|Home | About | Journals | Submit | Contact Us | Français|
During acute infections, a small population of effector CD8 T cells evades terminal differentiation and survives as long-lived memory T cells. We demonstrate that the transcriptional repressor Blimp-1 enhances the formation of terminally differentiated CD8 T cells during LCMV infection, and Blimp-1 deficiency promotes the acquisition of memory cell properties by effector cells. Blimp-1 expression is preferentially increased in terminally differentiated effector and “effector memory” (TEM) CD8 T cells, and gradually decays after infection as central memory (TCM) cells develop. Blimp-1-/- effector CD8 T cells show some reduction in effector molecule expression, but primarily develop into memory precursor cells that survive better, and more rapidly acquire several TCM attributes, including CD62L and IL-2 expression and enhanced proliferative responses. These results reveal a critical role for Blimp-1 in controlling terminal differentiation and suppressing memory cell developmental potential in effector CD8 T cells during viral infection.
Memory (M) T cells are a major component of long-term immunity because of their longevity and other unique properties that include the ability to maintain a high proliferative potential in order to expand robustly upon secondary infection, rapidly re-express cytotoxic proteins and cytokines upon restimulation, and sustain memory T cell homeostasis in the absence of infection by undergoing cytokine-dependent cell division (Kaech and Wherry, 2007). A small population of CD8+ effector (E) cytotoxic lymphocytes (CTLs) cells survive the E→M transition and acquire these hallmark memory properties, while the majority (~80-90%) of effector cells undergo a program of terminal differentiation in which they maintain effector functions but lose memory cell potential, longevity, and strong proliferative responses (Fearon et al., 2001; Klebanoff et al., 2006; Lanzavecchia and Sallusto, 2002). A major goal in the field of memory CD8 T cell biology has been to understand the factors that determine the balance between memory cell potential and terminal differentiation.
During some infections, surface protein expression can be used to distinguish the subpopulations of antigen-specific effector and memory cells that achieve a more terminally differentiated state from those that are more likely to acquire the memory cell properties outlined above. For example, during LCMV, Listeria, T. gondii, and MCMV infections, terminally differentiated short-lived effector cells that express high levels of the NK cell maker KLRG1 (killer cell lectin-like receptor G1) and low levels of IL-7Rα (referred to as IL-7R) have a significantly shorter lifespan and reduced proliferative capacity in response to the homeostatic cytokines IL-7 and IL-15 or secondary antigenic challenge compared to effector cells with a KLRG1lo IL-7Rhi phenotype (Huster et al., 2004; Joshi et al., 2007; Kaech et al., 2003; Sarkar et al., 2008; Schluns et al., 2000; Snyder et al., 2008; Voehringer et al., 2001; Wilson et al., 2008). In contrast, the KLRG1lo IL-7Rhi effector cells survive the E→M transition better, have a longer lifespan, and acquire the ability to proliferate well in response to both homeostatic cytokines and antigenic re-challenge (Hikono et al., 2007; Joshi et al., 2007; Sarkar et al., 2008). In addition, memory cells that express high amounts of CD62L (referred to as central memory, TCM), CD27, and CXCR3, which also produce IL-2 and express low levels of CD43 and cytotoxic molecules, are more “functionally mature” memory cells (i.e., cells with enhanced proliferative capacity and longevity). These cells accumulate and dominate the memory population with time in lymphoid tissues while end-stage cells with a CD62Llo (termed “effector memory,” TEM) KLRG1hi IL-7Rlo CD27lo CXCR3lo CD43hi phenotype typically decay, but are often better maintained in peripheral tissues such as the liver and lung (Bachmann et al., 2005; Badovinac and Harty, 2007; Hikono et al., 2007; Kaech et al., 2002; Wherry et al., 2003). Decreasing the duration or intensity of exposure of activated T cells to antigen and inflammation at the time of priming generates an effector cell population that shares many phenotypic and functional characteristics with a more mature memory CD8 T cell population (Badovinac et al., 2005; Badovinac et al., 2004; Badovinac and Harty, 2007; Cannarile et al., 2006b; Heffner and Fearon, 2007; Ichii et al., 2004; Intlekofer et al., 2008; Joshi et al., 2007; Sarkar et al., 2007). Some key signals involved in this process are the cytokines IFNγ and IL-12 and the transcriptional regulators T-bet, Id2, Bmi-1, Bcl6 and Bcl6b (Badovinac et al., 2005; Badovinac et al., 2004; Badovinac and Harty, 2007; Cannarile et al., 2006b; Heffner and Fearon, 2007; Ichii et al., 2004; Intlekofer et al., 2008; Joshi et al., 2007; Pearce and Shen, 2007; Sarkar et al., 2007; Takemoto et al., 2006).
