Lymphotoxin β Receptor Induces Sequential Activation of Distinct NF-κB Factors via Separate Signaling Pathways*

Lymphotoxin β receptor (LTβR)-induced activation of NF-κB in mouse embryo fibroblasts was mediated by the classical pathway and by an alternative or second pathway. The classical pathway involved the IκB kinase (IKK)β- and IKKγ-dependent degradation of IκBα and resulted in the rapid but transient activation of primarily RelA-containing NF-κB dimers. The alternative or second pathway proceeded via NF-κB-inducing kinase (NIK)-, IKKα-, and protein synthesis-dependent processing of the inhibitory NF-κB2 p100 precursor protein to the p52 form and resulted in a delayed but sustained activation of primarily RelB-containing NF-κB dimers. This second pathway was independent of the classical IKK complex, which is governed by its central IKKγ regulatory subunit. The sequential engagement of two distinct pathways, coupled with the negative feedback inhibition of RelA complexes by NF-κB-induced resynthesis of IκBα, resulted in a pronounced temporal change in the nature of the NF-κB activity during the course of stimulation. Initially dominant RelA complexes were replaced with time by RelB complexes. Therefore, the alternative activation path mediated by processing of p100 was necessary for sustained NF-κB activity in mouse embryo fibroblasts in response to LTβR stimulation. Based on the phenotype of mice deficient in various components of the LTβR-induced activation of p100 processing, we conclude that this pathway is critically involved in the function of stromal cells during the generation of secondary lymphoid organ microarchitectures.

the exact nature of the initial innate response, NF-B factors then help to develop the appropriate adaptive responses by lymphocytes. In the final phase of the immune response, NF-B factors have important roles during the expansion and differentiation of lymphocytes involved in the adaptive response. Beyond the innate and adaptive antipathogenic responses, NF-B factors are also essential during development and maintenance of lymphoid organ structures (6,7), and they make important contributions during the development of hematopoietic cells, including B cells and osteoclasts (8,9). To carry out its diverse physiologic roles, NF-B factors not only help to induce expression of various factors and effectors, but depending on the cellular context, they also transcriptionally induce proteins that function to protect cells from apoptosis and that help to stimulate proliferation (1,2,5,10).
NF-B is a collective term for a family of dimeric complexes comprised of combinations of five polypeptides, RelA, c-Rel, RelB, p50/NF-B1, and p52/NF-B2. p50 and p52 are the Nterminal parts of the longer p105/NF-B1 and p100/NF-B2 proteins, respectively, and they are generated by proteolytic processing (1,2,4,5). High levels of p50 are produced constitutively by a cotranslational mechanism. In contrast, usually only small amounts of p52 exist in cells, but higher amounts may be induced by select signals.
To activate NF-B, appropriate environmental signals must bring about the release of NF-B dimers from their bound cytoplasmic inhibitors, in particular from the prototypical inhibitor IB␣ and its close relatives, IB␤ and IB (1,2,4,5). NF-B factors are in addition subject to various direct and indirect mechanisms that modulate their ability to stimulate transcription, dependent also on promoter context (1,2,5), but the release from the inhibitors is a first and necessary step in the activation process. Most of the NF-B activation signals, and in particular inflammatory cytokines, such as TNF␣ and IL-1, induce the phosphorylation of the IBs followed by the rapid ubiquitin-and proteasome-mediated degradation of the inhibitors, thus freeing NF-B dimers to migrate to the nucleus to initiate gene transcription (1,2,4,5). IBs are phosphorylated on two conserved serines by the IB kinase (IKK) complex. IKKs consist of the catalytic subunits, IKK␣ and IKK␤, and the regulatory subunit IKK␥ (also known as Nemo).
Most signals have been shown to activate NF-B by the classical, IKK-dependent pathway and, in particular, to be dependent on the IKK␤ catalytic and IKK␥/Nemo regulatory subunit to bring about the degradation of small IB inhibitors (1,2,5,11). In addition to the small IBs, the long forms of the NF-B1 and NF-B2 proteins, p105 and p100, can also act as cytoplasmic inhibitors of bound Rel proteins due to the presence of IB-like inhibitory ankyrin domains in their C-terminal halves (1,2,4,5). p105 may be completely degraded in response to some signals in a manner similar to that of small IBs, including IKK␤/IKK␥-induced phosphorylation of two serines * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In the present report, we demonstrate that physiologic signaling via the lymphotoxin ␤ receptor (LT␤R) in stromal cells induced the degradation of IB␣ via the classical pathway, and it induced processing of p100 via an alternative pathway. p100 processing was shown to be dependent on NIK and IKK␣ but independent of IKK␤ and IKK␥/Nemo. Therefore, the p100 processing pathway was entirely independent of the IKK complex, not just of the IKK␤ kinase subunit. We also demonstrate that transient activation of the classical pathway caused the transient activation of p50-RelA dimers, whereas the delayed and protein synthesis-dependent p100 processing led to the delayed and sustained liberation of p50-RelB and p52-RelB complexes. We also provide an explanation and supporting evidence for how p100 processing liberated p50-RelB complexes.

