Survival of Igα-Deficient Mature B Cells Requires BAFF-R Function

The BCR is a multiprotein complex expressed on the surface of transitional, immature, and mature B cells. The Ag-binding component of the receptor is the membrane-bound Ig molecule composed of two identical H chains (HCs) and two identical Ig L chains (LCs) that are interconnected by disulfide bonds. The membrane-bound Ig complex is noncovalently associated with the signal-transducing Igα/Igβ heterodimer in a 1:1 stoichiometry (1, 2). The Igα and Igβ proteins are encoded by the B lymphocyte–specific genes mb-1 (Cd79a) and B29 (Cd79b), respectively (3–6). Proper assembly of these four BCR proteins in the endoplasmic reticulum (ER) is required for the transport of the BCR complex onto the B cell surface (7). On resting B cells, the BCR forms autoinhibited oligomers that counteract activation signals from the BCR (8, 9). These BCR oligomers reside with other B cell surface proteins in nanoscale protein islands (10).

The cytoplasmic tail of Igα and Igβ each carries a single ITAM with two conserved tyrosines (11). Upon B cell activation, spleen tyrosine kinase (Syk) phosphorylates and then binds to these tyrosines (12). The resulting BCR/Syk complex connects the BCR to several downstream signaling pathways, leading to the activation, proliferation, and differentiation of B cells (13–15). For example, phosphorylation of the SLP-65/BLNK adaptor protein by Syk leads to an increased calcium flux in activated B cells (16–19), whereas activation of the ERK2 pathway results in upregulation of the B cell activation marker CD86 (20). Once phosphorylated, Syk also becomes a binding partner of the p85 adaptor component of PI3K and can promote the activity of the PI3K signaling pathway (21).

In pre-B cells, proliferation and differentiation are controlled by the pre-BCR, an autonomously signaling receptor comprising a membrane-bound HC, a surrogate LC, and the Igα/Igβ heterodimer (22–24). A loss of a pre-BCR component, such as the HC, Igα, or Igβ, results in an arrest of B cell development at the pro-B cell stage (25–32). Once developing B cells have successfully assembled the HC and LC genes, they express an IgM-BCR on their surface. These cells migrate from the bone marrow (BM) to the spleen and then progress through transitional stages (T1 and T2) to become immature and then mature follicular (FO) B cells (33). During their final maturation steps, B cells pass several check points that ensure the deletion of autoreactive cells from the B cell pool (34). A loss of any of the BCR components at this stage results in the elimination of these immature B cells (35–38).

The maintenance of mature B cells in vivo is mainly attributed to two receptors: the BCR and the receptor for the B cell–activating factor belonging to the TNF family BAFF-R (also known as BR3) (39). The B cell–activating factor belonging to the TNF family (BAFF; also known as BLyS)–BAFF-R interaction plays a fundamental role in the survival of mature B cells (40, 41). The in vivo abrogation of BAFF-R expression leads to a partial block at the T1 to T2 transition in the periphery and strongly reduced mature B cell numbers (42–45). Mature B cells also require the expression of the BCR for their fitness and long-term survival in the periphery. This was first discovered in a study of mice with an inducible deletion of the HC gene (46, 47). Mature B cells that lose the HC rapidly undergo apoptosis but can be rescued by continuous PI3K signaling (48). Furthermore, abrogation of the BCR signaling component Igα or deletion of parts of the Igα tail also resulted in the loss of the affected cells (46). These findings suggest that the resting BCR sends a tonic or maintenance signal promoting the survival of mature B cells. In this study, we revisited this issue using a mouse model in which the inducible abrogation of Igα is accompanied by the de novo expression of a GFP marker. We found that, in contrast to HC-deficient B cells and B cells with a truncated Igα tail, the Igα-deficient B cells survived longer in the periphery and could still respond to BAFF. Furthermore, we showed that the increased apoptosis of HC-deficient B cells is associated with an ongoing unfolded protein response (UPR).

To delete the mb-1 gene, encoding the Igα protein in B cells, the mouse strain Igα/BAP31-eGFP^inv (25, 26) was mated with the mb1-CreER^T2 strain (16), which expresses a tamoxifen-inducible form of the Cre recombinase under the control of the mb-1 promoter region. The mb-1 gene is expressed specifically in B cells throughout B cell maturation. In the resulting cmb1 mice, Igα expression is abrogated in all B cell subsets within a period of 10 d following the induction with 6 mg tamoxifen.

To delete the Igh gene, mb1-CreER^T2 mice were crossed with the previously described B1-8 ^fl/JhT (47) mice. To generate Igα tail–truncated B cells, mb1-CreER^T2 mice were crossed with the previously described Igα^C1f mice (46). All animal studies were conducted in mice aged 8 to 12 wk and were carried out in accordance with the German Animal Welfare Act, having been reviewed by the regional council and approved under license #G-09/103.

Mice were treated once by gavage of 6 mg tamoxifen (Ratiopharm) dissolved in 20% ClinOleic (Baxter) and sacrificed within 7 to 30 d after the treatment, depending on the experiment.

cmb1 mice were treated with tamoxifen as described above. After 15 d, groups of three mice were either left untreated or were injected once with 0.5 mg nonblocking anti–BAFF-R isotype control Ab 5A12 or with the blocking anti–BAFF-R Ab 9B9. Analysis was performed after 15 additional days. The 5A12 and 9B9 Abs were kindly provided by Prof. Antonius Rolink (Department of Biomedicine, University of Basel, Basel, Switzerland).

