IgG1 B cell receptor signaling is inhibited by CD22 and promotes the development of B cells whose survival is less dependent on Igα/β

Starting from ES cells in which the Cδ gene segment replaced that of Cμ (9), we constructed an IgH locus whose C gene cluster was modified to contain only the Cμ and Cγ1 gene segments (Fig. S1). These gene segments were organized in opposite orientations. The new minilocus, which was successfully transmitted into the germ line, was flanked by loxP sites in opposite orientations, such that Cre-mediated recombination would either switch B cells in vivo from IgM to IgG1 expression (or vice versa) or, when induced in the germ line, generate strains of mice exclusively expressing either IgM or IgG1 in the B cell lineage. As our original intention was to compare IgM- and IgG-mediated antigen presentation by B cells, we deleted the polyA sites controlling the production of the secreted forms of the μ and γ1 chains, respectively. Depending on the C_H gene segment in the upstream position, the mutant IgH loci were designated IgH^μγ1 or IgH^γ1μ. As expected, homozygous IgH^μγ1 mice produced a B cell compartment consisting exclusively of B cells expressing the membrane form of IgM, with no IgM being detectable in the blood (unpublished data). In contrast, homozygous IgH^γ1μ mice exclusively produced IgG1-expressing B cells, as further described below. However, contrary to our expectation, large amounts of IgG1 were detectable in the blood of these animals, presumably because of the usage of an alternative polyA site (Fig. 1 E). B cell development and maintenance in this latter mouse strain is the subject of this study.

The IgH^γ1μ locus is schematically depicted in Fig. 1 A, and the populations of B lineage cells in the spleen, BM, and peritoneal cavity of homozygous IgH^γ1μ mice are depicted in Fig. 1 (B–D). A survey of the composition of the various B cell compartments in mutant and control animals is given in Tables I–III,²³. In the spleens of adult mutant mice, transitional, mature B2, and marginal zone (MZ) B cells could be detected on the basis of expression of typical surface markers, and all B cells expressed IgG1 instead of IgM on the surface. In comparison to the controls, the absolute numbers of splenic B cells were normal, but we found a reduction in the fractions of transitional, in particular T1, cells and a two- to threefold increase in the fraction of MZ B cells (Fig. 1 B and Table II). These deviations from the normal situation indicated an impairment of B cell development in the mutant animals that became more apparent when B cell development in the BM was analyzed by flow cytometry. Staining of the cells for B220, c-Kit, CD43, and CD25 revealed a partial block at the pro– to pre–B cell transition (Fig. 1 C and Table I). Accordingly, although the total cell numbers in the BM of IgH^γ1μ/γ1μ and WT mice were comparable (11.9 ± 1.7 and 11.4 ± 1.6 million cells per femur, respectively), the numbers of immature B cells in the BM and the spleen, identified by low levels of B220 and high levels of AA4.1, respectively, were also somewhat reduced in the mutant animals in comparison with the controls (Fig. 1, B and C; and Tables I and II).

In the peritoneal cavity, both B1 and B2 cells were detectable in the mutants, but the fraction of B1a (CD5⁺CD43⁺CD19^high) was substantially reduced compared with the controls (Fig. 1 D).

We conclude that expression of γ1 instead of μ chains is compatible with the development of B cells of all subsets, but that the development of B1 and B2 cells is compromised.

