Antigen-specific clonal expansion and cytolytic effector function of CD8⁺ T lymphocytes depend on the transcription factor Bcl11b

Bcl11b is expressed in mature CD4⁺ and CD8⁺ T cells (Fig. 1) in addition to thymocytes (Albu et al., 2007). Its removal with the CD4-Cre deleter caused a block in positive selection with marked reduction of T cells in the periphery, which presented developmental abnormalities (Albu et al., 2007), therefore impeding the study of Bcl11b role in mature T cells. To circumvent this problem, we crossed Bcl11b^F/F mice with the dLck (Lck distal) promoter–iCre mice, taking into account that the dLck promoter is active in mature T cells (Wildin et al., 1995; Zhang et al., 2005). We observed efficient removal of the floxed Bcl11b allele in CD8⁺ T cells, whereas the removal in CD4⁺ T lymphocytes was less efficient (Fig. S1), which is consistent with previous observations showing higher activity of the dLck-iCre in CD8⁺ T cells (Zhang et al., 2005; D’Souza et al., 2008). Quantitative evaluation by flow cytometry indicated that the floxed allele was removed in ∼90% of the CD8⁺ T cells (Fig. 1 A), which is similar to previous results for other genes (Zhang et al., 2005; D’Souza et al., 2008). Bcl11b was still present in DP thymocytes, as well as in CD4 and CD8 SP thymocytes of Bcl11b^F/F/dLck-iCre mice (Fig. 1 A and Fig. S1).

CD8 and CD4 SP thymocytes were present in Bcl11b^F/F/dLck-iCre mice in similar percentages to those of WT (Fig. 1 B), and the TCR-β levels were similar to those of the WT cells, indicating that these cells reached maturity (Fig. S2). There was a modest but significant reduction in the percentage and absolute numbers of CD8⁺ T cells in the lymph nodes and spleens of Bcl11b^F/F/dLck-iCre mice, whereas the reduction in the CD4⁺ T cell numbers was less remarkable (Fig. 1, C and D; and not depicted).

The efficient removal of BCL11B in the CD8⁺ T lymphocytes of Bcl11b^F/F/dLck-iCre mice, and the fact that these mice still have substantial numbers of CD8⁺ T cells in the periphery which are not affected by developmental defects, makes Bcl11b^F/F/dLck-iCre mice an appropriate model to study the CD8⁺ T cell immune response in the absence of Bcl11b. For this purpose, we crossed the Bcl11b^F/F/dLck-iCre mice onto an OT-1 background (Fig. 1 C), in which the majority of CD8⁺ T cells express a TCR (Vα2Vβ5) which recognizes the OVA peptide residues 257–264 presented in the context of H-2K(b) (Hogquist et al., 1994). We further used an adoptive transfer system and infection with Listeria monocytogenes expressing Ova (Lm-Ova; Shen et al., 1995) to evaluate the immune response. Specifically, we cotransferred equal numbers of naive CD62L^hiCD44^lo CD45.2/OT-1 Bcl11b-deficient and CD62L^hiCD44^lo CD45.1/2/OT-1 WT CD8⁺ T cells in CD45.1 recipient mice. After the transfer, we infected the recipient mice with Lm-Ova and evaluated the expansion of Bcl11b-deficient and WT CD8⁺ T cells, based on the CD45.1 and CD45.2 allelic differences. We found that CD45.1/2 WT CD8⁺ T cells expanded significantly, reaching a peak at day 7 of infection, and then declined in number during the contraction phase (Fig. 2, A and B). The peak of expansion for Bcl11b-deficient CD8⁺ T cells was also at day 7; however, the magnitude of the expansion was dramatically reduced (Fig. 2, A and B). In addition, the numbers of CD45.2 Bcl11b-deficient CD8⁺ T cells remained lower compared with the numbers of WT cells during the contraction phase of the immune response (Fig. 2, A and B). Similar results were observed when knockout and WT OT-1 CD8⁺ T cells were transferred separately in recipient mice (unpublished data), suggesting that the defect in expansion was not a result of the competition by WT cells.

