IL-21 Restricts Virus-driven Treg Cell Expansion in Chronic LCMV Infection

Foxp3+ regulatory T (Treg) cells are essential for the maintenance of immune homeostasis and tolerance. During viral infections, Treg cells can limit the immunopathology resulting from excessive inflammation, yet potentially inhibit effective antiviral T cell responses and promote virus persistence. We report here that the fast-replicating LCMV strain Docile triggers a massive expansion of the Treg population that directly correlates with the size of the virus inoculum and its tendency to establish a chronic, persistent infection. This Treg cell proliferation was greatly enhanced in IL-21R−/− mice and depletion of Treg cells partially rescued defective CD8+ T cell cytokine responses and improved viral clearance in some but not all organs. Notably, IL-21 inhibited Treg cell expansion in a cell intrinsic manner. Moreover, experimental augmentation of Treg cells driven by injection of IL-2/anti-IL-2 immune complexes drastically impaired the functionality of the antiviral T cell response and impeded virus clearance. As a consequence, mice became highly susceptible to chronic infection following exposure to low virus doses. These findings reveal virus-driven Treg cell proliferation as potential evasion strategy that facilitates T cell exhaustion and virus persistence. Furthermore, they suggest that besides its primary function as a direct survival signal for antiviral CD8+ T cells during chronic infections, IL-21 may also indirectly promote CD8+ T cell poly-functionality by restricting the suppressive activity of infection-induced Treg cells.


Introduction
The immune system has to efficiently eliminate pathogens but simultaneously needs to avoid the potential self-damage and immunopathology caused by excessive immune activation. Therefore, a tight regulation of immune responses is critical for host survival. The subset of CD4 + CD25 + regulatory T (Treg) cells exerts key negative regulatory mechanisms of the immune system that prevent autoimmunity and T cell mediated inflammatory disease [1]. Treg cells are best defined by expression of the signature transcription factor forkhead box P3, Foxp3 [2,3,4,5,6,7]. Their fundamental role in the maintenance of immune homeostasis and tolerance is well established [8,9,10] and unambiguously demonstrated by the severe multi-organ autoimmune disease, allergy and inflammatory bowel disease that develops in Foxp3-deficient mice or patients with immune dysregulation, polyendocrinopathy, enteropathy, X-linked (IPEX) syndrome [3,11,12,13]. However, the relevance of Treg cell responses for shaping adaptive immunity against pathogens, in particular in the context of chronic infections, refigmains much less understood.
Treg cells potentially have both beneficial and adverse effects on disease outcomes during viral infections. By dampening effector immune responses, Treg cell responses mitigate immunopathology resulting from exaggerated inflammation and tissue destruction during acute [14,15,16,17], or protracted infections [18,19,20,21,22]. In addition, Treg cells have been shown to support antiviral immunity by modulating T cell migration to the site of infection [15,23]. Conversely, Treg cells were shown to suppress CD8 + T cell responses in some infections [21,24], which may prevent immunopathology, but hampers effective pathogen control and ultimately promotes persistent infection [18,21,25,26]. Thus, while Treg cells favorably influence pathogen clearance in many acute infections [14,15,16,23], they seem to negatively regulate CD8 + T cell responses during chronic infections [18,19,20,24,26]. Furthermore, elevated numbers of Treg cells have also been associated with persistent viral infections in humans [27,28,29]. However, to date little is known as to whether Treg cell activation represents a mechanism of immune evasion that facilitates persistence, or about host factors that might regulate Treg cell responses during chronic viral infection.
Many characteristics of persistent viral infections in humans, such as HIV, HCV, or HBV, are also observed during chronic infection of mice with the arenavirus lymphocytic choriomeningitis virus (LCMV). Accordingly, murine models of chronic LCMV infection have been extensively studied and have provided key insights into the molecular mechanisms leading to virus persistence and the associated modulation of adaptive immunity in face of persistent infection. For example, infection with a high dose of the fast-replicating strain LCMV Docile (or Armstrong clone 13) results in virus persistence that is accompanied by a progressive functional impairment and -for some specificities -even leads to deletion of the virus-specific CD8 + T cell response, termed T cell ''exhaustion'' [30]. Exhausted antiviral T cells gradually lose their capacity to produce antiviral cytokines and to proliferate ex vivo [31,32,33]. Similar states of T cell exhaustion have been demonstrated in patients with chronic HIV or HCV infections [34]. CD8 + T cell exhaustion directly correlates with the antigen load, i.e. is enhanced during prolonged or high viral replication. Furthermore, CD8 + T cell exhaustion is also influenced by the sustained expression of several inhibitory receptors [35] and immunomodulatory cytokines, such as IL-10 [36,37] and TGFb [38]. Moreover, it can be aggravated by the loss of help or essential cytokines provided by CD4 + T cells. In particular, we and others [39,40,41] have recently identified the pro-inflammatory cytokine IL-21 as an essential CD4 + T cell-produced factor that prevents CD8 + T cell exhaustion during chronic viral infection.
