Regulation of late B cell differentiation by intrinsic IKKα-dependent signals

Abstract

NF-κB-inducing kinase (NIK)-mediated IKKα phosphorylation activates the alternative NF-κB pathway, which is characterized by nuclear translocation of p52:RelB heterodimers. This alternative pathway is initiated by a select few receptors, including LT-βR, BAFF-R, and CD40. Although NIK, IKKα, and p52 are all critical regulators of LT-βR signaling in stromal cells during humoral immune responses, lymphocytes require NIK, but not p52, for optimal Ig production. This disparity suggests that NIK possesses critical cell-type-specific functions that do not depend on NF-κB. Here we use mice bearing targeted mutations of the IKKα activation loop Ser^176/180 (IKKα^AA) to address the B cell-intrinsic functions of NIK–IKKα signaling in vivo. We find that IKKα^AA B cells mount normal primary antibody responses but do not enter germinal centers. This defect likely derives from ineffective early T–B cell collaboration and leads to impaired generation of humoral memory and relatively short-lived, low-affinity antibody production. Our findings contrast with those obtained by using p52^−/− B cells, which mount normal Ig responses, and alymphoplasia (NIK mutant) B cells, which produce very little primary Ig. Thus, the NIK–IKKα–p52 axis is not as linear and exclusive as previous studies suggest, and IKKα possesses critical NF-κB-independent functions in B cells.

High-affinity Ig responses involve the formation of specialized germinal centers (GCs) in secondary lymphoid follicles (1). During GC initiation and maintenance, collaboration among rare antigen-specific lymphocytes, stromal cells, and myeloid cells promotes B cell differentiation into high-affinity Ig-secreting plasma cells and memory cells (reviewed in ref. 2). These cellular interactions are facilitated by chemokines, adhesion molecules, TNF family members, Toll-like receptor ligands, and antigen. Recognition of these various ligands by a subset of corresponding receptors can induce gene transcription via the NF-κB family of transcription factors (c-Rel, RelA, p50, p52, and RelB) (3–5). Although NF-κB signaling clearly potentiates B cell activation ex vivo, its intrinsic importance for humoral immunity in vivo remains poorly defined.

In the resting state, NF-κB heterodimers are held inactive in the cytoplasm via association with regulatory IκB subunits (reviewed in ref. 6). Receptor ligation initiates phosphorylation of IκB proteins by IκB kinase (IKK) complexes and subsequent IκB degradation. p52 and p50 are generated from respective p100 (NF-κB2) and p105 (NF-κB1) precursors, which contain carboxyl-terminal IκB-like regulatory domains that are cleaved after IKK activation. Two separate pathways promote NF-κB target gene activation. The classical pathway requires IKKβ kinase activity (7, 8), whereas the alternative pathway involves selective nuclear translocation of p52:RelB dimers after NF-κB-inducing kinase (NIK)-mediated phosphorylation of IKKα (but not IKKβ) (9–11).

Interestingly, the classical and alternative pathways are initiated by distinct extracellular stimuli. Although IL-1, TNF, LTα1β2, CD40L, antigen receptors, and LPS all induce NF-κB activation through the classical pathway (4, 6), the alternative pathway is more selectively initiated after anti-CD40, BAFF, or LTα1β2 treatment (10–14). However, in the context of inflammatory responses the classical pathway likely influences alternative pathway function via secondary mechanisms, e.g., inflammation-induced BAFF, CD40L, or p100 expression (15–18), and BAFF-mediated autoimmunity requires both the alternative and classical pathways (19). Nonetheless, because B cells are the only cells known to express both BAFF-R and CD40, and LTα1β2:LT-βR are critical for optimal Ig and GC responses, the alternative pathway seems selectively suited to regulate B cell differentiation.

