Drs Vandenbark and Dr Offner have a significant financial interest in Orchestra Therapeutics, a company that has a commercial interest in the results of this research and technology.
Dr Arthur A. Vandenbark, Neuroimmunology Research, R&D-31, Veterans Affairs Medical Center, 3710 SW U.S. Veterans Hospital Road, Building 101, Room 408, Portland, OR 97239, USA. Email: firstname.lastname@example.org Senior author: Halina Offner, email: email@example.com
The identification of CD4+ CD25+ Foxp3+ regulatory T (Treg) cells as natural regulators of immunity in the periphery and tissues has stimulated tremendous interest in developing therapeutic strategies for autoimmune diseases. In this review, the site of origin, antigen specificity, homing markers and cytokine profiles of Treg cells were evaluated in autoimmune colitis and type 1 diabetes, two examples in which Treg cells were effective as therapy. These studies were compared with studies of Treg cells in experimental autoimmune encephalomyelitis and multiple sclerosis, where successful therapy has not yet been achieved. Antigen-specific Treg cells appear to have more potent activity than polyclonal Treg cells and therefore hold more promise as therapeutic agents. However, Treg cells specific for the pathogenic T effector cells themselves have largely been overlooked and deserve consideration in future studies.
CD4+ CD25+ Foxp3+ regulatory T (Treg) cells are now considered to be ‘master regulators’ of the immune system.1 In normal mice, Treg cells limit inflammation and inhibit a number of autoimmune diseases.2–5 Initial descriptions indicated that Treg cells expressed very high levels of the interleukin-2 (IL-2) receptor, CD25, and consequently these cells are often referred to as CD4+ CD25+ or CD4+ CD25bright. The forkhead/winged helix transcription factor gene, Foxp3, is strongly linked to the regulatory function of CD4+ CD25+ Treg cells,6–8 and has become a useful intracellular marker for their identification. Although a normal complement of Treg cells specific for self tissue determinants may maintain self-tolerance,9 it is now appreciated that an overabundance of Treg cells may impede immunosurveillance against autologous tumour cells10 and may suppress the ability of CD4+ CD25− effector T cells to eliminate parasites.11 Taken together, these findings document the importance of the CD4+ CD25+ Treg cell subpopulation in regulating autoreactive as well as protective T effector (Teff) cells in vivo.
The fundamental regulatory activity that can be consistently demonstrated by Treg cells in vivo and in vitro has stimulated great interest in developing novel strategies for treating ongoing inflammatory conditions. There are many excellent review articles available that speak directly to possible clinical applications of Treg cells,12–19 and it is not the purpose here to reiterate this discussion. Rather, this review will focus on papers that present data relevant to the issue of the ability of Treg cells to reverse an ongoing inflammatory response. The source, antigen specificity, manipulations and homing of the Treg cells are all likely to be critical elements that will determine their therapeutic potential. Initially, two key examples that demonstrate the therapeutic activity of Treg cells will be reviewed, followed by discussion of a number of papers directed at the function of Treg cells in preventing and possibly treating central nervous system (CNS) inflammation. Finally, the key points from these papers will be summarized and new directions will be outlined for future studies.
Two success stories using Treg cells to treat ongoing autoimmune diseases
Against the backdrop of literally thousands of papers devoted to studying the characteristics, basic functional properties, and possible mechanisms associated with CD4+ CD25+ Foxp3+ Treg cells, there are relatively few that speak to the possible effects of Treg cells as therapy for established autoimmune disease processes. The following examples describe two notable successes in this regard.
Treg cells can reverse established colitis and cure intestinal inflammation
In a landmark paper, Mottet et al.20 provided the first evidence that Treg cells have therapeutic potential. In the murine colitis model, clinical disease can be induced in immunodeficient [severe combined immunodeficient (SCID) or recombinase activation gene knockout (RAG KO)] mice 3–4 weeks after transfer of naïve CD4+ CD45RBhigh cells. Disease induction is characterized by piloerection, hunching, anal inflammation, diarrhoea and weight loss. In this study, these colitic mice received two injections of 106 CD4+ CD25+ or CD4+ CD25− CD45RBlow splenic T cells 4 weeks apart. Mice receiving CD4+ CD25+ T cells (of undefined specificity) started to recover within 2 weeks of transfer, with gradual disappearance of clinical and histological signs, restored colonic architecture, including reappearance of goblet cells, and enhanced survival rate (91%) at 14 weeks. In contrast, mice injected with CD4+ CD25− T cells or control mice without a second cell transfer continued to lose weight, did not show clinical improvement, had a much lower survival rate at 14 weeks (22%) and had worsening inflammation in the lamina propria, depletion of goblet cells, and epithelial cell hyperplasia and ulceration. The CD4+ CD25+ T cells were found to proliferate in the mesenteric lymph nodes and within the inflamed colon, where they maintained contact with both CD11c+ dendritic cells (DC) and pathogenic T cells.
Uhlig et al.21 further characterized CD4+ CD25+ Foxp3+ T cells in wild-type mice and in mice cured from experimental colitis, and in both cases, the Treg cells accumulated in the colon and secondary lymphoid organs. Successful treatment of colitic mice with Treg cells could be reversed upon treatment of the mice with anti-IL-10 monoclonal antibody (mAb), but the IL-10 appeared to be derived from both Treg and non-Treg sources. Of importance, Treg cells secreting IL-10 could be found within the colon not only in mice cured of colitis but also under steady-state conditions. Immunohistological studies in human subjects with inflammatory bowel disease indicated the presence of IL-10+ Treg cells in the colon mucosa and intestinal lymphoid follicles, demonstrating that the observations in mice have relevance to human intestinal diseases.