Here we have investigated the role of Treg cell responses in chronic LCMV infection and found that the persistence-prone strain Docile (DOC) induces a substantial dose-dependent expansion of the Treg cell population that directly correlated with the magnitude of the virus inoculum. The expanded Treg cell population massively impaired the functionality of antiviral CD8 + T cells, interfered with virus clearance, and thereby promoted chronic viral infection. Strikingly, enforced expansion of the Treg cell population with IL-2/anti-IL-2 immune complexes (IL-2ic; [42]) permitted the persistence of LCMV-DOC already at much lower infection doses and even predisposed to chronic infection with an LCMV-strain that generally fails to establish persistence in immunocompetent mice. Importantly, this infection-triggered  expansion of the Treg cell population was greatly inhibited by  IL-21R signaling in Treg cells, thus defining a novel role for IL-21  in preventing T cell exhaustion and viral persistence via limiting virus-driven Treg cell proliferation.

Results
Dose-dependent expansion of the Treg cell population in response to chronic viral infection We examined the role of Foxp3 + Treg cell responses in chronic viral infections in mice infected with the fast-replicating strain of LCMV-DOC. LCMV-DOC is characterized by its potential to establish a chronic, persistent infection depending on the size of the virus inoculum. Although low doses of LCMV-DOC (2610 2 -2610 3 PFU) induce potent CD8 + T cell-mediated immunity and are cleared in immunocompetent hosts, infection with intermediate and high doses (2610 4 -2610 6 PFU) results in virus persistence due to the exhaustion of the virus-specific CD8 + T cell response [30,31]. By choosing different virus inocula, we determined dynamics of the Treg cell population during an acute, resolving infection and a chronic infection. After an initial decline of Treg cell numbers between days 0-7 that was independent of the dose of infection and the LCMV strain used (i.e. DOC or WE) ( Figure 1A and data not shown), we observed a striking dosedependent expansion and recovery of the Treg cell population in LCMV-DOC infected mice ( Figure 1B, C) that directly correlated with the ability of the virus to establish persistence ( Figure 1D).  Figure 1D and data not shown); while mice infected with persistence-inducing doses (2610 4 -2610 6 PFU) of LCMV-DOC still exhibited high viral titers in blood, spleen, liver and kidney, and subsequently failed to control the infection ( Figure 1D and data not shown). As expected, we also detected a dose-dependent reduction in the frequency of gp33-specific CD8 + T cells, which was indicative of the progressing exhaustion of the T cell response, particularly in the spleen ( Figure 1E), and the resulting inability to resolve the infection ( Figure 1D).
Consumption and bioavailability of IL-2 by Treg cells has been suggested to restrict IL-2-dependent effector T cell differentiation and expansion [43,44,45,46]. Conversely, IL-2-driven expansion of CD8 + T cell expansion during an immune response can occur at the expense of Treg proliferation/survival [47,48]. Moreover, IL-21 has been suggested to interfere with Treg-mediated suppression by inhibition of IL-2 [49]. To monitor IL-2 and IL-21 production during acute and chronic infection, we took advantage of Il2-emGFP-Il21-mCherry dual reporter transgenic mice [50] that were infected with low dose (i.e. 200 PFU) LCMV-WE or high dose (i.e. 2610 5 PFU) LCMV-DOC. Both Il2 (GFP) and Il21 (mCherry) were predominantly expressed by CD4 + compared to CD8 + T cells and frequencies increased from days 7-15 post infection with low dose LCMV-WE ( Figure 1F-H). Interestingly, chronic infection with high dose LCMV-DOC potently Author Summary T cell exhaustion represents a state of T cell dysfunction associated with clinically relevant diseases, such as persistent viral infections or cancer. Although the molecular signature of exhausted T cells has been characterized in detail at the functional and transcriptional level, the immunological mechanisms that lead to T cell exhaustion during chronic infections remain poorly understood. Our present study reports two major findings that illustrate a pathway that contributes to T cell exhaustion during viral infection, and indicate its modulation by both, the pathogen and the host. First, we show that a persistence-inducing virus triggers the massive proliferation of Foxp3 + regulatory T (Treg) cells and demonstrate the potential of Treg cells to promote T cell exhaustion and chronic infection. Second, we identify IL-21 as a crucial host factor that antagonizes this virus-driven expansion of the Treg population in a cell intrinsic manner independent of IL-2. Thus, in addition to its known pre-dominant direct positive effects on antiviral T cells, IL-21 can also alleviate the suppressive activity of Treg cells. Together, these results suggest enhanced Treg cell responses as a mechanism of immune evasion that could be therapeutically targeted with IL-21.