To better understand memory cell development and maturation, we focused on B-lymphocyte induced maturation protein-1 (Blimp-1; Prdm1), a transcriptional repressor that is required for the terminal differentiation of B cells into plasma cells, as well as the terminal differentiation of other diverse cell types (Horsley et al., 2006; Magnusdottir et al., 2007; Martins and Calame, 2008; Ohinata et al., 2005; Shapiro-Shelef et al., 2003; Turner et al., 1994). Because it is a critical regulator of terminal differentiation, we hypothesized that Blimp-1 might positively regulate the formation of terminally differentiated CD8 T cell subsets and potentially negatively regulate the process of memory cell maturation. Recently, Blimp-1 was found to play an important role in T cell homeostasis (Kallies et al., 2006; Martins et al., 2006). Blimp-1-deficiency in T cells during development results in a lethal inflammatory disease associated with excess accumulation of peripheral CD44hi CD62Llo activated CD4 and CD8 T cells, and greater numbers of HSV-specific memory CD8 T cells form in these mice (Kallies et al., 2006; Martins et al., 2006). However, the precise role of Blimp-1 in the differentiation of antigen-specific effector and memory CD8 T cells could not be properly assessed in these studies because of the added complications of ongoing inflammatory disease and poor regulatory T cell function in mice lacking Blimp-1-/- expression in the entire T cell compartment. Another study demonstrated that Blimp-1 mRNA is preferentially expressed in the short-lived effector cell subset of LCMV antigen-specific CD8 T cells, although no functional role for Blimp-1 in CD8 T cell differentiation has been described (Intlekofer et al., 2007). In this study, we define a critical role for Blimp-1 in regulating the balance between terminal differentiation and memory cell development by showing that Blimp-1 not only controls the terminal differentiation of antigen-specific CD8 T cells, but also that it affects the rate at which they acquire hallmark memory cell properties.
Using “P14 chimeric” mice, which contain a small number (~5,000) of Thy1.1+ P14 CD8 T cells that recognize the DbGP33-41 epitope of LCMV, we previously compared the gene expression profiles of naïve and LCMV-specific effector and memory CD8 T cells using DNA microarrays (Wherry et al., 2007). This analysis showed that Blimp-1 mRNA was upregulated in day 8 effector CTLs compared to naïve CD8 T cells, but decayed as memory T cells formed between days 40-100 post infection (p.i.). Using quantitative real-time PCR (qRT-PCR), we observed a similar pattern of Blimp-1 mRNA levels in LCMV-specific P14 CD8 T cells (Figure 1A). To investigate Blimp-1 expression at the single-cell level, we infected BAC tg reporter mice that express YFP under the control of Blimp-1 regulatory elements with LCMV. While less than 1% of CD44lo CD62Lhi naïve CD8 T cells expressed Blimp-1:YFP, Blimp-1 was upregulated in ~80% of LCMV-specific effector CD8 T cells and then progressively declined thereafter, both in terms of the percent of cells expressing Blimp-1:YFP as well as the amount expressed on a per-cell basis (i.e. the median fluorescence intensity (MFI) of YFP) (Figure 1B). Together, these data showed that Blimp-1 mRNA was induced in most effector CD8 T cells during clonal expansion, but its expression gradually declined during the E→M transition as the memory T cell population functionally matured.
Next, we explored whether Blimp-1 expression was differentially regulated between different subsets of effector and memory CD8 T cells during LCMV infection. In agreement with previously published data, Blimp-1 mRNA, protein, and YFP reporter expression were increased ~2-4 fold in terminally differentiated KLRG1hi IL-7Rlo short-lived effector cells (termed SLECs) compared to KLRG1lo IL-7Rhi memory precursor effector cells (MPECs) (Figures 1C-F) (Intlekofer et al., 2007). Interestingly, heightened Blimp-1 expression continued to be maintained even in the memory population in KLRG1hi IL-7Rlo CD27lo CD62Llo (TEM) cells (Figures 1C-F). In contrast, the KLRG1lo IL-7Rhi and CD27hi CD62Lhi (TCM) subsets became preferentially enriched with Blimp-1:YFPlo cells between days 10-40 p.i. (Figure 1E, note red highlighting in quadrant 3-left panels and quadrant 2-right panels). These data show that Blimp-1 expression is extinguished in the population as functionally mature subsets of CD8 TCM cells that have increased longevity and proliferative responses to both antigen and homeostatic cytokines form (Hand et al., 2007; Hikono et al., 2007; Joshi et al., 2007; Wherry et al., 2003).
