Baff Mediates Survival of Peripheral Immature B Lymphocytes

Over the past 30 years, research on B lymphocyte development has concentrated on the early stages, in the bone marrow (BM), where lymphoid progenitor cells give rise to fully committed B lymphocytes after a process of sequential recombination and assembly of Ig gene components 1. The BM is also an important site for induction of immune tolerance, where self-reactive B lymphocytes are eliminated if they encounter membrane-bound self-antigen during their development 2,3,4.

In contrast, little is known about the signals driving the maturation of peripheral immature IgM⁺ B cells into mature, recirculating B cells. Several studies have shown that immature B cells leaving the BM, in contrast to mature B cells, do not proliferate in response to stimulation of their B cell receptor (BCR), but rather undergo death by apoptosis 5. This feature is thought to allow negative selection of emerging autoreactive B cells after encounter with self-antigen in the periphery 6,7.

Immature B cells found in the periphery are referred to as immature transitional B cells to distinguish them from their BM counterparts 6. The phenotype of transitional B cells varies from study to study but is generally thought to feature high levels of surface IgM (IgM^hi) and CD24 (heat stable antigen [HSA]; references 5, 6, 8–10). The lack of a clear definition of immature transitional B cells may have arisen from possible heterogeneity in this population. This idea is supported by a recent study demonstrating the existence of two transitional B cell stages in the spleen. Transitional type 1 (T1) B cells are IgM^hi and negative for IgD (IgD⁻), L-selectin (L-selectin⁻), CD23 (CD23⁻), and CD21 low (CD21^lo), and transitional type 2 (T2) B cells are IgM^hi, IgD^hi, CD21^hi, CD23⁺, and L-selectin⁺ 11. Immature B cells leaving the BM are thought to differentiate into T1 B cells before reaching the spleen. B cell maturation is an active process that appears to take place in the spleen, as T1 B cells do not express L-selectin, preventing them from homing to LNs, and T2 B cells are only found in the spleen 11. Adoptive transfer experiments in recombination activating gene (RAG)-2^−/− mice using purified T1 or T2 B cell subsets have shown that T1 are the precursors for T2 and mature B cells and T2 B cells are the direct precursors of mature B cells 11. It is still unclear whether T1 B cells can differentiate directly into mature B cells. The maturation of B cells through the T1 and T2 stages depends on signals through the BCR 11,12,13.

B cell activation factor from the TNF family (BAFF; also called TNF and apoptosis ligand–related leukocyte-expressed ligand 1 [TALL-1], TNF homologue that activates apoptosis, nuclear factor κB, and c-Jun NH₂-terminal kinase [THANK], B lymphocyte stimulator [BlyS], and zTNF4) is a member of the TNF family expressed by T cells, monocytes/macrophages, and dendritic cells 14,15,16,17,18. BAFF can be found both as a membrane-bound factor and as a soluble agent 14,17. BAFF specifically binds to B cells and promotes their proliferation in the presence of anti-μ 14,15,16,17. Recently, two known orphan TNF receptor–like molecules, B cell maturation antigen (BCMA [19, 20]) and transmembrane activator and calcium-modulating and cyclophilin ligand (CAML) interactor (TACI [21]), were identified as specific receptors for BAFF 18,22,23. Expression of BCMA is B cell specific 19,20,24,25, whereas that of TACI can be found on B cells as well as on subsets of activated T cells 21. Injection of BAFF in normal mice leads to the disruption of the splenic B and T cell structures and results in elevated levels of IgM in the serum 17. BAFF transgenic (Tg) mice have an elevated number of B lymphocytes in the periphery, secrete various autoantibodies, and develop an SLE-like condition leading to severe glomerulonephritis 18,26,27. The number of marginal zone (MZ) B cells is increased in BAFF Tg mice 26. In contrast, B cell development in the BM is not affected by the BAFF transgene 18,26,27. Treatment of normal mice with blocking soluble BCMA-Ig fusion protein led to a marked reduction in B cell numbers in the periphery 22. Treatment of mice with TACI-Ig inhibits T cell–dependent and –independent immune responses 23, abolishes germinal center formation 28, and was shown to prevent proteinuria and prolong the life of mice in a model of SLE 18.