Together, these data demonstrate a direct correlation between the size of the virus inoculum, CD8 + T cell dysfunctionality, virus persistence and the expansion of Treg cells, thus indicating a potential contribution of Treg cells to the impaired T cell function and the induction of viral persistence.

IL-21 antagonizes the virus-induced Treg cell expansion during chronic infection
IL-21 receptor (IL-21R) signaling is essential for the maintenance and sustained functionality of antiviral T cell responses in chronic infections [39,40,41]. As a consequence, IL-21R 2/2 animals have an impaired control of LCMV-DOC, and exhibit much higher virus loads [40]. In LCMV-DOC infected WT mice, the main Treg cell expansion was observed between days 10 to 20 post infection, peaking at 15 days before returning to almost naïve levels by five weeks post infection (Figure 2A, B). Strikingly, this virus-driven Treg cell expansion was much more pronounced and longer lasting in mice lacking IL-21R expression, which suggested that the pro-inflammatory cytokine IL-21 restricted the proliferation of Treg cells in viral infections (Figure 2A, B). Indeed, the Treg cells of naïve and LCMV-DOC infected WT mice expressed the IL-21R as assessed by flow cytometry ( Figure 2C). We thus investigated whether the increased Treg expansion observed in IL-21R 2/2 mice represented a direct inhibitory effect of IL-21 on Treg cells or rather was related to the increased viral replication in these mice. To address this issue, we generated mixed bone marrow (BM) chimeras by reconstituting lethally irradiated WT (CD45.1 + ) mice with a 1:1 ratio of WT (CD45.1 + ) and IL-21R 2/2 (CD45.2 + ) BM, and evaluated their Treg cell responses to infection with LCMV-DOC. Analysis of naïve BM chimeras confirmed similar reconstitution efficacy of the CD8 + and CD4 + T cell populations including Treg cells from both WT and IL-21R 2/2 donor BMs at 8 weeks after BM transfer ( Figure 2D, and data not shown). However, upon infection the population of IL-21R 2/2 Treg cells expanded 3-fold over that of WT Treg cells to represent 30% versus 10% of all CD4 + T cells, respectively ( Figure 2E-G). Since this augmented proliferation of IL-21R 2/2 Treg cells was detected side by side to WT Treg cells in the same animals and identical viral loads, it clearly established the direct inhibitory effect of IL-21 on Treg cells in vivo. Nevertheless, we considered that the higher Treg cell numbers in infected IL-21R 2/2 mice could not only result from the absence of inhibitory IL-21R signaling but might also indicate a compensatory proliferation to overcome a potential functional deficit of the IL-21R 2/2 Treg cells. To exclude the latter possibility, we isolated CD4 + CD25 + GFP + Treg cells from WT DEREG and IL-21R 2/2 DEREG mice by FACS-sorting 15 days post infection with 2610 6 PFU LCMV-DOC, and compared their suppressive activity in a classical T cell inhibition assay. As shown in Figure 2H, both WT and IL-21R 2/2 Treg cells comparably inhibited the proliferation of anti-CD3/CD28-stimulated naïve CD25 neg CD4 + T cells in vitro, suggesting a normal function of IL-21R 2/2 Treg cells. Thus, the increased expansion of IL-21R 2/2 Treg cells in LCMV-DOC infected mice highlighted an important inhibitory role of IL-21 in restraining Treg cell expansion during chronic LCMV infection. Except for a slightly reduced CD25 expression, IL-21R 2/2 Treg cells were comparable to WT Treg cells with respect to the expression or characteristic Treg cell surface markers ( Figure 2I). We did not detect any IL-10 producing or gp61-specific Treg cells in infected mice, suggesting that the suppressive activity of Treg cells in LCMV-DOC infection involved neither IL-10-mediated suppression nor virus-specific Treg cells ( Figure 2J and data not shown).