Antigen-specific Foxp3+ Treg cells can be induced in the periphery through the use of transforming growth factor-β (TGF-β). Coombes et al.22 demonstrated that oral administration of ovalbumin (OVA) in T-cell receptor transgenic (TCR Tg) mice specific for OVA on the SCID background (where new T-cell specificities could not be induced) could convert naïve OVA-specific T cells in mesenteric lymph nodes to become Foxp3+, with smaller numbers also present in the spleen and lamina propria. Moreover, in the presence of exogenous TGF-β, CD103+ DC from the mesenteric lymph nodes promoted this conversion to Treg in vitro much more efficiently than CD103− DC. The CD103+ DC induce expression of gut homing receptors (including integrin α4/β7) on T cells through release of the vitamin A metabolite, retinoic acid, which is converted by a retinal dehydrogenase, aldh1a2, that is expressed by CD103+ DC but not CD103− DC. A key finding was that retinoic acid receptor inhibitors blocked the spontaneous induction of Foxp3 on naïve T cells by CD103+ DC. The ability of CD103+ DC to metabolize vitamin A is therefore likely to account for their ability to drive the spontaneous conversion of naïve T cells into Foxp3+ T cells in the absence of TGF-β or other exogenous factors, as well as to promote the expression of gut homing receptors.
Taken together, these findings support the contention that the gut constitutes a unique environment that can promote the induction of antigen-specific Treg cells through interactions with CD103+ DC and retinoic acid, and successful treatment of colitis may involve both transferred natural IL-10-secreting Treg cells and locally converted antigen-specific induced Treg cells, as discussed in a recent review of this subject by Izcue and Powrie.19
Tang and colleagues23 published another landmark paper using in vitro expanded antigen-specific Treg cells to prevent and treat type 1 diabetes. One large hurdle for using Treg cells in therapy is cell numbers. The solution to this problem given by Tang et al. was to use fluorescence-activated cell sorting CD25+ CD62 ligand-positive (CD62L+) cells from non-obese diabetic (NOD) or TCR Tg mice specific for islet antigens, and to expand these cells in vitro using anti-CD3/CD28 mAb plus high concentrations of IL-2, resulting in approximately 200-fold increases in Treg cell numbers. Sorting and expanding such populations from TCR Tg mice, including BDC2.5 mice specific for an islet antigen, resulted in highly enriched antigen-specific Treg cells that expressed typical Treg markers and secreted TGF-β and IL-10, but not IL-2 or interferon-γ (IFN-γ). From this paper, it was not clear that all of the Treg cells expanded from the TCR Tg mice were specific for the islet antigen, because the initial cell mixture probably included Treg cells with non-transgenic TCR. The expanded Treg cells inhibited CD4+ responder cells in vitro and proliferating Treg cells could be detected in recipient mice in spleen and lymph nodes 7 days after transfer. Of particular interest were the Tregs from the BDC2.5 TCR Tg mice that selectively underwent three or four rounds of cell division in the pancreatic lymph nodes, most likely in response to DC presentation of islet antigens, and down-regulated expression of CD62L. Moreover, cotransfer of the islet-specific BDC2.5 Treg cells with diabetogenic Teff cells from NOD mice prevented the onset of diabetes, even at cell ratios as low as 1 : 9 Treg : Teff cells. This was highly potent suppression compared to polyclonal Treg cells expanded from NOD mice that could at best delay the onset of diabetes. Suppression could be observed both in lymphopenic and non-lymphopenic mice, indicating that the antigen-expanded Treg cells functioned in vivo in the presence of a fully developed pathogenic T-cell response. The most impressive finding of this paper was the ability of the antigen-expanded Treg cells to reverse ongoing diabetes. In one set of experiments, the Treg cells were cotransferred along with a NOD islet cell transplant into recipient mice with total endogenous islet cell destruction. These recipient mice reverted to normoglycemia within 24–48 hr, whereas recipients of NOD Treg cells did not revert. In a second setting, transferred BDC2.5 Treg cells reversed diabetes in 60% of new-onset diabetic NOD mice.
Taken together, these results demonstrated that expanded and passively transferred Treg cells specific for islet antigens were more potent at preventing and reversing type 1 diabetes than polyclonal Treg cells that were not selected for antigen specificity.
Treg cells in experimental autoimmune encephalomyelitis and multiple sclerosis
Experimental autoimmune encephalomyelitis (EAE) is an inflammatory disease of the CNS induced by injection of defined myelin antigens in complete Freund’s adjuvant (CFA) or by transfer of activated myelin-reactive CD4+ Teff cells. EAE is commonly used as a model for multiple sclerosis (MS) because of certain similarities in disease course, the involvement of encephalitogenic myelin-specific T cells, and the resulting pathology that includes perivenular lesions in the CNS accompanied by extensive demyelination and axonal damage. Although many variations of EAE are now available, the three most popular models include C57BL/6 mice (I-Ab) that develop monophasic or chronic EAE after injection of myelin oligodendrocyte glycoprotein (MOG)-35-55 peptide/CFA/Ptx; B10.PL mice (I-Au) that develop monophasic or chronic EAE after injection of myelin basic protein (MBP)-Ac1-11 peptide/CFA/Ptx; and SJL/J mice (I-As) that develop relapsing EAE after injection of proteolipid protein (PLP)-139-151 peptide/CFA. Additionally, TCR Tg mice have been developed in which the great majority (> 95%) of CD4+ T cells are specific for MOG-35-55, MBP-Ac1-11, MBP-85-99 or PLP-139-151 peptides. These TCR Tg mice are particularly valuable because they may develop spontaneous EAE after being bred onto a RAG KO background, which allows evaluation of pathogenic and regulatory mechanisms in the absence of the strong proinflammatory environment provided by CFA.
