Regulatory T cells in HIV infection: pathogenic or protective participants in the immune response?

Abstract

Introduction Suppressor or regulatory T cells (Tregs) have enjoyed a checkered history since the first report of in-vivo suppressor T-cell activity in 1971 [1]. Widespread acceptance of Tregs as a bona fide T-cell subset came only with the identification of a surface marker, CD25 (the alpha chain of the IL-2 receptor), that was constitutively expressed by CD4 T cells with suppressive ability in vivo[2,3]. The demonstration that purified CD4+CD25+ T cells could suppress the proliferation of CD4+CD25− T cells in vitro provided a crucial confirmation of the link between phenotype and function [4]. Importantly, the in-vitro suppression assay could also be applied in human disease [5]. The past decade has seen an explosion of research interest in Tregs, as the tools of modern cellular and molecular immunology have been applied to understanding immune regulation in mouse and humans [6–10]. In particular, the discovery that the transcription factor Foxp3 is highly expressed by Tregs in both species, and is required for their development and function [11–14], has opened the way for elegant studies of Treg biology in mouse models [15] and for more reliable identification of Tregs in the human [16]. This review will briefly describe the current state of understanding of normal Treg biology, most of which is still based on murine models, and will then detail how Tregs are affected by HIV infection, and how they may influence the course of HIV disease. Identification of regulatory T cells Murine studies As mentioned above, the expression of CD4 and CD25 was initially used to define Tregs in the mouse [3]. As activated CD4 T cells also express CD25, purified CD4+CD25+ Tregs were likely to be contaminated with an unknown proportion of activated conventional T cells. Such contamination did not present a major problem in animals raised in pathogen-free conditions, as the number of conventional CD4 T cells expressing CD25 was usually small compared with the number of Tregs. Murine studies have continued to use CD25 as the principal marker for the isolation of viable Tregs, whereas FoxP3 expression is more commonly regarded as the most definitive marker after cell fixation (Table 1).Table 1: Regulatory T cell phenotype in mouse and human.Human studies In contrast to the mouse, human circulating effector/memory CD4 T cells express CD25 at a level only slightly lower than that of Tregs [17], and this finding has presented a major obstacle to the accurate identification of human Tregs. To avoid massive contamination of purified human Tregs with memory cells, investigators initially chose to study the relatively pure minority Treg subpopulation expressing the highest level of CD25 [5]. Human Tregs have also been identified by the expression of a number of molecules that, although not exclusive to the Treg subset, are either more highly or more uniformly expressed when compared with conventional CD4 T cells (Table 1). These molecules include members of two families of co-stimulatory molecules: the CD28 superfamily, represented by cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) and inducible co-stimulator protein, and the tumor necrosis factor receptor superfamily, represented by CD27, OX40, 4-1BB and glucocorticoid-induced tumor necrosis factor receptor (GITR) family-related protein. In addition, human Tregs have been subdivided according to their activation state. Most Tregs in neonates are CD45RA-positive, whereas the majority in adults express CD45RO. The presence of a discrete population of CD45RA-positive Tregs in adults has also recently been described by several groups [17–19]. These naive Tregs probably comprise that part of the peripheral Treg repertoire that has not received sufficient T-cell receptor stimulation to convert to the expression of CD45RO. In general, CD45RA-positive Tregs express CD25 and co-stimulatory molecules at lower levels than CD45RO-positive Tregs, indicative of a lower ‘activation’ status [17]. Two recent technical advances have improved the identification and isolation of the majority of human Tregs that do not express the highest levels of CD25. The generation of anti-FoxP3 monoclonal antibodies allowed Tregs to be identified after fixation and permeabilization [16], whereas the surface CD127loCD25+ phenotype was shown to be highly correlated with intracellular Foxp3 expression, providing a means of purifying viable human Tregs [20,21]. It should be noted, however, that FoxP3 expression is not always indicative of regulatory status within human CD4 T cells [17,40–44]. Mechanism of regulatory T cell function Murine studies Most of our knowledge of in-vivo Treg function comes from murine models. Tregs are involved in virtually every stage of the adaptive immune response. They thus