|Home | About | Journals | Submit | Contact Us | Français|
T cell exhaustion is common during chronic infections and can prevent optimal immunity. While recent studies have demonstrated the importance of inhibitory receptors and other pathways in T cell exhaustion, the underlying transcriptional mechanisms are unknown. Here, we define a role for the transcription factor Blimp-1 in CD8 T cell exhaustion during chronic viral infection. Blimp-1 repressed key aspects of normal memory CD8 T cell differentiation and promoted high expression of inhibitory receptors during chronic infection. These cardinal features of CD8 T cell exhaustion were corrected by conditionally deleting Blimp-1. While high expression of Blimp-1 fostered aspects of CD8 T cell exhaustion, haploinsufficiency indicated that moderate Blimp-1 expression sustained some effector function during chronic viral infection. Thus, we identify Blimp-1 as a transcriptional regulator of CD8 T cell exhaustion during chronic viral infection and propose that Blimp-1 acts as a transcriptional rheostat balancing effector function and T cell exhaustion.
Following acute viral infections, antigen-specific CD8 T cells can undergo a memory differentiation program that results in the development and maintenance of robust, functional memory CD8 T cells. Once a pathogen is cleared, memory CD8 T cell differentiation follows a characteristic differentiation program where a subset of IL-7Rαhi effector CD8 T cells persists into the memory pool, increases expression of IL-7Rα and lymphoid homing molecules such as CD62L and CCR7 and gains the ability to produce IL-2 (Kaech and Wherry, 2007; Williams and Bevan, 2007). These memory CD8 T cells also acquire several cardinal properties such as the ability to rapidly respond and reactivate effector functions upon antigen re-exposure, high proliferative potential and long-term antigen-independent maintenance via IL-7 and IL-15 driven self-renewal (Kaech and Wherry, 2007; Williams and Bevan, 2007).
In contrast to acute infections, during chronic viral infections virus-specific CD8 T cells undergo an altered pattern of differentiation and become exhausted. CD8 T cell exhaustion is a transcriptionally altered state of T cell differentiation distinct from functional effector or memory CD8 T cells (Wherry et al., 2007). While some persisting infections such as EBV and CMV induce functional T cells that control these infections in healthy people, many other chronic infections in animal models and humans, particularly those with sustained viral replication, are associated with T cell exhaustion (Shin and Wherry, 2007). In these situations, exhausted CD8 T cells undergo a hierarchical loss of function, ultimately resulting in virus-specific CD8 T cells with severely compromised effector function, and in some cases these cells are physically deleted (Fuller and Zajac, 2003; Wherry et al., 2003). In addition, unlike normal memory CD8 T cells generated following acute infections, during chronic infections virus-specific CD8 T cells remain IL-7Rαlo, IL-15Rβlo, CD62Llo and IL-2lo (Shin and Wherry, 2007). These virus-specific CD8 T cells have major defects in homeostatic proliferation and often become dependent on persisting antigen rather than IL-7 and IL-15 for maintenance (Shin et al., 2007).
Another cardinal feature of exhausted CD8 T cells is the sustained expression of multiple inhibitory receptors (Blackburn et al., 2009; Kaufmann et al., 2007; Nakamoto et al., 2008; Wherry et al., 2007). While the precise role of inhibitory receptors in the initiation of CD8 T cell exhaustion remains unclear, considerable evidence indicates that the expression of these receptors is important for regulating multiple functional aspects of CD8 T cell exhaustion once a chronic infection is established. The PD-1 pathway has received considerable attention because exhausted CD8 T cells have substantially higher expression of this molecule and blockade of this pathway in vivo re-invigorates antiviral CD8 T cell responses (Barber et al., 2006; Sharpe et al., 2007; Velu et al., 2009). However, additional pathways, including LAG-3, also have a major role in limiting the effectiveness of exhausted CD8 T cells during chronic viral infection (Blackburn et al., 2009). Other known or potential inhibitory pathways that have also been associated with T cell dysfunction during chronic viral infections include 2B4, CD160, PirB, GP49B, CTLA-4, and Tim3 (Blackburn et al., 2009; Jones et al., 2008; Kaufmann et al., 2007; Nakamoto et al., 2008; Wherry et al., 2007) and different inhibitory receptors may regulate distinct aspects of functional exhaustion (Blackburn et al., 2009). Thus, the factors controlling expression of inhibitory receptors could represent a fundamental control mechanism for T cell exhaustion.
