Developmental Acquisition of the Lyn-CD22-SHP-1 Inhibitory Pathway Promotes B Cell Tolerance¹

Bcell receptor hyperresponsiveness is associated with a breakdown in tolerance and the development of autoantibodies in several genetically modified mouse strains (1, 2). A good example of this is the Lyn-deficient mouse (3). Lyn is a Src family kinase (SFK)³ that, like two other SFKs expressed by B cells, Blk and Fyn, phosphorylates ITAMs on BCR Igα/Igβ chains following Ag binding (4, 5, 6). Lyn also functions to phosphorylate ITIMs on inhibitory receptors that negatively regulate BCR signaling, including CD22 (5, 7, 8, 9), FcγRIIb (7, 8), and perhaps others (10, 11, 12, 13). In this way, Lyn facilitates recruitment of the Src homology region 2 domain-containing phosphatase-1 (SHP-1) and SHIP phosphatases to the plasma membrane, which down-modulate BCR signaling (5, 8, 9). In the absence of Lyn, BCR signaling is supported by Blk and Fyn, but inhibitory receptors are ineffective at down-regulating BCR signaling, thereby leading to BCR hyperresponsiveness (3).

Lyn-deficient mice exhibit increased plasma cell numbers and serum Ig levels. Surprisingly, these mice also produce autoantibodies to nuclear Ags (4, 14, 15). Why elevated BCR signaling would lead to a loss of tolerance in lyn^−/− mice is not entirely self-evident, because whereas BCR signaling drives activation of mature B cells, it also promotes tolerance-inducing mechanisms in immature B cells. For example, binding of Ag by immature B cells in the bone marrow (BM) induces expression of RAG 1 and 2, which can generate L chain rearrangement and thereby change the specificity of the BCR (receptor editing) (16, 17, 18). Continued recognition of Ag by the BCR may induce apoptosis of self-reactive B cells (clonal deletion) (16). Therefore, genetic alterations that enhance BCR signaling should increase the sensitivity of immature B cells to self Ags, causing more thorough removal of autoreactive BCR specificities from the newly formed B cell repertoire. For this reason, it seems paradoxical that autoimmunity develops in the lyn^−/− mouse and in other mice with genetic alterations increasing the strength of BCR signaling.

To account for the break in B cell tolerance observed in Lyn-deficient mice, we hypothesized that the level of Lyn-mediated inhibitory signaling might change during the course of B cell development. Interestingly, we found that B cells become less sensitive to BCR engagement as they proceed through development. This reduction in BCR sensitivity was mediated by Lyn and CD22, because Lyn or CD22 deficiency strongly increased BCR signaling in follicular B cells, but only weakly increased signaling in transitional stage 1 (T1) B cells. Consistent with these findings, Lyn deficiency had a modest effect on events associated with receptor editing, whereas it more strongly enhanced the sensitivity of mature B cells to BCR-induced proliferation. These results indicate that inhibition of BCR signaling by the Lyn-CD22-SHP-1 pathway is up-regulated as B cells mature in the spleen, and suggest that this developmental change is involved in the maintenance of peripheral immune tolerance of B cells.

Mice aged 7–12 wk were used for most experiments. Four-week-old mice were used in experiments using motheaten-viable mice (Ptpn6^me-v) backcrossed onto C57BL/6 background (The Jackson Laboratory). The lyn^−/− mice were used as described (4), and backcrossed at least 15 generations onto C57BL/6 background. The cd22^−/− mice (19) backcrossed at least 16 generations onto C57BL/6 background were obtained from E. Clark (University of Washington, Seattle, WA). MD4 transgenic Ig_Hel mice were obtained from J. Cyster (University of California, San Francisco, CA). All animals were housed in a specific pathogen-free facility at University of California, according to University and National Institutes of Health guidelines. Animal use was approved by the University of California Institutional Animal Care and Use Committee.

