BAFF-R signaling aids the differentiation of immature B cells into transitional B cells following tonic BCR signaling

Abstract

BAFF is an important pro-survival cytokine for mature B cells. However, previous studies have shown that the BAFF receptor, BAFF-R, is already expressed at the immature B cell stage, and that the pro-survival protein Bcl-2 does not completely complement the B cell defects resulting from the absence of BAFF-R or BAFF. Thus, we hypothesized that BAFF also functions to aid the differentiation of non-autoreactive immature B cells into transitional B cells and to promote their positive selection. We found that BAFF-R is expressed at higher levels on non-autoreactive than autoreactive immature B cells and that its expression correlates with that of surface IgM and with tonic BCR signaling. Our data indicate that BAFF-R signaling enhances the generation of transitional CD23⁻ B cells in vitro by increasing cell survival. In vivo, however, BAFF-R signaling is dispensable for the generation of CD23⁻ transitional B cells in the bone marrow, but is important for the development of transitional CD23⁻ T1 B cells in the spleen. In addition, we show that BAFF is essential for the differentiation of CD23⁻ into CD23⁺ transitional B cells both in vitro and in vivo through a mechanism distinct from that mediating cell survival, but requiring tonic BCR signaling. In summary, our data indicate that BAFF-R and tonic BCR signals cooperate to enable non-autoreactive immature B cells to differentiate into transitional B cells and to be positively selected into the naïve B cell repertoire.

Introduction

Cytokines act on cells of the immune system to regulate and coordinate their survival, differentiation and activity. Over the last decade, the cytokine BAFF (also known as BLyS) has been defined as a critically important and specific factor that promotes the survival of transitional T2, follicular, and marginal zone B cells (1–6). The ability of BAFF to promote B cell survival is mediated specifically through its binding to BAFF-R (also known as BR3), as indicated by the similar phenotypes of mutant mice lacking BAFF or BAFF-R, and the differences from those lacking other BAFF receptors such as TACI and BCMA (5, 7–9). Moreover, evidence indicates that BAFF-R signaling mediates B cell survival by preventing TRAF2/TRAF3 from inhibiting the alternative NF-κB pathway (10–12).

In recent years it has been established that the B cell antigen receptor (BCR) generates a ligand-independent tonic signal that is also important for the survival of B cells (13–15). This signal, moreover, synergizes with those of cytokine receptors such as IL-7 receptor (IL-7R) and BAFF-R, to promote the survival of B lymphocytes at different stages of differentiation. Specifically, during early B cell development the pre-BCR and the IL-7R synergize to promote survival and proliferation of pre-B cells (16, 17). In naïve mature B cells both tonic BCR and BAFF-R signals are necessary for B cell survival, as demonstrated by the lack of mature B cells in the absence of either signal (9, 13). Immature B cells are the first developing B cells to express a mature BCR, in the form of IgM, on the cell surface. While immature B cells still express the IL-7 receptor, they appear to respond minimally to IL-7 (18). In the bone marrow, immature B cells undergo a selection process that eliminates self-reactive specificities and generates the naïve mature B cell repertoire in the periphery (reviewed by (19–21)). Interestingly, no cytokine so far has been found to be necessary for the survival of immature B cells during these selection events, nor for aiding their differentiation into peripheral transitional and mature B cells.

Newly developed IgM⁺IgD⁻CD21⁻CD23⁻ bone marrow immature B cells undergo differentiation into mature peripheral B cells through an intermediate step in development called the transitional stage (22). Transitional B cells are found in both bone marrow and spleen and retain high levels of CD24 and CD93 expression seen on immature B cells, but also display variable expression of IgD, CD21 and CD23 (22–25). In fact, splenic transitional B cells have been further classified into three sub-populations based on differential level of expression of these markers (26, 27): T1 (IgM^highIgD^lowCD21^−/lowCD23⁻), T2 (IgM^highIgD⁺CD21⁺CD23⁺), and T3 (IgM^lowIgD⁺CD21⁺CD23⁺), with T1 cells being developmental precursors of both T2 and T3 cells. While a similarly precise definition of transitional B cells in the bone marrow is lacking, cells resembling splenic transitional T1 and T2 B cells are present also in this tissue (22, 25, 26). B cell dependency on BAFF is currently considered to begin at the transitional T2 B cell stage of development in the spleen (4, 5, 9), a stage that displays a noteworthy increase in the expression of BAFF-R (28). However, development of BAFF-R-deficient B cells has been only minimally assessed in competition with wild-type B cells (29), which is a more stringent test of B cell development (30, 31). In fact, some evidence suggests a potential role for BAFF and BAFF-R at B cell stages earlier than T2. For instance, a small reduction in the number of transitional T1 B cells was observed in the spleen of BAFF- and BAFF-R-mutant mice, although this difference was not statistically significant (5, 9, 32). Additionally, BrdU incorporation studies have demonstrated a reduced export of bone marrow immature B cells into the spleen of BAFF-R-deficient mice (7). BAFF was also suggested to promote either the in vitro differentiation of bone marrow immature B cells, or to increase the survival of newly generated transitional B cells (10). Overall, these data suggest that BAFF may also function at the immature and transitional T1 B cell stages.

It has been shown both in mice and humans that 30–50% of the newly generated immature B cells are non-autoreactive and presumably undergo positive selection into the spleen, while as many as 50–70% are autoreactive and potentially subjected to mechanisms of negative selection (33, 34). Interestingly, other studies have shown that only approximately half of wild-type immature bone marrow B cells express BAFF-R and bind BAFF (28, 35), suggesting that BAFF may play a role in the process of immature B cell selection. In the studies presented here we tested the hypothesis that BAFF-R is expressed by non-autoreactive, but not autoreactive, immature B cells. In addition, we evaluated whether BAFF-R expression and signaling promote the differentiation of non-autoreactive immature B cells into transitional B cells, thus contributing to the positive selection process.

We found that BAFF-R is indeed expressed at higher levels on non-autoreactive than autoreactive immature B cells, and that BAFF aids the positive selection and differentiation of non-autoreactive immature B cells only when these cells also receive sufficient tonic BCR signals. Moreover, we show that BAFF-R is important for the differentiation of immature B cells into CD23⁻ as well as CD23⁺ transitional B cells, but potentially via different molecular pathways.

Materials and Methods

Mice

The 3-83Igi, H-2^d (Igh^3-83/3-83Igk^3-83/3-83H-2^d/d) (36, 37), 3-83Igi, Rag1^−/−,H-2^b (Igh^3-83/3-83Igk^3-83/3-83Rag1^−/−H-2^b/b) (37, 38), B1-8/3-83Igi, H-2^b or H-2^d (Igh^B1-8/3-83Igk^3-83/3-83) (38), 3-83Igi-low (Igh^3-83/3-83Igk^3-83/3-83H-2^d/dmb-1^{-/mb-1-mEGFPinv}) (39), and 3-83Igi-mb1^−/− (Igh^3-83/3-83Igk^3-83/3-83H-2^d/dmb-1^−/−) (39) mice have been previously described and were all on a BALB/c genetic background. BAFF-R^−/− (tnfrsf13c^−/−) mice on a C57BL/6 genetic background have been previously described (9) and were obtained from Jackson Laboratories. A/WySnJ, C57BL/6, C57BL/6.Ly5.1 (C57BL/6 mice congenic for Ly5.1, also known as B6SJL), BALB/c and CB17 mice were purchased from Jackson Laboratories and either used immediately or bred in our facility. All mice were bred and maintained in specific pathogen free rooms at the Biological Resource Center at NJH, Denver, CO. All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC).