Despite recent work on inhibitory receptors and on the transcriptional profiles of exhausted CD8 T cells (Wherry et al., 2007), the fundamental transcriptional mechanisms underlying this type of T cell dysfunction are unknown. In the present study, we have examined the role of the transcriptional repressor Blimp-1 in CD8 T cell exhaustion during chronic viral infection in mice. Blimp-1 is a zinc finger-containing transcriptional repressor perhaps best known for governing fate decisions in memory B cell differentiation. In the germinal center, expression of Blimp-1 promotes terminal differentiation of plasma cells while repressing the transcriptional program of memory B cells (Calame, 2006; Shaffer et al., 2002; Shapiro-Shelef et al., 2003). Blimp-1 also has been shown to regulate fate decisions and cellular differentiation in other hematopoietic and non-hematopoietic cells (John and Garrett-Sinha, 2008). Here, we show that Blimp-1 has an important role as a transcriptional regulator of CD8 T cell exhaustion. Elevated expression of Blimp-1 in virus-specific exhausted CD8 T cells during chronic viral infection was associated with repression of memory T cell differentiation and elevated expression of inhibitory receptors. Both of these features of CD8 T cell exhaustion were reversed when Blimp-1 was conditionally deleted. However, haploinsufficient mice controlled chronic viral infection better than either wt or full conditionally deficient mice. Thus, our studies indicate that Blimp-1 is a transcriptional regulator of CD8 T cell exhaustion and suggest a model where Blimp-1 is a transcriptional rheostat regulating the balance between CD8 T cell effector function when moderately expressed and CD8 T cell exhaustion when highly expressed.
To investigate the role of Blimp-1 in CD8 T cells during chronic viral infection, we used lymphocytic choriomeningitis virus (LCMV). LCMV infection of adult mice is a well-established model with which to examine virus-specific T cell differentiation during acute versus chronic infection. Infection with the Armstrong strain of LCMV (Arm) results in an acute infection with clearance of infectious virus by day 8−10 p.i. and generation of functional effector and memory T cells (Wherry et al., 2003). In contrast, infection with the clone 13 strain of LCMV results in a chronic infection with viremia for 2−3 months, long-term viral persistence in some tissues and T cell exhaustion (Ahmed et al., 1984; Wherry et al., 2003). To determine the kinetics of Blimp-1 expression in virus-specific CD8 T cells during acute versus chronic LCMV infection, we first used quantitative RT-PCR (qRT-PCR) (Fig 1a). DbGP33-specific CD8 T cells were sorted from the spleens of Arm or clone 13 infected mice on d8, 15 and 30 p.i. and Blimp-1 mRNA expression was examined (Fig 1a). At d8 p.i., Blimp-1 expression was upregulated to a similar degree in DbGP33-specific CD8 T cells from either Arm or clone 13 infected mice (Fig 1a). After the first week of infection, Blimp-1 mRNA decreased modestly in DbGP33-specific CD8 T cells from LCMV Arm infected mice. In contrast, Blimp-1 was highly upregulated in virus-specific CD8 T cells from clone 13 infected mice by d15 p.i. and remained highly expressed for at least one month (Fig 1a).
Blimp-1 expression in virus-specific CD8 T cells was also examined during acute versus chronic LCMV infection using a Blimp-1 YFP reporter mouse. After infection with either Arm or clone 13, YFP expression in the reporter mice was elevated in CD44hi CD8 T cells, but remained low in naïve (CD44lo) CD8 T cells (Fig 1b). DbGP33-tetramer+ CD8 T cells were almost uniformly YFP+ (i.e. Blimp-1+) at d8 p.i. (Fig 1b) and the MFI of YFP was similar in DbGP33-specific CD8 T cells between acute and chronic viral infection (Fig 1c). However, after d8 p.i., YFP MFI increased in DbGP33-specific CD8 T cells during chronic LCMV infection, but decreased after d8 following acute infection (Fig 1c). Blimp-1-driven YFP expression was not limited to a specific tissue microenvironment and was elevated in antigen-specific CD8 T cells in both lymphoid and non-lymphoid tissues during chronic viral infection (Fig 1d). There were also differences in Blimp-1 YFP expression in different epitope-specific CD8 T cell populations during chronic infections. DbNP396-specific CD8 T cells undergo exhaustion early during chronic infection and are often physically deleted (Fuller and Zajac, 2003; Wherry et al., 2003). These DbNP396-specific CD8 T cells expressed higher Blimp-1 YFP than the DbGP33-specific CD8 T cells that also become exhausted but persist during chronic infection (supplemental Fig 1). Thus, Blimp-1 expression was substantially higher in antigen-specific CD8 T cells during chronic viral infection than following acute viral infection. Moreover, the pattern of Blimp-1 expression suggested a correlation between Blimp-1 expression and T cell dysfunction and/or terminal differentiation.
