Functional plasticity in human FOXP3⁺ regulatory T cells

Abstract

CD4⁺ regulatory T (T_reg) cells expressing the Foxp3 transcription factor are critical for the induction and maintenance of immune homeostasis and self-tolerance in experimental rodents and humans. Foxp3⁺ T_reg cells constitute a unique CD4⁺ T cell subset with potent suppressive properties, and their functional and homeostatic stability is essential to ensure dominant tolerance in a variety of inflammatory settings. Interestingly, recent evidence points to the inherent potential of T_reg cells to adapt to environmental cues and consequently manifest functional plasticity by downregulating Foxp3 expression, and reprogramming into inflammatory T cells. The potential for suppressive Foxp3⁺ T_reg cells to undergo functional plasticity and gain inflammatory properties is of concern when one considers the ex vivo manipulation or generation of such cells for therapeutic purposes in various autoimmune or chronic inflammatory disorders. Collectively, the experimental evidence accumulated so far on the modalities of this plasticity can provide valuable cues as to strategies that can be implemented to control it, potentially allowing to facilitate the path to efficient and safe Treg-based therapy.

Background

A breakdown in self-tolerance is thought to underlie many autoimmune disorders, and de novo tolerance induction to allogeneic grafts is required for successful transplantation. In recent years, CD4⁺CD25⁺ Treg cells expressing the FOXP3 transcription factor have emerged as a key mechanism of peripheral tolerance in rodents and humans. They represent 1–10% of circulating CD4⁺ T cells, and are capable of dampening the activation and function of various immune cell subsets.¹ “Natural” FOXP3⁺ Treg (nTreg) cells can arise from the thymus as a defined cellular lineage, or Treg cells can differentiate from conventional CD4⁺ T cells in the periphery (iTreg) by upregulating FOXP3 upon antigenic stimulation in specific tolerogenic contexts.² In particular, several factors have been identified, such as TGFβ or IL-2, that can promote the expression of FOXP3 by causing the demethylation of CpG motifs within evolutionary conserved regions of the foxp3 promoter.³ Stable FOXP3 expression and function are critical for the genetic programming of Treg cell fitness and function.⁴ Experiments performed in murine models of autoimmunity have shown that a developmental or functional deficiency of Foxp3⁺ Treg cells breaks down self-tolerance, and provokes autoimmunity. Moreover, natural disease progression is accompanied by a functional and/or quantitative waning of Treg cells in mouse models of autoimmunity.⁵ In humans, discrepant observations have been reported regarding the progressive dysfunction of Treg cells throughout disease development, mostly due to the lack of appropriate cellular markers, confounding variables imparted by active disease activity, ongoing immunosuppressive therapies and methodology.⁶ However, a diminished frequency or dysfunction of Treg cells has been reported in many human diseases, including systemic lupus erythematosus (SLE), rheumatoid arthritis, type 1 diabetes (T1D), multiple sclerosis (MS), and graft-vs.-host disease (GVHD), as well as transplant rejection.⁷

Given the outstanding role of Treg cells in the pathophysiology of various diseases, multiple therapeutic avenues aiming to restore or maintain the functionality of Treg cells have been investigated in murine models. However, very few could be tested in humans, even less met with success.

A major concern has recently arisen from the discovery of the functional plasticity of Treg cells. Indeed, it has been shown that, unlike what was previously thought, FOXP3⁺ Treg cells may not be a stably committed cell subset, as they can lose FOXP3 expression and adopt the pro-inflammatory functions of either Th1, Th2 of Th17 cells.⁸ As Treg cell transfer is one of the main therapeutic avenues explored and currently tested in human auto- and allo-immunity, the discovery that Treg cells also manifest functional plasticity is potentially alarming. Although it remains to be determined to which extent this phenomenon actually contributes to the onset/progression of autoimmunity, the potential reprogramming of Treg cells into inflammatory cell lineages casts a shadow on the eventual use of Treg-based immunotherapy. A deeper understanding of the factors that modulate this phenotypic and functional plasticity in FOXP3⁺ Treg cells should be attained in order to implement Treg-cell based therapies in autoimmune disease. To what extent does Treg plasticity pose a threat to the perspective of Treg-based cellular therapy?

