labeled ex vivo expanded Tregs were tracked in vivo. Our preliminary results showed that a dose of ∼350 × 10⁶ led to a 3%–5% increase of the circulating Tregs. It is clear that ex vivo expansion is needed to dramatically increase the number of Tregs (JA Bluestone, MK Hellerstein, SE Gitelman et al., unpubl.).

Specificity

In mouse models of Treg therapy in transplantation, a consistent feature is that Tregs purified from tolerant hosts are more effective in transferring tolerance to new hosts than Tregs from naïve hosts. The improved efficacy is likely owing to an increased frequency of donor alloantigen-reactive Tregs induced by various tolerance-inducing protocols (Bushell et al. 1995; Cobbold et al. 2004; Karim et al. 2004; Ochando et al. 2006; Yates et al. 2007; Verginis et al. 2008; Francis et al. 2011). There are two types of alloantigen-reactive T cells. The “direct” alloantigen-reactive T cells recognize intact alloantigen expressed on donor cells, and the “indirect” alloantigen-reactive T cells recognize processed donor alloantigens presented by host antigen-presenting cells (Gould and Auchincloss 1999; Rogers and Lechler 2001). Interestingly, allograft tolerance is primarily mediated through the indirect pathway (Hara et al. 2001; Yamada et al. 2001; Callaghan et al. 2007; Sanchez-Fueyo et al. 2007; Gokmen et al. 2008), and many tolerance-inducing protocols expand Tregs with the indirect specificity (Wise et al. 1998; Ochando et al. 2006; Verginis et al. 2008). Two independent studies compared the efficacy of direct Tregs with Tregs of mixed direct and indirect specificities, and both found that additional indirect specificity improved efficacy (Joffre et al. 2008; Tsang et al. 2008). Indirect Tregs alone confer some protection against rejection, but the protection is very limited (Golshayan et al. 2009; Tsang et al. 2009), not much better than protection conferred by direct Tregs with a single specificity from T-cell receptor (TCR) transgenic mice (Brennan et al. 2011). Because current Treg manufacturing technology is only capable of large-scale production of polyclonal and direct alloreactive Tregs (see below for more details), we have compared the relative potency of direct and polyclonal Tregs. We found that when combined with deletion of 80% donor-reactive T cells, Tregs of direct alloreactivity could induce long-term islet allograft survival. Polyclonal Tregs were also capable of inducing long-term graft acceptance, but five times more cells were needed (K Lee and Q Tang, unpubl.). It has been thought that the impact of direct Tregs would be short lived because their activation and function depends on short-lived donor-derived antigen-presenting cells. However, there is some evidence that host antigen-presenting cells can acquire intact donor human leukocyte antigen (HLA) antigens expressed by the grafts and present alloantigens in a “semidirect” fashion, thus professional direct alloantigen presentation, and hence direct Treg function, may persist long after transplantation (Sagoo et al. 2012). Alternatively, direct Treg may create a tolerogenic milieu in the graft tissue to promote indirect Tregs through infectious tolerance, thus achieving long-term protection. Taken together, we believe combining direct and indirect alloreactive Tregs may be optimal in inducing graft survival when limited Treg numbers and/or minimal immunosuppression are used. When combined with adequate preconditioning and given in sufficient numbers, both direct and polyclonal Tregs may be able to induce long-term graft survival.