Blimp-1 has recently been described to play a role in T cell activation and homeostasis (Kallies et al., 2006; Martins et al., 2006). Based on the substantial increase in Blimp-1 expression during chronic viral infection, we questioned whether Blimp-1 might have a particular relevance during CD8 T cell exhaustion. To examine this issue, we used conditional deletion of prdm1, the gene encoding Blimp-1. To avoid any potential complications due to the action of Blimp-1 in the early events of T cell activation, we crossed prdm1flox/flox mice (Shapiro-Shelef et al., 2003) to mice with cre recombinase expression driven by the human granzyme B promoter (gzmB-cre) (Jacob and Baltimore, 1999). Hereafter, we refer to these prdm1f/f × gzmB-cre mice as conditional knockout mice (CKO). To determine when gzmB-cre was active following infection, we crossed the gzmB-cre mice to Rosa26-f/stop/f-YFP mice, which have a floxed stop site upstream of the gene encoding YFP in the Rosa26 locus. Granzyme B is expressed in activated CD8 T cells 1−2 days post-activation (Chang et al., 2007) and accordingly, YFP fluorescence was present in activated CD25+ CD8 T cells by 3 days after LCMV clone 13 infection of gzmB-Cre × Rosa26-f/stop/f-YFP mice (Fig 2a). By d6 p.i., YFP was expressed by the majority of activated, CD44hi and tetramer+ CD8 T cells, indicating cre-mediated recombination had occurred in most responding CD8 T cells at this time (Fig 2a, data not shown), though in the prdm1f/f × gzmB-cre mice it remains possible that some Blimp-1 mRNA and/or protein could persist temporarily following gene deletion.
Blimp-1 CKO and wildtype (wt) littermates were infected with LCMV clone 13 and viral control and T cell responses examined. Through the first ~20−30 days of infection, viral load in the serum and tissues was similar between wt and CKO mice (Fig 2b). The frequency of antigen-specific CD8 T cells in the blood was also similar between the Blimp-1 CKO and wt groups, although the contraction phase appeared delayed in the Blimp-1 CKO mice (Fig 2c). Spleens of CKO mice tended to contain higher absolute numbers of CD8 T cells specific for three different LCMV epitopes compared to wt mice at d30 p.i. (Fig 2c). There were also more DbGP33-specific CD8 T cells in the inguinal lymph nodes of the Blimp-1 CKO mice, but similar numbers in the bone marrow and liver of CKO and wt mice (Fig 2c). Among the different virus-specific CD8 T cell populations examined, we observed a larger increase in the number DbNP396-specific CD8 T cells in the CKO mice over wt (3-fold) compared to DbGP33 or DbGP276-specific CD8 T cell populations (2.2-fold and 1.8-fold, respectively), suggesting that deletion of Blimp-1 had the greatest impact on the accumulation of the most severely exhausted CD8 T cells (Fig 2c).
The phenotype of Blimp-1 CKO antigen-specific CD8 T cells was distinct from wt antigen-specific CD8 T cells. Many of the Blimp-1 CKO antigen-specific CD8 T cells expressed CD127 and CD62L, two molecules associated with memory CD8 T cell differentiation following acute infections that are not normally expressed by exhausted CD8 T cells, (Fig 2d). Even DbNP396-specific CD8 T cells expressed CD127 and CD62L in the absence of Blimp-1 (supplemental Fig 2a). Despite the expression of some memory-like markers by Blimp-1 deficient CD8 T cells, the Blimp-1 CKO CD8 T cells had only modest changes in cytokine production. IFNγ production was similar between CKO and wt CD8 T cells and MIP-1α was slightly reduced in the CKO CD8 T cells (Fig 2e). While production of TNF was slightly elevated in the Blimp-1 deficient T cells, both DbGP33-specific and DbNP396-specific CD8 T cells from the CKO mice were still considerably less functional than memory CD8 T cells from immune mice, in which 80−95% of the antigen-specific CD8 T cells co-produced IFNγ and TNF (Fig 2e, supplemental Fig 2) (Wherry et al., 2003). In agreement with previous reports (Martins et al., 2008; Martins et al., 2006), IL-2 production was elevated in the Blimp-1 CKO antigen-specific CD8 T cells to levels that were comparable to memory CD8 T cells after acute viral infection (Fig 2e, supplemental Fig 2). Thus, loss of Blimp-1 in antigen-specific CD8 T cells using gzmB-cre did not alter the initial viral load during chronic infection nor did it substantially improve effector cytokine production. Conditional deletion of Blimp-1 did, however, result in a trend toward increased numbers of antigen-specific CD8 T cells, restored some key aspects of normal memory CD8 T cell differentiation, including CD127 and CD62L expression, and led to the partial restoration of antigen-specific CD8 T cell populations that were otherwise terminally differentiated and deleted during chronic viral infection (e.g. DbNP396).
We next examined whether there was any relationship between the expression of inhibitory receptors and Blimp-1 during chronic LCMV infection. We have previously found that two subsets of exhausted CD8 T cells can be identified during chronic LCMV infection. One subset expresses intermediate amounts of PD-1 (PD-1int) and is capable of ‘revival’ upon PD-1 pathway blockade, while the other expresses high amounts of PD-1 (PD-1hi) and is more terminally differentiated (Blackburn et al., 2008; Nakamoto et al., 2008); similar observations have been made during HCV infection in humans (Nakamoto et al., 2008). We sorted virus-specific exhausted PD-1int and PD-1hi CD8 T cell subsets from chronically infected mice and assessed Blimp-1 mRNA expression by qRT-PCR (Fig 3a). The PD-1hi subset had 2−3 fold higher Blimp-1 mRNA than the PD-1int subset, suggesting a potential role for Blimp-1 in the regulation of PD-1 expression, CD8 T cell exhaustion, and/or terminal differentiation (Fig 3a).