Heterozygous IgH^γ1μ mice allowed us to study the development of γ1 and μ/δ chain–expressing B cells in competition. As expected from the compromised development of IgG1-expressing cells (Fig. 1 C), the peripheral B cell compartment was dominated in this situation by IgM-expressing B cells by a factor of 6 in the spleen and lymph nodes and >10 in the peritoneal cavity (Fig. 2 A, left). Both IgM- and IgG1- expressing cells were long lived, but the IgG1-expressing cells persisted as well as and perhaps even slightly better than their IgM-expressing counterparts over 30 d of treatment with anti–IL-7 antibody, which blocks B cell generation in the BM (Fig. 2 A, right) (21). The longevity of the cells was also determined in IgH^γ1μ homozygous mice by measuring the decay of cells having incorporated BrdU, after a feeding period of 1 mo, over time (Fig. 2 B). Strikingly, the apparent half-life of the mutant cells, averaged over the population of CD19⁺CD21⁺ B cells in the blood, was longer than that of WT cells by a factor of 1.5. This difference does not necessarily indicate an intrinsic property of these cells distinguishing them from their IgM- and IgD-expressing counterparts but could simply reflect the diminished output of B cells from the BM in the mutant animals. Nevertheless, the mutant cells were clearly long-lived, and this explains how IgH^γ1μ mice can generate a peripheral B cell compartment similar in size to that of WT mice despite a defect in B cell generation. It may also explain why the apparent counterselection of the IgG1- expressing cells is stronger at the level of immature and transitional T1 cells as compared with mature B cells (Table IV).

Like mature IgM- and IgD-expressing B cells, the mature IgG1-expressing B cells in heterozygous IgH^γ1μ mice were dependent on BAFF–BAFF-R interaction, as shown by the loss of mature (B220⁺AA4.1⁻), but not transitional (B220⁺AA4.1⁺), splenic B cells in IgH^γ1μ/γ1μ and WT mice

on a BAFF-R–deficient background (Fig. 3 B). Interestingly, the IgG1-expressing B cells in homozygous IgH^γ1μ mice expressed lower levels of BAFF-R than WT B cells (Fig. 3 A), supporting the notion that there is a link between BCR signaling and BAFF-R expression (22).

The B cells of the mutant animals responded vigorously to mitogenic stimuli addressing TLRs or the BCR by proliferation. Although the proliferative responses of the IgG1- expressing B cells through TLRs and CD40 ligation were not notably different from those of control B cells, we observed a distinctly enhanced fraction of highly proliferating mutant cells upon BCR cross-linking (Fig. 4 A). These differences were more pronounced when low doses (1–10 μg/ml) of the cross-linking anti-kappa antibody were used (Fig. 4 A and not depicted).

Although IgG1-expressing cells responded to BCR cross-linking by normal or dampened—rather than enhanced— total tyrosine phosphorylation and c-Jun N-terminal kinase (JNK), extracellular signal–related kinase (ERK), Akt, nuclear factor of activated T cells 2, and IκBα phosphorylation (Fig. 4 B and not depicted), a clear difference between IgG1-expressing and WT cells in the stimulation experiments was enhanced Ca²⁺ mobilization upon BCR cross-linking (Fig. 4 C and Fig. S2). This enhanced response was also seen in mature B cells that had been acutely switched from IgM to IgG1 expression in vivo through induced Cre-mediated recombination (Fig. 4 E). The enhanced Ca²⁺ response in IgG1-expressing B cells is not caused by increased levels of BCR expression (Fig. S3). To address whether it results from an impairment of CD22 inhibition (7), IgH^γ1μ/γ1μ mice were crossed to CD22 knockout animals, and anti-kappa–induced Ca²⁺ mobilization was measured in B cells from the compound mutants. Anti-kappa–induced Ca²⁺ signaling was strongly increased in IgH^γ1μ/γ1μ CD22^−/− B cells compared with the response of IgH^γ1μ/γ1μ B cells, despite the fact that the latter is already considerably stronger than that of WT cells (Fig. 4 C). This increase in the Ca²⁺ response was found over a wide range of anti-kappa antibody concentrations (from 0.4 to 15 μg/ml; unpublished data). Thus, CD22 deficiency causes an increase of Ca²⁺ flux, irrespective of whether the cells express an IgM and IgD or an IgG1 BCR.