Flow cytometry and antibodies

Single cell suspensions were stained with fluorescent or biotinylated monoclonal antibodies against B220 (RA3-6B2), CD21 (7G6), CD23 (B3B4), CD24 (M1/69), Ly5.2 (104.2.1), CD90.1/Thy1.1 (OX-7), CD43 (S7), CD2 (RM2-5), IgM^a (DS-1), IgM^b (AF6-78), IgD (11-26c-2a), that were purchased from either BD Pharmingen or eBioscience. Biotinylated goat polyclonal anti-mouse BAFF-R antibodies were purchased from R&D. Antibodies against IgM (R33-24.12, (40)), IgD (1.3–5, (41)), and CD19 (1D3, (42)) were generated in house. Biotinylated antibodies were revealed with fluorochrome-conjugated streptavidin (Molecular Probes). Propidium iodide (PI, 1.25 μg/ml) was added in some experiments to exclude dead cells. Stained cells were run on a Facscan (BD), Facscalibur (BD) or a Cyan analyzer (DakoCytomation). Flow cytometric analyses were performed on live B220⁺ or CD19⁺ lymphocytes based on incorporation of PI and/or forward and side scatter with FlowJo software (TreeStar).

Cell sorting and microarray analysis

Bone marrow cells were isolated from femurs, tibias and pelvises of 3-83Igi, H-2^d and 3-83Igi, Rag1^−/−, H-2^b groups of mice (8 mice per group), pooled and stained with anti-CD43, anti-IgD, anti-CD23, and anti-B220 antibodies. Stained cells were sorted for B220⁺CD23⁻CD43⁻IgD⁻ (immature B) cells at ~7000 events/second using a MoFlo cell sorter (Beckman). Sorted cells (approximately 1–5×10⁷) were analyzed for purity by comparing to unsorted samples using a FACSCalibur (BD Biosciences). The purity was >95% in all sorted cell samples. Cell pellets were flash frozen in liquid nitrogen for 15 sec, and stored at −80°C until shipment to Miltenyi Biotech. RNA preparation, amplification, and hybridization to Agilent Whole Mouse Genome Oligo Microarrays 4×44K were performed by Miltenyi Biotech. The microarray data have been deposited in NCBI’s Gene Expression Omnibus (GEO, (43)) and are accessible through GEO Series accession number GSE22802 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE22802).

Retroviral constructs and production of retroviral particles

The murine tnfrsf13c cDNA (encoding BAFF-R) was amplified with primers BAFFR-5-NotI (TATGCGGCCGCTGTCCCAGCTGCATGAG) and BAFFR-3-SalI (CGCGTCGACGTTCCAGCCTCCACTGC) and AccuTaq Polymerase (SIGMA) from cDNA synthesized from BALB/c RNA. The PCR product was cloned between NotI and SalI in the retroviral pMSCV-IRES-GFP vector (44) to generate the pMSCV-BAFFR-IRES-GFP plasmid. The pMSCV-Flag-Bcl2-IRES-Thy1.1 (45) and pMSCV-GFP-IRES-hN-RasG12D (39) retroviral vectors were described before. All retroviral vectors encoded replication deficient viruses. Retroviruses were produced in Phoenix cells as previously described (39).

Generation of bone marrow chimera and retrogenic mice

Recipient mice were irradiated as described (46). For non-retrogenic bone marrow chimeras, donor bone marrow cells depleted of red blood cells were mixed at the indicated ratios and a total of 1–2×10⁶ cells in 100 μl PBS were injected in a tail vein of each recipient. When generating retrogenic mice, donor mice were injected i.p. with 3.75 mg 5-FU 4 d prior to bone marrow harvest. Bone marrow cells isolated from 5-FU-treated donor mice were cultured overnight at 1×10⁶ cells/ml in complete IMDM (Gibco) (100 U/ml penicillin, 100 μg/ml streptomycin, 0.05 mM β-mercaptoethanol, 2 mM Glutamax, 0.1 mM non-essential amino acids, 10% heat-inactivated FBS) supplemented with recombinant cytokines (IL-3, IL-6, and stem cell factor, a gift of Dr. Yosef Refaeli, NJH). After 24 h, the cells were resuspended in 0.5 ml complete IMDM, 1 ml Phoenix cell supernatant containing retrovirus particles, 3.2 μg (2.1 μg/ml final) polybrene, and recombinant cytokines and centrifuged at 1124 × g for 1.5 h at RT. Cells were cultured overnight in fresh complete medium with recombinant cytokines and transduced a second time as described above. Following transduction, recipient mice received a total of 1–5×10⁵ cells in 100 μl PBS via tail vein injection. Mice were analyzed 5–8 weeks later.

In vitro B cell differentiation and immature B cell transduction

Bone marrow cells were cultured in complete IMDM in the presence of IL-7 for 3–4 days to enrich for IgM⁺ immature B cells. Cells were then washed twice with PBS to remove IL-7 and plated at 2–4×10⁶ cells/ml with 10 ng/ml recombinant murine BAFF (R&D Systems) or as otherwise indicated. On subsequent days, cells were stained with antibodies against B220, CD21, CD23, IgM and IgD to determine differentiation state. For immature B cell transduction, bone marrow cells were cultured in IL-7 as above for 2–3 days, and subsequently resuspended at 4–6×10⁶ cells/ml in a cocktail consisting of 0.5 ml complete IMDM, 1 ml retroviral supernatant, 3.2 μg polybrene (2.1 μg/ml final) and IL-7. Cells were centrifuged at 1124 × g for 1.5 h at RT and cultured overnight in fresh complete medium containing IL-7.

Statistical analysis

Statistical significance was calculated with Graphpad Prism software using a one-tailed Students t-test with equal variance (using Welch’s correction when appropriate); P < 0.05 was considered significant. Data are represented as means ± standard deviation (SD). N.S. = non significant.

Results

BAFF-R is differentially expressed in autoreactive and non-autoreactive immature B cells

To test the hypothesis that BAFF-R is differentially expressed by autoreactive and non-autoreactive developing B cells, we analyzed immature B cells from mouse models that generate a monoclonal population of either non-autoreactive or autoreactive immature B cells (Table S1). The 3-83Igi mice have been previously established as a model to analyze development and selection of immature B cells with pre-defined antigen reactivity (37, 47). In these heavy and light chain immunoglobulin (Ig) gene targeted mice, all B cells develop expressing the 3-83 Ig. When maintained on an H-2^d genetic background, newly developed 3-83Ig⁺ (3-83H/3-83κ) immature B cells are non-autoreactive and develop into mature B cells that largely (>90%) retain the 3-83 specificity (36, 37, 48). In contrast, newly developed 3-83Ig⁺ B cells from 3-83Igi, H-2^b mice on a Rag1^−/− genetic background (3-83Igi, Rag1^−/−, H-2^b) represent a uniform population of autoreactive immature B cells because of the high avidity binding of 3-83 Ig to the MHC class I protein K^b (37, 47) and the absence of receptor editing due to the deficiency of Rag1 (37, 38). Bone marrow cells from non-autoreactive 3-83Igi, H-2^d and autoreactive 3-83Igi, Rag1^−/−, H-2^b mice were analyzed for B220, IgD and BAFF-R expression. Fig. 1A shows BAFF-R expression on B220⁺IgD⁻ gated lymphocytes, which mostly represent immature B cells in 3-83Igi mice as pro-B cells are relatively few, and Ig heavy and light chain gene targeted mice do not generate pre-B cells (47). As shown in the histogram representation, non-autoreactive immature B cells uniformly expressed BAFF-R on the cell surface. The specificity of the anti-BAFF-R antibodies was indicated by the absence of staining on B220⁻ non-B lineage cells (Fig. 1A) and BAFF-R-deficient B cells (data not shown). In accordance with our hypothesis, we found that BAFF-R was not detected, or detected only at very low levels, on the surface of autoreactive immature B cells (Fig. 1A, histogram). On average, the expression of BAFF-R on autoreactive immature B cells was approximately 3.8-fold (±1.1; n=3) lower than that detected on non-autoreactive immature B cells, and only slightly above the background level observed on non-B lineage cells. Differential expression of BAFF-R by immature B cells was also confirmed at the gene transcription level. Non-autoreactive and autoreactive immature B cells were sorted from bone marrow of 3-83Igi, H-2^d and 3-83Igi, Rag1^−/−, H-2^b mice, respectively, as B220⁺CD23⁻CD43⁻IgD⁻ cells. After RNA isolation and cDNA preparation, their cDNAs were hybridized to Agilent whole mouse genome microarrays. Transcript levels of tnfrsf13c, the gene encoding BAFF-R, were found to be approximately 4-fold higher in non-autoreactive than autoreactive immature B cells (Fig. 1B), and in accordance with the observed differences in surface protein levels. Thus, autoreactive and non-autoreactive immature B cells differentially express BAFF-R at both transcript and protein levels.