To examine the relationship between Blimp-1 and inhibitory receptor expression in more detail, Blimp-1 YFP mice were infected with LCMV clone 13 and on d30 p.i., DbGP33 tetramer+ CD8 T cells were examined. Blimp-1 driven YFP expression in DbGP33+ CD8 T cells correlated with PD-1 expression in multiple tissues, but not with other markers such as CD127 (supplemental Fig 3). Next, we gated on exhausted DbGP33+ CD8 T cells that expressed high versus intermediate/low expression of PD-1, LAG-3, 2B4, or CD160 (Fig 3b). The inhibitory receptorhi subsets consistently had a higher Blimp-1 YFP MFI than the inhibitory receptorint/lo subsets regardless of which inhibitory receptor was examined (Fig 3b). This difference was greatest for 2B4, with a 7.8-fold difference in Blimp-1 YFP MFI in 2B4hi versus 2B4lo exhausted CD8 T cells. There was also a robust difference in Blimp-1 YFP expression in PD-1hi versus PD-1lo (3.3-fold), LAG-3hi versus LAG-3lo (3.3-fold) and CD160hi versus CD160lo (1.6-fold). Furthermore, using multiparameter flow cytometry to examine co-expression of 3 inhibitory receptors (PD-1, 2B4 and LAG-3), it was clear that higher Blimp-1 expression corresponded to an increase in the number of inhibitory receptors co-expressed by the same cell (Fig 3c).
Given the correlation between Blimp-1 YFP expression and inhibitory receptor upregulation we next examined inhibitory receptor expression in Blimp-1 CKO mice during chronic LCMV infection. On d30 p.i., DbGP33-specific CD8 T cells from CKO mice expressed considerably less PD-1 and LAG-3, and little CD160 or 2B4 compared to DbGP33+ + CD8 T cells from wt mice (Fig 3d). DbNP396-specific CD8 T cells also exhibited a decrease in PD-1 expression in CKO mice as compared to wt mice (supplemental Fig 2c). In addition, while nearly half of the DbGP33-specific CD8 T cells from wt mice co-expressed all four inhibitory receptors, only a small fraction of the Blimp-1 CKO DbGP33+ CD8 T cells expressed all four of these molecules simultaneously and more than half of the CKO cells expressed zero or only one inhibitory receptor (Fig 3d). This difference in expression of multiple inhibitory receptors in the CKO mice was not restricted to the DbGP33 specific CD8 T cell population. The overall population of CD44hi CD8 T cells from chronically infected wt mice, which contains other LCMV specific CD8 T cell populations, had higher expression of PD-1, LAG-3, 2B4 and CD160, compared to CD44hi CD8 T cells from Blimp-1 CKO mice (Fig 3d). Together, these data suggest an important role for Blimp-1 in regulating the elevated expression and co-expression pattern of inhibitory receptors on exhausted CD8 T cells.
The decrease in inhibitory receptor expression might be expected to improve viral control in the CKO mice. However, Blimp-1 is known to regulate expression of granzyme B (Gong and Malek, 2007) and is involved in generating terminally differentiated (but functional) effector CD8 cells following acute infection (Rutishauser et al. submitted and Kallies et al, submitted, accompanying manuscripts). While the wt mice controlled viremia by ~d60 p.i., Blimp-1 CKO mice continued to have high viral titers in the serum 2 months p.i. despite reduced inhibitory receptor expression (Fig 4a). In addition to wt and CKO mice, we also infected mice with only one intact copy of the Blimp-1 gene (prdm1f/+ × gzmB-Cre; conditional hets). Conditional haploinsufficiency resulted in mice that controlled viremia more rapidly than full CKO mice (Fig 4a). Unexpectedly, these conditional het mice also controlled viremia faster than the wt mice and had lower viral load in some tissues than either CKO or wt mice (Fig 4a). We next tested whether deletion of only one copy of prdm1 was sufficient to impact expression of inhibitory receptors. Indeed, these conditional het mice, like the CKO mice, had significantly lower MFI of PD-1 compared to virus-specific CD8 T cells from the wt mice at d15 p.i., a time point when viral load was similar in all three sets of mice (Fig 4b and c). Thus, while mice completely deficient in Blimp-1 had poor long-term viral control compared to wt mice, conditional het mice had lower PD-1 expression and controlled chronic infection more rapidly than wt mice.