IL-6 has been suggested to regulate the balance between Treg and pro-inflammatory Th17 cell responses [51,52] similar to IL-21 [53]. While we have previously shown that IL-17-producing CD4 + T (Th17) cells were barely detectable in LCMV-DOC infection [40], it remains possible that IL-21 inhibits Treg cell expansion by regulation of IL-6. However, comparing IL-6 2/2 and WT mice we found no differences in Treg cell expansion ( Figure 2K), antiviral CD8 + and CD4 + T cell responses ( Figure 2L, M), and virus titers ( Figure 2N) up to day 30 post infection with 2610 4 PFU LCMV-DOC. Together with the Treg cell-intrinsic negative IL-21R 2/2 signaling observed in WT:IL-21R 2/2 mixed BM chimeras, these data suggest that IL-21 restricted the virusdriven Treg cell expansion in LCMV-DOC infection independently of IL-6 signaling.

Depletion of Treg cells partially restores CD8 + T cell functionality and increases disease severity during chronic viral infection
To directly evaluate the impact of Treg cells on viral persistence, we next sought to analyze antiviral T cell responses and viral clearance in the absence of Treg cells. We therefore studied LCMV-DOC infection using the DEREG mouse model, in which Treg cells can be ablated by diphtheria toxin (DT) treatment due to transgenic expression of a high affinity DT receptor under control of the Foxp3 promoter [54]. DEREG and nontransgenic WT control mice were treated with DT and infected with 2610 5 PFU LCMV-DOC on day 0, and the DT treatment was continued throughout the experiment ( Figure 3A). Since a single DT injection depleted Treg cells in naïve mice for 3 days, we injected DT every 3 days to achieve complete Treg cell ablation. While this treatment effectively depleted all Foxp3 + Treg cells in naïve DEREG mice ( [54], and data not shown), we consistently observed in LCMV-DOC infected DEREG mice the emergence of a residual GFP neg Foxp3 + Treg cell population not depleted by DT, presumably due to lacking expression of the GFP-DTR fusion protein ( Figure 3B). Compared to DT-treated WT controls, Treg cell-depleted DEREG mice exhibited a greatly enhanced morbidity in response to high dose LCMV-DOC infection, as indicated by an increased weight loss ( Figure 3C). As a result, a significant number of DT-treated, LCMV-DOC-infected DEREG mice had to be prematurely removed from the experiment and euthanized, whereas no equivalent morbidity was observed in DT-treated WT mice ( Figure 3D). Thus, Treg cell depletion -even if not absolute -substantially aggravated the disease severity in LCMV-DOC-infected DEREG mice. Depletion of Treg cells did not affect the numbers of virus-specific CD8 + T cells, as indicated by the percentage of gp33-specific CD8 + T cells in spleens and livers ( Figure 3E). However, Treg celldepletion to some extent restored the functionality of the antiviral CTL and significantly increased the frequencies of gp33-specific and overall splenic CD8 + T cells producing IFN-c upon restimulation in vitro ( Figure 3F). In comparison, Treg celldepletion did not enhance frequencies of IFN-c-producing virusspecific CD4 + T cells ( Figure 3G). In spite of partly restoring CD8 + T cell cytokine responses, Treg cell depletion did not influence virus control, and LCMV-DOC replicated to comparably high levels in the blood and organs of DT-treated WT and DEREG mice at 15 days post infection ( Figure 3H). Thus, Treg cells appeared to inhibit the functionality rather than the expansion of antiviral T cells. Consistent with the observed onset of immunopathology, the effect of Treg cell depletion on the antiviral T cell response was more pronounced at 10 days post infection ( Figure 3I-M). Depletion of Treg cells resulted in higher percentages of gp33-specific CD8 + T cells ( Figure 3J) and increased percentages of cytokine-producing antiviral CD8 + and CD4 + T cells ( Figure 3K, L), yet did not affect viral titers ( Figure 3M).
We next assessed the impact of increased Treg cell numbers in IL-21R 2/2 mice and treated both IL-21R 2/2 and WT mice with DT between days 8 to 15 post infection with 2000 PFU LCMV-DOC (Fig. 4A). Although Treg cells were considerably (but not entirely) depleted, frequencies of gp33-specific CD8 + T cells remained unchanged ( Figure 4B, C) similar to the results obtained by depletion through the entire course of infection ( Figure 3B, E). However, Treg cell depletion partially restored frequencies of IFNc-producing gp33-specific cells in IL-21R 2/2 mice to levels found in WT mice ( Figure 4D). Furthermore, Treg cell depletion lowered virus titers significantly in liver and lung of IL-21R 2/2 mice, although viral loads in spleen and kidney remained unaffected ( Figure 4F).
