SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References

Immunological tolerance is one of the fundamental concepts of the immune system. During the past decade, CD4+CD25+-regulatory T cells have emerged as key players in the development of tolerance to autoantigens as well as to foreign antigens. Still many questions remain illusive regarding the basic properties of CD4+CD25+-regulatory T cells. This review aims to recapitulate some of the current understandings about the phenotype, function and clinical relevance of murine and human CD4+CD25+-regulatory T cells.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References

The main role of the immune system is to protect the individual from pathogens, and one of its fundamental qualities is the ability to distinguish between self and nonself and between antigens encountered in harmful and nonharmful contexts. In the thymus, potentially self-reactive T cells are deleted, resulting in the generation of a peripheral T-cell repertoire that is largely self-tolerant. Despite this, some self-reactive T cells are present in most individuals. Nevertheless, autoimmune diseases only occur infrequently, which suggests that autoreactive T cells are controlled in the periphery. Peripheral tolerance is sustained by several mechanisms such as deletion, anergy and ignorance [1, 2]. In addition, there is compelling evidence for the existence of more ‘active’ mechanisms of tolerance that operate through the generation of immune-regulating T cells. Several types of regulatory T cells, such as γδ T cells, NKT cells, CD8+ and CD4+ T cells, have been described [3]. CD4+-regulatory T cells can be further divided into induced regulatory cells that secrete interleukin-10 (IL-10) and transforming growth factor-β (TGF-β) such as Tr1 cells [4] and T-helper 3 (Th3) cells [5] and the so-called naturally occurring CD4+CD25+-regulatory T cell (CD25+Treg), which is the focus of this review.

T-cell-mediated suppression of autoimmune disease was first described by Nishizuka and Sakakura more than 30 years ago [6]. They discovered that thymectomy on day 3 of life (d3Tx) results in organ-specific autoimmunity. However, disease did not develop in mice thymectomized as early as day 2 or as late as day 7 of life. Further studies showed that d3Tx animals could be rescued from disease if reconstituted with thymocytes or splenocytes from a normal adult animal but not from an adult animal that had been d3Tx [7]. This indicated that murine T cells that exit the thymus before d3Tx are qualitatively different from cells that emigrate later on. However, the cells responsible for the inhibition of autoimmune disease were not discovered until the mid-1990s when Sakaguchi and coworkers identified a subpopulation of CD4+ T cells expressing the IL-2 receptor α-subunit (CD25) [8]. Depletion of CD25+ T cells from adult splenocytes followed by transfer of CD25 T cells to immune-deficient hosts resulted in a spectrum of organ-specific autoimmune diseases similar to d3Tx. In addition, cotransfer of CD25+ T cells prevented induction of autoimmunity in the cell-transfer model as well as in the d3Tx model [9]. Owing to intense research during the past decade, CD25+Treg has emerged as a central T-cell population for preserving peripheral tolerance, not only to autoantigens but also to foreign antigens, in mice as well as in humans [10, 11].

Phenotype of CD25+-regulatory T cells

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References

Naturally occurring regulatory T cells express CD25 constitutively, and this marker has proven to be very useful for isolation of these cells in mice. However, CD25 is not an optimal marker as it is upregulated upon activation of T cells. This is especially apparent when investigating human CD25+Treg. In the naïve mouse, CD4+CD25+ T cells are seen as a distinct population of cells easily distinguished from CD4+CD25 T cells that comprise between 5 and 10% of peripheral CD4+ T cells. The isolation of murine CD25+Treg is therefore fairly straightforward unless the animal is suffering from ongoing inflammation. Among human CD4+ T cells, approximately 30% express CD25 (Fig. 1). The majority of these cells express CD25 with low-to-intermediate intensity (CD25int) and only between 1 and 3% of the CD4+ T cells express CD25 with high intensity (CD25high) [12]. In vitro studies of sorted CD25int and CD25high cells have shown that it is the CD25high population that functions as suppressor cells [13]. Consequently, CD25int cells are most likely memory cells with CD25 expression resulting from encounter with foreign antigens. The continuous expression pattern of CD25 on CD4+ T cells from adult peripheral blood has made the isolation of human CD25+Treg with high purity difficult, which should be remembered when assessing data regarding suppressive potential. It should also be noted that there have been several reports on CD4+CD25 cells with regulatory properties [14]. This implies that naturally occurring CD4+-regulatory T cells are not necessarily confined to the CD4+CD25+ T-cell population.

image

Figure 1. Expression of CD25 on CD4+ T cells from adult peripheral blood. CD4+ T cells with the highest expression of CD25, but not those with low or intermediate expression of CD25, display a Treg phenotype and suppress T-cell responses in vitro.

Download figure to PowerPoint

Other CD25+Treg-surface markers

In mice, surface expression of markers other than CD25 have been useful in isolating CD4+-regulatory T cells, including CD45RB, CD38 and CD62L [15–17]. Notably, in vitro all freshly isolated murine CD25+ T cells suppressed the proliferation of CD4+CD25 T cells irrespectively of their expression of these markers [18]. However, this does not exclude that the suppressive property in vivo could be affected. Indeed, CD4+CD25+CD62L+ but not CD4+CD25+CD62L splenocytes were shown to delay the onset of diabetes in NOD mice [19]. However, these markers do not necessarily identify human CD25+Treg. We investigated mononuclear cells from adult peripheral blood, cord blood and thymus in order to see whether markers that identified CD25+Treg in mice could also be used to identify the human population of CD25+Treg [12]. We found that CD25highTreg in adult blood expressed intracellular cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4/CD152) and CD122, while CD25 and CD25int T cells were negative for CTLA-4 and expressed low levels of CD122. Furthermore, similar CD25+ T cells were identified in cord blood and in thymus. Both CTLA-4 and CD122 are markers that are expressed by murine CD25+Treg [8] (E. Suri-Payer, unpublished observation). In addition, CD25+Treg from adults displayed a memory phenotype as they were CD45RARO+, CD45RBlow and expressed both CD62L and CD38 with low intensity, and this phenotype was largely shared with the CD25int T cells. In contrast, CD25+Treg derived from cord blood had a naïve phenotype and were mainly CD45RA+RO as were the CD25 T cells in cord blood. Notably, Jonuleit et al. showed that in adult peripheral blood, only the CD25+CD45RO+ cells have suppressive ability [20]. However, we found that cord blood CD25+Treg, which are mainly CD45RA+, were able to suppress proliferation induced by anti-CD3Ab [21]. Furthermore, CD45RA+CD25+ T cells expressed twofold higher levels of Foxp3 mRNA than CD45RACD25+ T cells did [22]. This suggests that expression of CD45RO is an indicator of antigen experience and of limited use for identification of CD25+Treg in humans.

