A Novel Mutation in the Nfkb2 Gene Generates an NF-κB2 “Super Repressor”¹

The NF-κB transcription factors are critical regulators of the development and function of the immune system. Mammalian cells contain five NF-κB members (RelA, RelB, c-Rel, NF-κB1, and NF-κB2), each of which has an N-terminal Rel homology domain with sequences for dimerization, DNA binding, and nuclear localization (1). In resting cells, NF-κB dimers are maintained in an inactive state by interaction with IκB proteins, which mask the nuclear localization of NF-κB proteins, thereby preventing their nuclear migration. Activation of the NF-κB pathway by cytokines and other ligands leads to proteolytic processing of IκB proteins, allowing NF-κB dimers to translocate to the nucleus and activate the transcription of target genes (1).

NF-κB1/p50 and NF-κB2/p52 are synthesized as precursor proteins p105 and p100, respectively. The C-terminal regions of p105 and p100, containing ankyrin repeat domains, function as IκB and retain NF-κB proteins in the cytoplasm (2). Processing of the precursors into their active forms p50 and p52 is required for nuclear migration (1). In contrast to p105, which undergoes constitutive processing (3), p100 processing is tightly regulated and is triggered by external ligands (4). Activation by ligand increases the stability of NF-κB-inducing kinase (NIK),³ which recruits IκB kinase (IKK)α (5). IKKα phosphorylates p100 at serines 866 and 870 at the C terminus, enabling recruitment of the E3 ligase β-transducin repeat-containing protein (βTrCP) and partial processing by the proteasome to yield p52 (6).

The “canonical” NF-κB pathway has a central role in the inflammatory response. This pathway predominantly activates p50:RelA dimers and is triggered by a variety of proinflammatory ligands, including TNF-α, LPS, and viral proteins (2). By contrast, a limited number of TNF family cytokines activate the “noncanonical” NF-κB pathway; lymphotoxin (LT), receptor activator of NF-κB ligand (RANKL), CD40L, and B cell activating factor (7). Although the noncanonical pathway mostly activates p52:RelB dimers, both RelA and RelB containing NF-κB dimers are activated by these stimuli (8). The noncanonical pathway regulates the development and organization of secondary lymphoid organs (via LT signaling), medullary thymic epithelial cell differentiation (RANKL), thymocyte emigration (LT), B lymphocyte survival and maturation (CD40L and B cell activating factor), and bone homeostasis (RANKL) (9, 10, 11, 12, 13, 14).

Mice harboring mutations that affect NF-κB2 expression or function have revealed a complex range of physiological roles for this molecule. Nfkb2^−/− mice, which lack both p100 and p52, have disorganized splenic architecture, small or absent peripheral lymph node (LN), and a reduced number of mature B cells (9, 10, 15, 16, 17). Mice engineered to process p100 constitutively display abnormal lymphocyte proliferation and enlarged spleen and LN, and die during the perinatal period (18). Similar defects leading to unregulated p52 production are associated with the development of lymphoma in humans (19, 20). Collectively, these studies indicate that p52 levels are strictly controlled, and loss of this regulation results in severe immune abnormalities.

NIK plays a pivotal role in p100 phosphorylation and processing. The phenotype of Nik^−/− mice or alymphoplasia (aly) mice, which have a mutant form of NIK, is similar to that of Nfkb2^−/− mice, but the mesenteric LNs are absent in addition to peripheral LNs (21, 22). In this respect, these mice more closely resemble mice in which both the canonical and noncanonical NF-κB pathways are compromised, including LTα-deficient mice (Lta^−/−) (23), LTβ receptor-deficient mice (Ltbr^−/−) (11), and mice lacking both NF-κB1 and NF-κB2 (17). Like Nfkb2^−/− mice, Nfkb1^−/− mice have partial defects in LN formation, lacking inguinal LNs but with the mesenteric LNs retained (17). The more severe LN defect seen when both pathways are perturbed suggests that multiple NF-κB molecules participate in the development of LN and to some extent can compensate for the loss of the other. Therefore, the complete absence of LN in Nik^−/− mice, which retain p100 but lack p52, raises the possibility that the IκB function of p100 may not be restricted to regulating p52:RelB activity, but may also serve to regulate other NF-κB molecules.