To investigate the function of Blimp-1 in effector and memory CD8 T cell development, we generated Blimp-1 “conditional knockout” (CKO) mice by crossing mice containing a loxP-flanked Prdm1 gene (Prdm1flox/flox) (Martins et al., 2006) to transgenic mice expressing the Cre recombinase under the control of the human Granzyme B promoter (GzB-cre) (Jacob and Baltimore, 1999) (GzB-cre+; Prdm1flox/flox animals = CKO, littermate controls - GzB-cre+; Prdm1+/+ or GzB-cre-; Prdm1flox/flox = WT). Unlike the mice in which Blimp-1 is deleted during thymic development (Kallies et al., 2006; Martins et al., 2006), in this system Blimp-1 is expressed normally until CD8 T cells are activated and uninfected CKO mice did not demonstrate abnormal T cell activation (data not shown). Additionally, deletion efficiency reaches ~85-95% of virus-specific CD8 T cells during LCMV infection (Figure S1) (Chappell et al., 2006; Maris et al., 2003).
Using these mice, we found that Blimp-1 deficiency significantly increased effector cell survival after infection. Eight days after LCMV infection, similar frequencies of LCMV-specific CD8 T cells were observed in the spleens of CKO and WT mice (Figure 2A). The CKO mice also generated slightly, but not significantly, increased numbers of LCMV-specific effector CD8 T cells in the spleen, similar numbers in several other tissues, and significantly more effector cells in the lymph nodes (LN) and bone marrow (BM) as assessed by staining for both MHC class I tetramers and intracellular IFNγ after in vitro stimulation (Figure 2A-C). Interestingly, the effector CD8 T cell death that normally follows viral clearance was substantially decreased in the CKO mice; initially ~5 fold more LCMV-specific memory CD8 T cells persisted in the spleen and other tissues (especially the LN and BM) of the CKO animals (Figures 2B,D). BrdU labeling studies during these time periods did not show more proliferation in CKO cells compared to WT cells, indicating that the CKO effector cells survived better than WT cells (data not shown). After 5 months, the number of LCMV-specific CKO CD8 T cells declined to that in WT mice (except in BM and LN), indicating that the initial enhancement in survival was overridden by other, Blimp-1-independent, homeostatic control mechanisms (Figure 2B,E).
Because Blimp-1 was preferentially expressed in more terminally differentiated subsets of effector and memory CD8 T cells, we hypothesized that it might play a role in the formation of these cells. Indeed, CKO mice formed profoundly fewer effector cells with a KLRG1hi IL-7Rlo (SLEC) phenotype and there was a corresponding increase in both the percentage and number of KLRG1lo IL-7Rhi (MPEC phenotype), CD27hi and CD62Lhi (TCM) effector CD8 T cells in the spleen, liver, and lungs (Figure 2F and data not shown). Normally, the WT population gradually accumulates CD62Lhi TCM cells, but interestingly, the CKO mice accumulated a larger percentage of TCM cells (and KLRG1lo IL-7Rhi cells) at a faster rate than WT mice (Figures 2F,G). Effector and memory CD8 T cells from the CKO animals had higher expression of CD95 (Fas) as previously described (Kallies et al., 2006), but similar levels of CD44, CD43, bcl-2, and CD122 compared to WT (data not shown). Thus, in the absence of Blimp-1, several phenotypic changes that normally occur during the E→M transition were accelerated and a more mature-looking memory CD8 T cell population with an increased number of TCM cells developed at earlier time points after infection.
