FOXP3+ regulatory T cells: Current controversies and future perspectives



Regulatory T cells (Treg) provide protection from autoimmune disease, graft-versus-host disease, transplant rejection and overwhelming tissue destruction during infections. Conversely, high Treg numbers enable cancer cells to evade the host immune response. Thus, Treg are seen as an important tool to manipulate the immune response. However, as the immunological community is trying to move this knowledge from mice to humans, contradictory results regarding the number and function of Treg in various diseases are appearing. This problem arises because we cannot clearly define Treg populations on the basis of expression of CD25 and other cell surface markers in humans. This review addresses the utility of the FOXP3 forkhead transcription factor for the identification of Treg populations and summarizes recent data on the expression of FOXP3 in lymphomas. It is crucial to really understand Treg biology before attempting therapies, including (i) the injection of expanded Treg to cure autoimmune disease or prevent graft-versus-host disease or (ii) the depletion or inhibition of Treg in cancer therapy. For instance, new data arising from the study of haematological malignancies highlight the additional complexity of systems where malignant cell populations may also be direct Treg targets.


effector T cell


The immune system must be able to distinguish self from non-self yet still respond to attacks by pathogens. While negative selection in the thymus plays a key role in eliminating self-reactive T cells, some autoreactive cells are released into the periphery or generated de novo. The concept of specialized suppressor cells dominantly controlling immune responses and protecting from self reactivity was revived 10 years ago by the finding of Sakaguchi and colleagues 1 that CD4+ T cells expressing IL-2Rα (CD25) could prevent autoimmune diseases. This CD4+CD25+ population emigrates from the thymus 2 and has been called regulatory T cells (Treg). The aims of this mini-review are to discuss (i) if we can identify these Treg in human disease, especially focussing on FOXP3 as a Treg marker and (ii) to address recent studies describing newly identified Treg functions in haematological malignancies.

Immunophenotype of Treg

Although the CD4+CD25+ T cell population is enriched for suppressive activity, this phenotype has proven to be insufficient for the homogenous purification of Treg. Not only do some regulatory cells express CD8 rather than CD4 3, but CD25 has limitations as it is not T cell restricted and cannot be used to distinguish Treg from activated effector T cells (Teff). While the murine CD4+CD25+ population is relatively homogenous and is highly enriched in Treg, in humans CD25+ cells contain both Treg and Teff populations (Fig. 1). To obtain enriched Treg with little Teff contamination, it is necessary to gate on the CD4+CD25high population that has regulatory activity 4. Since it is conceivable that in acute inflammatory diseases increased numbers of CD25+ Teff are present in the blood, it is very difficult to isolate pure Treg. Thus, a significant problem in comparing clinical studies in humans is the lack of a defined cut-off for CD25high expression, resulting in experimental Treg populations containing variable proportions of Teff. This in turn leads to highly variable numbers of Treg being reported, especially in patients with autoimmune disease. Furthermore, some studies conclude that Treg are defective in their suppressive activity, which may in fact simply reflect their contamination with Teff 5.

Figure 1.

Intensity of CD25 expression on Treg (TR) of naïve or memory phenotype. PBMC (A and B) and cord blood lymphocytes (C) were stained with mAb against CD4, CD25, FOXP3, CD45RO, CD45RA and CD95 for six-colour FACS analysis. Dot plots for CD25 versus CD4 expression are depicted. To show the differential CD25 expression level FOXP3+ subpopulations have been coloured black. In (A) FOXP3+, CD45RAhigh, CD45ROlow, CD95low naïve-like Treg from adult blood, in (B) FOXP3+, CD45RAlow, CD45ROhigh, CD95high memory-like Treg from adult blood and in (C) FOXP3+ Treg from cord blood are marked in black. The latter overwhelmingly have the CD45RAhigh, CD45ROlow, CD95low naïve phenotype. It can be appreciated from (A) that naïve-like Treg express intermediate levels of CD25 that overlap mostly with CD25+ Teff (grey population). A significant portion of memory-like Treg (B) are CD25high and these cells can be sorted without Teff contamination. Cord blood Treg are well separated from the other T cells as hardly any CD25+ Teff are present. Figure courtesy of B. Fritzsching and E. Suri-Payer. Local Ethical Committee Approval and informed patient consent was obtained for these studies.

A number of other cell surface Treg markers exist, including GITR, CTLA-4, CD45RB, CD62L, Nrp1, CD103 and LAG-3. However, these have not enabled homogenous Treg purification as most of them are also up-regulated during T cell activation and are thus present on Teff. A negative selection step to remove PD-1+-activated Teff and Treg from PD-1 resting Treg has recently been proposed as a method for isolating a subset of more highly purified Treg for therapeutic applications 6.