Treg cells can prevent spontaneous (Sp)-EAE
Although the literature regarding Treg activity in EAE is still relatively sparse, the work that has been reported provides strong evidence that Treg cells do impact the course of disease. In an early elegant series of papers, Juan Lafaille’s group demonstrated that CD4+ T cells from normal mice could protect against Sp-EAE occurring in MBP TCR Tg mice bred onto the RAG-1 KO background (T/R− mice).24,25 Although these studies preceded the discovery of Foxp3 as the lineage marker for Treg cells, the transferred CD4+ T cells undoubtedly contained Treg cells as well as other types of regulatory T cells.26
With the realization that CD4+ CD25+ T cells provided systemic tolerance against autoantigens and protection from autoimmune diseases, their role in regulating Sp-EAE gained renewed interest. Hori et al.,27 using Juan Lafaille’s model system, evaluated the ability of TCR Tg CD4+ CD25+ MBP-Ac1-17-specific T cells to protect T/R− mice from developing Sp-EAE. The MBP-specific CD4+ CD25+ Treg cells, which could be identified using a clonotypic antibody to the specific TCR, could be found only in T/R+, but not T/R− mice, and were present in spleen, lymph nodes and single positive thymocytes. However, the MBP-specific Treg cells constituted only ∼ 40% of the CD4+ CD25+ subset in the periphery and ∼ 80% in the thymus, with the remaining clonotype-negative T cells representing other unknown regulatory specificities. The mixed T-cell population could suppress CD4+ CD25− MBP-specific T-cell responses, and this suppression appeared to be MBP-specific and mediated by Tg CD4+ CD25+ T cells, because deletion of these cells with the clonotypic mAb abrogated the suppression of responses induced by MBP but not anti-CD3 mAb. Importantly, transfer of only 2 × 105 mixed CD4+ CD25+ T cells from T/R+ mice almost completely prevented the onset of Sp-EAE, whereas transfer of non-Tg CD4+ CD25+ cells (MBP clonotype negative) from the T/R+ mice had greatly reduced suppressive capacity. These data clearly demonstrated that MBP-specific Treg cells contributed to suppression, but also implicated some CD4+ CD25+ Treg cells of different specificities in disease protection. In this vein, the development of MBP-specific CD4+ CD25+ cells in T/R+ mice was shown to depend upon the expression of endogenous non-Tg TCR-α chains that may rearrange in RAG-1+ mice even in the presence of the MBP TCR transgenes. Interestingly, the MBP-specific CD4+ CD25+ subset was enriched for cells coexpressing two TCRs: an endogenous TCR-α chain paired with Tg TCR-β (unknown regulatory specificity?), and the complete Tg TCR specific for MBP. For this particular discussion, it is noteworthy that we previously demonstrated in TCR BV8S2 single Tg mice that regulatory T cells specific for self BV8S2 TCR determinants utilized the Tg BV8S2 chain in combination with a rearranged AV2 chain that differed from MBP-Ac1-11-specific T cells by only the CDR3-AJ region.28 Taken together, these findings suggest that the regulatory dual TCR-expressing Treg cells from T/R+ mice recognize both MBP (allowing thymic selection/commitment) and a different self antigen, possibly including TCR determinants, that mediate specific regulatory function.
Again using this same model system, Hori et al. addressed the question of whether the transferred CD4+ CD25+ Treg cells from recombinase-positive wild-type mice could recruit Tg MBP-specific T cells to regulatory function in Sp-EAE-protected T/R− mice.29 The protected T/R− mice developed comparable levels of CD4+ CD25+ T cells as T/R+ mice, but with many fewer cells that were MBP-specific. Moreover, the clonotype-positive Tg MBP-specific CD4+ CD25+ T cells from the protected T/R− recipients were unable to suppress the responses of CD25− MBP-specific T cells to MBP presented by antigen-presenting cells, indicating that these cells had no suppressive function in vitro. Most importantly, only the non-clonotypic CD4+ CD25+ T cells could transfer protection against Sp-EAE in naïve T/R− recipients, whereas MBP-specific CD4+ CD25+ donor cells were not effective. MBP-specific Treg cells were therefore not recruited from Sp-EAE-protected T/R− recipients, and protection was probably the result of other specificities present in the wild-type donor cells. This result was confirmed when protected T/R− mice all developed EAE within 6–30 days after antibody depletion of all donor cells. Taken together, these experiments demonstrated that continuous presence of transferred Treg cells was required to maintain tolerance in the protected T/R− mice.
Our own studies with this model30 investigated more precisely which cell populations in the T/R+ mice could transfer protection against Sp-EAE to T/R− recipients. To enhance diversity and facilitate expansion of possible regulatory phenotypes, we immunized the T/R+ mice with MBP-Ac1-11 peptide/CFA to induce EAE. This step decreased the Tg CD4+ BV8S2+ splenocytes from 44% to 28% and increased non-Tg CD4+ BV8S2− splenocytes from 1% to 6%. Moreover, there were increases in CD4− CD8+ BV8S2+ T cells (5%) and CD4− CD8− BV8S2+ T cells (7%). All four T-cell subpopulations retained reactivity to MBP, although the CD4+ BV8S2− cells had the lowest responses. Transfer of 0·5 million cells of each of these T-cell subpopulations to T/R− mice before the onset of Sp-EAE revealed that CD4+ BV8S2− T cells could completely protect the recipient mice, and CD4− CD8− BV8S2+ and CD4+ BV8S2+ T cells were only partially protective (delayed onset and reduced clinical severity of Sp-EAE). In contrast, CD4− CD8+ BV8S2+ T cells had no protective capability. The highly protective CD4+ BV8S2− subpopulation was CD25+, contained non-clonotypic TCRs, and uniquely expressed the CCR4 chemokine receptor that was previously implicated in trafficking of Treg cells.31 Protected recipient T/R− mice had increased CD4+ CD25+ Treg-like cells, retention of the pathogenic MBP-specific cells in the spleen, and markedly reduced CNS inflammation. The partially protective phenotypes appeared to be mainly clonotypic T cells with altered functional properties. Although some protective activity remained in the clonotypic MBP-specific T-cell population (in agreement with Hori et al.27 discussed above), the major regulatory subpopulation appeared to have the highest content of non-MBP-reactive T cells.