In the GzB-cre system, Prdm1 is deleted in a proportion of granzyme B-expressing NK and virus-specific CD4 T cells (~40% and ~10%, respectively) (Maris et al., 2003). Therefore, to specifically isolate the Blimp-1 deficiency to CD8 T cells, we examined two additional experimental systems. First, Prdm1 was deleted in P14 CD8 T cells using CD4-cre, and small numbers of these cells were adoptively transferred into congenic B6 mice to generate CD4-cre WT and CD4-cre CKO P14 chimeric mice that were subsequently infected with LCMV. The results from these experiments confirmed many of the phenotypes observed in the GzB-cre system, indicating that the aforementioned phenotypes were intrinsic to the Blimp-1-/- CD8 T cells (Figure S2A). However, genetic incompatibilities prevented long-term studies in these animals due to deletion of donor P14 cells. Therefore, we generated an additional experimental system involving mixed bone marrow chimeras (BMC) whereby bone marrow from Ly5.1+ CD8α-/- animals was mixed in a 90:10 ratio with bone marrow from either Ly5.2+ GzB-cre-; Prdm1flox/flox (BMC WT) or Ly5.2+ GzB-cre+; Prdm1flox/flox (BMC CKO) and used to reconstitute lethally irradiated Ly5.1+ recipient mice. Two months later, these mice were infected with LCMV. The phenotypes of antigen-specific effector and memory CD8 T cells from these mice re-confirmed those found in the above systems (Figure S3A, data not shown). As in the GzB-cre system, the CKO effector cells from the mixed BMC also showed reduced cell death between days 7 and 35 p.i. compared to WT cells (Figure S3B).
To better understand the mechanisms by which Blimp-1 regulates the balance between terminal differentiation and maintenance of memory cell potential, we compared the gene expression profiles of WT and CKO LCMV-specific effector CD8 T cells using Illumina BeadChips. Of the ~24,000 genes analyzed, 128 genes were significantly differentially expressed, and of these, ~70% (87 genes) showed increased mRNA in the CKO effector CD8 T cells compared to the WT cells (Table 1 and Table S1), in agreement with Blimp-1 being a transcriptional repressor (Huang, 1994). To further analyze the gene expression profile of the CKO effector cells, we categorized several of the differentially expressed genes into distinct functional groups (Table I). Some potential Blimp-1 gene targets, such as Bcl6, Tcf7, Tbx21 (T-bet), and Kit were then confirmed using qRT-PCR or FACS (Figure 3). A number of the differentially expressed genes contain predicted Blimp-1 binding sites (BS), some are regulated by Blimp-1 in other cell types (Id2, Xcl1, and CD83), and others are known direct targets of Blimp-1 (Bcl6, Id3, and Prdm1 itself) (Table 1, Column E) (G.M., K.C. and M. Cimmino unpublished observations, Martins and Calame, 2008 and Gong and Malek, 2007). These data suggest that control of terminal differentiation by Blimp-1 may share common features across different cell types.
Lastly, we compared the Blimp-1 CKO effector cell gene signature to that of MPECs, as previously described by our lab using Affymetrix GeneChips (Joshi et al., 2007) (Table 1, compare columns A vs. D). Although the magnitude of the differential expression cannot be directly compared across the two different microarray platforms, similarity in terms of the relative increase or decrease in gene expression can be assessed. Of the 128 genes differentially expressed in Blimp-1 CKO effector cells, ~90% showed the same trend in expression in MPECs compared to SLECs (Table 1 and data not shown). Thus, it appears that Blimp-1 represses many of the genes that control memory CD8 T cell potential and the development of memory CD8 T cell properties.
Because the mRNA transcripts of certain effector molecules were downregulated in CKO vs. WT cells, we next sought to establish whether Blimp-1 deficiency might perturb CTL function during viral infection. First, we examined the ability of effector and memory WT and CKO CD8 T cells to produce IFNγ, TNFα and IL-2 after 5 hour in vitro stimulation with peptide (Figure 4A). The numbers of IFNγ-producing CD8 T cells corresponded closely to those obtained with MHC class I tetramer staining between the two groups (compare Figures 2A and and4A),4A), indicating no differences in the frequency of IFNγ+ virus-specific CD8 T cells. In addition, similar frequencies of TNFα-producing CD8 T cells were observed between the two groups (data not shown). However, a considerably larger percentage of CKO effector CD8 T cells produced IL-2 compared to WT cells (Figure 4A), which translated into an overall increase in the percentage of polyfunctional, triple cytokine-producing (IFNγ, TNFα, IL-2) cells in the CKO animals. Notably, though, the overall amount of IFNγ produced on a per-cell basis (MFI) was reduced in effector and memory CKO CD8 T cells (Figure 4A).