The current paradigm of alternative NF-κB signaling was defined largely by using primary fibroblasts and suggests a linear and exclusive kinase–substrate relationship among NIK, IKKα, and p100 processing (3, 8). Accordingly, both alymphoplasia (aly/aly) mice, which bear an inactivating NIK mutation, and NF-κB2^−/− mice are unable to mount GC responses because of impaired maturation/function of nonhematopoietic stromal cells (20–22), likely because of suboptimal LT-βR-mediated gene transcription (7, 23). Interestingly, NIK activity is also required in hematopoietic cells for normal Ig responses (21, 22), but NF-κB2^−/− and RelB^−/− hematopoietic cells can respond normally to thymus-dependent immunogens (20, 24). These findings indicate that the alternative NF-κB pathway (induction of p52:RelB heterodimers) is not absolutely required in hematopoietic cells for humoral immunity and suggests that NIK likely has signaling functions distinct from p100 processing. Consistent with this latter possibility, inhibiting NIK expression affects classical NF-κB activation pathways (25), aly/aly B cells display diminished proliferative responses to LPS (which does not induce p100 processing) (22, 26), and, in T cells, NIK may directly phosphorylate the transactivation domain of c-Rel (27). Whether these p100-independent effects of NIK depend on IKKα has not been evaluated. Moreover, although IKKα was initially thought to exclusively activate NF-κB, recent studies suggest that it has additional targets such as histone H3 and E2F1, which are phosphorylated after nuclear IKKα translocation (28, 29). Thus, although an exclusive kinase–substrate relationship among NIK, IKKα, and p100 may exist in stromal cells, the situation is likely more complicated in hematopoietic cells where NIK (and potentially IKKα) may also possess critical functions independent from the alternative NF-κB (p52:RelB) pathway.

To further understand the relationship among NIK, IKKα, and p100 during humoral immune responses we performed an in-depth analysis of IKKα^AA knockin mice, which bear mutations in the IKKα activation loop at Ser¹⁷⁶ and Ser¹⁸⁰ that prevent its phosphorylation by NIK (and therefore p52:RelB activation) but do not affect induction of classical NF-κB heterodimers or kinase-independent IKKα functions in keratinocyte development (10, 11, 30). This model is useful for examining cell-type-specific requirements of the alternative NF-κB pathway and is also optimal for determining which, if any, of the p100-independent roles of NIK depend on IKKα activation. Unlike IKKα^−/− fetal liver-derived B cells, which do not mature past the transitional T₂ stage (13), IKKα^AA B cells are able to reach maturity (11). We previously reported that IKKα Ser^176/180-dependent signals facilitate thymus-dependent responses by promoting LTβ-R-mediated maturation of follicular dendritic stromal cells (FDCs) (10) (consistent with results from aly/aly and NF-κB2^−/− chimeric mice). Herein we report the unexpected finding that IKKα Ser^176/180-dependent signals are also required intrinsically in B cells for GC induction, but not for initial production of specific antibodies after primary immunization. This finding contrasts with the observation that NIK inactivation in hematopoietic cells severely limits primary antibody production (21, 22). Collectively, these results reveal that the NIK/IKKα axis acts coordinately in both FDCs and B cells to effect high-affinity humoral immunity and that in B cells NIK plays important signaling roles that may be independent of IKKα activation.

Results and Discussion

Impaired GC Formation and Plasma Cell Accumulation, but Normal Ig Production, After Primary Immunization of μMT/IKKα^AA Chimeras.