Adjunct Immunosuppression

Experimental data show that Tregs cannot prevent rejection as a stand-alone therapy. Treg induction of long-term graft survival requires short-term adjunct immunosuppression to create a therapeutic window. Most immunosuppressive drugs used in the transplant clinics today were selected based on their ability to block immune responses before the existence of Tregs and their role in transplant tolerance was established. Some of these immunosuppressive drugs antagonize Treg function and survival, whereas others are less harmful to Tregs or even beneficial (Table 1). Therefore, selection of adjunct immunosuppression will have a significant impact on the efficacy of Treg therapy. Currently planned clinical trials have to rely on available drugs to combine with Treg therapy, and developing new “Treg-supportive” immune-modulatory drugs will be instrumental for improving efficacy of Treg therapy. For example, activation of antigen-presenting cells via CD40 can abrogate Treg-mediated suppression (Serra et al. 2003), and blocking CD40 and CD40L interaction in mouse and nonhuman primate models of transplantation has shown promising results (Li et al. 2008). However, anti-CD40L antibodies had unexpected thrombotic complications in humans owing to platelet activation. Using nonthrombogenic alternatives or targeting CD40 may solve the problem. Given the differential impact of PI3 kinase-AKT-mTOR signaling in Tregs and conventional T cells, it is also possible to preferentially inhibit conventional T-cell activation by antagonizing the PI3 kinase-signaling pathway (Han et al. 2012). In addition, directly boosting Tregs may be an effective strategy to combine with Treg therapy. IL-2 therapy induces significant expansion of Tregs, but can also expand CD8⁺ T cells and NK cells, particularly when higher doses of IL-2 are used (Bluestone 2011; Koreth et al. 2011; Saadoun et al. 2011; Long et al. 2012). In mouse models, it was possible to preferentially deliver IL-2 to Tregs using anti-IL-2 antibodies and repair defects of Tregs in a mouse model of autoimmune diabetes (Boyman et al. 2006; Tang et al. 2008). The efficacy of such strategy in humans remains to be determined. Last, histone deacetylases destabilize Tregs by promoting FOXP3 degradation, and histone deacetylase inhibitors have been shown to improve Treg homeostasis and induce allograft tolerance (Tao and Hancock 2007; Tao et al. 2007). New discoveries and understanding in Treg biology and transplantation tolerance have led to ever expanding therapeutic opportunities. However, it is extremely unlikely that any single agent will be effective against transplant rejection or capable of inducing tolerance. The challenge is to design an optimal combinational regimen and delivery program to steer the immune system away from rejection and into tolerance.

Table 1.

Impact of immunosuppressive drugs on Tregs

Drug	Mechanism of immunosuppression	Impact on Tregs	References
Corticosteroids	Binds to nuclear receptor to inhibit AP1 and NF-κB and expression of proinflammatory cytokines	May support Tregs by reducing inflammation	Karagiannidis et al. 2004; Xu et al. 2009
CNI	Inhibits calcineurin, calcium signaling pathway, NFAT activation, IL-2 production	Detrimental to Treg function and survival	Baan et al. 2005; Pascual et al. 2008; Zeiser et al. 2008; Demirkiran et al. 2009
Rapamycin	Inhibits mTOR, protein synthesis, proliferation	Spares Tregs, thus increases percentage of Tregs	Baan et al. 2005; Pascual et al. 2008; Zeiser et al. 2008; Demirkiran et al. 2009
MMF	Inhibits purine biosynthesis, T- and B-cell proliferation	Likely neutral	Baan et al. 2005; Pascual et al. 2008; Zeiser et al. 2008; Demirkiran et al. 2009
ATG	Deletes T cells, NK cells, B cells	Deletes Tregs less efficiently, thus increases percentage of Tregs	Lopez et al. 2006; Morelon et al. 2010
Anti-CD25	Deletes CD25-expressing cells	Deletes Tregs	Bluestone et al. 2008; Toso et al. 2009
CTLA4-Ig	Blocks CD80 and CD86 costimulation of CD28 and T-cell clonal expansion	Spares Tregs when used at subsaturating dose	Bluestone et al. 2008
Anti-LFA-1	Blocks LFA-1 and ICAM-1 interaction and T-cell activation, blocks leukocyte trafficking	Dramatically increases circulating Treg percentage	Posselt et al. 2010

CNI, calcineurin inhibitor; MMF, mycophenolate mofetil; ATG, antithymocyte globulin.