EXPERIMENTAL PROCEDURES
Cell Culture-IKK␣ Ϫ/Ϫ and IKK␤ Ϫ/Ϫ mouse embryonic fibroblasts (MEFs) were kindly provided by Drs. Q. Li and I. M. Verma, and IKK␥ Ϫ/Ϫ MEFs were kindly provided by Drs. M. Pasparakis and K. Rajewsky. NF-B1 Ϫ/Ϫ and NF-B2 Ϫ/Ϫ MEFs were kindly provided by Dr. E. Claudio. To prepare embryonic fibroblasts from wild-type and aly/aly mice, 12-day-old embryos were dissected, heads and inner organs were removed, and remaining parts were minced, filtered, and subjected to trypsin (0.25%) digestion for 10 min at 37°C. The resulting cells were filtered and washed in Dulbecco's modified Eagle's medium (Invitrogen). Fibroblasts were grown in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum and antibiotics.
Transfections and Western Analyses-Fibroblasts were plated into 6-well plates at 10 5 cells/well 24 h prior To Whom It May Concern: stimulation. After treatment, cells were lysed in 100 mM Tris, pH 6.8, 4% SDS, 20% glycerol, sonicated, and subjected to SDS-PAGE. MEFs were transfected with LipofectAMINE 2000 (Invitrogen). Cells were analyzed 24 h after transfection. Expression vectors for IKK␣, IKK␤, and NIK were kindly provided by Drs. R. Geleziunas and W. C. Greene (15). Human p100 was excised from a previously described expression vector (16) and inserted into pcRSV (Invitrogen).
Nuclear/Cytoplasmic Extracts and Electrophoretic Mobility Shift Assays (EMSAs)-For each nuclear preparation, 5 ϫ 10 5 cells were plated 24 h prior to stimulation. Stimulation was done under serum-free conditions. Following stimulation, nuclear and cytoplasmic extracts were prepared essentially as described (17). Briefly, fibroblasts were mechanically removed, washed twice in phosphate-buffered saline, and resuspended in 400 l of low salt buffer (10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, supplemented with protease and phosphatase inhibitors). After a 15-min incubation on ice, Triton-X-100 was added to a final concentration of 0.6%, and the suspension was vigorously vortexed for 10 s. The nuclei were pelleted, and the supernatant served as cytoplasmic extract. The pelleted nuclei were resuspended in 50 l of high salt buffer (20 mM Hepes, pH 7.9, 400 mM NaCl, supplemented with protease and phosphatase inhibitors) and incubated for an additional 15 min on ice. 2.5 l of this preparation were used in DNA binding reactions. An NF؊B-binding site from the light chain enhancer was used as a probe: 5Ј-AGTTGAGGGGACTTTCCCAGGC-3Ј (Promega, Madison, WI). Oct1 oligonucleotides were used in controls: 5Ј-TGTCGAATGCAAATCACTAGAA-3Ј. Complementary and annealed oligonucleotides were end-labeled with [␥-32 P]ATP. Approximately 20,000 cpm of probe were used per assay. The binding reaction was carried out at room temperature for 15 min in a total volume of 25 l containing 10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM MgCl 2 , 0.5 mM dithiothreitol, 50 g/ml poly(dI-dC)⅐poly(dI-dC), 4% glycerol. For supershift analyses, nuclear extracts were preincubated with antibodies for 30 min on ice prior to adding the probe. For Western blot analyses, 20 l of each preparation were subjected to SDS-PAGE. The efficiency of the nuclear-cytoplasmic fractionation was confirmed in several ways, including by the fact that entry of factors was dependent on stimulation and by the fact that RelA did not enter nuclei in NEMO-deficient cells, whereas RelB did (see Fig.6, A and C).