To determine whether similar defects in expansion were also observed in a polyclonal CD8⁺ T cell response, we next investigated the ability of Bcl11b-deficient CD8⁺ T cells to expand in response to infection with influenza virus. In these experiments, irradiated CD45.1 recipient mice were reconstituted with Bcl11b^F/F/dLck-iCre (CD45.2) and WT (CD45.1/2) bone marrow cells, and the mixed chimeras were infected with a sublethal dose (1,000 egg infectious units [EIU]) of A/PR8/34 influenza 8 wk after reconstitution. The frequency of influenza nucleoprotein NP-specific, as well as influenza acidic polymerase (PA)–specific, CD8⁺ T cells was evaluated in mediastinal lymph nodes. The results show that the frequency of NP-specific or PA-specific Bcl11b-deficient CD8⁺ T cells was >10-fold decreased compared with WT (Fig. 2 C). These results show that even on a polyclonal background, Bcl11b-deficient CD8⁺ T cells have a poor expansion in response to infection, which is similar to the results obtained with the OT-1/Lm-Ova model.

To further investigate the cause of the reduced expansion of the Bcl11b-deficient CD8⁺ T cell, we first evaluated the incorporation of BrdU in transferred OT-1 cells on day 6 after the Lm-Ova infection to determine whether their Ag-specific proliferation was altered. We found that the percentage of CD8⁺ T cells that had incorporated BrdU was reduced in Bcl11b-deficient OT-1 CD8⁺ T cells compared with the WT OT-1 cells (Fig. 3 A). In addition, we evaluated the forward scatter of the CD8⁺ T cells as a measurement of T cell size, which increases after activation and before division (Grumont et al., 2004). Our results clearly indicate that the population with high forward scatter was reduced in the Bcl11b-deficient CD8⁺ T cells compared with WT CD8⁺ T cells (Fig. 3 B), which is consistent with fewer proliferating cells in this population.

To determine whether the reduction in the CD8⁺ T cell numbers during the course of infection was caused by enhanced cell death, we evaluated the apoptosis of Bcl11b-deficient and WT CD8⁺ T cells by measuring the levels of cleaved caspase-3. We found that a similar percentage of Bcl11b-deficient CD8⁺ T cells stained positively for cleaved caspase-3 compared with WT on days 6, 7, 9, or 12 after infection (Fig. 3 C and Fig. S3), demonstrating that the reduction in the number of Bcl11b-deficient CD8⁺ T cells during the course of infection is not a result of increased cell death.

To directly measure the proliferative response of the OT-1 CD8⁺ T cells to antigenic stimulation, we labeled Bcl11b-deficient and WT naive OT-1 CD8⁺ T cells with CFSE, incubated them with APCs loaded with OVA_257-264 peptide, and monitored their proliferation by dilution of CFSE. Bcl11b-deficient CD8⁺ T cells showed decreased proliferation in response to Ag-specific stimulation compared with WT (Fig. 4 A). In addition, the up-regulation of the early activation marker CD69, a measurement of TCR activation, was reduced in response to stimulation with OVA_257-264 peptide-loaded APCs (Fig. 4 B). We further investigated the ability of Bcl11b-deficient CD8⁺ T cells to proliferate in response to stimulation with α-CD3/CD28 antibodies and found that though the proliferation was substantially improved, it did not reach the same levels as in WT (Fig. 4 C). However, when the cells were stimulated with PMA/ionomycin, proliferation of Bcl11b-deficient CD8⁺ T cells reached the WT levels (Fig. 4 D). These results suggest that the reduced proliferation of Bcl11b-deficient CD8⁺ T cells is likely to be caused by altered TCR signaling, with a potential defect upstream of CD3 engagement, and a second defect upstream of PKC-Θ. The levels of both TCR and CD28 were equivalent in Bcl11b-deficient OT1 CD8⁺ T cells (unpublished data), although on a polyclonal background we observed a slight reduction in TCR and CD3 levels (unpublished data). Interestingly, the CD8 coreceptor levels were reduced in Bcl11b-deficient CD8⁺ T cells (Fig. 5 A and Table S1). The reduction in the CD8 coreceptor levels was not only observed in Bcl11b^F/F/dLck-iCre mice but also when Bcl11b was removed from CD8⁺ T cells with the use of the ER-Cre deleter in vivo (Fig. S4). We conducted quantitative (q) RT-PCR and demonstrated that CD8α and β coreceptor mRNAs were lower in the absence of Bcl11b (Fig. 5 B), suggesting that the reduction in the coreceptor levels is caused by a decrease in the coreceptor gene expression.