Treg MANUFACTURING

Treg Isolation

The first step in Treg expansion is the isolation of Tregs. A two-step magnetic activated cell sorting (MACS) protocol has been proposed for good manufacturing practice (GMP)-compliant isolation of human Tregs (Hoffmann et al. 2006; Peters et al. 2008b). However, Tregs isolated using this protocol are often contaminated with conventional T cells. This is because CD25 expression on human CD4⁺ cells is not restricted to Tregs and there is no clear separation between Tregs and non-Tregs based on the level of CD25 expression. Selecting only the top 2% CD25^hi Tregs increases purity at the expense of low yield. The additional use of CD127 allowed high-yield and high-purity recovery human Tregs based on the cell-surface phenotype of CD4⁺CD25⁺CD127^lo/− (Liu et al. 2006), and on average 1 × 10⁶ Tregs can be isolated from 100 mL of blood using this approach. Tregs selected based on CD4⁺CD25⁺CD127^lo/− markers have been found to be more effective than CD4⁺CD25⁺ Tregs in controlling alloimmune-mediated arterial atherosclerosis in a humanized mouse model (Nadig et al. 2010). It has also been reported that human Tregs can be identified on the basis of CD4, CD25, and CD45RA expression with naïve Tregs having the CD4⁺CD25^midCD45RA⁺ phenotype and antigen-experienced effector Tregs having a CD4⁺CD25^hiCD45RA⁻ phenotype (Miyara et al. 2009). It has been shown that CD45RA⁺ Tregs are more stable during repeated in vitro stimulation with anti-CD3 and anti-CD28, whereas CD45RA⁻ Tregs progressively lost FOXP3 expression after each round of stimulation (Hoffmann et al. 2009). We have found that CD4⁺CD25⁺CD127^lo/− Tregs contain both CD45RA⁺ and CD45RA⁻ populations and the CD45RA⁺ subset preferentially expands in culture (Putnam et al. 2009), as others have reported (Miyara et al. 2009), and the presence of the CD45RA⁻ subset does not negatively impact the purity of the Tregs at the end of the expansion. Selecting Tregs with three or more cell-surface markers improves purity and yield, but makes it challenging to isolate the cells using MACS. Fluorescence-activated cell sorting (FACS) is capable of identifying cell subsets based a panel of three or more markers, however, currently, there is no GMP-compliant FACS and the process is slow, which poses a limit on the number of Tregs that can be purified and manufactured. An instrument that combines the bulk processing capability of MACS with the high precision multiparameter-based FACS would be ideal for fast isolation of highly pure Tregs.

Polyclonal Treg Expansion

Tregs can be readily expanded using anti-CD3 and anti-CD28-coated beads supplemented with IL-2 (Levings et al. 2001; Herold et al. 2002; Hoffmann et al. 2004; Earle et al. 2005; Putnam et al. 2009). Conventional CD4⁺ T cells and CD8⁺ T cells expand better than Tregs using this protocol, therefore high purity of the starting population is essential for producing highly pure expanded Tregs. This is particularly problematic for MACS-purified CD4⁺CD25⁺ Tregs, as they are often contaminated with CD25⁺FOXP3⁻ conventional cells. The addition of rapamycin improves the purity of the culture because rapamycin suppresses the proliferation of conventional T cells and Treg growth is less affected (Battaglia et al. 2005). Despite these advances, large-scale manufacturing of Tregs remains challenging because even highly pure Tregs lose FOXP3 expression with repeated stimulation even in the presence of rapamycin (Hoffmann et al. 2009; Hippen et al. 2011b). The loss of FOXP3 is likely owing to destabilization of FOXP3 expression in Tregs instead of outgrowth of a few contaminating conventional T cells (Hoffmann et al. 2009). The cellular and molecular basis for Treg destabilization during in vitro stimulation is presently unclear. Nonetheless, it will be important to determine an optimized expansion protocol to maximize yield without compromising purity.