A number of studies have identified additional markers that are expressed at higher levels on CD25+Treg compared to CD25 T cells [14]. Some of these markers are inhibitory costimulatory receptors like PD1 or members of the tumour necrosis factor receptor (TNFR) superfamily, such as GITR (glucocorticoid-induced TNFR-related protein), OX40, 4-1BB and TNFRII. Yet others are chemokine receptors, Toll-like receptors or homing receptors such as CD103 (αEβ7 integrin) (Fig. 2). Other recently discovered markers are LAG-3, an MHC class II-binding CD4 homologue [23] and neuropilin (Nrp1), which is involved in axon guidance, angiogenesis and T-cell activation [24]. However, the majority of these markers should be used with caution since most surface markers, although expressed on CD25+Treg, are upregulated also on CD25 T cells after stimulation. In addition, several of these markers were identified on CD25+ T cells isolated from mice and the expression has been difficult to confirm in humans. Currently, none of these molecules have proven to be fully responsible for the suppressive function of CD25+Treg and questions regarding their functional significance remain.

image

Figure 3. Proposed interactions of effector T cells and CD25+Treg with GITR-L. (A) Modulation of the immune response in the steady state. Interaction of GITR-L and GITR on effector T cells makes them refractory to suppression. Signalling through GITR in the presence of IL-2 results in the expansion of CD25+Treg. (B) Activated APC downregulate GITR-L, which facilitates suppression. APC, antigen-presenting cells.

Download figure to PowerPoint

Foxp3

Recent studies have revealed the gene Foxp3 to be central in the development and function of CD25+Treg [25]. The importance of Foxp3 was discovered when the underlying defect in scurfy mice was investigated. Scurfy mice suffer from a spontaneous X-linked mutation, which leads to fatal lymphoproliferative disease associated with multiorgan infiltrates and early death by 3–4 weeks of age in hemizygous males [26]. Lately, the genetic defect in scurfy mice has been identified as a mutation in Foxp3 (forkhead box p3), a gene coding for a member of the forkhead/winged-helix family of transcriptional regulators [27]. The Foxp3 protein (also known as scurfin) contains a C-terminal forkhead domain, a single C2H2 zinc finger and a leucin zipper motif. Scurfy mice have a two base-pair frameshift insertion resulting in a truncated product, which lacks the forkhead domain and the nuclear localization signal, and this suggests that it is unable to function as transcriptional regulator [27, 28]. FOXP3, the human orthologue of the murine Foxp3, has been found to be mutated in patients suffering from IPEX (immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome), a severe and fatal autoimmune/allergic syndrome, which reminds to a great extent of the condition of scurfy mice. The majority of the mutations in the human patients have been found to be located to the forkhead region, further showing the importance of this domain in the function of FOXP3[29].

The fact that scurfy mice are hyperresponsive to TCR stimulation and that overexpression of Foxp3 in cells induce poor proliferation and limited production of IL-2 prompted the investigation of the relationship between Foxp3 expression and CD25+Treg [30]. These studies showed that both mRNA and protein levels of Foxp3 were confined to the CD4+CD25+ subset [31–33]. In mice, Foxp3 mRNA is detected in peripheral CD4+CD25+ T cells and in CD25+CD4+CD8 thymocytes, whereas other thymocytes/T cells or B cells do not express Foxp3. In contrast to the previously mentioned markers for CD25+Treg, Foxp3 was not induced in T cells after TCR stimulation [31]. Importantly, retroviral transduction of both murine and human naïve CD25 T cells with Foxp3 converts them to regulatory cells with functional characteristics similar to naturally occurring CD25+Treg. Interestingly, as a result of the transduction, cell-surface molecules associated with CD25+Treg were upregulated, including CD25, CTLA-4, GITR and CD103 [31, 34]. In addition, Rudensky et al. performed experiments in which mice were reconstituted with both Foxp3 and Foxp3+ bone marrow cells. These chimeric mice had CD4+ T cells that were mosaic with respect to expression of either the wildtype or the mutant allele but remained healthy, which is also seen in female carriers of IPEX. Importantly, CD25+CD4+CD8 that developed in the thymus and in the periphery of these animals all originated from the Foxp3+ bone marrow. Also, adoptive transfer of wildtype CD25+CD4+ T cells but not CD25CD4+ T cells rescued scurfy mice from disease, which shows that Foxp3 is required for the development of CD25+Treg [32]. Investigations of Foxp3 in humans have largely confirmed the expression pattern seen in mice, and CD4+CD25+FOXP3+ T cells have been identified in thymus as well as in adult peripheral blood and cord blood [22, 34–38]. In contrast to the murine studies, human CD4+CD25 T cells have been reported to express FOXP3 and to acquire suppressive ability after stimulation in vitro with plate-bound anti-CD3 and anti-CD28 MoAbs [35]. Recently, CD4+CD25 T cells in human intestinal lamina propria were reported to contain FOXP3+ cells [39]. These cells were anergic upon TCR stimulation in vitro but did not display suppressive function when cultured in a one-to-one ratio with responder T cells, which possibly was due to a low frequency of FOXP3+ cells. Of note, TGF-β which is abundant in the intestine has been shown to induce FOXP3 expression in both murine and human CD4+CD25 T cells [40]. Conclusively, costimulation with TGF-β might be a key for development of peripheral regulatory T cells. Overall, Foxp3 plays a vital part in the generation CD25+Treg and is the most specific marker currently available. Since Foxp3 is a nuclear protein, it is of limited value as a tool for isolation of CD25+Treg ex vivo. However, it will no doubt be very useful in the search for cell-surface markers specific for CD25+Treg.