For immunoprecipitation, cytoplasmic and nuclear preparations were generated from 1.5 ϫ 10 6 cells/condition. These extracts were adjusted to 150 mM NaCl and mixed with anti-RelB antibodies (SC-848) or anti-NEMO antibodies (SC-8330) (Santa Cruz Biotechnology, Santa Cruz, CA) as a negative control that had been conjugated to agarose beads (Pierce). Following a 2-h incubation at 4°C, the agarose beads were washed four times in 150 mM NaCl, 25 mM Hepes, pH 7.3, 10% glycerol, 1% Triton-X-100, and the immunoprecipitated preparations were subjected to SDS-PAGE.

LT␤R Induces Processing of p100 and Degradation of IB␣ in
Mouse Embryo Fibroblasts-NF-B2-deficient mice are impaired in their splenic microarchitecture, they lack Peyer's patches, and they are severely impaired in lymph node formation, primarily due to defects within the stromal compartment (6,7,18,19). These deficiencies in secondary lymphoid organs, which also include loss of follicular dendritic cell networks, in turn contribute to defective immune responses in these mutant mice. Similar deficiencies have been noted in aly/aly mice (20,21) and in mice lacking lymphotoxin ␤ (LT␤) receptor or its ligands (LT␣␤ and Light) (20,(22)(23)(24)(25), members of the TNF receptor/ligand family. aly/aly mice are mutated in NIK (26). Overexpression of the wild-type form of NIK induces processing of the p100 protein of NF-B2 to p52, but its mutant form (aly) does not (13). Furthermore, splenocytes from aly/aly mice contain much less of the p52 protein of NF-B2 than aly/ ϩ mice while maintaining normal levels of p100 (21). NIK-induced processing in B cells depends on IKK␣ (14), and thus IKK␣deficient B cells contain much less p52 protein (14). Finally, although IKK␣-deficient animals die perinatally, which limits their analysis, it could nevertheless be shown that these mutant mice are deficient in Peyer's patches organogenesis (27). Based on these data, we hypothesized that critical functions of the LT␤ receptor on stromal cells depend on signaling via NIK, IKK␣, and NF-B2 and thus may involve processing of p100.
To test for this possibility, we subjected MEFs to an agonistic antibody directed against the LT␤R. We investigated with Western analyses for the expression of the NF-B2 proteins p100 and p52, as well as the NF-B1 proteins p105 and p50, the inhibitor of NF-B (IB)␣, IB␤, RelA, RelB, and c-Rel at six time points during an 8-h stimulation with anti-LT␤R antibodies ( Fig. 1). We also tested for expression of these proteins at 15 min and 8 h of stimulation with TNF␣. The experiments revealed a marked decrease in p100 and a concomitant increase in p52, beginning just before 4 h and maximal by 8 h of stimulation via the LT␤R. No such changes in p100 and p52 levels were seen with TNF␣ stimulation. TNF␣ instead caused the nearly complete degradation of IB␣ and the partial degradation of IB␤ by 15 min of stimulation; IB␣ levels were partly restored by 8 h of stimulation. LT␤R stimulation induced only a partial degradation of IB␣, which showed a delayed onset when compared with that induced by TNF␣. The amounts of IB␣ began to increase again after 2 h of stimulation via the LT␤R and were above starting levels by 8 h. Most likely this was due to increased synthesis in response to activated NF-B, in the absence of continued degradation of this inhibitor (see below). We failed to observe consistent changes in the amounts of the other proteins analyzed in Fig. 1, with the exceptions of RelB, whose amounts were increased, and p105, whose amounts were modestly decreased after 8 h of stimulation with TNF␣. In contrast to p52, the amounts of the p50 form of NF-B1 did not increase after LT␤R stimulation. These data indicated that LT␤ receptor stimulation in MEFs caused processing of p100 to generate p52 and a more modest degradation of the IB␣ inhibitory protein, implying engagement of two pathways to activate NF-B.
LT␤R-induced Processing of p100 Depends on Protein Synthesis, NIK, and IKK␣ but Not IKK␤ or IKK␥-Light and the membrane-bound LT␤ 2 ␣ are natural ligands of the LT␤ receptor. In addition to the agonistic antibody, Light could also be shown to induce processing of p100 to p52 in MEFs, whereas platelet-derived growth factor did not (Fig. 2). Given the delayed onset of processing, we asked whether the underlying mechanisms might involve intermediate steps requiring protein synthesis. Light-induced processing of p100 was sensitive to the protein synthesis inhibitor cycloheximide (Fig. 2), and this result was confirmed when cells were stimulated with the agonistic antibody to the LT␤ receptor (data not shown). Thus, signal-induced processing of p100 required the new or continued synthesis of a protein, which could explain the slow onset of processing upon stimulation.
