GIMAP1 Is Essential for the Survival of Naive and Activated B Cells In Vivo Free

The first identified member of the family of GTPases of the immunity associated proteins (GIMAPs) was GIMAP1, which was reported as a malaria-induced gene in mouse splenocytes (1). Since then, genetic association studies implicated human GIMAP genes in autoimmune diseases, including systemic lupus erythematosus (2), Behçet’s disease (3), and type I diabetes (4, 5). Furthermore, their deregulated expression was reported in lymphomas (6–11). Eight or nine GIMAP family members have been identified in mammals (12). They are a family of septin-related guanine nucleotide-binding G proteins that bear strong resemblance to dynamins (13). Mammalian GIMAPs are expressed prominently within lymphoid compartments, suggesting a role in lymphocyte function (12, 14–19). In vivo and in vitro studies implied a role for GIMAPs in lymphoid homeostasis and survival (20–30).

GIMAP5 is the most studied GIMAP family member. A mutation in Gimap5 was found to be the cause of lymphopenia seen in the Biobreeding diabetes-prone rat strain (14, 15). In GIMAP5-deficient rats, T cell development appears to occur normally within the thymus, but there are few T cells in the periphery (14, 15, 24, 31, 32). This was attributed to spontaneous apoptosis of T cells, although the mechanism by which this occurs remains unclear (24, 32, 33). Recent work suggested that T cell death may result from the inability of their mitochondria to sequester Ca²⁺ following capacitative entry (28). A similar paucity of peripheral T cells is seen in GIMAP5-deficient mice, which develop spontaneous colitis, resulting in early mortality (23, 26, 27). Gimap5 deficiency in mice affects various hematopoietic cell types (23, 27, 34) and can lead to a progressive multilineage failure of bone marrow hematopoiesis (34). Knowledge of the extent to which these effects are cell intrinsic awaits the use of conditional alleles in the study of Gimap5.

GIMAP5’s close relative, GIMAP1, is also required for the maintenance of peripheral T cells. Previously, we showed that conditional deletion of Gimap1 from lymphocyte progenitors using Gimap1^f/fCD2Cre⁺ (CD2Cre) mice resulted in normal lymphocyte development but severe reductions in peripheral T cell numbers (22). Surprisingly, we also found a profound deficit of mature peripheral B cells. This study did not address GIMAP1 function in activated B cells. The role that GIMAPs might play in the survival of activated lymphocytes remains unresolved. Although GIMAP5-deficient rat T cells can be activated successfully via their AgRs, GIMAP5-deficient mouse T cells were reported to be unable to proliferate in response to ex vivo stimulation (24, 27, 35). More recently, other studies suggested an important role for GIMAP1 in mature B cells, highlighting its potential role in B cell lymphomas. Diffuse large B cell lymphomas (DLBCLs) show hypomethylation at the Gimap locus, resulting in overexpression of GIMAP1 (10). In addition, the Gimap cluster is found within an early replication fragile site hotspot (6). Early replication fragile site hotspots are proposed to play a mechanistic role in some of the most common genome rearrangements during B cell lymphomagenesis. These studies prompted us to examine, in greater depth, the role that GIMAP1 plays in B cell function. We used a combination of transgenic mice in conjunction with in vivo and in vitro techniques to show that GIMAP1 is required for the maintenance of B cell numbers in the resting peripheral pool, as well as throughout mature B cell activation and differentiation.

Mice were bred and maintained in specific pathogen–free conditions at The Babraham Institute. Husbandry and experimentation complied with existing United Kingdom Home Office and European Union legislation, as well as local standards, as approved by the Babraham Institute Animal Welfare and Ethical Review Body. Gimap1^f/f mice [described previously (22)], bearing a “floxed” Gimap1 allele, were crossed with Cd79a^cre/+ mice (obtained from Michael Reth, Max Planck Institute for Immunobiology and Epigenetics, Freiburg, Germany) to generate Gimap1^f/fCd79a^cre/+ mice, allowing conditional ablation of Gimap1 in the B cell lineage (36). Gimap1^f/f mice were also crossed with ER^T2Cre⁺ mice (obtained from Thomas Ludwig, Columbia University, New York, NY) to generate Gimap1^f/fER^T2Cre⁺ mice, enabling conditional ablation of Gimap1 upon administration of tamoxifen (37). To conditionally delete Gimap1 in germinal center (GC) B cells, Gimap1^f/f mice were crossed with AicdaCre⁺ mice (38) (obtained from M. Busslinger, Research Institute of Molecular Pathology, Vienna, Austria) to generate Gimap1^f/fAicdaCre⁺ animals. Gimap1^f/fCd2^cre/+ mice [described previously (22)] were crossed with Eμ-bcl-2-36–transgenic mice expressing human Bcl2 (39) to generate Gimap1^f/fCd2^cre/+Bcl2^tg mice.

