B Cell Receptor Signal Transduction in the GC Is Short-Circuited by High Phosphatase Activity

See allHide authors and affiliations

Science  01 Jun 2012:
Vol. 336, Issue 6085, pp. 1178-1181
DOI: 10.1126/science.1213368


Germinal centers (GCs) generate memory B and plasma cells, which are essential for long-lived humoral immunity. GC B cells with high-affinity B cell receptors (BCRs) are selectively expanded. To enable this selection, BCRs of such cells are thought to signal differently from those with lower affinity. We show that, surprisingly, most proliferating GC B cells did not demonstrate active BCR signaling. Rather, spontaneous and induced signaling was limited by increased phosphatase activity. Accordingly, both SH2 domain–containing phosphatase-1 (SHP-1) and SH2 domain–containing inositol 5 phosphatase were hyperphosphorylated in GC cells and remained colocalized with BCRs after ligation. Furthermore, SHP-1 was required for GC maintenance. Intriguingly, GC B cells in the cell-cycle G2 period regained responsiveness to BCR stimulation. These data have implications for how higher-affinity B cells are selected in the GC.

T cell–dependent immune responses result in the selection of high-affinity B cells, a process that occurs in the germinal center (GC) and depends on high-rate somatic mutation of V regions to generate variants. Resultant GC B cells can differentiate into either memory or plasma cells, which confer lasting humoral immunity (1). During this process, the B cell receptor (BCR) promotes the selective survival or expansion of higher-affinity GC cells, but how this occurs is unclear. It is possible that BCRs on higher-affinity GC B cells transduce a stronger, more sustained or qualitatively different signal. A second possibility is that higher-affinity BCRs more effectively capture antigens (Ag), which are subsequently presented to helper T cells, resulting in higher-affinity B cells obtaining more T cell–derived survival or proliferative signals (1, 2). Though BCR function is central to the process of GC B cell selection, BCR signaling in the GC is not well understood.

In vivo BCR signaling in GC B cells is of great interest, as these cells are activated and undergo continuous selection based on BCR affinity. Such study is complicated by the fact that GC B cells are rare, transient, and heterogeneous. Furthermore, some GC B cells express an immunoglobulin G–containing BCR, which mediates different signaling than the immunoglobulin M (IgM) BCR (35). Heterogeneity may also confer distinct signaling phenotypes on GC BCRs, which would be obscured in experiments using assays of populations rather than single cells.

To overcome these issues, we have used an IgM BCR B1-8 transgenic (Tg) mouse (6, 7). The Tg encodes a germline Vh186.2 rearrangement that is common in the anti-nitrophenyl (anti-NP) response when combined with the 2 to 3% of Tg B cells expressing Vλ1. Such B cells in the Tg mice mount a vigorous GC response to NP–chicken gamma globulin (CGG) immunization (6, 7). These B cells undergo mutation in the Vλ1 chain, but because of the dominant role of the heavy chain in determining affinity, light-chain mutations have very limited effects on the affinity (6). Thus, these mice represent a source of large numbers of relatively homogeneous GC B cells expressing only IgM, thus obviating differential signaling by isotype-switched BCRs (3); moreover, the unimmunized Tg mice contain the exact naïve precursors of these GC B cells.

We first examined B cell signaling in freshly isolated splenic Ag-specific (i.e., Vλ1+) GC (peanut agglutinin, PNA+) and non-GC (PNA) cells that were immediately fixed, followed by flow cytometric analysis of phosphorylated proteins (fig. S1, A and B) (8). In immunized mice, non-GC cells are mainly nonresponding bystander cells and serve as an internal control, used in addition to naïve splenocytes. Naïve cells demonstrated basal tyrosine phosphorylation of the tyrosine kinase Syk (p-Syk, Tyr352) and its substrate BLNK (p-BLNK, Tyr84) (Fig. 1A), both of which are proximal signal transducer elements of the BCR. Results were consistent with genetic and inhibitor studies (9, 10). GC B cells, however, had little detectable p-Syk or p-BLNK and much reduced total phosphotyrosine (p-Tyr) compared with either non-GC or naïve Ag-specific B cells (Fig. 1A). In contrast, p-p38 (Thr180/Tyr182), p-ribosomal S6 (Ser235/Ser236), and p-Akt (Thr308), were present in GC cells at similar or higher levels compared to control cells (fig. S1C).