Figure 1. BAFF-R is differentially expressed by non-autoreactive and autoreactive immature B cells.

(A) Bone marrow cells from non-autoreactive 3-83Igi, H-2^d (non-aut) and autoreactive 3-83Igi, Rag1^−/−, H-2^b (aut) mice were analyzed by flow cytometry for expression of B220, IgD and BAFF-R. Dot plots show expression of B220 and IgD on live lymphoid cells. Histograms display BAFF-R levels on gated B220⁺IgD⁻ cells (representing mostly immature B cells) from 3-83Igi, H-2^d (non-aut, intact line) and 3-83Igi, Rag1^−/−, H-2^b (aut, dashed line) mice, and on B220⁻ non-B cells (filled histogram). Data are representative of three mice per group analyzed in three independent experiments. (B) Bone marrow B220⁺IgD⁻CD23⁻CD43⁻ immature B cells from two groups each of non-autoreactive 3-83Igi, H-2^d (black bar) and autoreactive 3-83Igi, Rag1^−/−, H-2^b (gray bar) mice were sorted via Moflo to >95% purity and subjected to whole genome microarray analyses. The relative tnfrsf13c mRNA levels (arithmetic mean of normalized processed signals from two groups of mice) in the immature B cell samples are shown. Differences between the signals had P values <10⁻¹².

BAFF-R expression correlates with IgM expression and tonic BCR signaling in immature B cells

The low expression of BAFF-R observed in autoreactive immature B cells could indicate suppression by autoantigen-mediated BCR signaling, or lack of induction in the absence of tonic BCR signals. To assess these two possibilities we analyzed BAFF-R levels on immature B cells from additional mouse strains. The B1-8/3-83Igi, H-2^b mouse strain generates immature B cells each co-expressing autoreactive (3-83H/3-83κ) and non-autoreactive (B1-8H/3-83κ) BCRs (Table S1). As a consequence, these dual BCR-expressing B cells receive both autoantigen-mediated and tonic BCR signals (38). In contrast, when the B1-8/3-83Igi mice are on an H-2^d background, both BCRs are non-autoreactive and B cells receive only tonic BCR signals (Table S1). When assessed by flow cytometry we found that IgD⁻ immature B cells that co-express autoreactive and non-autoreactive BCRs (B1-8/3-83Igi, H-2^b) displayed surface BAFF-R at levels that were 2.3-fold (±0.6, n=3) reduced relative to those of non-autoreactive B1-8/3-83Igi, H-2^d cells (Fig. 2A, bottom panel). Importantly, BAFF-R levels were higher on dual BCR-expressing (B1-8/3-83Igi, H-2^b) than single BCR-expressing (3-83Igi, Rag1^−/−, H-2^b) autoreactive immature B cells, but nevertheless lower than those on non-autoreactive (3-83Igi, H-2^d and B1-8/3-83Igi, H-2^d) immature B cells. Also of significance, IgD⁻ B1-8/3-83Igi, H-2^b immature B cells express, on average, half the amount of surface IgM when compared with dual BCR-expressing non-autoreactive immature B cells from B1-8/3-83Igi, H-2^d mice (Fig. 2A, top panel, and (38)), a level potentially resulting in diminished tonic BCR signaling (39). These data suggest that tonic, and not autoantigen-mediated, BCR signals regulate BAFF-R levels on immature B cells. To test this idea further we used another mouse strain, 3-83Igi-low (Table S1), that is a hypomorphic Ig-α strain in which 3-83Ig⁺ non-autoreactive immature B cells express reduced amounts of surface IgM (Fig. 2B, top panel, and (39)). We found that BAFF-R surface levels were, on average, 1.6-fold (±0.2, n=4) reduced on 3-83Igi-low (BCR-low) relative to 3-83Igi, H-2^d (BCR-normal) IgD⁻ immature B cells (Fig. 2B, bottom panel). Thus, immature B cells from both B1-8/3-83Igi, H-2^b and 3-83Igi-low mice express BAFF-R at levels intermediate between those of non-autoreactive (B1-8/3-83Igi, H-2^d or 3-83Igi, H-2^d) and autoreactive single BCR-expressing (3-83Igi, Rag1^−/−, H-2^b) immature B cells. Additionally, and in agreement with previous reports (28, 35), BAFF-R expression is further upregulated as cells mature (Fig. 2A, spleen). These results suggest that BAFF-R expression on immature B cells is not inhibited by antigen-mediated signals, but rather generally correlates with surface levels of IgM and, potentially, tonic BCR signaling.

Figure 2. Expression of BAFF-R correlates with that of surface IgM and tonic BCR signaling.

(A) Flow cytometric analysis of IgM (top) and BAFF-R (bottom) expression on B220⁺IgD⁻ bone marrow immature B cells that carry two types of non-autoreactive BCRs (B1-8/3-83Igi, H-2^d mice, dual-non-aut, black intact line) or autoreactive and non-autoreactive BCRs (B1-8/3-83Igi, H-2^b mice, dual-aut, black dashed line). Expression on B220⁻ non-B cells (non-B, filled histograms) is shown as a negative control. Expression of BAFF-R on spleen CD19⁺ B cells from B1-8/3-83Igi, H-2^d mice (dual-non-aut, SP, gray line) is shown as reference (spleen B cells from BALB/c mice had identical levels; data not shown). Data are representative of three mice per group collected in three independent experiments. (B) Flow cytometric analysis of IgM (top) and BAFF-R (bottom) expression on B220⁺IgD⁻ bone marrow immature B cells that express normal levels (3-83Igi, H-2^d mice, Normal, intact line) or low levels (3-83Igi-low mice, Low, dashed line) of non-autoreactive BCRs. Expression on B220⁻ non-B cells (non-B, filled histograms) is shown as a negative control. Data are representative of four mice per group collected in three independent experiments. (C) Representative flow cytometric analysis of BAFF-R expression on immature B cells generated in culture with IL-7 and then transduced or not. Shown are BAFF-R levels on BCR-normal non-transduced cells from 3-83Igi, H-2^d mice (Normal, filled histogram), and on BCR-low GFP control transduced (Low+GFP, dotted line) and N-RasD12 transduced (Low+RasD12, black line) cells from 3-83Igi-low mice. Immature B cells were gated as B220⁺IgD⁻ (Normal) or B220⁺IgD⁻GFP⁺ (Low+GFP and Low+RasD12). (D) Geometric mean fluorescence intensity (MFI, ± SD, n = 4, from 4 independent experiments) of BAFF-R expression on immature B cells analyzed as described in (C). P-values of <0.05 (*) and <0.005 (**) are indicated. N.S. = not significant.

