Reactivation of IgG-switched memory B cells by BCR-intrinsic signal amplification promotes IgG antibody production

Secondary antibody responses are marked by faster kinetics, improved antibody affinity and a switch from IgM to other immunoglobulin isotypes, most notably IgG, compared with primary responses. These changes protect from reinfection and represent the principle of most vaccination strategies. Yet, the molecular mechanisms that underlie B-cell memory responses are unclear. Here we show, by inactivating the immunoglobulin tail tyrosine (ITT) signalling motif of membrane-bound IgG1 in the mouse, that the ITT facilitates maintenance and reactivation of IgG-switched memory B cells in vivo. The ITT motif equips IgG-switched cells with enhanced BCR signalling capacity, which supports their competitiveness in secondary immune reactions and drives the formation of IgG-secreting plasma cells even in the absence of T-cell help. Our results demonstrate that ITT signalling promotes the vigorous production of IgG antibodies and thus provide a molecular basis for humoral immunological memory.

T he production of antibodies by B lymphocytes is initiated on recognition of extracellular antigen by clonotypic antigen receptors expressed on the B-cell surface. On secondary antigen encounter, antibody responses are more vigorous and dominated by the production of immunoglobulin G (IgG) antibodies. However, the mechanisms that encode such improved memory responses remain unclear. Binding of antigen to B-cell antigen receptors (BCRs) initiates a complex series of signalling events that, in combination with costimulatory signals, promote cellular activation and differentiation into antibodyproducing plasma cells 1 . A monomeric BCR unit is composed of a membrane-bound immunoglobulin (mIg) molecule of one of the five classes, mm, dm, gm, em, or am, that non-covalently associates in a 1:1 stoichiometry with a heterodimer of mIgassociated a and b proteins (Iga/b, CD79a/b) on the cell surface 2,3 . Antigen-mediated reorganization of BCR complexes triggers the activation of cytoplasmic protein tyrosine kinases (PTKs) that phosphorylate immunoreceptor tyrosine-based activation motifs (ITAMs) in the cytoplasmic domains of Iga and Igb (refs 4-7). Phosphorylated ITAM tyrosines serve as docking sites as well as allosteric activators for PTKs of the spleen tyrosine kinase (Syk) family 8,9 . In addition, Iga contains an evolutionary conserved non-ITAM tyrosine residue (Y204) that on phosphorylation recruits the central adaptor protein Src-homology 2 (SH2) domain containing adaptor protein of 65 kDa (SLP65, also called BLNK) via its SH2 domain 10 . Syk-mediated phosphorylation of SLP65 then enables the formation of a supramolecular signalling platform from which the activation of ubiquitous signalling cascades is initiated 11 .
The importance of individual ITAM and non-ITAM tyrosine residues in Iga and Igb for the development of the B-cell compartment and antibody production has been analysed in a set of mouse strains expressing tyrosine-to-phenylalanine (YF) mutant versions of Iga or tyrosine-to-alanine mutant Igb (refs 12-14). Inactivation of either ITAM motif alone has little impact on B-cell development and overall antibody production. Combined inactivation of both BCR ITAMs, however, causes an almost complete block in B-cell development at the proB-to preB-cell transition 13 . The SLP65-binding non-ITAM tyrosine motif in Iga is dispensable for the development of B2 B cells but has a specific function in supporting T-independent antibody responses 14 .
Immunization with T-dependent antigens initiates the formation of germinal centres in which activated B cells can undergo affinity maturation and immunoglobulin class switching. The cellular outputs of germinal centres are antibody-secreting plasma cells and memory B cells that are poised to differentiate into plasma cells on antigen reencounter 15 . Immunoglobulin class-switch recombination to IgG or IgE isotypes equips B cells with altered BCRs that in addition to the canonical Iga/b signalling subunit contain signalling motifs in the cytoplasmic domains of the respective mIg heavy chains, which are absent in mIgM-and mIgD-BCRs of naive cells. On antigen binding, these immunoglobulin tail tyrosine (ITT) motifs become phosphorylated and serve as docking sites for the ubiquitous cytoplasmic adaptor protein growth factor receptor-bound 2 (Grb2). Recruitment of Grb2 and its binding partner Bruton's tyrosine kinase (Btk) into the signalosome of mIgG-and mIgE-containing BCRs amplifies BCR-proximal signalling events, including phosphorylation of SLP65 and subsequent Ca 2 þ mobilization [16][17][18] . The essential importance of the entire cytoplasmic domains of mIgG and mIgE for antibody responses of the respective isotype was demonstrated in genetically engineered mouse models [19][20][21] . Whether or not the ITT motif is responsible for the function of the cytoplasmic tails of mIgG and mIgE in vivo, remained, however, unknown.
