NF-κB1 and c-Rel cooperate to promote the survival of TLR4-activated B cells by neutralizing Bim via distinct mechanisms

Different microbial components activate cells of the innate and the adaptive immune system by binding specific Toll-like receptors (TLRs).¹  Bacterial lipopolysaccharide (LPS) is a TLR4 ligand that promotes both B-cell proliferation and differentiation.²  Signaling networks downstream of TLR4 in B cells include the Rel/nuclear factor-κB (NF-κB) and mitogen-activated protein kinase (MAPK) pathways. Rel/NF-κB transcription factors are dimers composed of 5 related subunits (c-Rel, RelA, RelB, p50NF-κB1, and p52NF-κB2). In the absence of stimulatory signals, Rel/NF-κB factors are retained within the cytosol by IκB proteins.³  Signals emanating from cell surface or intracellular receptors promote Rel/NF-κB nuclear translocation by engaging an IκB kinase (IKK) that phosphorylates IκBs, targeting them for degradation.⁴  In mature B cells,^5-7  both c-Rel and NF-κB1 are required for B-cell proliferation and survival^8,9  after B-cell receptor (BCR) or TLR4 stimulation.

Little is known about the role of MAPKs in TLR4-activated B cells. MAPK signaling uses kinase cascades that activate the effector serine/threonine kinases extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinase (JNK), and p38. MAP kinase kinase kinases (MAP3Ks) are at the apex of these pathways.¹⁰  TLR4 uses distinct MAP3Ks to activate different MAPK effectors; transforming growth factor–activated kinase 1 (TAK1) induces JNK and p38, whereas Tpl2 activates ERK.^11,12  IKK is essential for both TLR4 activation of Rel/NF-κB and ERK.¹³  In unstimulated cells, Tpl2 forms a complex with p105NF-κB1, the p50NF-κB1 precursor.¹⁴  TLR4 activation of IKK2 targets p105 for proteasomal degradation, leading to the release of Tpl2, which then engages the mitogen-activated protein kinase kinase (MEK)/ERK pathway.¹³ nfkb1^−/− macrophages lack detectable Tpl2 because of its rapid turnover,¹²  resulting in a failure to activate ERK after TLR4 stimulation.^11,12 

Given microbial products can induce apoptosis,^15,16  higher organisms have evolved mechanisms to prevent this cell death. In many instances, activated B cells use the intrinsic survival pathway,¹⁷  which is regulated through interactions between members of 3 subgroups of Bcl-2 proteins: prosurvival proteins (Bcl-2, Bcl-x_L, Bcl-w, Mcl-1, and A1), multi-BH (Bcl-2 homology) domain proapoptotic proteins (Bax, Bak, and Bok), and proapoptotic BH3-only proteins (Bim, Bad, Bid, Bik, Bmf, Hrk, Noxa, and Puma).¹⁸  Bax and Bak, which trigger apoptosis by disrupting mitochondrial outer membrane integrity,¹⁹  are kept in check by Bcl-2 prosurvival homologs. BH3-only proteins activated via various transcriptional and posttranslational mechanisms²⁰  are thought to initiate apoptosis either by binding Bcl-2 prosurvival homologs associated with Bax and Bak, thereby leading to their liberation and activation,^18,21,22  or in the case of BH3-only proteins such as Bim, by directly binding to and activating Bax or Bak.²³  Irrespective of which model of Bax/Bak activation is correct, increased or decreased levels of Bcl-2-like prosurvival and BH3-only proteins, respectively, tip the balance in favor of survival, whereas increases in BH3-only protein expression or reductions in Bcl-2–like prosurvival proteins induce apoptosis.