Treg cells can prevent induced EAE and contribute to genetic EAE resistance
The first study that directly addressed the role of CD4+ CD25+ Treg cells in induced EAE was from Adam Kohm in Steve Miller’s group.32 In this report, CD4+ CD25+ Treg cells obtained from lymph nodes of naïve C57BL/6 mice inhibited antigen-induced proliferation and IFN-γ secretion by encephalitogenic MOG-35-55-specific T cells in vitro and reduced the severity of clinical and histological EAE when transferred in vivo before active or passive disease induction. Although the number of lymph node and spleen cells secreting IFN-γ and tumour necrosis factor-α (TNF-α) were unchanged in Treg-protected mice, there was a reduction in IL-2-secreting cells and increases in cells secreting IL-4 and IL-5. Inflammation of the CNS and cellular infiltration were strongly reduced in Treg-protected mice, but the transferred Treg cells could be found only in the lymph nodes but not in the CNS, indicating a differential homing pattern. The Treg cells in lymph nodes selectively expressed both p-selectin (CD62P) and intercellular adhesion molecule type 1 (ICAM-1) which probably promoted functional interactions with target T cells. These results suggest that EAE-protective lymph node-derived Treg cells migrated back to lymph nodes in vivo and carried out their regulatory programme by local promotion of T helper type 2 (Th2) cytokines that inhibited the expansion of encephalitogenic T cells, in part through down-regulation of IL-2.
The strong resistance to EAE induction mediated by transferred lymph node Treg cells appears also to play an important role in genetic resistance to EAE. As reported by Reddy et al.33 (Vijay Kuchroo’s laboratory), PLP-139-151-specific T cells occurred at about the same frequency in lymph nodes of unimmunized EAE-susceptible SJL/J mice and EAE-resistant B10.S mice, as measured by PLP-specific tetramers. After immunization and restimulation ex vivo with PLP-139-151 peptide/CFA, SJL/J mice had a higher frequency of PLP-139-151-specific T cells, including those secreting IFN-γ. However, of critical importance, EAE-resistant B10.S mice had a higher frequency of PLP-specific CD4+ CD25+ Treg cells present in the naïve T-cell subpopulation than EAE-susceptible SJL/J mice. Depletion of Treg cells with anti-CD25 antibody enhanced Th1-biased PLP responsiveness and allowed induction of clinical EAE in ∼ 30% of otherwise-resistant B10.S mice after immunization with PLP-139-151 peptide/CFA. Although perhaps not the only factor, PLP-specific Treg cells clearly contributed to the genetic resistance to EAE in these B10.S mice.
In a contemporary report, Zhang et al.34 further demonstrated that anti-CD25 antibody pretreatment enhanced the mortality and severity of PLP-139-151-induced EAE in SJL/J mice. This treatment decreased the percentage of CD4+ CD25+ Treg cells in blood, peripheral lymph nodes and spleen that was associated with an increase in production of IFN-γ and a decrease in IL-10 in lymph node cells stimulated with PLP-139-151 peptide ex vivo. Of importance, transfer of wild-type but not IL-10-deficient CD4+ CD25+ Treg cells from unimmunized SJL/J donors partially protected recipient mice from PLP-139-151-induced EAE. Although previous work from this group has demonstrated the importance of IL-10 in the prevention of EAE,35 this was the first study to implicate IL-10 directly in Treg cell protection against EAE.
EAE-protective Treg cells manifest both antigen-specific activation and non-specific suppression
A critical point of discussion regarding Treg suppression is whether or not effects are antigen specific (based on TCR recognition) or relatively antigen non-specific (based on broadly-reactive Treg effector mechanisms). To evaluate this question, Yu et al.,36 used aggregated immunoglobulin-PLP1, an immunoglobulin carrying PLP-139-151 peptide, to induce high frequencies of Treg cells specific for the PLP-139-151 peptide in TCR Tg 5B6 mouse on the RAG-2 KO background that prevented rearrangement of Treg cells bearing different TCRs. These induced Treg cells expressed CD25, CTLA-4, Foxp3 and IL-10, and suppressed EAE induced by PLP-139-151 in both SJL/J and F1 (SJL/J × C57BL/6) mice. However, the PLP-139-151-specific Treg cells did not inhibit EAE induced by either MBP-85-99 or MOG-35-55 peptides, indicating that the Treg activity in vivo was antigen-specific. Further experiments demonstrated that PLP-139-151-specific Treg cells that were preactivated with antigen-presenting cells pulsed with the cognate PLP-139-151 peptide gained some suppressive activity that was capable of partially inhibiting EAE induced with MBP-85-99, MOG-35-55 and PLP-178-191 peptides, indicating a broad and non-specific regulatory effect. These insightful experiments placed into context the relative contributions of antigen-driven activation versus non-specific regulatory mechanisms that characterize functional aspects of Treg cells.
Treg cells from CNS can prevent but not treat ongoing EAE
It is now known that Treg cells within the CNS can contribute to protection and natural recovery from EAE. McGeachy et al.37 demonstrated the critical importance of CD4+ CD25+ Foxp3+ Treg cells in the resolution of monophasic MOG-35-55-induced EAE in C57BL/6 mice. Although a previous study from this group using chimeric mice implicated IL-10-secreting B cells in recovery from EAE,38 further investigation demonstrated that most of the cells that accumulated in the CNS during EAE were CD4+ T cells and CD11c+ DC. Splenic B cells from recovered mice could protect wild-type mice from EAE, but there were relatively few B cells in CNS. In fact, CD25+ CD4+ T cells that could be induced with phorbol 12-myristate 13-acetate/ionomycin to secrete IL-10 increased in number over the course of EAE and their appearance correlated well with recovery. These T cells expressed typical Treg markers, including Foxp3, GITR, CTLA-4, CD44, CD69 and CD103 that apparently guide Treg cells to sites of inflammation, and the CD4+ CD25+ Treg cells possessed a potent ability to inhibit in vitro proliferation and cytokine expression of CNS CD4+ CD25− Teff cells. Importantly, CNS-derived CD4+ CD25+ T cells were ∼ 100 times more potent than lymph node Treg cells at inhibiting EAE disease severity after transfer into wild-type mice just before induction of EAE. Finally, antibody depletion of CD25+ cells before the induction of EAE abrogated natural recovery, and depletion after recovery restored susceptibility of the mice to reinduction of disease.