Activation of CD25+-regulatory T cells

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References

Freshly isolated CD25+Treg are not able to suppress T-cell responses and only exert inhibitory function after stimulation via the TCR. Antigen-specific as well as polyclonal TCR stimulation activates CD25+Treg and induces suppressive function in vitro, whereas irrelevant antigens do not [41, 42]. CD25+Treg are very sensitive to stimulation with antigen and are suppressive at antigen doses 10–100 times lower than those needed to activate CD25 T cells [41]. They have a diverse TCR repertoire and are therefore able to respond to many different antigens, including food, microbial, allo- and autoantigens [21, 25, 43, 44].

Target cells for CD25+Treg mediated suppression

CD25+Treg have been most thoroughly studied with regard to their effects on T cells. Once CD25+Treg have been activated with specific antigen and IL-2, they inhibit the IL-2 production of their target cells and suppress both CD4+ and CD8+ T-cell responses of proliferation and cytokine production in an antigen nonspecific manner [18, 41, 42, 45]. The effect of CD25+Treg on antigen-presenting cells (APCs) including DC is controversial. CD25+Treg have been found to downregulate the expression of costimulatory molecules and reduce the stimulatory capacity of both human and murine DC [46–48]. Others have reported that CD25+Treg act directly on the target cells since suppression can be detected using in vitro culture systems devoid of APC [18, 45, 49]. This does, however, not exclude that CD25+ Treg influence the stimulatory capacity of APC in vivo. The effect may not be directly implicated in contact-mediated suppression on target cells but may contribute to regulation of the immune response in general. The inhibitory effect of CD25+Treg on B cells is not clear but class switching to IgE has been shown to be inhibited in vivo after adoptive transfer of CD25+ T cells, but whether this is a direct effect on the B cells is not known [50]. In addition, both natural killer T cells (NKT) and NK cell functions have been reported to be downregulated by CD25+Treg [51, 52].

The importance of IL-2

A distinctive feature of CD25+Treg in vitro is that they are hyporesponsive to stimulation and do not proliferate and produce either no or low levels of cytokines [53, 54]. Accordingly, they are dependent upon the cytokines that the effector cells produce. IL-2 seems to be particularly important since mice deficient for IL-2, IL-2Rα or IL-2Rβ have very few or no CD25+Treg and prematurely succumb to severe lymphoproliferative and autoimmune syndromes. Administration of IL-2 or transfer of IL-2-producing cells to IL-2 deficient animals restores the production of CD25+Treg and lymphoid homeostasis. Furthermore, thymic expression of IL-2Rβ in the thymus of IL-2Rβ–/– mice restores the production of CD25+Treg and prevents lymphoproliferation and lethal autoimmunity [55]. This indicates that IL-2 has an important role in the generation of CD25+Treg in the thymus and is crucial for peripheral homeostasis. The importance of IL-2 is also described by the fact that stat5–/– mice have few CD25+Treg, while mice transgenic for the active form of STAT5 display a greater frequency of these cells [56]. Studies of CD25+Treg activation in vitro have shown that IL-2 is needed for induction of suppressive ability. Murine CD4+CD25+ T cells cultured with plate-bound CD3 Ab in the absence of IL-2 resulted in both poor recovery and suppressive ability [57]. More importantly, the addition of anti-IL-2 completely abrogated the suppressive effect of CD25+Treg on IL-2 mRNA transcription. Of note, it appears that IL-4 can substitute for IL-2 in vitro, as the inhibition of IL-2 mRNA remained when IL-4 was added to the cultures in the presence of anti-IL-2 [58].

Need for costimulation?

Activation of naïve CD4+CD25 T cells requires two signals: one through the TCR by MHC–peptide interaction and a second via costimulation delivered by APC. The primary costimulatory signal is mediated when CD28 binds to CD80 or CD86 expressed on the APC. The requirements of costimulation for the homeostasis and induction of suppressive function in CD25+Treg are not fully determined. NOD mice deficient for CD80/86 or CD28 have decreased numbers of CD25+Treg and display exacerbated spontaneous diabetes, which can be reversed upon transfer of WT CD25+ T cells [59]. CD28 has been shown to be important for the thymic development and also for the peripheral homeostasis of CD25+Treg. Conceivably, CD28 mediates a direct effect on the CD25+Treg by enhancing the TCR signalling and an indirect effect by inducing IL-2 production of effector cells [60]. In vitro studies of human and mouse CD25+Treg have shown that stimulation with anti-CD28 Ab in addition to a strong TCR stimulus breaks the suppression of both proliferation and transcription of IL-2 mRNA [13, 57, 61]. This suggests that suppression by CD25+Treg could be abrogated during a strong inflammatory response in vivo. A recent in vitro study showed that CD25+Treg were fully competent suppressors, irrespectively of whether they were preactivated in the presence of wildtype APC, CD80–/– APC or CD86–/– APC [57]. This indicates that CD28 signalling in CD25+Treg may be crucial for the development and homeostasis but not necessary for induction of suppression.