Recently we have shown that BCR-low immature B cells display reduced levels of tonic BCR signaling and that this deficiency can be complemented by the expression of the constitutively active form of N-Ras, N-RasD12 (39). Thus, to examine further whether tonic BCR signaling regulates BAFF-R levels we tested if N-RasD12 expression could re-establish normal BAFF-R expression on 3-83Igi-low immature B cells. BCR-low bone marrow immature B cells from 3-83Igi-low mice were enriched and transduced in cultures containing IL-7 as previously described (39). Cells were transduced with retroviruses encoding either N-RasD12 or GFP only control and assayed for BAFF-R expression by flow cytometry thereafter. We found that expression of N-RasD12 augmented BAFF-R levels on the surface of BCR-low IgD⁻ immature B cells, to a level similar to that of BCR-normal (3-83Igi, H-2^d) immature B cells, while gfp control transduction did not (Figs. 2C,D).

Overall these data indicate that BAFF-R is expressed on bone marrow IgM⁺IgD⁻ immature B cells and its expression level correlates with that of surface IgM and tonic BCR signaling.

BAFF augments the generation of transitional B cells in vitro

It is known that cytokine and antigen receptor signaling can synergize to guide lymphocyte survival and differentiation. To test whether BAFF-R has a function in immature B cell development, we investigated its role in the differentiation of immature B cells into transitional B cells using an in vitro system that we have recently developed ((39), and Fig. 3A). In this system, bone marrow IgM⁺IgD⁻CD21⁻CD23⁻ immature B cells are generated in IL-7 cultures and then re-cultured in the absence of IL-7 and in the presence of BAFF where they differentiate into cells that retain high levels of CD24 and variably express IgD, CD21 and CD23 ((39), and data not shown). Although it was proposed that BAFF regulates expression of CD21 and CD23 (28), most data indicate that it does not control CD23 (or IgD) expression (9, 32, 49), and only slightly affects that of CD21 (50). Thus, IgD, CD21, and CD23 can still be used as developmental markers in these analyses. Transitional B cells in the bone marrow have yet to be fully characterized into clear subpopulations, although both newly formed transitional T1- and T2-like B cells have been described, ((22, 25, 26), and Fig. S1). For simplicity, in this study we classify immature B cells as IgM⁺IgD⁻CD21⁻CD23⁻, and transitional B cells in bone marrow and spleen as cells expressing high levels of IgM and CD24, and low levels of IgD and CD21, with additional CD23 and higher IgD expression discriminating transitional T2 from T1 in spleen and T2-like from T1-like in bone marrow.

Figure 3. BAFF augments the differentiation of immature B cells into transitional B cells in vitro.

(A) Schematic of the system to study the differentiation of immature B cells in vitro. Briefly, bone marrow cells are cultured with IL-7 for 3–4 days and then transferred in cultures with or without BAFF for two days. (B) Representative flow cytometric analysis of CD21, CD23 (dot plots) and IgD (histograms) expression on 3-83Igi, H-2^d non-autoreactive immature B cells that express normal levels of BCR, at the beginning (day 0) and end (day 2) of culture with or without BAFF. Cells from 3-83Igi-mb1^−/− mice were used as a negative control as the B cell population of these mice contains only pro-B cells and is deficient of immature B cells. Numbers represent frequencies of live B220⁺ B cells in indicated gates. Frequencies in histograms are shown only for 3-83Igi, H-2^d cells. (C) Mean frequency (± SD, n = 11 mice from 11 independent experiments) of CD21⁺CD23⁻ (T1-like), CD21⁺CD23⁺ (T2-like), and IgD⁺ (T1-like and T2-like) B cells at days 0 and 2 of culture with or without BAFF as described in (B). (D) Absolute cell numbers (± SD, n = 5 mice from 2 independent experiments) of CD21⁻CD23⁻ (immature), CD21⁺CD23⁻ (T1-like), and CD21⁺CD23⁺ (T2-like) B cells as described in (B). P-values of <0.05 (*), <0.005 (**), and <0.001 (***) are indicated. N.S. = not significant.

To determine the role of BAFF in the immature to transitional B cell differentiation process, non-autoreactive (3-83Igi, H-2^d) immature B cells generated in the presence of IL-7 (Fig. 3B, day 0) were subsequently cultured with or without BAFF for 2 days and then assessed by flow cytometry for differentiation. Ig-α-deficient bone marrow cells from 3-83Igi-mb1^−/− mice were used as a negative control as B cell development in these mice does not progress beyond the B220⁺IgM⁻ pro-B cell stage (Fig. 3B, and (39)). In accordance with our previous results (39), we observed significant generation of T1- (IgD⁺CD21⁺CD23⁻) and T2- (IgD⁺CD21⁺CD23⁺) like transitional B cells after culture of BCR-normal immature B cells in the presence of BAFF (Figs. 3B,C,D). Cultures with BAFF (and without IL-7) were also characterized by significant death of CD21⁻CD23⁻ immature B cells (Fig. 3D). Importantly, although some T1-like B cells were generated even in the absence of BAFF, their frequency and numbers were significantly increased in cultures containing BAFF (Figs. 3B,C,D). In contrast, generation of T2-like B cells, as measured by expression of CD23 and high levels of IgD, was strongly dependent on the presence of BAFF (Figs. 3B,C,D). These results suggest that BAFF augments, but is not required for, the differentiation of non-autoreactive IgD⁻CD21⁻CD23⁻ immature B cells into transitional IgD⁺CD21⁺CD23⁻ T1-like B cells in vitro, and that BAFF is essential for the generation, in addition to the maintenance, of transitional IgD^highCD21⁺CD23⁺ T2-like B cells.

Bcl-2 can replace BAFF for the generation of transitional T1-like, but not T2-like, B cells in vitro

BAFF mediates its effects by activating Akt and the alternative NF-κB pathways, leading to higher expression of pro-survival Bcl-2 family members, among other changes (51). In fact, a Bcl-2 transgene appeared to genetically complement BAFF-R-mutant mice to restore normal frequencies of follicular mature B cells (9, 50). In TACI-Ig transgenic mice, however, a Bcl-2 transgene only partially restored T2 and follicular B cell numbers, while it did not restore T3 and marginal zone B cell subsets, suggesting that BAFF does not act solely as a survival factor, but also participates in B cell maturation (52). Indeed, these previous analyses did not discriminate between the survival and the accumulation of mature B cells as a result of Bcl-2 expression, and it remains possible that the absence of BAFF-R signaling hinders B cell differentiation.

To further explore whether BAFF aids the maturation of immature B cells in addition to mediating B cell survival, 3-83Igi, H-2^d non-autoreactive immature B cells generated in vitro in the presence of IL-7 were transduced with Bcl-2 or control Thy1.1 encoding vectors, and transduced B cells were assessed for differentiation in cell cultures that contained BAFF or not. As shown in Fig. 4, Bcl-2 expression promoted the generation of transitional CD21⁺CD23⁻ T1-like B cells in the absence of BAFF at a frequency similar to that induced by BAFF in control cells. However, CD21 levels were lower in the absence of BAFF (1.36 ±0.17 fold lower, on average) even upon expression of Bcl-2 (1.29 ±0.23 fold lower, on average), and in accordance with previous studies indicating that BAFF is required for optimal CD21 expression (9, 50, 52). In contrast to T1-like B cells, the generation of transitional CD21⁺CD23⁺ (and IgD^high) T2-like B cells only occurred in the presence of BAFF, which could not be substituted by Bcl-2 (Fig. 4, and data not shown).

Figure 4. BAFF is required for the differentiation of CD23⁻ into CD23⁺ transitional B cells via a mechanism distinct from that mediating cell survival.