Because the B cell tropic receptors CD40 and BAFF-R both induce NIK-mediated IKKα activation, we sought to determine whether, in addition to their established role in FDCs, IKKα Ser^176/180-dependent signals are critical in B cells. To investigate this possibility we generated chimeric mice (hereafter referred to as μMT/WT and μMT/AA) by adoptively transferring WT or IKKα^AA bone marrow into sublethally irradiated B cell-deficient μMT recipients. By using this approach, B cells in reconstituted mice derive exclusively from donor stem cells, whereas other hematopoietic compartments display mixed chimerism, providing an appropriate model system for evaluating the B cell-intrinsic role of NIK–IKKα association in vivo. After reconstitution, we immunized chimeric mice with NP-KLH/alum and monitored the accumulation of plasma cells, GC B cells, and NP-specific antibodies. μMT/AA chimeras contained normal numbers of B cells, and we observed normal accumulation of splenic inflammatory myeloid cells and maturation of FDC networks after immunization [Fig. 1A and supporting information (SI)]. The normal appearance of stromal cell networks indicates that FDCs were derived from μMT host (and not IKKα^AA) cells as expected. However, using flow cytometry we found reduced generation of splenic CD138⁺ plasma cells and no evidence of GL-7⁺ GC B cell accumulation in μMT/AA mice at day 14 (d14) after immunization (Fig. 1 A and B). The deficit in splenic IKKα^AA plasma cell accumulation (which was also apparent in the bone marrow; data not shown) was not paralleled by a significant decrease in primary hapten-specific antibody titers (Fig. 1C). Thus, B cell-intrinsic IKKα Ser^176/180-dependent signals are required for GC formation and splenic CD138⁺ plasma cell accumulation during primary immune responses but are dispensable for induction of primary antigen-specific Ig.

Fig. 1.

IKKα Ser^176/180-dependent signals are required in B cells for GC formation and accumulation of CD138⁺ plasma cells, but not for antigen-specific primary Ig production. μMT/WT and μMT/AA chimeras were generated as described in Materials and Methods and immunized with NP-KLH/alum. (A) Expression of B220, IgM, CD4, CD35, and PNA-binding antigen was assessed via fluorescence microscopy in splenic cryosections on d14 after immunization. (B) The accumulation of splenic GC phenotype (B220⁺ GL-7⁺) B cells and CD138⁺ IgM⁺ plasma cells was analyzed on d14 via flow cytometry. Shown are high-resolution 5% contour plots that were first gated on live CD3⁻ events. Numbers indicate the average percentage of live CD3⁻ events that were IgM⁺ CD138⁺ (Upper) or B220⁺ GL-7⁺ (Lower) ± 1 standard deviation from three immunized chimeric (or naïve) mice. (C) The relative abundance of NP-specific IgM and IgG1 was assessed in sera collected from immunized μMT/WT and μMT/AA chimeras on the indicated days after immunization. All results are representative of three independent experiments.

Diminished Long-Lived Hapten-Specific Ig Titers and Impaired Recall Responses in μMT/AA Chimeras.

The normal induction of primary antibody upon immunization suggests that, without NIK-mediated IKKα activation, B cells can differentiate into Ig-secreting cells but that the resulting effector cells display an atypical cell-surface phenotype (are CD138⁻), are short-lived, or are too infrequent for detection via non-antigen-specific flow cytometry. Interestingly, histologic examination of splenic cryosections revealed the accumulation of extrafollicular IgM^bright cells in the red pulp, which are characteristic of pre-GC short-lived plasmablast responses and may be responsible for normal primary antibody production (SI Fig. 7). To further investigate whether antigen-specific Ig produced in μMT/AA chimeras reflected induction of short-lived immature CD138⁻ plasma cells or the accumulation of long-lived plasma cells in organs other than the spleen or bone marrow, we assessed long-lived NP-specific IgG1 titers after primary immunization with NP-KLH/alum. Although anti-NP IgG1 titers in μMT/AA chimeras were normal on d14 after primary challenge, by d28 they had diminished to nearly preimmune levels (Fig. 2A). In contrast, NP-specific serum IgG1 titers were maintained in μMT/WT chimeras on d28 at levels similar to those observed near the peak of the response (d14). Thus, B cell-intrinsic IKKα Ser^176/180-dependent signals regulate the longevity of antigen-specific Ig titers after primary immunization, suggesting that the plasma cells producing anti-NP Ig on d14 in μMT/AA chimeras were short-lived.

Fig. 2.