The role of CTLA-4

Cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4; CD152) is a CD28 homologue, which also binds to CD80/86. CTLA-4 is induced upon T-cell activation and then functions as a negative regulator of activation. Interestingly, the only cells in naïve animals or human cord blood that express CTLA-4 in the absence of activation is CD4+CD25+ T cells [12, 62]. This raises questions regarding the role of CTLA-4 for the inhibitory mechanism and induction of suppressive capability in CD25+Treg. Several groups have investigated the involvement of CTLA-4 in the suppressive mechanism. The addition of CTLA-4 Ab or Fab fragments to in vitro cocultures of murine CD4+CD25 and CD25+ T cells neutralizes the inhibitory effect [62], and administration of CTLA-4 MoAb abolished the protective ability of CD25+Treg in the murine inflammatory bowel disease (IBD) model [63]. The fact that CD4+CD25 T cells also express CTLA-4 after stimulation causes problems when interpreting these results. In a recent study, CD25+Treg were preactivated in the presence of anti-CTLA-4 Fab fragments before they were added to cultures with CD25 T cells. This experiment showed that CD25+Treg were fully competent suppressors despite blocking of CTLA-4 [57]. With regard to human in vitro studies, it was recently shown that suppression by CD4+CD25+CTLA-4+ T cells was partly inhibited by the addition of anti-CTLA-4 Ab [64]. However, the majority of investigations have not been able to establish a role for CTLA-4 in the suppressive function [13, 20, 65]. Furthermore, CTLA-4–/– mice develop a fatal lymphoproliferative disease, but CD25+Treg development and homeostasis appear normal and CD25+Treg displayed uncompromised suppressive ability in vitro. Interestingly, CTLA-4-deficient CD25+Treg expressed increased levels of TGF-β and suppression was decreased by neutralizing TGF-β. This implies that lack of CTLA-4 induces compensatory mechanisms of suppression in CD25+Treg which could explain the current contradictions in the literature [66].

The role of GITR

Freshly isolated CD25+Treg from normal mice constitutively express high levels of GITR, while CD4+CD25 T cells, CD8 T cells, B cells or DC and macrophages express GITR at low levels, which increases after activation [10, 67, 68]. The high expression level on CD25+Treg suggested an important role for GITR on this subset. Original reports on the relationship between GITR and CD25+Treg showed that TCR stimulation in combination with cross-linking of GITR with anti-GITR Ab abrogated CD25+Treg-mediated suppression, while Fab fragments did not. This indicates the need for signal transduction and not mere blocking of the receptor. In addition, administration of anti-GITR Ab resulted in induction of autoimmune disease in normal mice [10, 67]. The abrogation of suppression could depend on anti-GITR Ab signalling on the CD25+Treg as well as on the CD25 T cells since CD25 T cells upregulate the expression of GITR when activated. However, anti-GITR Ab was shown to break the anergic state of the CD25+Treg, while the proliferation of CD25 T cells was unaffected [10]. Furthermore, anti-GITR Ab reactive only with mouse cells were shown to abrogate suppression in cocultures of CD25+Treg from mice together with CD25 T cells from rats [67]. The authors concluded that the abrogation of suppression was dependent on signalling on the CD25+Treg and not the CD25 T cells, which leads to both inhibition of suppressive activity and reversal of nonresponsiveness to IL-2 of the CD25+Treg. Recent data have resulted in reassessment of this deduction [69]. By using GITR–/– mice, the authors were able to show that suppression was only abrogated when anti-GITR Ab was added to cocultures of GITR–/– CD25+Treg with GITR+/+ CD25 T cells and not vice versa, which suggests that GITR engagement provides CD25 T cells with a signal that raises the threshold for suppression by CD25+Treg. In support of this, experiments using rat CD25 T cells and mouse CD25+Treg with CFSE labelling as readout system for suppression revealed that the rat CD25 T cells remained suppressed in the presence of anti-GITR Ab while the mouse-derived CD25+Treg proliferated [69]. It has previously been shown that CD25+Treg proliferate after engagement of GITR in the presence of exogenous IL-2 [10]. GITR–/– mice have been shown to have a similar frequency of CD25+Treg in thymus as normal animals but the frequency in lymph nodes and spleen is reduced by approximately 30% [69]. This indicates that GITR signalling could be partly responsible for the homeostasis of CD25+Treg. Interestingly, the ligand for GITR (GITR-L) is expressed on resting APC and is downregulated by triggering the B-cell receptor, CD40 and different Toll-like receptors (TLR). This raises the possibility that resting APC can stimulate a nonspecific expansion of CD25+Treg by GITR/GITR–L interaction in the presence of IL-2 secreted by effector T cells early in the course of an immune response (Fig. 3). Such an early expansion of CD25+Treg might be important in order to control the effector cells in the late phase of the reaction although further studies are needed.

Bacterial stimulation

A recent report showed that CD25+Treg express several members of the TLR family, such as TLR-4 [70]. TLRs recognize certain components, so-called pathogen-associated molecular patterns, which are shared by most microbes, and also certain endogenous molecules that are released during inflammation. Caramalho et al. found that in vitro stimulation of murine CD25+Treg with a high dose of lipopolysaccharide (LPS) through TLR-4 induced proliferation and enhanced their survival and suppressive capacity even in the absence of APC. This indicates that LPS acts directly on TLR-4 molecules expressed by CD25+Treg and improves their capacity to suppress. Another group has reported opposite results. They showed that stimulation of DC with LPS or CpG reversed the CD25+Treg-mediated suppression, and this reversal was partly dependent on IL-6 production by the DC [71]. It is possible that both scenarios could take place in vivo and such things as the concentration of the stimulating agents or the local cytokine milieu may regulate the outcome. Undoubtedly, bacterial stimulation influences the maturation of the immune system. Several epidemiological studies have shown a correlation between improved hygienic conditions and the development of IBD, allergies and autoimmune diseases [72]. It is possible that a reduced exposure to microbes affects the development of tolerance and also the homeostasis of the CD25+Treg pool.