Current information clearly demonstrates that BAFF is a very important factor controlling several aspects of B cell biology with the potential to break immune tolerance when overexpressed. Here we show that, in vitro, BAFF preferentially supports the survival of a subset of immature (T2) and some MZ B cells found solely in the spleen of normal mice. The T2 and MZ B cell subsets are enlarged in BAFF Tg mice, and some survive 72 h in vitro without exogenous stimuli. Our results suggest that excess survival signals provided to peripheral immature B cells in BAFF Tg mice reduce sensitivity of autoreactive B cells to BCR-mediated negative selection, resulting in autoimmunity. Our work underscores the importance of BAFF for B cell maturation, immune tolerance, and maintenance of B cell homeostasis.

Full-length murine BAFF was expressed in Tg mice using the liver-specific α1 antitrypsin promoter with the apolipoprotein (Apo) E enhancer as described previously 26. C57BL/6 and C3H/HeJ mice were purchased from Animal Resource Center. BAFF Tg mice are maintained as heterozygotes for the transgene by backcrossing onto C57BL/6 mice. BAFF Tg mice are screened for the presence of the transgene, both by PCR and Southern blot analysis using genomic DNA isolated from 2–3-mm long tail snips. We used mice from two separate lines of BAFF Tg mice issued after 8–10 backcrossings to C57BL/6. In experiments including BAFF Tg mice, negative littermates were used as control. Animals between 6 and 12 mo old were used, except for experiments including histological procedures for which younger 3–4-mo-old animals were selected. Animals were housed under conventional barrier protection and handled in accordance with the Animal Experimentation and Ethic Committee, which complies with the Australian code of practice for the care and use of animals for scientific purposes.

Flag-tagged soluble human BAFF (amino acids 83–285) was expressed by Escherichia coli and purified as described previously 14,22. Human recombinant BAFF was shown to efficiently stimulate mouse B cells 17,22. We found that the optimal concentration for BAFF in our assays was 2 μg/ml (data not shown). BAFF was also inactivated by boiling for 30–45 min and used as control. Polymyxin B and LPS from E. coli were obtained from Sigma-Aldrich and used at a final concentration of 5 and 10 μg/ml, respectively. Polyclonal rabbit anti-BAFF serum was obtained from rabbits immunized with human flag-tagged BAFF (amino acids 83–285) as described previously 14. The serum collected from bleeds made before immunization of these rabbits with BAFF was used as control.

Mice were killed by cervical dislocation and lymphoid organs were collected under sterile conditions. Spleen, thymus, and LNs were dissociated by grinding between frosted glass slides (Menzel-Glaser). Cells from the BM were collected after flushing mouse femurs with RPMI. PBLs were isolated by density gradient centrifugation of EDTA-treated mouse blood over Ficoll-Paque™ PLUS (Amersham Pharmacia Biotech). Cells were filtered though a 70-μm nylon cell strainer (Falcon; Becton Dickinson), and erythrocytes were removed by osmotic lysis with red blood cell lysis solution (8.34 g/liter ammonium chloride, 0.84 g/liter sodium bicarbonate, and 1 mM EDTA, pH 8.0). Cultures were conducted in glutamine-containing RPMI 1640 supplemented with 10% FCS and 100 U/ml penicillin/streptomycin (Life Technologies). Lymphocytes (3 × 10⁶/ml) were routinely stimulated for 72 h in culture with 2 μg/ml recombinant soluble human flag-tagged BAFF. Polyclonal rabbit anti-BAFF and rabbit control sera were used at the final concentration of 5%. Specific goat anti–mouse μ chain antibody was purchased from Southern Biotechnology Associates, Inc. and was used in culture at 10 μg/ml.