NIK-mediated IKKα activation is required for the generation of long-lasting antigen-specific Ig and enhanced antigen responsiveness upon antigen recall. μMT/WT and μMT/AA chimeras were immunized as in Fig. 1. (A) NP-specific serum IgG1 titers were monitored on the indicated days after primary immunization (left arrow). Mice received a second dose of NP-KLH/alum on d28 (right arrow). NP₃₀-BSA-coated plates were used to measure total anti-NP IgG1, whereas binding to NP₃-BSA-coated plates was used to enrich for high-affinity Ig. Endpoint dilutions were defined as the least concentrated serum sample displaying an A₄₀₅ of >0.2 (4-fold above the background A₄₀₅). Circles represent values from individual mice. LD, limit of detection; n.s., not significant. Lines show sample averages. ∗, P < 0.05 (Student's t test). (B) The accumulation of splenic B220⁺ GL-7⁺ GC B cells was monitored on d7 after secondary immunization (d35 after primary challenge) as in Fig. 1B. The bar graph shows the average percentage + 1 standard deviation of CD3⁻ cells displaying the indicated phenotype from four μMT/WT and five μMT/AA chimeras on d7 after secondary immunization.

Because Ig affinity maturation and memory B cell production are the major outcomes of GC reactions, we predicted that μMT/AA chimeras, which lack detectable GCs, would be hyporesponsive upon secondary NP-KLH/alum challenge and produce relatively low/moderate-affinity Ig. Although μMT/WT chimeras rapidly (within 7 days) mounted robust NP-specific Ig responses after secondary challenge that exceeded titers observed during the primary response (Fig. 2A, d14 vs. d35), secondary NP-specific Ig production in μMT/AA chimeras did not exceed that of the primary response. This finding indicates that functional humoral memory generation depends on NIK–IKKα signaling in B cells and is consistent with impaired primary GC formation in μMT/AA chimeras. In addition, we did not observe GC B cell or CD138⁺ plasma cell induction after secondary challenge of μMT/AA chimeras (Fig. 2B). Finally, differences in serum anti-NP titer between immunized μMT/WT and μMT/AA chimeras were most apparent when NP₃-BSA (which enriches for high-affinity antibodies) was used to capture specific Ig instead of NP₃₀-BSA. In fact, upon capture with NP₃, we observed significantly lower titers in μMT/AA chimeras during the primary response. Together, these data indicate that B cells must initiate NIK–IKKα signaling to undergo affinity maturation and memory B cell differentiation in the GC.

IKKα^AA B Cells Proliferate and Express Activation Markers Normally ex Vivo but Display Impaired Plasma Cell Differentiation.

Both GC responses and high-affinity Ig production are severely impaired in mice lacking CD40/CD40L signals (31, 32). Thus, we postulated that impaired B cell responsiveness to this receptor ligand pair accounts for the lack of GCs observed in μMT/AA chimeras. CD40 ligation induces B cell proliferation, costimulatory molecule expression, and, in the presence of cytokines, differentiation into antibody-secreting cells (reviewed in ref. 33). Interestingly, we found that IKKα^AA B cells proliferated normally after stimulation with anti-CD40 alone or in combination with IL-4 or BAFF (Fig. 3A). These data contrast with reports of suboptimal CD40-mediated proliferation of aly/aly B cells (22, 26). Because BCR ligation has not been reported to induce p100 cleavage, we were not surprised to find that IKKα^AA B cells also proliferated normally after BCR ligation (Fig. 3A). Thus, NIK-mediated IKKα activation does not significantly influence B cell proliferation after CD40 ligation.

Fig. 3.

IKKα Ser^176/180-dependent signals regulate plasma cell differentiation ex vivo but not mitogenic responses or induction of MHC class II expression. WT and IKKα^AA B cells were purified, CFSE-labeled, and cultured for 3 or 6 days as described in Materials and Methods. (A) Numbers indicate the percentage of live cells that proliferated (were CFSE^low). Results are representative of at least four independent experiments. (B) Plasma cell differentiation was assessed by monitoring the frequency of CD138⁺, proliferated (CFSE^low) B cells. (C) The mean frequency (+ 1 standard deviation) of plasma cells (Left) and dead (7-amino-actinomycin D⁺) plasma cells (Right) in duplicate samples. Results in B and C are representative of two independent experiments.