We also compared tyrosine phosphorylation of CD22 and Src homology domain 2–containing protein tyrosine phosphatase (SHP-1) recruitment between control (IgM/IgD-expressing) and IgG1-expressing B cells after BCR stimulation. Anti-kappa–stimulated splenic B cells were analyzed by immunoprecipitation of CD22, and equal tyrosine phosphorylation of CD22 was found in splenic B cells of control IgH^γ1μ/γ1μ mice. We also could not detect any quantitative differences in SHP-1 recruitment to CD22 in the two types of cells when several experiments were quantified (Fig. 4 D). This is in contrast to the published results from cell line studies in which IgG2a B cells did not show any CD22 tyrosine phosphorylation and SHP-1 recruitment after stimulation with antigen (7). In that report, it was also postulated that there may be diminished association of CD22 to BCRs of the IgG2a isotype. Therefore, we performed immunoprecipitation with Igκ-specific antibodies and looked for the amount of coprecipitated CD22. We found that the same amount of CD22 protein could be coprecipitated from unstimulated and stimulated B cells of control and IgH^γ1μ/γ1μ mice and that the coprecipitated CD22 exhibited comparable tyrosine phosphorylation (Fig. 4 D). We conclude that in IgG1- expressing B cells of IgH^γ1μ/γ1μ mice, normal activation of CD22-dependent pathways takes place. The higher Ca²⁺ response of IgH^γ1μ/γ1μ mice can also not be attributed to changed B cell subpopulations in their spleens, because a similarly increased Ca²⁺ response was detected in lymph-node B cells (Fig. S2).

We also addressed the question of whether CD22 deficiency affects B cell subsets in IgH^γ1μ/γ1μ mice. In the BM, a similar composition of B cells was found in IgH^γ1μ/γ1μ CD22^−/− and IgH^γ1μ/γ1μ mice. Although CD22^−/− mice show a characteristic reduction of mature recirculating B cells (B220^hi IgM⁺), a reduction of B220^hi IgG1⁺ cells was not seen in IgH^γ1μ/γ1μ mice on a CD22-deficient background (Fig. S4 A). In the spleen, a considerably higher number of IgG1⁺ B cells was present in heterozygous IgH^γ1μ/+, CD22-deficient mice compared with their CD22-proficient counterparts (Fig. S4 B). The MZ B cell compartment was enlarged in both CD22-proficient and -deficient IgH^γ1μ/γ1μ mice, contrasting the reduction of MZ B cells in CD22-deficient mice carrying WT IgH loci (however, note that differences in genetic backgrounds may play a role in this case; Fig. S4 C) (23). Similar to IgH^γ1μ/γ1μ mice, IgH^γ1μ/γ1μ CD22^−/− mice had a strongly reduced B1a cell population in the peritoneal cavity (Fig. S4 D).

Normal B cell development is severely compromised in mice carrying a mutation in the mb-1 gene that leads to a partial ablation of the cytoplasmic tail of the Igα signal transducer. Such animals generate <1% of normal mature B cell numbers, and these cells do not accumulate, presumably because of a shortened life span (24, 25). We wondered whether expression of the γ1 chain in the BCR would rescue B cell development and maintenance in mice carrying a truncated Igα chain. Therefore, double-mutant mice homozygous for the Igα truncation mutation (Igα^Δc1) and homo- or heterozygous for IgH^γ1μ were generated. The Igα truncation clearly impeded B cell development in the IgH^γ1μ mice, as reflected by reduced fractions of immature and mature B cells in the BM and spleen (Fig. 5, A and B; and Tables I and II). Like in WT mice, an Igα-null mutation led to a complete block of B cell development in IgH^γ1μ/γ1μ mice (Fig. S5). However, in comparison to mice carrying the Igα truncation on a WT background, Igα^{Δc1/ Δc1} IgH^γ1μ/γ1μ double mutants displayed a 10-fold larger fraction of mature B cells in the spleen (Fig. 5 B and Table II), which were partly organized into small B cell follicles (Fig. 5 E) and had a life span indistinguishable from that of their Igα-sufficient counterparts (Fig. 2 B). In accord with these results, when the Igα mutation was combined with IgH^γ1μ/+ heterozygosity, we observed a dramatic preponderance of cells expressing IgG1 BCRs over those expressing BCRs associated with IgM (Fig. 5 D). We conclude that the presence of γ1 chains in the BCR makes the maintenance, but not the development, of the cells independent of the Igα cytoplasmic tail.