Freshly prepared or cultured lymphocytes were resuspended in FACS^® buffer (1% BSA, 0.05% sodium azide in PBS) at a concentration of 5 × 10⁶ cells/ml. Three- to four-color fluorescence surface staining was done using various combinations of FITC-, PE-, Cy5-, and Cychrome™-labeled antibodies. Fluorescent-labeled anti–mouse antibodies anti-CD4 (L3T4), anti-CD8α (Ly-2), anti-CD45R/B220 (RA3-6B2), anti-CD1 (1B1), anti-IgD (11-26c.2a), anti-IgM R6-60.2), anti-CD69 (H1.2F3), anti-CD62L (L-selectin), anti-CD5, anti-Fas, anti-CD23 (IgE Fc receptor), anti-CD24 (HSA, 30F1), and anti-CD21 were supplied by BD PharMingen. Cy5-conjugated anti-IgM antibody was purchased from Jackson ImmunoResearch Laboratories. FITC-labeled antibodies were used diluted 1:100, whereas other fluorochrome-labeled antibodies were used at a 1:200 final dilution. Annexin V-FITC Apoptosis Detection Kit 1 (BD PharMingen) was used according to the manufacturer's instructions. For flow cytometry we acquired 30,000 events per sample.

For BAFF binding experiments, sorted T1 and T2 B cells were incubated with 2 μg/ml of flag-tagged human BAFF for 1 h on ice in FACS^® buffer. Cells were washed twice and stained with a biotinylated anti-Flag antibody M2 (Sigma-Aldrich) diluted 1:500. Cells were washed twice and the signal was detected using PE-labeled streptavidin (Jackson ImmunoResearch Laboratories).

Data were collected on a FACSCalibur™ flow cytometer and analyzed using CELLQuest™ software (Becton Dickinson).

For cell sorting of splenic B and T cells, 5–10 × 10⁸ freshly isolated splenocytes were stained with PE-labeled anti-B220 and FITC-labeled anti-CD3 in PBS plus 10% FCS. Staining with biotin-labeled anti-HSA (revealed using Cychrome™-labeled streptavidin) and FITC-labeled anti-CD21 was used to sort T1 and T2 B cells instead of anti-IgM or anti-IgD to prevent potential activation signals through the BCR. The Mo-Flo cell sorter (Cytomation) at the Microbiology and Immunology Department of the University of New South Wales (New South Wales, Australia) was used to sort gated B220⁺ B cells, CD4⁺ or CD8⁺ T cells, HSA^hiCD21^lo/− T1 B cells, and HSA^hiCD21^hi T2 B cells. Reanalysis of sorted B or T cell populations demonstrated >98% purity. The purity for sorted populations of T1 and T2 B cells was 90 and 95%, respectively.

Frozen sections of spleen were subjected to immunohistochemical analysis as described previously 29. Biotin-labeled goat anti–mouse IgM antibodies and horseradish peroxidase–streptavidin were purchased from Jackson ImmunoResearch Laboratories.

Previous studies 17,18,22,26,27 have suggested a potential role for BAFF in B cell survival. We tested the survival effect of BAFF in vitro by incubating mouse splenocytes for 72 h with or without BAFF, and made a first assessment by flow cytometry using forward light scatter (FSC)/side light scatter (SSC) plots (Fig. 1 A). Freshly prepared splenocytes displayed higher FSC levels and are shown in the R2 population (Fig. 1 A, top). After 72 h in culture, untreated splenocytes died and display lower FSC levels, shown as the R1 population (Fig. 1 A, middle). Interestingly, many cells were detected in the live R2 gate in 72-h cultures of BAFF-stimulated splenocytes (Fig. 1 A, bottom). The FSC level of cells in R2 was similar to that of fresh splenocytes analyzed at the same time using the same instrument settings (Fig. 1 A, top and bottom). The R2 cell population was markedly reduced when BAFF was blocked using a blocking polyclonal rabbit anti-serum, but not when a control rabbit serum or BAFF inactivated by boiling was used (Fig. 1 B). We showed previously that blocking BAFF with BCMA-Ig, but not a control Ig fusion protein, was also able to reproduce this result 22. This phenomenon was not blocked by polymyxin B at concentrations which in parallel inhibited the effect of 10 μg/ml E. coli–derived LPS on splenocytes (data not shown). Moreover, the R2 population was also seen in BAFF-stimulated LPS-hyporesponsive splenocytes from C3H/HeJ mice (data not shown). These control experiments allowed us to rule out the possibility of endotoxin contamination from the E. coli–derived BAFF preparation as the factor responsible for the presence of the R2 population. BAFF alone at the concentrations used in our assays does not promote B cell proliferation 14,15. In conclusion, treatment of splenocytes for 72 h with BAFF specifically leads to the presence of cells in the R2 gate, with FSC levels suggesting that these cells might be surviving. We confirmed the viability of BAFF-stimulated cells present in the R2 population after 72 h in culture by showing that these cells were negative for annexin V binding and propidium iodine (PI) staining by flow cytometry (data not shown).