In addition, because aly/aly B cells are hypoproliferative after LPS stimulation (which does not induce p100 cleavage), we determined whether IKKα is important for p52-independent NIK-mediated B cell activation. Unlike aly/aly and IKKα^−/− B cells (13, 22), IKKα^AA B cells proliferated normally in response to LPS (Fig. 3A). We also found that IKKα^AA B cells induced MHC class II expression normally after all tested activation stimuli (Fig. 3A). These findings are consistent with the normal LPS- or anti-CD40-mediated proliferation of NF-κB2^−/− B cells (34) and suggest that the reported role of NIK in ex vivo B cell clonal expansion is independent of IKKα and p52:RelB heterodimers (22, 26). Notably, we observed normal LT-β expression in anti-CD40-treated IKKα^AA B cells (data not shown) and FDC maturation in μMT/AA chimeras (SI Fig. 7), indicating that the impaired responsiveness of IKKα^AA B cells does not derive from reduced delivery of LT-αβ to FDCs.

Based on the impaired plasma cell accumulation and short-lived primary antibody titers observed in μMT/AA chimeras, we assessed the ability of IKKα^AA B cells to undergo plasma cell differentiation ex vivo. Consistent with radiation chimera data in Fig. 1, we observed impaired accumulation of CD138⁺ IKKα^AA plasma cells in response to LPS and CD40 ligation (Fig. 3B). This deficit was likely due to enhanced plasma cell death because an enhanced percentage of CD138⁺ IKKα^AA cells bound the cell death marker 7AAD (Fig. 3C). Thus, both in vivo and ex vivo data support the conclusion that NIK–IKKα signals regulate plasma cell longevity.

Critical IKKα Ser^176/180-Dependent Effector Functions in B Cells Cannot Be Provided in Trans by WT Cells.

Because GC formation involves collaboration between B cells and activated T cells, we monitored the accumulation of splenic CD62L^low ICOS⁺ CD4⁺ T cells after immunization of μMT/WT or μMT/AA chimeras. We found that ICOS⁺ T cells accumulated in μMT/WT, but not μMT/AA chimeras, indicating that IKKα Ser^176/180-dependent signals in B cells regulate T cell activation in vivo (Fig. 4A). In addition, CD4⁺ cell recruitment to FDC zones was impaired in immunized μMT/AA mice (Figs. 1A Right and 4B). By flow cytometry, >95% of all splenic CD4⁺ cells also expressed CD3 (data not shown), strongly suggesting that FDC zone-resident CD4⁺ cells observed in immunized chimeras were T cells. Together, these findings suggest that B cell-intrinsic IKKα signals control differentiation of recently described ICOS⁺ CXCR5⁺ follicular T_H (T_FH) cells (35) and are consistent with the observation that interaction with B cells affects the differentiation of T_FH cells ex vivo (36). However, μMT/AA chimeras were not completely devoid of ICOS⁺ CD4⁺ cells (Fig. 4A) or FDC-associated CD4⁺ cells (Figs. 1A and 4B), indicating that B cell-intrinsic IKKα Ser^176/180-dependent signals act in concert with additional factors to facilitate optimal T_FH generation.

Fig. 4.

IKKα signals control the accumulation of activated splenic T cells and the localization of CD4⁺ cells to FDC networks upon immunization. (A) The accumulation of activated (CD62L^low ICOS⁺) splenic CD4⁺ T cells was assessed in μMT/WT and μMT/AA chimeras via flow cytometry on d14 after immunization as described for GC B cells in Fig. 1B. Numbers are the percentage (+ 1 standard deviation) of CD4⁺ T cells with an activated phenotype from three mice of the indicated genotypes. (B) Splenic cryosections were stained with antibodies directed against CD35, CD4, and B220 as in Fig. 1A Right. The relative accumulation of CD4⁺ cells in splenic FDC zones was assessed in immunized μMT/WT and μMT/AA chimeras. Data were collected from three separate immunized chimeric mice of each genotype, and each data point represents the relative CD4⁺ cell frequency in a single FDC zone.