Total mRNA was isolated from whole mature B cells using the Quick-RNA MicroPrep Kit (Zymo Research). Purified B cells (1 × 10⁵) were lysed according to the manufacturer’s instructions. Aliquots of 0.1 μg mRNA were reverse transcribed using the First-Strand cDNA Synthesis Kit (Fermentas/Thermo Scientific), according to the manufacturer’s instructions. Amplification of Xbp-1s/u cDNA was achieved by semiquantitative PCR using the following primers: Xbp1 3s: 5′-A AAC AGA GTA GCA GCG CAG ACT GC-3′ and Xbp1 12As: 5′-TC CTT CTG GGT AGA CCT CTG GGA G-3′. The PCR products were separated on a 3% agarose gel and visualized by ethidium bromide staining. Hprt transcripts were used as an endogenous control and were amplified using the following primers Hprt-F: 5′-GCT GGT GAA AAG GAC CTC T-3′ and Hprt-R: 5′-CAC AGG ACT AGA ACA CCT GC-3′. Quantification of Xbp1s relative band intensity was performed using ImageJ software. The relative intensity of Xbp1s of each sample displaying the band was divided by the relative intensity of Hprt of the same sample and then normalized to the mb1-CreER^T2 + DTT control, which was equal to 1.

Quantitative RT-PCR (qRT-PCR) was performed using Maxima SYBR Green Master Mix 2X (Thermo Scientific) in a 7500 Fast real-time system light cycler (Applied Biosystems) at an annealing temperature of 60°C for 40 cycles. The following primers were used for amplification: Xbp1 total-forward (F): 5′-AAG AAC ACG CTT GGG AAT GG-3′ and Xbp1 total-reverse (R): 5′-ACT CCC CTT GGC CTC CAC-3′; Xbp1s-F: 5′-GAG TCC GCA GCA GGT G-3′ and Xbp1s-R: 5′-GTG TCA GAG TCC ATG GGA-3′; Edem F: 5′-ACT GAT TCC AAA CAG CCC TT-3′ and Edem R: 5′-GGA TCC CTG TCT TGG GTG TTT-3′; DnaJ9B F: 5′-GCC ATG AAG TAC CAC CCT GA-3′ and DnaJ9B R: 5′-CTT TCC GAC TAT TGG CAT CC-3′; Hspa5 F: 5′-TAT TGG AGG TGG GCA AAC CAA G-3′ and Hspa5 R: 5′-CGC TGG GCA TCA TTG AAG TAA G-3′; Ddit3 F: 5′-GCG ACA GAG CCA GAA TAA CA-3′ and Ddit3 R: 5′-ACC AGG TTC TGC TTT CAG GT-3′; and H3 F: 5′-GTG AAG AAA CCT CAT CGT TAC AGG CCT GGT-3′ and H3 R: 5′-CTG CAA AGC ACC AAT AGC TGC ACT CTG GAA-3′ (49). To amplify the transcripts of genes encoding the BCR components, the following primers were used: Iga S1: 5′-ACC GCA TCA TCA CAG CAG AAG G-3′ and Iga AS2: 5′-TCC TGG TAG GTG CCC TGG A-3′; Igb AS-1: 5′-CTT CAC CAT GGA GCT CCG CTT T-3′ and Igb S1: 5′-GCT GTT GTT CCT GCT GCT GC-3′; Vkdeg: 5′-3GGCTGCAGSTTCAGTGGCAGTGGRTCWGGRAC-3′; and Ck17R: 5′-GGAAGGATCCAGTTGGTGCAGCATCAG-3′.

Flow cytometry (FACS) analysis was performed on 1 × 10⁵–2 × 10⁶ cells in FACS Buffer (PBS; 3% FCS (PAN Biotech); 0.05% NaN₃) on ice for 20 min. Before each staining, the cells were incubated with the anti-FcR–blocking Ab (CD16/32, clone 2.4G2) on ice for ≥5 min. The following Abs were used for the staining: PE- and Alexa Fluor 647–anti-CD19 (eBio1D3, eBioscience; 6D5, BioLegend, respectively); FITC- and PE–anti-IgD (11-26c.2a; BioLegend); Alexa Fluor 647– and PE-Cy7–anti-CD23 (B3B4, BioLegend; 2G8, Southern Biotech); biotin–anti-BAFF-R (polyclonal; R&D Systems); eFluor 450–anti-IgM (eB121-15F9), PE- and allophycocyanin–anti-CD93 (AA4.1), PE–anti-CD180 (RP/14), PE–anti-MD-1 (MD14), PE–anti-CD38 (90), PE–anti-MHC class II (M5/114.15.2), PE–anti-CD86 (GL1), and PerCP-Cy5.5–anti-CD5 (53-7.3) (all from eBioscience); and PE–anti-CD43 (S7), PE–anti-MHC class I (34-2-12), PE–anti-Fas (Jo2), and anti-Igβ (HM79-16) (all from BD Pharmingen). Anti-Igβ was biotinylated using a LYNX conjugation kit (AbD Serotec), according to the manufacturer’s instructions, and visualized with PE–anti-Biotin (BK-1/39; eBioscience).

The residual unbound Abs were removed by addition of 700 μl FACS buffer to the cells and centrifugation at 4°C and 300 × g for 5 min.

FACSCalibur, LSR II, and LSRFortessa (Becton Dickinson) cell analyzers were used for data acquisition. Analysis was performed using FlowJo software (TreeStar).

Intracellular cell staining was performed using the ADG FIX & PERM Kit (Dianova). Cells were stained extracellularly for the expression of CD19, IgM, or IgD, fixed with paraformaldehyde-based buffer A for 10 min at room temperature, and washed with 1 ml PBS. All centrifugation steps were performed at 4°C and 300 × g for 5 min. Abs for intracellular analysis were dissolved in saponin-based Buffer B for 15 min at room temperature. Subsequently, the cells were washed with 0.5% saponin buffer (PBS; 0.5% saponin [Sigma]; 0.2% BSA [PAA Laboratories]; 0.02% NaN₃ [Sigma]), centrifuged as described, and washed again with FACS buffer (PBS; 3% FCS [PAN Biotech]; 0.05% NaN₃) before analysis. Abs used for intracellular analysis were Alexa Fluor 647–anti-Igα (24C2.5, eBioscience; detects the cytoplasmic tail), anti-Igβ (B29/123, Santa Cruz), PE–anti-κLC (187.1, Southern Biotech), and DyLight 649–anti-IgM [F(ab′)₂, polyclonal, Jackson ImmunoResearch]. Anti-Igβ Ab was directly labeled using the Zenon Alexa Fluor 647 mouse IgG2b labeling kit (Life Technologies), according to the manufacturer’s instructions.