When the expression of granzyme B (GzB) protein was examined, we found that ~50-60% of the CKO effector CD8 T cells initially (day 5 p.i.) expressed GzB (compared to ~90% in WT effector cells) but this rapidly decayed to near undetectable levels by day 7 p.i. (Figure 4B). Despite this profound reduction in GzB expression, cytotoxic activity of CKO CD8 T cells was only marginally affected based on in vivo cytotoxicity assays, indicating that other cytotoxic pathways were unaffected by Blimp-1 deficiency (data not shown). The decreased GzB and increased IL-2 expression was also recapitulated in the P14 CD4-cre and BMC CKO systems (Figures S2B, S3A), although the depressed IFNγ staining was not, perhaps due to differences in the polyclonal TCR repertoire vs. that of the high affinity P14 TCR (data not shown). Despite the modest effects on CTL function in the GzB-cre CKO animals, the clearance of LCMV by plaque assay technique was slightly delayed by 1-2 days in the majority of CKO animals (Figure 4C and data not shown; note, virus was not detected in the spleen, liver or kidney of CKO mice at day 15 p.i.). Nonetheless, the CKO memory CD8 T cells were capable of providing a protective immune response, because LCMV-immune CKO mice cleared a 2° infection with a more virulent strain of LCMV (cl. 13) that normally causes chronic infection and cannot be cleared in the absence of functional memory CD8 T cells (as measured by serum viral titers at day 5 p.i., data not shown). Interestingly, restimulation of CKO memory CD8 T cells did not “reverse” their phenotypes because the 2° CKO effector CD8 T cells maintained a IL-7Rhi CD27hi CD62Lhi KLRG1lo GzBlo phenotype compared to WT cells (Figure 4D).
The ability to proliferate robustly in response to secondary infection and homeostatic cytokines is a hallmark property of memory cells (Bachmann et al., 2005; Badovinac and Harty, 2007; Hikono et al., 2007; Kaech et al., 2002; Wherry et al., 2003). Because Blimp-1 was preferentially expressed in memory T cell subsets that have relatively reduced proliferative responses, we first confirmed that Blimp-1 itself marked cells with reduced proliferative capacity by transferring equal numbers of sorted Blimp-1:YFP+ and Blimp-1:YFP- P14 memory T cells from the same chimeric mice into naïve congenic recipient mice that were subsequently infected with recombinant Listeria expressing GP33-41 (rLM33) and analyzing their numbers 7 days later (Figure 5A). As predicted, the Blimp-1:YFP+ cells expanded less than the Blimp-1:YFP- population, indicating that Blimp-1 expression directly corresponds to memory T cells with decreased proliferative capacity.
Next, we tested whether the absence of Blimp-1 also impacted the proliferative responsiveness of effector cells to antigen in vitro and in vivo. To do this, we first examined cell division by CFSE-dilution in purified effector cells from WT and CKO animals incubated in vitro with peptide for 48 hours. In these experiments, CKO CD8 T cells underwent more extensive cell division than WT cells with less than 15% of WT effector cells dividing compared to >50% of CKO effector cells (Figure 5B). To test if similar results were found in vivo, equal numbers of day 8 LCMV WT or Blimp-1-deficient P14 effector cells sorted from “chimeric” mice were transferred into new naïve congenic recipients that were then infected with recombinant vaccinia virus expressing GP33-41 (rVV33). On a per cell basis, the secondary expansion of Blimp-1-deficient effector cells was ~4 fold greater than that of the WT effector CD8 cells, indicating that Blimp-1 can suppress the ability of effector CD8 T cells to proliferate when restimulated with antigen (Figure 5C). However, when we performed similar experiments with the WT and CKO memory CD8 T cells found ~90 days p.i., their recall responses were not significantly different (Figure 5D). This result was not entirely unexpected since, by this time, WT and CKO memory CD8 T cells had more similar phenotypes (Figures 2F,G and and4A).4A). These results suggest that Blimp-1 acts to limit the proliferative potential of effector CD8 T cells as they terminally differentiate, but as Blimp-1 expression naturally decays over time, the WT memory CD8 T cells “catch up” to and attain similar proliferative responses to antigenic challenge as CKO cells.