These observations raised the possibility that IKKα^AA B cells might be able to participate in GCs but that impaired activation of T cells or FDCs in μMT/AA chimeras precludes GC initiation. Moreover, B cells produce factors such as TNF and lymphotoxin that are important for FDC maturation during GC initiation (37). Thus, it seemed plausible that IKKα^AA B cells might respond normally in the presence of an alternative source of these “trans acting” factors and/or optimally activated T cells. To address these possibilities, we created mixed chimeras by reconstituting CD45.2⁺ μMT recipients with bone marrow cells from CD45.1⁺ WT and CD45.2⁺ IKKα^AA mice combined at a 1:1 ratio (data not shown). After reconstitution, we observed a moderate competitive disadvantage for IKKα^AA B cells, which comprised ≈30% of the total μMT mixed chimera B cell pool (data not shown). Strikingly, flow cytometric examination of splenocytes at d14 after immunization with NP-KLH/alum revealed that GC B cells in μMT/WT-AA mixed chimeras were exclusively derived from CD45.1⁺ WT precursors (Fig. 5A). Similarly, we observed a paucity of splenic CD138⁺ CD45.2⁺ IKKα^AA plasma cells in immunized mixed chimeras. As expected, the presence of WT B cells was sufficient to promote normal accumulation of ICOS⁺ CD4⁺ T cells in μMT/WT-AA mixed chimeras upon immunization (SI Fig. 8A). Histologic analyses supported flow cytometric data, revealing that IKKα^AA B cells (CD45.2⁺ CD3/CD11b⁻ cells) could differentiate into red pulp-resident IgM^bright plasma cells but did not enter GCs to undergo proliferation (i.e., become Ki67⁺) and selection (Fig. 5B and SI Fig. 8C). Thus, although B cell-intrinsic IKKα Ser^176/180 signals control T cell activation in vivo (Fig. 4), even in an environment containing optimally activated T cells and robust GCs, B cells cannot participate in the GC response without intact NIK–IKKα signaling.

Fig. 5.

IKKα^AA B cells are unable to enter GCs in the presence of responding WT B cells in vivo. Mixed μMT/WT-AA chimeras were created and immunized as described in Materials and Methods. In these μMT/WT-AA mice the B cell compartment displayed “two-way” mixed chimerism, comprising both CD45.1⁺ WT and CD45.2⁺ IKKα^AA B cells. Non-B cell hematopoietic compartments, however, displayed “three-way” mixed chimerism, being derived from CD45.1⁺ WT, CD45.2⁺ endogenous (μMT), and CD45.2⁺ IKKα^AA stem cells. (A) The percentage of WT and IKKα^AA B cells displaying a plasma cell (CD19⁺ CD138⁺) or GC (B220⁺ GL-7⁺) phenotype was assessed via flow cytometry at d14 after immunization. Lower graph shows the average percentage of B cells with the indicated phenotype (error bars depict 1 standard deviation). (B) Splenic cryosections from immunized μMT/WT-AA mixed chimeras (prepared on d14 after immunization) were stained with PNA and antibodies directed against CD45.2, CD3ε, and CD11b. Because CD45.2 was expressed on IKKα^AA B cells and non-B hematopoietic cells, IKKα^AA B cells were identified in situ based on lack of CD3ε or CD11b expression. No CD45.2⁺ CD3/CD11b⁻ cells were visible in cryosections from immunized chimeric mMT/CD45.1 WT mice (which contained no CD45.2⁺ B cells) (SI Fig. 8B).