Mice were immunized i.p. with 100 μg 4-hydroxy-3-nitrophenylacetyl NP-19–keyhole limpet hemocyanin (KLH; Biosearch Technologies, Nocata, CA) adsorbed to alum in saline. They were bled at the indicated time points before boosting with soluble Ag. In adoptive-transfer experiments, lymph node cells from Gimap1^f/fER^T2Cre⁺ and ER^T2Cre⁺ mice were stained with CFSE and CellTrace Violet (CTV; Life Technologies), respectively, and then mixed in a 1:2 ratio (Gimap1^f/fER^T2Cre⁺/ER^T2Cre⁺) prior to i.v. injection of 5 × 10⁶ cells/mouse into B6.SJL-Ptprca Pepcb/BoyJ mice. Mice were treated with 200 μg tamoxifen/g body weight or vehicle control i.p. on days 1 and 2 following adoptive cell transfer. On day 13 after cell transfer, mice were killed, and the numbers of transferred cells present in peripheral blood and spleen were determined on the basis of anti-CD45.1, anti-CD45.2, CFSE, CTV, and anti-B220 staining.

Single-cell suspensions were prepared from lymphoid tissues and peripheral blood. Abs were directed against the following surface markers: CD93, B220, CD38, GL7, Fas, IgG1, CD138, IgM, CD45.1, and CD45.2. DAPI and biotinylated nitrophenyl-4-hydroxy-5 iodophenacetic acid (NIP) cells were analyzed using an LSR II or Fortessa (BD Biosciences, Oxford, U.K.), and data were analyzed using FlowJo software (TreeStar, Ashland, OR). To purify GC and follicular B cells, splenocytes from animals immunized 7 d previously with NP-KLH in alum were depleted of CD43⁺ and CD11c⁺ cells using biotinylated anti-CD43 and anti-CD11c Abs, followed by anti-biotin beads (MACS; Miltenyi Biotec), prior to depletion on an AutoMACS (Miltenyi Biotec), resulting in >95% CD19⁺ cells. To fractionate CD19⁺ cells into GC and follicular B cells, cells were stained with B220, GL7, and CD95 and sorted for GC cells (B220⁺CD95⁺GL7⁺) and for follicular B cells (B220⁺CD95⁻GL7⁻). To measure levels of apoptosis, splenocytes were stained for B220, GL7, and CD95, and active caspase-3 was determined using a CaspGLOW Fluorescein Active Caspase-3 Staining Kit (BioVision, Milpitas, CA). For 5-ethynyl-2′-deoxyuridine (EdU) incorporation, mice were injected i.p. with 0.5 ml 2 mg/ml EdU in PBS, and splenocytes were harvested 90 min later. EdU incorporation was determined using the Molecular Probes Click-iT EdU Flow Cytometry Assay Kit (Thermo Fisher Scientific).

ELISA plates (Nunc, Paisley, U.K.) were coated with NP23-BSA or NP2-BSA (10 or 2 μg/ml, respectively) as previously described (40). For ELISPOT, MultiScreen-HA mixed cellulose ester plates (Millipore, Watford, U.K.) were coated with NP23-BSA or NP2-BSA in PBS, washed, and blocked before applying serially diluted cells and incubating overnight. Detection was performed using biotinylated anti-IgG1 and anti-IgM (SouthernBiotech, Birmingham, AL), alkaline phosphatase–conjugated streptavidin (R&D Systems), and 5-bromo-4-chloro-3-indoyl phosphate/nitro-blue tetrazolium substrate (R&D Systems), and the number of Ab-secreting cells (ASCs) was determined using a CTL ELISPOT reader and CTL ImmunoSpot Software (Cell Technologies, Shaker Heights, OH).