Among BH3-only proteins, Bim plays a prominent role in hemopoietic cell apoptosis.^24,25  Bim_EL and Bim_L are the most abundant isoforms^26,27  and bind with high affinity to all Bcl-2–like prosurvival proteins.²⁸  Bim proapoptotic activity is regulated by transcriptional and posttranslational mechanisms.²⁰  In the latter instance, Bim can be regulated by phosphorylation,²⁹  with ERK inactivating Bim by targeting it for ubiquitination and proteasomal degradation,^30-32  whereas JNK enhances Bim proapoptotic activity.³³ 

Here we show p105NF-κB1 promotes B-cell survival after LPS stimulation by a novel nontranscriptional mechanism involving Tpl2-dependent ERK activation that leads to Bim phosphorylation and degradation. LPS stimulation also leads to a c-Rel–dependent increase in A1 and Bcl-x_L, which bind Bim. Consistent with both NF-κB1 and c-Rel promoting TLR4 mediated B-cell survival, loss of Bim prevented the elevated death of LPS-stimulated nfkb1^−/− and c-rel^−/− B cells. Furthermore, LPS caused more apoptosis in nfkb1^−/−c-rel^−/− B cells than either nfkb1^−/− or c-rel^−/− B cells, demonstrating that optimal survival after TLR4 stimulation requires Bim to be neutralized by the c-Rel–dependent induction of A1 and Bcl-x_L, plus p105NF-κB1/Tpl2/ERK–mediated Bim phosphorylation and degradation.

All experimental mice are on a C57BL/6 background and were 7 to 12 weeks of age. nfkb1^−/−,⁷ c-rel^−/−,⁶ nfkb1^−/−c-rel^−/−,³⁴  and vav-bcl-2 transgenic (bcl-2^T)^35,36  mice were maintained as inbred strains. nfkb1^−/−bim^−/− and nfkb1^−/−bad^−/− mice were generated by intercrossing nfkb1^−/− and bim^−/−24  or nfkb1^−/− and bad^−/−37  mice, whereas nfkb1^−/−bcl-2^T, c-rel^−/−bcl-2^T, and nfkb1^−/−c-rel^−/−bcl-2^T mice were generated by intercrossing the mutant and bcl-2^T mice. All animal experiments were conducted in accordance with guidelines of the National Health and Medical Research Council (Australia).

Antibody sources were: ERK and c-Raf phosphospecific antibodies (Cell Signaling Technology, Danvers, MA), ERK, Tpl2, and actin antibodies (Santa Cruz Biotechnology, Santa Cruz, CA), monoclonal mouse anti–Bcl-x_L and hamster anti–mouse Bcl-2 antibodies (BD Biosciences, San Jose, CA) and monoclonal rat anti-Bim antibody (clone 3C5), rat anti-A1 (clones 56B3 and 51B2), and anti-Mcl1 (clone 14C11-20) antibodies (Dr L. O'Reilly, The Walter and Eliza Hall Institute of Medical Research [WEHI]) Melbourne, Australia). PD98059 was from Calbiochem (San Diego, CA), 4-hydroxy-tamoxifen (4-HT), and Escherichia coli LPS was from Sigma-Aldrich (St Louis, MO).

Viral stocks of pMy-internal ribosomal entry site (IRES)–green fluorescent protein (GFP)³⁸  and pMy-IRES-GFP-Raf/Estrogen Receptor (ER) were generated as described.¹¹  Bone marrow chimeras were made by injecting irradiated (2 × 550 rad) C57BL/6 Ly5.1 mice with approximately 10⁶ virus-infected Ly5.2⁺wt or nfkb1^−/− hemopoietic progenitors.

Splenic B cells were isolated by MACS purification (Miltenyi Biotec, Auburn, CA) or cell sorting.³⁹  B cells prepared using both protocols were routinely of more than 95% purity. Splenic B cells from chimeras containing progenitors transduced with GFP-expressing retroviruses were purified by staining with phycoerythrin (PE)-conjugated anti-B220 antibody and sorting for GFP⁺PE⁺ cells.

Purified B cells seeded at 3 × 10⁵ cells/mL in Dulbecco modified Eagle medium/10% fetal bovine serum/50 μM 2-mercaptoethanol were unstimulated, treated with LPS (20 μg/mL), affinity-purified goat antimouse IgM (Fab)₂ fragments (25 μg/mL; Jackson ImmunoResearch Laboratories, West Grove, PA), or both stimuli. PD98059 and 4-HT were used at 20 μM and 50 nM, respectively. Cell proliferation was quantified by ³H-thymidine uptake. B-cell survival in culture was assessed by flow cytometric analysis of cells stained with propidium iodide.