not only prevent the activation of self-reactive cells, but downregulate responses to foreign antigen at the expansion and effector stages (Table 2). Blocking Treg function in vivo has been shown to stimulate rejection of otherwise tolerated tumors [45] and to allow the clearance of chronic infections under some experimental conditions [46]. Despite their potent in-vivo activity, the mechanism of murine Treg suppressive activity remains unclear, with evidence both for and against the involvement of transforming growth factor beta (TGF-β), IL-10, CTLA-4, target cell killing and the induction of indoleamine 2,3-dioxygenase (IDO) production by dendritic cells and macrophages [25,26,47–52]. IDO is an enzyme that degrades tryptophan to kynurenine. IDO activity not only deprives T cells of an essential amino acid, but drives the production of immunosuppressive tryptophan metabolites [53].Table 2: Potential effects of regulatory T cells on immune responses and on HIV infection in particular.Suppression of CD25− responder T-cell proliferation to an artificial polyclonal stimulus such as anti-CD3 monoclonal antibody has been widely adopted as the definitive assay for murine Treg function. In these cell co-culture assays, the secretion of TGF-β and IL-10 by Tregs can usually be detected. In-vitro suppression is not, however, dependent on soluble factors, but requires physical contact between Tregs and responder T cells, involving an as yet undefined molecular mechanism [54]. On the basis of these findings, it was surprising that murine Tregs mediating in-vivo regulatory function were recently shown not to contact responder T cells [55]. To add to concerns about the validity of the in-vitro suppression assay, a number of recent reports have detected normal in-vitro suppressive function mediated by Tregs derived from gene knockout mice with major deficits in self-tolerance and in-vivo regulatory function [56,57]. In-vitro interactions between murine Tregs and dendritic cells have been reported to reduce dendritic cell expression of co-stimulatory molecules, which could explain how they downregulate multiple aspects of immune activation and differentiation [58–60]. Whether this is achieved by means of soluble mediators or cell–cell interaction remains an open question. Treg–dendritic cell interaction mediated via CTLA-4 ligation of CD80/86 has been shown to induce IDO production by dendritic cells [52]. This mechanism accounts at most for only part of the spectrum of Treg functions, because IDO knockout mice do not manifest an autoimmune phenotype [61], in contrast to FoxP3 knockout mice with no Tregs [62]. Human studies Human studies of Treg function have of necessity been limited to in-vitro testing. Human Tregs can suppress polyclonal stimulation of conventional T cells (usually CD25− cells stimulated with anti-CD3 with or without anti-CD28 or allogeneic non-T cells as antigen-presenting cells). Like murine Tregs, human Tregs make soluble immunosuppressive mediators such as TGF-β and IL-10, but require cell–cell contact with target cells to function in vitro. Once again, the molecular mechanism underlying suppression is still highly controversial. In the early studies of Baecher-Allen et al.[5] reliable suppressive activity was limited to the CD25high population. With the removal of contaminating CD25intCD45RO+ conventional T cells, we have now shown that CD25int Tregs have equally potent suppressive activity [20], as have CD45RA-positive naive Tregs [17]. It seems counterintuitive that naive and effector memory Tregs should have the same functional activity, given the clear in-vitro differences between conventional naive and effector memory cells. Furthering our understanding of how Tregs function at a cellular and molecular level is clearly one of the major challenges currently facing the field. Selection and specificity of regulatory T cells Murine studies Elegant genetic studies using green fluorescent protein to mark Foxp3+ Tregs in vivo have shown that Tregs arise in the thymus after selection for intermediate affinity recognition of self-antigen, which drives the expression of CD25 and then Foxp3 [15]. Tregs can also be generated de novo from naive peripheral CD25−CD4+ T cells after prolonged low avidity stimulation [63] or in skin graft recipients treated with non-depleting anti-CD4 antibodies [64], but the ongoing contributions of these peripheral pathways to the adult repertoire of Tregs remain unknown. In vitro, TGF-β can drive the differentiation of naive murine CD25−CD4+ T cells to a Foxp3+ suppressive phenotype [65] and recent in-vitro and in-vivo evidence supports a role for a combination of TGF-β and all-trans retinoic acid, a metabolite of vitamin A, in de novo Treg differentiation, particularly in the gut [66–69]. The antiself specificity of peripheral Tregs has not been formally demonstrated in