Curiously, combining the IgH^γ1μ with the Igα^Δc1 mutation promoted B1a cell development in both the peritoneal cavity and spleen beyond what is seen in IgH^γ1μ/γ1μ mice (Fig. 5 C and Tables II and III). The fraction of these cells was strongly reduced in IgH^γ1μ mice, but when the majority of the B2 cells were depleted through the Igα mutation, they became the dominant fraction in the peritoneal cavity. We also noted that the splenic B cells developing in the double-mutant mice had increased surface levels of IgG1 BCRs, with some increase being detectable even in animals heterozygous for the Igα truncation (Fig. 5 D). However, the levels of BAFF-R on these cells were unaffected by the Igα truncation (unpublished data).

We describe a mouse strain whose IgH locus is engineered such that either membrane-bound γ1 or μ chains can be produced through Cre-mediated inversion of a loxP-flanked C_H region gene minilocus containing the Cγ1 and Cμ gene segments in opposite orientation. This strain can be used either for switching B cells from IgM to IgG1 expression, and vice versa, through induced Cre recombinase expression or to generate strains in which B cell development relies on the exclusive usage of membrane-bound μ or γ1 chains, respectively. In this paper, we use this system to study B cell development and maintenance in the latter situation. A main advantage of the present approach over earlier experiments, in which the developmental potential of IgH transgenes encoding H chains of various isotypes was analyzed, is that we can study B cell development under conditions in which the cells go through V_H region diversification, like in WT mice, instead of starting from a prerearranged IgH transgene encoding a single V_H region. We have indeed observed that homozygous IgH^γ1μ mice produce a highly diverse repertoire of IgH chain–variable regions in their mature B cell compartment, indistinguishable in a first approximation from that of WT mice (unpublished data). The present system also avoids the serious complications of copy number and position- dependent variations of transgene expression (15), in that IgH expression is controlled by the very same V_H promoters that are used in normal B cell development.

In homozygous IgH^γ1μ mice, all B cells indeed express IgG1, and in the blood of the animals we detect IgG1 but no IgM antibodies. Although B cells of all subsets are generated and the overall size of the peripheral B cell compartment is similar to the one for WT mice, B cell development in the BM is clearly compromised, with a partial developmental block at the pro– to pre–B cell transition. The nature of this block, whose elucidation will require the generation and analysis of pro–B cell lines from the mutant animals, likely lies in the inadequate assembly or signaling capacity of the pre-BCR (pBCR) in the mutant animals, given that the pBCR plays a crucial role at this very point in development (26, 27). Indeed, in pBCR-deficient mice, B cell development is largely blocked at the pro– to pre–B cell transition, at which the cells go through a pBCR-dependent phase of rapid proliferation. The developmental block seen in the IgH^γ1μ animals is clearly much milder, in that not only do many more B cells develop in these animals over time but also a normal fraction of large, cycling cells is seen in the pre–B cell compartment (unpublished data), which are lacking in pBCR-deficient mice (28). We therefore speculate that a γ1 chain–containing pBCR-like molecular complex is assembled in the animals but is functionally compromised. This problem is currently under investigation.

As predicted by this scenario, the mutant cells compete poorly in development with their WT counterparts in heterozygous IgH^γ1μ mice. This effect is more dramatic at the level of immature B cells in the BM and transitional T1 B cells in the spleen than that of mature splenic B cells. The accumulation of IgG1-expressing mature B cells in these animals could be explained by prolonged average life spans for these cells as compared with those of WT, IgM-, and IgD-expressing cells. At face value, this is indeed independently suggested by the BrdU chase data in Fig. 2 B, but one should bear in mind that these data could mainly reflect the dependence of B cell turnover on B cell output from the BM. It is well established that the average half-life of B cells is dramatically prolonged when B cell generation in the BM is blocked (29).