Here we demonstrate that inactivation of the ITT by a YF substitution in mIgG1 leads to substantially impaired IgG1 antibody responses in the mouse. This is due to the reduced generation of IgG1-secreting plasma cells despite normal memory cell formation. In heterozygous animals, cells expressing YFmutant mIgG1 have a strong competitive disadvantage in entering the plasma cell compartment compared with cells expressing wild-type mIgG1. Furthermore, the ITT improves the reactivation of memory B cells in a T-cell-independent manner. Our results show that ITT signalling is essential for B-cell memory in vivo.

Results
ITT phosphorylation enhances BCR-proximal signalling. To investigate the function of the ITT signalling motif in vivo, we introduced a point mutation into the genome of C57BL/6 mice that causes a YF substitution in the cytoplasmic domain of mIgG1 (mIgG1-YF; for details see Supplementary Figs 1 and 2). First, we compared the signalling capabilities of YF-mutant and wild-type mIgG1-BCRs in primary cells and used phosphorylation of the adaptor protein SLP65 and the serine/threonine kinase v-akt murine thymoma viral oncogene homologue 1 (Akt, also called protein kinase B, PKB) as readouts for the activation of BCRregulated signalling pathways ( Fig. 1a,b, respectively, and Supplementary Fig. 3A). In line with our previous observations in transfected B cells 16,17 inactivation of the ITT caused a substantial reduction in phosphorylation of SLP65 under optimal and suboptimal stimulation conditions (Fig. 1a). By contrast, the activation of Akt was not affected by the ITT mutation (Fig. 1b), showing that the ITT specifically improves the activation of Ca 2 þ -regulating components in IgG-switched cells. As control, splenic B cells of both genotypes were treated with F(ab') 2 fragments to IgM, which demonstrated their equal signalling capacities ( Supplementary Fig. 3B). Next, we tested whether the enhanced phosphorylation of SLP65 was associated with augmented Ca 2 þ mobilization in IgG1-switched B cells. To this end, purified splenic B cells from immunized mice were loaded with Indo-1 and stained with FITC-conjugated non-activating monomeric anti-IgG1 Fab fragments. Subsequently, mIgG-BCRs were stimulated with polyclonal anti-IgG F(ab') 2 fragments. This setting allowed us to monitor BCR-induced Ca 2 þ mobilization selectively in mIgG1-positive cells ( Fig. 1c and Supplementary  Fig. 4A). Inactivation of the ITT caused a strong reduction in BCR-induced Ca 2 þ mobilization as previously shown in transfected B cells. As control, splenic B cells of both genotypes were treated with F(ab') 2 fragments to IgM ( Supplementary  Fig. 4B). In conclusion, phosphorylation of the ITT motif greatly improves mIgG-BCR signalling by promoting BCR-induced Ca 2 þ signalling even under conditions of suboptimal stimulation. This improved reactivity of mIgG-BCRs is brought about by the recruitment of a Grb2/Btk signalling module to the ITT motif, which facilitates activation of phospholipase C-g and subsequent second messenger generation 16,17 .
Germinal centre B cells do not require ITT signalling. Membrane-bound IgG-containing BCRs are expressed for the first time in the life of a B cell following successful immunoglobulin class-switch recombination, which predominantly occurs during germinal centre reactions. To test whether B cells require ITT signals to prevail in the competitive environment of a germinal centre, we investigated the cellular composition of germinal centres in mIgG1-YF mice during a primary immune response against the hapten 4-hydroxy-3-nitrophenylacetyl (NP) coupled to the carrier protein keyhole limpet haemocyanin (KLH). We followed the generation of mIgG1-expressing cells during the first 15 days of the primary response and used CD38 as a marker to differentiate between memory (CD38-positive) cells and activated germinal centre (CD38-negative) B cells 22 ( Supplementary Fig. 5). The kinetics of mIgG1-positive germinal centre B-cell generation during the primary response was very similar in mIgG1-YF and wild-type mice, even though there was a trend towards a somewhat higher production in wild-type mice at days 11 and 15 (Fig. 2a). Also, mIgG1-expressing cells specific for the hapten NP were produced in similar amounts in both genotypes ( Fig. 2b and Supplementary Fig. 6) as were NP-specific mIgM-expressing cells (Fig. 2c). Furthermore, we analysed the generation of antibody-secreting cells during the primary response to NP-KLH. The numbers of IgG1-secreting cells in the spleens of both genotypes were comparable during the first 9 days of the primary antibody response. After that period the numbers of IgG1-producing cells in wild-type animals were twoto threefold higher than in mIgG1-YF animals (Fig. 2d). The numbers of IgM-secreting cells were the same in both animal groups throughout the experiment (Fig. 2e). Thus, ITT signalling does not seem to be a major driving force for the generation of IgG-switched B cells in the highly selective environment of a germinal centre but seems to favour the formation of IgGsecreting cells already in the primary immune response.