Mechanism of suppression

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References

The mechanism of suppression by CD25+Treg is poorly understood. The majority of murine and human in vitro studies have concluded that CD25+Treg mediate suppression by a yet unknown cell-contact-dependent mechanism, which is cytokine independent. Suppression cannot be abrogated by neutralizing IL-4, IL-10 or TGF-β, and CD25+Treg cultured with CD25 T cells in a transwell system are unable to suppress the proliferation of the responder cells [20, 41, 42]. Interestingly, human CD25+Treg fixed with paraformaldehyde remained suppressive as long as they had been activated before fixation [73, 74]. Collectively, this suggests the involvement of a surface-bound molecule that is upregulated on CD25+Treg upon activation and mediates a suppressive signal to the responder cell. However, no such agent has yet been identified even though CTLA-4 has been proposed as a candidate. Another suggested mechanism of cell-contact-dependent suppression is by TGF-β bound to the cell surface of CD25+Treg [75, 76]. These findings have been corroborated by some authors, as at least suppression by human thymic CD25+Treg seems to be partly dependent on TGF-βin vitro[77] (our unpublished observations). In contrast, others have had difficulties reproducing the results by Nakamura et al. [78]. The potential role of TGF-β remains controversial as CD25+Treg from TGF-β1-deficient mice suppress CD25 T cells in vitro[78], while adoptive transfer of TGF-β1-deficient CD25+Treg did not protect recipients from colitis in the SCID transfer model in vivo[76]. These results suggest that TGF-β produced by CD25+Treg is of particular importance in regulation of intestinal inflammation. However, in vitro it might be one of several mechanisms and in this setting other mechanisms of suppression could be of greater importance.

Mechanisms of suppression in vivo

There is a marked contrast with regard to the importance of immunosuppressive cytokines in vivo as compared to CD25+Treg suppression in vitro, and several cytokines have been implicated as mediators of inhibition. In the murine model of IBD, neutralizing Ab to IL-10 or TGF-β were shown to abolish the protective effect of the CD4+CD45RBlow cells [15, 79]. Similar results have been obtained using adoptive transfer of IL-10–/– CD25+Treg. In contrast, IL-10-deficient CD25+Treg are able to inhibit the development of gastritis [80]. One important difference between autoimmune gastritis and IBD is the requirement for intestinal bacteria for induction of IBD, as transfer of CD25 T cells to germ-free mice does not result in disease [81]. In vivo, it is likely that cell-contact-dependent suppression by CD25+Treg is needed but, during inflammation in the intestine, IL-10 and TGF-β are also required to control the response. Still, in conditions less dependent on bacterial presence, for instance in autoimmune thyroiditis, protection from disease is reversed by neutralizing Abs to IL-4 and TGF-β[82]. Similarly, the protection of NOD mice by transferred CD4+CD25+CD62L+ is abrogated after treatment with anti-TGF-β Ab [17]. Overall, these data indicate that more than one mechanism of CD25+Treg suppression is operating in vivo. One possibility is that there are different subsets of CD25+Treg that either suppress by cell-contact-dependent mechanisms or suppress via production of cytokines. Alternatively, one CD25+Treg might suppress by more than one mechanism depending on the situation. Factors such as the intensity of the immune response, the local cytokine environment and pathology of the disease might be very important in order to optimise the suppressive response by CD25+Treg.

The possibility for a third mechanism of action was raised by two groups who simultaneously showed that human CD25+Treg are able to induce suppressive properties in CD4+CD25 T cells when cultured in vitro. This ‘infectious tolerance’ rendered the CD25 T cells anergic and they subsequently started to produce TGF-β[74] or IL-10 [73]. The primary culture of CD25+Treg together with CD25 T cells required cell contact for induction of anergy. However, when the anergized T cells were transferred to fresh cultures, they were shown to suppress naïve T cells in a cytokine-dependent and cell-contact-independent manner. Interestingly, these two studies reached opposite mechanistic conclusions possibly, due to different culture systems, stimuli and length of contact with the CD25+Treg. This mechanism of infectious tolerance could clarify the discrepancy in the in vivo data and might also explain how a small population of cells can regulate a much larger population of responder T cells in vivo.

Development of CD25+Treg

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References

Several findings point to the thymus as central in the generation of CD25+Treg. First, d3Tx mice develop a multitude of organ-specific autoimmune diseases such as gastritis, thyroiditis and pancreatitis depending on the genetic background [83]. Second, CD4+CD8 thymocytes depleted of CD25+ cells induce autoimmune diseases when transferred into BALB/c athymic nu/nu mice, while the unseparated cell population does not [84]. Third, studies of human and murine SP CD4+CD25+ thymocytes have revealed suppressive function in vitro both to polyclonal and antigen-specific stimuli [22, 77, 84, 85]. In addition, both human and murine CD4+CD8CD25+ thymocytes are phenotypically similar to peripheral CD25+Treg and they express Foxp3/FOXP3[12, 30, 32, 36, 62, 67]. Therefore, it appears that the thymus not only contributes to self-tolerance by deleting autoreactive T cells but also by producing CD25+Treg.

Thymic selection of CD25+Treg

The development of CD25+Treg in the thymus is still under investigation and it is not clear how they are selected as compared to conventional CD4+ T cells. Papiernik et al. [86] were the first to show that CD25+Treg originate in the thymus as flourochrome labelling showed that CD4+CD25+ T cells emigrate from the thymus to populate the periphery. Since thymic CD25+Treg are identified in the single positive T-cell pool in the thymus, it has been hypothesized that they might be educated during an altered negative selection. This was tested in a study in which TCR-transgenic mice expressing a receptor specific for influenza-virus haemagglutinin (HA) were crossed with transgenic mice that expressed the antigen [87]. Strikingly, the percentage of CD25+Treg in thymus and peripheral lymph nodes rose dramatically and was found to constitute 30 and 50% of CD4+ cells, respectively. Further, T cells equipped with a low-affinity TCR for the specific peptide did not develop into CD25+Treg. The selection of CD25+Treg was shown to be dependent on radio-resistant elements of the thymus such as the thymic epithelium and not bone marrow-derived DCs [87]. The conclusion from the study was that CD25+Treg are selected on high-affinity TCR interactions with target antigen, which possibly is expressed by medullary epithelial cells (mTECs). Interestingly, mTECs have been shown to express a variety of organ-specific proteins, such as myelin oligodendrocyte glycoprotein (MOG), glutamic acid decarboxylase 65 (GAD65) and acetylcholine receptor (AchR) [88]. This promiscuous gene expression is partly under the control of a putative transcription factor called AIRE [89]. Patients with mutation in AIRE leading to a loss of function develop multiorgan autoimmune diseases (APECED, autoimmune polyendocrinopathy). Expression of peripheral organ-specific proteins in the thymus might be used for elimination of self-reactive T cells or alternatively for selection of CD25+Treg. However, aire-deficient mice have normal numbers of CD25+Treg in their lymphoid organs [89] and display suppressive function both in vitro and in vivo[90]. These results indicate that mTECs are not necessary for development of CD25+Treg in the thymus. Indeed, other studies have shown that commitment to the CD25+Treg lineage occurs already in the thymic cortex. K14-transgenic mice, which exclusively express MHC-II on the cortical epithelium and not on mTECs or bone marrow-derived APC, have normal CD25+Treg. This shows that cortical epithelium is sufficient for differentiation of CD25+Treg [91].