normal mice, because of the technical difficulty inherent in detecting these intermediate affinity T-cell receptor interactions. Foxp3+ Tregs with defined specificity for foreign antigen have, however, been isolated from peripheral lymphoid and non-lymphoid tissues. In particular, murine Tregs specific for Leishmania major[46,70], and allo-major histocompatibility complex antigens [71] have been well characterized. These cells could be derived from self-reactive thymic Tregs that crossreact against foreign antigen, or from peripheral Foxp3− cells. It was recently demonstrated that at least some of the Leishmania-specific FoxP3+Tregs are the descendants of peripheral CD25+CD4+ T cells [70], but whether they first expressed FoxP3 in the thymus or in the periphery remains unknown. Human studies FoxP3+ T cells are present in human thymus and cord blood [17], suggesting a similar differentiation and selection mechanism to that in the mouse. Whereas FoxP3 expression appears to lag behind the expression of CD25 in the mouse, approximately half the FoxP3+ cells in the human infant thymus are CD25−[17], once again underlining the differences in FoxP3 expression between the species. Self-reactive specificities within human Treg populations have not been demonstrated experimentally, but several studies have documented reactivity against well-defined foreign antigens, including hepatitis C virus (HCV) [72]. Once again, whether antiforeign reactivity represents a crossreactive specificity of a Treg with antiself specificity, or whether it reflects differentiation from a peripheral antiforeign conventional T cell is not known. In-vitro differentiation from naive human CD4 T cells in the presence of TGF-β [73], vitamin D [74] or prostaglandin E2 [75] induces FoxP3 expression and suppressive function, and these pathways may also be used to generate Tregs in vivo, particularly during ongoing responses in the gut. Our own unpublished data (Fazekas de St Groth, Smialkowski and Dervish) derived from paired biopsies of involved versus uninvolved bowel mucosa in ulcerative colitis patients have shown a positive correlation between the presence of an inflammatory infiltrate and both the percentage and absolute number of CD4+FoxP3+ T cells. These data are consistent with a model in which the increases in FoxP3+ T cells within the gut lamina propria in conditions such as inflammatory bowel disease are secondary to the inflammatory process, operating as a negative feedback loop to control the adverse effects of inflammation (Table 2). Molecules involved in regulatory T cell biology in vivo Studies using genetically modified mice have provided many insights into the factors required to generate and maintain normal populations of Tregs. Understanding these factors is essential if we are to understand how Tregs are involved in immune responses to pathogens such as HIV. IL-2 Murine studies IL-2 is required for normal Treg function in addition to its role in the maintenance of Treg numbers. Mice lacking IL-2 [76], IL-2Rα [77] or IL-2Rβ [78] have profound regulatory abnormalities, although CD4+Foxp3+ T-cell numbers are only modestly reduced [79]. The expression of CD25 by Foxp3+ cells is markedly decreased in IL-2-deficient mice, and the administration of IL-2 rapidly normalizes CD25 expression via the Stat5 pathway [80] and increases Foxp3 expression per cell [79]. IL-2 is essential for normal Treg function in vitro, as demonstrated by experiments in which the addition of anti-IL-2 monoclonal antibodies blocked suppression in co-cultures of Treg and conventional Tcells [81,82]. Human studies Recent clinical trials involving the administration of IL-2 have reported increases in circulating Treg numbers, particularly within the CD45RA compartment, indicating that human Treg viability is also supported by IL-2 [83–87]. IL-2 also increases FoxP3 expression within sorted CD4+CD25+ T cells in vitro via the classical STAT3/STAT5 signalling pathway [87]. Transforming growth factor beta Murine studies The role of TGF-β in Treg biology is complex. Mice deficient in TGF-β or its receptor develop multiorgan inflammation and die at a young age [88]. Both conventional and regulatory T cells make TGF-β and the normal maintenance of self-tolerance requires that both subsets also express functional TGF-β receptors [48]. Whether TGF-β is directly involved in mediating Treg function remains controversial [39,57]. Certainly TGF-β treatment of murine Foxp3− cells can induce both Foxp3 expression and regulatory function in vitro[65]. Human studies TGF-β treatment can also convert human Foxp3− cells to a suppressive, FoxP3+ phenotype in vitro[73]. Consistent with the high TGF-β levels in the gut, increased numbers of Tregs in the gut mucosa have been documented in human inflammatory bowel disease (IBD) [89] and in HIV infection (see below). As mentioned