The ITT improves the maintenance of IgG-switched B cells. Even though ITT signals were not essential for the generation of IgG-switched cells during germinal centre reactions, the possibility remained that their survival in the periphery may be regulated by the ITT. Hence, to test whether the ITT motif contributes to the maintenance of memory B cells, we determined the frequency of NP-specific mIgG1-positive B cells in the spleens of wild-type and ITT-mutant mice 11 weeks and 36 weeks after immunization with NP-KLH ( Supplementary Fig. 7). Whereas the frequency of antigen-specific mIgG1-expressing cells was similar in both genotypes 11 weeks after immunization (Fig. 3a), their number was roughly four times higher in wild-type animals 36 weeks after immunization (Fig. 3b). We conclude that the presence of intact ITT motifs in the cytoplasmic domains of membrane-bound IgG molecules improves the lifespan of IgGswitched memory B cells, presumably by enhanced tonic signalling of ITT-containing mIgG-BCRs 23 .
ITT signalling promotes production of IgG antibodies. To investigate the influence of ITT signals on IgG production, we analysed the kinetics of IgG1 antibody formation and affinity maturation against NP in ITT-mutant and wild-type animals. Total IgG1 antibody titres in mIgG1-YF mice increased much slower than in wild-type animals (Fig. 4a), and were reduced approximately by half on day 28 of the primary response (Fig. 4b). These differences in IgG1 production during the primary response are consistent with the differences in early plasma cell formation (Fig. 2d). The isotypes IgG2b and IgG2c were moderately elevated in the sera of mIgG1-YF mice (Fig. 4c,d) indicating the existence of compensatory mechanisms as observed before 19 . Next, we compared the kinetics of total and high-affinity NP-specific IgG1 antibodies in mIgG1-YF and wild-type mice during primary and secondary immune responses. Note that reimmunization was done at a time point when numbers of antigen-specific mIgG1-positive memory B cells were roughly the same in both genotypes (Fig. 3a). Again, the production of IgG1 in ITT-mutant mice was strongly diminished compared with wild-type animals. IgG1 titres during recall responses in mIgG1-YF animals were reduced to approximately one-third of what we observed in wild-type controls and thus did not even reach the level of primary response titres of wild-type mice (Fig. 4e). Analysis of high-affinity NP-specific IgG1 antibodies gave similar results (Fig. 4f). The ratio of high-affinity to total NP-specific IgG1 titres was moderately elevated in wild-type animals during primary responses, yet increased over time in a comparable manner in both genotypes (Fig. 4g). As control, we determined the production of high-affinity IgG2c antibodies, which was similar in both genotypes (Fig. 4h). These results show that in vivo ITT-induced signals are essential to generate proper IgG antibody responses.

The ITT improves the competitiveness of IgG-switched cells.