Several recent studies have provided additional evidence for cross-talk between the canonical and noncanonical NF-κB pathways. In a study of osteoclastogenesis in NIK-deficient mice, p100 was shown to bind p50:RelA complexes and prevent their nuclear translocation (24). Using genetic mutants lacking the three canonical IκB proteins, p100 was shown to interact with and regulate p50:RelA complexes in response to noncanonical stimuli (25). Similarly, in naive T cells p100 forms complexes with p50:RelA to regulate T cell activation (26).

In this study, we describe the characterization of mice harboring a novel mutation in Nfkb2. This mutant allele encodes a nonprocessible form of p100, preventing p52 formation. Furthermore, the inhibitory precursor form of NF-κB2 inhibits the activation of RelA, leading to more severe defects than those found in Nfkb2^−/− mice. We provide evidence in vivo that NF-κB2 function is not limited to the regulation of p52:RelB activation, and that a key function of p100 is to interact with and regulate RelA-containing complexes.

Male BALB/c mice were treated with N-ethyl-N-nitrosourea (ENU) as described (27). ENU-treated mice were mated with isogenic females to yield G₁ progeny, which were bled at 7 wk of age, and the number of PBLs determined using an Advia 120 Automated Hematological Analyzer (Bayer). Mice were routinely housed in clean, conventional facilities and were used for experiments at 8–14 wk of age. All experiments with mice were approved by St. Vincent’s Hospital Animal Ethics Committee.

Mice heterozygous for the Lym1 mutation on a BALB/c background were mated with C57BL/6 wild-type (wt) mice. F₁ mice were identified at 7 wk of age by elevated PBL counts and intercrossed to produce F₂ mice. DNA was prepared from the liver of F₂ mice and simple sequence length polymorphisms spaced evenly throughout the genome were amplified and analyzed. The Lym1 mutation was localized to chromosome 19, and the candidate interval was refined via analysis of additional Mit marker and in-house CA repeat marker in the region.

DNA was prepared from tail biopsy and each exon of Nfkb2 was PCR-amplified and sequenced on an automated sequencer (Applied Biosystems). Genotyping was performed by PCR amplification of exon 23 of Nfkb2, followed by digestion with Bsu 36I (New England Biolabs). In the presence of the Lym1 mutation, a wt fragment (233 bp) is digested into two smaller fragments of 144 and 89 bp.

Total RNA was isolated from thymocyte suspensions using TRIzol reagent (Invitrogen Life Technologies). Real-time quantitative PCR analysis was performed with primer and probe Assay-on-demand sets (Applied Biosystems) for Aire and β-actin as a housekeeping reference gene. Analysis was performed on a Rotor-Gene RG-3000 cycler (Corbett Research). Results represent mean + SD of triplicates from three to four mice of each genotype.

Single cell suspensions were stained as described (28) with Abs specific for CD23 (clone B3B4), CD21 (7G6), B220 (RA3-6B2), CD4 (GK1.5), CD8 (53-6.7), CD69 (H1.2F3) and CD62L (Mel-14), and analyzed on a LSR, FACSCalibur, or FACSAria flow cytometer (BD Biosciences).

Small, resting B cells were isolated from spleen suspensions using a combination of Percoll gradient centrifugation and anti-B220 MACS beads (Miltenyi Biotec). B cells were stimulated for 3 days in RPMI 1640 with 5% FCS and 10⁻⁴ M 2-ME containing either an optimal dilution of baculovirus-derived CD40L, 20 μg/ml LPS (Difco) or 10 μg/ml F(ab′)₂ goat anti-mouse μ chain (Jackson ImmunoResearch Laboratories). Viability of cells cultured in medium alone was determined by flow cytometry at 24-h intervals by propidium iodide exclusion. For quantification of cell numbers, 2 × 10⁴ Calibrite beads (BD Biosciences) were added to each well before harvest.