Next, we investigated the mechanism underlying LT␤R-induced processing by taking advantage of mutant mice impaired or lacking in various signaling components. We generated MEFs from aly/aly mice, which carry a mutation in NIK. Agonistic antibodies to the LT␤R failed to induce processing of p100 to p52 in aly/aly MEFs (Fig. 3A). Therefore, NIK was required for LT␤R-mediated processing of p100, consistent with the ability of NIK to induce processing in transfected cells and the impaired LT␤R-induced NF-B transcriptional activity in NIK-mutated and NIK-deficient MEFs (27,28).
NIK was shown previously to depend on IKK␣ to induce processing in B cells, and consistent with this, LT␤R signaling also failed to induce processing in MEFs from IKK␣-deficient mice ( Fig. 3A; the 8-h-stimulated lane for these mutant cells contained slightly more protein, but the ratio of p100 to p52 did not change). Interestingly, MEFs from mice lacking IKK␤ or those lacking IKK␥ (also known as Nemo) were permissive for LT␤R-induced processing of p100, as were MEFs deficient in NF-B1 (Fig. 3A). (IKK␣/IKK1-, IKK␤/IKK2-, and Nemo/IKK␥deficient mice are described in Refs. 29 -31.) The regulatory subunit IKK␥ and the two catalytic subunits IKK␣ and IKK␤ together constitute the classical IKK core complex. Therefore, although LT␤R-induced processing did require the IKK␣ subunit, it was nevertheless independent of the classical, IKK␥ (Nemo)-containing IKK complex that controls degradation of the small IB inhibitors in response to many signals.
By comparison with wild-type MEFs, the amounts of p100 appeared to be somewhat reduced in NF-B1-and IKK␤-deficient MEFs but were especially reduced in IKK␥-deficient MEFs. Nevertheless, processing still occurred in response to LT␤R stimulation, resulting in a more substantial depletion of p100 in the mutant versus wild-type cells (a longer exposure of the IKK␥-deficient MEFs is shown in Fig. 3A, lower panel). Basal and LT␤R-induced activation of NF-B via the classical Wild-type MEFs were stimulated with TNF␣ or with agonistic anti-LT␤ receptor antibodies for the times shown. Total extracts were prepared by SDS lysis and subjected to SDS-PAGE followed by Western analysis with antibodies against NF-B and IB proteins as indicated.
FIG. 2. LT␤ receptor-induced processing of p100 depends on de novo protein synthesis. Wild-type MEFs were stimulated for 8 h with anti-LT␤ receptors, Light, or platelet-derived growth factor (PDGF), or they were preincubated with cycloheximide (CHX) and stimulated for 8 h with Light or left without added stimulus, as indicated. Whole cell extracts were analyzed for the presence of the p100 and p52 proteins of NF-B2 using SDS-PAGE. ns denotes a nonspecific protein that is variably detected.
FIG. 3. IKK␣ and NIK are required for p100 processing, but IKK␤, IKK␥ (Nemo) or NF-B1 are not. As shown in A, MEFs deficient in IKK␣, IKK␤, IKK␥, or NF-B1 or mutated in NIK (aly/aly) were stimulated with anti-LT␤ receptor antibodies for 8 h or not treated as indicated. p100 and p52 proteins of NF-B2 were detected by Western analysis of whole cell lysates. Because expression levels of NF-B2 were low in the IKK␥-deficient cells, a long exposure of the Western blot for these mutant cells is also shown in the lower panel. ns denotes a nonspecific protein that is variably detected. As shown in B, an inhibitor of IKK␤ reduces resynthesis of p100 but does not interfere with p100 processing in LT␤ receptor-stimulated wild-type MEFs. Wild-type MEFs were preincubated for 30 min with increasing amounts of PS1145 and then stimulated with TNF␣ for 15 min to detect IB␣ degradation (lower panel), or they were stimulated for 8 h with anti-LT␤ receptor antibodies to detect p100 and p52 proteins (upper panel) by Western analysis of whole cell lysates as indicated. Untreated cells are shown in the first lane.