Homozygous IgH^γ1μ/γ1μ mice showed characteristic shifts in peripheral B cell populations, with reduced numbers of B1 cells and higher splenic MZ B cell counts. Interestingly, a companion study by Horikawa et al. using transgenic mice that express a μ/γ1 transgenic BCR also found an increased MZ B cell population (see Horikawa et. al [30] on p. 759 of this issue). Although the MZ B cell expansion may be related to a decreased cellular output from the BM, the B1 cell defect is likely caused by improper signaling properties of the IgG1 BCR. Our biochemical data from splenic B cells argue that these are not based on an uncoupling of the inhibitory CD22 receptor from the BCR. The B1 cell defect is likely also CD22 independent, because CD22 has been shown not to modulate the Ca²⁺ responses of B1 cells (31).

Mature B cells expressing an IgG1 BCR exhibited enhanced Ca²⁺ mobilization upon BCR cross-linking as compared with WT B cells, irrespective of whether the cells had gone through development as γ1-expressing cells or had their BCR switched from IgM to IgG1 after maturation. Although the biochemical basis of this enhanced signaling capacity of the IgG1 BCR remains to be elucidated, it clearly does not reflect an inability of the inhibitory CD22 receptor to dampen the response of the mutant cells, as earlier work on B cell lines expressing an IgG2a BCR had indirectly suggested (7). This is also born out in the accompanying study by Horikawa et al. (30) in which antigen-specific IgG1- and IgM-transgenic mice are compared with respect to their BCR responses and the role of CD22 inhibition. In the work of Wakabayashi et al. (7), an antigen-specific γ2a chain was stably expressed in various B cell lines. It is therefore possible that the results obtained in that study reflect a peculiarity of the particular IgH chain used or of the B cell clones selected. In the present system, as well as in that of Horikawa et al. (30), mice are used that express a diverse repertoire of γ1 chains in the B cell compartment in vivo. As γ2a and γ1 chains are very similar and no different functions have as yet been attributed to their cytoplasmic tails (32), we speculate that the present results may reflect common signaling properties of IgG BCRs in normal physiology.

Although the enhanced Ca²⁺ response to BCR cross-linking in B cells expressing IgG1 BCRs may reflect the unique signaling properties of the latter, we have so far been unable to identify signaling cascades that are selectively activated through the BCR in IgG1-expressing cells. Indeed, phophorylation of known players in BCR signaling like JNK, ERK, Akt, nuclear factor of activated T cells 2, and IκBα was normal or dampened, rather than enhanced, in these cells compared with their IgM- and IgD-expressing counterparts. Total tyrosine phosphorylation was also not enhanced in IgG1 cells. These issues clearly need further investigation, particularly in cells acutely switched from IgM to IgG1 expression.

We did not observe enhanced proliferation of the mutant cells upon TLR ligation in vitro, but upon BCR cross-linking, the fraction of highly proliferating cells was increased. The latter effect is likely not caused by an increased fraction of MZ B cells in the mutant mice, because follicular B cells proliferate better than MZ B cells upon BCR cross-linking (33). Our results are in line with earlier experiments by Martin and Goodnow (6) in which B cells expressing a transgenic IgG1 of defined antigenic specificity were triggered by antigen. The mutant cells in the latter study generated strikingly larger numbers of progeny cells than antigen- activated cells expressing transgenic IgM and IgD because of their enhanced survival. In the present experiments, the enhanced in vitro proliferation upon BCR cross-linking, as well as the accumulation of IgG1-expressing cells in the mature B cell compartment in the heterozygous mutant animals (i.e., when the cells compete with IgM- and IgD-expressing cells), may similarly reflect an enhanced efficiency by which survival signals are delivered to the cells via the BCR upon isotype switching, as it typically occurs in the generation of memory B cells. It will be interesting to see whether we can use the present experimental system to bestow enhanced in vivo survival to IgM-expressing cells by switching them to IgG1 expression through induced Cre recombinase activity and, thus, to assess whether expression of the IgG1 BCR as such increases the life span of these cells.