We next examined the functional properties of virus-specific CD8 T cells from wt, conditional het, and CKO mice. The ability of antigen-specific CD8 T cells from the conditional het mice to co-produce IFNγ and TNF was not dramatically different from wt or CKO cells, though the MFI of IFNγ was higher in WT and conditional hets compared to CKO cells (Fig 5a, b). Co-production of IFNγ and MIP-1α was slightly increased in conditional het compared to CKO CD8 T cells, but conditional het and wt CD8 T cells were similar in this regard (Fig 5a). IL-2 production was high only in the CKO mice, suggesting that even one copy of prdm1 was sufficient to repress IL-2 during chronic infection (Fig 5a). Thus, production of antiviral cytokines by exhausted CD8 T cells was similar regardless of the number of copies of prdm1.
Blimp-1 expression could be particularly relevant for cytotoxic activity since Blimp-1 has been implicated in the expression of granzyme B (Gong and Malek, 2007). Virus-specific CD8 T cells from wt, conditional het, and CKO mice were equally capable of degranulating based on CD107a surface expression following peptide stimulation (Fig 5b). However, Blimp-1 CKO mice were severely deficient in expression of granzyme B compared to both conditional het and wt mice (Fig 5c). The ability of antigen-specific CD8 T cells from CKO mice to lyse peptide-coated targets was also substantially impaired by d8 p.i. and was essentially absent by d30 p.i., while killing was similar by CD8 T cells from the conditional het and wt mice (Fig 5d). Normally, during chronic LCMV infection cytotoxicity declines as the infection progresses (Wherry et al., 2003). Indeed, the level of killing by wt and CKO cells decreased by d30, but this residual killing was still higher than that observed in the complete absence of Blimp-1 (Fig. 5d). In contrast, LCMV Arm immune mice sustain the ability to robustly reactivate killing activity (supplemental Fig. 4) (Barber et al., 2003). These observations suggest that while high expression of Blimp-1 was detrimental to CD8 T cell responses and was associated with substantial upregulation of inhibitory receptors, retention of one intact copy of prdm1 and likely intermediate expression of Blimp-1 was required to induce and/or maintain cytotoxic potential. Thus, Blimp-1 conditional deficiency indicated a key role for this transcription factor in controlling some aspects of T cell exhaustion (increased inhibitory receptors, repression of memory T cell differentiation), but complete loss of Blimp-1 also compromised the low cytotoxic activity found in wt or conditional Blimp-1 het CD8 T cells during chronic LCMV infection.
Blimp-1 appears to be a good candidate for transcriptional control of some aspects of CD8 T cell exhaustion during chronic viral infection. The data described above, however, do not rule out the possibility that Blimp-1 deficiency in other cell types, or changes in environmental factors such viral load and/or pathogenesis of infection, could have an impact on antigen-specific CD8 T cells. To examine the cell-intrinsic role of Blimp-1 in CD8 T cell exhaustion we generated mixed bone marrow (BM) chimeras by injecting equal numbers of T- and B-cell depleted bone marrow cells from Ly5.2+ Blimp-1 CKO and Ly5.1+ wt animals into lethally irradiated Ly5.1+ recipient mice (Fig 6a). After reconstitution, the mixed BM chimeras were infected with LCMV clone 13 and analyzed 4 weeks after infection (Fig 6a).
At one month p.i., the differentiation state and function of virus-specific wt and CKO CD8 T cells from viremic (data not shown) mixed BM chimeras was examined. As observed in separate wt and CKO mice, a higher proportion of Blimp-1 deficient antigen-specific CD8 T cells in the mixed chimeras expressed CD127 and CD62L compared to the wt antigen-specific CD8 T cells (Fig 6b). In addition, the functional profiles of wt and CKO cells in the mixed chimeras paralleled observations in individual wt and CKO mice. Production of IL-2 and TNF were elevated, and MIP-1α was slightly decreased in Blimp-1 CKO CD8 T cells compared to wt CD8 T cells (Fig 6c). In the mixed BM chimeras, the virus-specific CKO cells also had defective granzyme B expression, while the wt cells sustained expression of this protein (Fig 6d). Finally, CKO antigen-specific CD8 T cells in the mixed BM chimeras expressed lower amounts of inhibitory receptors compared to wt antigen-specific CD8 T cells in the same animals (Fig 6e). Although we did not observe as drastic a difference in LAG-3 expression in the mixed chimeras as we did in individual mice, PD-1 expression was decreased in Blimp-1 deficient antigen-specific CD8 T cells compared to wt antigen-specific CD8 T cells in the same host (Fig 6e). The CD160hi and 2B4hi subsets present in wt virus-specific CD8 T cells were also dramatically reduced or absent from the CKO antigen-specific CD8 T cell populations (Fig 6e). The pattern of co-expression of multiple inhibitory markers was also clearly distinct for CKO and wt DbGP33-specific CD8 T cells in the mixed chimeras. Whereas less than half of the CKO DbGP33-specific CD8 T cells co-expressed 3 or more inhibitory receptors, nearly two thirds of the wt cells in the same mice co-expressed 3 or more inhibitory receptors (Fig 6e). In parallel experiments, we also generated mixed chimeras using BM from wt and prdm1f/f × CD4-cre mice and observed a similar impact on the pattern of memory T cell differentiation, function (data not shown) and expression of inhibitory receptors (Fig 6f). This difference in inhibitory receptor expression between the CD4-cre CKO CD8 T cells versus wt CD8 T cells was apparent in both the expression level of individual inhibitory markers as well as co-expression patterns (Fig 6f). These observations suggest that the changes observed in the gzmB-cre CKO CD8 T cells were not due to the timing of cre-mediated recombination or deletion of Blimp-1 in granzyme B-expressing non-CD8 T cells. Thus, the mixed BM chimeras indicate that Blimp-1 has a cell intrinsic role in regulating inhibitory receptor expression and other central features of T cell exhaustion, and identify Blimp-1 as a transcriptional regulator of CD8 T cell exhaustion during chronic viral infections.
