Nature 426, 671 - 676 (11 December 2003); doi:10.1038/nature02067
Development and maintenance of B and T lymphocytes requires antiapoptotic MCL-1
JOSEPH T. OPFERMAN1, ANTHONY LETAI1, CAROLINE BEARD1, MIA D. SORCINELLI1, CHRISTY C. ONG1 & STANLEY J. KORSMEYER1
Howard Hughes Medical Institute, Dana Farber Cancer Institute, Department of Pathology and Medicine, Harvard Medical School, Boston, Massachusetts 02115, USA
Correspondence and requests for materials should be addressed to S.K. (stanley_korsmeyer@dfci.harvard.edu).
Regulated apoptosis is essential for both the development and the subsequent maintenance of the immune system1, 2. Interleukins, including IL-2, IL-4, IL-7 and IL-15, heavily influence lymphocyte survival during the vulnerable stages of VDJ rearrangement and later in ensuring cellular homeostasis, but the genes specifically responsible for the development and maintenance of lymphocytes have not been identified3-8. The antiapoptotic protein MCL-1 is an attractive candidate, as it is highly regulated9, appears to enhance short-term survival10 and functions at an apical step in genotoxic deaths11. However, Mcl-1 deficiency results in peri-implantation lethality12. Here we show that mice conditional for Mcl-1 display a profound reduction in B and T lymphocytes when MCL-1 is removed. Deletion of Mcl-1 during early lymphocyte differentiation increased apoptosis and arrested the development at pro-B-cell and double-negative T-cell stages. Induced deletion of Mcl-1 in peripheral B- and T-cell populations resulted in their rapid loss. Moreover, IL-7 both induced and required MCL-1 to mediate lymphocyte survival. Thus, MCL-1, which selectively inhibits the proapoptotic protein BIM, is essential both early in lymphoid development and later on in the maintenance of mature lymphocytes.
The proapoptotic 'BH3-only' members of the BCL-2 family respond to selective death signals and trigger activation of the 'multidomain' death effectors BAX and BAK, which constitute an obligate gateway to the intrinsic death pathway. Conversely, antiapoptotic BCL-2 members have an important role in binding and sequestering BH3-only molecules, thus preventing activation of BAX and BAK13, 14. BH3 domains can be subdivided into those that 'activate' (for example, BID and BIM) by inducing oligomerization of BAX and BAK, or those that 'sensitize' (for example, BAD and BIK) by occupying the pocket of antiapoptotic family members15. Individual BCL-2 family members have specific roles. Haematopoiesis is initially normal in Bcl-2-deficient animals, but over time lymphocytes are vulnerable to apoptosis, particularly following activation stimuli16. Bcl-XL-deficient lymphocytes show that it is critical for the survival of CD4+CD8+ double-positive (DP) thymocytes, but not for mature T cells17, 18.
We generated a conditional Mcl-1 allele by targeting loxP sites upstream of the ATG start codon and between exons 1 and 2 (Fig. 1a). This Mcl-1flox(f) allele was transmitted through the germ line, and matings of Mcl-1f/wt mice yielded viable Mcl-1f/f offspring at the expected mendelian frequency (Supplementary Fig. 1a–d). Both Mcl-1f/wt and Mcl-1f/f cells express levels of MCL-1 protein comparable to wild-type (wt) cells (Fig. 1b and Supplementary Fig. 3a). We also generated an Mcl-1null allele in which portions of Mcl-1 exons 1 and 2 were replaced with an internal ribosome entry site allowing the expression of a neomycin–-galactosidase fusion protein (Fig. 1a). Mcl-1f/null murine embryonic fibroblasts (MEFs) after Cre-mediated deletion, Mcl-1deleted/null (/null), lack detectable MCL-1 protein (Fig. 1b).
Figure 1 Conditional deletion of Mcl-1 results in loss of peripheral lymphocytes. Full legend
High resolution image and legend (61k)
To restrict deletion of the Mcl-1f allele to T-cell development, we introduced the lymphocyte-specific protein tyrosine kinase–Cre (Lck–Cre) transgene19. Lck–CreMcl-1wt/null mice were crossed with Mcl-1f/f mice and the progeny analysed. Peripheral T lymphocytes were decreased in both spleen and lymph nodes of Lck–CreMcl-1f/null mice when compared with littermate control mice (f/wt, f/null and Lck–CreMcl-1f/wt all showed normal numbers) (Fig. 1c and Supplementary Fig. 2a, c). Both CD4+ and CD8+ subsets were depleted in these Lck–CreMcl-1f/null mice (Fig. 1c and Supplementary Fig. 2b). To determine whether the remaining T cells survived without MCL-1 or represent a minority population that failed to delete, we used polymerase chain reaction (PCR) assays that distinguish Mcl-1f, Mcl-1 and Mcl-1null alleles (Fig. 1a, d). The deleted allele, Mcl-1, was detectable only in CD3+ T cells bearing a Mcl-1wt allele and was absent in the residual T cells of Lck–CreMcl-1f/null mice (Fig. 1d). Consistent with this observation, the intact Mcl-1f allele was present in this residual T-cell population. This suggests that T cells lacking both copies of Mcl-1 are non-viable, and that the residual T cells found in Lck–CreMcl-1f/null mice exist because they have escaped Cre-mediated deletion.