To compare the performance of wild-type and ITT-mutant B cells directly in the same animals, we bred mIgG1-YF BL/6 mice expressing ITT-mutant mIgG1 b with Balb/c mice expressing wild-type mIgG1 a . Owing to allelic exclusion, B cells of  Figure 1 | The ITT amplifies BCR-induced Ca 2 þ signalling. Splenic B cells from naive wild-type (wt) and ITT-mutant (YF) mice were isolated by magnetic cell sorting and stimulated with either low (2 mg ml À 1 , left panels) or high (10 mg ml À 1 , right panels) doses of anti-IgG F(ab') 2 fragments. Unstimulated cells served as controls (filled histograms). Subsequently, cells were stained with phospho-specific antibodies to either SLP65 (a) or Akt (b) wt: (black curves), YF: mIgG1-YF (grey curves). (c) Splenic B cells from three age-matched male mice of each genotype that had been immunized with sheep red blood cells for 11 days were pooled, loaded with Indo-1-AM and stained with FITC-labeled monomeric Fab fragments against IgG1. Ca 2 þ mobilization of mIgG1-positive cells was monitored before and after stimulation with anti-IgG F(ab') 2 fragments (20 mg ml À 1 , indicated by an upward arrow) in the presence of 1 mM extracellular CaCl 2 . Data are representative of three to four experiments.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9575 ARTICLE heterozygous F1 animals expressed either wild-type (allotype a) or YF-mutant (allotype b) mIgG1. Animals that had both allotypes in wild-type form served as controls. Allotype-specific analysis of IgG1 production in animals that were immunized with ovalbumin (OVA) in aluminium hydroxide (alum) showed that wild-type mIgG1 a -expressing B cells contributed up to six times more to the production of soluble IgG1 than their mIgG1 bexpressing ITT-mutant counterparts both during primary and secondary responses (Fig. 5a,c, grey graph). In control animals, both allotypes of IgG1 were produced in equal amounts (Fig. 5b) and, hence, were present in an almost 1:1 ratio during the entire course of the experiment (Fig. 5c, black graph). In conclusion, these results reveal that ITT-mediated signal amplification increases the competitiveness of mIgG-expressing cells, which facilitates their contribution to antibody production.
ITT signals promote plasma cell formation. To test whether reduced antibody production by ITT-mutant B cells was linked to impaired plasma cell generation, we determined the frequency of OVA-specific IgG1 antibody forming cells in heterozygous F1 animals at day 21 of the recall response by means of allotype-specific ELISPOT assays (Fig. 6a). In accordance with the diminished IgG1 b serum titres (Fig. 5) we observed an approximately fourfold reduction in the number of YF-mutant IgG1 b -producing antibody forming cells compared with IgG1 aproducing cells both in the bone marrow ( Fig. 6a,b,d) as well as in the spleens of heterozygous F1 animals (Fig. 6c,d). We conclude that ITT signalling supports the generation of plasma cells from the memory cell compartment.
ITTs promote T-cell-independent activation of memory B cells.
The reactivation requirements of memory B cells, particularly the need for T-cell help, are not fully understood. However, reactivation of memory B cells by antigen, especially viruses, can occur independently of T cells [24][25][26][27] , yet seems to require specific niches in secondary lymphoid tissue 28   cytomegalovirus (hCMV) (Fig. 7a) and were adjusted to contain equal numbers of glycoprotein B-specific mIgG1-positive cells ( Supplementary Fig. 8). Consistent with our previous observation (Fig. 3b) the frequency of glycoprotein B-specific mIgG1-expressing cells 20 weeks after the last booster immunization was lower in ITT-mutant mice than in wild-type animals ( Fig. 7b and Supplementary Fig. 8A). Therefore, we transferred roughly 1.5 times more B cells expressing other Ig isotypes from mIgG1-YF mice ( Supplementary Fig. 8B). Subsequently, recipient animals were challenged with virus-like particles that contained glycoprotein B in their envelope, which triggers memory B-cell reactivation but does not generate a primary response 26 . During the entire course of the experiment, glycoprotein B-specific IgG1 titres were consistently three times higher in animals that had received wild-type mIgG1-expressing B cells than in animals that had received ITT-mutant cells (Fig. 7c). As experimental control, we measured the simultaneous production of IgG2c antibodies against glycoprotein B, which was enhanced in recipients that had received B cells from mIgG1-YFmutant mice (Fig. 7d). Since we transferred slightly more cells from ITT-mutant mice to obtain equal numbers of mIgG1expressing cells, the enhanced IgG2c production might be due to a higher number of transferred IgG2c-positive cells and/or compensatory mechanisms as observed before ( Fig. 3 and 19 ). At any rate, the results show that ITT-mediated improved reactivation of mIgG-positive memory B cells occurs also in the absence of T-cell help and thus is a B-cell-intrinsic feature.