The impaired induction of ICOS⁺ T cells in μMT/AA chimeras (and the associated deficit in follicular T cell localization; Fig. 4B) suggests that the primary defect in IKKα^AA B cells is a failure to effectively collaborate with T cells during early stages of the primary response. As a result, GCs are not initiated in μMT/AA chimeras, and antibody-secreting IKKα^AA effector cells are short-lived and express relatively low affinity Ig. Ex vivo results (Fig. 3) suggest that impaired lymphocyte activation in μMT/AA chimeras is not a consequence of suboptimal CD40-mediated clonal B cell expansion.

Our observation that IKKα^AA B cells are excluded from GCs even in the presence of responding WT B cells and optimally activated T cells suggests two possibilities. First, IKKα Ser^176/180-dependent signals may be critical in B cells during initial cognate T–B interaction, as well as during later stages of nascent GC formation such as migration into follicles, clonal expansion, and/or interaction with FDCs (37). Alternatively, initial T–B collaboration involves the formation of motile, monogamous T–B cell conjugate pairs (38). Because mature GCs are dominated by the cellular descendents of a common clone, the prevailing theory is that GCs are initiated by individual (or perhaps fewer than three) B cells (1, 39). It is possible that the CD62L^low ICOS⁺ CD4⁺ T cells that accumulate in μMT/WT-AA mixed chimeras derive exclusively from early monogamous T–WT B cell conjugate pairs. In this mixed chimera environment, IKKα^AA B cells may not gain access to activated T cells within an appropriate timeframe, precluding their entry into or initiation of GCs, and limiting their differentiative capacity to the formation of plasmablastic extrafollicular foci. The finding that IKKα signaling in B cells is critical for GC formation and long-lived Ig titers, but dispensable for primary antibody production, reveals that NIK (which itself is critical for primary Ig responses) possesses important IKKα-independent functions and underscores the importance of further evaluating the p100-independent activities of IKKα.

Materials and Methods

Mice and Immunizations.

μMT, IKKα^AA, and C57BL/6 (WT) mice were maintained in a specific pathogen-free environment at the Burnham Institute for Medical Research. Chimeric mice were generated by administering a sublethal dose (500 rad) of γ-radiation to μMT recipients, followed by immediate i.v. transfer of ≈5 × 10⁶ donor bone marrow cells (previously depleted of erythrocytes by using hypotonic ammonium chloride) from WT or IKKα^AA mice. Mixed chimeras described in Fig. 4 received 2.5 × 10⁶ WT and 2.5 × 10⁶ IKKα^AA bone marrow cells in a single injection. Chimeras were aged for at least 6 weeks to allow for complete reconstitution and subsequently immunized i.p. with 50 μg of NP-KLH precipitated in alum. Secondary immunizations described in Fig. 2 were also performed with 50 μg of NP-KLH/alum.

ELISA.

Sera were collected from all recipients before immunization and on the indicated days after challenge. Ninety-six-well high-binding-capacity plates were coated with NP₃-BSA or NP₃₀-BSA (Biosearch Technologies, Novato, CA) for 18 h at 4°C. Plates were blocked for 20 min at room temperature with PBS containing 0.5% BSA. Sera were serially diluted (beginning at 1:300) 2-fold in PBS/0.5% BSA and incubated in NP-BSA-coated wells for 2 h at room temperature. Plates were extensively washed and further incubated with alkaline phosphatase-conjugated anti-mouse IgM or anti-mouse IgG1 secondary antibodies for 1 h at room temperature. After additional washing, phosphatase substrate (Sigma, St. Louis, MO) was added to samples and the A₄₀₅ was measured. Data in figures indicate the least concentrated dilution displaying an A₄₀₅ of >0.2 (a value of 100,000 indicates that serum less concentrated than 1:100,000 yielded an A₄₀₅ of <0.2).

Flow Cytometry.