CD19–CreMcl-1f/null mice with Cre recombinase under the control of a B-lineage-restricted gene20 exhibited a profound reduction in peripheral B cells when compared with littermates with intact Mcl-1 (Fig. 1e and Supplementary Fig. 2d, e). Mcl-1 deletion reduced the number of splenic B cells to 12% of the levels in Mcl-1f/wt mice, whereas lymph node B cells were barely detectable. Hemizygosity for Mcl-1 (Mcl-1f/null, CD19–CreMcl-1f/wt) resulted in reduced peripheral B220+ cells, indicating that Mcl-1 dosage has more effect on B-cell lineages than it does on T-cell lineages (Fig. 1e and Supplementary Fig. 2e). The few remaining peripheral B cells in CD19–CreMcl-1f/null mice escaped deletion as they displayed detectable Mcl-1f alleles but no Mcl-1 alleles by PCR (Fig. 1f). This suggests that MCL-1 is necessary for the development or maintenance of mature B cells.
We next examined whether MCL-1 was required at earlier stages of lymphocyte development. The total number of thymocytes was markedly reduced in Lck–CreMcl-1f/null mice (Fig. 2a; <20% of littermate Mcl-1f/wt mice). The numbers of CD4+CD8+ DP cells, as well as CD4+ or CD8+ single-positive (SP) cells, were all decreased (Fig. 2a and Supplementary Fig. 3b). This warranted an examination of earlier CD4-CD8- double-negative (DN) stages of thymocyte development that can be subdivided into four serial subclasses. Developing pro-T cells are initially CD44+CD25- DN stage 1 (DN1), progress to CD44+CD25+ DN stage 2 (DN2), followed by CD44-CD25+ DN stage 3 (DN3) and finally CD44-CD25- DN stage 4 (DN4) before becoming DP thymocytes21. The total number of DN1 and DN2 thymocytes was relatively normal in Lck–CreMcl-1f/null mice (Fig. 2b). By contrast, the DN3 and DN4 subsets were markedly reduced on deletion of Mcl-1 (Fig. 2b and Supplementary Fig. 3c). The deleted Mcl-1 allele was present in all four DN stages of development (Fig. 2c), consistent with studies placing the initiation of Lck–Cre-mediated deletion at the onset of CD44 expression19. Because DN3 and DN4 thymocytes are markedly depleted on deletion of Mcl-1, we examined the cell fate of the previous DN2 thymocytes and found a significant increase in the percentage of apoptotic (Annexin-V+) DN2 cells (Fig. 2d). Although deletion of Mcl-1 starts by DN1, the most prominent block manifests at DN2/DN3. It should be noted that DN2 cells undergo T-cell receptor rearrangement and are highly dependent on cytokine signalling for survival22.
Figure 2 Lck–Cre-mediated deletion of Mcl-1 blocks thymocyte development. Full legend
High resolution image and legend (59k)
The three main stages of early B-cell development are: pre-pro-B (CD43+B220+CD19-; subfraction A), pro-B (CD43+B220+CD19+; subfractions B and C) and pre-B (CD43-B220+CD19+; subfraction D)23. The percentage and absolute number of pre-pro-B cells in CD19–CreMcl-1f/null bone marrow were comparable to those in littermates that retained Mcl-1 (Fig. 3a, b). However, the pro-B- and pre-B-cell populations were markedly reduced after Mcl-1 deletion (Fig. 3b and Supplementary Fig. 4). This includes the earliest of the pro-B-cell stages, subset B, which was depleted, as were all subsequent stages (Fig. 3a,b and Supplementary Fig. 4). Although the pro-B, subfraction A cells had not deleted Mcl-1, stages B, C and D all showed Mcl-1 (Fig. 3c). Subfraction B pro-B cells with Mcl-1 showed more apoptotic (Annexin-V+) cells, consistent with an increased death of pro-B cells contributing to the loss of subsequent stages (Fig. 3d).