Alloantigen-Reactive Treg Expansion

The frequency of direct alloreactive Tregs has been estimated to be between 1% and 10% (Lin et al. 2008; Veerapathran et al. 2011). Proof-of-principle experiments have shown that alloantigen-reactive Tregs can be expanded using donor antigen-presenting cells such as dendritic cells, B cells, and unfractionated peripheral blood mononuclear cells (Peters et al. 2008a; Chen et al. 2009; Sagoo et al. 2011; Veerapathran et al. 2011; Tran et al. 2012). Collectively, these studies reported that alloantigen-expanded Tregs were 5–32 times more potent at suppressing alloantigen-stimulated proliferation in vitro than polyclonal Tregs. These results suggest that 5–32 times less alloantigen-expanded Tregs may be sufficient to achieve the same therapeutic efficacy as polyclonal Tregs. Based on our estimate that 5 × 10⁹ polyclonal Tregs would be sufficient to induce tolerance when combined with 90% deletion of endogenous T cells, 150 × 10⁶ to 1 × 10⁹ alloantigen-reactive Tregs would be needed to achieve similar efficacy. Thus, clinical translation of alloantigen-reactive Treg therapy will require reliable manufacturing of more than hundreds of millions cells under GMP conditions. In this regard, we have developed a process using CD40L-activated allogeneic B cells to selectively stimulate the expansion of alloantigen-reactive Tregs (Putnam et al. 2013). After primary expansion, the cells are restimulated with anti-CD3 and anti-CD28-coated beads to increase cell yield. Using this approach, we are able to achieve 200- to 4000-fold expansion in 16 d. The cells are highly donor reactive and have demethylated the Treg-specific demethylation region (TSDR). Clinical trials applying Tregs manufactured with this protocol in kidney and liver transplantations are planned to start in 2014.

Although most successes in expanding human alloantigen-reactive Tregs have been in generating direct Tregs, efforts have been made to expand human indirect Tregs with less spectacular results (Jiang et al. 2003; Veerapathran et al. 2011). This is likely because of the 100 times lower frequency of indirect alloreactive Tregs in the Treg pool (Veerapathran et al. 2011). Given the experimental evidence of improved efficacy in combining direct and indirect Tregs to induce tolerance, developing protocols to manufacture indirect alloreactive Tregs is an important future direction. Alternative to selective expansion of indirect Tregs from the existing repertoire, forced expression of TCRs with indirect alloreactivity during expansion of direct alloreactive Tregs can generate dual specificity Tregs with improved ability to protect grafts in mouse models (Tsang et al. 2008). Genetic engineering Treg specificity has been successfully shown with human Tregs (Brusko et al. 2010). In addition to TCRs, other desirable features can be introduced into the Tregs such as traceable markers, tunable TCRs, chemotactic receptors to synthetic ligands, and drug-inducible suicidal enzymes (Lim 2010). These “designer” features would allow monitoring of the infused Tregs, controlling their activities and trafficking patterns, and eliminating them when needed.

Ex Vivo Induction of Tregs

Alternative to isolating preexisting Tregs for ex vivo expansion, it has been proposed that conventional CD4⁺ T cells can be converted to Tregs during ex vivo expansion with the addition of TGF-β together with rapamycin or all-trans retinoic acid (Lu et al. 2010; Hippen et al. 2011a). These in vitro-induced Tregs acquire some features of Tregs such as expression of CD25 and FOXP3, reduced expression of effector cytokines, and ability to suppress in vitro and in vivo in a humanized mouse model of xenogeneic graft-versus-host disease (GvHD). Activated human T cells transiently express FOXP3; therefore FOXP3 expression alone does not distinguish activated T cells from Tregs. One distinction between ex vivo-isolated Tregs and in vitro induced Tregs is the DNA methylation status of the TSDR, which is important for Treg commitment and stability (more details in the paragraph below). Ex vivo-isolated Tregs have a demethylated TSDR, whereas in vitro-induced Tregs have fully methylated TSDR suggesting that they are not committed Tregs (Baron et al. 2007; Hippen et al. 2011a). Inhibiting or knocking down the DNA methyltransferase promotes demethylation of the FOXP3 locus and may drive commitment and stabilization of induced Tregs (Kim and Leonard 2007; Lal et al. 2009; Sanchez-Abarca et al. 2010). Further experimental evidence on the commitment and stability of ex vivo induced Tregs is needed before they can be considered as a viable source of therapeutic Tregs for humans.