Serum Abs were measured using isotype-specific ELISA, as described (28). Briefly, serum Abs were captured with goat anti-mouse isotype-specific sera (Southern Biotechnology Associates) and revealed with isotype-specific, HRP-conjugated goat anti-mouse sera (Southern Biotechnology Associates). Purified myeloma proteins were used as controls.

Frozen spleen sections were stained with unlabeled anti-B220 (clone RA3-6B2), anti-CR2/CR1 (clone 7G6; BD Biosciences) or anti-IgD (clone 11-26c) Abs, or biotinylated anti-CD3 (clone KT3), IgM (clone 331.12), and MOMA-1 Abs. Unlabeled Abs were detected with a monoclonal anti-rat Ab conjugated to HRP (clone G16-510E3; BD Biosciences) and biotinylated Abs with streptavidin-alkaline phosphatase (Southern Biotechnology Associates). Alkaline phosphatase was visualized with the Fast Blue kit (Vector Laboratories), with endogenous phosphatases blocked by the addition of 2 mM levamisole (Sigma-Aldrich). HRP was visualized with the 3-amino-9-ethylcarbazole substrate kit (Vector Laboratories). For lung sections, organs were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned and stained with H&E. Immunohistochemistry was performed as described, using Abs specific for CD45R, CD3, or F4/80 (Serotec) (29).

B cells purified from mouse spleen using anti-B220 MACS beads (5 × 10⁶/sample) were either untreated or stimulated for 4 h at 37°C with 10 μg/ml anti-CD40 Ab (clone FGK45). Cytoplasmic and nuclear fractions were prepared using the Qproteome cell compartment kit (Qiagen). Macrophages were derived from the bone marrow of mice by culture in L cell-conditioned medium for 7 days. Bone marrow macrophages were stimulated for 3 h with 1 μg/ml LPS followed by elution with Cell Dissociation buffer (Invitrogen Life Technologies). Lysates from equivalent numbers of cells were resolved by SDS-PAGE and immunoblotted using Abs specific for RelA (F-6), NF-κB1 (E10), or NF-κB2 (C5) (Santa Cruz Biotechnology), Apaf-1 (cytoplasmic) a gift from Dr. D. Huang (Walter and Eliza Hall Institute (Melbourne, Australia), and poly(ADP-ribose) polymerase (nuclear) (clone 42; BD Biosciences). Mouse embryonic fibroblast cell lines were generated from E13 embryos by trypsin dissociation of tissue, as described (30). For immunoprecipitation, mouse embryonic fibroblasts were stimulated for 15 h with 1 μg/ml agonist LTβR-specific Ab (AF.H6; BD Biosciences) and whole cell lysates (1 × 10⁶/sample) were prepared in lysis buffer (20% glycerol, 0.2 mM EDTA, 0.5% Nonidet P-40, 150 mM NaCl) and immunoprecipitated overnight at 4°C with 2 μg of anti-RelA Ab (F-6). SDS-PAGE and Western blotting were performed essentially as described (31). Band intensity on films was quantitated using Scion Image software.

The osteoclastogenic potential of murine bone marrow was determined in vitro as previously described (32). Briefly, 1 × 10⁵ bone marrow cells were incubated with 100 ng/ml RANKL and 25 ng/ml M-CSF, with change of medium and mediators at day 3. After 7 days, cells were fixed and stained histochemically for tartrate-resistant acid phosphatase. Tartrate-resistant acid phosphatase-positive cells with three or more nuclei were counted as osteoclasts. Histomorphometric analysis was conducted on L4 lumbar vertebrae as previously described (33).

Significant differences were determined by Student’s two-tailed t tests for independent events or one-way ANOVA followed (where significant) by Fisher’s protected least significant difference post hoc test to identify the significant pairwise differences (see Fig. 5, B–D).