Functional Plasticity in Foxp3⁺ Treg Cells: Lessons from the Mouse Models

Although long considered a stable and terminally differentiated lineage, it has recently been shown that murine Treg cells can lose Foxp3 expression in vitro and in vivo.⁹ The ensuing Foxp3^- population adopts the phenotype of conventional CD4⁺CD25^-Foxp3^- T cells (Tconv) which have the potential to produce pro-inflammatory cytokines such as IFNγ and IL-17.⁹ The loss of FOXP3 expression by Treg cells has been primarily observed following their adoptive transfer into lymphopenic mice.^9,10 In this context, as much as 80% of Treg cells have been reported to lose the expression of FOXP3, with this proportion varying depending on the anatomical site (mucosal vs. lymphoid organ), immune state (inflammation vs. immune quiescence), and the lympho-competence of the host.⁹ As such, we and others have shown that the injection of Treg cells into T cell-deficient mice results in their reprogramming into Th1/Th17-like Foxp3^- T cells in peripheral lymph nodes (pLN) of the host’s T cell compartment, while the emergence of these cells is augmented in mesenteric LN (mLN) and lamina propria (40% in pLN vs. 80% in mLN).¹⁰ Importantly, these Foxp3^- cells express high levels of inflammatory cytokines and carry a pathogenic potential similar to that of Tconv cells, as shown by the chronic inflammation they cause in vivo. It has also been proposed that a subset of the Foxp3⁺ Treg cell pool is susceptible to functional reprogramming, although Treg cells maintaining Foxp3 expression after lymphopenic transfer can further lose Foxp3 expression upon retransfer in new hosts.¹⁰

Importantly, all Foxp3⁺ Treg cell subsets are not equally prone to undergo functional plasticity. Indeed, in vitro studies in mice have shown that iTreg cells are more prone to FOXP3 loss than are nTreg cells, possibly due to their already plastic origin. Indeed, the determining difference identified between these two subsets is the level of methylation of the foxp3 promoter, whereby the promoter of foxp3 is fully demethylated in nTreg cells, and only partially demethylated in iTreg cells in mice and humans.³ In addition, Treg cells of thymic or peripheral origin do constitute distinct subsets, as shown by their distinct gene expression signatures.¹¹ Differences within the iTreg cell subset also exist depending on the site of isolation and/or their mode of induction. This heterogeneity of the Treg cell population has led to the hypothesis that Treg plasticity is limited to a minor subset of peripheral Treg cells in mice.¹² Thus, both homeostatic mechanisms and differentiation status may direct the phenomenon of Treg plasticity.

In humans, it has been shown that Treg cells isolated as CD4⁺CD25⁺CD127^Low T cells progressively lose regulatory potency upon in vitro culture, and increasingly secrete inflammatory cytokines such as IL-2, IL-17 and IFNγ.¹³ We and others have also reported that a significant proportion of human CD4⁺CD25⁺ T cells (10–30%) can suddenly lose FOXP3 expression within 48hrs after in vitro restimulation.¹⁴ Interestingly, the naïve, CD45RA⁺ sub-population of Treg cells is consistently less prone to the loss of FOXP3 than memory, CD45RO⁺ Treg cells.^13-15 As naïve Treg cells are by definition not induced but of thymic origin, these observations suggest that human nTreg cells, like their murine counterpart, are more capable of maintaining their lineage than are peripherally induced, antigen-experienced iTreg cells.³ Ex vivo, it has been shown that FOXP3⁺ Treg cells can exhibit a certain pluripotency, whereby they can co-express FOXP3 and inflammatory cytokines such as IFNγ and/or IL-17.¹⁶ However, these cells seem to retain their suppressive potency, and not cross-differentiate into fully pathogenic Tconv-like T cells. Hence, the extent of Treg plasticity that can be expected from human Treg cells in vivo remains elusive.

Current Treg-Targeting Therapeutic Approaches

The main therapies that have been explored revolve around two concepts: (1) supporting the fitness and homeostasis of the endogenous Treg population; and (2) infusion of Treg cells, either autologous or allogeneic, in vitro induced or genetically engineered.