Cultured splenocytes were also double stained with anti-B220 and anti-CD4, or CD8 antibodies to detect B and T lymphocytes, respectively, by flow cytometry. The staining showed that 80% of cells in R2 are B cells. Surviving B cells represent 20–30% of the total BAFF-stimulated splenocyte culture, whereas only 2–3% of B cells survived in 72-h cultures of nonstimulated splenocytes. Some T cells (6.1 ± 1.8 vs. 1.2 ± 0.5% in controls), mainly CD4⁺ T cells, survived in this assay; however, this effect might result from some secondary support from surviving B cells. Evidence for that lies in the fact that T cells, highly purified by cell sorting, do not survive after 72-h stimulation with BAFF (Fig. 2 A). In contrast, 27% of purified B cells survive upon BAFF stimulation in the same assay (Fig. 2 A). Viability of the R2 population was always confirmed in parallel using the annexin V/PI staining for each experiment (data not shown). In conclusion, BAFF specifically induces the survival of B lymphocytes independently of the presence of T cells in vitro.

To test the potential survival effect of BAFF on B cells from other lymphoid organs, we purified lymphocytes from mouse blood (PBLs), peripheral LN (PLNs), mesenteric LNs (MLNs), and spleen, and incubated these cells with or without BAFF for 72 h. Significant BAFF-induced cell survival, as assessed by flow cytometric analysis of the annexin V/PI staining, was observed for cultured splenocytes (Fig. 2 B). Little cell survival was detected with BAFF-treated lymphocytes from LNs compared with untreated controls (Fig. 2 B). Unstimulated PBLs survived better in culture compared with lymphocytes from other sources, and the effect of BAFF on survival of these cells was variable but overall not highly significant (Fig. 2 B). No BAFF-induced survival was seen in mouse BM, thymus, or PBLs from human blood (data not shown). This result indicates that BAFF induces the survival of predominantly spleen-resident B cell subpopulations in vitro.

The spleen contains various subpopulations of B cells such as immature transitional B cells, which are divided into the T1 and T2 subgroups 11. The mature B cell subset is predominant, expressing low levels of IgM (IgM^dull), high levels of IgD (IgD^hi), intermediate levels of CD21 (CD21^int), and low to intermediate levels of HSA (HSA^lo/int; reference 1). Finally, MZ B cells, localized in the MZ, a structure delineating the red pulp from the white pulp, like T2 B cells are IgM^hi and CD21^hi but do not express either CD23 or IgD 30.

To define which of these B cell subpopulations survived after stimulation with BAFF, BAFF-treated splenocytes were stained with anti-IgM, anti-IgD, anti-B220, anti-HSA, anti-CD23, anti-CD21, and anti–L-selectin antibodies. By flow cytometry, cells were gated on the R2 surviving population (using FSC/SSC plots) and analyzed for expression of IgM, IgD, B220, HSA, and L-selectin in comparison with normal fresh splenic B cells, which were analyzed at the same time using the same instrument settings (Fig. 3 A). The surviving R2 B cell population contained IgM^hi and IgD⁺ cells (Fig. 3 A). These cells expressed very high levels of B220, suggesting that they were naive B cells (Fig. 3 A). R2 surviving B cells also expressed high levels of L-selectin, indicating that they were not T1 B cells (Fig. 3 A). Staining of control fresh splenocytes with HSA gave rise to two separate populations (Fig. 3 A), HSA^lo/int and HSA^hi cells, corresponding to mature and immature B cells, respectively. Interestingly, the expression of HSA on R2 surviving B cells was higher than that of mature B cells from fresh splenocytes, a sign of immaturity indicating that surviving B cells might be immature transitional B cells (Fig. 3 A).