While several transcription factors have been shown to regulate effector and memory T cell differentiation following acute infections, the transcriptional mechanisms of T cell exhaustion during chronic infections have remained unclear. In this study, we identify an important role for the transcriptional repressor Blimp-1 in regulating several defining features of CD8 T cell exhaustion during chronic viral infection. Exhausted CD8 T cells had substantially higher expression of Blimp-1 compared to functional effector or memory CD8 T cells generated following acute infection. This higher expression of Blimp-1 correlated with upregulated inhibitory receptor expression and repression of memory T cell properties. Conversely, conditional deletion of Blimp-1 resulted in reduced inhibitory receptor expression by exhausted CD8 T cells and demonstrated the central involvement of Blimp-1 in preventing normal memory T cell differentiation (i.e. CD127 and CD62L expression and IL-2 production) during chronic viral infection. These studies, however, also indicated that while high expression of Blimp-1 was associated with key aspects of T cell exhaustion, some Blimp-1 was essential for sustained T cell function including cytolysis, which was in turn necessary for eventual control of chronic viral infection. Thus, Blimp-1 appears to act as a transcriptional rheostat controlling T cell functionality at low amounts and T cell exhaustion when highly expressed.
CD8 T cell exhaustion is a common feature of many chronic viral infections in both animal models as well as in humans and is a likely reason for poor control of infection in these settings (Shin and Wherry, 2007). For example, while T cell exhaustion has been perhaps most extensively studied during LCMV infection in mice, T cell dysfunction, including expression of inhibitory receptors, occurs during infections such as SIV, HIV, HBV and HCV (Shin and Wherry, 2007). Thus, there has been considerable interest in defining the molecular mechanisms of T cell exhaustion. High and sustained expression of PD-1 has emerged as a hallmark of T cell exhaustion, and blocking the PD-1:PD-L pathway can re-invigorate immune responses during persisting infections (Freeman et al., 2006; Sharpe et al., 2007). In the current study we found that Blimp-1 expression was 2−3 times higher in the more terminally differentiated PD-1hi subset of exhausted CD8 T cells compared to the PD-1int/lo subset, which can be ‘revived’ by antibody blockade (Blackburn et al., 2008; Nakamoto et al., 2008). Other inhibitory receptors, including LAG-3, 2B4 and CD160, are also upregulated by exhausted CD8 T cells and these pathways cooperate to negatively regulate CD8 T cell responses during chronic viral infection (Blackburn et al., 2009; Wherry et al., 2007). Blimp-1 appeared to cell-intrinsically regulate expression of these additional inhibitory receptors since the absence of Blimp-1 resulted in reduced expression of PD-1, LAG-3, 2B4 and CD160 by virus-specific CD8 T cells during chronic infection. While the expression of these different inhibitory receptors has become emblematic of functional exhaustion, it should be noted that inhibitory receptors are unlikely to be the only regulatory pathways involved. Key roles for elevated IL-10 and the loss of IL-21 have been described (Brooks et al., 2006; Ejrnaes et al., 2006; Elsaesser et al., 2009; Frohlich et al., 2009; Yi et al., 2009) and T cell exhaustion is associated with dramatic changes in global gene expression profiles (Wherry et al., 2007). It is interesting that despite lower expression of inhibitory receptors in the absence of Blimp-1, these CD8 T cells remained poor cytokine producers. One possibility is that other negative regulatory pathways compensate in this setting. Alternatively, Blimp-1 could positively regulate expression of antiviral cytokines, either directly or indirectly. A third possibility is that Blimp-1 controls one “module” of the transcriptional program of T cell exhaustion which includes inhibitory receptor expression and memory repression. While inhibitory receptors are clearly linked to T cell function, it is possible that another layer of transcriptional control also influences expression of antiviral cytokines. Such an idea of overlapping transcriptional modules is emerging for Foxp3+ Tregs (Koch et al., 2009; Zheng et al., 2009). In vivo reversal of exhaustion by inhibitory receptor blockade is almost always accompanied by extensive proliferation (Barber et al., 2006; Sharpe et al., 2007; Velu et al., 2009) and this cellular division could be associated with selective expansion (Blackburn et al., 2008) as well as transcriptional reprogramming. Future studies are necessary to dissect additional transcriptional pathways associated with T cell exhaustion and Blimp-1-independent regulation of cytokine production. It will also be important to compare functional changes in exhausted CD8 T cells that occur following antibody-mediated inhibitory receptor blockade versus temporal deletion of Blimp-1.
