An Essential Role for BAFF in the Normal Development of B Cells Through a BCMA-Independent Pathway

See allHide authors and affiliations

Science  14 Sep 2001:
Vol. 293, Issue 5537, pp. 2111-2114
DOI: 10.1126/science.1061964


The B cell activating factor BAFF (BlyS/TALL-1/zTNF4) is a tumor necrosis factor (TNF)–related ligand that promotes B cell survival and binds to three receptors (BCMA, TACI, and the recently described BAFF-R). Here we report an absolute requirement for BAFF in normal B cell development. Examination of secondary lymphoid organs from BAFF-deficient mice revealed an almost complete loss of follicular and marginal zone B lymphocytes. In contrast, mice lacking BCMA had normal-appearing B lymphocyte compartments. BAFF therefore plays a crucial role in B cell development and can function through receptors other than BCMA.

B cell development is a temporally and spatially regulated process that begins in the bone marrow, where common lymphoid progenitors differentiate into pro–B cells, pre–B cells, and later, B lymphocytes (1, 2). After B cell receptor expression, rearrangement, and deletion of autoreactive clones (3, 4), a fraction of the cells migrate to secondary lymphoid organs. There, they may encounter antigen and undergo clonal selection and deletion in a complex series of steps (5, 6). In the spleen, newly formed (B220+, IgMhi) cells acquire more mature phenotypes with down-regulation of IgM and up-regulation of molecules including CD21, CD23, and IgD.

A new TNF ligand thought to play a central role in B cell development is BAFF (7), also known as BlyS, TALL-1, THANK, or zTNF4 (8–11). BAFF specifically binds to B lymphocytes, costimulates their proliferation, and promotes the survival of splenic B cells in vitro (12). Overexpression studies in mice established that BAFF can cause a systemic lupus erythematosis (SLE)–like syndrome (13,14). Because circulating levels of BAFF are elevated in some lupus patients (15, 16), there has been great interest in further understanding the possible roles of this ligand in human disease and B cell development.

Three TNF receptors are known to specifically bind BAFF. The B cell maturation antigen, BCMA (17), and a transmembrane activator and calcium-modulating cyclophilin ligand interacting protein, TACI (18), have been previously described. The BAFF receptor, BAFF-R, was only recently identified (19). Derivatives of BCMA and TACI have been implicated in reducing peripheral B cell numbers, prolonging survival in mouse SLE models (11, 20), and inhibiting humoral immune responses (21). However, their mechanism of action has been unclear for several reasons. (i) In contrast to BAFF-R, both BCMA and TACI have also been identified as receptors for the more broadly expressed TNF ligand APRIL (22, 23). (ii) Only low levels of BCMA on developing B lymphocytes have been demonstrated (24). (iii) TACI was originally reported to be present not only on B cells, but also on activated T lymphocytes (18).

To better determine the importance and mechanism of BAFF function, we generated mice that were deficient either in BAFF or in BCMA (25). All were outwardly normal and survived to at least 6 to 8 months of age without unusual morbidity. At necropsy, all major organs including thymus, spleen, and lymph node were present, although average spleen weights of BAFF−/− animals were significantly reduced (26).

Immunohistochemical studies of secondary lymphoid tissues from BAFF−/− mice revealed severe losses of B220+cells (Fig. 1A) (27). In the spleen, those that remained were in regions directly adjacent to the periarterial T cell zones, which appeared largely normal. Simultaneous detection of MOMA-1 and B220 (Fig. 1B) suggested that BAFF−/− mice have significantly fewer marginal zone and follicular B cells than wild-type animals; this was confirmed by fluorescence-activated cell sorting (FACS) analysis (see below). By contrast, staining for other antigen-presenting cell markers including CD11c, MAdCAM-1, and FDC-M1 demonstrated essentially normal splenic architecture outside the B cell compartment (26). In lymph nodes from mutant mice, the few remaining B220+ cells tended to localize at the subcapsular locations normally occupied by follicular B cells (Fig. 1C).

Figure 1

Immunohistochemistry. Micrographs showing wild-type (WT) or BAFF−/− (BAFF KO) tissue sections, as indicated at left. Antigens detected are shown below the panels (27). (A) Spleen B cells were stained for B220 (brown); T cells were stained for CD4 and CD8 (blue). (B) Spleen metallophilic macrophages were stained for MOMA-1 (brown, ring-shaped structures); B cells were stained for B220 (blue). (C) Lymph node B cells were stained for B220 (brown); T cells were stained for CD4 and CD8 (blue).