To identify genes important for immune regulation, we conducted a forward genetic screen using the chemical mutagen ENU. A cohort of mutant mice was generated by crossing male ENU-treated BALB/c mice with BALB/c females. The resultant G₁ progeny were bled at 7 wk of age, and the different hemopoietic cell subsets were quantified using an automated blood analyzer.

A pedigree of mice (Lym1) was found in which approximately half the members exhibited an increased number of circulating lymphocytes (Fig. 1,A). By mating affected Lym1 mice to BALB/c wt mice, we confirmed that the lymphocytosis was inherited as a dominant trait (data not shown). Lym1 heterozygous mice were healthy and fertile. A comprehensive histopathological analysis of these mice, however, revealed an absence of peripheral LN (inguinal, brachial, axillary, cervical) and Peyer’s patches (Fig. 1 B and data not shown). Mesenteric LN were present, but were reduced in size and cellularity (data not shown).

The Lym1 mutation was genetically mapped by mating affected Lym1 mice on a BALB/c background with C57BL/6 wt mice and then intercrossing the F₁ progeny. DNA from both affected and unaffected F₂ mice was typed for a genome-wide panel of polymorphic simple sequence length polymorphism markers. The mutation was localized to a 6-Mb interval on chromosome 19 (Fig. 1 C). The Nfkb2 gene, located within this region, was considered a strong candidate as LN development is perturbed in Nfkb2^−/− mice (9, 10).

Sequence analysis of the 23 exons of Nfkb2 revealed a thymine to adenine conversion of base 2854 that was present only in affected Lym1 mice (Fig. 1,D). This base change is predicted to cause the substitution of a stop codon for Y868, truncating the protein in the C-terminal phosphorylation site (Fig. 1,E). This led to an NF-κB2 precursor protein with a predicted mass of 96 kDa, termed p100^Lym1 (Fig. 1,F). The Lym1 mutation introduced a Bsu 36I restriction enzyme recognition site, enabling Lym1 mice to be genotyped using RFLP analysis (Fig. 1 G).

Upon dissection, it was found that Nfkb2^Lym1/Lym1 mice lacked mesenteric LNs in addition to peripheral LNs and Peyer’s patches (data not shown), suggesting that the Lym1 mutation led to a more severe phenotype in the homozygous state. Nfkb2^Lym1/Lym1 mice also displayed reduced fertility compared with either wt or Nfkb2^Lym1/+ mice, with less frequent litters and smaller litter sizes (data not shown).

Disorganized splenic architecture is a hallmark of mouse models in which the noncanonical NF-κB pathway is disrupted, including Nfkb2^−/− mice (9, 10, 22). In spleen sections from Nfkb2^Lym1/Lym1 mice, T and B cells were not segregated into discrete compartments and the number of follicular dendritic cell clusters and marginal zone B cells were severely reduced (Fig. 2, A–C). In spleen from wt mice, marginal metallophillic macrophages, defined by staining with MOMA-1, formed a ring-like zone defining the marginal zone, but this structure was not evident in Nfkb2^Lym1/Lym1 spleen (Fig. 2 D). The abnormalities in splenic architecture in Nfkb2^Lym1/+ mice were intermediate between that of wt and Nfkb2^Lym1/Lym1 mice (data not shown). Thus, although peripheral blood lymphocytosis is inherited as a dominant trait, disruptions to LN and spleen development are inherited as semidominant phenotypes.

Histological analysis showed the presence of inflammatory cell infiltrates in the lung and liver of mice carrying the Lym1 mutation (Fig. 2 and data not shown). In Nfkb2^Lym1/Lym1 mice, these infiltrations were extensive with large foci composed of T and B cells, and macrophages (Fig. 2, E–H). Smaller foci of similar composition were present in the lung and liver of Nfkb2^Lym1/+ mice (data not shown).