Whether the differences in antiviral CD8 + T cells were too small to result in better virus control or whether the failure to detect differences in viral titers ( Figure 3H, M) has to be attributed to the suppressive activity of the residual Treg cells that resisted DT depletion ( Figures 3B and 4B) remains to be clarified. Regardless, our results establish a link between the functional impairment of the CD8 + T cell response and the elevated Treg cell levels observed during chronic infection in absence of IL-21.

IL-2 driven Treg cell expansion inhibits antiviral T cell responses and promotes persistent viral infection
We next examined the potential of Treg cells to limit the antiviral immune response and to promote virus persistence in a gain of function approach. For this purpose, we injected immune complexes (ic) comprised of recombinant IL-2 and the anti-IL-2 antibody JES6-1 [42], to selectively expand the subset of Foxp3 + CD4 + Treg cells in vivo ( Figure 5A). In naïve mice, three injections of IL-2ic drastically expanded the Treg population to represent 40-50% of all CD4 + T cells in blood, spleen and liver within 5 days after the first injection ( Figure S1 and Figure 5B). Similarly, IL-2ic treatment triggered a pronounced expansion of Treg cells in mice infected with 2610 3 PFU LCMV-DOC, established stably elevated levels of Treg cells in blood for at least 30 days (Figure 5B), and thus appeared comparable to that observed during high dose LCMV-DOC infection. The Treg cell population of infected, IL-2ic-treated mice was fully comparable to that of untreated, infected mice, with respect to cell surface expression of FR4, GITR, CD103 and CD25 ( Figure 5C) as well as TCR-Vb profiles (data not shown). The IL-2ic-stimulated expansion of the Foxp3 + Treg cell population profoundly interfered with generation and maintenance of gp33-specific CD8 + T cells, CD62L downregulation, and their capacity to produce IFN-c and TNF-a as measured in spleen and liver at days 15, 30, and 65 post infection ( Figure 5D-G, and Figure S2), which is reminiscent of the state of exhaustion that usually coincides with viral persistence in high dose LCMV-DOC infection. Accordingly, this long lasting impairment of antiviral T cells in presence of the enhanced Treg cell expansion prevented IL-2ic-treated animals from controlling a low dose LCMV-DOC infection. While infectious virus was readily cleared from the blood and most organs in control mice within 15 days, IL-2ic-induced Treg cell expansion resulted in a failure to clear virus in spleens, livers, kidneys and lungs for more than 2 months ( Figure 5H-J). Notably, IL-2ic expanded Treg cells also impaired antiviral CD8 + T cell effector responses and viral clearance of low dose LCMV-WE infection, which is otherwise rapidly cleared irrespectively of the viral inoculate ( Figure S3).
Taken together, the experimental expansion of Treg cells recapitulated both the long lasting functional impairment of the antiviral T cell response and the viral persistence that characterize high-dose LCMV-DOC infection in the setting of a low dose inoculum, and thus emphasized the remarkable potency of Treg cells to facilitate persistent viral infections. While Sprent and colleagues have clearly shown that IL-2:IL-2mAb (JES6-1) primarily target high affinity IL-2R + Treg cells and have minimal effects on low affinity IL-2R + naïve and memory CD8 + T cells [42], we cannot completely rule out the possibility that the IL-2:IL-2mAb also target high affinity antiviral effector CD8 + T cells resulting in terminal differentiation and exhaustion. However, this scenario appears   [42], and the normal expansion of antiviral CD8 + T cells until day 7.
The above data demonstrate the differential regulation of the Treg cell population during viral infection by IL-2 and IL-21. To better understand counter-regulation of Treg cells by IL-2 and IL-21 in the absence of confounding virus dynamics, we delivered the IL-21 gene by hydrodynamic injection to IL-2ic treated mice in the absence of viral infection ( Figure 6). Indeed, IL-21 significantly inhibited IL-2ic driven expansion of Treg cells. Similar results were obtained by co-injection of an engineered IL-21-Fc fusion protein together with IL-2ic. Together, these data indicate that IL-21 interferes with IL-2 driven expansion of Treg cells to optimize antiviral effector T cell responses.