A mixture of Igα⁺ and Igα^neg splenic B cells was obtained by MACS-based negative selection using the B Cell Isolation Kit (Miltenyi Biotec), according to the manufacturer’s instructions. The cells were then separated on an AutoMACS (Miltenyi Biotec). For the ex vivo activation and survival assays, Igα^neg B cells were separated from Igα⁺ cells by flow cytometer–based cell sorting (FACS) using the following Abs: PE–anti-CD43 (S7, BD Pharmingen), allophycocyanin–anti-Thy1.2 (53-2.1, eBioscience), and PE–anti-CD93 (AA4.1, eBioscience). Mature B cells were selected by gating on the double-negative population and were separated using GFP expression. Igα^∆c B cells were sorted using a similar protocol. To separate HC⁺ and HC^neg B cells, anti-CD19 (eBio1D3, eBioscience) and F(ab) fragments of IgM (polyclonal, Jackson ImmunoResearch) and IgD (digested lo-md, AbD Serotec) Abs were included in the staining. F(ab) fragments were used to avoid possible activation by whole-molecule Abs. For RT-PCR analysis and proximity ligation assay (PLA), Igα⁺, Igα^neg, HC⁺, HC^neg, and Igα^∆c B cells, as well as B cells with one mb1-CreER^T2 allele, were sorted as follows. Alexa Fluor 647–anti-CD19 (eBio1D3) and PE–anti-CD93 (AA4.1, both from eBioscience) were used to exclude possible contamination with non-B cells and transitional B cells, and Brilliant Violet–anti-CD138 (281-2) and Brilliant Violet 421–anti-CXCR4 (12G5, both from BioLegend) were applied to exclude plasma cells and F(ab) fragments of IgM (polyclonal, Jackson ImmunoResearch) and IgD (digested lo-md, AbD Serotec) Abs. A FACSAria (Becton Dickinson) cell sorter was used to separate the cells. After the separation, the B cells were analyzed for purity and viability using anti-CD19, IgM, and IgD Abs and 7 aminoactinomycin D ([7-AAD], eBioscience).

Ca²⁺ influx in MACS-purified mature B cells was measured by flow cytometry using the intracellular fluorescent dye Indo-1. Intracellular Ca²⁺ influx was measured on the LSR II flow cytometer (Becton Dickinson). Igα^neg B cells were distinguished from Igα⁺ B cells based on GFP expression. The cells (0.5 × 10⁶–1 × 10⁶) were suspended in 1 ml plain Iscove’s cell culture medium supplemented with 1% FCS. Indo-1 dye was prepared 5 min prior to incubation with the cell samples and consisted of 25 μl Indo-1 (Life Technologies, dissolved in DMSO), 25 μl Pluronic F-127 (Life Technologies), and 113 μl FCS (PAN Biotech). Each sample was stained with 15 μl the Indo-1 dye. The cell samples were incubated in the dark at 37°C for 45 min. Cells were centrifuged at 300 × g and 4°C. After removing the supernatant, the pellet was resuspended in 500 μl Iscove’s cell culture medium supplemented with 1% FCS. The following stimuli were used to mobilize the Ca²⁺ flux: anti-IgM F(ab′)₂ (goat polyclonal #115-006-075, 10 μg/ml; Jackson ImmunoResearch), anti-κLC F(ab′)₂ (polyclonal, 5 μg/ml; Southern Biotech), anti-Igβ (HM79-16, 10 μg/ml; BD Pharmingen), and latrunculin A (1 μM; Cayman).

Splenic B cells were purified by negative selection and separated based on GFP expression, as described above. Igα⁺ and Igα^neg B cells (1 × 10⁵) were cultured separately in 0.2 ml complete Iscove’s cell culture medium (Biochrom) supplemented with 10% FCS (PAN Biotech). The cells were left unstimulated or incubated with the following stimuli: polyclonal anti-IgM F(ab′)₂ (goat polyclonal #115-006-075, 10 μg/ml; Jackson ImmunoResearch), anti-CD180 (RP/14, 10 μg/ml; eBioscience), anti-CD38 (90, 10 μg/ml; eBioscience), anti-CD40 (1C10, 10 μg/ml; eBioscience), mouse IL-4 (10 ng/ml ImmunoTools), unmethylated CpG oligonucleotides (ODN 1826, 2.5 μM; InvivoGen), LPS (kindly provided by Prof. Marina Freudenberg, Center for Biological Signaling Studies; 10 μg/ml), and anti-Igβ (HM79-16, 10 μg/ml; BD Pharmingen). The cells were harvested 24 h after stimulation, and expression of the activation marker CD86 was analyzed by flow cytometry using anti-CD86 (GL1; eBioscience). The DNA-dye 7-AAD was used to discriminate viable and dead cells.

Chemical induction of the UPR was achieved using 1 mM DTT in complete Iscove’s cell culture medium at 37°C for 10 min.

B cells were sorted as described above and cultured in medium in the absence or presence of 100 ng/ml human recombinant or murine BAFF (ImmunoTools and R&D Systems, respectively) for 5 or 9 d. Every second day, the medium was exchanged and supplemented with fresh BAFF.

The principle underlying PLA was described (50, 51). In this study, sorted B cells in PBS without FCS were placed onto eight-well diagnostic microscope slides (Thermo Scientific) and left for 30 min at 4°C. The cells were then fixed with 4% paraformaldehyde supplemented with 2% sucrose at room temperature for 15 min. The slides were washed in PBS, and the cells were fixed at 37° for 30 min. PLA was performed on fixed and permeabilized B cells following the manufacturer’s instructions (Olink Bioscience). In brief, the primary Abs were detected with secondary anti-rabbit or anti-goat IgG Abs conjugated with PLA-PLUS and PLA-MINUS probes. The PLA probes were incubated with a ligation solution containing oligonucleotides and ligase. In the case of close proximity between the PLA probes, the oligonucleotides hybridized with the probes, forming a circle, which was then amplified in the following step by a polymerase diluted in the polymerization solution (Olink Bioscience). Fluorophore-labeled detection probes hybridized to the rolling circle amplification products, resulting in quantifiable signals. The cells were incubated with a pair of primary Abs: anti-PERK (T-19, goat polyclonal, dilution 1:130; Santa Cruz) and anti-PERK pThr980 (16F8, rabbit monoclonal, dilution 1:130; Cell Signaling). The cells were visualized using the Laser Scanning Microscope 780 (Zeiss; 63×/1.4 oil-immersion objective), and quantitative analysis was performed with BlobFinder software (Olink Bioscience).