IKK to IB degradation path may be required for optimal p100 expression. MEFs deficient in IKK␥ may have contained especially low levels of p100 because the classical activation pathway was completely blocked in these mutant cells, whereas residual activity may have persisted in the IKK␤-deficient mutants due to the presence of IKK␣. We also used an IKK␤specific inhibitor, PS1145, to provide further support for the suggested role of the classical activation route in maintaining p100 levels while having no role in processing. To control for the activity of this inhibitor, we confirmed a dose-dependent inhibition of TNF␣-induced degradation of IB␣ after 15 min of stimulation (Fig. 3B). Increasing amounts of PS1145 also decreased the amounts of p100 after 8 h of stimulation via the LT␤ receptor, presumably due to reduced new synthesis of p100, whereas processing to p52 was essentially unaffected (Fig. 3B). Therefore, optimal expression of p100 depended on basal and induced activation of the classical, IKK-mediated pathway for NF-B, but processing of p100 did not and instead only depended on the IKK␣ subunit.
It remained theoretically possible that the inability to process p100 in IKK␣-deficient and NIK-impaired MEFs in response to LT␤ stimulation was not directly related to loss of IKK␣ or NIK function. We therefore tested whether transfection of these mutant and wild-type MEFs with p100 together with IKK␣, IKK␤, or NIK could confirm the conclusions reached with Fig. 2. Overexpression of IKK␣ and especially of NIK in wild-type MEFs induced processing of p100 to generate p52, whereas IKK␤ did not (Fig. 4A). IKK␣ and NIK were similarly able to induce processing in IKK␥ (Nemo)-deficient (Fig. 4B), IKK␤-deficient (Fig. 4D), and NIK-impaired (aly/aly) (Fig. 4E) MEFs. This confirmed that the classical Nemo/IKK␤ pathway was irrelevant for processing and that the inability of aly/aly MEFs to allow processing could be overcome simply by supplying wild-type NIK or IKK␣. This latter result also placed IKK␣ downstream of NIK, which was confirmed by the fact that overexpression of IKK␣ in IKK␣-deficient MEFs resulted in processing, whereas overexpression of wild-type NIK did not (Fig. 4C). Therefore, we concluded that processing of p100 to p52 as induced by LT␤R stimulation depended on and proceeded via NIK and then IKK␣ but was independent of classical IKK␥ (Nemo) and IKK␤-dependent NF-B activation; classical IKK activity did, however, help maintain p100 levels.
LT␤R Induces Sequential Activation of RelB and RelA by Distinct Pathways-Next, we investigated LT␤R-initiated NF-B activation in EMSAs designed to determine the composition of activated NF-B. Wild-type MEFs contained some basal B DNA binding activity composed primarily of p50 homodimers and p50-RelA heterodimers (faster and slower migrating shifted bands, respectively), as assessed in EMSA supershift experiments with antibodies to the various NF-B subunits (Fig. 5A). Approximately equal amounts of extracts were loaded, and this was confirmed in separate EMSAs in which DNA binding activity to the cognate site for the Octamer-1 transcription factor was assessed (data not shown). LT␤R stimulation for 2 h resulted in increased amounts of DNA binding activity composed primarily of p50-RelA and, to a lesser degree, p50-RelB ( Fig. 5B; RelA supershifts marked). After 8 h of stimulation, the binding activity of p50-RelA dimers had decreased, whereas p50-RelB activity had increased further, and p52-RelB activity could be detected as well ( Fig. 5C; RelB and p52 supershifts marked). In addition, p50 homodimer binding activity appeared to have increased somewhat. No significant c-Rel DNA binding activity was noted in these MEFs, although this antibody was able to detect c-Rel binding in lymphoid cells (data not shown).
The EMSA supershifts with 8-h-stimulated extracts revealed predominantly p50-containing complexes with an apparently much smaller contribution of p52-containing complexes. One must consider, however, that the EMSA assays are not quantitative and do not necessarily correlate with the degree to which a given NF-B dimer has been released from cytoplasmic inhibition. Different dimers bind standard B DNA elements with varying strength, and p52-containing dimers in particular have not been carefully tested for preferred binding sites. Furthermore, the various antibodies used differ in strength of binding and may in addition be differentially affected in the supershift assay. It is possible, therefore, that these EMSA assays underestimated the amounts of p52-containing complexes in particular. This notion was supported by nuclearcytoplasmic fractionation experiments, which revealed considerable migration of p52 and RelB into nuclei starting by 2 h of LT␤R stimulation and increasing thereafter, whereas the amount of RelA in nuclei was highest after 2 h of stimulation, declining thereafter ( Fig. 6A; RelB migrated as two closely spaced bands with the upper band preferentially translocating to nuclei). In addition, RelB coimmunoprecipitation experiments confirmed that RelB was associated with p52 in the nucleus ( Fig. 6B; coimmunoprecipitated p52 is marked by an asterisk). Together the data revealed an early activation of p50-RelA dimers, which decreased after 2 h, whereas activation of RelB dimers continued to increase past 2 h.
