Activation by IKKα of a Second, Evolutionary Conserved, NF-κB Signaling Pathway

See allHide authors and affiliations

Science  24 Aug 2001:
Vol. 293, Issue 5534, pp. 1495-1499
DOI: 10.1126/science.1062677


In mammals, the canonical nuclear factor κB (NF-κB) signaling pathway activated in response to infections is based on degradation of IκB inhibitors. This pathway depends on the IκB kinase (IKK), which contains two catalytic subunits, IKKα and IKKβ. IKKβ is essential for inducible IκB phosphorylation and degradation, whereas IKKα is not. Here we show that IKKα is required for B cell maturation, formation of secondary lymphoid organs, increased expression of certain NF-κB target genes, and processing of the NF-κB2 (p100) precursor. IKKα preferentially phosphorylates NF-κB2, and this activity requires its phosphorylation by upstream kinases, one of which may be NF-κB–inducing kinase (NIK). IKKα is therefore a pivotal component of a second NF-κB activation pathway based on regulated NF-κB2 processing rather than IκB degradation.

Mammals express five NF-κB transcription factors: RelA, RelB, c-Rel, NF-κB1, and NF-κB2 (1). Unlike the Rel proteins, NF-κB1 and NF-κB2 are synthesized as large precursors (p105 and p100, respectively) that require proteolytic processing to produce their respective p50 and p52 NF-κB subunits (1). Mature NF-κB dimers are kept in the cytoplasm through interaction with inhibitory IκB proteins, and the major pathway leading to their activation is based on inducible IκB degradation (1, 2). This canonical pathway, triggered by proinflammatory cytokines, microbes, and viruses, requires activation of the IKK complex (2). Because the NF-κB1 and NF-κB2 precursors contain IκB-like ankyrin repeats in their COOH-termini, they can function as IκBs (3, 4). Unlike IκB degradation, processing of NF-κB1 is a constitutive process (5, 6). NF-κB2 processing, however, could be a regulated process because it is most active in mature B cell lines (7) and is defective in aly mice (8). The aly mutation, which maps to the gene encoding NIK, interferes with the development of primary and secondary lymphoid organs (9), as does a complete NIK deficiency (10). Interestingly, NIK induces ubiquitin-dependent processing of NF-κB2 (11) but is not required for induction of NF-κB DNA binding activity (12).

NIK was discovered as an NF-κB–activating kinase (12) and was later shown to phosphorylate and activate IKKα (13), one of the two catalytic subunits of the IKK complex (2). The other catalytic subunit, IKKβ, is 52% identical to IKKα (2), and in vitro both subunits exhibit IκB kinase activity (14). Despite these similarities, IKKα and IKKβ have distinct functions (2,5). IKKβ is essential for proper activation of NF-κB in response to proinflammatory stimuli and for prevention of tumor necrosis factor (TNF-α)–induced apoptosis (15–18), whereas IKKα is dispensable for IKK activation and induction of NF-κB DNA binding activity in most cell types (17, 19). IKKα, but not IKKβ, is essential for proper skeletal morphogenesis and differentiation of the epidermis (19,20). However, this function does not depend on IKK activity or NF-κB activation (21). These findings raise the question of whether IKKα has any NF-κB–related functions that are masked by the perinatal lethality of Ikkα–/–mice. Here, we provide evidence that IKKα kinase activity is required for B cell maturation, formation of secondary lymphoid organs, induction of a subset of NF-κB target genes, and inducible NF-κB2 processing. This function of IKKα is strikingly similar to that ofDrosophila IKK, which is required for processing of Relish, a NF-κB2–like precursor protein (22, 23). In addition to explaining the function of IKKα, these results shed new light on the mechanisms involved in the evolution of innate and adaptive immunity.

Analysis of bone marrow cells from wild-type,Ikkα–/–, andIkkβ–/– radiation chimeras (24) revealed complete absence of B cells inIkkβ–/––derived samples (25). By contrast, B cells were present in Ikkα–/–reconstituted bone marrow (25). Although these cells expressed normal levels of early B cell markers, a B220hiCD24lo population, representing circulating mature B cells, was absent (25). No differences in absolute numbers of thymocytes and peripheral T cell populations were found between wild-type and Ikkα–/–radiation chimeras (25). This is in marked contrast toIkkβ–/– radiation chimeras, which lack B and T cells (18). Analysis of B cell markers inIkkα–/– reconstituted spleen and lymph nodes revealed a reduction of the mature IgMloIgDhipopulation (Ig, immunoglobulin) relative to virgin IgMhi B cells (Fig. 1A). Whereas the splenic B to T cell ratio was almost normal 6 weeks after reconstitution withIkkα–/– stem cells, it was considerably reduced thereafter, mostly as a result of the loss of mature IgMloIgDhi B cells (25). This defect in B lymphopoiesis is cell-autonomous (25).