Immunoblot analysis was performed as described (16). The membrane was probed with anti–NF-κB2 (#4729) and anti-TRAF3 (#4882) Abs (both polyclonal; Cell Signaling). Detection of actin (sc-1616, I-19, polyclonal; Santa Cruz) served as a control for the loaded protein amount. Quantification of p52 and TRAF3 relative band intensities was performed using ImageJ software. The relative intensities of the p52 or TRAF3 bands for each sample were divided by the relative intensity of actin of the same sample and then normalized to those of the untreated HC⁺ control, which was equal to 1.

Unpaired two-tailed Student t tests or nonparametric Mann–Whitney tests (with n = 3–5 mice per group) were carried out using Prism software (GraphPad) to determine the statistical relevance of differences between groups.

To study the function of mature B cells lacking the Igα signaling component of the BCR, we generated the cmb1 mouse model, allowing the conditional deletion of the mb-1 gene in B cells. The cmb1 mouse carries the coding sequence for a tamoxifen-inducible CreER^T2 recombinase on one mb-1 allele (Fig. 1A). The second mb-1 allele harbors a “flip-flop” cassette with the coding sequences of exons II–IV of mb-1, followed by an inverted BAP31–enhanced GFP (referred to as GFP for simplicity) sequence encoding a membrane-bound form of GFP. To ensure proper expression of both targeted mb-1 alleles, the new sequences were inserted into exon II of the mb-1 gene. The flip-flop construct is flanked by loxP sites in a way that allows the Cre-mediated inversion and, thus, alternative expression of either Igα or BAP31-eGFP. This enables an easy tracking of mb-1–deficient (Igα^neg) B cells via their GFP expression (16, 25, 26, 52). Five days after receiving a single dose of tamoxifen, expression of GFP was observed in the blood of cmb1 mice. The B cells displayed an intermediate BCR expression in both GFP⁺ and GFP⁻ populations (Supplemental Fig. 1A). At day 10 after treatment, a distinct GFP⁺ population could be detected that no longer expressed IgD-BCR or IgM-BCR on the surface (Supplemental Fig. 1B). At 20 d, the spleen of cmb1 mice harbored 7–9% GFP⁺ Igα^neg B cells, all of which were IgM- and IgD-BCR–negative (Fig. 1B, 1C).

Interestingly, when we stained the splenic B cells of induced cmb1 mice for the maturation markers CD93, CD23, and CD21 (Supplemental Fig. 1C, 1D), GFP⁺Igα^neg B cells were detected in the CD93⁻CD23^hiCD21⁺ FO B cell subset but not in the T1 or T2 transitional or marginal zone B cell subsets (Fig. 1D). As demonstrated by the forward scatter analysis, GFP⁺Igα^neg FO B cells were similar in size to their Igα-sufficient (Igα⁺) counterparts (Fig. 1D). Thus, their general fitness seems to be unaltered. GFP⁺Igα^neg B cells were also found in the lymph nodes (LNs), BM, and peritoneal cavity (PC) of tamoxifen-induced cmb1 mice (Supplemental Fig. 1E–J). As expected, the LN-derived Igα^neg B cells lacked surface expression of IgM-BCR and IgD-BCR (Supplemental Fig. 1F). In the BM, GFP⁺Igα^neg B cells were detected in the mature recirculating subset, as demonstrated by their B220^hiCD24^loCD93⁻ phenotype (Supplemental Fig. 1G, 1H). In the PC, Igα^neg B cells were found in the B1a, B1b, and B2 subsets (Supplemental Fig. 1I, 1J).

To test the signaling function of Igα^neg B cells, we analyzed purified splenic B cells from tamoxifen-induced cmb1 mice for their expression of the activation marker CD86 after 24 h in culture with medium alone or supplemented with various B cell–stimulatory Abs or reagents (Fig. 2A). In contrast to the Igα⁺ controls, the Igα^neg B cells did not upregulate the CD86 marker upon exposure to anti-IgM F(ab′)₂ or anti-CD38 Abs and were only moderately stimulated by anti-CD180 (Fig. 2A). As a control, we verified that Igα^neg B cells still expressed CD180 and CD38 on their surface, although CD180 was expressed at slightly lower levels (Supplemental Fig. 2A). The unresponsiveness of Igα^neg B cells to anti-CD38 and their lower responsiveness to anti-CD180 Abs are in line with previous studies showing that CD38 signaling is dependent on BCR expression, and CD180 uses similar signaling elements as the BCR (53–55). However, other B cell stimuli, such as anti-CD40 Ab, IL-4, and CpG, resulted in CD86 upregulation on both Igα⁺ and Igα^neg B cells, indicating that signaling via the respective receptors does not require BCR expression (Fig. 2A). In a parallel approach, we stimulated cmb1-derived Igα⁺ and Igα^neg B cells with an anti-Igβ–specific Ab. We found that CD86 was expressed on the surface of these cells 24 h after stimulation, although it was minimally reduced compared with Igα⁺ B cells (Fig. 2A). This finding was unexpected because Igα^neg B cells did not express the BCR complex on their surface. Hence, we performed flow cytometric analyses to test whether Igβ alone was expressed on the surface of Igα^neg B cells because it was reported that Igβ could be detected on B cell surfaces upon BCR internalization (56). Indeed, we could detect Igβ on Igα^neg B cells but at lower amounts than on Igα⁺ control cells (Fig. 2B). As an alternative read-out for B cell activation, we monitored intracellular calcium (Ca²⁺) influx upon stimulation of Igα⁺ and Igα^neg B cells with anti-IgM F(ab′)₂, anti-κLC F(ab′)₂, anti-Igβ Abs, or the actin polymerization–inhibiting compound latrunculin A (Fig. 2C). Stimulation with anti-IgM F(ab′)₂, anti-κLC F(ab′)₂, or latrunculin A increased the intracellular Ca²⁺ concentration in Igα⁺ cells but not in Igα^neg B cells, confirming that BCR signaling is defective in Igα^neg B cells. Stimulation with anti-Igβ elicited a moderate Ca²⁺ influx in Igα^neg B cells, in line with the observation that lower levels of Igβ were expressed on the surface of these cells relative to the control (Fig. 2C). The finding that Igα^neg B cells do not respond to latrunculin A in this assay is in line with previous results showing that the calcium response induced by this drug is dependent on BCR signaling (57, 58). B cells lacking an HC have reduced amounts of MHC class I and II molecules and more Fas on their surface (47). We studied the expression of these proteins on Igα^neg B cell surfaces and found that Fas and MHC class I were expressed at similar levels on Igα⁺ and Igα^neg B cells, whereas the expression of MHC class II was decreased 5-fold on Igα^neg B cells (Supplemental Fig. 2B).