To test this hypothesis, BAFF-treated and fresh splenocytes were stained with anti-IgM, anti-CD21, and anti-CD23 antibodies, and T1, T2, mature, and MZ B cell subpopulations were analyzed as described previously 11. BAFF-stimulated splenocytes were gated on the R2 surviving population (Fig. 3 B). Both fresh and R2 B cells were gated on the CD23⁺ and CD23⁻ subpopulations to distinguish T2/mature B cells from T1/MZ B cells, respectively. The staining showed that a large proportion of the surviving cells were T2 B cells. Almost no T1 B cells were present after 72-h culture with BAFF (Fig. 3 B). Some survival was observed in the MZ and mature B cell compartment (Fig. 3 B and Table); however, cells falling into the mature B cell gate also expressed higher levels of HSA and IgM compared with control mature B cells (Fig. 3 A and see Fig. 4 B, and Table). Interestingly, the pattern of IgM/CD21/CD23 expression of cells in R2 was identical to that seen with splenocytes from CBA/N and CD45^−/− mice, in which B cell maturation is arrested at the T2 B cell stage 11. Therefore, some or all of the mature-like B cells may in fact be T2 cells expressing slightly lower levels of IgM. The results indicate that T2 B cells account for most of the surviving cells in R2 when compared with fresh splenocytes (Fig. 3 B and Table). The same results were obtained when the IgD⁺ and IgD⁻ subpopulations were gated instead of the CD23⁺ and CD23⁻ B cells (data not shown). Moreover, the level of HSA expression on R2 surviving B cells is similar to that of gated T2 B cells from fresh splenocytes (Fig. 3 C).

T1 and T2 B cells were purified from normal splenocytes by cell sorting and incubated with BAFF for 72 h. Only the T2, but not the T1, B cell subset survived (Fig. 3 D) and kept their original T2 B cell phenotype (data not shown). We analyzed by flow cytometry the level of flag-tagged BAFF binding on the surface of sorted T1 and T2 B cells. We observed a more intense binding of BAFF to T1 B cells compared with that to T2 B cells, yet T1 B cells do not survive in vitro after BAFF treatment (Fig. 3 D). These results indicate that BAFF, in vitro, supports survival of T2 and some MZ B cells but not of T1 or mature B cells

To investigate whether BAFF, together with signals through the BCR, may be required for maturation of T2 B cells, we designed an in vitro maturation assay. We purified by cell sorting HSA^hiCD21^hi T2 and HSA^lo/intCD21^int mature B cells and incubated them for 72 h with BAFF. Due to technical limitations, ∼2% of mature B cells inevitably contaminate preparations of sorted T2 cells. Anti-μ antibodies can promote proliferation of mature B cells in the presence of BAFF and can interfere with analysis of our maturation assay 14,15,16,17. To overcome this problem, we added anti-μ antibodies 50 h after T2 and mature B cell cultures were started, when most mature B cells had already died. Cells were stained with antibodies to HSA, CD21, and IgD to delineate populations of mature and T2 B cells. BAFF-treated mature splenic B cells did not survive and the addition of anti-μ after 50 h of culture did not rescue these cells and did not promote their proliferation in the presence of BAFF (Fig. 4 A). This control was important, as it indicates that the inevitable low level of contaminating mature B cells found in the sorted T2 B cells will not interfere with our maturation assay using this particular experimental format. Untreated sorted T2 B cells died after 72 h in culture (Fig. 4 B). No cell survival was detected in cultures of T2 cells treated with anti-μ, in contrast to treatment with BAFF, which allowed T2 cells to survive (Fig. 4 B). Interestingly, in cultures of sorted T2 cells incubated for 72 h with BAFF and treated with anti-μ after 50 h in culture, we detected the presence of HSA^int/lo, CD21^int, and IgD⁺ cells corresponding to the phenotype of mature B cells (Fig. 4 B). However, fewer cells were detected in the surviving R2 population (Fig. 4 B, top) compared with BAFF treatment alone. This possibly indicates that many T2 B cells died after anti-μ plus BAFF stimulation and only a few of them were able to respond to the differentiating signals promoted by these reagents. No differentiation into MZ B cells was detected in these assays and T1 B cells did not survive in any of the culture conditions described here (data not shown). These results indicate that, in vitro, BAFF can induce B cell maturation when signals are triggered through the BCR of some T2 B cells.