Blimp-1 regulates cell fate decisions and differentiation in a variety of settings. In the germinal center, Blimp-1 controls the terminal differentiation of B cells into long-lived plasma cells while repressing the development of memory B cells (Calame, 2006). Blimp-1 also has a key role in determining cell fate for primordial germ cells (Hayashi et al., 2007; Ohinata et al., 2005), in the sebaceous gland (Horsley et al., 2006), in developing zebrafish (Roy and Ng, 2004) and endomesodermal differentiation in Xenopus (de Souza et al., 1999). Following activation during an acute infection, antigen-specific effector CD8 T cells can adopt one of two fates: terminally differentiated effector CD8 T cells or memory CD8 T cell precursors (Chang et al., 2007; Joshi et al., 2007; Sarkar et al., 2008). Tbet and Blimp-1 have been identified as transcription factors influencing the differentiation of terminally differentiated effector CD8 T cells versus memory precursors (Intlekofer et al., 2007; Joshi et al., 2007; Sarkar et al., 2008), and other transcription factors such as Eomesodermin (Intlekofer et al., 2005), Id2 (Cannarile et al., 2006), Bcl6 and Bcl6b (Ichii et al., 2004; Manders et al., 2005), and Bmi1 (Heffner and Fearon, 2007) have also been shown to regulate memory CD8 T cell differentiation after acute infection. During chronic infection, antigen-specific CD8 T cells differentiate into a population transcriptionally distinct from effector and memory CD8 T cells present following acute infection (Wherry et al., 2007). The role of Blimp-1 in other biological settings suggests that Blimp-1 could regulate the differentiation of exhausted CD8 T cells. During chronic LCMV infection, CD8 T cell exhaustion becomes established progressively after the effector phase (Wherry et al., 2003; Wherry et al., 2007). Important events in this progression, such as the appearance of the PD-1hi subset of exhausted CD8 T cells (Barber et al., 2006; Wherry et al., 2007), correspond to the further upregulation of Blimp-1 observed 2−4 weeks post infection. The upregulation of Blimp-1 in virus-specific CD8 T cells by 1 month of chronic viral infection therefore might represent a developmental switch that promotes some aspects of the transcriptional program of CD8 T cell exhaustion. It will be important to examine the role of Blimp-1 at multiple stages of CD8 T cell exhaustion to further dissect how fate decisions and differentiation of exhausted CD8 T cells occur.
The amount of Blimp-1 expressed also appeared to have a crucial impact on CD8 T cell differentiation and exhaustion. Mice with one intact copy of the prdm1 gene were not intermediate between wt and CKO mice, but rather achieved more efficient control of infection than wt or CKO mice. Many transcription factors function as transcriptional on/off switches (Louis and Becskei, 2002). In contrast, our data suggest that Blimp-1 could act as a molecular rheostat in CD8 T cells during chronic infection, mediating different cell fates or transcriptional events at different expression levels. Other transcription factors can function in a graded fashion and some molecules that possess this property include master regulators of cell fate and differentiation such as Nanog, Sox2 and Oct-3/4 (Rizzino, 2008). For example, small quantitative changes in expression of Oct-3/4 control three distinct cell fate decisions for embryonic stem cells (Niwa et al., 2000). The transcription factor PU.1 also acts in a graded manner during hematopoietic differentiation (Laslo et al., 2006). Accumulating evidence also points to the importance of quantitative changes in the expression of transcription factors such as Tbet and Eomes in memory CD8 T cell differentiation following acute infection (Intlekofer et al., 2007; Intlekofer et al., 2005; Joshi et al., 2007). Our data on both temporal expression patterns and full versus partial conditional deficiency suggest that low or intermediate expression of Blimp-1 is required for some effector functions and could be important in fate decisions between memory and terminal effector cell differentiation following acute infection, while the high expression of Blimp-1 that occurs during chronic viral infection promotes CD8 T cell exhaustion and represses memory differentiation.