As the BCR acquires a new signaling module in terms of the cytoplasmic tail and, perhaps, other parts of the newly expressed IgH chain upon isotype switching, it was of interest to ask whether the cytoplasmic tails of the Igα/β heterodimer, which mediate signal transduction through IgM and IgD BCRs, are required for the maintenance of B cells expressing an IgG1 BCR. Indeed, we had previously shown that the Igα cytoplasmic tail is required for the survival of mature B cells in WT mice (24, 25). We now demonstrate that, in contrast to this result, the partial truncation of the Igα cytoplasmic tail does not impede the survival of B cells expressing an IgG1 BCR. The simplest interpretation of this finding is that, in accord with the considerations above, the γ1 chain, likely through its conserved cytoplasmic tail, indeed delivers a survival signal to the cells, and that this signal contributes to the longevity of memory B cells expressing class-switched BCRs. However, we also note that in the compound mutant animals, B cell development is profoundly compromised, in that the animals produce only one tenth of the normal numbers of B cells. Although this is still roughly 10 times more cells than in animals carrying a WT IgH locus in combination with the Igα truncation (24, 25), we cannot exclude that the cells generated in these animals are a selected population of B cells that are not representative of the B cell population generated in WT mice.

Igα^ΔC1/ΔC1, Igα^ΔTm/ΔTm, BAFF-R^−/−, and CD22^−/− mice were described previously (24, 34, 35) and kept on C57BL/6 (Igα^ΔTm/ΔTm and BAFF-R^−/−), 129 (CD22^−/−), or mixed 129 × C57BL/6 (Igα^ΔC1/ΔC1) genetic backgrounds. Mice were analyzed at the ages of 6–16 wk. The generation of the IgH^μγ1 and IgH^γ1μ mouse strains is described in Fig. S1. Both strains were generated from BALB/c-derived ES cells and kept on a BALB/c background. The sequence of the mutant locus is available from GenBank/EMBL/DDBJ under accession no. EF495199. Mice were bred and maintained under specific pathogen-free conditions. Animal care and experiments were conducted according to protocols approved by the Harvard University Institutional Animal Care and Use Committee and by the CBR Institute for Biomedical Research.

Preparations of cell suspensions from lymphoid organs and cell-surface stainings were performed as previously described (36). Data were acquired on a FACSCalibur (BD Biosciences) and analyzed using CellQuest (BD Biosciences) and FlowJo (TreeStar Inc.) software. Dead cells were labeled with propidium iodide or TO-PRO-3 iodide (Invitrogen) and excluded from the analysis. Monoclonal antibodies R33-24.12 (anti-IgM), 1.3-5 (anti-IgD), RA3-6B2 (anti-B220), R33-18-10 (anti-Igκ), and M5/114 (anti-IA^b) were prepared and conjugated in our laboratory. Monoclonal antibodies to CD5, CD19, CD21/35, CD22, CD23, CD24, CD25, CD43, CD69, CD86, CD95, CD117, Igλ, IgG1, and H2Kb were purchased from BD Biosciences. The monoclonal antibody to AA4.1 was purchased from eBioscience. AA4.1-PE (anti-C1qR_p) was a gift from M.P. Cancro (University of Pennsylvania School of Medicine, Philadelphia, PA). Rabbit polyclonal antibodies to mouse BAFF-R were generated as previously described (34). Ig serum concentrations were determined by ELISA, as described previously (36).

0.8 mg/ml BrdU (Sigma-Aldrich) was given in the drinking water for 31 d. Lymphocytes were surface stained for CD19 and CD21, and BrdU incorporation was determined with the BrdU Flow Kit (BD Biosciences).