Figure 3 CD19–Cre-mediated deletion of Mcl-1 blocks B-cell development. Full legend
High resolution image and legend (89k)
The developmental stages at which T- and B-cell development is impaired in Mcl-1 conditional knockouts is similar to defects observed in interleukin-7 (IL-7) or IL-7 receptor (IL-7R)-deficient mice4, 5. To assess whether IL-7 can affect MCL-1 expression, we cultured recombination-activating gene 2 (Rag-2)-deficient thymocytes (enriched for DN1/DN2 owing to a lack of T-cell receptor rearrangement) in the presence or absence of IL-7 for 16 h (ref. 24). Treatment with IL-7 resulted in a 6-fold increase in MCL-1 protein levels in these thymocytes (Fig. 2e). Moreover, when real-time PCR was used to quantify messenger RNA, addition of IL-7 (10 µg ml-1) resulted in a 5-fold increase in Mcl-1 mRNA (Fig. 2f).
We addressed the question of whether mature lymphocytes also require MCL-1 for their maintenance. MCL-1 protein is present in peripheral B220+, CD4+ and CD8+ lymphocytes (Supplementary Fig. 5). The viability of wild-type T cells was promoted best by culture in IL-7, to a lesser extent with IL-15 and least with IL-2 (Fig. 4a). An increase in MCL-1 protein levels was most marked for T cells cultured in IL-7, followed by IL-15 and then IL-2 (Fig. 4b). Real-time PCR assay showed a quantitative increase in Mcl-1 mRNA on exposure of mature T cells to IL-7, indicating that there is a transcriptional component to this response (Fig. 4c). To assess whether Mcl-1 is required for IL-7-signalled survival, we turned to the MxCre recombinase model25. Interferon- (IFN-)-induced deletion of Mcl-1 in cultured CD3+ T cells from MxCreMcl-1f/null mice revealed that the IL-7-mediated survival of mature T cells requires MCL-1 (Fig. 4d).
Figure 4 MCL-1 is required for mature lymphocyte survival. Full legend
High resolution image and legend (73k)
To address whether MCL-1 is required for the maintenance of existing, mature lymphocytes in vivo, we administered polyinocinic-polycytidylic acid (pI-pC) to MxCreMcl-1f/null mice, which induces deletion of Mcl-1. Both B and T cells were depleted from the spleen 7 days after administration of pI-pC (Fig. 4e). As MxCre deletes Mcl-1 in multiple cell types, we performed an adoptive transfer to assess whether the loss of mature B and T cells represented a cell-autonomous event. CD3+ and B220+ peripheral lymphocytes were enriched by bead depletion of myeloid and natural killer cells from the spleen and lymph nodes of MxCreMcl-1f/null mice. Mature lymphocytes (5 106) were adoptively transferred to Rag-2-deficient recipients and allowed to engraft over 7–10 days. pI-pC was then administered to induce deletion of Mcl-1 in the donor lymphocytes. Within 2 days of pI-pC injection, the percentage of both T and B lymphocytes in spleen and lymph nodes was reduced by 60% for MxCreMcl-1f/null cells compared with Mcl-1f/wt cells (Fig. 4f). Peripheral B and T cells were progressively lost over the 2-week time course following deletion of Mcl-1 (Fig. 4f). This adoptive transfer regimen indicates an intrinsic requirement for MCL-1 in the maintenance of peripheral B and T cells, which are rapidly lost on Mcl-1 deletion.
To examine why MCL-1 is singularly required at cytokine-dependent steps, we assessed its capacity to bind selected BH3-only members. Two BH3-only members, BIM and BAD, are involved in cytokine-dependent survival pathways. Bim-deficient cells display resistance to death due to cytokine deprivation26; by contrast, Bad-null mice and knock-in mice bearing a non-phosphorylatable BAD3SA molecule show an altered apoptotic threshold for cytokine deprivation, confirming a 'sensitizer' role for BAD27, 28. Recombinant MCL-1 protein showed no substantial binding to the BAD-BH3 peptide (>2,900 nM), but did display a strong interaction with BIM-BH3 (74 2 nM) (Fig. 2g). By contrast, BCL-2 showed comparable binding to both BAD and BIM-BH3 peptides15 (Fig. 2g). Consistent with these determined affinities, immunoprecipitation of proteins from solubilized thymocytes revealed that MCL-1 interacts with BIM not BAD (Fig. 2h). This was confirmed with glutathione S-transferase (GST)–MCL-1 protein, which pulled down substantially more BIM from thymocyte extracts than did comparable GST–BCL2; however, MCL-1 again did not associate with BAD (Fig. 2i). This suggests that there are meaningful subsets of antiapoptotic BCL-2 members based on their selective interactions, just as there are biologically relevant subsets of proapoptotic members. A rationale for the importance of MCL-1 would be its selective sequestration of the 'activator' BH3-only BIM, thus preventing the activation of BAX and BAK.