Fig. 1

Spontaneous and ligand-induced BCR signaling in GC, non-GC, and resting B cells. (A) BCR-linked basal signaling in gated populations of GC, non-GC, and naïve B cells from instantly fixed total splenocytes harvested from day-13 NP-CGG immunized mice (fig. S1). Fixed cells were treated with or without calf intestinal phosphatase (CIP) and then labeled with antibodies specific for phosphorylated proteins. (A) Histograms show representative results of tyrosine phosphorylation from CIP-treated (red) and untreated (blue) cells, overlaid for direct comparison. The y axes show relative cell numbers. Bar charts show net median fluorescence intensity (MFI) indicating basal phosphorylation calculated by subtracting MFI of CIP-treated from CIP-untreated cells. Error bars denote SEM from at least five independent experiments, each using cells pooled from spleens of at least two mice for each group. *P < 0.05 and **P < 0.01 MFI of GC or non-GC compared to naïve cells. (B) Response of GC, non-GC, and resting B cells to BCR ligation. Total splenocytes from day-13 NP-CGG immunized mice were stimulated ex vivo with anti-μ (15 μg/ml) for 5 min. Levels of Syk, Blnk, Tyr, p38, and Erk phosphorylation in gated GC and non-GC cells were measured. Profiles of GC-unstimulated (red filled area), GC anti-IgM–stimulated (red open trace), non-GC–unstimulated (blue filled area), and non-GC anti-IgM–stimulated (blue open trace) cells are overlaid for direct comparison.

The lack of BCR signaling could be explained by low in vivo Ag exposure or inherent resistance to BCR signals. To distinguish these possibilities, we stimulated splenocytes from day 13 postimmunization (fig. S2A) with the monoclonal antibody to IgM (anti-IgM), b.7-6 (Fig. 1B), F(ab′)2 anti-IgM, or NP–bovine serum albumin (fig. S2, B and C). In contrast to non-GC cells, GC B cells demonstrated little if any induction of several phosphoproteins downstream of the BCR, suggesting that they were inherently antigen-refractory.

To evaluate BCR down-regulation, we stimulated GC B cells directly with fluorescently labeled anti-IgM, which allowed us to electronically gate the analysis on cells with equivalent surface Ig levels (fig. S3A). Such GC cells again showed little induction of phosphoproteins compared with the non-GC cells with equivalent BCR expression. At 15 min poststimulation, GC B cells still did not contain elevated levels of p-Syk, excluding kinetic differences (fig. S3B). More than 97% of the cells were still viable at the end of stimulatory cultures (fig. S4A); consistent with this, GC cells were not inert—they generated p-Erk and p-p38 in response to phorbol 12-myristate 13-acetate (PMA)/ionomycin stimulation, which bypasses the BCR (fig. S2D). Furthermore, resistance to BCR-mediated generation of p-Syk was not a general property of activated B cells (fig. S4B).

To assess more proximal BCR signaling events, we evaluated basal phosphorylation of CD79, which is part of the BCR complex. GC cells had about half the amount of total CD79 compared with resting B cells (fig. S5, A to C) as expected, because GC cells have less surface Ig (fig. S3A). Upon stimulation with anti-IgM, only non-GC and naïve B cells showed an increase in the amount of p-CD79, revealing a defect at the earliest measurable events of BCR signaling (fig. S5, A to C).

Heightened phosphatase activity could explain the failure of GC B cells to accumulate increased tyrosine-phosphorylated CD79, Syk, and BLNK (11, 12); hence, we examined effects of Tyr phosphatase inhibition with H2O2. Exposure of GC B cells to H2O2 (fig. S6A) resulted in increased phosphorylation of Syk in non-GC cells, as reported by Wienands et al. (10). p-Erk, which is presumably downstream of Syk activation, was similarly affected, as were total p-Tyr levels. Notably, H2O2 rescued substantial p-Syk, p-Erk, and p-Tyr expression in GC cells. Phosphorylation was somewhat higher in the non-GC cells, however, particularly at early time points. These data suggest that established phosphatase activity restrains spontaneous phosphorylation in GC B cells.