To block B cell lymphopoiesis in the BM of IgH^g1m/+ heterozygous mice, 1 mg anti–IL-7R monoclonal antibody (A7R34) was injected i.v. every second day.

2–10 × 10⁶ B cells/ml were incubated with carboxyfluorescein succinimidyl ester (CFSE; Invitrogen) in RPMI 1640 (Invitrogen) containing 5 mM CFSE for 5 min at 37°C. CFSE-labeled cells were cultured in 200 μl complete DMEM (Invitrogen) at a concentration of 4 × 10⁶ cells/ml in 96-well microtiter plates at 37°C in the presence of different mitogenes. Cells were analyzed by flow cytometry. For activation, anti-RP105 antibody (eBioscience), LPS (Salmonella minnesota Re 595; Sigma-Aldrich), IL-4 (PeproTech), anti-CD40 (BD Biosciences), CpG (phosphothioate-stabilized CpG-ODN [ TCCATGACGTTCCTGATGCT]; TIB MOLBIOL), or anti-IgM F(ab′)₂ fragments (Jackson Immuno Research Laboratories) or anti-kappa (Southern Biotechnology Associates, Inc.) were used.

Splenocytes were resuspended in RPMI 1640 supplemented with 1% FCS (5 × 10⁶ cells/ml) and incubated with 4.5 μM Indo-1 and 0.003% Pluronic F-127 (Invitrogen) at 37°C for 45 min. Afterward, splenocytes were stained on ice with anti–Mac-1–FITC and anti-CD5–PE (BD Biosciences), followed by acquisition with a FACSVantage (BD Biosciences) and stimulation with anti-kappa at the concentrations indicated in Fig. 4. Data were analyzed using FlowJo software.

After red blood lysis, splenic B cells were negatively selected using CD43 magnetic beads (Miltenyi Biotec) and stained with B220, CD21, and CD23 antibodies. Follicular B cells (B220⁺, CD21⁺, CD23^bright) were sorted with a FACSVantage and resuspended in complete B cell medium at a concentration of 2 × 10⁶ cells/ml. After preincubation for 1 h at 37°C, cells were stimulated with 20 μg/ml anti-kappa and lysed in RIPA lysis buffer (40 mM Tris, pH 7.6, 150 mM NaCl, 2 mM EDTA, 1% NP-40, 0.25% sodium deoxycholate, 5 μg/ml aprotinin, 2 μM pepstatin, 10 μg/ml leupeptin, 1 mM PMSF, 5 mM NaF, 0.5 μM okadaic acid, 1 mM Na₃VO₄). Protein concentrations were determined by a Protein Assay Kit (Bio-Rad Laboratories), and 6 μg of protein per lane was separated by a 10% SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Millipore). The membranes were immunoblotted with phospho-ERK1/2 (T202/Y204), phospho-JNK1/2 (T183/Y185), phospho-Akt (S473), phospho-IκBα (S32/36), anti-ERK1/2, anti-JNK1/2, anti-Akt (Cell Signaling), anti-IκBα (Santa Cruz Biotechnology, Inc.), and anti–β-actin (Sigma-Aldrich) antibodies.