Inflammatory foci are seen in other mouse models with perturbed noncanonical pathway signaling (11, 34, 35, 36), which may reflect inadequate self-tolerance due to reduced Aire expression and disorganized medullary thymic epithelial cell networks (13, 14, 36). Indeed, Aire expression was significantly reduced in the thymus of Nfkb2^Lym1/Lym1 mice, suggesting that the inflammatory lesions evident in the lung and liver of these mice are due to autoimmune processes (Fig. 2,I). There was a progressive increase in severity of the inflammation as Nfkb2^Lym1/Lym1 mice aged (data not shown), which likely contributed to their reduced lifespan (Fig. 2 J).

T cell development in the thymus of Nfkb2^Lym1 mice was analyzed by flow cytometry. Overall, the thymus was enlarged in young Nfkb2^Lym1/Lym1 mice compared with wt, but became smaller in size and cellularity compared with controls as the mice age and became moribund (Fig. 3,A and data not shown). The total number of CD8 single positive (SP) cells was unchanged in Nfkb2^Lym1/Lym1 thymus compared with wt controls (Fig. 3,A). Double positive, double negative, and CD4 SP cells, however, were significantly increased in number (Fig. 3,A). An increased proportion of the CD4 SP cells had the appearance of recent thymic emigrants (CD69⁻, CD62L⁺) (13, 37), suggesting that thymic export may be impaired in Nfkb2^Lym1/Lym1 mice (Fig. 3 A, bottom).

The blood of Nfkb2^Lym1/Lym1 mice contained a significantly increased number of CD4 and CD8 T cells compared with wt, and a small increase in B cell number, although this increase was not significant (Fig. 3,B). There was no major difference in the proportion of CD69⁻ CD62L⁺ peripheral blood or splenic CD4 cells between wt and Nfkb2^Lym1/Lym1 mice (Fig. 3 C and data not shown). There was no difference in the expression of CD25 between wt and Nfkb2^Lym1/Lym1 T cells in all organs tested, but the expression of CD44 was slightly reduced, suggesting that the T cells were more naive in Nfkb2^Lym1/Lym1 mice (data not shown).

T cells were represented in normal number in the spleen, but spleens were enlarged in Nfkb2^Lym1/Lym1 mice (Fig. 3,C). This splenomegaly did not correlate with a significant increase in total splenocyte number (Fig. 3 C) and became more pronounced as the mice age (data not shown). Histological staining showed an increase in erythropoiesis in the spleen of Nfkb2^Lym1/Lym1 mice (data not shown), suggesting that splenomegaly resulted from an expansion of the splenic red pulp.

We next analyzed B cell development in the bone marrow and spleen. The proportion and number of pro-B, pre-B, and immature B cells in the bone marrow was unchanged, suggesting that early B cell development is normal in Nfkb2^Lym1/Lym1mice (Fig. 4,A). The number of mature recirculating B cells was significantly reduced, however, in Nfkb2^Lym1/Lym1 bone marrow compared with wt (Fig. 4,A). The number of transitional stage 1 B cells was significantly increased and there was a profound reduction in the number of transitional stage 2 B cells, marginal zone, and follicular B cells in Nfkb2^Lym1/Lym1 spleens (Fig. 4,B). Measurement of serum Ig in resting Nfkb2^Lym1/Lym1 mice showed a significant reduction in the basal levels of IgM, IgG1, IgG2b, and IgA compared with wt (Fig. 4,C). Proliferation of purified B cells from Nfkb2^Lym1/Lym1 mice in response to CD40L, LPS, or BCR ligation was significantly reduced (Fig. 4,D). Induction of isotype switching and plasma cell differentiation in vitro in response to these same stimuli was also reduced in Nfkb2^Lym1/Lym1 B cells (data not shown). These defects are more likely a consequence of decreased B cell proliferation or survival (Fig. 4 E) rather than an intrinsic inability to differentiate, as Nfkb2^Lym1/Lym1 B cells were capable of differentiation in vitro, but at reduced frequency (data not shown). Similar defects in B cell development and function are characteristic of Nfkb2^−/− and aly mice (9, 10, 15, 38).