Assessing Treg Identity and Purity after Expansion

Tregs are typically CD3⁺CD4⁺ and express the transcription factor FOXP3. Although CD3 and CD4 are clearly not unique for Tregs, conventional T cells transiently express FOXP3 after activation; therefore FOXP3 expression does not distinguish Tregs from conventional T cells either (Allan et al. 2007). It has been proposed that Tregs are best defined by their ability to suppress conventional T-cell proliferation in the “classical” in vitro suppression assay. It is important to remember that activated human conventional CD4⁺ T cells show dose-dependent inhibition of T-cell proliferation similar to the suppression curve obtained using Tregs (Walker et al. 2003, 2005). On the other hand, there are numerous examples in mouse models that cells with normal suppressive activities measured in the in vitro suppression assay fail to function in vivo (Zhou et al. 2008; Wohlfert et al. 2011; Ouyang et al. 2012). In addition, in vitro suppression assays cannot detect contaminating Tconv cells. Typically, Tregs suppress maximally when used at 1:2 to 1:4 Treg to conventional T-cell ratios, which means this assay cannot detect the presence of as many as 50% contaminating T cells. Collectively, the in vitro suppression assay can misidentify cells as Tregs and is not an appropriate approach for assessing the identity or the purity of a Treg product.

To properly identify Tregs, it is helpful to know how Tregs develop and what defines their lineage. Treg development in the thymus is a multistep process controlled by cytokines and signaling through T-cell receptor and CD28, which culminates in FOXP3 expression and epigenetic modification of the FOXP3 locus (Burchill et al. 2008; Lio and Hsieh 2008; Long et al. 2009; Lio et al. 2010; Ohkura et al. 2012). The TSDR of the FOXP3 locus is demethylated in cells that have committed to the Treg lineage to ensure stable inheritance of FOXP3 expression in dividing cells and the lineage stability of Tregs (Floess et al. 2007; Huehn et al. 2009; Josefowicz and Rudensky 2009; Zheng et al. 2010; Ohkura et al. 2012). Although FOXP3 expression is necessary for Treg development and function, Treg lineage specification and stabilization also depends on coordinated activities of up to 300 proteins to ensure persistent high expression of FOXP3 (Hill et al. 2007; Fu et al. 2012; Rudra et al. 2012; Samstein et al. 2012). This suggests that coexpression of FOXP3 and a collection of proteins essential for Treg lineage specification may define a Treg. However, the list of these essential proteins is not defined currently and there is evidence that many of these proteins have redundant roles in defining the Treg lineage (Fu et al. 2012) and there may never be an unambiguous definition for Tregs based on patterns of protein expression. Because the function of these proteins is reflected in the epigenome of a Treg to ensure inheritable high expression of FOXP3 and simultaneous repression of IL-2 locus and other genes characteristic of effector cells (Ohkura et al. 2012), an epigenetic profile most accurately defines a bona fide Treg (Hori 2008). Until a comprehensive epigenomic fingerprint of a Treg is defined, demethylated TSDR can be used currently to determine the purity and identity of Treg products (Wieczorek et al. 2009).

EARLY EXPERIENCE OF REGULATORY T-CELL THERAPY IN HUMANS

As of April 2013, there have been four reported clinical trials of Treg therapy in humans, three in GvHD and one in type 1 diabetes (Trzonkowski et al. 2009; Brunstein et al. 2010; Di Ianni et al. 2011; Marek-Trzonkowska et al. 2012). In addition, there are seven Treg trials registered in www.clinicaltrials.org, four in GvHD, one in type 1 diabetes, one in liver transplantation (NCT01624077), and one in kidney transplantation (NCT01446484). Some features of the published trials are summarized in Table 2 with emphasis on the Treg product infused. All four reported trials showed acceptable safety and promising efficacy of the treatment. Although none of these trials are in solid organ transplantation, information from these trials is instrumental for the design of future solid organ trials. For example, in the Brunstein et al. (2010) trial, because of the mismatch of HLA of the Treg cord blood donors from the host and the conventional T-cell donors, persistence of Tregs can be tracked. The infused Tregs could be detected between 4 hr and 7 d after infusion. The increase in infused Tregs in circulation was dramatically less after a second infusion of cryopreserved Treg products, suggesting cryopreservation may compromise the viability or stability of the Tregs, although other factors may also contribute. In addition, a clear trend of more consistent overall increase in Tregs was observed in rapamycin-treated patients when compared with cyclosporine-treated patients. In the Marek-Trzonkowska et al. trial, we learn that infusion of 20 × 10⁶/kg body weight polyclonally expanded Tregs in type 1 diabetic children between the ages of 8 and 16 is safe and the treatment is associated with immediate doubling of the percentage of circulating Tregs and a trend of increase at 2 wk. In the Di Ianni trial, therapeutic Tregs enabled the infusion of a higher dose of conventional T cells and better immunity against opportunistic infections. Particularly, the investigators observed markedly improved prevention against cytomegalovirus (CMV) disease. Five out of seven patients who received influenza vaccine after Treg infusion achieved protective antibody titers. All these results suggest that Treg infusion did not impede protective immunity against infections or lead to global immunosuppression. Overall, experience in clinical application of Treg therapy thus far shows that it is feasible, safe, and potentially effective.