above, our own data from IBD patients indicate that Tregs increase in proportion to the degree of inflammation in the bowel mucosa, suggesting that they are responding to inflammatory stimuli. Increases in Tregs are, however, not a universal accompaniment of bowel inflammation, as Treg numbers in the gut are reduced in graft versus host disease [90]. Whether human gut mucosal Tregs are induced in situ from Foxp3− cells or derived from circulating Foxp3+ cells remains to be determined, although recent reports concerning the effects of TGF-β and retinoic acid in the murine gut (above) strongly suggest in-situ induction. Cytotoxic T-lymphocyte-associated protein 4 and indoleamine 2,3-dioxygenase Murine studies CTLA-4-deficient mice develop a rapidly fatal lymphoproliferative syndrome that can be overcome by the provision of wild-type bone marrow in mixed chimeras, indicating that CTLA-4-deficient Tregs cannot mediate normal regulatory function in vivo[91]. This finding is paradoxical, because CTLA-4 transmits a negative rather than a positive signal, so its absence should enhance rather than inhibit the function of cells that express it. It has been postulated that CTLA-4 must therefore serve different functions in Tregs and conventional T cells. One such alternative function of CTLA-4 is to induce dendritic cells to express IDO, thus inhibiting T-cell proliferation as mentioned above [52]. Recent in-vitro evidence indicates that IDO may also function to induce de novo FoxP3 expression in CD4 T cells, suggesting that CTLA-4 and IDO may be part of a positive feedback loop between dendritic cells and Tregs [92]. Human studies Anti-CTLA-4 antibody is in clinical trial as a stimulant of antitumor immune responses [93]. Potentially life-threatening immune-related adverse events (IRAE) such as colitis, hypophysitis and adrenal insufficiency have been noted in a substantial number of patients in those trials, consistent with the postulated role of CTLA-4 as a mediator of Treg function [94,95]. Fortunately, these IRAE are generally reversible with standard anti-inflammatory therapy. A requirement for CTLA-4 expression during TGF-β-dependent induction of Foxp3 in naive human CD4 T cells has recently been demonstrated in vitro[96], linking CTLA-4 with the peripheral pathway of Treg generation in addition to its role in the function of Tregs differentiated in the thymus. There is also evidence for the bi-directional link between CTLA-4 and IDO mentioned above. CTLA-4 has thus been shown to induce the production of IDO by human dendritic cells [97,98] and conversely IDO can convert human CD4+CD25− T cells to CD4+CD25−FoxP3+ cells [99]. PD-1 and regulatory T cells PD-1 is an inhibitory receptor expressed on the surface of differentiated effector T cells in both the mouse and human [100,101]. It has recently been shown that the expression of PD-1 by CD8 T cells in chronic infections such as HIV is associated with decreased cytotoxic activity [102–104]. The expression of PD-1 in Foxp3+ Tregs is restricted to the cytoplasm, but surface expression can be induced by activation in vitro[105]. The role of PD-1 expression in Treg function is currently not known. Regulatory T cells as part of a normal feedback loop in inflammation There is increasing evidence that FoxP3+ Tregs are overrepresented as a proportion of CD4 T cells at inflammatory sites in both the mouse and human. The best evidence for such increases is in IBD [89]. As mentioned above, whether such cells are derived directly from the thymus or are induced in the periphery is unknown. Certainly, IDO expression is also upregulated at these sites and may be part of the positive feedback loop described above. Do regulatory T cells prevent clearance of infectious organisms? Murine Leishmania Healed Leishmania skin lesions in resistant mice contain a high proportion of Tregs and a few viable organisms [46]. The transfer of Leishmania-specific T cells depleted of Tregs can produce sterilizing immunity in immunodeficient mice, suggesting that Tregs prevent full clearance under normal conditions [46]. Importantly, however, Tregs neither prevent healing of the skin lesion, nor are they implicated in lesion spread in susceptible mouse strains. Human hepatitis C virus It has been suggested that excessive Treg activity may be associated with the induction and maintenance of the HCV carrier state, based on evidence that Treg numbers are increased in the peripheral blood of HCV-infected patients [106] and that the specificity of Tregs for HCV antigens can be demonstrated [72]. Among HCV-infected individuals, however, liver enzyme levels and hepatic damage correlate with lower rather than higher Treg numbers, consistent with a protective rather than pathogenic role of Tregs in HCV [107]. In chimpanzees, Tregs are equally high in chronic HCV carriers and spontaneously recovered animals [108]. It thus appears unlikely that Tregs are entirely responsible for the carrier state in HCV infection. Regulatory T cells in HIV It has been suggested that excessive Treg activity may be responsible for establishing a carrier state in HIV-infected individuals. HIV infection is accompanied by an early loss of immune function affecting multiple cell types [109], which could conceivably be caused by abnormally high regulatory activity. Regulatory T-cell numbers in HIV infection The number of Tregs detected in samples from HIV-infected patients is influenced by a number of factors, including the tissue sampled, the stage of infection, and whether antiretroviral therapy (ART) has been instituted. Table 3 summarizes a number of reports of Treg numbers in peripheral blood, tonsils, lymph nodes and duodenal mucosa of HIV-infected patients and SIV-infected rhesus macaques. The picture has been complicated by the use of different Treg detection methods in different laboratories, each with their individual sources of error (Table 1). For example, the use of a CD4+CD25+ gating strategy would result in different levels of contamination by activated CD4 T cells, depending on the clinical status of the patient. Alternatively, the CD4+CD25hi gate underestimates total Treg numbers by up to sixfold in normal individuals [20] and by an unknown proportion in HIV-infected individuals. Until recently, Foxp3 expression was estimated only by polymerase chain reaction for Foxp3 messenger RNA, usually without full quantitation. Characteristic Treg markers such as CTLA-4 and GITR have also been used as indirect measures of Treg number and activity, but once again are also expressed by a subset of activated CD4 T cells. The use of standardized multicolor antibody cocktails for Treg identification and purification (Table 1) should rapidly increase the precision with which we can estimate Treg numbers, phenotype and function in HIV infection.Table 3: Published studies of regulatory T cell number in HIV and SIV infection.Infected tissues A consensus is beginning to emerge as more studies of Tregs in HIV are published and compared with the data from the SIV model. In lymphoid tissues such as the tonsil [114,122], a large increase in Tregs, detected by immunohistochemical staining for Foxp3 expression, has been reported (Table 3). Expression of CD25 by these Foxp3+ cells is lower than in blood or mucosa [114,122]. Like the tonsil, lymph nodes from untreated HIV-infected individuals contain an unusually high number of Tregs and dendritic cells with a semi-mature phenotype and the ability to induce Foxp3 expression in vitro[117]. In the gut, which is a site of primary HIV and SIV infection associated with early and prolonged CD4 T-cell depletion [123,124], the absolute number of CD4+Foxp3+ Tregs increases during infection by HIV [119] and SIV [118]. Up to 50% of total mucosal CD4 T cells express Foxp3 in some treatment-naive individuals [119]. Several mechanisms that may explain HIV-related increases in Tregs have recently been reported (Fig. 1). Zaunders et al.[125] have provided evidence that CD4+CD127lo T cells are relatively spared in HIV infection, compared with conventional CD4+CD127hi T cells. Survival of Tregs may be a consequence of the ability of FoxP3 to repress retroviral transcription from the HIV promoter, partly by blocking the activation of nuclear factor kappa B [126]. Nilsson et al.[122] recently showed that HIV gp120 could increase Foxp3 expression in cultured CD4+Foxp3+ T cells, while inhibiting their apoptosis. Hsieh et al.[127] have suggested that one of the effects of HIV-1 Tat on dendritic cells is to increase CTLA-4 expression and the suppressive potency of Tregs. A fourth possibility is that the increase is a natural response to inflammation at the site, analogous to the situation in IBD (see above).Fig. 1: Postulated mechanisms whereby HIV may increase regulatory T cells during primary infection at sites of viral entry and early replication such as the reproductive and gastrointestinal tracts. Red arrows indicate an effect that increases regulatory T cell (Treg) activity and green arrows an effect that decreases Treg activity. Multiple mechanisms may lead to relative or absolute increases in Tregs. Tregs have fewer copies of HIV DNA than CD4+CD127hi cells [125,126], and may therefore survive preferentially at the site of infection (a). HIV gp120 increases Treg FoxP3 expression and inhibits apoptosis of Tregs via a CD4-gp120-dependent pathway [122] (b). In addition, HIV Tat increases Treg expression of cytotoxic T-lymphocyte-associated protein 4 and suppressive activity [127] (c), whereas immune activation and inflammation induce Treg proliferation as part of a natural feedback loop involving transforming growth factor beta (TGF-β) (d). Finally, Tregs induce indoleamine 2,3-dioxygenase (IDO) expression in