It has been demonstrated that the CD8 coreceptor plays a critical role in CD8⁺ T cell activation and immune response by amplifying the Ag-specific TCR response and Ag-specific proliferation (Veillette et al., 1989; Arcaro et al., 2001; Holler and Kranz, 2003). To further demonstrate that that the altered expression of CD8 coreceptor causes diminished TCR signaling, we activated the CD8⁺ T cells by treatment with α-TCR/CD8/CD28 and evaluated the phosphorylation of Zap70. Our results demonstrate that Zap70 phosphorylation was diminished after TCR/CD8 corecptor/CD28 engagement in Bcl11b-deficient CD8⁺ T cells (Fig. 5 C), suggesting that one of the causes of the reduced proliferation of Bcl11b-deficient CD8⁺ T cells is decreased CD8 coreceptor levels.

The reduced levels of CD8 coreceptor could not be the sole cause of the reduced proliferation of Bcl11b-deficient CD8⁺ T cells because stimulation with α-CD3/CD28 antibodies also resulted in reduced proliferation but substantially improved. Considering that PMA/ionomycin treatment brought the expansion of Bcl11b-deficient CD8⁺ T cells to WT levels (Fig. 4 D), we hypothesized that there might be a second defect in Bcl11b-deficient CD8⁺ T cells, upstream of PKC-Θ. We found that an additional gene with central role in TCR signaling presented altered mRNA levels in Bcl11b-deficient CD8⁺ T cells, specifically Plcγ1 (phospholipase C gamma 1; Wilde and Watson, 2001; Bonvini et al., 2003; microarray analysis, unpublished data), and demonstrated by qRT-PCR that the Plcγ1 mRNA levels were reduced in Bcl11b-deficient CD8⁺ T cells (Fig. 5 D). Calcium flux was attenuated in Bcl11b-deficient CD8⁺ T cells after activation of CD8⁺ T cells with α-CD3/CD28 antibodies, supporting the observation with regard to Plcγ1 expression (Fig. 5 E). All these data collectively demonstrate that deregulated expression of Plcγ1, together with the reduced CD8 coreceptor expression in Bcl11b-deficient CD8⁺ T lymphocytes, are likely to contribute to the reduced Ag-specific proliferation and expansion in vivo and in vitro.

Regulation of CD8 expression by Bcl11b raised the question of whether Bcl11b directly controls the expression of CD8 coreceptor genes by binding to the CD8 enhancers. We conducted chromatin immunoprecipitations (ChIPs) to determine whether Bcl11b associates with any of the CD8 enhancers, including E8I, E8II, E8III, EIV, and E8V. Our ChIP assays demonstrated that Bcl11b associated with E8I, EIV, and E8V enhancers but not with E8II and E8III enhancers (Fig. 6). E8I and E8II control the CD8 coreceptor levels in mature CD8⁺ T cells. In addition, E8II and E8III together control the CD8 coreceptor levels in DP thymocytes (Ellmeier et al., 1997, 1998; Hostert et al., 1997; Kioussis and Ellmeier, 2002; Feik et al., 2005). However, maintenance of the CD8 coreceptor genes expression in the periphery seems to be complex, as even in the absence of E8I and E8II, CD8⁺ T cells still maintain much of their coreceptor expression (Ellmeier et al., 2002). It has been suggested that other enhancers, such as EIV and EV, and/or other mechanisms, might be involved in the maintenance of CD8 expression in the periphery (Kioussis and Ellmeier, 2002). Our results show that Bcl11b is present on several enhancers of the CD8 coreceptor genes, including E8I, E8IV, and E8V and through these multiple associations perhaps contributes to the maintenance of optimal expression levels of the CD8 coreceptor genes in the periphery. The remaining expression of CD8 coreceptor in the absence of Bcl11b may be attributed to the E8II enhancer, to which Bcl11b did not bind (Fig. 6), as well as to other transcription factors that control the E8I, E8IV, and E8V enhancers. Overall, our data demonstrate that Bcl11b is one of the transcription factors implicated in the maintenance of CD8 coreceptor expression in peripheral CD8⁺ T cells, potentially through direct association with some of the CD8 gene enhancers.

Because effective CTLs are required for the control of L. monocytogenes, we next evaluated the bacterial clearance mediated by Bcl11b-deficient CD8⁺ T cells. In these experiments Bcl11b-deficient and WT OT-1 CD8⁺ T cells were transferred to separate recipient mice, which were then infected with Lm-Ova. We found that bacterial titers were much higher in recipients of Bcl11b-deficient OT-1 CD8⁺ T cells than in recipients of normal OT-1 CD8⁺ T cells (Fig. 7 A). Bacterial clearance was also impaired when Bcl11b^F/F/dLck-iCre/OT-1 and WT OT-1 mice were directly infected with Lm-Ova (unpublished data).