For co-immunoprecepitation experiments, splenic B cells were purified after erythrocyte lysis by a complement lysis of T cells. 10⁷ B cells/ml were stimulated with anti-kappa at 37°C for the times indicated in the figures. After stimulation, cells were lysed in lysis buffer (50 mM Tris [Roth], pH 7.5, 6.6 mM NaCl [Roth], 0.5 mM EDTA [Applichem], 1 μg/ml aprotonin [Roche], 5 μg/ml leupeptin [Roche], 1 mM PMSF [Roche], 1 mM Na₃VO₄ [Sigma-Aldrich], 1% Brij-58 [Sigma-Aldrich]) and incubated for 30 min on ice. The lysates were centrifuged and immunoprecipitated with rabbit anti-CD22 (a gift from P. Crocker, University of Dundee, Dundee, UK) and protein A–sepharose (GE Healthcare) or anti-kappa antibody and protein G–sepharose (GE Healthcare). Immune-complexed coated beads were collected after overnight incubation and washed three times with ice-cold lysis buffer. After boiling the beads for 3 min in a reducing sample buffer, supernatants were separated by a 7.5% SDS-PAGE and transferred to nitrocellulose membranes (BioTrace NT; Pall). Membranes were incubated with primary antibodies 4G10 (Upstate Biotechnology), SHP-1 (Biozol), and anti-kappa overnight at 4°C. Secondary antibodies were goat anti–mouse IgG–horseradish peroxidase (HRP; Jackson ImmunoResearch Laboratories) and goat anti–rabbit IgG-HRP (Santa Cruz Biotechnology, Inc.). Membranes were developed with ECL (GE Healthcare). Blots were scanned, and band intensities were quantified using ImageJ software (National Institutes of Health).

To induce in vivo inversion of the IgH^μγ1 allele with subsequent expression of IgG1 instead of IgM in B cells, IgH^μγ1/JHT Cre-ER^T2 (37) mice were fed for 3 d with 1–5 mg tamoxifen (Sigma-Aldrich) per mouse by gavage (depending on the weight of the recipients). 3 d after discontinuation of tamoxifen, splenocytes from treated mice were harvested, and B cells were negatively selected using CD43-MACS beads (Miltenyi Biotec). After purification, cells were stained with Cy5-conjugated anti-IgM Fab fragments, washed twice, and labeled with Fluo-3 and Fura Red (Invitrogen). Ca²⁺ flux of B cells resuspended at 2 × 10⁶ cells/ml in Krebs-Ringer solution (1 mM Ca²⁺) was assayed by flow cytometry (FACSCalibur), measuring the fluorescence emission of Fluo-3 in the FL-1 channel and Fura Red in the FL-3 channel. Data were analyzed using FlowJo software.

For histological staining of B cells and T cells, frozen 6-μm sections were thawed, air dried, fixed in acetone, and stained for 1 h at room temperature in a humidified chamber with rat anti-CD19 (ID3; BD Biosciences) and biotinylated anti-CD3 (145-2C11; BD Biosciences), followed by HRP-conjugated goat anti–rat IgG and alkaline phosphatase–conjugated streptavidin (Southern Biotechnology Associates, Inc.).

Fig. S1 shows the targeting strategy to generate IgH^γ1μ or IgH^μγ1 alleles. Fig. S2 and Fig. S3 demonstrate enhanced Ca²⁺ responses in lymph-node B cells of IgH^γ1μ/γ1μ mice or acutely switched splenic B cells, respectively. Fig. S4 shows that CD22 deficiency does not affect B cell development in IgH^{γ1μ/ γ1μ} mice but results in accumulation of a higher number of IgG1⁺ cells in heterozygous mice. Fig. S5 demonstrates that IgG1⁺ B cells do not develop in the absence of Igα.

We thank Kevin Otipoby and Gunther Galler for discussions, Christine Patterson for critical reading of the manuscript, Claudia Uthoff-Hachenberg, Angela Egert, Christian Linden, and Jing Wang for excellent technical help, and John P. Manis for help with planning the targeting construct and probes.

This work was supported by the FP6 Marie Curie Research Training Network (grant MRTN-CT-2004-005632 to A. Waisman), the Deutsche Forschungsgemeinschaft (grant SFB 243 to K. Rajewsly, grant SFB 490 to A. Waisman, and grant SFB 466 to L. Nitschke), and the National Institutes of Health (grant 1 R37 AI054636-01). J. Seagal was supported by a Dorot fellowship and a fellowship from the Cancer Research Institute, J. Song was supported by the Center for Molecular Medicine Cologne, Y. Sasaki was supported by the Uehara Memorial Foundation, and D. Melamed was supported by a fellowship from the International Union Against Cancer.

The authors have no conflicting financial interests.