Consistent with higher Tyr-phosphatase activity, maximal phosphorylation in the GC required a higher concentration of H2O2 compared with non-GC B cells (Fig. 2A). To determine if H2O2 directly disinhibited BCR signaling, we pretreated splenocytes of immunized animals with a suboptimal concentration of H2O2 before anti-IgM stimulation. Such pretreatment synergistically enhanced BCR-induced Syk phosphorylation compared with anti-IgM or H2O2 alone (Fig. 2B). Treatment with anti-IgM and H2O2 (Fig. 2, C and D) also synergistically elicited more Ca2+ flux in both GC and non-GC cells compared with either anti-IgM or H2O2 alone; anti-IgM alone stimulated little Ca2+ flux in GC cells. The response to ionomycin showed that GC cells are at least as equipped to flux Ca2+ as non-GC cells (Fig. 2E). These results indicate that phosphatase activity limits the ability of GC B cells to carry out Tyr phosphorylation and Ca2+ flux in response to BCR stimulation.

Fig. 2

Analysis of phosphatase-dependent regulation of BCR signaling in GC cells. (A) Total splenocytes from day-13 NP-CGG immunized mice were stimulated with 5 mM (green) or 10 mM (blue) H2O2, followed by detection of p-Syk and p-Blnk by flow cytometry in gated GC and non-GC Ag-specific cells. These panels are representative of three independent trials each of three mice. (B) Assessment of interaction between BCR ligation and phosphatase inhibition. Total splenocytes from immunized mice as in (A) were stimulated with 15 μg/ml anti-IgM (red), 5 mM H2O2 (blue), or both (green), and generation of p-Syk was assessed by flow cytometry. Panels are representative of three or more experiments. (C) Indo1AM-loaded total splenocytes from immunized mice were stimulated with 15 μg/ml anti-IgM (red), 5 mM H2O2 (blue), or both (green). Stimuli were added after acquiring events for 5 min to establish a basal level. Profiles of stimulants in gated GC (top) and non-GC (bottom) populations are overlaid. The y axis shows the indo1 violet-to-blue fluorescence ratio, an indicator of intracellular Ca2+ levels. (D) Compiled responses of GC and non-GC cells treated as in (C). Background-subtracted MFI was calculated from gates drawn before (180-s time, background) and after (200-s time, beginning at approximately the initial peak of response to anti-μ) stimulation. Error bars show mean + SEM of net MFI from four independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001. (E) Ca2+ flux to ionomycin, used as a positive control indicating equal responsiveness of GC and non-GC cells.

To assess this, we measured both Tyr and Ser/Thr phosphatase activity in lysates from purified GC, non-GC, and naïve B cells. As predicted, there was more Tyr and Ser/Thr phosphatase activity in the GC than in naïve cells (fig. S6B). The finding of increased Ser/Thr phosphatase activity may be connected to impaired GC B cell generation of p-Erk and pp38 in response to BCR ligation (Fig. 1B). Although this could have been due to a block in proximal BCR signaling, there was also less Erk and p38 phosphorylation evident upon PMA/ionomycin stimulation in GC B cells (fig. S2D), consistent with increased Ser/Thr phosphatase activity.

We next examined whether SH2 domain–containing phosphatase-1 (SHP-1), a Tyr phosphatase known to regulate BCR signaling upon BCR ligation (1315), may be responsible for reduced GC BCR signaling. Possible SHP-1 substrates include CD79, Syk, Vav, BLNK, and CD22 (16). Furthermore, we investigated SH2 domain–containing inositol 5 phosphatase (SHIP-1), which also regulates BCR signaling (17, 18), as well as Src family members because they can exert negative effects on BCR signaling (19). The activity of these proteins is increased by phosphorylation (20, 21). We found more p-SHIP-1 (Tyr1020), p-SHP-1, and p-SRC (Tyr416) in unstimulated GC compared with non-GC and naïve B cells (Fig. 3, A and B, and fig. S6C). Ex vivo stimulation with anti-IgM increased phosphorylation of SHP-1, SHIP-1, and Src proteins (most likely Lyn) in non-GC and naïve B cells, reflecting normal regulation of signaling (15, 19, 20). However, p-SHIP-1, p-SHP-1, and p-Src did not rise in anti-IgM–stimulated GC B cells; in fact, they consistently decreased upon BCR ligation.