(A) Representative flow cytometric analysis of CD21 and CD23 expression on Thy1.1 control and Bcl-2 transduced non-autoreactive BCR-normal immature B cells from 3-83Igi, H-2^d mice, at the beginning (day 0) and end (day 2) of culture with or without BAFF. Numbers represent frequencies of live B220⁺Thy1.1⁺ B cells in indicated gates. (B) Mean frequency (± SD, n = 2 pooled mice in each of 4 independent experiments) of CD21⁺CD23⁻ (T1-like) and CD21⁺CD23⁺ (T2-like) transitional B cells in the transduced Thy1.1⁺ population of Thy1.1 control (black bars) and Bcl-2 (white bars) 3-83Igi, H-2^d B cells at day 2 of culture with or without BAFF, as described in (A). P-values of <0.05 (*) are indicated. N.S. = not significant.

Overall, these data indicate that BAFF functions to promote both B cell survival and differentiation. Moreover, they indicate that differentiation of CD21⁻CD23⁻ immature B cells into transitional CD21⁺CD23⁻ T1-like B cells largely depends on the pro-survival effect of BAFF, while that of transitional CD21⁺CD23⁻ T1-like B cells into transitional CD21⁺CD23⁺ T2-like B cells relies on a pro-maturation function of this cytokine.

Increased expression of BAFF-R does not rescue differentiation of BCR-low immature B cells

Recently we have shown that BCR-low immature B cells are strongly impaired in their differentiation into transitional B cells both in vitro and in vivo (39). As shown above (Fig. 2B), BCR-low immature B cells express subnormal levels of BAFF-R, in addition to low levels of surface IgM. Thus, we asked whether the reduced differentiation of BCR-low immature B cells was in part due to the low expression of BAFF-R and, consequently, low response to BAFF. To test this hypothesis we generated a retroviral construct that encodes mouse BAFF-R. The functionality of the BAFF-R retroviral construct was confirmed by its ability to partially restore mature B cell generation in A/WySnJ mice, which express a co-dominant natural mutant form of BAFF-R (8, 50) (Fig. S2). We next transduced BCR-low immature B cells from 3-83Igi-low mice with retroviruses encoding either BAFF-R or GFP only and tested the ability of transduced immature B cells to differentiate in vitro in the presence of BAFF. Importantly, transduction with the BAFF-R-encoding retrovirus significantly increased levels of BAFF-R on BCR-low IgD⁻ immature B cells relative to GFP control transduction (Fig. 5A). In accordance with our previous studies (39), we found that most BCR-low immature B cells were arrested in differentiation at the IgD⁻CD21⁻CD23⁻ stage, and only a very small frequency (0.5–5%) of CD21⁺CD23⁻ T1-like and CD21⁺CD23⁺ T2-like transitional B cells were generated after 2 days of culture with BAFF (Figs. 5B,C, GFP, and data not shown). However, increased expression of BAFF-R did not promote normal generation of transitional BCR-low B cells (as compared to 3-83Igi, H-2^d B cells in Fig. 3), but only slightly increased their frequency to 1–8% (Figs. 5B,C, BAFF-R). Increased BAFF-R expression was also unable to restore normal generation of BCR-low transitional and mature B cells in vivo (Fig. S3), and similar results were obtained when BAFF-R expression was enforced on autoreactive 3-83Igi, Rag1^−/−, H-2^b immature B cells (Fig. S4).

Figure 5. Increased levels of active N-Ras, but not of BAFF-R expression, rescue the differentiation of BCR-low immature B cells.

(A) Representative flow cytometric analysis of BAFF-R expression on BCR-low immature B cells from 3-83Igi-low mice generated in cultures with IL-7 and transduced with either GFP only control (filled histogram) or BAFF-R (black line). Immature B cells were gated as live B220⁺IgD⁻GFP⁺. (B) Representative flow cytometric analysis of CD21 and CD23 expression on GFP control or BAFF-R transduced 3-83Igi-low B cells before (day 0) and after (day 2) culture with BAFF. Numbers represent frequencies of live B220⁺GFP⁺ B cells in indicated gates. (C) Mean frequency (± SD, n = 2 pooled mice in each of 3 independent experiments) of CD21⁺CD23⁻ (T1-like) and CD21⁺CD23⁺ (T2-like) transitional B cells in GFP control (white bars) or BAFF-R (black bars) transduced (GFP⁺) 3-83Igi-low B cell populations at days 0 and 2 of culture with BAFF, as described in (B). (D) Representative flow cytometric analysis of CD21 and CD23 expression on GFP control or N-RasD12 transduced 3-83Igi-low B cells before (day 0) and after (day 2) culture with or without BAFF. Numbers represent frequencies of live B220⁺GFP⁺ B cells in indicated gates. (E) Mean frequency (± SD, n = 2 pooled mice in each of 7 independent experiments) of CD21⁺CD23⁻ (T1-like) and CD21⁺CD23⁺ (T2-like) transitional B cells in GFP control (white bars) or N-RasD12 (black bars) transduced (GFP⁺) 3-83Igi-low B cell populations at days 0 and 2 of culture with BAFF, as described in (D). P-values of <0.05 (*) and <0.005 (**) are indicated. N.S. = not significant.

As mentioned above, expression of N-RasD12 was able to complement suboptimal tonic BCR signaling in BCR-low immature B cells, promoting higher levels of BAFF-R expression (Figs. 2C,D), and differentiation into transitional and mature B cells (39). However, it remained unclear whether the differentiation process mediated by N-RasD12 was still dependent on, or rather independent of, the presence of BAFF. Therefore, we transduced BCR-low immature B cells with N-RasD12 or GFP only encoding retroviruses and assessed their differentiation in vitro in the presence or absence of BAFF. In contrast to the poor differentiation attained with increasing only BAFF-R expression (Fig. 5C), optimal in vitro differentiation of BCR-low CD21⁻CD23⁻ immature B cells into CD21⁺CD23^−/+ transitional B cells was achieved upon reinstatement of tonic BCR signaling via expression of N-RasD12 (Figs. 5D,E), in accordance with what we have previously reported (39). In addition to these data, we found that generation of transitional CD21⁺CD23⁺ T2-like B cells in N-rasD12-transduced BCR-low cell cultures was further increased by the presence of BAFF (Figs. 5D,E), suggesting that the signaling pathway activated by N-RasD12 could not completely substitute for the absence of BAFF-R signaling. Overall these data indicate that reduced tonic BCR signaling is responsible for the reduced differentiation of BCR-low CD21⁻CD23⁻ immature B cells into CD21⁺CD23^−/+ transitional B cells, and that the role of BAFF-R signaling in transitional B cell generation is secondary to that of tonic BCR signaling.

BAFF-R expression aids the development of transitional T1 B cells in the spleen but not T1-like B cells in the bone marrow

Studies of BAFF- and BAFF-R-deficient mice have clearly shown a requirement for this cytokine for the survival and maintenance of spleen transitional T2 and mature B cells, while minimal and statistically insignificant effects have been observed at the transitional T1 B cell stage in vivo (4, 5, 9). Our in vitro data, however, suggest that BAFF operates also on IgD⁻CD21⁻CD23⁻ immature B cells and IgD⁺CD21⁺CD23⁻ T1 B cells to augment their differentiation into transitional IgD⁺CD21⁺CD23⁻ T1-like and IgD^highCD21⁺CD23⁺ T2-like B cells, respectively. Thus, we hypothesized that a significant role for BAFF in the generation of transitional T1 B cells could be found in vivo in a competitive environment.