Fluorophore-conjugated Abs directed against the following molecules were used: B220 (RA3-6B2), CD23 (B3B4), IgM (II/41), Igλ_1–3 (R26-46), CD16/CD32 (2.4G2) CD22.2 (Cy34.1), and CD72 (K10.6) all from BD Pharmingen; CD24 (M1/69) from BioLegend; CD93 (AA4.1) and CD5 (53-7.3) from eBioscience; Igκ (187.1) and Igλ (JC5-1) from Southern Biotechnology Associates; and IgM (goat polyclonal Fab monomer, μ-chain specific) (Jackson ImmunoResearch Laboratories). Cells were analyzed on an LSR-II or FACSCalibur (both from BD Pharmingen). Lymph node B cells were purified by negative selection using CD43 microbeads (Miltenyi Biotec), according to the manufacturer’s protocol, and passage through an autoMACS Separator (Miltenyi Biotec). For cell sorting, splenocytes or BM cells were stained for CD23, CD93, and IgM Fab monomer for 30 min and either B220 or a non-B cell mixture (CD4, CD8, CD11b, Gr1, NK1.1, and Ter119) in HBSS, 1% FCS, and 0.5% BSA. Cells were then sorted on a MoFlo sorter (DakoCytomation). Dead cells were excluded by propidium iodide (BioChemika) uptake. All FACS data were analyzed with FlowJo version 6.4.1 (Tree Star software).

For measurements of intracellular-free calcium levels, splenocytes or BM cells were loaded with Indo-1 AM (Molecular Probes/Invitrogen) and then labeled for CD93 (allophycocyanin), CD23 (PE), IgM (Fab monomer) (FITC), and B220 (PE-Cy7), or a mixture of Abs recognizing non-B cells (CD4, CD8, CD11b, Gr1, NK1.1, Ter¹¹⁹) (PE-Cy7). Cells were resuspended in RPMI 1640 supplemented with 1% BSA and 20 mM HEPES and warmed to 37°C for 3 min, and analysis was initiated with flow cytometry. After the baseline was established for 30–45 s, cells were stimulated with 0.5–50 μg/ml goat anti-mouse IgM F(ab′)₂ (Jackson ImmunoResearch Laboratories), 5 ng to 100 μg of 0.2-μm filtered hen egg lysozyme (HEL; Sigma-Aldrich), or 16 μg/ml ionomycin (Sigma-Aldrich). Median intracellular calcium concentration as indicated by the ratio of fluorescence 405 nm emission to 530 nm emission was measured over time by flow cytometry. Dead cells were excluded by propidium iodide uptake.

For intracellular measurement of phospho-ERK, 3 × 10⁶ splenocytes or 2 × 10⁶ BM cells were warmed to 37°C for 30 min in RPMI 1640 medium with 1% BSA and 20 mM HEPES. During the final 10 min of warming, the cells were labeled with anti-IgM Fab FITC. Some samples were treated with 10 μM MEK inhibitor U0126 (Sigma-Aldrich) during the warming period. Cells were then stimulated with 0–50 μg/ml goat anti-mouse IgM Fab′)₂ (Jackson ImmunoResearch Laboratories) or 1 μg/ml PMA (Sigma-Aldrich). After the desired period of stimulation, cells were fixed with 4% paraformaldehyde (Electron Microscopy Sciences) and permeabilized with ice-cold 100% methanol (Electron Microscopy Sciences). Cells were labeled with phospho-p44/42 (Thr²⁰²/Tyr²⁰⁴) rabbit mAb (197G2; Cell Signaling Technology) that recognizes phosphorylation sites required for ERK activity. Cells were then labeled with donkey anti-rabbit IgG allophycocyanin (Jackson ImmunoResearch Laboratories) as well as Abs to B220 (PE-Cy7), CD23 (PE), and CD24 (Pacific blue). (The fixation process prevented labeling for CD93.) Cells were analyzed by flow cytometry on an LSR-II.

Sorted splenic B cells from four mice of each genotype were suspended at a concentration of 2 × 10⁶ cells/ml in RPMI 1640 with 10% FCS, 20 mM HEPES, 2 mM glutamine, and 1 mM sodium pyruvate, and stimulated with 0–15 μg/ml goat anti-mouse IgM F(ab′)₂ (Jackson ImmunoResearch Laboratories) in 96-well tissue culture plates (Corning Glass) at 37°C, 5% CO₂. For proliferation assay, cells were incubated for 48 h. A total of 1 μCi/well [³H]thymidine (Amersham) was added during the final 4 h of culture, and incorporation was measured by scintillation counting.