Figure 1

Defective B cell maturation and GC formation in Ikkα–/– radiation chimeras. (A) Splenocytes (SP) and lymph node cells (LN) of wild-type (Ikkα+/+) andIkkα–/– radiation chimeras were analyzed by flow cytometry 6 weeks after engraftment for expression of B (B220, IgM, IgD) and T (Thy1.2) cell markers. Percentages of positive cells in indicated regions are shown. Dot plots are representative of six independent experiments. (B) The turnover of mature (IgMloIgDhi) B cells was determined by means of BrdU incorporation. Eight weeks after engraftment, mice were administered BrdU and splenocytes (SP) were analyzed by flow cytometry after three-color staining for BrdU, IgM, and IgD. Left panel: BrdU incorporation into mature splenic B cells ofIkkα+/+ (gray) andIkkα–/– (black) radiation chimeras after 7 days of labeling. Right panel: Kinetics of BrdU incorporation over a 7-day period. Values represent means ± SD for three mice in each group and time point. (C) Frequencies of mature IgD+ lymph node B cells in Ikkα+/+and Ikkα–/– radiation chimeras (n = 5 each group, mean ± SD). (D) CD43 resting splenic B cells were purified and cultured without stimulation for 72 hours. At 24-hour intervals, percentages of apoptotic cells were determined by 7-amino–actinomycin D staining. Initially, more than 97% of the cells were viable (n = 3 each group, mean ± SD). (E)Ikkα–/– and wild-type radiation chimeras were injected with DNP-KLH. After 12 days, their spleens were stained with PNA and counterstained with hematoxylin. Almost no GC formation could be detected in Ikkα–/– reconstituted spleens.

Examination of B cell turnover by means of bromodeoxyuridine (BrdU) labeling (26) revealed thatIkkα–/– B cells incorporated more BrdU than did wild-type B cells (Fig. 1B). The high turnover ofIkkα–/– B cells is likely to account for the lower fraction of circulating mature B cells (Fig. 1A) and the lower frequency of mature IgD+ B cells inIkkα–/– lymph nodes (Fig. 1C). The increased turnover of Ikkα–/– B cells correlates with higher rates of spontaneous apoptosis, seen in vitro (Fig. 1D) and in vivo (25). Ikkα–/– B cells also exhibit defective mitogenic responses to antibody to IgM and especially to lipopolysaccharide (LPS) (25). To test whether immune responses dependent on cellular interactions are functional, we immunized mice with dinitrophenol–keyhole limpet hemocyanin (DNP-KLH), and 12 days later we evaluated the formation of germinal centers (GCs) (27). Wild-type spleens exhibited numerous GCs characterized by B cell areas that bound peanut agglutinin (PNA). In contrast, the spleens of Ikkα–/– radiation chimeras contained very few PNA-stainable cells (Fig. 1E).

The absence of IKKα can be compensated by IKKβ in liver and keratinocytes, leading to normal IKK and NF-κB activation by proinflammatory stimuli (19, 21). Biochemical analysis of purified resting and stimulated B cells from radiation chimeras yielded similar results. Basal IKK activity was elevated inIkkα–/– B cells (Fig. 2A). This was associated with lower IκBα levels and somewhat higher basal NF-κB DNA binding activity (Fig. 2B). However, no major differences in LPS-induced IKK and NF-κB DNA binding activities were detected betweenIkkα+/+ and Ikkα–/–B cells (Fig. 2, A and B). These results were highly reproducible and were seen upon analysis of at least five different pairs of radiation chimeras (25). We also compared the expression of individual NF-κB proteins and found similar amounts of RelA, c-Rel, RelB, and NF-κB1 p105 and p50 (Fig. 2, C and D). Strikingly, however,Ikkα–/– B cells exhibited defective processing of NF-κB2 and contained very little p52 and increased amounts of p100 (Fig. 2D). These defects were observed in bone marrow, spleen, and lymph node–derived B cells, regardless of their NF-κB2 expression level.