We next monitored the numbers of GFP⁺Igα^neg B cells in the spleen of cmb1 mice over time after the termination of Igα expression and found that they remained stable for the next 7–20 d (Fig. 3A, Supplemental Fig. 3A). However, at day 60 after tamoxifen treatment, a reduced, but distinct, population of GFP⁺Igα^neg B cells was detected in the spleen of cmb1 mice (Supplemental Fig. 3A, lower panels). We even detected some GFP⁺Igα^neg B cells in the blood 200 d after tamoxifen treatment of cmb1 mice (Supplemental Fig. 3B). The long-term survival of Igα^neg B cells is in contrast to the rapid death of BCR-negative (BCR^neg) B cells in mice whose Igh or mb-1 gene was deleted using Mx-Cre or CD21-Cre (46–48). To analyze the difference in the behavior of BCR^neg B cells in greater detail, we crossed homozygous mb1-CreER^T2 mice with the B1-8^fl mouse strain, allowing for the conditional deletion of the Igh gene (47). This B1-8^fl mouse strain carries a floxed B1-8 V_HDJ_H gene on one allele of the Igh locus and a deletion of the whole J_H region (J_HT) on the second Igh allele. B1-8^fl/mb1-CreER^T2 mice obtained were treated with the same tamoxifen-application protocol and analyzed in parallel to the tamoxifen-treated cmb1 mice. The HC-deficient (HC^neg) B cells generated with the mb1-CreER^T2 deleter allele were short-lived, and their relative and absolute B cell numbers decreased within 20 d after tamoxifen treatment (Fig. 3B, Supplemental Fig. 3C). Thus, the difference in the in vivo survival of Igα^neg and HC^neg B cells is not due to the use of different Cre constructs or an altered tamoxifen-application protocol.

To learn more about the difference between Igα^neg and HC^neg B cells, we analyzed the intracellular expression of the BCR signaling components Igα and Igβ, as well as the κLC by flow cytometry and qRT-PCR. As expected, Igα protein was absent in Igα^neg B cells, whereas Igβ and κLC protein were expressed inside these cells at identical and slightly reduced levels relative to that of their Igα⁺ counterparts, respectively (Fig. 3C). However, inside HC^neg B cells, the expression of protein for all three BCR components was strongly reduced in comparison with that of HC⁺ B cells (Fig. 3D). These differences in intracellular protein levels were not due to a reduced gene expression in BCR^neg B cells. qRT-PCR analysis showed that, with the expected exception of Igα (Iga), the Igβ (Igb) and κLC (Igk) genes were identically expressed in Igα⁺ and Igα^neg B cells (Fig. 3E). In HC^neg B cells, the transcripts of all three genes were even slightly more abundant than in HC⁺ B cells (Fig. 3F). A parallel analysis of the intracellular protein expression of μHC and δHC confirmed that these two proteins were lacking in HC^neg B cells and were present at nearly identical levels in Igα⁺ and Igα^neg B cells (Supplemental Fig. 4A, 4B). Furthermore, λLC was not expressed in Igα^neg or HC^neg B cells (Supplemental Fig. 4C).

The lower amount of intracellular BCR components in HC^neg B cells can be due to secretion of LCs (59, 60), a reduction in translational efficiency, or an increase in protein degradation. A block of protein translation is associated with an ongoing UPR involving activation of stress sensors in the ER and the phosphorylation of the kinase PERK (61, 62). To test for the induction of UPR in BCR^neg B cells, we monitored the phosphorylation of PERK in isolated B cell populations with an intramolecular PERK:p-PERK PLA. Although the p-PERK level in Igα^neg B cells was unchanged relative to their Igα⁺ counterparts (Fig. 4A, 4B), the levels of p-PERK were elevated in HC^neg B cells compared with the HC⁺ control (Fig. 4C, 4D).

Another UPR component is the inositol-requiring enzyme 1 (IRE1)α, an ER-resident receptor whose endonuclease activity is activated by UPR and mediates the unconventional splicing of the mRNA encoding the transcription factor X-box protein 1 (XBP1). The protein encoded by spliced Xbp1 (Xbp1s) mRNA enters the nucleus where it induces the transcription of several UPR target genes (63). The generation of Xbp1s mRNA can be assessed by a semiquantitative RT-PCR assay (63). No Xbp1s transcripts were detected in freshly isolated Igα^neg, Igα⁺, HC⁺, and B cells carrying one allele of mb1-CreER^T2 (Fig. 4E, lanes 1, 3, 7, and 9), whereas Xbp1s was present in HC^neg B cells (Fig. 4E, lane 5) and in all samples in which UPR was induced using the reducing agent DTT (Fig. 4E, lanes 2, 4, 6, 8, and 10). Notably, HC^neg B cells displayed the most abundant Xbp1s upon DTT treatment compared with other B cell types (Fig. 4E, lane 6).