Collectively, both the study by Jordan et al. and the study by Bensinger et al. suggest that, at least in a TCR-transgenic system, CD25+Treg require stronger interaction of their TCR with self-peptide/MHC-II than what is needed for positive selection of other T cells but lower than the threshold for negative selection. Recently, van Santen et al. investigated if such a window of avidity exists by using a model in which the expression of the TCR-specific peptide was driven by a promotor that is inhibited in a dose-dependant way in the presence of tetracycline. They found that the antigen dose that gave a relative increase of CD25+ Treg simultaneously resulted in the deletion of CD25 T cells. In fact, the actual number of CD25+ T cells was stable when the antigen expression was increased and no conversion of CD25 T cells to a CD25+Treg phenotype took place. Rather, it appears that CD25+Treg are more resistant to clonal deletion than the CD25 T cells [92]. Additional studies are needed to clearly determine how, when and where T cells commit to the CD25+Treg lineage and in what way this is linked to Foxp3 expression.

Antigen specificity of CD25+Treg

In vitro CD25+Treg require activation via their TCR to suppress and one of their characteristics is their anergic state. The anergy can be broken by addition of rhIL-2 or anti-CD28 Ab but once these agents are removed the CD25+Treg retain their anergic phenotype [18, 41, 93]. This is a distinctive feature of CD25+Treg since other T cells, which have become anergic in vitro by antigenic stimulation without costimulation, may have suppressive capacity but these cells will not spontaneously resume to an inhibitory state once anergy has been broken [25]. This unique ability has raised hope for the potential use of expanded CD25+Treg as a clinical treatment, but it is also highly feasible that expansion of CD25+Treg is a physiological event in vivo. Several studies have generated data, which imply that CD25+Treg are organ-specific and that they need to interact with organ-specific antigens and possibly expand to remain in the regulatory T-cell pool. Experiments in the d3Tx model have revealed that spleen cells from donors of the same sex are better at preventing orchitis and oophoritis than spleen cells from ovariectomized mice or female mice in the case of oophoritis or orchitis, respectively [94, 95]. In addition, CD25+Treg from rats ablated of their thyroid are unable to protect from thyroiditis, while the capacity to protect against other autoimmune diseases remained [96]. How is this peripheral pool of putatively antigen-specific CD25+Treg maintained? In contrast to the lack of proliferation in vitro, CD25+Treg actively proliferate to antigen stimulation in vivo in nonlymphopenic normal hosts [97–99]. Notably, CD62LhighCD25+Treg from HA-TCR transgenic mice were found to proliferate extensively in the pancreatic lymph node (LN) but not in other LN when transferred into ins-HA-transgenic mice, which express the model antigen HA in the pancreatic islets [97]. This suggests a continuous activation and proliferation of autoreactive CD25+Treg in the LN draining the target organ. Indeed, CD25+Treg from the pancreatic LN but not from other LN efficiently protect from autoimmune diabetes [100].

Both murine and human CD25+Treg have as diverse TCR repertoires as CD4+CD25 T cells as judged by the expression of various TCR α/β gene segments [41, 101, 102]. This suggests that CD25+Treg are capable of responding to a wide spectrum of antigens. Whether they are biased towards responding to self-antigens or are as broad in their repertoire as CD25 T cells is not known. A very recent study on murine CD25+Treg indicates that the TCR repertoire of CD25+Treg and CD25 effector T cells, although being similarly diverse, recognize only partly overlapping antigens. Furthermore, particular CD25+Treg have substantially higher avidity for MHC class II-bound peptides from peripheral self than the CD25 T cells [103]. Additionally, a few studies of human CD25+Treg have shown that they suppress proliferation and cytokine production to both self-antigens and foreign antigens, including MOG, hHSP60, Helicobacter pylori antigens, beta-lactoglobulin (beta-LG) and pollen extract in vitro[21, 43, 102, 104–106]. This exemplifies the diversity of the CD25+Treg pool in unmanipulated individuals. Indeed, if the primary function of CD25+Treg is to prevent the activation of potentially hazardous T cells then it would be an advantage to have a similarly composed TCR repertoire as the potentially reactive CD25 T cells.

Autoreactive T cells can be identified in most healthy individuals, but still very few succumb to autoimmune diseases. We have shown that healthy human adults have CD25+Treg that are able to suppress the proliferation and the IFN-γ production of CD4+CD25 T cells, which react to MOG, a neural autoantigen involved in the pathogenesis of multiple sclerosis [21]. Interestingly, in order to detect a similar suppressive effect from CD25+Treg derived from cord blood, the number of cells used for the suppression assay had to be fourfold higher [21, 22]. This suggests that an expansion of MOG-reactive Treg takes place in the periphery after birth. Overall, when considering data obtained in vitro and in vivo from humans and mice, it is conceivable that small numbers of antigen-specific CD25+Treg are selected on self-peptide in the thymus, which then seeds the periphery where naïve CD25+Treg will reside in the LN in a resting condition. In the steady state in the draining LN, patrolling APC will present various self and noninflammatory foreign peptides depending on location and thereby induce activation, proliferation and possibly also migration of specific CD25+Treg. This would explain why the removal of an organ results in the elimination of the CD25+Treg-mediated protection of that organ and why CD25+Treg derived from the pancreatic LN are better suppressors of diabetes in NOD mice than CD25+Treg derived from other LN.