Unexpectedly, MCL-1 showed dual requirements in the well-defined differentiation pathways of lymphocytes. Mature T and B cells have a striking dependence on MCL-1 for their survival. In addition, loss of MCL-1 increased apoptosis at DN2 and arrested the development of DN3 thymocytes, as well as arresting development of comparable-stage pro-B cells (subfraction B). Note that both DN2 and pro-B cells depend on the cytokine IL-7 signalling through common -chain and IL-7R chain pathways3-5. Overexpression of antiapoptotic BCL-2 has been shown to rescue T-cell development in IL-7R knockout mice29, 30, indicating that one of the main effects of IL-7 is to promote cell survival. IL-7 substantially increased the levels of MCL-1 in immature thymocytes and mature T cells, and enhanced their survival. Early lymphocyte development and mature lymphocyte homeostasis share the common characteristic of a cytokine-dependent step that is totally reliant on MCL-1 to block apoptosis. The evidence supports a model in which MCL-1 is required at a distinct apical step11 to counter BIM selectively, a checkpoint that could prove essential for multiple death/survival pathways.
Methods
Targeting the Mcl-1 genomic locus The Mcl-1 genomic locus was targeted by placing a loxP site upstream of the ATG start codon. A loxP-flanked cytomegalovirus (CMV) promoter driving the expression of a hygromycin–thymidine kinase fusion protein was inserted in the intron between exons 1 and 2. RW4 embryonic stem cells (129SvJ) were targeted with the construct, followed by transient transfection with CMV–Cre. Embryonic stem cell clones were selected and screened for successful recombination, and those containing two loxP sites were injected into C57Bl/6 blastocysts.
Cell sorting and viability Thymus, spleen, lymph nodes and bone marrow suspensions were prepared from animals that were killed at 6–12 weeks of age. Stained cells were analysed on a FACSCalibur (Becton Dickinson). For high-speed cell sorting, cells were analysed and collected on a MoFlow cell sorter (MoFlow). Cell-surface stains were purchased from BD-Pharmingen, with the exception of anti-CD127 (Ebioscience). Bead enrichment (>90%) for desired peripheral lymphocytes was performed by positive selection using magnetic beads (Dynal and Miltenyi). Viability was assessed by flow cytometric analysis of Annexin-V and propidium iodide staining (BioVision).
Analysis of MCL-1 protein Rag-2-deficient thymocytes (6–8 weeks of age) were cultured with or without recombinant mouse IL-7 (R&D). After 16 h of culture, cells were lysed and proteins were separated by 10% SDS–PAGE (Invitrogen); they were then transferred and immunoblotted using an MCL-1-specific rabbit polyclonal generated against MCL-1 peptide (sequence, SPEEELDGCEPEAIGKRPAV) and anti--actin (Chemicon).
Analysis of cytokine treatment on Mcl-1 mRNA expression Rag-2-deficient thymocytes or purified T lymphocytes were starved in complete RPMI (Roswell Park Memorial Institute) media with 10% fetal calf serum for 2 h. IL-7 was then re-administered to the cultures and cells were harvested at 30, 120 and 160 min. Total RNA was extracted (Trizol) and complementary DNA synthesized (Promega) to act as template for TaqMan real-time PCR analysis (Applied Biosystems, Prism). PCR primers were Mcl-1 cDNA (forward, AGAGCGCTGGAGACCCTG; reverse, CTATCTTATTAGATATGCCAGACC) and hypoxanthine phosphoribosyl transgerase (HPRT) (forward, GTTGGATACAGGCCAGACTTTGTTG; reverse, GAGGGTAGGCTGGCCTATAGGCT). These primer sets were used with Sybr-Green (Applied Biosystems) to measure Mcl-1 relative transcript levels and HPRT to control for input cDNA.
In vitro MxCre deletion CD3+ peripheral lymphocytes were enriched from MxCref/wt plus MxCref/null mice. The cells were cultured with or without IFN- (PBL Biomedical) or IL-7, or with both. After 48 h, the cells were analysed by flow cytometry.
Peripheral Mcl-1 deletion and analysis Lymphocytes were enriched by bead depletion of myeloid and natural killer cells from the spleen and lymph nodes of MxCreMcl-1f/wt and MxCreMcl-1f/null mice. Lymphocytes (5 106) were adoptively transferred to Rag-2-deficient recipients (6–12 weeks of age). After 7–10 days of engraftment, mice were injected with pI-pC (400 µg) to stimulate IFN-, inducing Cre. Recipient mice were killed and analysed.
BH3-only interaction experiments Affinities of interaction between GST fusion proteins and BH3 peptides were determined as previously described by fluorescence polarization15. Co-immunoprecipitations were performed on whole-cell, wild-type thymocytes lysed in 0.2% NP-40 using anti-BAD (clone 48, BD-Transduction Labs) and anti-BIM (clone 14A8, Chemicon), then immunoblotted as indicated. GST pull-down experiments were performed on thymocyte lysates (0.2% NP-40) then immunoblotted.
Supplementary information accompanies this paper.