FACS analysis revealed very marked reductions in specific populations of peripheral B cells in BAFF−/− animals. Staining for mature follicular (CD21hi, CD23hi) and marginal zone (CD21hi, CD23lo) splenocytes showed almost complete loss of these cells (Fig. 2A). The remaining B lymphocytes mostly exhibited staining similar to that of recently described T1 transitional B cells (Fig. 2B). Although these cells (CD21lo, IgMhi) were normal in number, there were almost no cells of a T2 phenotype (CD21hi, IgMhi) thought to represent the immediate next stage of B cell development (6,28). Other hematopoietic cell lineages were not affected in BAFF−/− mice. Bone marrow cells were present in nearly normal numbers; only the normal population of recirculating mature B lymphocytes was absent (26). In addition, there were no statistically significant abnormalities in the proportions of CD4-, CD8-, and CD69-expressing cells in thymus (20) or spleen (Table 1).

Figure 2

FACS analysis. Cells from 7-week-old mice of the genotypes shown above each panel were stained for the indicated antigens, and then analyzed after electronic scatter gating. wt: wild-type (mixed 129/Sv, C57BL/6) control cells, which exhibited staining similar to cells from the A/J strain (parental to A/WySnJ). The numbers within each panel are percentages of gated events. (A) Reduced follicular B cells (CD21+, CD23+; circular regions) and marginal zone B cells (CD21+, CD23lo; rectangular regions) from BAFF−/− and A/WySnJ spleens are compared to wild-type and BCMA−/− cells. (B) Reduced T2 B cells (CD21+, IgMhi; upper rectangular regions) and preserved T1 B cells (CD21, IgMhi; lower rectangular regions) from BAFF−/− and A/WySnJ spleens are compared to wild-type and BCMA−/− cells. (C) Preserved B1 B cells (B220+, CD23) and reduced B2 B cells (B220+, CD23+) in peritoneal lavage from BAFF−/− mice are compared to wild-type cells. After gating on B1 cells, CD5 and IgM staining reveals B1a and B1b cells, as indicated. Results represent stainings from eight mice of each genotype.

Table 1

B and T cells from wild-type and BAFF−/−mice. Cell numbers are ×10−6. Cells from six wild-type and six BAFF−/− mice were counted and analyzed by FACS to enumerate the indicated populations of cells. B220+ B cells include the following categories: follicular = CD21+, CD23+; marginal zone = CD21+, CD23lo; newly formed = CD21, CD23; T1 = IgMhi, CD21; T2 = IgMhi, CD21+; pro-B = CD43+, IgM; pre-B = CD43, IgM+; recirculating B = IgD+, IgMlo; immature B = IgD, IgM+. P values were determined by Student's t test; *P ≤ 0.05.

View this table:

These findings in BAFF-deficient mice bear marked similarities to previous descriptions of the mutant A/WySnJ mouse strain (29). These mice were reported to exhibit an ∼10-fold loss of splenic B cells (30) and were recently shown to be mutant in the newly identified BAFF receptor, BAFF-R (19). We therefore directly compared their splenocytes to BAFF−/−splenocytes, using FACS analysis (Fig. 2, A and B). Follicular and marginal zone B cells from homozygous A/WySnJ mice were substantially reduced as compared with either mixed wild-type (129/Sv, C57BL/6) or parental (A/J) animals of identical age, but less severely affected than in mice lacking BAFF. By contrast, cells from BCMA−/− mice were similar in number and in staining to wild-type controls (Fig. 2) (26).

Lymph node and peritoneal B cells obtained from BAFF−/−mice exhibited further abnormalities consistent with those observed in their spleens. Although lymph nodes were present and were readily identified, they exhibited an ∼10-fold reduction in B cell numbers (20). Of note, peritoneal lavage of these animals returned normal-appearing B1 B cells (31) but included markedly fewer cells of the B2 lineage (B220+, CD23hi) (Fig. 2C). Again similar to the A/WySnJ strain, mice lacking BAFF exhibited no abnormalities in the ratios of B1a and B1b cells recovered.

To establish the function of the remaining B cells in BAFF−/− mice, we determined their serum antibody levels at baseline and in response to both T-dependent and T-independent antigens (32). The results from assays of resting wild-type, heterozygous, and knockout mouse sera are shown in Fig. 3A. Homozygous BAFF knockout mice exhibited a profound reduction (10-fold) in total serum immunoglobulin and in each subclass with the exception of immunoglobulin A (IgA), which was only moderately reduced in knockout animals. Also of note was the finding that heterozygous (BAFF+/−) mice exhibited a reproducible, roughly twofold baseline reduction in IgG subclasses and IgM present in their sera (26).