As BAFF preferentially induces the survival of T2 and some MZ B cells from normal splenocytes in vitro, we analyzed whether these populations were affected in Tg mice overexpressing BAFF. Freshly prepared control and BAFF Tg mice–derived splenocytes were stained with anti-IgM, anti-CD21, and anti-CD23 antibodies as described above, and subpopulations of mature, T1, T2, and MZ B cells were analyzed by flow cytometry. Absolute cell numbers for each B cell subset were higher in spleens from BAFF Tg mice than from control littermates, and reflected the splenomegaly observed in these animals (26; Table). However, comparison of the proportion for each B cell subsets revealed a dominant expansion of the T2 and MZ B cells in the spleen of BAFF Tg mice compared with that of control mice (Fig. 5 A and Table). Expansion of the MZ was also evident through histological analysis of spleen sections stained with anti-IgM (Fig. 5 B). In contrast, the proportion of typical mature B cells decreased (Table). The proportion of T1 B cells in BAFF Tg mice remains essentially similar to that of control splenocytes (Fig. 5 A and Table). These results were reproduced in another laboratory (Woodcock, S.A., Biogen Incorporated, personal communication). Interestingly, the level of CD21 expression on T1 B cells from BAFF Tg mice is higher and that of IgM lower compared with levels on T1 B cells from control spleen. Although not conclusive, this observation suggests faster kinetics of B cell maturation in BAFF Tg mice (Fig. 5 A).

CD1^hi B cells have been shown to be the main source of autoantibodies in a murine model of SLE 31. We analyzed each splenic B cell subset for the presence of CD1^hi B cells in control and BAFF Tg mice. Levels of CD1 expression were generally slightly lower on B cells from BAFF Tg mice compared with control splenocytes (Fig. 5). Interestingly, in both control and BAFF Tg mice, T2 and MZ B cells express high levels of CD1 (Fig. 5 C). However, the CD1^hi B cell compartment accounts for >30% of all splenocytes in BAFF Tg mice versus only 9% in control mice (Fig. 5 D).

We also analyzed B cells in PLNs, MLNs, and blood from BAFF Tg mice. Few MZ-like cells could be detected in the PBLs and PLNs of control littermates (Fig. 5 E). Surprisingly, we found increased numbers of B cells displaying a T2 and MZ phenotype in these lymphocyte preparations (Fig. 5 E). Immaturity of the T2-like cells found in these preparations was confirmed by their expression of HSA, which was higher than that of gated mature B cells (data not shown). Expression levels of L-selectin on freshly prepared B cells from BAFF Tg and control mice were similar (data not shown) and do not account for aberrant homing of T2 and MZ B cells to the LNs. In conclusion, overexpression of BAFF in vivo leads to augmented numbers of all B cells in the spleen with the preferential expansion of the CD1^hi T2 and MZ B cell compartments but also expansion of T2 and MZ-like B cells in the blood and LNs.

Splenocytes from control littermates were incubated in medium supplemented or not with BAFF. Splenocytes from BAFF Tg mice were cultured in nonsupplemented medium. As shown previously, survival in control splenocyte cultures is only apparent (R2 population) if BAFF is added to the culture (Fig. 6 A). In contrast, cultures of BAFF Tg–derived splenocytes in nonsupplemented medium gave rise after 72 h to a surviving population of cells in the R2 gate (Fig. 6 A). Surviving cells were mainly B cells with the same phenotype as surviving B cells after 72-h BAFF treatment of C57BL/6-derived splenic cells (Fig. 3 A), displaying IgM^hi, IgD^hi, B220^hi, and HSA^hi levels (Fig. 6 B), and therefore were mainly T2 B cells. These surviving cultured cells also displayed higher levels of L-selectin compared with fresh splenocytes from BAFF Tg mice (data not shown). T2 B cells from BAFF Tg mice, isolated by cell sorting, also survived for 72 h in vitro, indicating that the survival status of these cells is intrinsic rather than promoted by BAFF produced by splenic T cells in mixed cultures (data not shown). These results suggest that T2 B cells were “programmed/sensitized” by BAFF in vivo leading to extended, stimulation-independent survival of these cells ex vivo.