Because reduced bacterial clearance may be caused by the reduced Ag-specific expansion of Bcl11b-deficient CD8⁺ T cells, we further evaluated ex vivo, on a per cell basis, the Ag-specific killing activity of CTLs purified from LM-Ova–infected mice. Specifically, we compared equal numbers of Bcl11b-deficient and WT OT-1 CTLs purified from Lm-Ova–infected OT-1 mice for Ag-specific cytotoxic activity against MEC.B7.SigOva target cells, expressing OVA_257-264 peptide via MHC class I (van Stipdonk et al., 2001). Our results clearly show that in the absence of Bcl11b, CTLs killed OVA_257-264-expressing target cells less efficiently on a per cell basis (Fig. 7 B), suggesting that Bcl11b-deficient CD8⁺ T cells might have defects in CTL activity.

One cause for the poor Ag-specific cytolytic activity of Bcl11b-deficient CTLs might be defective differentiation into short-lived effector cells, which up-regulate Klrg1 and down-regulate CD127, during the immune response (Kaech et al., 2003). Our results show that Bcl11b-deficient OT-1 CD8⁺ T cells differentiated normally into Klrg1^hiCD127^lo effector population (Fig. 8 A); however, the absolute numbers of effector cells was reduced overall (not depicted). Indeed, the reduced numbers of effector cells may account for the increased bacterial burden. However, in the ex vivo killing assays, equal numbers of CD8⁺ T cells from LM-Ova–infected Bcl11b^F/F/dLck-iCre/OT-1 and WT OT-1 mice were used. Even in these conditions, Bcl11b-deficient CTLs presented reduced Ag-specific killing activity, demonstrating that the reduced killing activity is not only a result of reduced numbers of CTLs. Thus, the reason for poor cytolytic activity could be caused by a reduction in the cytolytic effector molecules, perforin, and granzyme B. We found that the levels of granzyme B were reduced in Bcl11b-deficient CTLs at the peak of infection with LM-Ova (Fig. 8 B). In addition, the CD8⁺ T cells purified from LM-Ova–infected Bcl11b^F/F/dLck-iCre/OT1 mice presented a major reduction in the perforin levels, compared with equal numbers of OT1 WT cells (Fig. 8 C). We also investigated the perforin and granzyme B mRNA levels in equal numbers of CD8⁺ T cells purified from LM-Ova–infected Bcl11b^F/F/dLck-iCre/OT1 and OT1 mice and found lower levels of both mRNAs in the absence of Bcl11b (Fig. 8 D).

We further determined, by ChIP assays, whether Bcl11b associates with the genomic regions upstream of the transcription start sites (TSSs) of perforin and granzyme B genes in CD8⁺ T cells purified from LM-Ova–infected mice. Our data demonstrated that Bcl11b associated with the genomic region between −452 and −1000 upstream of the perforin gene TSS (Fig. 8 E) but not to the distal CTL-specific DNase I hypersensitive site (Pipkin et al., 2007; Cruz-Guilloty et al., 2009; Fig. S5). Bcl11b also associated with the granzyme B genomic region from −13 to −1309 upstream of TSS (Fig. 8 F) but not with a genomic region localized 1 kb downstream of the granzyme B gene stop codon (Fig. S5). The ratios of the CD8 mean fluorescence intensity (MFI) between Bcl11b-deficient and WT CD8⁺ T cells during T cell activation, as well as in CD8⁺ T cells isolated from LM-Ova–infected mice, remained similar to the ratio of CD8⁺ T cells from naive mice (Table S2), suggesting that it is unlikely that further reduction in CD8 levels during the activation might contribute to the reduction in effector molecules.

All these results collectively demonstrate that Bcl11b-deficient CD8⁺ T cells have reduced cytolytic activity in vivo in response to infection and, ex vivo, against Ag-specific targets. The reduced cytolytic activity in vivo correlates with decreased ability of Bcl11b-deficient CTLs to produce perforin and granzyme B during LM-Ova infection.