B-cell lysates were fractionated on isoelectric focusing gels (ZOOM IPG strips; Invitrogen, Carlsbad, CA), after which gel strips were resolved in the second dimension on NuPAGE MOPS/SDS ZOOM gels (Invitrogen). Proteins were transferred onto nitrocellulose membranes and Western blotting performed as described.¹¹ 

For 3-color FACS analysis of B lymphocytes,³⁴  10⁶ splenocytes were incubated with FITC-conjugated anti-CD23, APC-conjugated anti-CD21, and PE-conjugated anti-IgM antibodies and viable cells examined using a FACSCalibur (BD Biosciences).

mRNA was isolated from FACS-purified IgM⁺wt and nfkb1^−/− splenic B cells (10⁶ cells/population) before and after LPS stimulation. bim mRNA levels were measured by real-time PCR⁴⁰  using cDNA generated from equivalent amounts of RNA. bim mRNA levels were normalized by performing quantitative PCR for gapdh mRNA. Primer sequences are available on request.

Total cell extract isolated from 5 × 10⁷ magnetic-activated cell separation purified wt and nfkb1^−/− sIgM⁺ splenic B cells for each time point were lysed¹¹  and a sample of each subjected to Western blotting using anti-HSP70 antibodies to quantify protein content in each lysate. The remainder of the samples was incubated with rat anti-Bim monoclonal antibody (3C5), followed by immunoprecipitation (IP) with protein G-Sepharose beads. After washing, samples were eluted, fractionated on 12% sodium dodecyl sulfate-polyacrylamide gels and the proteins transferred onto nitrocellulose membranes. Filters were incubated with either mouse monoclonal anti–BCl-x_L, hamster anti-mouse Bcl-2, or monoclonal rat anti-A1 antibodies and bound antibody revealed by enhanced chemiluminescence (GE Healthcare, Little Chalfont, United Kingdom) using horseradish peroxidase-conjugated goat anti-mouse, goat anti-hamster or goat anti-rat antibodies (Southern Biotechnology Associates, Birmingham, AL).

Mature nfkb1^−/− B cells exhibit heightened spontaneous and LPS-induced death in culture.⁸  Initially, we determined whether perturbation of the cell intrinsic survival pathway⁴¹  was responsible for the abnormal death of nfkb1^−/− B cells. This was assessed using nfkb1^−/− mice expressing a hemopoietic lineage-driven bcl-2 transgene (bcl-2^T). Consistent with previous reports,^8,9  loss of NF-κB1 led to enhanced spontaneous B-cell death in culture over 48 hours (Figure 1A). LPS stimulation further heightened the death of nfkb1^−/−, but not wt B cells. bcl-2^T, expressed at equivalent levels in wt and nfkb1^−/− B cells (Figure S1, available on the Blood website; see the Supplemental Materials link at the top of the online article) blocked enhanced spontaneous plus LPS-induced nfkb1^−/− B-cell apoptosis (Figure 1B). Although bcl-2^T expression rescued TLR4-stimulated nfkb1^−/− B-cell survival, proliferation remained defective (Figure 1B). These data show that NF-κB1 independently regulates survival and proliferation of TLR4-stimulated B cells, with death that occurs in the absence of NF-κB1 mediated through the cell-intrinsic pathway.

NF-κB control of transcription for genes encoding Bcl-2 prosurvival family members^42,43  prompted a comparison of Bcl-2, Mcl-1, Bcl-x_L, and A1 expression in wt and nfkb1^−/− B cells (Figure 1C). Bcl-2 and Mcl-1 were expressed at similar levels in unstimulated wt and nfkb1^−/− B cells. LPS treatment did not alter Mcl-1 expression, whereas Bcl-2 expression increased marginally (∼2-fold) in wt but not nfkb1^−/− B cells after 12 hours. Basal levels of Bcl-x_L and A1, plus their induction by LPS, which is c-Rel dependent,^42,43  was normal in nfkb1^−/− B cells. This indicates that the abnormal death of nfkb1^−/− B cells was not the result of major differences in the expression of Bcl-2 prosurvival family members.