The mechanisms responsible for the differentiation of immature B cells emerging from the BM into mature B cells found in the periphery are poorly understood. It is, for instance, still unclear why only 10% of newly formed IgM⁺ B cells reach the spleen, and why only a very small proportion of these achieve maturation 32. Many reports indicate that selection of specific transitional B cells takes place during B cell maturation in the periphery, and signaling through the BCR is a critical event in this process 11,12,33. Paradoxically, it is also clear that most immature transitional B cells die after stimulation of their BCR, a mechanism thought to eliminate autoreactive B cells 5,6,8,9. The exact nature of the positive signal driving maturation of B cells, although unknown, is either highly specific and/or requires concomitant signals apart from those through the BCR to allow transitional B cells to survive and differentiate into mature B cells.

We tested in vitro whether BAFF could be this extra factor needed to push immature B cells towards survival and differentiation. In our in vitro assays, we have been able to demonstrate that BAFF promotes the survival of immature T2 B cells and the differentiation of some of them into mature B cells when anti-μ antibodies were added to trigger signals through the BCR. Therefore, in vitro, BAFF synergizes with signals through the BCR to promote maturation of T2 B cells and as such is a critical new element for our understanding of this process.

The general expansion of the B cell compartment in the spleen of BAFF Tg mice, which is characterized by increased B cell numbers in all subsets, suggested that BAFF might act on all peripheral B cells. The discrepancy of this result with our in vitro results may be explained by environmental/tissue-specific signals acting in association with BAFF in vivo which the in vitro system could not mimic. However, the vastly unequal expansion of these subsets, favoring strongly the T2 and MZ B cell compartments, indicated that overexpression of BAFF directly or indirectly modified the dynamics of the entire B cell maturation and differentiation pathway in BAFF Tg mice. These results tend to indicate that BAFF might be a general survival factor for all B cells in vivo, with the specific potential to alter the process of B cell maturation.

Experiments in which BCMA-Ig was used to block BAFF in normal mice led to a marked and general reduction in B cell numbers in the periphery 22. Similarly, previous studies have shown that shutting off the expression of the BCR on the surface of B cells led rapidly to a reduced number of mature B cells in the periphery 34. The exact nature of the BCR-mediated survival signal in mature B cells is unknown. It is therefore possible that, in vivo, constant signals from BAFF and through the BCR are needed to keep the mature B cells alive. The dynamics of B cell homeostasis are very complex, and other studies have suggested that immature B cells may also compete with recirculating B cells for survival signals 35. We believe that normal recirculating B cells are probably not self-renewing and can only be replaced through the differentiation of immature B cells recruited in the periphery. Therefore, inhibition of BAFF, like the suppression of BCR signaling, could also affect B cell maturation and prevent the renewal of the mature B cell pool. 11-d treatment with BCMA-Ig markedly (although still incompletely) reduced B cell numbers in the periphery 22. A longer treatment with BCMA-Ig, reflecting the life span of recirculating B cells, might allow most mature B cells to naturally disappear from the periphery without being replaced. Supporting this model, BAFF does not promote survival of naive mature B cells, in vitro, despite expression of both BAFF receptors on these cells 18,22,23. BAFF might only be an important cofactor for proliferation once mature B cells have been activated 14,15,16,17,18. Nevertheless, we cannot formally exclude the possibility that, in vivo, BAFF directly supports survival of mature B cells. Adoptive transfer of B cell subsets in mice deficient for BAFF is the next strategy to obtain a final confirmation on whether BAFF is a general survival factor for all B cells or a specific survival and maturation factor for peripheral immature B cells. IL-4 and CD40 ligand are also factors involved in B cell survival; however, gene targeting experiments have shown that both factors are not required for B cell maturation 36,37. TACI and BCMA are also the receptors for another member of the TNF family, a proliferation-inducing ligand (APRIL), and it is at this point unclear whether or not this factor may also play a role in B cell survival 38.