Lysates were prepared from FACS-purified GC and follicular B cells, and PCR for GIMAP1 and GIMAP8 was performed as previously described (22, 30).

FACS-purified GC and follicular B cells were lysed in Nonidet P-40 lysis buffer, and GIMAP1 and actin protein were detected by Western blot, as previously described (22).

Follicular B cells were purified as described above and cultured on fibroblasts (expressing CD40L and BAFF) with IL-4 in the presence of vehicle control (DMSO) or 50 nM 4-hydroxytamoxifen (4-OHT) for 4 d, prior to reculture in IL-21 for an additional 4 d, as described (41). Cells were counted using a CASY cell counter (Innovatis) and stained for flow cytometric analysis, as described above, using anti-CD138 and anti-IgG1.

We first determined the lineage-intrinsic nature of the B cell requirement for GIMAP1 by crossing our Gimap1^f/f allele with CD79a-Cre, which is only active in the B cell lineage (36). Splenocytes, bone marrow, and peritoneal cells were enumerated and stained to determine the numbers of different B cells, as previously described (22). The B cell phenotype of these mice mirrored that of Gimap1^f/fCD2Cre⁺ mice, in which numbers of pro-B, pre-B, and immature B cells were normal (Fig. 1A–C), but mature peripheral B cells were significantly reduced (Fig. 1D–I). In addition, type 2 (T2) and type 3 transitional B cells were reduced, as previously observed in GIMAP1^f/fCD2Cre⁺ mice (Fig. 1J–L) (22). Previous studies suggested that GIMAPs might regulate survival by manipulation of Bcl-2 family members (25, 29, 34, 42). To address this, we crossed Gimap1^f/fCD2Cre⁺ mice (in which Gimap1 is ablated in the lymphoid lineage) with Eμ-bcl-2-36–transgenic mice, which overexpress human Bcl2 (39). To our surprise, hBcl-2 expression did not overcome the effect of Gimap1 deletion on the numbers of mature B cells in bone marrow, spleen, and peritoneum of Gimap1^f/fCD2Cre⁺ mice (Fig. 2). More interestingly, when expression of the hBcl-2 transgene caused an increase in cell numbers in control GIMAP1-sufficient mice (Gimap1^f/f), this increase was not seen in GIMAP1-deficient animals (Gimap1^f/fCD2Cre⁺). Signals through the BCR and BAFF receptor are also key for B cell survival. We found no differences in the expression of either of these receptors on GC B cells (Supplemental Fig. 1).