Fig. 3

Differences in phosphorylation and localization of SHIP-1 and SHP-1 in GC and naïve B cells after and without BCR ligation. (A) (Top) SHIP-1 phosphorylation (Tyr1020) was determined by Western blot of lysates from unstimulated and anti-IgM stimulated (5 min) fluorescence-activated cell–sorted GC, non-GC, and naïve B cells. β-actin was used as a loading control. (Bottom) Quantitation of blot shown in top is MFI of gated p-SHIP-1 bands relative to total SHIP-1. Blots shown are representative of three similar experiments. (B) (Top) Tyrosine phosphorylation of SHP-1 was determined by immunoprecipitation of total SHP-1 in fluorescence-activated cell–sorted GC and naïve B cells both unstimulated and after 2 min of anti-IgM stimulation, followed by blotting with anti-pTyr (4G10). (Bottom) Quantitation of blots shown in top is MFI of gated p-SHP-1 relative to β-actin, which was determined by Western blot on parallel samples of the same lysates. Data representative of four similar experiments. (C and D) High-throughput imaging cytometric analysis of SHP-1/BCR association. Total splenocytes were either left unstimulated or stimulated with anti-IgM (b.7-6) for 2, 5, 10, 15, and 30 min (see fig. S7A for 15- and 30-min summary data). (C) Representative images of GC (top row) and non-GC (bottom row) B cells captured by the Amnis Imagestream X (Amnis, Seattle, Washington). (D) Colocalization of SHP-1 and BCR was measured in gated non-GC (λ1+PNAlo) and GC (λ1+PNAhi) as similarity scores in GC (shaded) and non-GC (open) B cells at multiple time points with respect to BCR ligation (all time points summarized in fig. S7A). The box depicts a gate drawn at <1.2 similarity, which was used to calculate the percentage of GC and non-GC cells demonstrating substantial SHP-1/BCR dissociation at various times after BCR ligation (fig. S7B). (E and F) Effect of deletion of SHP-1 in B cells on the ongoing GC response. The strategy for tamoxifen-induced deletion using a new B cell–specific inducible Cre enzyme (hCD20-TamCre) and SHP-1fl/fl mice is detailed in fig. S8. SHP-1fl/fl mice with or without (control) the Cre Tg were immunized with NP-CGG, treated from days 9 to 12 with tamoxifen, and then analyzed at day 14. (E) Representative flow cytometric analysis of splenocytes from experimental (left) and control (right) mice, detecting Ag-specific GC cells as PNA+/NIP+ among gated B220+ B cells. Numbers are percentages of B cells in the gate. (F) Data from three independent experiments showing loss of GC B cells (B220+PNA+NIP+) upon SHP-1 deletion in B cells.

SHP-1 (15) and SHIP-1 (22) are constitutively associated with the BCR in resting cells,. Unstimulated GC B cells demonstrated more SHP-1/BCR (Fig. 3, C and D, and fig. S7, A and B) and SHIP-1/BCR (fig. S7C) colocalization than did non-GC cells. BCR ligation resulted in brisk dissociation of both SHP-1 and SHIP-1 from BCR in non-GC cells. In these cells, BCR and SHP-1 relocalized to opposite poles of the cell, consistent with biochemical analysis of resting B cells (15). This dissociation was sustained for SHP-1 but transient for SHIP-1. In contrast, GC B cells barely showed any dissociation upon BCR ligation, though BCR/SHP-1 colocalization was reduced in a minor subset, albeit with slower onset and less sustained duration than in non-GC cells (Fig. 3D, fig. S7B, and see below). The differential localization of phosphatases in GC versus non-GC cells is further consistent with the regulation of BCR signaling in GC B cells by phosphatases.

To determine the importance of SHP-1 in B cells during GC responses, we crossed SHP-1fl/fl (23) with a new Tg strain (hCD20.TamCre) that allows B cell–specific inducible Cre action (fig. S8, A to C). B cell SHP-1 expression was reduced by injecting tamoxifen at days 9 to 12 after immunization (fig. S8, A to C), which resulted in marked reduction of GC B cell frequencies (Fig. 3, E and F), indicating that SHP-1 and, presumably, regulation of BCR signaling is required for maintaining the GC reaction.