Other intriguing features of BAFF Tg mice are the dramatic expansion of the MZ B cell compartment, its possible relation with the parallel expansion of the T2 B cells compartment, and the autoimmune phenotype displayed in these mice. The MZ B cell population is an obscure subset, which contains memory as well as virgin B cells 39,40. Previous studies have shown that both CBA/N and CD45^−/− mice have their B cell development blocked at the T2 B cell stage, yet these mice have a perfectly normal population of MZ B cells 11,41,42. This observation led to the hypothesis that some T2 B cells, under specific conditions, may directly differentiate into MZ B cells without going through a mature B cell stage. It is therefore conceivable that in BAFF Tg mice, B cell maturation is aberrantly skewed toward the MZ compartment. Recently, studies using various Ig heavy chain Tg mice have shown that, depending on their BCR composition, some newly formed mature B cells could be positively selected into the MZ compartment 40. Moreover, some B cells selected to migrate to the MZ are suspected to be potentially autoreactive 43. In addition, recent work showed that CD1^hi B cells in the spleen of NZB × NZW mice, a model of SLE, are the major source of autoreactive B cells 31. We showed that in normal mice, T2 and MZ B cells express high levels of CD1 compared with mature and T1 B cells, confirming previous reports 44. We also showed that the population of CD1^hi B cells is dramatically enlarged in BAFF Tg mice. Our findings led us to a model described in Fig. 7. BAFF promotes the survival of T2 B cells, which in normal conditions will lead to the maturation of highly selected cells into the mature follicular B cell compartment. In contrast, in BAFF Tg mice, overexpression of BAFF may trigger excessive survival signals in T2 B cells, including autoreactive cells, which may fail to respond to censoring death signals. The autoreactive nature of their receptors might also force their massive assignment to the MZ compartment and subsequent expansion of this compartment, as observed in BAFF Tg mice. These B cells may become activated by autoantigens at this site, contribute to the formation of numerous germinal centers observed in these mice 26, and differentiate into memory B cells, some of which may also localize in the MZ (Fig. 7). BAFF may also be important for the survival of germinal center B cells 28, and we do not exclude the possibility that autoreactive B cells may also emerge at this stage, due to aberrant survival of self-reactive B cells created after errors during affinity maturation.

Recent work described the selective absence of MZ B cells in PyK-2–deficient mice and a possible defect in MZ-specific chemotaxis 45. It is also possible that BAFF stimulates the production of an MZ-specific chemokine or a chemokine receptor on B cells driving most of these cells into the MZ of BAFF Tg mice. It is unclear whether the presence of T2 B cells in LNs of BAFF Tg mice also reflects an aberrant homing process or possible BAFF-enhanced B cell maturation in LNs. L-selectin expression was normal on freshly isolated splenocytes from BAFF Tg mice and cannot be evoked to explain this result. Previous studies indicated that B cell maturation takes place in the spleen 6,11, yet immature B cells have also been detected in LNs 46. B cell maturation outside of the spleen has not been investigated, but the relatively well-sustained numbers of B lymphocytes in splenectomized individuals raises the question of potential alternative sites for B cell maturation.

BAFF is a pivotal survival factor during B cell maturation, a process which is as yet poorly understood. Our results suggest that excessive BAFF-induced survival might induce the breakdown of immune tolerance at a critical stage of peripheral B cell maturation. Therefore, the highly selective recruitment of immature B cells into the mature B cell pool might be the result of active mechanisms of self-tolerance, possibly controlled by BAFF.

We would like to thank Susan Kalled, Liz Musgrove, Kieran Scott, Charles Mackay, Christine Ambrose, Phillip Hodgkin, Rita Carsetti, and Alison Saunders for critical reading of the manuscript. We also thank Jenny Kingham, Eric Schmied, Anthony Chaplin, and Julie Ferguson for help with maintenance of the colony of BAFF Tg mice, and Stephen A. Woodcock for help with histology and the original two lines of BAFF Tg mice. We thank Robert Wadley and the University of New South Wales for help with cell sorting.

This work was supported by the Glazebrook Trust and Biogen Incorporated.