To examine the role of GIMAP1 in mature B cells in vivo, we generated mice, in which the Gimap1 gene can be inducibly deleted, by crossing Gimap1^f/f mice with mice expressing Cre recombinase under the control of the estrogen receptor (ER^T2Cre⁺) to create Gimap1^f/fER^T2Cre⁺ mice. This allowed conditional deletion of Gimap1 upon administration of tamoxifen or its synthetic derivative, 4-OHT. Deletion of Gimap1 was seen within 2 d of in vitro 4-OHT treatment of lymphocytes derived from Gimap1^f/fER^T2Cre⁺ mice (P. Datta, unpublished observations). To control for Cre-mediated toxicity, ER^T2Cre⁺ mice were used alongside Gimap1^f/fER^T2Cre⁺ mice (43). In preliminary experiments, we administered tamoxifen to Gimap1^f/fER^T2Cre⁺ and ER^T2Cre⁺ mice. We observed reductions in T and B cells in peripheral lymphoid organs (data not shown). To eliminate the possible effect of Gimap1 deletion in nonlymphoid cells, we went on to use an adoptive transfer model in which only 5 × 10⁶ lymph node cells were injected into replete hosts. Using this system, only a small proportion of circulating lymphocytes is susceptible to Gimap1 deletion, which could be tracked by staining cells prior to transfer and by their expression of CD45.2. We adoptively transferred a mixture (1:2 ratio) of Gimap1^f/fER^T2Cre⁺ lymphocytes (stained with CFSE) and ER^T2Cre⁺ lymphocytes (stained with CTV) into a replete, congenic CD45.1 host and administered tamoxifen to induce deletion of Gimap1 in transferred Gimap1^f/fER^T2Cre⁺ cells. Thirteen days later, splenocytes and peripheral blood were harvested and analyzed. Tamoxifen (but not vehicle) treatment resulted in the loss of nearly all Gimap1^f/fER^T2Cre⁺ B cells, with little effect on ER^T2Cre⁺ B cells (Fig. 3). Fig. 3A and 3B show representative flow cytometry plots of CD45.2⁺B220⁺ splenocytes taken on day 13 after tamoxifen or vehicle treatment; transferred Er^T2Cre⁺ and Gimap1^f/fEr^T2Cre⁺ cells are stained with CTV or CFSE, respectively. It shows the disappearance of Gimap1^f/fER^T2Cre⁺ B cells and the relative retention of Er^T2Cre⁺ B cells. Unstained cells shown in these FACS plots are transferred cells that had presumably lost their label because they were CD45.2⁺B220⁺. These data are representative of three independent experiments, and we found no differential effect of labeling with CTV or CFSE on lymphocyte survival; similar results were obtained when Gimap1^f/fER^T2Cre⁺ cells were labeled with CTV and ER^T2Cre⁺ cells were labeled with CFSE (data not shown). This supports the view that the disappearance of Gimap1^f/fER^T2Cre⁺ B cells is due specifically to the loss of GIMAP1. The percentages of input Gimap1^f/fER^T2Cre⁺ and ER^T2Cre⁺ B cells found in spleen were determined for each mouse (Fig. 3C, 3D). The low frequency of transferred cells in the spleen is typical for this type of experiment and reflects the dispersal of lymphocytes throughout lymphoid organs and multiple other tissues and organs of the body. We also show the ratio of recovered Gimap1^f/fER^T2Cre⁺/ER^T2Cre⁺ B220⁺ cells from blood and spleen following tamoxifen or vehicle treatment (Fig. 3E, 3F). These results show that peripheral B cells intrinsically require continuous GIMAP1 expression for their maintenance.

GIMAP1 expression in GC B cells has not been analyzed. Upon activation, Ag-specific B cells differentiate into long-lived memory B cells and plasma cells (PCs), which are generated within the specialized microenvironment of the GC. Microarray studies from the Immunological Genome Project (https://www.immgen.org) show that GC B cells express GIMAP1 mRNA (Fig. 4A). We looked at GIMAP1 protein expression in GC B cells (Fig. 4B). Although mRNA levels of GIMAP1 are reduced in GC B cells compared with follicular B cells, we found similar levels of GIMAP1 protein expressed in the two cell types, suggesting posttranscriptional regulation of GIMAP1 mRNA in B cells. To examine the role that GIMAP1 plays in B cell activation and differentiation, we specifically ablated Gimap1 in activated B cells by crossing Gimap1^f/f mice with AicdaCre⁺ mice to generate Gimap1^f/fAicdaCre⁺ mice. Aicda is expressed within GC B cells and was used successfully to elucidate the roles of many genes in the GC reaction (38, 44). We purified GC and follicular B cells from these animals and from Gimap1^f/f controls by FACS sorting 7 d after NP-KLH immunization. Specific deletion of the Gimap1 gene was evident only in GC B cells (defined as B220^+veGL7^+veFas^+ve) from Gimap1^f/fAicdaCre⁺ mice (Fig. 5A). The nearby Gimap8 gene (analyzed as a control) was intact in GC and follicular B cells from both strains of mice.