Whole-cell lysates were prepared in SDS lysis buffer from sorted splenic B cells or lymph node B cells purified by autoMACS negative selection. Proteins were separated on a NuPAGE bis-Tris gel (Invitrogen) with MOPS buffer and immunoblotted on Immobilon-FL transfer membrane (Millipore). Abs were directed against phospho-Src family kinase Tyr⁴¹⁶ (Cell Signaling Technology; 2101s), phospho-Lyn Tyr⁵⁰⁷ (Abcam ab53122), Lyn (Abcam ab1890), and BAP31 (Abcam ab37120) as a loading control. These Abs were detected with fluorescently labeled secondary Abs using the Odyssey Infrared Imaging System (LI-COR Biosciences).

Total RNA was isolated from sorted B cells using the RNeasy Micro kit (Qiagen), and cDNA was transcribed using the iScript cDNA synthesis kit (Bio-Rad), according to the manufacturers’ instructions. Equivalent amounts of cDNA were used in quantitative PCR on an ABI Prism 7700 sequence detection instrument (Taqman; PerkinElmer Applied Biosystems). Primer and probe sequences for RAG1 and RAG2 (20), and GAPDH (21) were used as published.

Genomic DNA was isolated from sorted Igλ⁻ B cell populations using the Gentra PureGene tissue kit (Qiagen). Quantitative PCR was performed, as described (22). The amount of Vκ-RS product in each sample was normalized to the amount of β-actin product and compared with the normalized target value in wild-type (WT) C57BL/6 follicular B220⁺ IgM⁺ κ⁺ splenocytes to determine a relative quantity (comparative cycle threshold method).

Statistical analyses, including unpaired two-tailed t test and four-parameter logistic equation (linear regression analysis), were performed with Prism v.4.0 (GraphPad software).

It has previously been suggested that immature B cells are more sensitive than mature B cells to limiting amounts of BCR stimulation (23, 24), but other studies have not agreed with this conclusion (25, 26). These earlier studies have used various manipulations to generate sufficient numbers of immature B cells to analyze signaling. To better evaluate this issue, we analyzed calcium elevation in unmanipulated B cells taken directly ex vivo from young adult mice. B cells at various developmental stages were identified by a combination of mAbs, as described by Allman et al. (27) (Fig. S1 in the supplemental data).⁴ We found this Ab combination, which included nonstimulating anti-IgM Fab FITC, had minimal effects on BCR-induced calcium signaling (Fig. S2).⁴ When WT splenic B cells were stimulated with subsaturating concentrations of anti-IgM F(ab′)₂, T1 and T2 B cells exhibited greater increases in cytoplasmic calcium levels (intracellular Ca²⁺ concentration ([Ca²⁺]_i)) than did transitional stage 3 (T3) and mature follicular B cells (Fig. 1, A–C). All of these B cell subsets had similar maximal calcium responses when stimulated with a saturating concentration of anti-IgM F(ab′)₂ (50 μg/ml). Although it has been proposed that T3 B cells represent an anergic population of B cells (28), we observed that T3 B cells responded to anti-IgM by increasing [Ca²⁺]_i similarly to follicular B cells. These data indicate that mature follicular B cells require more BCR stimulation than immature T1 and T2 B cells to induce comparable BCR-induced calcium signaling.

As B cells mature, IgM levels are down-regulated and IgD levels are up-regulated (29). We therefore wondered whether treatment of B cells with anti-IgM accurately measured the sensitivity of B cells to BCR engagement at each stage of development. To address this issue, we therefore gated on the subpopulation of follicular B cells with levels of membrane-bound Ig (mIg)M comparable to the level of mIgM on T1 B cells. These cells still exhibited decreased sensitivity to anti-IgM for the elevation of intracellular-free calcium compared with the T1 B cells (data not shown). To investigate whether reduction in mature B cell sensitivity to Ag is averted during development through up-regulation of surface IgD levels, we measured the sensitivity of B cells to Ag engagement using Ig transgenic B cells specific for hen-egg lysozyme (HEL). Similar to the results obtained when an anti-IgM reagent was used, immature transitional B cells exhibited greater increases in cytoplasmic calcium levels than did follicular B cells when treated with subsaturating concentrations of HEL, whereas their responses to higher doses of HEL were similar (Fig. 2). These data are consistent with those reported in another BCR transgenic system (23) and confirm that immature T1 B cells are in fact more sensitive to BCR stimulation than mature follicular B cells. These data also show that stimulation of B cells with anti-IgM accurately reflects the relative responses of different B cell populations to BCR engagement by Ag.