The noncanonical NF-κB pathway is known to be critical for the differentiation of osteoclasts via RANKL stimulation (24). To assess the effect of the Lym1 mutation on this process, we cultured the bone marrow of mutant mice in RANKL and M-CSF to stimulate osteoclast formation. The ability of Nfkb2^Lym1/+ cells to form osteoclasts in response to RANKL stimulation was significantly reduced, and Nfkb2^Lym1/Lym1 bone marrow was almost completely unresponsive to RANKL with osteoclast formation <0.5% of that seen with wt (Fig. 5,A). Consistent with this response, histomorphometric analysis of Nfkb2^Lym1/Lym1 bones showed a mild osteopetrosis, with significantly increased trabecular bone volume (Fig. 5,B) and number (Fig. 5,C) but no change in trabecular thickness (Fig. 5 D), indicating a corresponding defect in basal osteoclastogenesis in vivo. There was, however, no obvious defect in tooth eruption in Nfkb2^Lym1/Lym1 mice (data not shown).

The premature stop codon introduced by the Lym1 mutation truncates p100^Lym1 at Y868 in the C-terminal phosphorylation sequence (Fig. 1 D). Previous studies have shown that S866 and S870 comprise the docking site for the upstream kinase, IKKα (39), and S872 is phosphorylated by IKKα (39). Although S866 is present in p100^Lym1, both S870 and S872 are predicted to be absent, suggesting that p100^Lym1 is unable to be phosphorylated at the C terminus. Phosphorylation of p100 leads to recruitment of the E3 ubiquitin ligase β transducin repeat-containing protein βTrCP (40), which transfers ubiquitin to K855, thus targeting the precursor p100 form of NF-κB2 for processing to p52 by the proteasome. Thus, it appeared likely that processing of the NF-κB2 precursor protein would be perturbed in mice carrying the Lym1 mutation.

To test this likelihood, we stimulated purified splenic B cells with anti-CD40 Ab. Cytoplasmic and nuclear fractions were prepared and Western blotted to analyze the activation and nuclear translocation of NF-κB proteins. Both p100 and p52 were detectable in cytoplasmic extracts from untreated and stimulated wt B cells (Fig. 6,A). Cytoplasmic p100 levels were reduced after stimulation, suggesting that p100 was processed into p52, which migrated to the nucleus (Fig. 6,A, bottom). In Nfkb2^Lym1/+ cells, little, if any, p52 was detectable in lysates from either untreated or stimulated cells and there was no apparent reduction in precursor levels upon stimulation. In Nfkb2^Lym1/Lym1 cells, however, p100^Lym1 accumulated in response to anti-CD40 stimulation, and no p52 was observed in either the presence or absence of stimulation, even with longer film exposures (Fig. 6 A and data not shown).

Immunoblotting of nuclear extracts from wt B cells showed both p52 and RelA were recruited to the nucleus following stimulation (Fig. 6,B). In contrast, only minor increases in nuclear p52 or RelA were seen in Nfkb2^Lym1/+ cells following stimulation (Fig. 6,B). In resting mouse embryonic fibroblasts, p100 associated with RelA in both wt and Nfkb2^Lym1/Lym1 cells (Fig. 6,C). After stimulation of wt cells with an agonist Ab specific for LTβR, this association was substantially reduced, most likely as p100 was processed to p52. In contrast, the amount of p100^Lym1 associated with RelA in Nfkb2^Lym1/Lym1 cells increased upon stimulation (Fig. 6,C). Similarly, nuclear transport of RelA and p50 in response to LPS stimulation through the canonical NF-κB pathway appeared to be inhibited in Nfkb2^Lym1/Lym1 cells, as cytoplasmic RelA levels failed to decrease and p50 levels accumulated in response to LPS stimulation, in contrast to wt cells (Fig. 6 D).