CD25+-regulatory T cells in human disease

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References

CD25+Treg seem to play a fundamental role for the balance of more or less all immune responses (Fig. 4). A central question is whether the onset of, for example, autoimmunity or allergy reflects developmental or functional deficiencies in CD25+Treg, which will favour excessive activation and development of disease. Such a deficiency might not simply be seen as reduced numbers of CD25+Treg but could also depend on other factors, such as gaps in the TCR immune repertoire of the CD25+Treg or genetic polymorphisms leading to defective activation, function, survival or homing properties of the regulatory cells. Further investigations are needed to understand the mechanisms involved in the onset of autoimmunity and allergy and how CD25+Treg function may be enhanced in susceptible individuals.

image

Figure 4. Summary of immune responses that are inhibited by CD25+Treg. The actions of CD25+Treg have beneficial effects on several immune responses including autoimmunity, allergy and transplantation tolerance. However, responses to bacterial and tumour antigens are also suppressed by CD25+Treg and by blocking the effects by CD25+Treg, it is possible to enhance the immune response to these antigens.

Download figure to PowerPoint

Autoimmunity

The existence of CD25+Treg was first elucidated using murine models of autoimmunity and thus there is an abundance of reports that show how crucial CD25+Treg are for protection against development of autoimmunity in mice [54]. Ever since human CD25+Treg were identified, a key issue has been to assess whether they play a role in the development of autoimmune diseases in humans. There are two potential problems associated with the study of CD25+Treg in humans. First, since CD25 currently is the most useful marker for identification and isolation, it is more or less impossible to avoid contamination of effector cells, particularly in patients. Second, the composition of cells in the blood stream might not be comparable to the conditions at the site of inflammation. Several recent studies have reported a decrease in numbers of CD25+Treg in peripheral blood of patients suffering from systemic lupus erythematosus [107], type 1 diabetes [108] and rheumatoid arthritis (RA) [109, 110], while others found them to be unaltered in RA [111]. On the other hand, synovial fluid from patients with ongoing rheumatic disease was found to contain increased numbers of CD25highTreg as compared to the levels in peripheral blood [109, 110, 112]. These studies found that the CD25highTreg were functional in vitro and suppressed both proliferation and production of cytokines, including IFN-γ and TNF-α. This might suggest that the decreased levels of CD25+Treg in peripheral blood found in some studies reflect a shift of CD25+Treg from circulation to the site of inflammation. In contrast to the studies mentioned above, Ehrenstein et al. [111] showed that while CD25+Treg from RA patients could suppress proliferation they did not suppress IFN-γ and TNF-α and they were not able to convey a suppressive phenotype to effector CD4+CD25 T cells by the so-called infectious tolerance. Interestingly, this ability was restored in patients treated with anti-TNF-α and was accompanied by increased numbers of CD25+Treg in peripheral blood. Since anti-TNF-α treatment is shown to reduce the levels of proinflammatory cytokines, it is possible that this creates an environment that is beneficial for the CD25+Treg.

A recent study assessed the function of CD25+Treg in patients suffering from multiple sclerosis [11]. They found no difference in the frequency of CD25highTreg in the peripheral blood of patients compared to healthy controls, but detected functional differences with regard to suppressive ability in response to stimulation with anti-CD3 Ab. Furthermore, CD25highTreg from patients weakly suppressed the proliferation of CD25 T cells derived from either patients or healthy controls, while CD25highTreg from the controls suppressed both effector T-cell populations. These results suggest that the decrease in suppressive ability was due to a defect in the CD25highTreg rather than the possibility that responder CD25 T cells were refractory to suppression. This is in contrast to results obtained from studies of RA where synovial CD25+Treg were found to be better suppressors of synovial CD25 T cells than CD25+Treg from peripheral blood [110, 113]. This indicates that the effector cells in the synovium are harder to suppress, probably due to the ongoing inflammation, and that the CD25+Treg are adjusting accordingly. Conclusively, since the function of human CD25+Treg has been studied during ongoing disease, it is difficult to decipher whether the onset of disease actually depended on flaws in the CD25+Treg population or whether the detected deficient regulatory function is a result of the disease. Further research is needed in order to understand the complex mechanisms of genetic background and environmental factors that control the development of autoimmunity. Finally, a detailed understanding on how CD25+Treg activation is regulated might reveal novel strategies in the treatment of autoimmune diseases.