Table 2.

Summary of published Treg trials in humans

References	Indication (No. of patients)	Tregs
		Isolation	Expansion	% FOXP3+	Dosing/kg
Trzonkowski et al. 2009	GvHD (n = 2)	FACS	2–4 Wk, weekly stimulations	40%–90%	0.1–3 × 106
Brunstein et al. 2010	GvHD (n = 23)	MACS	18±1 d, Anti-CD3/CD28 on day 0	31%–96%	0.1–6 × 106
Di Ianni et al. 2011	GvHD (n = 28)	MACS	None	69%±14%	2–4 × 106
Marek-Trzonkowska et al. 2012	Type 1 diabetes in children (n = 10)	FACS	14 d, Anti-CD3/CD28 on days 0 and 7	90%–97%	10–20 × 106

GvHD, graft-versus-host disease; FACS, fluorescence-activated cell sorting; MACS, magnetic-activated cell sorting.

It is worthwhile to review the quality of the Treg products in these trials closely, so that lessons learned from these early experiences in humans can be applied to the planned trials in solid organ transplantations. The investigator reported a wide range of Treg purity ranging from 31% to 97% FOXP3⁺. Two factors contributed to the variability of Treg purity. In the Di Ianni trial when no ex vivo expansion was used, purity below 50% was reported. In the Brunstein trial, the percentage of FOXP3⁺ cells after MACS isolation varied between 20% and 80% despite the use of cord blood as source materials. These results illustrate the problem with the current MACS technology. Additionally, loss of FOXP3 during repeated stimulation of Tregs was evident in the Trzonkowski trial that reported the percentage of FOXP3⁺ decreasing from 90% to 70% and 40% after successive weekly stimulations. It is encouraging that by combining FACS purification and limiting to two rounds of stimulations in the Marek-Trzonkowska trial, the purity of the expanded Tregs was consistently above 90%, however, the authors noted an insufficient Treg yield in four out of 10 patients. We have completed infusion of 10 type 1 diabetic patients of up to 320 × 10⁶ expanded Tregs at UCSF. By using FACS purification, and two rounds of anti-CD3 and anti-CD28 stimulations, we are able to produce on average 2 × 10⁹ Tregs (range 0.2–3.1 × 10⁹) with an average of 92% FOXP3⁺ (range 76%–97%) from 1 unit of blood donation (J Bluestone, unpubl.). In the GvHD setting, up to 70% of FOXP3⁻ conventional T cells are tolerated because conventional T cells are needed for improving hematopoietic stem cell engraftment, restoring immunity, and mediating graft-versus-leukemia effect. In the solid organ transplant setting, high-purity Tregs will be needed to ensure safety, potency, and consistency of results.

CONCLUDING REMARKS

Taming a powerful multifaceted immune response against transplanted organ requires equally potent and versatile therapy without compromising the overall immune competence of the patient. Tregs have the desired specificity, versatility, and adaptability and decades of research has shown their therapeutic efficacy in transplantation. However, Tregs do not have sufficient potency as a stand-alone therapy for transplantation, and factors critical to the efficacy of Treg therapy in transplantation are dose, specificity, and adjunct immunosuppression. Although early clinical trials will be mainly focused on safety, it is important to design these trials with efficacy end points in mind so that the safety of the right dose with the desired specificity administered with optimized adjunct therapy can be evaluated. Central to the successful implementation of Treg therapy is the reliable production of consistently high-quality Tregs and the challenge is in balancing the yield and the purity of the products. The future of Treg therapy depends on effective and informative clinical trial designs, technological advancement in Treg manufacture, and better mechanistic understanding of Treg biology and transplantation tolerance in humans.