Subsequent work by Mann et al.39 clarified how B cells regulate Treg cells and IL-10 expression during recovery from EAE. Transfer of MBP-Ac1-11 peptide-specific T cells into B-cell-deficient B10.PL mice resulted in chronic EAE marked by the delayed appearance of Foxp3+ Treg cells and attenuated expression of IL-10 in the CNS. Chimeric B-cell knockout mice reconstituted with wild-type B cells recovered from EAE and had normal IL-10 and Foxp3 expression. However, knockout mice reconstituted with B7-deficient B cells did not, indicating that B7 expression by B cells is required to mediate peak expression of Foxp3 and IL-10 in the spinal cord during EAE. Interestingly, the source of Foxp3 and IL-10 expression was localized in non-encephalitogenic T cells. These data suggest that B cells regulate activation/mobilization of endogenous Treg cells within the CNS that mediate recovery from EAE.
An even more difficult question – the ability of Foxp3+ Treg cells to reverse ongoing clinical EAE – was addressed in a recent elegant study by Thomas Korn et al.40 in Vijay Kuchroo’s laboratory (Harvard University, Boston, MA). Using Foxp3gfp knock-in mice (in which all Foxp3+ Treg cells expressed green fluorescent protein) and tetramers for tracking MOG-35-55-specific T cells, this group was able to determine relative numbers of pathogenic and regulatory T cells present in the CNS during the natural course of EAE. MOG-specific Treg cells expanded in the periphery and migrated into the CNS, but their presence in the affected organ could not prevent disease onset or progression. A tremendous advantage of the model used for these studies was the ability to determine if any of the MOG-specific Treg cells present in the CNS originated from GFP− cells that could be induced by specific antigen or the inflammatory process rather than from local expansion of GFP+ natural Treg cells. The answer to this question was clearly ‘No’, because transferred CD4+ Foxp3/GFP− cells sorted by fluorescence-activated cell sorter did not convert to a GFP+ phenotype. However, immunization with MOG-35-55 peptide did expand GFP+ MOG-specific Treg cells in both the CNS and lymph nodes, and CNS Treg cells could be distinguished by expression of CCR5 and CD103, markers that have been associated with homing of Treg cells into sites of tissue inflammation. Of critical importance to the goals of this study, GFP/Foxp3+ Treg cells from the CNS were unable to suppress the activation of MOG-35-55-specific T cells from CNS, even though these cells could inhibit naïve MOG-35-55-specific cells. Surprisingly, splenic Treg cells also failed to inhibit the CNS Teff cells, indicating resistance of these encephalitogenic T cells to Treg-mediated regulation. A cytokine analysis revealed that the MOG-specific Teff cells from CNS but not the periphery secreted high levels of both IL-6 and TNF-α. These factors in combination were shown to completely reverse Treg-mediated suppression of proliferation and IL-17 secretion of MOG-specific Teff cells. As explained by Prod’homme et al.,41 in the News and Views that accompanied this remarkable study, this lack of Treg effect was not simply a difference in relative numbers of MOG-specific Teff cells to MOG-specific Treg cells (although the ratio did reach 13 : 1 at the peak of disease before diminishing to 4 : 1 during disease recovery). Rather, the production of TGF-β by the Treg cells in combination with IL-6 and TNF-α from Teff cells apparently provided a cytokine environment that promoted Th17 differentiation while blocking Treg cell development.42 Thus, there was no defect in the MOG-specific Treg population – only a transient but overpowering effect of the Teff cells. Resolution of EAE was reflected by a reduction in MOG-specific Teff cells without an increase in MOG-specific Treg cells. Not emphasized in this article was the predominance of tetramer-negative CNS Foxp3/GFP+ Treg cells that constituted between 10 and 25% of total mononuclear cells and were > 20-fold more numerous than the MOG-specific Treg cells. These cells, with TCRs specific for antigens other than MOG-35–55, increased by 50% between the peak of disease and recovery, were responsible for the dramatic increase in IL-10 production in the CNS during recovery, and may have been the major contributors in the recovery process. The specificity of these natural Treg cells is not known, but could be directed at self TCR determinants or activation antigens expressed by MOG-specific Teff cells.43
Taken together, these reports elucidate several important properties of Treg cells as they relate to EAE and CNS inflammation: (1) Treg cells contribute significantly to genetic resistance, natural recovery, and resistance to reinduction of EAE; (2) B cells can modulate Treg function, limiting the course of EAE in vivo through mechanisms involving IL-10; (3) Pretreatment with Treg cells before the induction of EAE could prevent or reduce clinical and histological signs of EAE, whereas treatment of ongoing EAE was not effective, due in part to resistance mechanisms (including secretion of IL-6 and TNF-α) by Teff cells that in combination with TGF-β promoted pathogenic activity of Th17 cells and blocked Treg cell activation; (4) The specificity of Treg cells need not be the same as that of the encephalitogenic T cells, although it remains unknown if such specificity confers greater suppressive activity in EAE.