FOXP3 represents a single definitive Treg marker

A significant advance in more precisely defining the Treg population occurred when the forkhead transcription factor FOXP3 was identified as being necessary for Treg development. Mutations in the Foxp3 (murine)/ FOXP3 (human) gene were identified as the sole genetic defect underlying the phenotypes of the fatal autoimmune diseases of the scurfy mouse 7 and human X-linked neonatal diabetes mellitus, enteropathy and endocrinopathy (IPEX) 8 and X-linked autoimmune and allergic dysregulation syndromes (XLAAD) 9. These autoimmune diseases arise from a lack of functional Treg and it was subsequently demonstrated that Foxp3 gene transfer converted naïve CD4+CD25 T cells into a regulatory population 1012.

Although transgenic mice exhibited Foxp3 expression in both B and T cells 12, subsequent studies using a Foxp3GFP knockin mouse 13 and immunophenotyping studies of single FOXP3+ human cells 14 did not detect B cell FOXP3 expression. Furthermore, specific silencing of Foxp3 expression in the T cell lineage was sufficient to induce a murine phenotype indistinguishable from that of the scurfy mouse 13. In both mice and humans, FOXP3 expression is also detectable in a small population of CD8+ T cells and a small subpopulation of CD25 cells 13, 14. Significantly, murine CD25–/lowFoxp3GFP+ cells were shown to have a similar gene expression profile and equivalent potency to the CD25highFoxp3GFP+ population in in vitro suppression assays, indicating that Foxp3 expression rather than CD25 is critical for the regulatory phenotype 13.

It should be emphasised that in humans FOXP3+ cells can also be either CD25low or CD25high (Fig. 1). Due to differential susceptibility to CD95-induced apoptosis, we found that Treg from adult individuals can be divided into two subpopulations. One population can be isolated from adult or cord blood and is characterized by a CD45RAhigh, CD45ROlow, CD95low phenotype characteristic for naïve T cells. We therefore refer to these cells as naïve-like Treg. They express CD25 levels in the range of Teff and are resistant to apoptosis. The other population resembles memory cells because of their CD45RAlow, CD45ROhigh, CD95high phenotype. Most of these memory-like Treg cells express very high levels of CD25 and they are sensitive to CD95-mediated apoptosis 15. Thus, on the basis of CD25 expression, naïve-like Treg cannot be separated from Teff cells. This has only been possible when analysing the total CD4 cell population by multicolour FACS of permeabilised cells including CD25 and FOXP3. However, sorting of pure Treg for functional analysis has still relied on CD25high expression and, therefore, only a fraction of the total human Treg population has been analysed.

Recent studies have demonstrated that down-regulation of cell surface IL-7 receptor (CD127) expression can be used to distinguish Treg from activated T cells 16, 17. The CD4+CD25+CD127low/– population contained approximately 80% of the FOXP3+ cells and was significantly larger than the CD4+CD25high population 16. CD127low/– expression, in combination with CD4+CD25+, thus enables the sorting of a more highly purified Treg population incorporating the FOXP3+CD4+CD25int population without significant Teff contamination. Overall the available data support the designation of FOXP3 as a lineage marker for Treg that identifies a broader regulatory population than that defined by CD4+CD25+ or CD4+CD25+CD127low/– expression alone. The recent improvement in Treg purification through reduced CD127 expression should facilitate our understanding of Treg function, Treg numbers and Treg defects in human disease.

In humans, FOXP3 is alternatively spliced to generate an isoform lacking exon2 that is still capable of repressing transcription from the native human IL-2 promoter 18. However, the expression of this isoform in Treg populations may be functionally different from the long isoform as exon2-deleted FOXP3-transduced T cells were less hyporesponsive and produced more IL-2 in vitro18. In addition, exon2 mediates interaction with the orphan retinoic acid nuclear receptor 19. Although Treg populations express both isoforms, it has not been investigated whether single cells or subpopulations in specific diseases express only a single isoform. Thus when selecting antibodies to detect human FOXP3 protein, it may prove critical to know whether, like our 150D/E4 FOXP3 mAb (B. Fox and A. H. Banham, unpublished data), they bind exon2. Nevertheless, more research is needed to clarify if the two different isoforms indeed possess different functions.

FOXP3 induction by activation

Controversy over the value of FOXP3 as a Treg marker has arisen from conflicting data derived from in vitro experiments on its activation-induced expression in CD4+CD25 cells. On activation by T cell receptor agonists, murine CD4+CD25 T cells acquire surface CD25 but do not express Foxp3 or acquire regulatory function 10, 12. In contrast, several studies have reported that FOXP3 mRNA is expressed by in vitro human activated non-regulatory T cells 14, 2022, which may gain regulatory activity 20, 22. Recent studies on the FOXP3 promoter identify a non-repressive chromatin structure that together with a specific transcriptional regulation programme may explain the ability of activation to induce FOXP3 expression in CD4+CD25 T cells 23. Gavin and colleagues 24 have recently reported that activation-induced FOXP3 expression in human T cells, as measured by flow cytometry, is transient and insufficient to induce Treg, proposing that sustained high level FOXP3 expression is required for Treg phenotype and function. Activation of CD4+CD25 T cells in the presence of TGF-β (reviewed in 25), as well as various in vivo regimes that target protein antigens to immature dendritic cells 26, have shown that CD4+CD25 T cells can be converted into Treg cells that stably express high levels of Foxp3/FOXP3. These data suggest that T cells stably expressing FOXP3 are Treg, independent of their origin. Thus, the high numbers of FOXP3+ cells that have been detected by immunohistochemistry in inflamed skin and in cancer tissue most likely represent Treg 27, 28.