To determine if Blimp-1 also regulated effector CD8 T cell proliferative responses to homeostatic cytokines, WT and CKO LCMV-specific effector CD8 T cells were CFSE-labeled and transferred into γ-irradiated mice, and donor cells were examined 6 days later in the spleen. In a third group of mice, WT memory CD8 T cells (from ~day 90 p.i.) were transferred for comparison. Similar to WT memory cells, CKO effector CD8 T cells proliferated more extensively than their WT counterparts (Figure 5E), even when the IL-7Rhi populations were compared directly (data not shown). In addition, larger numbers of donor CD8 T cells were recovered from animals that received CKO vs. WT effector cells (data not shown). Together, these data indicate that Blimp-1 antagonizes effector CD8 T cell proliferative potential in response to antigenic and cytokine stimulation, and that the decline of Blimp-1 expression in the E→M transition directly contributes to the progressive increase of TCM cells that acquire enhanced proliferative capacity and the ability to self-renew.
Formation of a long-lived functional memory CD8 T cell population is critical for protective immunity, and this population is maintained for extended periods of time by long-lived, non-senescent, self-renewing cells (Fearon et al., 2001; Kaech and Wherry, 2007; Lanzavecchia and Sallusto, 2002). How are T cells with these properties generated during an immune response? We propose that a small population of effector cells evades terminal differentiation during clonal expansion and acquires memory cell developmental potential. In this study, we asked whether Blimp-1, a transcriptional repressor known to regulate the terminal differentiation of diverse cell types, plays a functional role in regulating effector and memory CD8 T cell fate decisions (Horsley et al., 2006; Magnusdottir et al., 2007; Ohinata et al., 2005; Shapiro-Shelef et al., 2003). Data from our laboratory and others indicated that Blimp-1 is preferentially expressed in more terminally differentiated effector populations and TEM cells (Intlekofer et al., 2007), and our studies here demonstrate that effector CD8 T cells lacking Blimp-1 expression do not maintain normal expression of several cytolytic molecules, and develop almost entirely into memory cell precursors that more rapidly acquire a TCM cell phenotype and mature memory features such as enhanced survival, higher proliferative capacity, and increased IL-2 production. Thus, Blimp-1 suppresses the formation of both MPECs and TCM and, hence, its activity balances whether an activated CD8 T cell will terminally differentiate and acquire maximal CTL function or whether it will maintain memory cell potential.
In general, following many acute infections such as LCMV, Listeria, or Sendai virus, the memory CD8 T cell population evolves over time and slowly accumulates IL-7Rhi KLRG1lo CD62Lhi CD27hi TCM cells that display the functionally mature characteristics of enhanced survival and proliferation to antigen and homeostatic cytokines and increased IL-2 production compared to TEM (Bachmann et al., 2005; Hikono et al., 2007; Joshi et al., 2007; Pearce and Shen, 2007; Tripp et al., 1995; Wherry et al., 2003). In addition to demonstrating that this phenotypic and functional maturation of the memory population correlates with a loss of Blimp-1 expression, our studies suggest that Blimp-1 itself negatively regulates the acquisition of these mature memory features by effector cells. As suggested by our microarray data, which showed a high degree of similarity in global gene expression patterns between Blimp-1 deficient effector cells and natural memory precursor effector cells, several mechanisms probably account for the regulation of memory cell properties by Blimp-1. In terms of effector CD8 T cell survival, reduced cytotoxic molecule expression or increased expression of some protease inhibitors (such as Spi-6 and Spi2a) that block granzyme or cathepsin B activity have been proposed to enhance CTL survival (Ashton-Rickardt, 2005; Badovinac et al., 2000; Zhang et al., 2007; Zhang et al., 2006). Indeed, Blimp-1-/- CTLs express lower levels of granzyme B, granzyme A, fasL, and perforin and increased levels of Spi2a. In addition, CKO effector cell survival may also be enhanced by higher OX-40 expression because OX-40 signaling can augment memory T cell formation and its expression can be induced by IL-7 (Gaspal et al., 2005; Ruby et al., 2007). Finally, higher expression of lymph node homing receptors CD62L and CCR7 and retention in lymphoid organs may also contribute to the increased survival of the CKO cells by increasing their exposure to pro-survival cytokines at these sites (Link et al., 2007).