To test whether BAFF functions in the generation of transitional T1 B cells in vivo, we established mixed bone marrow chimeras by combining either C57BL/6 Ly5.2 BAFF-R-deficient or -sufficient bone marrow cells with C57BL/6 Ly5.1 congenic wild-type bone marrow cells (Fig. 6A and Table S1). These chimeric mice allowed us to compare the differentiation ability of BAFF-R-deficient B cells to that of BAFF-R-sufficient B cells side by side, thus representing a more stringent test for the function of BAFF-R in B cell development. Mixed bone marrow chimeras were analyzed by flow cytometry 6 weeks after their reconstitution to determine the frequency of BAFF-R-sufficient and -deficient Ly5.2⁺ donor cells in pro-B, pre-B, immature B, transitional B and follicular B cell populations in bone marrow and spleen tissues (gated as shown in Fig. S5A). In the spleen, transitional B cells were gated based on high levels of CD24 expression, which discriminates them from more mature B cells (22). The CD24^high B cell population of the spleen also contains some IgM⁺IgD⁻CD21⁻CD23⁻ immature B cells, but these cells represent only a small fraction (<20%, data not shown). In the bone marrow, instead, transitional B cells were gated based on high IgM and low IgD expression because CD24 is also expressed by IgM⁻ pro-B and pre-B cells (53). As shown in Fig. 6B (black bars), the frequency of BAFF-R-sufficient B cells remained fairly stable throughout B cell development, as indicated by the fact that similar frequencies of Ly5.2⁺ wild-type B cells were observed at all B cell stages. In contrast, the frequency of BAFF-R-deficient B cells was significantly decreased at the CD24^highCD23⁺ T2/T3 transitional B cell and CD24^lowCD23⁺ mature follicular B cell stages in the spleen of bone marrow chimeric mice relative to their precursor B cell populations in the same animals (Fig. 6B, white bars). These data indicate that BAFF-R plays a significant role in the generation and maintenance of T2, T3, and mature B cells, confirming previous observations (9). In addition to these anticipated results, we also found that the frequency of BAFF-R-deficient transitional CD24^highCD23⁻ T1 (and immature) B cells in the spleen was significantly reduced to half of that of their precursor bone marrow immature B cell population in the same mice (Fig. 6B, white bars). As shown through multi-parametric flow cytometric analysis (Fig. S5B), spleen CD24^highCD23⁻ T1 B cells also express low to high levels of CD21 and IgD, markers that distinguish them from newly generated IgD⁻CD21⁻CD23⁻ immature B cells. Thus, these data support the conclusion of our in vitro data that BAFF contributes to the differentiation of immature B cells into transitional T1 B cells.

Figure 6. BAFF is required for optimal differentiation of wild-type T1, as well as T2, splenic transitional B cells.

(A) Schematic for the generation of BAFF-R-deficient and -sufficient mixed bone marrow chimeras. Bone marrow cells from Ly5.1 and Ly5.2 congenic mice were used for these experiments. (B) B cells from mixed bone marrow chimeras generated as described in (A) were analyzed by flow cytometry for expression of the congenic marker Ly5.2. The bar graphs represent the mean frequencies (± SD, n = 6 chimera mice) of Ly5.2⁺ cells (BAFF-R^+/+, black bars; BAFF-R^−/−, white bars) in B cell populations from bone marrow and spleen tissues analyzed ex-vivo. Similar results were obtained in an additional independent experiment with three mice per group (data not shown). The following surface markers were used to discriminate B cell subsets within the total B220⁺ cell population in bone marrow: pro-B: IgM⁻CD43⁺; pre-B: IgM⁻CD2⁺; immature: IgM⁺IgD⁻; transitional T1-like: IgM^highIgD^lowCD23⁻; transitional T2-like: IgM^highIgD^lowCD23⁺; and in spleen: immature + transitional T1: CD24^highCD23⁻; transitional T2+T3: CD24^highCD23⁺; follicular (Fo.): CD24^lowCD23⁺. The gating strategy used in the analysis is depicted in Figure S5A. P-values of <0.05 (*), <0.005 (**), and <0.001 (***) are indicated. N.S. = not significant.

Cells displaying characteristics generally similar to those of splenic transitional T1 and T2 cells are also found in the bone marrow ((25, 26), and Fig. S1), and recent studies have suggested that immature B cells can differentiate into transitional B cells in both bone marrow and spleen tissues (25, 26). We used a multi-parameter flow cytometric analysis to distinguish donor-derived bone marrow T1-like (B220⁺IgM^highIgD⁺CD23⁻) and T2-like (B220⁺IgM^highIgD⁺CD23⁺) B cells, and spleen immature/T1 (B220⁺CD24^highCD23⁻) and T2/T3 (B220⁺CD24^highCD23⁺) B cells in BAFF-R-deficient and -sufficient mixed chimeras (Fig. S5). These analyses indicated that whereas the size of the transitional CD23⁺ B cell population was reduced in both spleen and bone marrow in the absence of BAFF-R, that of the transitional CD23⁻ B cell population was significantly decreased only in the spleen (Fig. 6C). In fact, the frequency of B cells within the bone marrow IgM^highIgD⁺CD23⁻ T1-like population was similar to that of those within the IgM⁺IgD⁻ immature B cell population, regardless of BAFF-R expression (Fig. 6C). These results indicate that BAFF is critical for the generation of transitional T1 B cells in the spleen, but not for that of transitional T1-like B cells in the bone marrow.

IL-7 cannot replace BAFF in the in vitro generation of transitional B cells

We have shown that optimal generation of transitional IgD⁺CD21⁺CD23⁻ T1-like B cells in vitro requires the pro-survival function of BAFF, which can be substituted by constitutive Bcl-2 expression (Fig. 4). Moreover, we have found that only spleen T1, but not bone marrow T1-like, CD23⁻ transitional B cells are affected by the absence of BAFF-R in vivo (Fig. 6). Thus, we questioned whether a different pro-survival cytokine could promote the generation of transitional T1-like B cells in the bone marrow.

The cytokine IL-7 has both pro-survival and differentiation functions in early B cell development (18). However, whether IL-7 operates on immature B cells is controversial (18), despite the fact that these cells still express IL-7R (54). Therefore, we decided to test whether IL-7 could replace BAFF, at least partially, in the differentiation of immature B cells into transitional B cells in vitro.

For this experiment, non-autoreactive (3-83Igi, H-2^d) immature B cells generated in vitro in an IL-7 culture were washed and re-cultured in the presence or absence of IL-7, or with BAFF as control, and analyzed for differentiation after two days. Neither low (pg/ml) nor high (ng/ml) concentrations of IL-7 promoted differentiation of IgD⁻CD21⁻CD23⁻ immature B cells into IgD⁺CD21⁺CD23^−/+ transitional-like B cells, as shown by the absence of significant CD21, CD23, and IgD expression (Figs. 7A,C and data not shown). In light of this result, and because it has been previously suggested that IL-7 may inhibit B cell differentiation (18), we asked whether IL-7 could actually suppress the differentiation of immature B cells mediated by BAFF. Thus, we compared differentiation of 3-83Igi, H-2^d immature B cells in the presence of BAFF and with or without IL-7. We found that IL-7 did not prevent BAFF-mediated differentiation of IgD⁻CD21⁻CD23⁻ immature B cells into transitional IgD⁺CD21⁺CD23⁻ T1-like and IgD^highCD21⁺CD23⁺ T2-like B cells, as indicated by significant expression of CD21, CD23 and IgD in cultures containing both BAFF and IL-7 (Fig. 7B and data not shown). A small but significant reduction was observed in the frequency of T1-like and T2-like B cells in cultures containing both BAFF and IL-7 (Fig. 7C), but these differences were caused by increased survival of non-differentiated immature B cells in response to IL-7 (Fig. 7D). In fact, cell counting indicated that T1-like and T2-like B cells were present in similar numbers in cultures containing BAFF regardless of IL-7 (Fig. 7D).

Figure 7. IL-7 neither promotes nor blocks the differentiation of immature B cells into transitional B cells in vitro.