In previous work, we were unable to address what role GIMAP1 might play in peripheral B cell function because Gimap1^f/fCD2Cre⁺ mice have very few mature B cells (22), and the lymphopenia within these mice is likely to affect B cell function. By immunizing Gimap1^f/fAicdaCre⁺ mice, we were able to determine whether GIMAP1 was required for peripheral B cell responses to a T-dependent Ag. Gimap1^f/fAicdaCre⁺ mice were generated, and their B cells were examined. Numbers and phenotypes of naive B cells were unaffected by the transgene (Supplemental Fig. 2). Mice were immunized with NP-KLH in alum, and B cell responses were analyzed 7, 14, and 35 d later. We used binding of biotinylated NIP to detect cells bearing an NP-specific BCR (Fig. 5B, 5C). NP-specific IgG1-switched splenic B cells were identified by flow cytometry (B220⁺IgM⁻IgD⁻NIP⁺IgG1^hi) (45) and enumerated (Fig. 5B, 5C). Gimap1^f/fAicdaCre⁺ mice mounted a very poor response, with relatively few NP-specific switched B cells detectable (Fig. 5B, 5C). We found a similar deficiency in GC B cells, defined as B220⁺GL7⁺CD95⁺ (data not shown). A defect in GC function in the absence of GIMAP1 was also evident when serum Ig levels in Gimap1^f/fAicdaCre⁺ and Gimap1^f/f mice were compared (Fig. 5D). At day 7, a small reduction in NP-specific IgG1 was seen in Gimap1^f/fAicdaCre⁺ mice compared with control mice. By day 14, this difference was much greater. This difference could reflect a slow turnover of functional GIMAP1 protein. In addition, we looked at affinity maturation by measuring Ab reactive with NP2 at 35 d postimmunization (p.i.) (Fig. 5D). Again, Gimap1^f/fAicdaCre⁺ mice had very low titers of high-affinity Abs. The number of ASCs was determined by ELISPOT on days 14 and 42 p.i. (Fig. 5E). As expected, no significant differences were seen in the numbers of IgM-secreting cells from the spleen or bone marrow. In contrast, numbers of IgG1-secreting cells specific for NP23 and NP2 from spleen and bone marrow were lower in Gimap1^f/fAicdaCre⁺ mice compared with control animals. In some mice, the number of ASCs was below the level of detection.

The impaired production of class-switched and high-affinity, Ag-specific Abs and the deficit of ASCs and long-lived PCs in Gimap1^f/fAicdaCre⁺ mice suggested a defect in GC B cells. We enumerated splenic GC B cells as NIP-binding, IgG1⁺ and CD38⁻ cells and looked at the development of the GC response on days 6, 8, and 10 p.i. The effect of Gimap1 deletion was evident by day 6 p.i. (Fig. 6A). By day 8 p.i., GC B cell numbers had decreased further in Gimap1^f/fAicdaCre⁺ mice. A slight recovery was seen at day 10 p.i., possibly reflecting the expansion of cell clones that had escaped Gimap1 deletion.

To investigate why GC B cells failed to develop in the absence of GIMAP1, we examined their proliferation and death. EdU incorporation into newly synthesized DNA was used to measure GC B cell proliferation on day 8 p.i. (Fig. 6B). There was a slight, but significant, decrease in the proliferation of these cells in Gimap1^f/fAicdaCre⁺ mice compared with controls. To look at cell death in GC B cells, we measured the percentage of GC B cells expressing active caspase-3 by flow cytometry. We found a small, but significant, difference in the percentage of GC B cells expressing active caspase-3 in cells from Gimap1^f/fAicdaCre⁺ mice (Fig. 6C).

To confirm our in vivo data, we made use of an in vitro system to generate induced GC (iGC) B cells (41). B cells were purified from spleens and cultured with fibroblasts expressing BAFF and CD40L. IL-4 was added to the culture for the first 4 d to generate iGC B cells and then IL-21 was added to induce their differentiation into PCs. B cells from Gimap1^f/fER^T2Cre⁺ and control ER^T2Cre⁺ mice were cultured for 8 d in the presence of 4-OHT or vehicle control. We were able to delete GIMAP1 by day 6 of culture and examine cell proliferation, GC development, differentiation to PCs, and cell death. Within 6 d of culture, all cells expressed GL7 and CD95, indicative of a GC-like phenotype (data not shown). The slope of the curves in vehicle-treated Gimap1^f/fER^T2Cre⁺ and ER^T2Cre⁺ cells was similar, suggesting that their rate of proliferation was the same. However, 4-OHT–treated Gimap1^f/fER^T2Cre⁺ cells, but not control ER^T2Cre⁺ cells, appeared to stop proliferating between days 4 and 8 (Fig. 7A). We found no difference in the percentage of cells that further differentiated into PCs (CD38⁺IgG1⁺; Fig. 7B), showing that lack of GIMAP1 has no effect on differentiation. In addition, we looked at DAPI uptake by these cells to determine the percentage of live cells. 4-OHT–treated Gimap1^f/fER^T2Cre⁺ cells had significantly fewer live (DAPI⁻) cells at day 8. This suggests that cells initiated apoptosis (evident on day 6) and stopped proliferating, resulting in a reduction in the number of iGC B cells and percentage of live cells on day 8 (Fig. 7C).