We speculated that the partial and transient degradation of IBs seen early after LT␤R exposure (see above) might be responsible for the early and transient increase in the p50-RelA binding activity, whereas the processing of p100 might be responsible for the late rise in p52-RelB binding activity and possibly also in p50-RelB binding activity. To test this theory, we investigated the activation of NF-B with EMSA assays in the mutant MEFs. LT␤R-mediated stimulation of IKK␤-and IKK␥ (Nemo)-deficient MEFs for 8 h resulted in strong activation of p50-RelB and, to an apparently lesser degree, p52-RelB, similar to what was observed in wild-type MEFs (Fig. 7, A and B, respectively; RelB supershifts marked). Contrary to results with wild-type MEFs, no activation of RelA-containing dimers (primarily p50-RelA) was observed (Fig. 7, A and B), not even after 2 h of stimulation (data not shown; see also Fig. 6C) These results indicated that the classical activation pathway via IKK␤/IKK␥ (Nemo) was not required for DNA binding activation of p50-RelB or p52-RelB but was required for activation of p50-RelA.
To confirm that LT␤R-mediated stimulation of IKK␥ (Nemo)-deficient MEFs caused translocation of both p52 and RelB into nuclei, we also performed nuclear-cytoplasmic fractionation experiments. As shown in Fig. 6C, RelB and p52 entered nuclei of these mutant cells and continued to do so during the course of stimulation, similar to what was seen with wild-type cells (Fig. 6A), whereas RelA failed to be translocated into nuclei, as expected for these mutant cells, which also served as a control for the experiments.
We next tested the relevance of the alternative pathway in the activation of B binding activity in response to LT␤R stimulation. IKK␣-deficient and NIK-impaired (aly/aly) MEFs failed to activate p50-RelB or p52-RelB after 8 h of stimulation (Fig. 7, C and D, respectively). As in wild-type MEFs, p50-RelA was activated after 2 h of stimulation in both mutant MEFs (data not shown) and was still clearly detected after 8 h (Fig. 7, C and D; RelA supershifts marked). Together these results demonstrated that activation of p50-RelA required IKK␤ and IKK␥ (Nemo) but not IKK␣ or NIK. On the other hand, activation of p50-RelB and p52-RelB required IKK␣ and NIK but not IKK␤ or IKK␥. We also note that activation of p50-RelA appeared to be somewhat prolonged in the absence of IKK␣ or NIK, whereas activation of p50-RelB and p52-RelB seemed to be slightly enhanced in the absence of IKK␤ and IKK␥ at early times (data not shown).
p100 Processing Activates p52-RelB and p50-RelB Complexes-The data described indicated that the alternative pathway of activation via NIK and IKK␣ and, by extension, processing of p100 were responsible for the activation of not only p52-RelB but also for the activation of p50-RelB. Although p52-RelB dimers could result from processing of p100-RelB complexes, it was less obvious how p50-RelB dimers might be activated. To investigate underlying mechanisms further, we analyzed LT␤R-induced activation in NF-B2-deficient MEFs since these cells lack the p100 inhibitor to begin with. In these mutant MEFs, p50-RelB was already basally activated in the absence of any added stimulus, and this binding activity was increased slightly further with stimulation via the LT␤R ( Fig.  8A; RelB supershifts marked). The basal activity suggested that it was not the act of processing of p100 per se but the absence of the p100 inhibitor that led to p50-RelB binding activity. RelB is reported to preferentially associate with p100 (32), and in the absence of p100, RelB is presumably free to associate with p50. We speculated that LT␤R stimulation of NF-B2-deficient MEFs may have led to a further increase above basal levels of p50-RelB DNA binding activity as a result of new synthesis of both RelB and NF-B1 induced via the classical NF-B activation route. In support of the theory that RelB complexes were easily activated in the absence of p100, we demonstrated that TNF␣ stimulation activated p50-RelB DNA binding activity to extremely high levels in NF-B2deficient MEFs (Fig. 8B, lower panel). By comparison, TNF␣stimulation of wild-type MEFs, which contain p100, did not result in such high activation of RelB complexes (Fig. 8B, upper panel; RelB supershifts in wild-type and NF-B2-deficient MEFs are marked). The RelB activity observed after long term stimulation of wild-type with TNF␣ (Fig. 8B, upper panel) was most likely due to the high amounts of RelB protein induced via the classical NF-B activation pathway, some of which may have escaped sequestration by p100 (Fig. 1). These results supported the notion that the absence or presence of p100 was the key to whether p50-RelB dimers were readily formed or not.  A and B) or NEMO/ IKK␥-deficient (C) MEFs were stimulated with anti-LT␤R antibodies for the times shown, and nuclear (N) and cytoplasmic (C) extracts were analyzed for the presence of RelA, RelB, and p52 by Western analyses in panels A and C. As shown in B, cytoplasmic and nuclear extracts were also immunoprecipitated (IP) with anti-RelB antibodies and analyzed for the presence of coimmunoprecipiated p52. Co denotes a polyclonal control antibody to NEMO and thus unrelated to RelB. Coimmunoprecipitated p52 is marked by an asterisk. The last lane represents a direct p52 Western analysis of the 8-h-stimulated cell lysate that was run on the same gel as the immunoprecipitated material to identify p52. The background bands in panel B are due to the immunoprecipitating antibodies present in the immunoprecipitates. The nuclear and cytoplasmic extract lanes contained approximately equal amounts of proteins, but on a per cell basis, the nuclear lanes represented ϳ8 times more cells than the cytoplasmic lanes. Together these results suggested that NIK and IKK␣ were required for activation of p50-RelB because they led to processing of p100 and thus removal of the preferred and inhibitory binding partner for RelB.