(A) Bone marrow cells from 3-83Igi, H-2^d mice were cultured in the presence of IL-7 for 3 days to generate BCR-normal non-autoreactive immature B cells. Immature B cells were then washed with PBS and re-cultured with or without BAFF or IL-7, as indicated. Dot plots are representative flow cytometric analyses of CD21 and CD23 expression on B cells at the beginning (day 0) and end (day 2) of culture with the indicated cytokines. Numbers represent frequencies of live B220⁺ B cells in indicated gates. IL-7 was used approximately at 250 pg/ml (low IL-7) and 10 ng/ml (high IL-7). (B) Immature B cells from 3-83Igi, H-2^d mice were generated in the presence of IL-7 as described in (A) and then re-cultured in the presence of IL-7 and BAFF, as indicated. Cells were analyzed as described in (A). (C) Mean frequency (± SD, n = 3 mice, one in each of 3 independent experiments) of CD21⁺CD23⁻ (T1-like) and CD21⁺CD23⁺ (T2-like) 3-83Igi, H-2^d transitional B cells at days 0 and 2 of culture with indicated cytokines, as described in (A) and (B). (D) Absolute cell numbers (± SD, n = 3 mice from 1 experiment) of CD21⁻CD23⁻ (immature), CD21⁺CD23⁻ (T1-like), and CD21⁺CD23⁺ (T2-like) B cells at days 0 and 2 of culture with indicated cytokines as described in (A) and (B). P-values of <0.05 (*), <0.005 (**), and <0.001 (***) are indicated. N.S. = not significant.

Thus, our data indicate that IL-7 cannot replace BAFF in the generation of CD21⁺CD23^−/+ transitional B cells in vitro. Importantly, however, IL-7 does not inhibit BAFF activity in this process. Moreover, other bone marrow factors that could potentially affect newly generated immature B cells in the bone marrow, such as APRIL (55, 56), hemokinin-1 (HK-1, (57)) and macrophage migration inhibitory factor (MIF, (58)), were also unable to promote differentiation of IgD⁻CD21⁻CD23⁻ immature B cells into IgD⁺CD21⁺CD23^−/+ transitional B cells in vitro (Fig. S6), suggesting that BAFF may be unique in this capacity.

Discussion

The studies presented here were developed to test whether BAFF-R is differentially expressed in autoreactive and non-autoreactive immature B cells, and if BAFF-R expression and signaling contribute to immature B cell selection. We found that BAFF-R is expressed at levels 4-fold higher by non-autoreactive than autoreactive immature B cells, both at the RNA and surface protein levels. Moreover, our data indicate that BAFF contributes to the selection of immature B cells by promoting the differentiation of immature B cells into transitional B cells upon normal tonic BCR signaling.

Previous studies have shown that expression of BAFF-R coincides with that of IgM during B cell development (28, 35). Our data extend those findings by indicating that BAFF-R expression is dependent on tonic BCR signaling and its level correlates with that of surface IgM. To support these conclusions, we show that immature B cells displaying low levels of IgM and tonic BCR signaling manifest low levels of BAFF-R expression. Moreover, reinstatement of tonic BCR signaling by expression of the constitutive active N-RasD12 protein leads to higher BAFF-R levels on BCR-low immature B cells. These data suggest that tonic BCR signaling leads to changes in the expression of nuclear proteins that promote transcription of the tnfrsf13c gene. A likely candidate for modulating tnfrsf13c gene transcription downstream of tonic BCR signaling is the NF-κB pathway, which may be activated by tonic BCR signaling and whose inhibition leads to reduced tnfrsf13c mRNA expression (59). In addition, previous studies have shown that tonic BCR signaling generates NF-κB p100, which is a substrate of the BAFF-R signaling pathway (60). Thus, it appears that immature B cells would be capable of responding to BAFF only upon optimal tonic BCR signaling, which increases both BAFF-R expression and function. Numerous publications have indicated that BAFF is an important factor for the survival of mature follicular and marginal zone B cells. Some reports, however, have suggested that BAFF may promote B cell differentiation as well. For instance, a reduced export of B cells from the bone marrow has been observed in BAFF-R mutant A/WySnJ mice (7). Moreover, the B cell defects seen in the absence of BAFF or BAFF-R cannot completely be corrected by constitutive expression of the pro-survival protein Bcl-2. Constitutive Bcl-2 expression does not rescue marginal zone B cells and only partially reinstates numbers of follicular B cells in TACI-Ig transgenic mice (52), and it does not reconstitute normal CD21 expression in BAFF-R mutant mice (9, 50), suggesting an underlying defect in B cell generation. On the other side, constitutive Bcl-2 was able to functionally substitute for BAFF in vivo, allowing a T-dependent immune response to a standard antigen, but one can argue that immature/transitional B cells could be drawn into an immune response in the absence of competition with mature B cells and in the presence of the anti-apoptotic Bcl-2 pathway (52). Our data, in fact, support a role for BAFF in the differentiation of B cells, and more explicitly in the differentiation of IgM⁺IgD⁻CD24^highCD21⁻CD23⁻ immature B cells into IgM⁺IgD⁺CD24^highCD21⁺CD23^−/+ transitional B cells. Specifically, our in vitro data show that the addition of BAFF to bone marrow-derived immature B cells enhances and promotes their differentiation into transitional-like B cells that express IgD, CD21 and, for some, CD23. A role for BAFF in the differentiation of immature B cells into transitional B cells is also supported by our in vivo data showing a significant reduction of BAFF-R-deficient CD24^highCD23⁻ immature/T1 and CD24^highCD23⁺ T2/T3 transitional B cell numbers in the spleen of mice in which BAFF-R-deficient B cells developed in competition with BAFF-R-sufficient B cells.

It has been proposed that BAFF regulates CD21 and CD23 expression (28). Thus, a justified question can be raised of whether our results indicate that BAFF promotes the differentiation of immature B cells or increases the expression of the CD21 and CD23 markers that we used for transitional B cell classification. We are confident that our analyses appropriately show that BAFF promotes the differentiation of immature into transitional B cells because of the following arguments. First, more than one study has challenged the idea that BAFF regulates CD23 expression (9, 32, 49). Thus, the use of CD23 to distinguish transitional T2 and T2-like B cells from transitional T1 and T1-like B cells in our studies seems appropriate both in the absence and in the presence of BAFF. Second, although CD21 expression is indeed influenced by BAFF, CD21 levels are only slightly (2-fold) reduced in the absence of BAFF (50). In our studies we used CD21 to characterize the presence of CD21⁺ transitional B cells developing from CD21⁻ immature B cells only in vitro. In these studies we were careful to gate CD21⁺ cells based on CD21⁻ cell controls, such as BCR-negative (3-83Igi-mb1^−/−) and BCR-low (3-83Igi-low) immature B cells. Thus, these analyses were able to distinguish CD21^low (transitional) from CD21⁻ (immature) B cells both in the presence and in the absence of BAFF. Third, IgD can also be used to distinguish transitional from immature B cells (22), and BAFF does not appear to affect IgD expression. In our studies we observed higher levels of IgD expression and increased frequency of IgD⁺ B cells in cultures containing BAFF, indicating an increase in B cell differentiation. Fourth, transitional T1 B cells in the spleen were gated based on high CD24 expression and the absence of CD23, markers that are not affected by BAFF. The fact that we observed a strong and significant reduction of BAFF-R-deficient transitional T1 B cells in the spleen of mixed bone marrow chimeras indicates that BAFF is important for the generation of these cells, and that BAFF-R signaling has an effect preceding the T2 B cell stage.

Splenic transitional B cells are classified as T1 when they are CD23⁻, and T2 or T3, when they express CD23 and depending on levels of IgM (23). In addition, all transitional B cells express variable amounts of IgD and CD21 (27). Although transitional B cells are also present in bone marrow (22, 25, 26), they are still not fully characterized in this tissue. Nevertheless, it was necessary in our studies to distinguish immature from transitional B cells in both bone marrow and spleen. Thus, we classified as transitional those cells that also upregulated expression of IgD and CD21 in bone marrow cultures and bone marrow tissues, discriminating transitional T1-like and T2-like B cells on the basis of CD23 expression. Our data indicate that BAFF-R and BAFF play a larger role in T2 and T3 than in T1 B cell differentiation. In fact, differentiation of immature B cells into transitional CD23⁻ B cells in bone marrow cultures, in bone marrow tissues and in spleen occurred to some extent even in the absence of BAFF, although these processes were enhanced by the addition of BAFF. On the other side, we show that the differentiation of transitional CD23⁺ B cells required BAFF both in vitro and in vivo. The differential dependency on BAFF among transitional B cell subsets may be due to the fact that BAFF-R expression and BAFF binding increase with their differentiation (28) and, in fact, T1 B cells are considered the developmental precursors of both T2 and T3 B cells (27).