Discussion
Immunoglobulin class-switch recombination alters the molecular composition of BCRs. While the canonical Iga/b signalling unit remains unchanged, mIgG-(and mIgE-) BCRs are equipped with two copies of an additional tyrosine-based signalling motif referred to as ITT. In this study, we demonstrate that the ITT is a key determinant of humoral immunological memory in vivo. ARTICLE This provides a molecular explanation for the phenotype of a number of mouse models in which the entire cytoplasmic tails of mIgG1 or mIgE were either deleted or transferred to mIgM, which indicated their importance for the production of IgG and IgE antibodies [19][20][21] . Furthermore, B cells of 'mIgG1-only' mice exhibit enhanced BCR signalling and a prolonged lifespan compared with mIgM-expressing cells 23 . Here we show that the biological activity of cytoplasmic mIgG tails is mainly mediated by the ITT signalling motif. Antigen-induced phosphorylation of the ITT creates a docking site for the versatile adaptor protein Grb2, the recruitment of which amplifies BCR-proximal signalling events as well as BCRinduced cell proliferation 16,17 . Consistently, B-cell-specific inactivation of the Grb2 gene in the mouse impairs reactivation of IgG-switched memory B cells, corroborating the importance of the ITT-Grb2 interaction for efficient antibody recall responses 17,29 . The most salient signalling effect of ITTmediated Grb2-recruitment into the BCR signalosome is the enhanced activation of phospholipase C-g2 (PLC-g2), concomitant with a greatly prolonged influx of Ca 2 þ across the plasma membrane. In line with this, homoeostasis of B-cell memory relies on the expression of PLC-g2 since its cell-type-specific ablation in mIgG1-expressing B cells causes reduced formation and survival of IgG1-switched memory B cells 30 . Furthermore, in B cells the phosphatase calcineurin, which controls the activation of transcription factor NF-AT, is specifically required for terminal differentiation into plasma cells 31 . Considering that the activity of calcineurin is stimulated by Ca 2 þ /calmodulin it appears possible that ITT-mediated prolongation of mIgG-BCR-induced Ca 2 þ mobilization augments the activity of calcineurin thereby supporting the differentiation of IgG-switched B cells into plasma cells.
Plasma cell differentiation is generally considered to be governed by two antagonizing groups of transcriptional regulators that either maintain the mature B-cell phenotype, such as Pax5 and Bcl-6, or induce the plasma cell differentiation programme like Irf4 and Blimp-1 (ref. 32). Expression of either set represses the other one and elimination of Bcl-6 and Pax5 expression seem prerequisite for plasma cell differentiation to occur. Signals from the BCR might tip the balance between these two sets of transcription factors in favour of the plasma cell differentiation programme in several ways. First, BCR-induced proteasomal degradation of Bcl-6 has been reported to occur in a MAP kinase-dependent manner 33 . Second, in a reciprocal way expression of Irf4 is induced on BCR stimulation 34,35 . Third, the transcription factor Stat3, which acts in concert with Irf4 to induce expression of Blimp1 (ref. 36), is activated on BCR stimulation 37,38 . Thus, ITT-mediated enhanced signalling of mIgG-BCRs may facilitate degradation of Bcl-6 and/or influence the activity of other components that govern plasma cell differentiation such as Irf4 and Stat3. Consistent with such a scenario, B-cell-specific deletion of stat3 results in a selective deficiency of IgG-producing plasma cells despite normal formation of germinal centres and memory B cells 39 . Besides improved BCR signalling, differential gene expression between memory and naive B cells has been reported and suggested to be involved in improved reactivation of memory B cells [40][41][42][43] . Furthermore, it has been proposed recently that the ability of both mIgM-and mIgG-expressing memory B cells to generate antibody-secreting cells on antigen challenge is primarily determined by their maturation stage that is reflected by expression of the cell surface receptors PD-L2 and CD80 (ref. 43). However, this conclusion was based on cell transfer experiments that did not reflect a physiological environment in which each (memory) B cell has to compete for antigen with antibodies as well as with antigen-specific (memory) B cells of other Ig isotypes that stem from the primary response 44,45 . Our data and that of other groups clearly show that the responsiveness of memory B cells is under control of the mIg isotype integrated into the BCR 19,21,45,46 . Consistently, previous in vivo studies showed that a cytoplasmic mIgG tail improves antigen-dependent as well as antigenindependent survival of B cells in the mouse 21,23 . In line with these earlier observations, inactivation of the ITT compromised the maintenance of IgG1-switched cells in our experiments. Even though we cannot say whether this was due to diminished homoeostatic proliferation, impaired survival or altered migration of ITT-mutant cells, data from Waismann et al. suggest that an mIgG cytoplasmic domain improves the survival of B cells in an antigen-independent manner 23 . Hence, it appears likely that the ITT motif not only improves antigen-induced signalling but probably also antigen-independent tonic signalling of mIgG-BCRs.