Our attention next turned to the proapoptotic BH3-only proteins, with a focus on Bim, given its prominence in B-cell death.⁴⁴  Bim involvement in nfkb1^−/− B-cell death was determined using nfkb1^−/−bim^−/− mice. Like the parental strains, nfkb1^−/−bim^−/− mice exhibit no gross developmental abnormalities (A.B., unpublished data, October 2006). A comparison of spleen cellularity, plus follicular and marginal zone (MZ) B-cell populations in wt, nfkb1^−/−, bim^−/−, and nfkb1^−/−bim^−/− mice (Table S1; Figure S2) revealed that increased follicular B-cell numbers that occur in the absence of Bim²⁴  were not further altered in nfkb1^−/−bim^−/− mice. However, reduced MZ B-cell numbers in the nfkb1^−/−45  and bim^−/−46  mutants were compounded in nfkb1^−/−bim^−/− mice. bcl-2^T expression had a similar effect on MZ B cells in nfkb1^−/− mice as loss of Bim (Figure S2).

With follicular B-cell development intact in nfkb1^−/−bim^−/− mice, spontaneous and LPS-induced death was examined in these mutant B cells (Figure 2). During the initial 24 hours in culture, the high spontaneous death of nfkb1^−/− B cells was abrogated by loss of Bim (Figure 2A, lanes 1-4). However, spontaneous nfkb1^−/−bim^−/− B-cell apoptosis after 48 hours (∼20%, lane 8), although less than for nfkb1^−/− B cells (∼50%, lane 6), was higher than that of bim^−/− B cells (∼8%, lane 7). Notably, bcl-2^T expression limited the spontaneous death of nfkb1^−/− B cells to approximately 10% (Figure 1A). This indicated that the elevated spontaneous death of nfkb1^−/− B cells involves Bim-dependent and -independent processes, with Bim-independent apoptosis operating predominantly at later times. In contrast, the higher levels of LPS-induced nfkb1^−/− B-cell death were completely inhibited by loss of Bim, at both early and late time points (Figure 2B).

The finding that Bim was critical for the enhanced spontaneous and LPS-induced death of nfkb1^−/− B cells prompted an examination of mechanistic links between NF-κB1 and Bim proapoptotic activity. First, a comparison of bim mRNA levels in wt and nfkb1^−/− B cells (Figure 3A) established that, despite modest increases in bim mRNA after LPS treatment, bim expression was equivalent in cells of both genotypes. Next, Bim levels were examined in LPS-activated wt and nfkb1^−/− B cells at 6-hour intervals over 18 hours (Figure 3B). Bim expression was equivalent in unstimulated wt and nfkb1^−/− B cells, gradually increasing in response to LPS, such that, by 18 hours, Bim in cells of both genotypes was 2- to 3-fold higher than in prestimulated cells.

In the absence of differences during sustained LPS activation, Bim expression was examined early during LPS stimulation (Figure 3C). Interestingly, Bim levels in LPS-stimulated wt but not nfkb1^−/− B cells was lower after 2 hours compared with B cells cultured without LPS. Given our bim mRNA expression data, this change in Bim levels probably reflected Bim turnover. Because ERK phosphorylation of Bim targets it for degradation,^29,30  we investigated the connection between ERK and the transient reduction in Bim after LPS stimulation. Initially, Bim levels were measured in wt B cells stimulated with LPS plus the MEK inhibitor PD98059 (Figure 3D). The transient decrease in Bim after 2 hours of LPS stimulation was blocked by PD98059, indicating that it was ERK-dependent.

p105NF-κB1 serves as a scaffold for the ERK-specific MAP3K, Tpl2,^12,14  with an absence of p105 in macrophages and fibroblasts leading to very low levels of Tpl2 that result in defective LPS-induced ERK activation.^11,12  Tpl2 was undetectable in nfkb1^−/− B cells (Figure 3E), and this coincided with a defect in LPS-induced ERK activation (Figure 3F). Consistent with LPS-induced ERK activation in B cells being dependent on Tpl2 and not c-Raf, the MAP3K that activates ERK downstream of many receptors, including the BCR,⁴⁷  S338 phosphorylation of c-Raf is absent in LPS-stimulated B cells (Figure 3F).