To determine whether memory B cell responses were also affected in Gimap1^f/fAicdaCre⁺ mice, we looked at secondary responses to NP-KLH. Thirty-five days after primary immunization, Gimap1^f/fAicdaCre⁺ mice and control mice were boosted with NP-KLH. Memory responses were analyzed 7 d later. The numbers of Ag-specific IgG1-switched B cells were significantly reduced in Gimap1^f/fAicdaCre⁺ animals; indeed, in some animals, none could be detected (Fig. 8A). NP-specific IgG1 Ab levels were also reduced, most markedly for high-affinity Ab (as determined by NP2 binding; Fig. 8B). When the numbers of IgG1 ASCs binding to NP23 and NP2 were determined, many Gimap1^f/fAicdaCre⁺ mice had numbers below our detection limit of 1 in 10⁶ cells (Fig. 8C).

Regulation of lymphocyte function and homeostasis is paramount for an effective and balanced immune system. Dysregulation of this balance frequently leads to autoimmune diseases, inability to fight infection, or cancer. Understanding the mechanisms maintaining this homeostasis may offer opportunities for the selective ablation of lymphocyte pools as a therapeutic strategy. Over the last 15 y, members of the GIMAP family have been implicated as important modulators of peripheral T lymphocyte homeostasis and survival. Very few studies addressed the role of GIMAPs in B cell biology. We showed that deletion of GIMAP8 results in a reduction in the number of recirculating B cells in the bone marrow (30). The generation and characterization of GIMAP1- and GIMAP5-deficient mouse strains further demonstrated that peripheral B cell survival is also influenced by GIMAPs (22, 23, 27, 34). As observed for T cells, deletion of Gimap1 appears to affect only mature cells within the B cell lineage (22). In contrast, GIMAP5-deficient mice show reductions in pro/pre-B cell progenitors, resulting in a more marked reduction in immature and mature B cells. This defect is thought to be due to the effect of GIMAP5 deficiency on hematopoietic stem cells (34). In a separate study, numbers of GIMAP5-deficient mouse B cells were also reduced in the periphery, and severe reductions in B cell responses to immunization were observed (27). As a caveat, there is an important methodological difference between our studies of GIMAP1 deficiency and those of GIMAP5; we used conditional, Cre-mediated ablation of Gimap1, whereas the Gimap5-related studies were conducted using germline mutant animals. In the absence of studies using conditional or elective deletion of the Gimap5 gene, it is not possible to assert whether the B cell changes reported in GIMAP5-deficient animals are B cell intrinsic or mediated via effects on other cells.

We have undertaken the development of mouse strains that enable specific and elective ablation of Gimap1 in B cells. Thus, the effects that we describe cannot be attributed to a “legacy” effect of GIMAP1 (i.e., the result of a function that GIMAP1 performs during B cell development that affects the functional capacity of the cells at a later stage) or to a cell-extrinsic effect acting on B cells. Adoptive transfer of B cells, followed by ablation of their Gimap1 gene, resulted in their disappearance from the periphery. This result reflects an intrinsic requirement for GIMAP1 in mature B cells, most likely to prevent their cell death in the periphery.

Importantly, our in vivo studies showed that B cells further require GIMAP1 during the development of Ab responses. To our knowledge, this is the first study to show the conditional ablation of a GIMAP protein during lymphocyte activation. Our data strongly suggest that B cells require GIMAP1 for their postactivation expansion because, in our in vitro iGC culture, we saw differentiation of cells into iGCs and PCs but not the expected subsequent expansion of these populations. Similar to the results of our in vitro experiments, GIMAP1-deficient GC B cells do not go through normal expansion in vivo. There is also reduced Ab production and affinity maturation, recall IgM response, and an inability to generate a long-lived memory response. These data establish GIMAP1 as absolutely essential for B cell peripheral function.

The deficit of GC B cells in Gimap1^f/fAicdaCre⁺ mice makes elucidation of the mechanism of GIMAP1 action challenging. We were able to detect an increase in active caspase-3 in GC B cells from Gimap1^f/fAicdaCre⁺ mice, indicative of apoptosis, and we also observed a reduction in the percentage of proliferating GC B cells. Although these differences were relatively small, the GC reaction is a rapidly evolving B cell response in which small changes in rates of proliferation and/or cell death can have a large impact on cell numbers (46). It is also likely that the few GC B cells that we detect are a mixture of GIMAP1-deficient GC B cells destined to die and GC B cells that have escaped Gimap1 gene deletion (47).