An important finding supporting a role for BAFF in the differentiation of immature into transitional B cells is the inability of Bcl-2 to fully substitute for BAFF in this function. In fact, enforced expression of Bcl-2 in immature B cells replaced BAFF for the development of transitional CD23⁻ T1-like, but not CD23⁺ T2-like, B cells in vitro. Moreover, addition of BAFF to bcl-2-transduced immature B cells further increased their differentiation into T1-like transitional B cells. Because enforced Bcl-2 expression does not reconstitute CD23⁺ transitional B cell generation in the absence of BAFF in vitro, these data strongly suggest that BAFF mediates the differentiation of CD23⁻ transitional T1 B cells into CD23⁺ transitional T2 B cells via a pro-maturation function. Development of CD23⁺ B cells has been observed in vivo in BAFF and BAFF-R deficient mice reconstituted with Bcl-2 transgenes (9, 50, 52). To reconcile our findings with those previous observations, we argue that Bcl-2 may have enhanced the survival and accumulation of the few transitional T2 and follicular CD23⁺ B cells that develop even in the absence of BAFF (9, 50). In particular, the results of our in vivo competition studies highlight the possibility that the differentiation of BAFF and BAFF-R-deficient CD23⁺ B cells observed with and without Bcl-2 transgenes (9, 50, 52) may be the product, at least in part, of the lack of competition with normal B cells. Thus, our data support the idea that BAFF-mediated signals lead to enhanced differentiation of immature B cells into transitional T1 B cells and are necessary for the differentiation of transitional T1 B cells into T2 B cells. However, it appears that the role of BAFF in these events is secondary to that of tonic BCR signaling. We have previously shown that tonic BCR signaling is required for the generation of both CD23⁻ and CD23⁺ transitional B cells (39). Moreover, our present data indicate that increasing BAFF-R expression in immature B cells with insufficient tonic BCR signaling does not reconstitute normal differentiation in response to BAFF, and that optimal BCR expression and function are essential for normal BAFF-R expression. Thus, our data support a model in which optimal tonic BCR signaling induces the differentiation of immature B cells into early transitional T1 B cells, by initiating the expression of CD21 and IgD, as well as increasing that of BAFF-R. BAFF binding, then, triggers BAFF-R signaling causing further differentiation into cells that express higher levels of IgD and CD21 (late T1) and of CD23 (T2). Thus, a synergy between BAFF-R and tonic BCR signaling appears necessary for optimal generation of transitional T1 B cells, for the differentiation of transitional T1 into transitional T2 B cells, and ultimately, therefore, for the production of mature B cells.

Immature B cells generated in the bone marrow can either migrate to the spleen for further differentiation, or first differentiate into transitional B cells within the bone marrow before migrating to the spleen (26). Our studies show that during in vivo B cell development BAFF-R signaling affects the generation of transitional T1 B cells in the spleen, but not that of T1-like B cells in the bone marrow. There are different potential explanations for this finding. For instance, it is possible that the B cells we have classified as transitional T1-like B cells in bone marrow do not belong to the same developmental stage as T1 B cells in the spleen. In support of this idea, we noticed that CD21 expression is higher on splenic T1 than on bone marrow T1-like B cells (data not shown). If so, tonic BCR signaling may be sufficient for the differentiation of immature B cells into transitional T1-like B cells in the bone marrow, while a cooperation of tonic BCR and BAFF-R signals may be required for the differentiation of immature and/or T1-like B cells into the splenic transitional T1 B cell population. Another possible explanation is one related to the environment. There may be a cytokine distinct from BAFF in the bone marrow that is able to promote the differentiation of immature B cells into T1-like B cells in cooperation with tonic BCR signaling. Our data show that increased cell survival, such as that mediated by constitutive Bcl-2 expression, can substitute for BAFF to enhance the differentiation of immature B cells into transitional T1-like B cells in vitro. Thus, another pro-survival cytokine in the bone marrow could potentially relieve the need for BAFF. We tested the cytokine IL-7 in this context because IL-7 permeates the bone marrow environment and promotes early B cell survival and differentiation. Our data show, however, that IL-7 does not enhance the generation of IgD⁺CD21⁺CD23⁻ T1-like B cells in vitro. Thus, a different bone marrow cytokine may be involved in this process. In preliminary studies we tested whether APRIL, HK-1 or MIF may be able to substitute for BAFF in promoting the differentiation of immature B cells into transitional B cells in vitro, as these cytokines have been shown to affect B cells and are present in the bone marrow tissue (55–58). However, none of these cytokines had any effect on the upregulation of IgD, CD21, and CD23 in culture and, thus, the differentiation of immature into transitional B cells, suggesting that BAFF may be unique in this function.

It has been reported that IL-7 hampers the differentiation of immature and mature B cells from their progenitors, although these data are controversial (18). Consequently, we assessed whether IL-7 could actually counteract the effects of tonic BCR and BAFF-R signaling in the differentiation of immature B cells into transitional B cells in culture. We found that tonic BCR and BAFF-R signaling were both still equally capable of promoting immature B cell differentiation in the presence or absence of IL-7. The absence of an inhibitory effect by IL-7 may be related to the fact that IL-7R expression and function are low on immature B cells (18, 54), although we have not tested whether this is the case in our system.

The differentiation of immature B cells into transitional B cells is a critical event for the proper generation of the naïve B cell repertoire. More than 50% of newly generated immature B cells are autoreactive (33, 34), and their further differentiation is unwarranted. However, nature has evolved a checkpoint that prevents positive selection of autoreactive immature B cells. Autoreactive immature B cells display low levels of surface IgM, as IgM is internalized following binding to autoantigen. Extrapolating from our analyses of BCR-low immature B cell development, the low IgM expression on autoreactive immature B cells would translate into low tonic BCR signaling that is insufficient to promote differentiation into transitional B cells, as also suggested by previous studies (61). In addition to displaying low tonic BCR signaling, we show here that autoreactive immature B cells are unable to upregulate expression of BAFF-R. Therefore, we suggest that the absence of both optimal tonic BCR and BAFF-R signaling prevent autoreactive immature B cells from further differentiation and thus contribute to their negative selection. We also envision that autoreactive immature B cells with different degrees of autoreactivity may progress to different stages of differentiation depending on levels of tonic BCR signaling and BAFF-R expression. Thus, low-avidity autoreactive immature B cells, such as the prototypical anergic B cells, may progress into transitional B cells because they receive low levels of both tonic BCR and BAFF-R signaling, but fail to reach the mature B cell stage because these signals are still reduced compared to those of non-autoreactive B cells. In contrast, high avidity autoreactive immature B cells, such as the 3-83Ig⁺ B cells we studied here, are absolutely blocked in differentiation because they do not receive any tonic BCR signaling. Our study indicates that tonic BCR and BAFF-R signals cooperate for the differentiation of normal IgM-expressing non-autoreactive immature B cells, by promoting their differentiation into transitional B cells. Therefore, tonic BCR and BAFF-R signaling pathways together guide the positive selection of non-autoreactive immature B cells into the peripheral mature B cell compartment, and are critical for shaping the naïve B cell pool such that it can mount a protective humoral immune response in the absence of autoreactivity.