Enhancement of the endogenous Treg population can be obtained in vivo by the selective activation, expansion and survival of Treg cells. The former can be achieved by triggering the expansion of Treg cells via the activation of their TCR, either polyclonally, or in an antigen-specific fashion using self-peptide vaccination. Both have proven to be efficacious in mouse models of autoimmunity.^17-21 These encouraging results have made vaccines and cellular transfer therapy important avenues currently pursued in clinical trials.²² Approaches selectively favoring the survival of Treg cells include the provision of low doses of IL-2, or the use of immunosuppressive pharmaceutical agents such as Cyclosporin A and Rapamycin, which have recently been found to selectively modulate biochemical pathways in Tconv or Treg cells.²³ Helpful in human hematopoietic stem cell transplantation, these approaches have so far only been proven beneficial for autoimmunity in mice.^24,25

The infusion of Treg cells is the second main type of therapy pursued in humans.²² Thus far, adoptive immunotherapy with Treg cells has been shown to be effective in mice in the prevention of experimental autoimmune encephalomyelitis (EAE), T1D, SLE, autoimmune gastritis, and inflammatory bowel disease (IBD).²⁶ Consequently, the opportunity of Treg cellular therapy for autoimmune and chronic inflammatory disorders as well as transplantation is currently the focus of several clinical trials.²⁷ However, Treg cells constitute a small population in PBMC, and obtaining sufficient amounts of cells for human-scale mass-injection has proven challenging. It requires either Treg isolation followed by in vitro expansion, or the manipulation of Tconv precursors into Treg cells. Indeed, Tconv cells can be reprogrammed into Treg cells in vitro either by lentiviral transduction of FOXP3 and its ectopic and stable expression,²⁸ or by inducing FOXP3 expression by stimulating Tconv cells in the presence of TGFβ and IL-2 thus producing iTreg cells.²⁹ Other soluble factors have been shown to improve this induction. Notably, retinoic acid has been shown to improve the stability of FOXP3 expression in these iTreg cells, and rapamycin has been shown to promote the survival of these cells as compared with those remaining FOXP3^-.^30,31

How Does Plasticity Affect the Perspective of Treg-Based Immunotherapy?

One major limitation to cellular therapy is therefore the potential lack of stability if the transferred Treg cells. The risk of loss of function of injected Treg populations, and their subsequent potential conversion into pathogenic T cells, has cast doubts over the future of Treg cell-based immunotherapy. The use of auto-antigen specific Treg cells, rather than total polyclonal populations, has also been proposed, and may even turn out to be a necessity for efficiency. With many Treg cells bearing autoreactive TCRs, stabilization of the Treg phenotype is essential to avoid potential complications of Treg cells serving as a reservoir of autoreactive Tconv cells. This possibility has become the main hurdle faced by the field of Treg therapy.

The current therapies that are potentially affected by Treg cell plasticity are mainly those relying on the isolation and expansion of Treg cells ex vivo, or on the in vitro conversion of Tconv cells into Treg cells. The studies establishing the capacity of Treg cell transfers to ameliorate or delay autoimmune diseases in mice pre-date the discovery of FOXP3 loss. As a result, the link between Treg plasticity and an ensuing pathogenicity is difficult to formally establish. Nonetheless, it is quite striking that lymphopenic mice injected with pure Treg cells, where 40–80% of the Treg cells lose FOXP3, do not develop disease despite these very high rates of Treg to Tconv conversion, whereas the transfer of Tconv in these mice leads to chronic bowel inflammation.^9,10 Similarly, autoimmunity-prone NOD mice do not develop diabetes when injected with autoantigen-specific, NOD.BDC2.5 Treg cells.²⁴ Moreover, the fact that Treg cell transfer can prevent autoimmunity in mouse experimental systems identical to those where the phenomenon of plasticity was demonstrated, points to the possibility that Treg plasticity does not stand in the way of Treg-based immunotherapy of autoimmunity. Some critical points may be taken from these experiments, in particular, the fact that lymphopenia-driven Treg plasticity, although extensive, seemingly does not lead to disease, and the fact that a Treg-based intervention has greater potential of success if performed very early in the pathogenesis (i.e., simultaneously to a transplantation, or soon after a diagnosis of autoimmunity).²²