dendritic cells, which in turn induces de novo FoxP3 expression in non-Tregs [92] (e). Circulation of Tregs between infected peripheral sites and lymphoid tissues such as lymph node and tonsils leads to secondary increases in Tregs in blood and lymphoid tissues (f). Antiretroviral therapy (ART) reduces the viral load, thereby reducing both the direct effects of HIV gp120 and Tat, and the secondary increase in immune activation and inflammation (g).The institution of ART has been shown to reduce markedly the number of Tregs in tonsils [114] and duodenal mucosa [119], but not in lymph nodes [117]. Whether this is a direct consequence of reduced viral load or a secondary effect of decreased inflammation is not yet clear. Peripheral blood In peripheral blood, a number of groups have reported minor increases in the percentage of Tregs within the CD4 T-cell compartment in subgroups of HIV-infected individuals [110–113,115,116,119,120], whereas others have seen no change [121] (Table 3). In general, individuals with the largest increases in Tregs as a proportion of CD4 T cells have CD4 T-cell counts below 200 cells/μl, and high viral loads, although the Treg percentages in this very lymphopenic group tend to be highly variable [110,113,114,116,120]. Blood Treg percentages tend to normalize after the institution of ART [119], consistent with the data from the gut and tonsils. Several studies have shown decreases in FoxP3 mRNA in peripheral blood T cells of patients with high viral loads or low CD4 T-cell counts [110,114–116], despite increases in the proportion of CD4+CD25hi T cells. A decrease in FoxP3 expression per cell suggests that regulatory function may be compromised. The decrease may be secondary to a relative deficiency in IL-2, which has been shown to control FoxP3 transcription (see above). FoxP3 mRNA levels normalize with ART [114,120]. Functional assays of regulatory T cells in HIV infection Initial studies of Treg function involved the comparison of in-vitro peripheral blood T-cell responses to HIV and to co-infections such as cytomegalovirus and Mycobacterium tuberculosis with and without the depletion of Tregs, and in some cases with the addition of purified CD4+CD25+ T cells [111,112,115,128,129] (Table 4). These studies indicated that Treg activity was retained in HIV infection, but did not determine its potency compared with the level in uninfected individuals. Early studies of peripheral blood regulatory activity suggested a decrease in patients with high viral loads and low CD4 T-cell numbers [111,115]. Similar conclusions were drawn from studies of infants, of whom those who were HIV infected had fewer Tregs and a smaller increase in HIV-specific CD8 T-cell responses after Treg depletion than exposed uninfected infants [130]. A recently published study has, however, provided definitive evidence that Tregs in lymph nodes retain potent suppressive activity even in late disease and in the face of high viral loads [131]. That study confirmed the decrease in FoxP3 expression and regulatory activity in the blood of infected individuals.Table 4: Published studies of Treg function in HIV infection.It is still unclear whether excessive Treg activity prevents an effective anti-HIV response at sites of primary infection such as the gut (Table 2). Further functional studies will be required to understand the impact of Tregs at different sites on the clearance of virus within the body as a whole. Regulatory T cells in animal models of HIV Correlations between changes in Treg activity and the pathogenicity of acute SIV infection of rhesus macaques, a susceptible species, and sooty mangabeys and African green monkeys, both resistant species, have failed to resolve the question of whether Tregs promote or prevent SIV-mediated disease. SIV infection in African green monkeys is accompanied by increased CD4+CD25+ T-cell frequency, FoxP3 expression and TGF-β secretion, compared with pathogenic SIV in macaques [135]. A comparison of SIV infection in macaques and sooty mangabeys, however, indicated a positive rather than a negative correlation between viral pathogenicity, immune activation, Treg activity, TGF-β levels and lymphatic tissue fibrosis [136]. Estes et al.