The mode of action of GIMAP1 remains enigmatic. GIMAP5 was reported to associate with Mcl1 and HSC70 and to promote stabilization of Mcl1 at the mitochondrial membrane (34). Mcl1 is critical for the survival of GC B cells and PCs (48, 49). It is possible that GIMAP1 also plays a role in Mcl1 stabilization that is not compensated by GIMAP5. However, we found no evidence of this in Gimap1-deficient GC cells. Furthermore, expression of the hBcl-2 transgene did not overcome the survival defect of mature B cells in Gimap1^f/fCD2Cre⁺ mice. An hBcl-2 transgene was shown to increase numbers of immature and mature B cell subsets (50). Interestingly, deletion of GIMAP1 overcame this effect of hBcl-2, suggesting that GIMAP1-induced lymphocyte survival uses a mechanism distinct from that of Bcl-2 family members or works downstream of Bcl-2 family members. We also looked at expression levels of IgM and BAFFR, both key mediators of B cell survival. However, no defects were detected in GIMAP1-deficient GC cells. In previous work, we showed that GIMAP5 associates with lysosomes, whereas GIMAP1 associates with the Golgi apparatus (51). Hence, it may be valid to speculate that B cells deficient in GIMAP1 succumb to lysosome-related cell death: lysosomal hydrolases are delivered to late endosomes via the mannose-6-phosphate pathway from the trans-Golgi network, where GIMAP1 is located. GIMAP1 loss may compromise the structural integrity of endosomes, enabling their release into the cytosol. Indeed, apoptosis of human GC B cells was shown to involve lysosomal destabilization, resulting in GC B cell death (52).

The fact that B cells can develop normally in the absence of GIMAP1, yet fail to maintain their peripheral numbers, suggests that the survival mechanisms of developing and mature B cells are distinct. Many proteins involved in B cell survival are required for both developing and mature B cells. However, members of the NF-κB family show B cell knockout phenotypes similar to Gimap1^f/fCD2Cre⁺ mice and are important for the survival of mature B cells, irrespective of their activation status (47, 53–57). In Gimap1^f/fCD2Cre⁺ mice, the B cell defect is first evident in the immature transitional T2 cells found in the spleen (22, 53). This implicates a problem in either the differentiation of type 1 cells into T2 cells or a failure of T2 cells to survive. A similar block is seen in mice deficient in members of the NF-κB family (53, 54). NF-κB activation is required for the differentiation of type 1 cells into T2 cells, suggesting that the BCR provides a survival signal mediated by NF-κB (53–57). This raises the possibility that tonic signaling through the BCR provides a survival signal to peripheral B cells that also relies upon GIMAP1. Perturbations in NF-κB signaling are seen in GIMAP5-deficient T cells, resulting in their deregulation (58). Dysregulation of the NF-κB signaling pathway is also seen in B cell lymphomas, with DLBCLs being strongly dependent on NF-κB activity (59–61). DLBCLs also have a hypomethylated region in the Gimap locus, resulting in overexpression of GIMAP1 and GIMAP5 (10). It is tempting to speculate that GIMAP1 and GIMAP5 regulate cell survival via the NF-κB pathway and that manipulation of either protein can tilt the balance between lymphocyte death and survival (58).

In summary, this work shows that GIMAP1 is required for the establishment and maintenance of the peripheral B cell pool and for all stages of postactivation B cell survival. In the absence of GIMAP1, mature B cells die, irrespective of their activation status or function. Together with our previous work, this establishes GIMAP1 as a key prosurvival factor for mature B lymphocytes and a potential target for the control of B cell–mediated diseases.

We thank Michael Reth for Cd79aCre⁺ mice, Thomas Ludwig for ER^T2Cre mice, Meinrad Busslinger for AicdaCre⁺ mice, members of the Babraham Institute Biological Services Unit for animal husbandry, members of the Babraham Institute Flow Cytometry Facility for assistance with flow cytometry and cell sorting, and Michelle Linterman for critical review of the manuscript.

The authors have no financial conflicts of interest.