We next used qRT-PCR to monitor the relative expression of total Xbp1 and Xbp1s, as well as the induction of several UPR target genes, including Dnajb9, Edem1, Hspa5, and Ddit3 encoding the ER stress response–associated proteins ERDJ4, EDEM1, BiP, and CHOP, respectively. The transcripts of all of these genes were more abundant in DTT-treated B cells, in which the UPR has been chemically induced, than in the untreated control (Fig. 4F). Although none of these transcripts was upregulated in Igα^neg B cells (Fig. 4G), HC^neg B cells showed significantly elevated levels of Xbp1, Xbp1s, Dnajb9, and Edem1 transcripts but only a small increase in the levels of Hspa5 and Ddit3 mRNA (Fig. 4H). In summary, these data show that the loss of BCR expression in HC^neg B cells results in UPR, whereas this is not the case in Igα^neg B cells.

Normal mature B cells require the prosurvival factor BAFF for their in vivo or ex vivo maintenance (42, 44, 45, 64, 65). To test whether this also holds true for Igα^neg B cells, we assessed their ex vivo survival in the absence or presence of BAFF. Igα^neg B cells were purified alongside their Igα⁺ counterparts from the spleen and cultured or not with BAFF over a period of 9 d. Similar to the Igα⁺ B cell population, Igα^neg B cells died in the absence of BAFF within 4 d, whereas in the presence of BAFF, ∼50% of the B cells survived in culture for ≥9 d (Fig. 5A, 5B). A parallel analysis of HC⁺ and HC^neg B cells showed that, in the presence of BAFF, 50% of the HC⁺ B cells, but only 20% of the HC^neg B cells, survived for 9 d in culture (Fig. 5C, 5D). Thus, in comparison with Igα^neg B cells, HC^neg B cells were less able to respond to BAFF. Importantly, BAFF-R was found at similar amounts on the surface of Igα^neg and HC^neg B cells, although at slightly reduced levels in comparison with the respective control cells (Supplemental Fig. 4D). Thus, the unresponsiveness of HC^neg B cells to BAFF is not due to the absence of BAFF-R expression. Immunoblot analysis revealed that HC^neg B cells had a defect in NF-κB signaling downstream of BAFF-R. After stimulation with BAFF, the cells displayed lower levels of the NF-κB2 cleavage product p52 compared with the HC⁺ control cells (Fig. 5E). TRAF3 was moderately elevated in HC^neg B cells (Fig. 5E). However, TRAF3 degradation, which is induced upon BAFF-R stimulation, might not be a reliable read-out for BAFF-R signaling defects; a recent study demonstrated that activation of alternative NF-κB signaling may be regulated in a TRAF3-independent manner in B cells (66).

The BAFF-dependent survival of B cells in the mouse can be tested with the help of the BAFF-R–blocking Ab 9B9 (67). Therefore, we injected cmb1 mice (15 d after tamoxifen treatment) with the 9B9 Ab or its isotype-matched control Ab, 5A12, which binds, but does not block, BAFF-R (67). The animals were analyzed by flow cytometry 15 d later for the presence of GFP⁺Igα^neg B cells in the spleen (Fig. 5F). Although treatment with the 5A12 isotype-matched Ab did not have a significant effect on the abundance of GFP⁺Igα^neg B cells compared with that of the untreated controls, treatment with the blocking 9B9 Ab resulted in a strong (3-fold) decrease in GFP⁺Igα^neg B cells in the spleen of cmb1 mice (Fig. 5F). The statistical analysis confirmed that the dependence of Igα^neg on BAFF-R signaling was indistinguishable from that of Igα⁺ B cells in the spleen and LNs of cmb1 mice (Fig. 5G, Supplemental Fig. 4E). A parallel analysis of Igα⁺ and Igα^neg mature recirculating B cells in the BM showed that both B cell populations were reduced by Ab 9B9 treatment (Supplemental Fig. 4F). However, in the PC, only the B2 B cells of Igα⁺ and Igα^neg populations were decreased in absolute numbers; the B1 fractions were not affected by Ab 9B9 treatment (Supplemental Fig. 4G). This is consistent with previously published results (67).

The survival of Igα^neg B cells in tamoxifen-induced cmb1 mice is surprising; a previously published study using the CD21-Cre deleter strain showed that B cells lacking either Igα or the Igα intracellular tail are short-lived (46). Thus, we also generated B cells expressing the tail-truncated Igα by crossing mb1-CreER^T2 and Igα^C1f mice (46). We analyzed the expression of IgM-BCR and IgD-BCR on the surface of Igα^C1f/mb1-CreER^T2 and cmb1-derived B cells before tamoxifen treatment. In comparison with wild-type (WT) B cells, Igα^C1f/mb1-CreER^T2–derived B cells carried lower amounts of IgM and IgD on their surface, and BCR expression was reduced even more on cmb1-derived B cells (Fig. 6A). Due to the CreER^T2 insertion, Igα^C1f/mb1-CreER^T2 and cmb1-derived B cells have only one functional mb-1 allele, explaining the lower expression of the BCR on these B cells. In addition, the cmb1-derived B cells carry an inserted Igα/BAP31-eGFP^inv cDNA cassette on the second allele that apparently reduces Igα production even further (38), resulting in roughly 10-fold reduced BCR levels in comparison with WT splenic B cells. We next treated Igα^C1f/mb1-CreER^T2 mice with the same tamoxifen protocol that we used for the cmb1 mice and monitored the loss of the Igα tail by an intracellular flow cytometric analysis. Five days after application of a single dose of tamoxifen, 93.8% of FO B cells carry the Igα tail truncation (Fig. 6B; designated Igα^∆c for simplicity). However, at day 15, only 45.7% of these mutant B cells were detected, and their relative number decreased further to almost undetectable levels over the next 11 d (Fig. 6B). This analysis suggests that B cells with a tail-truncated Igα are rapidly eliminated in the mouse, as previously described (46). Apparently, the total loss or a tail truncation of Igα has different consequences for the survival of mature B cells in the present setting, relating perhaps to the low levels of BCR expression on the B cells of cmb1 mice during their maturation (Fig. 6A). For further clarification of the rapid demise of Igα tail–truncated B cells, we assessed their survival in response to BAFF ex vivo. The cells were responsive to BAFF to a certain extent, because their viability was enhanced in comparison with untreated Igα tail–truncated B cells (Fig. 6C). However, compared with a WT control, only ∼35% of Igα tail–truncated B cells survived until day 5 after cultivation, whereas 65% of WT B cells were still viable at the same time point (Fig. 6C). In addition, we found that freshly isolated Igα tail-truncated B cells showed no evidence of the UPR, because no Xbp1s transcripts could be detected in these cells (Fig. 6D, lane 1). Interestingly, upon external induction of the UPR with DTT, the Igα tail–truncated B cells, as well as HC^neg B cells, displayed high levels of Xbp1s (Fig. 6D, lanes 3 and 5, Fig. 6E), whereas only low levels of Xbp1s were detected in cmb1-derived Igα^neg B cells and B cells carrying one mb1-CreER^T2 allele (Fig. 6D, lanes 4 and 6, Fig. 6E). This finding reveals major differences between cmb1-derived Igα^neg B cells and B cells with tail-truncated Igα.