Discussion
T cell exhaustion represents a state of T cell dysfunction associated with clinically relevant diseases, such as persistent viral weight loss and (D) percentage of surviving mice for groups of 8 mice. (E) Expansion of virus-specific CD8 + T cells and cytokine production of splenic (F) CD8 + and (G) CD4 + T cells as detected with gp33-specific tetramers or intracellular cytokine staining after restimulation with specific peptide or PMA/Ionomycin.   infections or cancer. Even though the molecular signature of exhausted T cells has been characterized in detail at the functional and transcriptional level [32,34], we are only beginning to understand the immunological mechanisms that support or counteract the development of T cell exhaustion during chronic infections [35]. In this study we report two major findings that establish a pathway of T cell exhaustion mediated by Treg cells during viral infection, and indicate its modulation by both, the pathogen and the host. First, we show that a persistence-inducing virus triggers the massive proliferation of Treg cells between days 9-20 post infection and demonstrate the potential of Treg cells to promote T cell exhaustion and chronic infection. Second, we identify IL-21 as a crucial host factor that antagonizes this virusdriven expansion of the Treg cell population. Together, these results suggest enhanced Treg cell responses as a mechanism of immune evasion that could be therapeutically targeted with IL-21.

(H) Virus titers in blood (left panel) and organs (right panel) of individual mice at 15 dpi. Dotted lines indicate the detection limit (DTL). (J) Percentages of gp33-specific CD8 + T cells in spleen, liver, and lungs 10 dpi. Frequencies of splenic IFN-c-producing CD8 + T cells (K) and CD4 + T cells (L) after restimulation with specific peptide (left graph) or PMA/Ionomycin (right graph). (M) Virus titers in blood and
Treg cells are essential for immune homeostasis and T cell tolerance [8,9], yet their contribution to anti-infectious immune responses is poorly defined. In the setting of persistent viral infections that was investigated here, Treg cells appear to downregulate antiviral T cell responses, and thus prevent the potentially lethal immunopathology caused by prolonged immune activation in presence of highly replicating virus (Figure 3). Supporting this notion, clinical studies have associated high numbers of Treg cells to chronic infection with HIV [29], HCV [28,55], or HBV [56], whereas low Treg cell numbers have been reported for elite HIV controllers [57], which collectively suggests that Treg cells modulate the equilibrium between host immune response and persisting virus [27] [34]. Our study has now recapitulated these observations in a murine model of persistent viral infection and applied experimental manipulation of the Treg cell response to define the direct link between virus-induced Treg cell expansion and T cell exhaustion. Our initial observation that titrated doses of LCMV-DOC induce graded degrees of Treg cell proliferation and T cell exhaustion allowed us to assess the impact of such virusinduced Treg cell expansion under ''non-persisting conditions'' using the identical LCMV strain. Although the expansion of Treg cells with IL-2ic achieved very high Treg cell numbers, it is important to note that this level of Treg cell expansion was fully comparable to that induced with LCMV-DOC at high doses or in IL-21R 2/2 mice. The Treg cell responses triggered by LCMV-DOC in presence or absence of IL-2ic treatment did not phenotypically differ with respect to their cell surface marker expression and exhibited TCRb profiles similar to that described for LCMV clone 13 infection [58]. Thus, the IL-2ic-elicited Treg cell proliferation appeared to truly mimic the physiological Treg cell expansion during chronic LCMV-DOC infection. These experiments exposed two facets of the antiviral Treg cell response in LCMV-DOC infected mice that are central for our understanding of the role and potential of Treg cells within antiviral and anti-tumoral responses. First, the increased morbidity in Treg-depleted DEREG mice infected with LCMV-DOC clearly revealed the critical role of Treg cells for preventing lethal immunopathology caused by potent immune responses to persisting antigen. Second, both the depletion as well as the gain of function approach demonstrated that Treg cells primarily modulate the functionality (e.g. cytokine response, antiviral activity) of antiviral T cells rather than influencing their priming consistent with an earlier report showing compromised cytolysis but no defect in priming, proliferation and motility of regulated CTLs [59]. This finding is especially promising, since it implies that the primed but exhausted antiviral T cells present in chronically infected subjects could be therapeutically rescued by removal of Treg cell-mediated suppression.