Because the first events associated with the establishment of B cell tolerance occur in the BM, it was important to determine whether B cells at various developmental stages in the BM respond to BCR stimulation in a manner similar to their splenic counterparts. We applied the same gating strategy for identifying B cell populations in the spleen to BM cells (30) (Fig. S1).⁴ Calcium responses to anti-IgM stimulation of BM B cell subpopulations were remarkably similar to those of corresponding splenic B cell populations (Fig. 1 D, and data not shown for 20 μg/ml anti-IgM). Splenic T1 cells responded nearly identically to immature newly formed (NF) cells, splenic T2 cells responded identically to BM-T2-like cells (data not shown), and splenic follicular B cells responded identically to mature naive B cells recirculating in the BM. As expected, the pro-B and pre-B cell populations had minimal or no response to anti-IgM (data not shown). These data indicate that WT B cells become less sensitive to low levels of BCR stimulation as they transition from immaturity to maturity, regardless of whether they reside in the BM or spleen.

To determine whether the changes in calcium signaling during B cell development were representative of changes in other BCR signaling reactions, we examined BCR-induced ERK activation as assessed by flow cytometry. This assay was largely specific because the response was nearly completely blocked by the MEK inhibitor U0126 (Fig. S3).⁴ Stimulation of WT spleen cells with anti-IgM led to significant increases in the levels of ERK phosphorylation in the various B cell subpopulations. Peak levels of phospho-ERK were detected in B cells after 2–3 min of stimulation (Fig. S3).⁴ As with calcium elevation, ERK activation after 2.5 min was induced by lower doses of anti-IgM in T1 B cells than in mature follicular B cells (Fig. 3, A–C). Interestingly, even at high doses of anti-IgM (50 μg/ml), follicular B cells activated ERK to a lesser degree than did immature T1 B cells. In contrast, stimulation with the diacylglycerol mimetic, PMA, induced higher levels of ERK phosphorylation in follicular B cells than in T1 B cells (Fig. 3,B). Phospho-ERK levels in stimulated BM B cell populations were similar to those in the corresponding stimulated splenic B cell populations (Fig. 3 C). Thus, data for both elevation of intracellular-free calcium and activation of ERK MAPKs showed that WT immature B cells are more sensitive to BCR stimulation than are WT follicular B cells.

Although Lyn-deficient B cells are known to have elevated signaling responses to BCR engagement (5, 7), it was not known whether this hypersensitivity is dependent on the maturation stage of the B cell. Spleens from lyn^−/− mice contained T1, T2, T3, and follicular B cell populations (Fig. S1),⁴ as previously reported (27). We found that the calcium responses of lyn^−/− follicular B cells were far more sensitive to IgM stimulation than were WT follicular B cells (Fig. 4, A–C), and their ability to respond to low doses of anti-IgM resembled WT immature B cells. Additionally, the peak amplitude of the calcium response by lyn^−/− follicular B cells was greatly exaggerated over that of WT follicular B cells even at saturating doses of anti-IgM (Fig. 4, A–C). These data indicate that Lyn serves an important role in limiting the magnitude of BCR signaling as well as dampening the sensitivity of mature B cells to IgM stimulation.

In contrast to the large effect of Lyn deficiency on mature follicular B cells, Lyn deficiency had a relatively minor effect on calcium responses by immature T1 B cells (Fig. 4, A–C). The peak amplitude as well as the sustained level of [Ca²⁺]_i following BCR stimulation was only slightly higher in lyn^−/− T1 cells compared with WT T1 B cells. Similarly, lyn^−/− newly formed BM B cells had only slightly higher elevations of [Ca²⁺]_i than their WT counterparts (Fig. 4 D). These data indicate that Lyn greatly inhibits BCR calcium signaling by follicular B cells, but surprisingly provides much less restraint on BCR calcium signaling by newly formed or T1 B cells.

To confirm the effects of Lyn deficiency on BCR signaling, we examined ERK phosphorylation in lyn^−/− B cells. Anti-IgM-stimulated lyn^−/− follicular B cells had much higher levels of phospho-ERK than stimulated WT follicular B cells (Fig. 3, D and E). In contrast, lyn^−/− T1 B cells had either slightly lower or similar levels of ERK phosphorylation following BCR stimulation compared with WT T1 B cells (Fig. 3, D and E). Thus, BCR-induced ERK activation and calcium elevation were both strongly enhanced in follicular B cells lacking Lyn, whereas these signaling events were not greatly affected in T1 B cells.