Our studies also point to a critical role for sustained cytolytic potential during chronic viral infections. Recent work has demonstrated the importance of cytotoxicity in long-term control of chronic infections in humans (Appay et al., 2000; Migueles et al., 2002; Migueles et al., 2008; Trabattoni et al., 2004). While exhausted CD8 T cells are known to have partial defects in killing compared to highly functional effector or memory CD8 T cells (Wherry et al., 2003), some residual cytotoxicity by antigen-specific CD8 T cells can be maintained in vivo (Agnellini et al., 2007). Blimp-1 conditional het mice controlled virus substantially faster than either the wt or CKO mice, and this difference corresponded to sustained cytolysis in the conditional het mice compared to the CKO mice. However, conditional het mice controlled virus in vivo more rapidly than wt mice as well, despite similar killing and cytokine production in vitro. It is possible that lower PD-1 expression by conditional het mice may lead to improvements in effector function in vivo that are not obvious in vitro, as has been observed with blockade of other inhibitory pathways such as LAG-3 (Blackburn et al., 2009). Conditional het mice also had slightly higher total numbers of antigen-specific CD8 T cells than wt mice including more NP396-specific CD8 T cells (data not shown). Higher numbers of virus-specific CD8 T cells and improved effector functions in vivo by conditional het mice could account for this more efficient control of infection compared to wt mice, but future studies are necessary to investigate these issue further.
In summary, we have identified Blimp-1 as a transcriptional regulator of functional exhaustion and repressor of memory differentiation in CD8 T cells during chronic viral infection. These studies provide a framework to begin dissecting Blimp-1 targets, regulation of Blimp-1 activity and additional transcriptional pathways involved in T cell dysfunction during chronic infection. While there are clearly additional transcription factors and transcriptional pathways that contribute to T cell exhaustion, our results identify Blimp-1 as a transcriptional regulator of CD8 T cell exhaustion during chronic viral infection.
Four to six week old C57BL/6 or C57BL/6 Ly5.2CR (Ly5.1+) mice were purchased from NCI. Rosa26-f/stop/f-YFP mice were purchased from Jackson Laboratories. BAC transgenic Blimp-1 YFP reporter mice were from Eric Meffre (Yale University, New Haven, CT). Prdm1f/f mice were from Kathryn Calame (Columbia University, New York, NY), granzyme B-Cre mice were from Joshy Jacob (Emory University, Atlanta, GA) and CD4-Cre mice were from Steven Reiner (University of Pennsylvania, Philadelphia, PA). Mouse strains were crossed and mice were bred and maintained in the Wistar Institute and were used in accordance with IACUC guidelines. Mice were infected with 2×105 plaque forming units (PFU) of LCMV Armstrong (Arm) i.p. or 2×106 PFU LCMV clone-13 (Cl-13) i.v. as described (Wherry et al., 2003). Virus was grown and titered as described (Wherry et al., 2003).
Lymphocyte isolation from lymphoid and non-lymphoid tissues, surface stains, intracellular cytokine stains (ICS) and CD107 assay were performed as previously described (Barber et al., 2006; Wherry et al., 2003). All antibodies were purchased from Biolegend except for CD127, CD160, TNF, IL-2 (eBioscience), 2B4 (eBioscience, BD Biosciences) LAG-3 (AbD Serotec), granzyme B (Caltag) and MIP-1α (R&D Systems). LIVE/DEAD dead cell stain, CFSE, streptavidin-APC and streptavidin-Quantum dot 655 were purchased from Invitrogen. MHC class I peptide tetramers were made and used as described previously (Wherry et al., 2004; Wherry et al., 2003). All flow cytometry data was acquired on an LSRII (BD Biosciences) and analyzed by FlowJo (Treestar). Pie charts were created using the Pestle and SPICE programs (Mario Roederer; Vaccine Research Center, NIAID, NIH).
Bone marrow cells from Ly5.1+ wt and CKO (either prdm1f/f × gzmB-Cre or prdm1f/f × CD4-Cre) mice were T and B cell depleted using MACS magnetic beads (Miltenyi Biotec) and adoptively transferred at a 1:1 ratio into lethally irradiated (1000 rad) C57BL/6 Ly5.2Cr (Ly5.1+) mice. Mice were fed antibiotic for two weeks and allowed to reconstitute for eight weeks before use.
Cells were sorted on a FACSAria (BD Biosciences). RNA extraction was performed using Trizol (Invitrogen). cDNA was generated using the High Capacity cDNA Archive Kit (Applied Biosystems). Relative quantification real-time PCR was performed on an ABI Prism 7000 using primers purchased from Applied Biosystems. HPRT was used as an endogenous control, and gene expression was measured as fold-increase over naïve CD8 T cells.
CD8 T cells were purified with MACS magnetic beads (Miltenyi Biotec) from spleens of CKO, het and wt mice according to the manufacturer's protocol. Target cells were either labeled with GP33 peptide or SIINFEKL peptide and CFSE labeled at two different concentrations. Equal numbers of DbGP33+ CD8 T cells were plated at a 2:1 ratio with the labeled target cells and total cell numbers were normalized with naïve splenocytes. The cells were incubated at 37C for 16−20 hrs and specific lysis was calculated as described (Hermans et al., 2004).
We thank Wherry lab members for insightful comments and Susan Kaech and Rachel Rutishauser for sharing unpublished data and helpful discussions. This work was supported by NIH (AI071309 and HHSN26620050030C), the Grand Challenges in Global Health Initiative, The Dana Foundation, and The Ellison Medical Foundation.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.