Treg cells may be reduced in peripheral blood mononuclear cells from subjects with multiple sclerosis
There has been longstanding interest in the possibility that subjects with MS may have reduced levels of functional regulatory cells that could allow the escape of pathogenic neuroantigen-reactive T cells. The first article that implicated CD4+ CD25+ Treg cells was from Viglietta et al.44 from David Hafler’s laboratory (Harvard University). In this study, blood cells were obtained from 15 patients with relapsing–remitting MS and 21 healthy controls (HC) and separated into CD4+ CD25high suppressor cells and CD4+ CD25− indicator cells. There were no differences in the percentages of CD4+ CD25+ or CD4+ CD25high cells in MS versus HC donors. The two cell subpopulations alone and 1 : 1, 1 : 2 and 1 : 4 ratios of suppressor to indicator cell types were stimulated with optimal and suboptimal concentrations of anti-CD3 in vitro and analysed for proliferation responses and release of IFN-γ and IL-10. This approach clearly demonstrated that even a ratio of 1 : 4 Treg cells to indicator cells from subjects with MS was significantly less inhibitory than Treg cells from healthy control donors. Mixing cells from MS and HC donors demonstrated that the defect in suppression resided in the CD4+ CD25+ Treg cells from MS subjects, and the defective Treg cell population was further characterized as being CD62L+. Moreover, the cloning efficiency of CD4+ CD25high cells and their suppressive function were significantly lower in subjects with MS versus HC donors, suggesting reduced responsiveness to activation signals.
This decreased functionality of CD4+ CD25+ T cells in patients with MS was confirmed by our studies in MS versus HC donors, and we further demonstrated that there was decreased expression of Foxp3 message and protein that correlated with decreased suppression in the CD4+ CD25+ Treg cell population.45 Moreover, our study was the first to demonstrate decreased Foxp3 expression in CD4+ CD25+ T cells from peripheral blood mononuclear cells (PBMC) of 19 MS subjects versus 19 age- and gender-matched HC donors. This study linked the deficit in functional suppression to reduced expression of Foxp3 in MS.
Further evaluation of human Treg cells by Baecher-Allan et al. revealed that major histocompatibility complex (MHC) class II expression by CD4+ CD25high T cells in vitro constituted a functionally distinct subset.46 These cells had high levels of Foxp3 and early contact-dependent target cell suppression, in contrast to MHC class II negative Treg cells that had an initial up-regulation of IL-4 and IL-10 secretion and later cell–cell contact-mediated suppression. Although the differences between MHC class II positive versus negative Treg cells could be distinguished in this in vitro culture system, the functional importance of these subsets is not known and cannot easily be studied in mice because of the lack of MHC class II expression on murine T cells, including Treg cells. An additional study by Kumar et al.47 detected an increase rather than a decrease in circulating CD3+ CD4+ CD25+ cells in patients with MS versus HC donors, but no difference in expression of Foxp3 in the Treg fraction from patients with MS. However, the Treg fraction did have a reduced ability to inhibit activation of CD4+ CD25− MBP-stimulated cells, and a similar trend for reduced activation of pokeweed mitogen-stimulated cells.
Extensive data on Treg cell phenotype and function in MS subtypes was presented by Venken et al.48 from the laboratory of Piet Stinissen and Jef Raus in Belgium. As in previous studies, no differences were found in the frequency of CD4+, CD4+ CD25+ and CD4+ CD25high T cells in PBMC from patients with relapsing–remitting MS (RR-MS) or secondary progressive MS (SP-MS) versus HC donors. However, suppressive activity of CD4+ CD25+ Treg cells from untreated RR-MS subjects was significantly decreased compared to Treg cells from SP-MS and HC. Crossover experiments established that the defect was associated with the MS Treg cells, not the indicator cells. This study also established that Treg function was associated with disease duration rather than the age of the patients, unlike our initial study that suggested an age-dependent decrease in Treg suppression,49 and was not influenced by gender. Phenotypically, CD4+ CD25+ Treg cells from patients with MS were not different from Treg cells from HC donors, with cells from both sources being CD95+, CD45RO+/CD45RA−, HLA-DR+, ICAM-1+, CD54+, GITRlow and CTLA-4moderate. Finally, low functional suppression observed in CD4+ CD25+ Treg cells from PBMC of RR-MS correlated with diminished levels of Foxp3 messengerRNA expression compared to HC donors. In contrast, Foxp3 message levels in SP-MS subjects were not different from those in HC donors. Although these Foxp3 expression studies were carried out on a relatively small cohort of subjects, the results confirmed our earlier findings, at least with respect to RR-MS patients.
The puzzling differences in suppression and Foxp3 expression patterns observed in RR-MS versus SP-MS subjects were addressed with greater precision and insight in two landmark papers by the same investigators. Using flow cytometry to evaluate Foxp3 expression at the single cell level, Venken et al.50 demonstrated a reduced number of CD4+ CD25+ Foxp3+ T cells in PBMC and less Foxp3 expression per cell in 55 untreated RR-MS patients compared to 40 HC donors, 15 SP-MS patients, 16 patients with ‘other neurological disease’ (OND) with inflammatory and non-inflammatory conditions, 10 subjects with rheumatoid arthritis, and five subjects with systemic lupus erythematosis. To be accurate, there was greater variability in Foxp3 expression in the SP-MS group than in RR-MS subjects and HC donors, with a single SP-MS donor expressing a much higher percentage of Foxp3+ T cells that may have skewed the SP-MS group mean and error. This variation may also have affected the group mean and error for the mean fluorescence intensity of Foxp3+ T cells in the SP-MS group that otherwise was quite similar to the RR-MS group. Given these considerations, however, the key issue is that there was a strong positive correlation between Foxp3 mean fluorescence intensity and % of CD25high cells (P < 0·001) when all 113 MS, OND and HC donors were evaluated and that both Foxp3 parameters correlated inversely with suppression in vitro. Conceivably, it may be possible to utilize individual Foxp3 expression levels as a marker of immunological competence in a clinical setting. Interestingly, longitudinal assessment demonstrated relatively stable Foxp3 levels over a 1-year period in RR-MS and HC donors. Analysis of phenotypic markers revealed an increase in expression of CD49d (VLA-4) and CD103+ cells in the Treg population from the RR-MS group, possibly suggesting an increased capacity of Treg cells to migrate into CNS tissue.