These data collectively indicate that Lyn has a stronger effect on modulating BCR signaling in mature B cells than in immature B cells. Because Lyn can negatively regulate BCR signaling through the recruitment of ITIM-bearing inhibitory proteins, we wondered whether the strong effect of Lyn on mature B cell signaling was due to increased expression of ITIM-bearing inhibitory receptors during B cell development. Consistent with this idea, we found that CD22 expression on the surface of B cells was increased 2.5-fold from the T1 stage to the follicular stage (Fig. 5,A; Table I) in agreement with earlier studies (31, 32). However, expression of CD16/32, which on B cells is primarily FcγRIIb, was largely unchanged, and two other negative regulators of BCR signaling, CD72 and CD5, were down-regulated during B cell maturation in the spleen (levels reduced by 35 and 37%, respectively) (Fig. 5,A; Table I).

Because CD22 expression increases substantially during B cell development, we hypothesized that CD22 deficiency would have disproportionate effects on mature B cells as compared with immature B cells. Indeed, cd22^−/− follicular B cells were much more sensitive to low-dose BCR stimulation and had much higher [Ca²⁺]_i than WT follicular B cells (Fig. 5,B). In contrast, CD22 deficiency had relatively minor effects on T1 B cells. Surprisingly, the calcium responses of cd22^−/− B cells were very similar in magnitude to the responses of lyn^−/− B cells in both mature and immature B cells (Fig. 5,B). Additionally, calcium responses of BM NF B cells from motheaten-viable mice, which have a partial loss of function mutation in the ptpn6 gene that encodes the tyrosine phosphatase SHP-1 (shp-1^me-v/me-v) (33), were slightly reduced compared with WT NF B cells (Fig. 5 C). These data indicate that the Lyn-CD22-SHP-1 inhibitory pathway has minor effects on limiting BCR signaling in immature B cells, but as B cells mature Lyn functions in conjunction with CD22 to mediate reduction in BCR sensitivity.

Because CD22 expression increases during B cell development, we investigated whether reduction of CD22 levels in mature follicular B cells could cause these cells to signal in a manner more similar to immature B cells. To test this, we studied follicular B cells from cd22^+/− mice, whose CD22 surface levels are similar to those of WT T1 B cells (Fig. 5,D). We found that calcium release by cd22^+/− follicular B cells was enhanced compared with WT, although it was not as robust as the calcium release of WT T1 B cells (Fig. 5 E). These data indicate that up-regulation of CD22 during B cell development has an appreciable role in reducing BCR sensitivity, but that additional factors are also likely to contribute.

We next investigated whether Lyn levels or function might change during B cell development to contribute to the acquisition by follicular B cells of the Lyn-CD22-SHP-1 inhibitory pathway. We sorted T1, T2, T3, and follicular B cells from spleens of WT mice and measured Lyn protein levels and levels of tyrosine phosphorylation at its activating and inhibitory regulatory sites by immunoblotting of the cell lysates. Lyn protein levels did not change substantially throughout B cell maturation in the spleen (Fig. 6,A, and data not shown), which is consistent with results from published mRNA microarray analysis of B cell developmental stages in the BM (34). Like other src family kinases, Lyn kinase activity is regulated by phosphorylation of two tyrosine sites (3). Lyn assumes an inactive conformational state when the C-terminal Y507 site is phosphorylated, whereas kinase activity is enhanced following phosphorylation of Y397 in the kinase domain. We found that Y507 phosphorylation was similar in splenic B cells at each stage of development and did not exhibit a substantial change upon BCR stimulation (Fig. 6,A). Interestingly, Y397 phosphorylation was elevated in unstimulated follicular B cells compared with immature transitional B cells. However, upon BCR stimulation, Y397 phosphorylation increased to similar levels in each of the B cell subsets analyzed (Fig. 6 A). These findings suggest that Lyn activity is enhanced in unstimulated follicular B cells, compared with immature B cells. Although these data do not directly assess Lyn function inside the B cell, they suggest that Lyn is in a more activated state in resting follicular B cells, potentially explaining in part the enhanced activity of the Lyn-CD22-SHP-1 inhibitory pathway in these cells compared with immature T1 B cells.