FOXP3 expression in tumour cells

mAb (such as 236A/E7) that enable detection of the FOXP3 protein at the single cell level in routinely fixed tissue sections 14 have been used to demonstrate that FOXP3 expression in malignant lymphoma cells is restricted to a poor prognosis subpopulation of patients with adult T cell leukaemia/lymphoma (ATLL) 29. ATLL has a CD4+CD25+ phenotype and strong immunosuppressive function, yet it is unclear whether the causal infection by human T lymphotropic virus type 1 (HTLV-1) targets Treg or induces the regulatory phenotype. Support for the latter hypothesis is provided by recent data demonstrating that HTLV-1 infection of CD4+CD25 T cells from healthy donors induces a Treg phenotype by up-regulating CD25 protein, FOXP3 mRNA expression and TGF-β secretion in CD4+ cells 30. While there have been reports of tumoural FOXP3 expression in an in vitro model of cutaneous T cell lymphoma 31, our data from primary lymphomas do not corroborate these findings, but indicate the presence of a high percentage of “natural” Treg in the skin of mycosis fungoides patients 27, 29.

Manipulation of Treg for cancer therapy

Increased Treg numbers have been reported in the blood, ascites or tumour tissue of patients with epithelial cancers and Curiel et al. 28, have shown that increased Treg numbers correlate with worse prognosis. For an in-depth analysis of Treg function in tumour immunity and immunotherapy please refer to a recent review 32. Briefly, studies in the murine system indicate that depletion of Treg leads to a better anti-tumour immune response. Nevertheless, there are only very few reports on tumour regression after Treg depletion in mice 32. Two particular problems arising from Treg depletion using anti-CD25 mAb are (i) an accelerated return of Treg through conversion of Teff to Treg 33 and (ii) the co-depletion of anti-tumour Teff 34.

Before attempting Treg depletion for the treatment of human cancer, one should also distinguish epithelial cancers from haematological malignancies. In patients with B cell-derived classical Hodgkin's lymphomas, high numbers of FOXP3+ cells and low numbers of TIA-1+ cells were correlated with improved survival and FOXP3+ T cell numbers were reduced on patient relapse 35. Further, we have recently demonstrated that high FOXP3+ Treg numbers predict improved survival of patients with follicular lymphoma, while a marked reduction is found on their transformation to a more aggressive diffuse large B cell lymphoma 36. Decreased number and function of FOXP3+ Treg have also been described in the plasma cell-derived malignancy multiple myeloma 37 and in Sezary syndrome, the aggressive leukaemic variant of cutaneous T cell lymphoma 27, 38. Since Treg can suppress B as well as T cell function 39, 40, the expansion of such tumours may be slowed by Treg. Therefore, immunotherapeutic strategies aiming to reduce Treg numbers, to improve the efficacy of cancer vaccines in haematological malignancies, may need to carefully balance advantages gained by improving T cell responses to tumour antigens with possible disadvantages arising from removing direct Treg suppression of the malignant population.

Another aspect of therapy involving Treg is to increase their numbers to combat autoimmunity or to prevent graft-versus-host disease. For the advances in expansion of auto- and alloantigen-specific Treg for future therapy please refer to recent reviews 41, 42.


The pivotal role of FOXP3 in Treg biology and its use as a Treg marker has greatly facilitated immunological research. Many of the contradictory results from in vivoversusin vitro experiments are likely to reflect the interplay between Treg and other cell populations and cytokines in the tumour microenvironment or during inflammation. The search for a specific cell surface marker for FOXP3+ cells, against which to raise mAb to facilitate both Treg purification and depletion, is still an active and important area of interest. Such reagents are needed to fully understand the number and function or inactivation of Treg in cancer and autoimmunity, respectively. The discovery of new cell surface molecules on Treg may also help in defining the as-yet-elusive mechanism of suppression. Finally, special care should be taken to further investigate the potentially opposing roles of Treg in patients with haematological malignancies before manipulating their numbers.


A.H.B. is supported by the Leukemia Research Fund. Many thanks to Karen Pulford, Philip Brown, Bridget Fox, Gaynor Bates and Benedikt Fritzsching for helpful discussions and/or critical reading of the manuscript. We also thank B. Fritzsching for providing Figure 1.


  1. 1