Allergy

Patients with IPEX, which are deficient of CD25+Treg, commonly suffer from organ-specific autoimmune diseases and in addition they develop severe dermatitis, high levels of serum IgE and sometimes eosinophilia. This suggests that CD25+Treg play an important part in the development of tolerance to allergens. We investigated whether the functions of CD25+Treg from birch allergic patients were comparable to those derived from healthy controls and whether their function was influenced by an ongoing allergic reaction [106]. We did not detect any differences in the numbers of either CD25highTreg or the total amount of CD4+CD25+ T cells in peripheral blood of birch allergic patients compared to healthy controls neither out of season nor during birch pollen season. Furthermore, both groups potently suppressed T-cell proliferation and IFN-γ production in response to birch allergen irrespectively of whether it was out of or during birch pollen season. The allergic immune response is characterized by the production of Th2 cytokines IL-4, IL-5 and IL-13, which leads to secretion of allergen-specific IgE and recruitment of eosinophils [114]. The ability of CD25+Treg to suppress production of Th2 cytokines would therefore be particularly desired. Interestingly, during season, CD25+Treg from birch allergics, but not from nonallergic controls, were unable to suppress birch pollen-induced IL-13 and IL-5 production, a difference that was not observed in cultures with soluble anti-CD3 Ab stimulation. In contrast, out of season, both allergics and nonallergics suppressed IL-13 and allergics also inhibited IL-5 production from CD25 T cells. Conclusively, this does not suggest a general deficiency of the CD25+Treg in birch allergic patients. However, in season, during an ongoing allergic reaction with already activated effector T cells present, CD25+Treg from birch allergics are less able to downregulate Th2 cytokines. It has been shown that increased levels of IL-4 obstruct the suppressive ability of CD25+Treg to suppress Th2 clones [115]. Since allergic individuals produce IL-4 when exposed to allergen, this might be one explanation for the deficient regulation of Th2 cytokines during birch pollen season. Alternatively, strongly activated birch-specific T cells might be more resistant to suppression [113]. In accordance with our results, others have shown that CD25+Treg derived from peripheral blood of patients allergic to cow milk or grass or birch pollen suppressed allergen-induced proliferation [104, 116]. Variable effects of CD25+Treg have been reported with regard to the suppression of Th2 cytokines in allergy. In support of our study, Bellinghausen et al. [116] found that CD25+Treg from a subpopulation of their allergic subjects, which produced higher levels of IL-4, were unable to suppress the production of IL-5 from CD25 T cells. Ling et al. [105] reported that patients suffering from hayfever were less capable of suppressing both proliferation and IL-5 production during the grass pollen season as compared to nonatopic individuals and that this defect remained also outside season although to lesser extent. CD25+Treg from nickel-allergic patients did not suppress nickel-specific T-cell proliferation to the same extent as healthy individuals did [117]. However, since the immunological mechanisms involved in nickel and pollen allergy are very different, it is not surprising that different results are generated from these two patient groups.

Tumour immunity

Research in the recent years has revealed that the activity by CD25+Treg is not always beneficial for the host. CD25+Treg have been shown to inhibit immune responses to tumour antigens, thereby promoting tumour growth [118]. Experiments in murine models of cancer have shown that depletion of CD25+Treg results in elimination of tumour cells and that this effect is even more enhanced after blocking CTLA-4 [119–121]. Studies have also been performed on human malignancies. Increased frequency of CD25+Treg with functional suppressive activity has been found in tumours and blood from patients with metastatic melanoma or cancer in pancreas, breast, lung and stomach [122–125]. These findings may have potential implications for the treatment of cancer patients and might explain the poor clinical response of cancer patients undergoing immunotherapy. Depletion of CD25+Treg in combination with tumour immunotherapy or vaccination could have the potential to enhance the chances of eradication of the malignancy. However, this should not be done on the expense of development of autoimmune diseases. It has been demonstrated that mere depletion of CD25+Treg with anti-CD25 Ab from animals does not lead to induction of autoimmune disease. However, in combination with strong TCR stimulation, absence of CD25+Treg may result in autoimmune disease [126], which is important to consider when attempting to inactivate or remove CD25+Treg as a means to enhance antitumour immune responses.

Chronic infection

Infection is another condition where CD25+Treg are found to counteract immunity. CD25+Treg have been shown to interfere with viral as well as bacterial and parasitic infections [127]. Adoptive transfer of CD25+Treg prevents lethal pneumonia in recombinant-activating gene-2-deficient mice infected with Pneumocystis carinii, but at the expense of deficient protective response and microbial clearance [128]. Similarly, CD25+Treg were shown to suppress Th1 responses in mice infected with Helicobacter pylori, thereby limiting the mucosal inflammation but with a higher bacterial load as a result [129]. Furthermore, CD25+Treg prevent sterilizing immunity to Leishmania infection [130] and the persistence of low numbers of microbes proved essential for the development of T-cell memory and prevention of reinfection. This indicates that the prevention of complete eradication of microbes may sometimes be beneficial for the host. Few attempts have been made to study CD25+Treg and infection in humans. However, Lundgren et al. [43] showed that CD25+Treg from carriers of Helicobacter pylori suppressed responses to H. pylori antigens in vitro and that H. pylori-infected individuals have increased frequencies of CD25high T cells in both the stomach and the duodenal mucosa relative to healthy controls [131]. Together, this indicates that CD25+Treg actively inhibit the eradication of the bacteria which contributes to the persistence of infection. In summary, depending on the pathogen in question, it might be necessary to reduce or enhance the activity of regulatory T cells to achieve an appropriate immune response.

Concluding remarks

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References

It is now firmly established that T-cell-mediated regulation of immune responses are essential for maintaining tolerance to both self-antigens and foreign antigens and among other kinds of regulatory T cells, CD25+Treg seem to play a key role. Research in the last several years have resulted in significant advances in the understanding of CD25+Treg and information has emerged regarding their origin, cellular and molecular interactions and their involvement in various conditions of illness. In spite of this, much still remains illusive. What is particularly needed in order to move forward is a good surface marker, or a combination of markers, that specifically denotes the CD25+Treg and separates them from activated cells and memory cells. The need for such markers becomes especially apparent when investigating the role of CD25+Treg in human disease. Another major issue is how CD25+Treg perform their suppressive function. Clearly, suppression can be mediated by both cell-contact-dependent mechanisms and immunosuppressive cytokines. However, particularly in vivo, it is unclear in what way the different mechanisms contribute to the induction of suppression and to what extent the so-called induced regulatory T-cell subsets take part. A detailed understanding of the relationships of various regulatory T-cell subsets might improve the knowledge of the events leading to tolerance in vivo. Eventually, what is desired are specific ways to manipulate either the naturally occurring or induced Tregs, by boosting or dampening their function according to circumstances.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References

Studies in our group have been supported by the Medical Faculty, Göteborg University, the Swedish Research Council, the Vårdal Foundation, Clas Groschinsky's Memory fund, Göteborg Medical Society, the Swedish Asthma and Allergy Association Research Foundation, FRF Stiftelse and Konung Gustav V:s 80 Årsfond.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Phenotype of CD25+-regulatory T cells
  5. Activation of CD25+-regulatory T cells
  6. Mechanism of suppression
  7. Development of CD25+Treg
  8. CD25+-regulatory T cells in human disease
  9. Concluding remarks
  10. Acknowledgments
  11. References