Although CD22 is generally considered to be downstream of Lyn, in that Lyn appears to be responsible for phosphorylating the ITIM of CD22 leading to SHP-1 recruitment to the plasma membrane, we observed that CD22 expression also affected the activity of Lyn. Lymph node B cells from CD22^−/− mice had a ∼3- to 4-fold reduction in Lyn Y397 phosphorylation levels, particularly in unstimulated B cells (Fig. 6,B). The levels of phosphorylation of Y507 were either slightly increased or unchanged (Fig. 6 B). These results suggest that whereas CD22 acts downstream of Lyn to recruit phosphatases to the plasma membrane and inhibit BCR signaling, it also acts upstream of Lyn to positively regulate Lyn activity.

Genetic deficiency in Lyn, CD22, or SHP-1 results in spontaneous autoantibody production (4, 14, 15, 35, 36), but it is not known whether this failure of tolerance occurs primarily during B cell development in the BM or later in the periphery. Because lyn^−/− T1 B cells in the spleen and BM had signaling responses to BCR stimulation that were only slightly greater than those of WT cells, we hypothesized that Lyn deficiency would either have no effect on events related to the establishment of B cell tolerance in the BM or would slightly exaggerate these events. It is thought that BCR signaling above a certain threshold in pre-B and immature B cells causes the induction of RAG-mediated recombination events at Ig L chain loci, leading to the expression of a new κ or λ L chain (receptor editing), which may result in loss of self-reactivity and thereby allow subsequent maturation (16). Therefore, to determine whether receptor editing is affected by Lyn deficiency, we measured the level of rag1 and rag2 mRNA in BM B cells. We found that there was little difference in the levels of rag1 and rag2 mRNA between lyn^−/− and WT small pre-B cells, but lyn^−/− NF B cells expressed several-fold higher levels of RAG1 and RAG2 than did their WT counterparts (Fig. 7,A). Because RAG induction causes recombination events that can lead to a switch from Igκ L chain expression to expression of Igλ, we also investigated whether Lyn deficiency affected the frequency of Igλ-expressing B cells. The lyn^−/− mice had slightly higher frequencies of Igλ⁺ newly formed cells than WT mice, although the frequencies of Igλ⁺ T1 B cells in the spleen were similar in WT and lyn^−/− mice (Fig. 7 B).

Finally, we also measured the frequency of B cells that had RS recombinations at the Igκ locus as an indication of receptor editing. Recombination at the RS sequence inactivates a functional Igκ locus, after which another L chain locus can be rearranged to generate a new L chain (16, 37). Purified Igλ⁻ small pre-B cells, newly formed, splenic T1, and follicular B cells were isolated by cell sorting from lyn^−/− and WT mice. The frequencies of RS recombination in these populations were measured by quantitative PCR of genomic DNA (22). This assay therefore measures the overall degree of L chain rearrangement in a population of B cells. The RS frequencies were similar in lyn^−/− newly formed and T1 B cells compared with WT (Fig. 7 C). Collectively, these data indicate that the mechanisms causing the removal of autoreactive BCRs in immature B cells are intact or slightly enhanced in lyn^−/− mice. This suggests that autoantibody production in lyn^−/− mice is not a consequence of impaired central tolerance.

We next asked whether Lyn-deficient mature follicular B cells are more easily activated through their BCR than their WT counterparts, as suggested by the signaling data. It has been reported that lyn^−/− B cells are hypersensitive to BCR-induced proliferation, although in those studies B cell populations were not fractionated (4, 14, 15, 38). To differentiate between B cells at different stages of development, we sorted splenic B cell populations and measured their ability to proliferate in response to different levels of BCR engagement. Not surprisingly, BCR stimulation did not induce WT or lyn^−/− T1 B cells to proliferate (Fig. 7,D), in agreement with previous reports (27, 39, 40). In contrast, lyn^−/− follicular B cells proliferated much more vigorously than did WT follicular B cells when stimulated with low doses of anti-IgM (2.5 μg/ml) (Fig. 7 E). However, lyn^−/− and WT follicular B cells incorporated similar amounts of [³H]thymidine when stimulated with a higher dose of anti-IgM (15 μg/ml). These data show that lyn^−/− follicular B cells are indeed more sensitive to anti-IgM-induced proliferation than WT follicular B cells, consistent with their increased sensitivity to anti-IgM-induced calcium release and ERK activation. These findings demonstrate that lyn^−/− mature B cells have a reduced threshold of activation, which therefore could contribute to the breakdown in tolerance to self Ags in the periphery of these mice.