The spleens of naïve or immunized mice were excised, and erythrocytes were depleted from single-cell suspensions by using hypotonic ammonium chloride. A total of 0.25–1.0 × 10⁶ cells were suspended in PBS/0.5% BSA and incubated with combinations of the following mouse-specific antibodies/reagents, which were all purchased from BD Biosciences Pharmingen (San Diego, CA): GL-7-FITC, anti-CD138-phycoerythrin (PE), anti-IgM-allophycocyanin, anti-ICOS-PE, anti-CD62L-allophycocyanin, anti-CD3-PEcy7, anti-CD11b-PEcy7, anti-CD11b-FITC, 7-amino-actinomycin D, and anti-CD45.1-allophycocyanin. Samples were washed once with PBS/0.5% BSA, and the expression of the indicated molecules was assessed by using a FACSCanto flow cytometer. Compensation between fluorescent channels was determined based on staining in single-stain control samples. Analysis and figure preparation were performed by using FlowJo software (Treestar, Ashland, OR).

Histology.

Spleens from immunized chimeras were embedded in Tissue-Tek O.C.T. Compound (Sakura Finetek, Torrance, CA) and frozen at −80°C. Sections (8 μm) were mounted to microscope slides, fixed for 10 min in cold acetone, blocked for 15 min with PBS/0.5% BSA, and stained with the following mouse-specific reagents from BD Biosciences for 1 h at room temperature: anti-B220-FITC, anti-IgM-allophycocyanin, anti-CD4-FITC, anti-CD3-PE, anti-CD11b-PE, anti-CD35-biotin, and anti-CD45.2-FITC. PNA-biotin for GC detection was from Vector Laboratories (Burlingame, CA). Sections were washed with PBS and incubated with streptavidin-Cy3 (Invitrogen/Zymed, Carlsbad, CA) or streptavidin-Cy5 (Jackson ImmunoResearch Laboratories, West Grove, PA) for 30 min. After extensive additional washing with PBS, samples were covered with Gel/Mount (Biomeda, Foster City, CA) and sealed with glass coverslips. Images were acquired by using a Zeiss Axiocam M1 microscope and Slidebook software (Intelligent Imaging Innovations, Denver, CO). For analysis of follicular CD4⁺ cell accumulation, the total pixel area of FDC zones was determined in Photoshop (Adobe Systems, San Jose, CA) by using the polygon lasso tool to define regions containing CD35^bright cells. The number of CD4⁺ cells was counted within defined FDC zones, and the relative cell accumulation was calculated according to the following equation: no. of FDC zone CD4⁺ cells/(FDC zone pixel area/1,000).

Ex Vivo B Cell Stimulation.

WT and IKKα^AA B cells were isolated from splenic single-cell suspensions by using MACS Technology to deplete CD43⁺ cells according to the manufacturer's recommended procedure (Miltenyi Biotec, Auburn, CA). To monitor proliferation, cells were incubated at 2 × 10⁷ per ml for 5 min at room temperature in PBS containing 2.5 mM CFDA-SE (Invitrogen/Molecular Probes, Carlsbad, CA). Samples were washed with PBS/0.5% BSA and cultured in RPMI medium 1640 containing 10% FCS. Cells were stimulated with anti-IgM F(ab)′₂ (10 μg/ml; Zymed), LPS (10 μg/ml), anti-CD40 (10 μg/ml IC10; eBiosciences, San Diego, CA), recombinant murine IL-4 (20 ng/ml; R & D Systems, Minneapolis, MN), and/or recombinant human BAFF (200 ng/ml; R & D Systems). Division-mediated dilution of CFSE and binding of anti-I-A^b (MHC class II)-PE (BD Biosciences) were monitored via flow cytometry as described above. Histograms were normalized by using FlowJo and depict the percent maximum cell number for each culture condition.

Abbreviation

dn

day n

FDC

follicular dendritic stromal cell

GC

germinal center

IKK

IκB kinase

NIK

NF-κB-inducing kinase

PE

phycoerythrin.