The cmb1 mouse model allows the easy identification and tracking of Igα^neg B cells with the GFP marker. We found that all GFP⁺Igα^neg B cells no longer carry a BCR on their cell surface and, unlike WT B cells, cannot be stimulated by anti-BCR, anti-CD38 Abs and are only minimally stimulated by an anti-CD180 Ab. However, these BCR^neg B cells still respond to other B cell stimuli, such as IL-4, CpG, and anti-CD40, as indicated by the upregulation of the activation marker CD86. Surprisingly, we found that, in contrast to HC^neg B cells, GFP⁺Igα^neg B cells can survive in vivo for >20 d. These two BCR^neg B cell populations were generated using the same mb1-CreER^T2 allele. Our finding that HC^neg B cells are short-lived in the induced B1-8^fl/mb1-CreER^T2 mouse is in line with previous data from B1-8^fl/Mx-Cre and B1-8^fl/CD21-Cre mice (46, 47). Clearly, mature B cells require HC expression for their long-term survival in vivo. However, according to our cmb1 mouse study, the loss of Igα and the BCR seems to be tolerated by mature B cells for some time.

When studied in parallel, we found that HC^neg B cells differ from GFP⁺Igα^neg B cells in terms of shape and size and show signs of ER stress and an ongoing UPR (68, 69). In mammals, three distinct, but intersecting, branches of the UPR were described: IRE1α, which mediates the splicing and activation of Xbp1, the ER kinase PERK, and the transcription factor ATF6 (63, 70–72). Although the UPR may be a secondary effect of HC inactivation, it may be responsible for the demise of HC^neg B cells. It is unknown how UPR initiation is triggered in the absence of the HC. It is possible that, immediately after HC ablation, unassembled LCs associate with BiP (73), an event that might sequester high amounts of the chaperone from PERK and IRE1α, instigating a series of events resulting in their activation (74). Indeed, the PERK and IRE1α UPR branches are activated in mature HC^neg B cells, as indicated by the increased PERK phosphorylation and unconventional splicing of Xbp1 transcripts, as well as the increased expression of XBP1s target genes. Activated PERK phosphorylates and inactivates the translation factor eIF2α, thus inhibiting protein translation (72). This is one of the mechanisms used to alleviate ER stress (62, 75, 76). The increased PERK activity could explain the reduced expression of LC, Igα, and Igβ, the BCR components remaining inside the mature HC^neg B cells. Importantly, chronic ER stress and a continuous UPR result in apoptosis via several pathways (77). For example, PERK can phosphorylate FOXO1, a modification that activates this transcription factor and induces apoptosis (78). Of interest in this context is the finding by Srinivasan et al. (48) that mimicking the PI3K pathway activation, which is achieved by expression of a constitutively active form of the PI3K's p110 catalytic subunit (p110*), as well as by ablating either the PI3K-antagonist PTEN or the AKT target FOXO1, promotes survival of mature HC^neg B cells. The increased survival of HC^negp110* B cells could be due to the multiple cross-talk and feedback mechanisms connecting the PI3K pathway to that of UPR. For instance, inhibition of AKT was shown to render cells prone to ER stress–induced apoptosis (79, 80). Indeed, AKT can directly phosphorylate and inhibit PERK, thus rescuing stressed cells from apoptosis (81). Thus, a constitutively active PI3K signal may inhibit PERK and other proapoptotic signaling pathways in HC^neg B cells, resulting in the survival and accumulation of HC^negp110* B cells, as reported (48).

Interestingly, we find that GFP⁺Igα^neg mature B cells without a BCR survive for longer times in tamoxifen-induced cmb1 mice than do mature B cells expressing a mutant BCR with an Igα tail truncation. The latter finding is in line with a previous study of Igα^C1f/CD21-Cre mice (46). It is not clear why the loss of the Igα tail induces B cell death. It is feasible that, inside the resting BCR, the Igα and Igβ tails inhibit each other and that the unpaired Igβ tail induces an apoptosis program. However, in the complete absence of Igα, Igβ, which is expressed on the surface of Igα^neg B cells, can form low-abundance dimers that may not be able to induce B cell death (82). If such a program is indeed induced, it does not seem to be accompanied by a general B cell activation, because the mutant B cells do not show upregulation of activation markers (46). We found no evidence of ongoing UPR in freshly isolated B cells with an Igα tail truncation. However, when the Igα tail-truncated, as well as the HC^neg, B cells were cultured ex vivo under ER stress conditions, they exhibited increased Xbp1s compared with cmb1-derived GFP⁺Igα^neg B cells, indicative of strong IRE1α activity.