Another critical parameter to consider for Treg infusions may be the relative iTreg/nTreg composition of the injected cells. As mentioned earlier, iTreg cells and nTreg cells differ both in nature and in stability. In the absence of a distinctive marker, it is not possible to determine what proportion of peripheral Treg cells are of either origin. However, because iTreg cells are believed to arise as a result of exposure to antigen, it seems likely that iTreg cells amount to a greater proportion of the peripheral Treg pool in humans than in SPF-inbred mice. Thus, as Treg cells commonly used in most cell transfer experiments in mice are isolated from peripheral lymphoid organs of naïve mice, they can be considered enriched for nTreg cells as compared with human Treg cells isolated from peripheral blood. Indeed, it has been shown that the human Treg cell compartment has a relatively short turnover, and is actively replenished by antigen-experienced, CD45RO⁺CD4⁺ T cells.³² It follows that human adult peripheral blood and umbilical cord blood, for instance, will greatly differ in their nTreg vs. iTreg content. Subsequently it has been proposed to isolate substantial amounts of FOXP3⁺ Treg cells from umbilical cord blood, thus obtaining an nTreg-enriched population of Treg cells, and expanded in vitro so as to obtain sufficient cell numbers.

Incidentally, clinical trials were initiated prior to the discovery of Treg instability, and involve the injection of Treg cells in the context of the prevention of GVHD following umbilical cord blood transplantation (UCBT) in patients suffering from hematologic malignancies.³³ The results have been encouraging, and strikingly reflect several highlights of the murine work. Indeed, the patients underwent conditioning, and as such were in a state of lymphopenia, and the cells injected were isolated from umbilical cord blood, thus most likely highly enriched in nTreg cells. Finally, the injection was concomitant to the reconstituting UCBT, and thus at the initiation stage of the targeted inflammatory disorder. The treatment was fairly successful, as the grade of GVHD was significantly reduced compared with historic control.³³ Most importantly, no detrimental side effect to the Treg infusion was reported. While the extent of Treg plasticity was not established, the outcome strongly suggests that it did not represent a threat to the patients.

The feasibility of transposing these findings to the benefit of autoimmune patients remains unknown. One major trial that is expected to bring insights into the feasibility, but also the safety, of Treg therapy, is the one initiated by the University of California (ClinicalTrials.gov Identifier: NCT01210664),²² where polyclonal, in vitro expanded, PBMC-derived Treg cells will be injected into early onset T1D patients. Notably, this protocol is not completely in line with the UCBT study, as the patients are recruited based on well-established disease, will not be conditioned, and the Treg cells will be of adult origin, and thus can be expected to comprise of a significant proportion of iTreg cells.

Strategies to Overcome the Problem

Strategies can be chosen to reduce the plasticity of Treg cells and maximize the potential of therapeutic efficiency of Treg-based immunotherapy. It will be of primordial importance to determine which markers will accurately define Treg populations for their isolation and ex vivo manipulation, as well as the functional parameters that will be used to assess success or failure of Foxp3⁺ T_reg cell-mediated disease protection. The time and frequency of injection, the lymphocompetence and genetics of the host are also important variables to monitor in these protocols.

The stability of transferred Treg cells can be greatly increased by the adjunction of factors that stabilize FOXP3 expression to the in vitro expansion conditions. These could include the use of retinoic acid, IL-10, and/or the use of methylation inhibitors such as azacytidine, which favor the demethylation of the foxp3 promoter. As such, it has been shown that methyl-transferase inhibitors could impact FOXP3 expression in vivo in mice.³⁴ Finally, there is a lack of consensus regarding the criteria of assessment of purity and stability of the injected cells. Indeed, rather than the expression of FOXP3 in itself, it would be interesting to use markers providing information on the expectable stability of the lineage. This is currently only feasible by the measurement of the methylation status of the foxp3 promoter.

Interestingly, Battaglia et al. recently proposed a “two-step” model of immunotherapy for patients suffering from T1D. While this protocol proposes to induce Treg cells in vitro from Tconv cells, it offers several novel avenues. In particular, they suggest to create a lymphopenic environment, an idea supported by the successful, yet transient, improvement of this disease by HSCT following total conditioning,³⁵ and to use Treg-stabilizing treatment.²² The potential results of such a clinical trial would certainly contribute a wealth of knowledge regarding the hopes of Treg-base immunotherapy, as it attempts to integrate a very large part of the current information, from both mice and men.

Conclusion

In conclusion, caution is required in the therapeutic use and manipulation of Treg cells in humans, as the factors that maintain their functional stability remain ill-defined. In particular, relying on the pre-clinical information gathered from mouse models will provide valuable guidance in the parameters that are at play, even though it remains uncertain how informative the mouse models are with respect to the human immune system. Nonetheless, the current evidence suggests that the plasticity of Treg cell subsets may constitute less of a hurdle than previously imagined, provided certain strategies are implemented to contain it.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.