We further used the ER-Cre system for inducible removal of Bcl11b in an in vitro culture system for CTL differentiation (Cruz-Guilloty et al., 2009). Specifically, we first activated CD8⁺ T cells from Bcl11b^F/F/ER-Cre and ER-Cre mice, after which the cells were grown in media containing OH-tamoxifen. Under these in vitro conditions, and with removal of the gene after the activation of the CD8⁺ T cells, the perforin and granzyme B mRNA levels were similar in Bcl11b^F/F/ER-Cre and ER-Cre CD8⁺ T cells (Fig. 8, G and H). Thus, removal of Bcl11b after in vitro activation of CD8⁺ T cells does not result in down-regulation of perforin and granzyme B mRNA levels, suggesting that the in vivo reduction of perforin and granzyme B and the reduced cytolytic activity might be secondary to suboptimal TCR activation, at least in part.

In this paper, we provide evidence that Bcl11b is a critical transcription factor for CD8⁺ T cell immune response by controlling two critical aspects of the CD8 immunity: Ag-specific expansion in response to infection, and cytolysis. Specifically, we demonstrate that Bcl11b-deficient CD8⁺ T cells failed to efficiently expand in response to infection with L. monocytogenes or influenza and produced less perforin and granzyme B. Bcl11b-deficient CD8⁺ T cells presented reduced Ag-specific proliferation, but not increased apoptosis, during the course of the immune response. TCR signaling after Ag-specific activation is required for naive CD8⁺ T lymphocytes to enter cell cycle and expand. After infection, the majority of Bcl11b-deficient CD8⁺ T lymphocytes were found blocked in G0/G1 (unpublished data). Furthermore, Bcl11b-deficient CD8⁺ T lymphocytes presented reduced up-regulation of the early activation marker CD69 after TCR activation, which is suggestive of altered TCR signaling. In terms of the mechanism of the reduced expansion during the immune response, we found that the CD8 coreceptor levels were reduced in the absence of Bcl11b. CD8 coreceptor genes not only play a critical role in the CD8 lineage development (Fung-Leung et al., 1991; Crooks and Littman, 1994) but also in proliferative response of CD8⁺ T cells (Cai and Sprent, 1994). CD8 coreceptor contributes to TCR binding to peptide–MHC class I complexes on APCs (Garcia et al., 1996), and it induces a conformational change in the CD3, which is important for TCR signaling (Gil et al., 2008). In addition, the coreceptor amplifies TCR signaling, through partitioning, together with the TCR complex, to the lipid rafts and association with p56lck (Arcaro et al., 2001). Bcl11b regulated CD8 coreceptor gene expression through association with the E8I enhancer, which is known to play a role in the CD8 coreceptor expression in mature CD8⁺ T cells (Ellmeier et al., 1997, 1998; Hostert et al., 1997; Kioussis and Ellmeier, 2002; Feik et al., 2005). In addition, Bcl11b was found to be present on E8IV and E8V enhancers, whose role in the control of CD8 expression is less characterized. Although the regulation of CD8 coreceptor expression during development has been well characterized (Ellmeier et al., 1997, 1998; Hostert et al., 1997; Kioussis and Ellmeier, 2002; Feik et al., 2005), not much is known about the regulation and maintenance of its expression in the peripheral CD8⁺ T cells. In the absence of Bcl11b, the levels of the CD8 coreceptor were reduced but not abrogated, which is consistent with the idea that other transcription factors play roles in controlling the activity of the E8I, E8IV, and E8V enhancers. In addition to this, Bcl11b might also regulate expression of CD8 coreceptors indirectly by regulating other transcription factors implicated in this process, such as MAZR (Bilic et al., 2006). Such a possibility is currently under investigation. The remaining CD8 coreceptor levels in the absence of Bcl11b may be attributed to the E8II contribution, which is also known to control CD8 coreceptor expression in CD8⁺ T cells, in addition to the DP thymocytes (Ellmeier et al., 1997), to which Bcl11b did not bind. In the absence of Bcl11b in DP thymocytes, CD8 coreceptor levels were not altered, suggesting that Bcl11b does not control expression of CD8 coreceptor in DP thymocytes (Albu et al., 2007). Therefore, our results demonstrate that Bcl11b is one of the transcription factors critical for the maintenance of the CD8 coreceptor levels in the periphery. In addition to the CD8 coreceptor, we found that Bcl11b regulates the expression of Plcγ1 in mature CD8⁺ T cells, which is another critical gene for TCR signaling (Wilde and Watson, 2001; Bonvini et al., 2003). In conclusion, our results suggest that TCR signaling in Bcl11b-deficient CD8⁺ T cells is impaired by at least two mechanisms: reduced expression of CD8, and probably, therefore, less optimal activation of CD8⁺ T cells, and reduced expression of Plcγ1 and, hence, reduced Ca2⁺ mobilization.