Figure 2

Normal NF-κB DNA binding but defective NF-κB2 processing in Ikkα–/– radiation chimeras. (A) Purified splenic B cells were stimulated with LPS (5 μg/ml). IKK activity and IκBα levels were determined at 0, 60, and 90 min, as indicated. IKK recovery was determined by immunoblotting (IB) with antibody to IKKγ. (B) NF-κB and NF-1 DNA binding activities in purified splenic B cells treated with LPS (5 μg/ml) were determined at 0, 60, and 90 min. Data in (A) and (B) are representative of at least five independent experiments; the average relative levels (rl) of NF-κB to NF-1 DNA binding activities in these experiments are indicated at the bottom of (B). (C) Expression of the indicated NF-κB proteins, NIK, and IKK subunits was examined by immunoblotting. Three individual wild-type and Ikkα–/– radiation chimeras were analyzed. (D) Splenic, lymph node (LN), and bone marrow (BM) B cells from groups of three Ikkα+/+ (wt) and three Ikkα–/––/–) radiation chimeras were pooled and analyzed by immunoblotting for expression of NF-κB1 (p105 and p50), NF-κB2 (p100 and p52), and RelA. Migration positions of molecular weight standards (in kilodaltons) are indicated at the right. (E) Splenic B cells were analyzed for p52-containing NF-κB complexes. Nuclear extracts from B cells pooled from three different wild-type andIkkα–/– radiation chimeras were incubated with a palindromic NF-κB binding site in the presence of either anti-p52 (06-413, Upstate Biotechnology) or a control antiserum that does not recognize p52. Extract quality was monitored by binding to an NF-1 probe. SS, supershifted p52-containing complex.

Having found defective NF-κB2 processing inIkkα–/– B cells, we examined whether NF-κB complexes in these cells were indeed deficient in p52. To optimize the detection of p52-containing NF-κB complexes, we used a palindromic NF-κB binding site (1) and an antibody capable of binding to such complexes. This analysis revealed smaller amounts of p52-containing NF-κB complexes in Ikkα–/–B cells (Fig. 2E).

To determine the role of IKKα phosphorylation by upstream kinases, such as NIK [which is also required for B cell maturation, GC formation, and NF-κB processing (8)], we generated IkkαAA knock-in mice in which the activating phosphorylation sites of IKKα, Ser176 and Ser180, were replaced by alanines (28). Although IkkαAA mice are viable, we wished to analyze the effect of this mutation on B cell development under the same conditions as theIkkα–/– null mutation. We therefore generated IkkαAA radiation chimeras and found a reduction in mature IgD+ B cells in their spleens and lymph nodes (Fig. 3, A and B). However, a more pronounced deficiency of mature B cells was detected in untreated 12-week-old IkkαAA mice (Fig. 3, A and B). In addition, untreated IkkαAA mice lacked distinct Peyer's patches and exhibited defective GC formation after immunization with DNP-KLH (Fig. 3C). Defects found inIkkαAA radiation chimeras were less severe (25). Untreated IkkαAA mice also exhibited defective NF-κB2 processing in purified splenic and lymph node B cells, as well as in bone marrow B cells, which express smaller amounts of NF-κB2 (Fig. 3D). The reduction in NF-κB2 processing (∼80%) was less severe than that caused by the complete IKKα deficiency. Although basal and LPS-induced NF-κB DNA binding activities were only slightly reduced inIkkαAA cells (Fig. 3E), the relative amount of p52-containing NF-κB complexes was substantially decreased (Fig. 3F).

Figure 3

Requirement of IKKα phosphorylation for B cell maturation and NF-κB2 processing. (A) The ratio of virgin (IgMhi) to mature (IgMloIgDhi) splenic B cells in untreated wild-type (wt) and IkkαAA (AA) mice and radiation chimeras was determined by flow cytometry. Data are means ± SEM for three animals in each group. (B) Relative frequencies of mature IgD+ lymph node B cells in untreated wild-type and IkkαAA mice and radiation chimeras. (C) Peyer's patches (top panels) and GCs (bottom panels) in wild-type and IkkαAAmice. Mice were immunized with DNP-KLH and examined 12 days later to assess GC formation. Peyer's patches (arrows) were visualized by staining with antibody to VCAM1. (D) Immunoblot analysis of NF-κB2 processing in purified B cells from spleen, lymph nodes, and bone marrow of untreated wild-type and IkkαAAmice. The ratio of p52 to p100 is indicated at the bottom. IKKα levels were used to verify equal loading. Data are representative of three independent experiments. (E) Splenic B cells fromIkkα+/+ and IkkαAAmice were stimulated with LPS, and NF-κB and NF-1 DNA binding activities were determined. Average relative levels (rl) of NF-κB to NF-1 DNA binding activities in five such experiments are indicated at the bottom. (F) The content of p52-containing NF-κB complexes in splenic B cells was determined as described in (E).