Another critical memory feature that we found to be regulated by Blimp-1 is proliferative capacity. Populations of memory cells that naturally express high levels of Blimp-1 have poorer proliferative responses to secondary recall challenge compared to populations with low Blimp-1 expression, and effector cells that genetically lack Blimp-1 expression proliferate more robustly to both antigen and homeostatic cytokines compared to their WT counterparts. In other cell types, Blimp-1 represses proliferative responses via direct transcriptional inhibition of the cell cycle regulator myc (Horsley et al., 2006; Lin et al., 1997). Although our microarray studies did not show a significant difference in Myc mRNA levels between WT and Blimp-1 CKO effector CD8 T cells, past analyses demonstrate an inverse relationship between Myc and Prdm1 expression in effector and memory cells, suggesting the possibility that these factors may interact in CD8 T cells (Wherry et al., 2007). A final feature of memory cells commonly observed is the ability to co-produce multiple cytokines (referred to as “polyfunctionality”), and we found that a greater percentage of Blimp-1 CKO effector and memory T cells were polyfunctional and co-produced IFNγ, TNFα, and IL-2 (albeit, average IFNγ expression was reduced in the GzB-cre CKO cells compared to WT cells, but no difference was observed in the CD4-cre P14 CKO cells). The increased IL-2 production is consistent with reports that Blimp-1 negatively regulates IL-2 expression in activated CD4 and CD8 T cells in vitro (Gong and Malek, 2007; Martins et al., 2008) and that the Il2 locus is a direct target of Blimp-1 repression in CD4 T cells (Martins et al., 2008).
One potential mechanism for the enhanced MPEC/TCM phenotype seen in the CKO effector cells could be that Blimp-1 normally represses other transcriptional regulators that promote memory cell development (and suppress terminal differentiation), and therefore in the absence of Blimp-1 these factors are expressed at higher levels. One such transcription factor (TF), a candidate from our microarray data and studies in B cells, is the proto-oncogene Bcl6, which is a direct target of Blimp-1 in both B and CD4 T cells (Cimmino et al., 2008; Shaffer et al., 2002). In germinal center B cells, Bcl6 and Blimp-1 mutually represses each other; Bcl6 inhibits Blimp-1 expression and promotes the formation of long-lived memory B cells (Martins and Calame, 2008; Shaffer et al., 2000; Toyama et al., 2002), while Blimp-1 inhibits Bcl6 expression and promotes the formation of short-lived and long-lived antibody secreting plasma cells (Shaffer et al., 2002; Shapiro-Shelef et al., 2003). In CD4 T cells, Blimp-1 and Bcl-6 also mutually repress one another with Bcl-6 enhancing Th1 differentiation and Blimp-1 promoting Th2 differentiation (Cimmino et al., 2008). In CD8 T cells, increased Bcl6 expression can augment both memory cell proliferative capacity as well as acquisition of a TCM phenotype (Ichii et al., 2002; Ichii et al., 2004). Mice deficient in Bcl6b, a close homologue of Bcl6, also demonstrate impaired secondary recall responses (Manders et al., 2005). Our gene expression studies found that Bcl6 is ~2-3 fold increased IL-7Rhi MPECs vs. IL-7Rlo SLECs (Table I and Figure 3A). Moreover, Bach2, another repressor of Blimp-1 in B-cells (Ochiai et al., 2006), is similarly upregulated ~2-4 fold in IL-7Rhi MPECs (Joshi et al., 2007). While it is not clear from these studies whether it is the level of Bcl6/Bcl6b or the level of Blimp-1 expression that is directly responsible for the phenotypes observed, taken together with our data these data clearly indicate that the Blimp-1/Bcl6 axis plays an important role in memory CD8 T cell maturation.
The results from our microarray study also lend insight into how Blimp-1 may interact with other transcriptional regulators that are known to promote terminal differentiation. In particular, the mRNAs of the TFs T-bet (Tbx21) and inhibitor of DNA binding 2 (Id2) were reduced in CKO effector cells compared to WT cells. Similar to CD8 T cells lacking Blimp-1, those lacking T-bet and Id2 form an increased frequency of IL-7Rhi and CD62Lhi cells (Cannarile et al., 2006a; Intlekofer et al., 2007; Joshi et al., 2007). Blimp-1 and T-bet appear to positively regulate each others’ expression in effector CD8 T cells because CD8 T cells lacking T-bet also produce less Blimp-1 mRNA (N.S. Joshi and S.M.K., personal communication, and (Intlekofer et al., 2005), although a different relationship may operate in activated CD4 T cells because T-bet levels are actually elevated in the absence of Blimp-1 (Cimmino et al., 2008). A recent study has also found that XBP-1, a critical regulator of plasma cell formation that is downstream of Blimp-1 activity in B cells, enhances the formation of terminally differentiated KLRG1hi effector CD8 T cells (Kamimura and Bevan, 2008).