After Ag-specific activation, naive CD8⁺ T lymphocytes expand and differentiate into effector cytotoxic T cells. The effector Klrg1^hiIL7R^lo CTL population was formed in the absence of Bcl11b, although in reduced numbers. The production of perforin and granzyme B, evaluated in equivalent numbers of CTLs from infected mice, was altered. Moreover, the perforin and granzyme B mRNAs were reduced in CD8⁺ T cells purified from LM-Ova–infected mice, and ChIP experiments demonstrated that Bcl11b bound to both perforin and granzyme B genomic regions upstream of their TSS in CTLs from infected mice. Unexpectedly, removal of Bcl11b after in vitro activation did not result in reduced perforin and granzyme B mRNA levels, suggesting that perhaps the reduction in perforin and granzyme B levels during infection is caused in part by suboptimal TCR activation. However, the fact that Bcl11b-deficient CTLs on a per cell basis had a reduced capacity to kill Ag-specific MEC.B7.SigOva target cells suggests that the reduced clonal expansion is not solely responsible for the altered cytolytic activity because in this assay equal numbers of Bcl11b-deficient and WT CTLs were compared. Thus, other possibilities for the differences that we observed between the in vivo versus the in vitro conditions include: growth of Bcl11b-deficient CD8⁺ T cells in vitro for several days might select a specific population, and/or might only favor survival of certain cells; the absence of specific conditions which are only present in vivo, but not in vitro; or in vivo removal of the gene with the dLckD-iCre before TCR activation, as opposed to in vitro removal after activation, could have different effects on CD8⁺ T cell activation/differentiation.

With regard to the presence of Bcl11b on genomic regions upstream of TSS of perforin and granzyme B genes, there are multiple possibilities: the presence of a transcription factor on promoter/enhancer does not always result in regulation of expression; or the presence of Bcl11b on genomic regions upstream of perforin and granzyme B TSS genes might only regulate perforin and granzyme B expression under specific conditions, which might differ between in vivo and in vitro experiments. Although at the moment it is hard to have a clear answer because of the discrepancy between the in vivo and in vitro results, it is possible that in certain conditions Bcl11b might regulate expression of perforin and granzyme B. Other transcription factors, such as Eomes and Runx3, were also found to control the expression of the CTL effector molecules (Pearce et al., 2003; Intlekofer et al., 2005; Chen et al., 2008; Cruz-Guilloty et al., 2009). Runx3 was found to associate with perforin and granzyme B promoters (Cruz-Guilloty et al., 2009). There are several Runx3 consensus sequences in the perforin promoter. One such site is closely located to a Bcl11b consensus site (unpublished data), suggesting the possibility that these two transcription factors may function together by binding to adjacent sites. The granzyme B promoter has a region of TGTG repeats which may potentially constitute binding sites for Runx3, and a Bcl11b consensus binding site resides very close to that (unpublished data), also suggesting the possibility that the two transcription factors may bind to the adjacent sites on the granzyme B promoter. Such possibilities are under investigation.

Bcl11b is expressed during T cell development, as well as in mature T cells, which suggests the possibility that it may regulate common genes in T cells and implicitly common genetic programs. Our observations suggest that, at least in DP thymocytes and CD8⁺ T cells, this is the case for certain genes (unpublished data); however, Bcl11b seems to differentially regulate specific genes in CD8⁺ T cells versus DP thymocytes, such as the CD8 coreceptor. How can Bcl11b regulate different genes in different T cell types, despite the fact that it is expressed at similar levels in these cells, rather than only regulating common genes? Bcl11b may cooperate with stage-specific transcription factors, and/or Bcl11b function may be modulated by specific signals, including TCR signaling, such that it can play a differential role in specific T cell subtypes. Such possibilities are under investigation. In conclusion, our data demonstrate that Bcl11b is a critical transcription factor for the CD8 immune response, controlling the expansion and execution of the effector function.

Mice carrying the floxed Bcl11b alleles (Bcl11b^F/F) were backcrossed on C57BL/6 background for nine generations and crossed with dLck-iCre C57BL/6 mice (gift from N. Killeen, University of California, San Francisco, San Francisco, CA) to generate conditional Bcl11b^F/F/dLck-iCre mice. To obtain OT-1 TCR/ Bcl11b^F/F/dLck-iCre, Bcl11b^F/F/dLck-iCre mice were crossed to OT-1 TCR transgenic mice (The Jackson Laboratory). CD45.1 mice were purchased from NCI or The Jackson Laboratory. Experiments were conducted with 5–10-wk-old mice unless specified. All mice except for the infection model were kept under specific pathogen-free conditions in the Albany Medical Center (AMC) Animal Facility, and all animal procedures were approved by the AMC Animal Care and Use Committee.