Dynamics of regulatory and effector T cell populations in homeostasis and during an immune response are very sensitive to the availability of IL-2. Competition for IL-2 between effector and regulatory T cells has been suggested to control tolerance and immunity or the outcome of infectious disease. Associated with the rapid expansion of virus-specific CD8 + T cells in the early phase of infection between days 0 to 8, we observed a remarkable drop in Treg cells below naïve levels irrespective of the LCMV inocula and strain used. This has also been observed in other infections and suggested to be due to consumption of IL-2 by expanding CD8 + T cells and required for efficient clearance of the invader [47,48]. While this appears feasible, it should be noted that expansion of CD8 + T cells in the acute phase of LCMV infection is independent of IL-2 [60,61] and that hyper-proliferation of Treg cells between days 10-20 in chronic LCMV infection was associated with potent suppression of IL-2 production by CD8 + T cells ( Figure 1F, G) arguing that IL-2 availability does not sufficiently explain cross-regulation of effector CD8 + T cells and Treg cell proliferation in acute and chronic LCMV infection. In vitro experiments suggested that IL-21 could inhibit Treg cells by suppression of IL-2 production in CD4 + effector T cells [49]. However, our results in the WT:IL-21R 2/2 mixed BM chimeric mice demonstrate that IL-21 inhibits Treg cell expansion directly in a cell intrinsic manner ( Figure 2E-G). The finding that expansion of Treg cells induced by IL-2ic treatment was impaired by simultaneous (hydrodynamic) overexpression of the IL-21 gene ( Figure 6) further supports this conclusion. Thus, IL-2 and IL-21 exert opposing activities on Treg cells, while they cooperate in driving effector and memory T cell responses, which adds another level of complexity to theoretical and experimental models addressing the dynamics of Treg cells and effector T cells [45].
Amongst the known Treg cell effector molecules, IL-10 has been shown to support the functional impairment of T cell responses during chronic infection with LCMV clone 13 [36,37]. However, we were unable to detect any IL-10-producing Treg cells in LCMV-DOC infected mice ( Figure 2J). Furthermore, IL-10 blocking antibodies or genetic IL-10-deficiency did not prevent T cell exhaustion and viral clearance in response to LCMV-DOC, in contrast to LCMV clone 13 infection [62]. The exhausted T cells in chronically LCMV infected mice have been shown to upregulate expression of several co-inhibitory receptors, e.g. PD-1 and Tim3, which contribute to T cell exhaustion in the LCMV model [63,64] and in human HIV patients [65,66]. It will thus be important to test whether these pathways are involved in the Treg cell-induced T cell exhaustion described in our study. Though a detailed characterization is beyond the scope of the current analysis, the experiments described in this manuscript will provide the framework for further mechanistic studies.
IL-6 and IL-21 have similar activities and interact in the crossregulation of inducible Treg and Th17 cell development in vitro and in vivo depending on the experimental model [51,52,53,67,68]. Interestingly, while IL-6 has recently also been shown to be essential for viral control by enhancing follicular T helper cell responses at late stages of chronic infection [69], virus titers and Treg numbers were comparable in LCMV-DOC infected IL-6 2/2 and WT mice up to day 30 post infection ( Figure 2K-N) [69]. Therefore, during chronic LCMV-DOC infection, the inhibition of virus-induced Treg cell expansion is a distinct function of IL-21, which is not accompanied by elevated IL-17 production of CD4 + T cells [40].
The importance of cell-intrinsic IL-21R signaling for the maintenance of CD8 + T cell functionality has been well documented [39,40,41] and is considered as the primary effect of IL-21 promoting the immune control of chronic viral infections. However, the present report clearly demonstrates that besides this direct function on CD8 + T cells, IL-21 also efficiently restricts the virus-induced expansion of the Treg population in a cell intrinsic manner. It remains to be clarified to which extent the direct effects of IL-21 on CD8 + T cells and its indirect effects on CD8 + T cells via the inhibition of Treg cells differentially contribute to the overall protective function of IL-21 in chronic viral infections considering that we only achieved limited recovery of T cell functionality (e.g. regain of cytokine production) and improved viral clearance only in lung and liver but not spleen and kidney of IL-21R 2/2 mice by Treg depletion. It should be noted, however, that, for reasons unknown, a fraction of (GFP neg Foxp3 + ) Treg cells resisted depletion by diphtheria toxin during LCMV infection in both IL-21R 2/2 and WT mice, although Treg depletion in naive mice was almost complete (.98%). It remains to be investigated whether the undeletable GFP neg Foxp3 + cells can compensate for the deleted Treg cells or represent a subpopulation of Treg cells that is responsible for maintenance of regulatory activity in LCMV-infected DEREG mice.
Regardless, our findings suggest a dual importance of IL-21 for preventing T cell exhaustion during chronic viral infections, and demonstrate that IL-21 in addition to its known direct effects on antiviral T cells [39,40,41] also partially alleviates the suppressive activity of Treg cells. Notably, in a model of acute lung infection, it was recently demonstrated that IL-21R 2/2 mice are protected from fatal lung immunopathology induced by pneumonia virus [70]. It is tempting to speculate that IL-21 might aggravate immunopathology by suppression of Treg cells in this infection model.