Another discrepancy between the present and the previous study (46) is that mature B cells without BCR survive for longer times in tamoxifen-induced cmb1 mice than in Igα^fl/CD21-Cre mice, in which the entire molecule of Igα is inactivated. An explanation for this discrepancy could be the lower abundance of the BCR on mature B cells of cmb1 mice and the fact that, after induction, the Igα/BAP31-eGFP^inv allele can be inverted several times as long as tamoxifen is present. It is feasible that both lower BCR levels on the cell surface and repeated inversion lead to an adaptation of B cells to reduced BCR signals, “priming” them in a way for the receptorless state. Thus, the survival of GFP⁺Igα^neg B cells in cmb1 mice might be an intrinsic effect of this mouse model.

Despite the differences listed above, the study by Kraus et al. (46) and this study both show that Igα^neg B cells, in general, seem to have an increased survival in comparison with HC^neg B cells. For example, in Igα^fl/CD21-Cre and B1-8^fl/CD21-Cre mice, the frequency of BCR^neg B cells is 33 and 14%, respectively (46). The low abundance of HC^neg B cells in CD21-Cre mice could also be due to an ongoing UPR and reduced fitness of these B cells. In terms of their normal biology, mature B cells are in greater danger of losing their HC and LC than the BCR signaling subunits Igα and Igβ. This is due to the fact that the activity of the mutator enzyme AID is focused on the V_H and V_L genes (83, 84). Thus, the somatic mutation process during a germinal center reaction can frequently generate HC^neg and LC^neg B cells. The UPR-mediated apoptosis program may have evolved to ensure that these signaling inert B cells are eliminated so as not to compromise the proper function of the immune system.

After the discovery that HC^neg mature B cells are short-lived, the research field rapidly adopted the model in which the BCR on resting B cells continuously sends out a maintenance or tonic signal required for the survival of mature B cells (37, 47, 85, 86). However, the discoverers of this phenomenon were more cautious in this respect and suggested that “other interpretations are possible, such as the BCR serving as a scaffold for another signaling structure or cell death because of improper processing of its components in the ER” (46). What further contributed to the confusion in the field is that the term “tonic signaling” is also used to describe ligand-independent signaling that is found in the case of the pre-BCR and autoaggregated form of the BCR on B cells derived from chronic lymphocytic leukemia (24, 87). We think that the resting BCR is structurally and functionally distinct from the activated BCR, and that this point needs to be considered in the discussion of these phenomena. Several studies showed that the resting BCR forms an autoinhibited oligomer that is not associated with Syk (8, 9, 88). However, Ag-dependent B cell activation requires interaction of the BCR with Syk, and the same is true for the autonomous signal that is emitted from the autoligated pre-BCR and B-CLL BCR (16, 24, 87). Furthermore, ligand-independent activation of B cells with oxidants, such as pervanadate, or with latrunculin also results in BCR oligomers opening and BCR–Syk interaction and, thus, is not an amplified BCR maintenance signal but rather mimics BCR activation. We think that the same applies to most chimeric AgRs or viral proteins carrying ITAM sequences. Therefore, we suggest that all BCR signaling events involving Syk should be regarded as part of the B cell–activation process and not a BCR maintenance signal.

The main function of the closed BCR oligomers on resting B cells is to set critical thresholds for B cell activation by preventing the uncontrolled interaction of Syk with BCR ITAMs. Whether it also sends out a survival signal is not clear. However, we think that the nature of the survival signal should be distinct from that of the activation signal. At least Syk is not associated with the resting BCR, and mature B cells can survive for extended times in vivo without Syk (9, 16). However, it is also possible that the closed BCR oligomers interact with other receptors on the B cell surface in a way that promotes the selection and survival of B cells. Recent studies suggest that, at nanometer distances, the B cell surface is more organized than previously appreciated (9, 57). For example, we found that, on resting B cells, IgM-BCR and IgD-BCR reside in different protein islands with average diameters of 120–200 nm (10). Interestingly, CD19 is found in close association with the IgD-BCR island (9). This organization could be established once the IgD-BCR is expressed at the T2 and immature B cell stage and may contribute to the selection and maintenance of B cells. However, although CD19 is a well-known activator of the PI3K signaling pathway and implicated in B cell survival (16, 89), it is not known whether CD19 within these structures is the origin of signals in resting B cells.

A well-established maintenance signal for mature B cells is transmitted by BAFF-R via the noncanonical NF-κB signaling pathway (67, 90). It was suggested that this signal is increased by cross-talk between the BCR and BAFF-R (91, 92). Additionally, BCR signaling can increase BAFF-R expression (93). Such a signaling connection could explain the slightly reduced BAFF-R levels on the surface of Igα^neg, as well as HC^neg, B cells. A recent study suggested that BAFF-R ligation can directly activate the ITAM/Syk module of the BCR and that such an interaction of the two receptors is required for BAFF-mediated B cell survival (94). Our findings that GFP⁺Igα^neg B cells display the same response and dependence on BAFF as do WT B cells in vitro and in vivo do not support this model. Clearly, on GFP⁺Igα^neg B cells, BAFF-R provides B cell–survival signals, although these cells do not carry a BCR on their surface. This finding is in line with our previous study showing that, in vivo, Syk is not required for the survival signal of BAFF-R (16). It is possible that BAFF-R signaling requires the expression of either intact BCR or Igβ molecules, which are possibly paired to form homodimers, as might be the case in GFP⁺Igα^neg B cells. This would explain why HC-deficient, as well as Igα tail–truncated, B cells are impaired in their response to BAFF to a certain extent, which might contribute to their rapid elimination. Together, these data show that BAFF-R can signal B cell survival independently of the BCR. Whether the BCR on resting B cells also provides a survival signal requires further studies on the structure and interactions of the autoinhibited oligomeric BCR complex.

We thank Prof. A. Rolink for the anti-BAFF-R Ab. In addition, we thank Drs. L. Leclercq, R. Pelanda, H. Jumaa, W. Römer, and P. Nielsen for reading the manuscript and for helpful scientific discussions.

The authors have no financial conflicts of interest.