In conclusion, our data show that the persistence and reactivation of IgG-switched B cells are under the control of the BCR and its cytoplasmic signalling machinery. The enhanced activation of this signalling machinery by the ITT may provide a clue as to why mIgG-expressing memory B cells are preferentially reactivated over mIgM-expressing cells in secondary immune responses 45,46 . Germinal centre-associated events, however, like generation of mIgG-expressing cells and antibody affinity maturation occur independently of ITT signals. In agreement with these findings, recent studies demonstrated that BCRs of germinal centre B cells are silenced by high phosphatase activity 47 and that it is the effectiveness of antigen presentation to T helper cells that determines the fate of a B cell in a germinal centre reaction [48][49][50] . These observations may explain the dispensability of the ITT in germinal centre processes. Nevertheless, the phenotype of ITT-mutant mice reveals that the BCR on IgG isotype-switched cells is endowed with the intrinsic ability to promote humoral immunological memory even in the absence of cognate T-cell help. Mice. mIgG1-YF knock-in mice were established using DNA from a C57BL/6 clone (RP23-331J13), in which the 19th codon of the transmembrane exon M2 of the g1 immunoglobulin locus was mutated from TAC to TTC causing an amino acid exchange from Y to F. In addition, a neomycin resistance cassette with flanking loxP sites was inserted into the EcoRI restriction site between the two transmembrane exons M1 and M2 of the g1 immunoglobulin locus. The targeting vector was electroporated into JM8.F6 ES cells (kindly provided by William C. Skarnes, Wellcome Trust Sanger Institute, Cambridge, UK), and recombinant ES cells were injected into eight-cell albino outbred embryos to generate chimeric offspring. After germline transmission, transgenic offspring were bred to EIIa-cre mice 51 to delete the resistance cassette. Mice were held under specific pathogen-free conditions in individually ventilated cages and kept on a pure C57BL/6 background. Wild-type C57BL/6 mice used in Figs 2,4 and 7 were purchased from Charles River (Sulzfeld, Germany). All other mice were bred in house. All animal experiments were approved by the state office for consumer protection and food safety of Lower Saxony and performed according to institutional and national guidelines.
Adoptive transfer of memory B cells. Transfer was performed as previously described 28 . Briefly, mice were immunized twice intraperitoneally with 5 mg glycoprotein B derived from hCMV (a gift from Sanofi Pasteur, Lyon, France) in Imject Alum at intervals of 5 weeks and with 5 mg glycoprotein B in PBS intravenously 5 weeks later and then rested for another 20 weeks. Subsequently, single-cell suspensions of splenocytes were depleted of T cells by complement lysis, followed by Ficoll gradient purification and enrichment of CD19 þ B cells by magnetic cell sorting (Miltenyi Biotec). For each donor the frequency of glycoprotein B-specific IgG1 þ memory B cells was determined by FACS, and total CD19 þ B lymphocytes adjusted to contain 200 glycoprotein B-specific IgG1 þ memory B cells were transferred into one or two Rag1-deficient recipient mice. One week after transfer, the recipients were boosted with 0.5 mg glycoprotein B-containing hCMV-dense bodies (virus-like particles) as described by Weisel et al. 28 and bled regularly.
ELISPOT. MultiScreen HTS filter plates (Millipore) were coated with either OVA or NP1-BSA, or NP14-BSA according to manufacturer's instructions. After incubation with 1-10 Â 10 5 cells for 4 h, captured antibodies were detected with the same detection antibodies and same dilutions as described for ELISAs and visualized with the AEC Chromogen Kit (Sigma-Aldrich). Wells were photographed with an ImmunoSpot analyzer (Cellular Technology Limited) and spots were either counted manually or using Bioreader software (BIO-SYS).
Ca 2 þ measurements. Splenocytes from three age-matched male mice immunized with sheep red blood cells for 11 days were pooled. B cells were enriched by magnetic CD43-depletion (Miltenyi Biotech) and stained with rat anti-mouse IgG1-FITC Fab fragments (clone A85-1, BD catalogue no. 553443). Fab fragments were generated by papain cleavage using the Pierce Fab Micro Preparation Kit (Thermo Scientific) according to the manufacturer's instructions. Subsequently, cells were loaded with Indo-1-AM as previously described 16 . mIgG-BCRs were stimulated with 20 mg ml À 1 goat anti-mouse IgG F(ab') 2 fragments (Jackson Immuno Research). For analysis of mIgG1-expressing cells, FITC-positive cells were gated.