In summary, our data support the concept of virus-induced Treg cell expansion as an active immune evasion strategy, and thus highlight a novel pathway by which viruses exploit regulatory mechanisms of the immune system to establish persistent infection.
In view of the relevance to human disease these results have direct therapeutic implications and suggest strategies that boost IL-21 signaling in T cells as novel treatment options for chronic viral infections and cancer.
Ethics statement: All animal experiments were approved by the local animal ethics committee (Kantonales Veterinä rsamt Zürich, licenses 217/2008 and 113/2012), and performed according to local guidelines (TschV, Zurich) and the Swiss animal protection law (TschG).

Cells and reagents
All cell lines were originally obtained from the American Tissue Culture Collection (ATCC). Chemicals were purchased from Sigma-Aldrich except were otherwise stated. PE-and APC-conjugated peptide-MHC class I tetramers (H-2D b /gp33-41) were generated as described [74] or kindly provided the NIH tetramer core facility. The LCMV-GP peptides gp33-41 (KAVYNFATM) and gp61-80 (GLNGPDIYKGVYQFKSVEFD) were bought from Mimotopes. The following antibodies (all eBioscience unless otherwise stated; clone names given in parentheses) were used for flow cytometry: FITC-

Manipulation of Treg cells
In depletion experiments, DEREG and WT control mice were i.p. injected with DT (Merck) diluted in PBS. After an initial dose of 200 ng DT, mice were treated with 100 ng DT every third day unless otherwise indicated. To boost Treg cells, mice received 3 daily i.p. injections of IL-2ic generated from carrier-free recombinant mouse IL-2 and anti-IL-2 mAb (JES6-1A12; both from eBioscience) as described [42].

Flow cytometry
Tetramer and antibody staining was performed on blood cells and single cell suspensions prepared from organs. Spleens and kidneys were passed through a 70 mM cell strainer to obtain single cell suspensions. Livers were first dissected into small pieces, and then passed through a cell strainer before lymphocytes were purified by Lympholyte M gradient centrifugation (Cedarlane Laboratories Ltd.). Blood samples were pretreated with red blood lysis buffer (155 mM NH 4 Cl, 10 mM KHCO 3 , 0.1 mM EDTA, pH 7) for 10 min at RT. Cells were incubated with anti-CD16/CD32 mAb (2.4G2) to block FccR. For surface staining, cells were incubated at RT with peptide MHC I tetramers in FACS buffer (FB; PBS containing 0.5% BSA) for 15 min followed by addition of the relevant surface antibodies and incubation for additional 20 min at 4uC. Cytokine-production by T cells was assessed using intracellular cytokine staining of single cell suspensions that had been stimulated in presence of 2 mg/ml Monensin with either 1 mM specific peptide or 100 ng/ml PMA and 1 mg/ml Ionomycin for 4 hours in vitro. The cells were surface-stained, fixed with 4% Formaldehyde in PBS and permeabilized with permeabilization buffer (FB containing 1% Saponin). Intracellular staining was then performed in permeabilization buffer at 4uC for 20 minutes. After 2 washes with permeabilization buffer, cells were resuspended in FB. All samples were acquired on a FACSCalibur with CellQuest software (both BD Biosciences) and analyzed using the FlowJo software (Tree Star Inc.).

Determination of virus titers
Blood samples were obtained from LCMV-infected mice at indicated times, diluted 5-fold in MEM (5% FCS) containing 50 U.I. of Liquemin (Drossapharm) and frozen. Organs were collected in 1 ml MEM (5% FCS) and smashed with a Tissue Lyser (Qiagen). Samples were stored at 280uC until further analysis by plaque forming assay [75].

Hydrodynamic gene delivery
Mouse IL-21 coding sequence was amplified by PCR and linked with hIgG1 Fc domain and cloned into pLIVE in vivo expression vector (Mirus Bio). Endotoxin-free plasmid DNA (100 mg) was injected i.v. in PBS in a volume equal to 10% body weight (0.1 ml/g) within 5 s. As a control, a hIgG1 expression vector was injected. To supplement IL21, mice received i.p. injections of 2 mg recombinant IL-21-hIgG1 fusion protein (kindly provided by Daniel Christ, Garvan Institute for Medical Research, Sydney, Australia) or PBS two times daily.

Statistical analysis
Data are shown as average 6SEM. Statistical analysis was performed with the unpaired two-tailed t-test (except for Fig. 1B-E) using the Prism 4.0 software (GraphPad Software). Differences were considered significant for p,0.05 and were denoted as *, p,0.05; **, p,0.01; ***, p,0.001.