We also examined whether IkkαAA B cells exhibit defective NF-κB–mediated gene induction. Splenic B cells were isolated from wild-type and IkkαAAmice that were or were not injected with LPS, and expression of known NF-κB target genes was analyzed by real-time polymerase chain reaction (PCR) (29). Transcription of several NF-κB target genes, including those encoding cyclin D2 and TNF-α, was similarly increased in wild-type and IkkαAA B cells, but expression of other target genes—including those for macrophage inflammatory protein (MIP)–1α and receptor activator of NF-κB (RANK) ligand—was clearly defective inIkkαAA cells (Table 1). Thus, IKKα phosphorylation is also required for increased expression of a subset of NF-κB target genes in B cells.

Table 1

Analysis of NF-κB target gene expression in wild-type and IkkαAA B cells. Wild-type andIkkαAA mice were injected with LPS (5 mg/kg ip) or phosphate-buffered saline. After 1 or 4 hours, splenic B cells were isolated and their RNA was extracted. Expression of the indicated NF-κB target genes was analyzed by RealTime PCR (Taq Man, PE Applied Biosystems) and normalized to the level of cyclophilin mRNA (29). Reverse transcription was done with 2 μg of total RNA, followed by 40 PCR cycles at 95°C for 15 s and 60°C for 1 min (Sybr Green Core Reagents, PE Applied Biosystems). Primer sequences are available upon request. The values represent change in mRNA abundance relative to the untreated sample of each genotype and are averages of two fully separate experiments. iNOS, inducible nitric oxide synthase.

View this table:

In addition to autophosphorylation by IKK, Ser176 and Ser180 of IKKα can be phosphorylated by NIK (13). Like IKKα, NIK is required for NF-κB2 processing (8, 11). Although NIK overexpression was shown to stimulate NF-κB2 phosphorylation (11), the results described above suggested that NIK may act via IKKα. Currently, the only known activator of NIK is the lymphotoxin β receptor (LTβR), which is not expressed in B cells (10). We therefore resorted to a surrogate system based on cotransfection of NIK and NF-κB2 expression vectors into wild-type, Ikkα–/–, andIkkβ–/– mouse fibroblasts. NIK induced NF-κB2 processing in wild-type and Ikkβ–/–cells, but not in Ikkα–/– cells (Fig. 4A). However, reexpression of IKKα inIkkα–/– cells restored NIK's ability to induce NF-κB2 processing.

Figure 4

Involvement of IKKα in NIK-induced NF-κB2 processing and NF-κB2 COOH-terminal phosphorylation. (A) Mouse fibroblasts derived from wild-type,Ikkα–/–, andIkkβ–/– embryos were transfected with an NF-κB2 p100 expression vector with or without NIK and IKKα expression vectors. After 36 hours, the processing of NF-κB2 was examined by immunoblotting with an NH2-terminal–specific p100 antibody. (B) Chromatographically pure IKKα and IKKβ (50 ng each) and immunopurified NIK, all produced in Sf9 cells using baculoviruses, were examined for phosphorylation of GST-IκBα(1-54) and GST-NF-κB2(754-900)(p100C) fusion proteins (11,14) as well as for autophosphorylation.

We also compared the abilities of recombinant NIK, IKKα, or IKKβ (14) to phosphorylate the regulatory domains of IκBα and NF-κB2. IκBα was phosphorylated much more efficiently by IKKβ than by IKKα and not at all by NIK (Fig. 4B). The COOH-terminal regulatory domain of NF-κB2, however, was phosphorylated most efficiently by IKKα. This clear difference in substrate specificities between IKKα and IKKβ correlates with their different biological functions.