The experiments of Fig. 8B also demonstrated that LT␤Rinduced activation of RelB DNA binding activity in wild-type MEFs was dependent on protein synthesis (upper panel), consistent with the dependence of p100 processing on protein synthesis. We observed some RelB binding activation even in the presence of the protein synthesis inhibitor cycloheximide when NF-B2-deficient MEFs were stimulated with the LT␤R (Fig. 8B, lower panel). In the absence of p100 (NF-B2-deficient MEFs), RelB proteins are likely to be associated with p105 and p50. It is possible that basal or induced turnover of the inhibitory p105 might have led to a further release and thus activation of RelB complexes, as p105 would not be replaced in this situation. LT␤R-induced activation of RelA complexes was noticeably enhanced in the presence of cycloheximide in wild-type (Fig. 8B, upper panel) and especially in NF-B2-deficient MEFs (lower panel), presumably due to the complete loss of relevant inhibitors in the absence of resynthesis. DISCUSSION The study presented here shows that LT␤R-mediated stimulation of mouse embryo fibroblasts resulted in the engagement of two separate signaling pathways, leading to activation of distinct NF-B complexes. We established that the initial or classical activation was mediated by IKK␤-and IKK␥-dependent degradation of IB␣ and subsequent liberation of primarily p50-RelA complexes, independent of NIK and IKK␣. We further demonstrated that the second or alternative activation was mediated by NIK-and IKK␣-dependent processing of the inhibitory p100 protein of NF-B2, independent of IKK␤ and IKK␥/Nemo. This latter pathway led to activation of primarily p50-RelB and p50-RelA dimers. These conclusions were derived from analyses of LT␤R-stimulated wild-type and mutant MEFs impaired in NIK or lacking IKK␣, IKK␤, IKK␥, or NF-B2. LT␤R-induced p50-RelA activation was modest and transient, consistent with an only partial and transient loss of IB␣ induced by the classical pathway. With time of stimulation, amounts of IB␣ increased to above starting levels to again inhibit RelA complexes. In contrast, DNA binding activity of RelB complexes increased and began to dominate the NF-B binding activity. Stimulation of the LT␤R on fibroblasts, therefore, initiated two separate pathways to sequentially activate different NF-B complexes. The change in the types of NF-B complexes activated is likely to result in a corresponding change in the genes targeted with time during the course of stimulation via LT␤R. The data presented here identify the LT␤R as a physiologic inducer of the second pathway of activation. The results furthermore clarify the importance of this pathway since long term B binding activity in response to LT␤R stimulation was entirely dependent on p100 processing.
A prior analysis of NIK-deficient MEFs discovered gene induction defects specific to stimulation via the LT␤R but not the TNF␣-receptor (28). Since the LT␤R-induced B binding activity analyzed at early times after stimulation was not impaired in the NIK-deficient MEFs (it is dominated by the classical activation path as shown here), it was speculated that LT␤Rinduced changes in transactivation potential might have been defective in the absence of NIK. Although it remains possible that NIK is also required to enhance transactivation in response to LT␤R stimulation, our data suggest that the lack of processing of p100 in NIK-deficient MEFs and thus the lack of sustained activation of RelB complexes in particular could also explain the loss of gene induction observed.