Venken et al. also investigated homeostatic parameters important in the development and function of naïve versus memory Treg cells in RR-MS versus SP-MS patients.51 This evaluation was enhanced by enrichment of functional Foxp3+ T cells in the CD25+ CD127lo subpopulation (obviating the need for intracellular staining of Foxp3), and involved isolation of naïve CD45RA+ (nTreg cells) and CD45RO+ (mTreg cells). The suppressive capacity of the nTreg cells was impaired in both patients with relapsing MS and those with chronic MS, whereas only the chronic patients had restored mTreg cells. Correspondingly, the frequency of nTreg cells was reduced in both patients with early MS (< 10 years) and those with chronic MS (≥ 10 years), whereas the frequency of mTreg cells was reduced only in the subgroup with early MS. The diminished level of nTreg cells was due apparently to decreased thymic output, because TCR excision circles (TREC) were reduced in the MS cohort. Early versus chronic MS patients had lower expression of CD31 (PECAM), an adhesion molecule that is down-regulated after homeostatic proliferation, on both nTreg cells and mTreg cells, indicating a high cell turnover perhaps as a result of the early inflammatory phase of MS. However, CD31 was expressed on mTreg cells from patients with chronic MS at normal levels, indicating that mTreg but not nTreg homeostasis recovers during disease progression. Importantly, coefficient analysis showed that age was the explanatory variable for CD31 expression of nTreg cells and disease duration for CD31 expression of mTreg cells. This remarkable paper thus provides an explanation for both the more obvious nTreg and mTreg deficit in early MS and for the partial nTreg deficit that remains in patients with SP-MS.
Increased Treg cells in cerebrospinal fluid from patients with MS
Although it is not yet possible to evaluate the presence of Treg cells in human CNS tissue, it is possible to quantify Treg cells in the cerebrospinal fluid (CSF). Haas et al.52 evaluated CD4+ CD25+ cells in PBMC and CSF of patients with MS and HC donors. As has been shown in previous studies, no differences were found in CD4+ CD25+ T-cell numbers in blood (n = 73/group) or CSF (n = 15 MS subjects). Moreover, no differences were observed in Foxp3 messenger RNA expression from CD4+ CD25+ PBMC (n = 13/group), in spite of reduced suppressive properties of these cells in the MS group versus CD4+ CD25− indicator cells reactive with recombinant MOG, anti-CD3/CD28 mAb or allogeneic cells. The functional defect was found to reside in the MS Treg cells and not the indicator cells, and was not affected by disease activity or sensitivity towards IL-2 deprivation cell death. Additionally, TREC expression was low and not different between MS patients and HC donors. It is not clear why the approach used to detect Foxp3 and TREC expression failed to discern differences between patients with MS and HC donors in the blood and CSF, but it may be that there are differences among various MS patient cohorts in the distribution of regulatory cell types, including Foxp3+ Treg cells, Tr1, Th2, Th3 and CD8+ subpopulations. Additionally, discrimination between Foxp3+ nTreg cells and mTreg cells might enhance the ability to discern differences in these subpopulations.
Using CD27, CD4 and CD25high as markers of Treg cells, Venken et al.50 demonstrated that Treg cells in CSF from patients with RR-MS were increased compared to paired samples from blood, whereas CSF Treg cells from patients with OND or SP-MS were lower than in blood. A similar comparison of Treg cells in CSF versus blood was reported by Feger et al.53 Unlike the earlier Huan and Venken reports,45,50 the Feger study did not observe significant differences in Foxp3 expression in PBMC of 36 patients with MS versus 40 HC donors, even though the same anti-Foxp3 antibody (PCH101 from eBioscience, San Diego, CA) was used. However, decreased suppression was observed in CD4+ CD25+ Treg cells from a small sampling of MS versus HC donors. The major finding of the Feger paper was the increased frequency of Treg cells in the CSF versus blood of 14 MS subjects representing various disease subtypes, with no such changes observed in nine OND subjects. Taken together, these two important studies indicate an enrichment of Treg cells in the CSF of patients with MS that appears to be related to the autoimmune inflammatory state. The presence of the migration markers, CD49d and CD103, in PBMC Treg cells from patients with RR-MS suggests that the Treg cells may be migrating from the blood to the CSF in these patients. Moreover, based on the EAE studies discussed above,37,40 one cannot dismiss the possibility that there could be even further enrichment of highly suppressive Treg cells within the CNS parenchyma. One might even postulate that the relative inability of peripheral Treg cells to inhibit neuroantigen responses47,52 might be a reflection of selective migration of these disease-activated Treg specificities into CNS inflammatory lesions. The disturbance of naive Treg cell development that occurs in patients with RR-MS51 may impair homeostasis and retard replacement of peripheral Treg cells, providing an explanation for the relatively low percentage of Treg cells in blood.