If GC BCRs do not signal, then how does affinity-based selection occur? We did observe a small fraction of GC B cells in which Syk is phosphorylated (fig. S5D) and SHP-1 relocalized (Fig. 3D and fig. S7B) upon BCR ligation. We thus tested whether BCR signal transduction was dependent on the cell cycle. BCR ligation induced p-Syk only in GC cells within the G2/M phase (Fig. 4, A and B). Commensurate with this, we found that dissociation of BCR and SHP-1 occurred in G2; the amount of total SHP-1 was also substantially reduced in this phase but returned to normal levels at M phase (Fig. 4, C to F, and fig. S7, D and E).

Fig. 4

BCR signal transduction in GC B cells at the G2/M phase of the cell cycle. Mice at day 13 postimmunization with NP-CGG were injected intravenously once with 3 mg of 5-bromo-2′-deoxyuridine (BrdU) and sacrificed 1 hour later. Splenocytes were isolated and either treated with anti-IgM or not for 5′, then fixed and stained as described in the supplemental materials and methods. (A) Flow cytometry gating to identify GC B cells (left) and then separate them into cell-cycle compartments based on 4′,6-diamidino-2-phenylindole (DAPI) and BrdU staining (right). (B) Phosphorylation of Syk in response to BCR ligation was measured in gated populations based on (A) and as labeled: G1 (left), S (center), and G2/M (right). Histograms of p-Syk staining in unstimulated cultures (gray-filled) and stimulated cultures (open) are shown. Data are representative of four independent mice from two independent experiments, all with similar results. (C to F) GC cells were analyzed on the Imagestream X for SHP-1 intensity and SHP-1/BCR colocalization during different phases of the cell cycle. GC B cells were prepared and treated as in (A) but analyzed on the Imagestream after staining as in Fig. 3. (C) DAPI and BrdU identify G1 (green), S (blue), and G2/M phases (red/black). G2 and M phases were separated based on nuclear area and aspect ratio. (D) Analysis of total SHP-1 expression as a function of cell cycle. (E and F) Analysis of SHP-1/BCR colocalization as a function of cell cycle during the G2 phase compared with the G1 and S phases. However, the SHP-1 level increased back at the M phase. Moreover, ex vivo stimulation resulted in time-dependent dissociation of SHP-1 from the BCR only during the G2 phase. Data are representative of two experiments.

The discovery that most highly proliferative GC B cells undergoing Ag-driven selection cannot execute BCR signaling was contrary to expectations. This posttranslational strategy for constraining signaling makes sense, however, as it could quickly be altered as cells progress through cycle or in response to other signals. These studies highlight a switch that gates early events in BCR signaling and that can be modified in GC B cells.

How, then, might affinity-based selection work? We suggest a model in which B cells integrate both T cell–derived and BCR signals in a cell-cycle–dependent manner. Studies suggest that individual BCRs are dedicated to either BCR signaling or Ag presentation, with only the former depending on canonical p-CD79 generation (24). GC BCRs, which do not generate p-CD79, may thus favor Ag presentation. Because BCR Ag capture is affinity dependent (25), more avid B cells would win out in a competition to gain T cell–dependent signals, for example, via CD40L. Such signals would likely promote survival over a longer time course (6) rather than instantaneously switch the outcome of BCR signaling by reversing phosphatase activity, given the known kinetics and molecular connections between CD40 and NF-κB signaling (26, 27).

Nonetheless, BCR signaling may be important at key times during GC B cell proliferative cycling. B cells must test their affinity with each round of somatic hypermutation (28). The testing of mutant phenotypes possibly occurs once per cell division, with the quality of BCR signaling determining the likelihood of completing mitosis and/or subsequent survival. Even in G2, the threshold for effective Ag sensing appears elevated, thus favoring selection of higher-affinity B cells.

Supplementary Materials

Materials and Methods

Figs. S1 to S8


References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank A. Haberman for critical reading of the manuscript; J. Bezbradica for assistance and advice on Western blotting, immunoprecipitation, and LiCor imaging; and G. Nolan for advice and protocols for flow cytometric detection of intracellular phosphoproteins. The data are tabulated in the main paper and in the supplementary materials. This work was supported by NIH grants AI43603 and AR44077 (to M.J.S.). A.M.K. performed experiments; A.M.K., J.C.C., and M.J.S. designed experiments, interpreted data, and wrote the manuscript; and J.C.C. contributed key reagents.
View Abstract

Navigate This Article