Central to the ability of B cells to discriminate between foreign and self Ags is the difference in the responses to Ag of immature vs mature B cells. Engagement of the BCR of immature B cells in the BM induces expression of the RAG1 and RAG2 genes and subsequent Ig L chain rearrangements with the goal of changing specificity away from self-reactivity, a process called receptor editing (16). If this loss of self-reactivity is not achieved or if self Ag triggers BCR signaling of T1 B cells in the spleen, then the cell undergoes apoptosis, resulting in clonal deletion. In contrast, strong BCR engagement of mature B cells induces proliferation, and in combination with cytokines or other signals can facilitate a productive immune response. A variety of studies has indicated that immature B cells are more sensitive to Ag than are mature B cells. We have used newly improved methods for measuring signaling reactions in individual cells from immune organs to compare the sensitivity of the various developmental stages of B cells in the spleen and BM for two key BCR-induced signaling responses, the elevation of intracellular-free calcium, and the activation of the ERK MAPK. These results confirm that T1 B cells are highly sensitive to BCR engagement and that follicular B cells are distinctly less sensitive. Moreover, we have shown that this change in signaling sensitivity of the BCR is due to a maturation-associated increase in the ability of the Lyn-CD22-SHP-1 pathway to attenuate BCR signaling. We propose that the low level of function of this inhibitory mechanism in immature B cells serves to permit efficient engagement of tolerance-promoting programs in these cells, whereas its greater activity in mature follicular B cells serves to create a threshold for activation that minimizes the activation of B cells in response to self Ags encountered in the periphery.

Previous studies have also compared the BCR responsiveness of immature and mature B cells, but have disagreed as to whether immature B cells are more sensitive (23, 24, 41), equally sensitive (25), or less sensitive (26, 28) to BCR stimulation relative to mature B cells. Our studies simultaneously analyzed immature and mature B cell populations taken immediately ex vivo from the BM or spleen of unmanipulated mice and used staining conditions that did not perturb signaling, and thus, we feel they are likely to represent the responses of these cell types in a reliable way. The difference in sensitivity of BCR signaling that we observed did not appear to result primarily from changes in the expression of mIgM and mIgD that occur as B cells mature in the spleen. T1 cells have high levels of mIgM and low levels of mIgD, whereas mature B cells have high levels of mIgD and diverse levels of mIgM, ranging from low levels to the high levels seen in T1 B cells. We found that there was still a considerable difference in sensitivity to anti-IgM treatment between T1 B cells and a gated subpopulation of follicular B cells that had levels of mIgM comparable to the expression level on T1 cells (data not shown). Moreover, when we used the Ag lysozyme to engage the BCR on MD4 antilysozyme Ig transgenic T1 and follicular B cells, we again found substantially greater sensitivity of the former compared with the latter. Similarly, Wen et al. (23) demonstrated that immature anti-Thy1 Ig transgenic B cells were more sensitive to Thy1 Ag than their mature B cell counterparts using techniques similar to ours. These findings indicate that mechanisms other than changes in surface Ig expression regulate B cell sensitivity to BCR engagement during maturation.