Previous experiments established the role of IKKβ in inducible IκB degradation and activation of NF-κB in response to infections and proinflammatory stimuli (15–18), but the role of IKKα in this process has remained enigmatic. Although not essential for the canonical NF-κB activation pathway, IKKα is required for proper patterning of the epidermis (19, 20), but this function is not mediated by IκB or NF-κB and does not require the kinase activity of IKKα (21). Our results demonstrate that IKKα is physiologically involved in NF-κB regulation, but instead of doing so through inducible IκB degradation, it exerts at least some of its NF-κB–related functions by regulating the processing of NF-κB2. Defective NF-κB2 processing inIkkα–/– or IkkαAA B cells is likely to interfere with their maturation and with the formation of secondary lymphoid organs. A similar defect in B cell maturation was recently described by Kaisho et al. (30), who reconstituted Rag2 –/–mice with Ikkα–/– stem cells. These authors attributed the defect to decreased NF-κB DNA binding activity inIkkα–/– splenic B cells (30). However, we found a slight reduction in LPS-induced NF-κB DNA binding activity in IkkαAA but not inIkkα–/– B cells. Similarly, the absence or inactivation of NIK, which may act upstream to IKKα, does not affect total NF-κB DNA binding activity (10) but does inhibit NF-κB2 processing (8, 11). Both the absence of NIK (10) and the IkkαAA mutation reduce the expression of certain NF-κB target genes.

Expression of NF-κB2 itself is increased in more mature B cell lines (7) and is up-regulated during B cell development (Figs. 2 and 3). However, no defect in NF-κB2 expression, other than its processing, was detected in IKKα-deficient B cells, regardless of their maturation state. Hence, the processing defect is not the consequence of defective B cell maturation. Although the complete knockout of the Nfκb2 gene also results in lymphoid organ defects, including the absence of GCs (31), the effects are not identical to those of the Ikkα mutations, which exert a more severe effect on B cell development. However, the Nfκb2 –/– mutation abolishes expression of both p100 and p52, whereas theIkkα mutations reduce p52 and increase p100 expression. Congruently, the specific ablation of p100 in the presence of p52 expression results in increased lymphocyte proliferation and enlargement of spleen and lymph nodes (32).

NIK overexpression was shown to enhance NF-κB2 phosphorylation and processing (11). Our results strongly suggest that NIK acts via IKKα. First, theIkkαAA mutation, which replaces the NIK phosphorylation sites of IKKα with alanines, inhibits NF-κB2 processing, B cell maturation, and formation of secondary lymphoid organs. Second, NIK fails to stimulate NF-κB2 processing in the absence of IKKα. Third, IKKα is more potent than NIK as a NF-κB2 COOH-terminal kinase. At this point it is not clear whether both NIK and IKKα phosphorylate the COOH-terminal regulatory domain of NF-κB2, or whether this activity is provided by IKKα alone, whose activation is NIK-dependent. NIK itself is thought to be activated by LTβR, a member of the TNF receptor family (9,10). Indeed, Ltβr –/– mice exhibit defects similar to those of aly mice, which encompass those caused by the IkkαAA mutation (33). However, LTβR is expressed in the stroma but not in lymphoid cells, and therefore it cannot activate the NIK–IKKα–NF-κB2 pathway in lymphoid cells. Adoptive transfer experiments suggest that IKKα phosphorylation, like expression of NIK (8) or NF-κB2 (31), is also required outside the B cell compartment for formation of secondary lymphoid organs.

Our findings illustrate a novel function for IKKα that depends on its protein kinase activity and cannot be compensated by the related IKKβ subunit. Although both IKK catalytic subunits are involved in the activation of NF-κB transcription factors, they do so via different mechanisms and substrates. IKKβ is the canonical activator of NF-κB in response to infection and inflammation, and IKKα is responsible for activation of a specific NF-κB factor required for B cell maturation and formation of secondary lymphoid organs. This function is exerted through processing of NF-κB2 and is remarkably similar to the function of the Drosophila IKK complex, which contains a single catalytic subunit that is similar to both IKKα and IKKβ (22, 23). The DmIKK/Ird5 protein does not phosphorylate the single IκB of Drosophila, Cactus, and instead leads to activation of antibacterial genes through phosphorylation-induced processing of theDrosophila NF-κB1/2 homolog, Relish (22, 23). Although in Drosophila the processing-dependent NF-κB pathway is the major provider of innate antibacterial immunity (23), in mammals this pathway has been assigned to a specific aspect of adaptive immunity: B cell maturation and formation of secondary lymphoid organs. Thus, the duplication of IKK catalytic subunits and their functional divergence correlates with the evolution of the immune system.


View Abstract

Navigate This Article