Key findings and future directions
There are several key issues that reappear consistently in the examples cited above. (1) It is functionally important whether the Treg cells are natural, constituitively Foxp3+, thymus-derived nTreg cells or induced Treg cells in which Foxp3 can be up-regulated by agents such as TGF-β, antigen stimulation, or oestrogen.54 This issue is of particular importance in translational studies. On the one hand, the nTreg cells appear to be quite similar in lineage, frequency and function in mice and humans. On the other, induced Treg cells from mice have more stable expression of Foxp3 and associated regulatory function than human Treg cells, which may have only transient expression of Foxp3 or Foxp3 expression without detectable regulatory activity.55,56 (2) Specificity of the Treg cells for target antigens in most cases confers more suppressive activity. In a review by Tang and Bluestone,12 it was estimated that islet antigen-specific Treg cells were ∼ 50 times more effective in controlling type 1 diabetes than polyclonal Treg cells and this probably applies to other diseases as well. This makes sense in that the specific Treg cells would likely be attracted to the same antigen-rich locations as pathogenic Teff cells, and might even be juxtaposed and activated simultaneously by the same antigen-presenting DC. This is important because Treg and Teff cells in the same proximity would allow T–T cell contact or short-range regulatory cytokines to inhibit target Teff cells. Moreover, it has been proposed that one of the most important mechanisms implemented by Treg cells involves the inhibition of the ability of DC to activate Teff,18 so DC–Treg interactions could prevent the activation of many specificities of Teff to multiple tissue antigens. (3) Induction of specific homing markers on Treg may direct organ localization of Treg cells. The Treg cells isolated from lymph nodes that are CD62L+ appeared to home mainly to lymph nodes,23,32 whereas Treg cells induced or conditioned in mesenteric lymph nodes by CD103+ DC up-regulated integrin α4/β7, which directed migration to gut mucosal tissue.22,57 Additionally, T-cell trafficking to the skin is mediated by P-selectin and E-selectin ligands as well as chemokine receptors, CCR4 and CCR10,57 and Treg cell trafficking to inflamed tissues is influenced by CD103. (4) Protection and recovery from disease (including colitis, type 1 diabetes and EAE) in several cases involved secretion of IL-10 in the inflamed tissues, either by antigen-specific Treg cells or bystander cells. (5) The local microenvironment may influence Treg cell induction and function. In mesenteric lymph nodes, therefore, TGF-β and retinoic acid may affect the expression of Foxp3 and gut homing receptors. (6) Teff cells may become resistant to regulatory strategies of Treg cells, as was the case for MOG-reactive Teff that produced IL-6 that appeared to redirect the Treg cells towards IL-17+ Teff cells.40
On a final note, it would appear that in the quest to define the activity of Treg cells specific for autoimmune target antigens, an important source of Treg cells has been largely overlooked. These Treg cells have been exposed primarily in TCR transgenic mice, where autoantigen-specific Treg cells can be identified using tetramers.40 The Treg cells in question are usually tetramer negative but still Foxp3+, and in most cases the specificity of these cells has not been identified. In the Korn study,40 these Foxp3+ cells represented > 90% of the Treg cells in the CNS, they increased in frequency as the mice recovered from EAE, and they could have been the source of IL-10 that was also highly increased during recovery. In the Tang study in type 1 diabetes, one unresolved issue was the exact specificity of the expanded Treg cells from the BDC2.5 Tg mice. This mouse is not on a RAG KO background, and so there will be a small percentage of T cells with TCRs to additional antigens besides the islet peptide that also would have been expanded by the anti-CD3/CD28+ IL-2 procedure, and could have contributed to the observed potent prevention and reversal of disease. As a third illustration, the Hori studies and our own study in Tg MBP-specific mice revealed CD4+ Tg-positive regulatory cells with dual TCR specificities27 as well a CD4+ Tg-negative population of T cells that had all the expected properties of Treg cells, and that prevented onset of Sp-EAE in 100% of recipient T/R− mice.29,30
The possibility exists that some Treg cells in these mice could be directed at the pathogenic specificity itself, as we have shown for self TCR peptide-specific T cells.58 In the type 1 diabetes study,23 expansion of the Treg cells with anti-CD3/CD28 would not discriminate between islet antigen-specific Treg cells and Teff-specific Treg cells, so both might well be found in the expanded population. Even expansion with MHC class II–peptide complexes may not prevent simultaneous expansion of Teff-specific Treg cells that would have an abundant source of specific antigen expressed on the Teff. As discussed by Tang and Bluestone,12 Treg cells that are in contact with target T cells through specific ligand interactions would probably be more potent than Treg cells activated separately, so having to exert regulation through bystander mechanisms. Relevant to this issue is the observation that mouse T cells lack expression of MHC class II molecules, so limiting the formation of an MHC–antigen structure to directly ligate the Treg TCR. Our experience (personal communication) indicates that unlike wild-type mice, Teff cells in Tg mice often have detectable but limited expression of MHC II, possibly providing some direct contact. However, the situation is entirely different in humans, in whom T cells abundantly up-regulate MHC class II after activation through the TCR. Therefore, human Teff cells can easily present a diverse array of self antigens, including self TCR sequences that can ligate TCRs of cognate T cells including Treg cells. Moreover, human TCR-specific T cells may be Foxp3+ upon direct isolation from blood, and can be expanded ∼ 100-fold in vivo after vaccination with TCR peptides in incomplete Freund’s adjuvant (Fig. 1), resulting in restoration of deficient Foxp3+ Treg cells (see Fig. 2 and ref. 59).
We propose that TCR-specific T cells, as well as other self-antigen-reactive T cells (e.g. those to IL-2, MHC II, etc.) represent a natural regulatory network of Foxp3+ nTreg cells that have been positively selected in the thymus.60 These cells emigrate to the periphery and can recognize Teff cells expressing the cognate peptides presented with MHC II molecules during T-cell activation (Fig. 3). Since TCR peptide treatment of Lewis rats can reverse the established clinical signs of EAE, we further propose that direct contact of TCR-reactive Treg cells with Teff cells can down-regulate their activation and so would be expected to inhibit ongoing disease. As thymic nTreg cells, these cells would probably have a naïve phenotype (CD45RA+). Activation in the periphery would lead to expansion of the Treg cells and an increased ability to migrate into the tissues to regulate target Teff cells, as well as bystander T cells. Thus, TCR peptide vaccination might be one approach that could expand potent Treg cells, which could reverse ongoing CNS inflammation.
This work was supported by National Institutes of Health grants NS23221 and NS23444, The Nancy Davis MS Center Without Walls, and the Department of Veterans Affairs Department of Biomedical Research.