Genetic evidence indicates that this reduction in BCR sensitivity during B cell maturation is due to an increase in the efficacy of the inhibitory pathway mediated by Lyn, CD22, and SHP-1. BCR signaling of T1 and T2 B cells was only modestly affected by Lyn deficiency, CD22 deficiency, or SHP-1 mutation, indicating that this inhibitory pathway is relatively inactive in immature B cells of the BM or spleen. In contrast, there was a dramatic enhancement in BCR signaling of mature B cells by Lyn and CD22 deficiencies, demonstrating the importance of this inhibitory pathway in limiting BCR sensitivity once B cells mature. One contributor to this developmental change in the ability of the Lyn-CD22-SHP-1 pathway to inhibit BCR signaling is likely to be the 2.5-fold up-regulation of CD22 during B cell maturation in the spleen. Follicular B cells from cd22^+/− mice, which express CD22 levels similar to those of WT T1 B cells, exhibited enhanced calcium responses compared with WT follicular B cells, but were distinctly less sensitive than WT T1 B cells. This indicates that other factors are present that mediate the developmental reduction in BCR sensitivity, besides up-regulation of CD22. Regulation of SHP-1, CD22, or Lyn activity could contribute to the developmental change in inhibitory signaling. For example, sialic acid acetylesterase has recently been shown to facilitate CD22 suppression of BCR signaling (42) and is up-regulated during B cell development (43). In contrast, Lyn levels did not change significantly between T1 and mature follicular B cells. However, we did see a higher level of Lyn activity in unstimulated mature B cells compared with unstimulated T1 B cells, as evidenced by phosphorylation of the active site Y397 residue. Although regulation of the Lyn Y397 site is not well understood, phosphorylation of Y397 might be enhanced in a lipid raft environment (44), and mature B cells appear to have a higher lipid raft content than transitional B cells (45). Interestingly, Lyn phosphorylation of Y397 was substantially decreased in unstimulated cd22^−/− mature B cells, which suggests that increasing CD22 levels during B cell development promotes Lyn activity, in addition to providing more substrate for Lyn.

Mice deficient in Lyn or CD22, as well as mice with a B cell-specific deletion of SHP-1 (36), are known to spontaneously produce autoantibodies against nuclear Ags, but the mechanism leading to this autoantibody production is unknown. We found that lyn^−/− mice had increased rag1 and rag2 mRNA expression in newly formed B cells, more modest increases in Igλ⁺ B cell frequencies, and no significant changes in RS rearrangements compared with their WT counterparts. These data suggest that Lyn deficiency does not compromise tolerance-related responses of immature B cells to self Ag and may even enhance them. Consistent with this conclusion, it has been shown that Lyn deficiency does not cause a significant change in the number of V_H3H9 Igλ⁺ DNA-reactive B cells in the spleen (46). Another study demonstrated that Lyn deficiency minimally affected the number of anti-HEL transgenic B cells in the BM of HEL-expressing mice, but did enhance the loss of anti-HEL transgenic mature B cells in the spleens of these mice (5). The implication of these results is that the Lyn-CD22-SHP-1 inhibitory pathway is not necessary for the establishment of tolerance to self during B cell development, in agreement with the slightly enhanced level of BCR signaling seen in immature B cells from mice genetically missing this pathway. In contrast, genetic mutations causing reductions in BCR signaling, and therefore reduced sensitivity to Ag, lead to enrichment of the B cell repertoire with autoreactive specificities in mice (47) and humans (48). Thus, it appears that Lyn deficiency leads to autoimmunity not through a defect in central tolerance induction, but rather due to a defect in the maintenance of tolerance in the periphery. In agreement with this idea, lyn^−/− mature follicular B cells exhibited substantially enhanced sensitivity to BCR-induced calcium release, ERK activation, and proliferation relative to their WT counterparts, which indicates that there is a reduction in the threshold of B cell activation in the absence of effective inhibition by the Lyn-CD22-SHP-1 pathway. This reduced threshold of activation could allow the recruitment of autoreactive B cells into an immune response, and perhaps in combination with abnormal myeloid cell function from Lyn deficiency, result in autoantibody production.

In summary, we have found that during maturation of B cells in the spleen, the sensitivity of the BCR to Ag stimulation was down-regulated. This developmentally acquired down-regulation in BCR sensitivity required Lyn and CD22, which act together with SHP-1 in an inhibitory pathway to suppress BCR signaling selectively in mature follicular B cells. Surprisingly, the Lyn-CD22-SHP-1 inhibitory pathway had a minor role in the regulation of BCR signaling in immature B cells. In agreement with the signaling results, Lyn deficiency had little effect on receptor editing-related events in immature B cells, which suggests that central tolerance mechanisms do not require this inhibitory pathway. On the basis of these observations, we propose that modulation of B cell sensitivity to Ag by the Lyn-CD22-SHP-1 inhibitory pathway during the course of B cell maturation is a critical factor in the maintenance of peripheral B cell tolerance to self Ags.

We thank Shuwei Jiang for cell sorting, and Arthur Weiss, Jason Cyster, Michelle Hermiston, and Clifford Lowell (all at University of California) and Ed Clark (University of Washington) for helpful discussions.

The authors have no financial conflict of interest.