Thus, in resting wt and Nfkb2^Lym1/Lym1 cells, p100 appears to bind RelA and inhibit its activation and nuclear translocation. In wt cells, this inhibition is relieved upon stimulation through the noncanonical NF-κB pathway as p100 is processed to p52. The inability to process p100^Lym1 in Nfkb2^Lym1/Lym1 cells, however, results in a “super repressor” form of p100 that not only prevents the formation of p52, but also continually inhibits RelA activation and nuclear translocation, regardless of stimulation with ligands that activate either the canonical or noncanonical NF-κB pathway.

NF-κB2 is a critical mediator of the noncanonical NF-κB pathway. The importance of NF-κB2 in the development and function of a variety of organs and cell lineages has been highlighted by the study of genetically modified mice in which NF-κB2 function has been ablated or modified (9, 10, 18, 21, 22, 24). In this study, we describe the phenotype of mice harboring a novel mutation in Nfkb2, which encodes a nonprocessible form of p100. We have shown that the abnormalities in Nfkb2^Lym1/Lym1 mice are more severe than those seen in mice completely lacking NF-κB2, indicating a role for NF-κB2 in the regulation of both the canonical and noncanonical NF-κB pathways.

Nfkb2^Lym1/Lym1 mice develop inflammatory lesions in the liver and lung that are reminiscent of those observed in mice deficient in RelB, LTβR, or functional NIK (11, 34, 41). We have shown that Aire expression is significantly reduced in the thymus of Nfkb2^Lym1/Lym1 mice, suggesting that a breakdown in peripheral tolerance is a contributing factor. A recent study has shown that RANK signals regulate the development of Aire-expressing medullary thymic epithelial cells, a process that is likely to be disrupted in Nfkb2^Lym1/Lym1 mice (14). Similar to other mice with defective NF-κB signaling (13), we noticed that the thymic medulla and cortex of Nfkb2^Lym1/Lym1 mice are poorly defined (data not shown), indicating that thymic structure is disorganized.

The differentiation of thymocytes is also perturbed, with an increased number of double negative, double positive, and particularly CD4 SP cells, present in the thymus. Despite having the appearance of recent thymic emigrants, suggesting a blockade in thymic emigration as suggested for Ltrb^−/− and aly mice (13), this expansion of CD4 T cells was also observed in peripheral blood, and together with an increase in peripheral CD8 T cells, appears to account for the lymphocytosis evident in the blood of Nfkb2^Lym1/Lym1 mice. Peripheral blood lymphocytosis has not been widely reported in other mouse models with abnormal NF-κB signaling, and it is unclear what is driving this defect, although it could reflect hyperactivity of Nfkb2^Lym1/Lym1 naive CD4 T cells in response to TCR stimulation, as was reported recently (26). Lymphocytosis did not appear to contribute to the development of splenomegaly in Nfkb2^Lym1/Lym1 mice, however, and similar to Relb^−/− mice, this appears to be due to increased erythropoiesis (34).

Similar to Nfkb2^−/− and aly mice, Nfkb2^Lym1/Lym1 mice have disrupted splenic architecture and B cell differentiation (9, 10). Unlike the defect in B cell differentiation in Nfkb2^−/− mice, which is first apparent after the transitional stage 1 B cell maturation in the spleen (15), there is a significant increase in the number of splenic transitional stage 1 B cells in Nfkb2^Lym1/Lym1 mice, suggesting that B cell differentiation may be blocked at this stage. The extent of B cell deficiency of more mature subsets, however, is similar between Nfkb2^−/− and Nfkb2^Lym1/Lym1 mice (15). The appearance of these phenotypes in Nfkb2^−/− mice suggests that p52 plays a critical and nonredundant role in the maintenance of splenic architecture and B cell differentiation, and that other NF-κB proteins cannot compensate for the loss of p52. In contrast, osteoclastogenesis and LN formation are only mildly perturbed in Nfkb2^−/− mice (16, 17, 24), suggesting that p50:RelA activation may contribute to these processes. In the presence of the nonprocessible NFkB2 allele in Nfkb2^Lym1/Lym1 mice, RelA activation is limited, reducing the extent of NF-κB signaling to below the threshold necessary for LN formation and osteoclastogenesis and hence abnormalities result. Alternatively, retention of RelB in the cytoplasm by p100^Lym1 may contribute to abnormalities in LN development because Relb^−/− mice lack all LN (34).