Our study provides a deeper understanding of the transcriptional network that governs effector cell fate decisions, and further exploration of the downstream targets and upstream regulators of Blimp-1 will help to better define this genetic pathway. Indeed, little is known about the upstream signals that regulate Blimp-1 expression in T cells. In B cells, IL-2, IL-5, IL-6, IL-10 and IL-21 can induce Blimp-1 expression, and STAT3 and STAT5, two transcription factors downstream of some of these cytokines also regulate Blimp-1 and Bcl6 expression (Calame, 2008). In T cells, IL-2, IL-4 and IL-12 have been shown to induce Blimp-1 expression under certain in vitro conditions, but it is not yet known how this treatment relates to CD8 T cell differentiation in vivo during immune responses (Gong and Malek, 2007). Thus, more work is needed to identify the signals that modulate Blimp-1 expression in vivo in both T and B cells. Improved understanding of Blimp-1 regulation could lead to new types of vaccines and treatments that modulate Blimp-1 activity to minimize certain aspects of Blimp-1-mediated effector CD8 T cell terminal differentiation, yet maximize other aspects of memory CD8 T cell differentiation.
C57BL/6 (B6) mice were obtained from the NCI (Frederick, MD). Thy1.1+ P14 TCR tg mice and generation of “P14 chimeric mice” via transfer of 50,000 naïve P14 CD8 T cells (~5,000 cells engraft) into B6 mice have been described previously (Joshi et al., 2007). Rag1-/- animals were from the Jackson Laboratories (Bar Harbor, ME). Blimp-1:YFP BAC tg mice were either directly infected or crossed to the P14 strain to create Blimp-1:YFP P14 chimeric mice (see Supplementary Methods for construct details). Prdm1flox mice were crossed to either lck-cre or CD4-cre mice (Cimmino et al., 2008; Martins et al., 2006) or Granzyme B-cre mice (Jacob and Baltimore, 1999). Mice were infected with 2 × 105 pfu LCMV-Armstrong (i.p.), 2 × 106 pfu of LCMV clone 13 (i.v.), 2 × 104 cfu recombinant L. monocytogenes (i.v.) expressing the LCMV epitope GP33-41 (rLM33, a gift from Dr. H. Shen) (Kaech et al., 2003) or 2 × 104 pfu vaccinia virus expressing GP33-41 (i.p.; rVV33), kindly provided by Dr. R. Ahmed. Viral titers were measured by plaque assay (Wherry et al., 2003). All animal experiments were done with approved Institutional Animal Care and Use Committee protocols.
DNA microarray analysis was performed on four independent samples of day 8 WT or CKO effector cells FACS-purified based on CD8 and MHC Class I tetramers DbGP33-41 and DbNP396-402 binding. RNA was isolated using TRIzol (Invitrogen, Carlsbad, CA) followed by RNeasy mini elute Kit (Qiagen, Valencia, CA). For hybridization and analysis information as well as qRT-PCR protocols see Supplementary Methods. Protein lysates from 1 × 106 sorted day 9 p.i. KLRG1hi IL-7Rlo or KLRG1lo IL-7Rhi effector CD8 T cells were processed and resolved by SDS-PAGE as described previously (Hand et al., 2007). Blimp-1 was detected by Western blotting using rat anti-Blimp-1 antibody at 1:200 (clone 6D3, Santa Cruz Biotechnology Inc, Santa Cruz, CA) and expression was normalized to expression of Grp94 (Hand et al., 2007).
Standard two-tailed t-tests assuming normal variance were used for all statistical calculations. All error bars and variances represent standard error of the mean, and * on all graphs represents p < 0.05.
We thank the members of the Kaech lab, Drs. J. Craft, R. Medzhitov, A. Poholek and T. Bondar for helpful comments and suggestions. We also thank Dr. Frank Constantini for the kind gift of ROSA26:EYFP mice. This work was supported by grants from the Burroughs Wellcome Fund and the NIH RO1-AI066232 and R21-AI077075 (SMK), RO1-AI50659 and RO1-AI43576 (KC), and MSTP TG 5T32GM07205 (RLR).
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.