Figure 3

Serum immunoglobulin (Ig) levels. (A) Resting total Ig levels from BAFF−/− (KO), BAFF+/− (HET), and wild-type (WT) mice were determined by enzyme-linked immunosorbent assay (32) and are shown for individual mice. Horizontal bars indicate the average for each genotype. (B) The NP-specific antibody responses of BAFF−/− and wild-type animals to a T-dependent antigen, NP-KLH, is shown. Bars represent the average of four mice for each genotype, and the error bars indicate +1 SD. (C) TNP-specific antibody responses to a T-independent antigen, TNP-Ficoll, are shown as described for (B).

Reduced serum antibody responses to intraperitoneal immunization with a T-dependent antigen, NP-KLH (33), were also apparent in BAFF−/− mice. Thus, BAFF knockout mice failed to produce a measurable specific antibody response at early time points, although they were able to mount a very small detectable response at later times (Fig. 3B) (20). Similar results were obtained after immunization of additional mice with the T-independent antigen TNP-Ficoll (Fig. 3C).

To better understand BAFF-dependent signaling, we also generated mice deficient in BCMA, one of three proposed receptors for BAFF. As for BAFF−/− mice, the bone marrow of BCMA−/−animals contains normal proportions of B cells and their readily identifiable precursors. However, they exhibited neither significant loss of splenic B cells (Fig. 2A) (26), nor loss of the T2 transitional B cells that occurred in mice lacking BAFF and in A/WySnJ mice.

Targeted mutation of the BAFF locus established that the encoded protein is necessary to support normal B cell development and function in vivo. Analysis of BAFF−/− spleens by both FACS and immunohistochemistry revealed loss of follicular and marginal zone B cells. Our results are consistent with the suggestion that T2 transitional B cells are a primary target for BAFF action (12). Alternatively, T1 (B220+, IgMhi, CD21lo) cells may require BAFF for progression to the T2 stage, and then beyond.

A notable feature of BAFF deficiency is its minimal effect on the bone marrow and on peritoneal populations of B1 cells. The bone marrow's nearly normal percentage of immature (B220+, CD43hi) cells suggests that marrow B lymphopoiesis proceeds in a BAFF-independent manner. Similarly, the peritoneal population of B1 B cells is far less sensitive to BAFF deficiency than are peritoneal B2 lymphocytes, consistent with both major views of B1 cell development (26, 34, 35).

These results illuminate several important aspects of the mechanism by which BAFF functions in vivo. The phenotype of BAFF-deficient mice is similar to, but more severe than, that of a previously identified mutant mouse strain, A/WySnJ (Fig. 2) (29). A/WySnJ mice express a mutated form of BAFF-R, a recently cloned member of the TNF receptor superfamily of proteins that binds only to BAFF (19). The phenotypes of BAFF−/− and A/WySnJ mice, combined with the binding specificity of BAFF-R, strongly suggest that BAFF primarily acts to promote survival and maturation of B2 cells through the newly identified receptor. This hypothesis is consistent with the more subtle B cell phenotypes that we and others have observed in animals lacking BCMA and TACI (36,37). The persistence of some splenic B cells in A/WySnJ mice might result from strain-specific effects, redundant signaling through TACI and BCMA, or mutated BAFF-R that reaches the surface of the cells but cannot properly signal.

It has been suggested that both ligand (BAFF and APRIL) and receptor (BCMA and TACI) redundancy may support B cell development (21). Our data demonstrate that there is no significant functional BAFF redundancy among TNF ligands in supporting B cell survival: APRIL cannot compensate for BAFF by signaling through TACI, BCMA, or any other receptor in BAFF-deficient mice.

Numerous details of the mechanism of BAFF signaling remain to be elucidated (26). In addition to a central role in regulating splenic B cell development shown here, BAFF could have further importance to events including secondary immune responses, memory, and mature B cell survival that occur after the T2 developmental stage. Also, crosses of BAFF-deficient mice with animals prone to other B cell disorders (e.g., malignancies) may soon validate clinical applications involving the BAFF pathway beyond those already envisioned in autoimmune diseases.

  • * Present address: National Jewish Medical and Research Center, 1400 Jackson Street, Denver, CO 80206, USA.

  • To whom correspondence should be addressed. E-mail: martin_scott{at}


View Abstract

Navigate This Article