The phenotype of Nfkb2^Lym1/Lym1 mice resembles that of mice in which loss of NIK function prevents NF-κB2 processing (aly or Nik^−/− mice) (21, 22). The similarities between Nik^−/− and Nfkb2^Lym1/Lym1 mice suggest that NIK function is largely limited to the induction of p100 processing. Some differences, however, are seen between these mouse models. Although RANKL-stimulated osteoclastogenesis in vitro is impaired in both models, basal osteoclastogenesis in vivo is normal in Nik^−/− mice, but is significantly perturbed in Nfkb2^Lym1/Lym1 mice (24). The cause of this discrepancy is unclear.

Nfkb2^Lym1/Lym1 mice also present some similarities with Nfkb1^−/−/ Nfkb2^−/− mice. Although partial defects in LN formation are seen in Nfkb2^−/− mice, all LNs are absent in Lta^−/−, Ltbr^−/−, Nik^−/−, and Nfkb1^−/−/Nfkb2^−/− mice (11, 17, 22, 23). Similarly, Nfkb1^−/−/Nfkb2^−/− mice have severe osteopetrosis, whereas single knockout mice have no obvious bone defects (42), providing further evidence that p100 regulates the activation of p50:RelA complexes. Indeed, p100 has been shown to interact with p50:RelA and regulate RelA activation in a number of previous studies (24, 25, 26, 43, 44). Consistent with this finding, we have shown that p100 associates with RelA in resting cells, preventing its nuclear translocation and activation. Stimulation through the noncanonical NF-κB pathway relieves this inhibition by processing p100 to p52 and thus releasing RelA.

The defects in Nfkb2^Lym1/Lym1 mice, however, are not as severe as in Nfkb1^−/−/Nfkb2^−/− mice, which die shortly after weaning and lack all mature B cells (42, 45). Similarly, Nfkb2^Lym1/Lym1 mice present a milder phenotype than either Rela^−/− (embryonic lethal) or Relb^−/− (multifocal inflammatory disease within 2 wk of birth) mice (34, 46), suggesting there is only a partial block in RelA and RelB signaling in Nfkb2^Lym1/Lym1 mice. The increased severity of the Nfkb2^Lym1/Lym1 phenotype compared with Nfkb2^−/− mice likely reflects the constitutive inhibitory effect of p100^Lym1 on canonical pathway signaling. Activation of canonical pathway signaling induces p100 expression, leading to accumulation of p100^Lym1 and retention of both canonical and noncanonical binding partners in the cytoplasm. Given the similarities in phenotype between Nfkb2^Lym1/Lym1 and Relb^−/− mice, cytoplasmic sequestration of RelB by p100^Lym1 is also likely to contribute to the observed defects (34).

Collectively, our analyses indicate that expression of a super-repressor form of p100 in Nfkb2^Lym1/Lym1 mice both prevents p52 generation and limits the activation of NF-κB complexes in response to both canonical and noncanonical stimuli. This mouse model, therefore, represents a unique tool for the study of cross-talk and compensation between NF-κB signaling pathways.

We thank Jason Corbin, Ailsa Frew, Lei Shong Lau, Ladina Di Rago, Ankita Goradia, and Hayley Croom for first-class research assistance, and Melanie Rowe, Kylie Gilbert and Erin Salt for expert animal husbandry.

We disclose that this work was supported in part by Murigen Pty Ltd for whom W.S.A., B.T.K., and D.J.H. consult and/or have a financial interest. The remaining authors have no financial conflict of interest.