The alternative activation path involving NIK, IKK␣, and p100 processing activated not only p52-RelB complexes, which could be generated directly from p100-RelB, but also activated p50-RelB. Activation of this latter complex was a consequence of removal of p100 via processing, the preferred and inhibitory binding partner for RelB, thus liberating RelB to instead associate with NF-B1 proteins, including the constitutively generated p50. The mere absence of p100 in NF-B2-deficient MEFs was already sufficient to lead to basal activation of p50-RelB (which was not observed in any other MEF), and this dimer was further activated in the NF-B2-deficient MEFs upon stimulation via the LT␤R, and in particular, upon stimulation with TNF␣. This was likely the result of new induced synthesis of RelB by the classical activation route, which was particularly noticeable after extensive stimulation with TNF␣ ( Fig. 1), and it may, in addition, have resulted from some stimulation-dependent degradation of p105 (12), the only known remaining inhibitor of RelB in the absence of p100 (small IBs are not thought to be physiologic inhibitors of RelB, although they could theoretically have contributed to some inhibition in the absence of p100). Based on these data, we conclude that the mere processing of p100 indirectly led to activation of p50-RelB complexes without the need of any additional NIK-and IKK␣-mediated signals in the process.
The observed sensitivity of p100 processing to protein synthesis inhibitors is intriguing. It suggested that induced or continued synthesis of one or more proteins was required. Such a mechanism is consistent with the observed long delay in p100 processing, in all experiments requiring 2 or more h of LT␤R stimulation before a clear increase in p52 was apparent. Continuous synthesis of p100 could be required, especially if processing is somehow linked to translation. Also, continuous or induced synthesis of NIK could be required, especially since cells may express only small amounts of NIK normally. Mechanisms that increase the amounts of NIK might be sufficient to induce processing, based on the fact that transfected NIK is highly active even in the absence of any added signals (Fig. 4). Nevertheless, induced or continued synthesis of other proteins cannot be ruled out.
MEFs in which the alternative pathway was impaired appeared to show a more sustained activation of RelA complexes, whereas MEFs in which the classical pathway was impaired appeared to have activated RelB complexes more rapidly. Although these quantitative assessments need to be independently confirmed, they do raise the interesting possibility that the two pathways may compete in wild-type cells. IKK␣ might be limiting and thus may not have been immediately available to the second pathway.
The results project a dynamic interplay of the pathways and complexes activated with time of stimulation. Loss of any of the components of the classical activation pathway (especially loss of IKK␥) appeared to reduce the amounts of p100/NF-B2 present in MEFs. In addition to NF-B2, RelB, c-Rel, and NF-B1 are also known targets of NF-B. Activation via the classical route may be needed to maintain NF-B components, including NF-B2 and RelB, to allow maximal effects of the alternative activation pathway. Different signals may involve the activation paths to different degrees and to different effects when viewed over the course of prolonged stimulation, depending on cell type.
Recently, we discovered that ligation of the B cell-activating factor (BAFF)-receptor by BAFF on B cells induces processing of the NF-B2 protein p100 to generate p52 (33). Although the requirements and effects of processing in B cells could not be investigated as thoroughly as was possible with MEFs here, p100 processing in B cells could be shown to depend on NIK and protein synthesis, to be independent of IKK␥ (Nemo), and to lead to activation of RelB complexes. In addition, in a recent study published after completion of our manuscript, Dejardin et al. (34) report findings similar to ours here with LT␤R-stimulated fibroblasts. Although LT␤R is able to activate the classical pathway as well as p100 processing (Ref. 34 and our study), BAFF does not appear to significantly stimulate the classical activation pathway in B cells, although it very effectively induces p100 processing. Engagement of this second or alternative activation path by BAFF in developing splenic B cells was shown to contribute to survival of these cells. Although the biologic roles of LT␤R-mediated activation are not fully characterized, they include differentiated functions of stromal cells in lymphoid organs. Based on overlapping defects present in mice deficient in LT␤R signaling, NIK function, IKK␣, and NF-B2, it is likely that LT␤R-mediated processing of p100 in stromal cells contributes to the communication between stromal cells and lymphoid cells and helps to organize secondary lymphoid organs. Downstream targets of p100 processing in stromal cells include the B lymphocyte chemoattractant (BLC) (CXCL13) chemokine (34), which is in agreement with previous reports demonstrating that induced expression of this chemokine is suppressed in NF-B2-deficient stromal cells (18) and that mice deficient in LT␤R signaling, NIK function, or NF-B2 are severely impaired in formation of B cell follicles (6, 7, 18 -25).