Naive CD8⁺ T lymphocytes were either sorted as CD62L^hiCD44^lo population or purified by depletion, using the CD8⁺ T cell depletion kit (Miltenyi Biotec). When using the Miltenyi depletion kit, the antibody cocktail was supplemented with 50 ng biotinylated anti–mouse CD25 and anti–mouse CD44 antibodies per 10⁷ cells to isolate mostly naive cells. The purity of CD44^loCD8⁺ T cells after purification was always >95%. The following antibodies (eBioscience) were used for surface and intracellular staining: anti-CD4 (clone GK1.5), -CD8α (clone 53–6.7), -CD8β (clone eBioH35-17.2), -CD45.1 (clone A20), -CD45.2 (clone 104), -CD69 (clone H1.2F3), -CD127 (clone A7R34), -Klrg1 (clone 2F1), –TCR-β (clone H57-597), -BrdU (clone PRB-1), –granzyme B (clone 16G6), and -perforin (clone eBioOMAK-D). Intranuclear staining of Bcl11b was conducted after staining for surface CD4 and CD8, followed by fixation and permeabilization with 1% Triton X-100 (Albu et al., 2007). Determination of apoptosis was conducted by intracellular staining with anti-cleaved caspase 3 antibodies (Cell Signaling Technology) after surface staining. Antibodies for Western blot analysis were: anti-Bcl11b (Bethyl Laboratories, Inc.), -perforin (Abcam), and P-Zap70 and Zap70 (Cell Signaling Technology).

1–2 × 10⁶ immune cells were stained with antibodies against surface markers for 30 min on ice. Cells were then fixed with freshly prepared Fixation/Permeabilization working solution (eBioscience) at 4°C for 30–60 min. After washing twice with permeabilization buffer (eBioscience), cells were incubated in permeabilization buffer at 4°C for 15 min, followed by addition of fluorophore-labeled anti–mouse granzyme antibodies (clone 16G6; eBioscience). Cells were then incubated on ice for 30 min in the dark, after which the cells were washed in the same permeabilization buffer and resuspended in staining buffer for flow cytometry analysis.

Flow cytometry analysis was conducted on FACSCalibur or FACSCanto flow cytometers (BD). Flow cytometry data were analyzed using FlowJo software (Tree Star, Inc.).

For influenza infection experiments, mixed bone marrow chimeras were generated by lethally irradiating CD45.1 recipient mice and reconstituting them with either a 1:1 mix of CD45.1 and CD45.2 C57BL/6 bone marrow (WT:WT chimeras) or a 1:1 mix of CD45.1 and CD45.2 Bcl11b^F/F/dLck-iCre bone marrow (WT:KO) chimeras. After reconstitution, mice were infected with 1,000 EIU of PR8. The frequencies of NP-specific and PA-specific CD8 T cells derived from each donor type were determined by flow cytometry in the mediastinal lymph nodes on day 10 after infection.

Sorted naive CD44^lo CD8⁺ T lymphocytes from CD45.2 Bcl11b^F/F/dLck-iCre/OT-1 and CD45.1/2/OT-1 mice (1:1; total 10⁴) were intravenously injected either together or separately into CD45.1 recipient mice. The next day after transfer, recipient mice were injected with 1.5 × 10⁴ CFU Lm-Ova (Shen et al., 1995). Bacterial burden was assessed in the livers and spleens of Lm-Ova–infected mice, separately transferred with Bcl11b-deficient or WT CD8⁺ OT-1 T cells on days 3, 4, and 5 after infection by plating homogenized liver and spleen tissues on tryptic soy broth agar plates. The next day, individual colonies were counted and bacterial burden was expressed as CFU/mg of tissue.

Bcl11b-deficient and WT CD8⁺ T lymphocytes were purified from spleens of LM-Ova infected mice at day 5 after infection and incubated with MEC.B7.SigOva cells expressing OVA_257-264 peptide via MHC class I (van Stipdonk et al., 2001). The lactate dehydrogenase activity released into the media after killing of the target cells was evaluated with the CytoTox96 kit (Promega).