[118] have reported dramatic increases in the frequency of CD4+CD25+Foxp3+ T cells in the T cell zones of lymph nodes in SIV infection of macaques, accompanied by increases in T cells producing TGF-β, IL10 and IDO. In comparison, more modest increases in Treg frequency and immune activation were seen in rhesus cytomegalovirus infection. A trial of anti-CTLA-4 monoclonal antibody treatment of rhesus macaques with acute SIV produced a decrease in viral load, an increase in effector function and a decrease in IDO and TGF-β levels suggestive of reduced Treg activity, prompting the conclusion that Tregs were exerting a detrimental effect [137]. Similarly, in a model in which an encephalitis sharing features with HIV encephalitis in humans was produced in immunodeficient mice reconstituted with human peripheral blood lymphocytes and given an intracranial injection of HIV-infected monocyte-derived macrophages, treatment with the IDO inhibitor, 1-methyl-D-tryptophan led to an increase in HIV-specific CD8 T cells and the elimination of infected macrophages, suggesting a role for IDO and, by extension, Tregs, in maintaining chronic infection [138]. Regulatory T cells and the design of future therapies Ablative therapies that directly target regulatory T cells Reducing Treg numbers or activity in vivo is likely to increase specific anti-HIV T-cell responses, whether or not Tregs are directly responsible for HIV-dependent immunodeficiency and the failure to control viral load. Testing treatments that target Tregs is most advanced in the field of cancer immunotherapy, with anti-CTLA-4 [93] and an IL-2 toxin conjugate [139] tested in clinical trials. Those studies have highlighted the potential for Treg ablation to precipitate severe autoimmune sequelae, including grade III/IV IRAE necessitating cessation of the therapy (see above and Weber [95]). The gut is the commonest site of major side effects, indicating the important role that Tregs play in maintaining normal immune homeostasis in the gut. Anti-CTLA-4 in particular appears to be both the most potent and the most likely to produce side effects, probably because it simultaneously reduces Treg activity and blocks intrinsic negative signals in activated high-affinity T cells. The positive effects of anti-CTLA-4 monoclonal antibody treatment of rhesus macaques with acute SIV [137] provide support for the application of anti-CTLA-4 therapy to HIV infection. Increasing effector T-cell activity by targeting PD-1 may also be effective, if it can be demonstrated that the activity of Tregs is not increased by this manoeuvre [105]. Treg ablation therapy could be designed for use at different stages of HIV infection, either alone or in combination with ART: in primary disease, in the hope of generating the type of immunity that operates in long-term non-progressors, combined with therapeutic vaccination, as in some of the cancer therapy trials, or with other novel agents such as antibodies directed to the co-stimulator 4-1BB [140]. Therapies that change the balance between regulatory T cells and other T-cell subsets IL-2 Both low and high-dose IL-2 therapy increase Treg numbers, particularly in the CD45RO-negative compartment, and increase the activation status of CD45RO-positive Tregs [83,84,86,87]. As mentioned above, IL-2 is known to increase the expression of Foxp3, which is associated with increased Treg activity in vitro. IL-2 is, however, also a potent stimulant of CD8 T-cell activity, particularly when conventional CD4 T-cell numbers are low. The effects of IL-2 may therefore differ at different stages of HIV infection, depending on the balance between regulatory and effector activity. IL-7 As most conventional T cells express high levels of IL-7 receptor whereas Tregs do not, IL-7 therapy offers the prospect of supporting T-cell reconstitution during ART without increasing Treg numbers or activity. IL-7 receptor expression by conventional T cells is, however, abnormally low in HIV infection [141], possibly as a result of downregulation in response to increased serum IL-7 levels [142]. In addition, IL-7 has been reported to increase HIV replication, although this may be secondary to increased T-cell viability [143,144]. In the first report of the effects of IL-7 therapy in cancer patients [145], both CD4 and CD8 T-cell numbers increased whereas the percentage of Tregs decreased. IL-7 therapy of SIV-infected rhesus macaques has also shown promise when combined with ART, supporting increases in CD4 and CD8 T-cell numbers in the naive and memory compartments [146]. The use of IL-7, possibly in combination with IL-2 or anti-CTLA-4, thus offers the prospect of resetting the balance of regulation in HIV disease. In conclusion, the interplay between Tregs and conventional T cells is a complex one. Tregs are involved in virtually every immune response and appear to function as a network of controlling and tuning circuits imposed above the primary activation circuits. Whereas the removal of Tregs reliably increases virtually every immune response so far tested, this increase come at a price – that of dysregulation, which may manifest as autoimmune disease or an increase in the immunopathology accompanying infectious disease. We need to understand more about the cause and effect relationships between Tregs, conventional T cells and HIV to judge whether Treg depletion will prove useful in the treatment of HIV, and whether it should be combined with other novel therapies such as the blockade of PD-1. Conflicts of interest: None.