Allelically different Bcl11b-deficient (CD45.2) and WT (CD45.1/2) naive CD8⁺ T lymphocytes were co–adoptively transferred in CD45.1 recipient mice which were then infected with LM-Ova. On day 6 after infection, mice were i.p. injected with 1 mg BrdU and sacrificed 16 h later for tissue harvesting. After surface staining, cells were permeabilized with ice-cold 95% ethanol and fixed, and then resuspended in freshly prepared DNase I (Roche) solution (50 Kunitz U/ml), stained with anti–BrdU-FITC antibody (eBioscience), and analyzed by flow cytometry.

In vitro proliferation was evaluated by CFSE dilution. Bcl11b-deficient and WT naive CD8⁺ T lymphocytes were purified from OT-1 mice as shown in the previous paragraph. The naive CD8⁺ T cells were labeled with 5 µM CFSE and incubated for 3 d with APCs pulsed with 0 or 0.1 µM OVA_257-264 peptide, for 2 d in media with 2 µg/ml of soluble α-CD28 and 0.5 µg/ml of cross-linked α-CD3, or in media with 10 nM PMA and 1 µM ionomycin. Cell divisions were assessed by tracking the dilution of CFSE by flow cytometry.

ChIP was conducted using CD8⁺ T cells purified on beads (Miltenyi Biotec), according to our previously published protocol (Cismasiu et al., 2006), with the difference that anti-Bcl11b antibodies were preloaded on anti–rabbit IgG-Dynabeads (Invitrogen), and the precipitated DNA was analyzed by qPCR. As a control, we used either anti-GFP antibodies or IgG. The primers for CD8 enhancers I, II, III, IV, and V were previously described (Sato et al., 2005). The transcription factor–bound DNA was calculated based on the following formula: 2^−ct of the immunoprecipitated sample divided by 2^−ct of the input, where ct is the threshold cycle, followed by corrections for the input dilution.

qRT-PCR from total RNA purified from naive or total CD8⁺ T lymphocytes, from naive mice, or from LM-Ova–infected mice was conducted as previously described (Cismasiu et al., 2006), always using primers that cross introns.

Differences between Bcl11b^F/F/dLck-iCre and control mice under all experimental conditions were analyzed by the two-tailed Student’s t test and expressed as the mean ± the SD. P ≤ 0.05 was considered significant for all analyses.

Fig. S1 shows conditional removal of Bcl11b in mature T cells with the use of the dLck-iCre system. Fig. S2 shows TCR-β on CD4 and CD8 SP thymocytes of Bcl11b^F/F/dLck-iCre mice. Fig. S3 shows histograms showing cleaved caspase-3 levels on gated CD45.1 recipient CD8⁺ T cell populations. Fig. S4 shows that acute removal of Bcl11b in vivo with the use of ER-Cre deleter results in reduced CD8 coreceptor levels. Fig. S5 shows that Bcl11b does not bind actin promoter, the distal CTL-specific DNase I hypersensitive site in the perforin locus, or a genomic region located 1 kb downstream of the granzyme B stop codon. Table S1 shows CD8 MFI in SP thymocytes and CD8⁺ T lymphocytes of Bcl11b^F/F/dLck-iCre and Bcl11b^F/F mice. Table S2 shows ratios of CD8 MFI of Bcl11b^F/F/dLck-iCre and Bcl11b^F/F CD8⁺ T lymphocytes.

We greatly acknowledge Dr. Nigel Killeen (University of California, San Francisco) for the dLck-iCre C57BL/6 mice, Dr. Hao Shen (University of Pennsylvania) for Lm-Ova, and Dr. Wilfried Ellmeier (Medical University of Vienna) for valuable information in regard to the CD8 enhancers. We thank members of Avram laboratory, Drs. Deb Fuller, Eric Yager, Michelle Lennartz (Albany Medical College), and Leszek Ignatowicz (Medical College of Georgia) for valuable suggestions and exciting discussions, Dr. Jennifer Torgersen, Ms. Jinbo Ma, and Mr. Rui Ou for technical assistance, Dr. Douglas Cohn and Ms. Betsy Bashaw for care of the mice, and Ms. Debbie Moran for secretarial assistance and help with the figures.

This work was supported by the National Institutes of Health (NIH) grants AI67846 and R01AI078273 to D. Avram, NIH grant AI61511 to T.D. Randall, and NIH training grant T32-HL-07194 to D. Albu.

The authors have no conflicting financial interest.