The link between impaired ERK activity in LPS-treated nfkb1^−/− B cells and increased apoptosis was reinforced by showing apoptosis in LPS-stimulated wt B cells was enhanced by PD98059 (Figure 4A). To confirm that impaired survival coincided with the ERK defect, B cells were isolated from mice engrafted with hemopoietic progenitors transduced with retroviruses coexpressing GFP and a c-Raf/estrogen receptor (Raf/ER) fusion protein that served as a 4-HT–inducible activator of ERK in B lymphoma cells (Figure S3A). Spontaneous (Figure 4B) and LPS-induced (Figure 4C) apoptosis was examined in these cells over 48 hours. Raf/ER activation diminished the spontaneous death of wt and nfkb1^−/− B cells after 24 hours (∼6% and ∼18%, respectively) compared with GFP⁺ control B cells (∼17% and 30% for wt and nfkb1^−/− cells, respectively). Sustained ERK activation had less impact on spontaneous apoptosis after 24 hours, with approximately 22% and approximately 42% death for Raf/ER expressing wt and nfkb1^−/− B cells, respectively (T48 hours) compared with approximately 40% and 51% for GFP⁺wt and nfkb1^−/− B cells.

In contrast, activated Raf/ER protected nfkb1^−/− B cells against LPS-induced apoptosis over 48 hours (Figure 4C). The reduced levels of apoptosis were not the result of increased viability resulting from proliferation, as sustained Raf/ER activation blocked wt and nfkb1^−/− B-cell division (Figure S3B), as previously reported for fibroblasts.⁴⁸  The finding that the Bim-dependent death of nfkb1^−/− B cells was associated with a defect in transient Bim degradation accompanying TLR4 activation, suggested that this death is normally prevented by ERK inhibiting Bim function during a critical period after LPS stimulation. This hypothesis was examined by monitoring apoptosis in LPS-treated nfkb1^−/− B cells expressing Raf/ER, where delayed ERK activation was achieved by adding 4-HT to cultures at specific times after LPS stimulation (Figure 4D). Remarkably, activation of Raf/ER within 4 hours of initiating LPS stimulation was necessary to prevent the enhanced death of nfkb1^−/− B cells; activating Raf thereafter had no significant impact on apoptosis. Collectively, these biochemical and genetic data indicate that the abnormally high death of LPS-stimulated nfkb1^−/− B cells is a consequence of a defect in ERK activation.

Despite many signals inducing ERK-dependent phosphorylation of Bim,²⁰  this remained to be established for TLR signaling. Two-dimensional gel electrophoresis combined with Western blotting was used to examine Bim modification in LPS-stimulated B cells. Before stimulation, Bim_EL, the major form of Bim expressed in B cells,⁴⁴  exhibited an identical pattern of 3 isoforms (Figure 5A, labeled 1-3) in wt and nfkb1^−/− B cells. After LPS stimulation for 2 hours (Figure 5A), the time when Bim levels normally decrease, 3 additional negatively charged Bim_EL isoforms (4-6) were detected in wt cells, whereas only isoforms 1 to 4 were seen in nfkb1^−/− B cells. ERK activation was required for the induction of isoforms 5 and 6 given they were absent in LPS plus PD98059-treated wt B cells (Figure 5B). The LPS-induced ERK-dependent isoforms were detectable within 60 minutes, peaking in abundance at approximately 2 hours and then gradually decreased until they were undetectable by 6 hours (Figure S4A). This reinforced that the kinetics of LPS-induced Bim phosphorylation reflect the transient activity of ERK. We also confirmed the relationship between Bim phosphorylation and ERK in W231 B lymphoma cells using the 4-HT induction of Raf/ER (Figures S3A, S4B). Together, these results establish that nfkb1^−/− B cells responding to LPS are defective in ERK-dependent Bim phosphorylation.

Engaging ERK via an alternate MAP3K should promote the survival of LPS-activated nfkb1^−/− B cells. Because BCR cross-linking activates ERK via c-Raf,⁴⁹  we examined whether BCR and TLR4 costimulation overcame the Bim phosphorylation and survival defects seen in LPS-treated nfkb1^−/− B cells. We first confirmed that c-Raf activation (S338 phosphorylation) and ERK phosphorylation were normal in nfkb1^−/− B cells after BCR cross-linking (Figure 5C). Survival was then determined for wt and nfkb1^−/− B cells activated with anti-IgM antibodies, LPS, or both stimuli (Figure 5D). BCR signals prevented the LPS-induced death of nfkb1^−/− B cells, with the pattern of Bim isoforms after BCR plus TLR4 costimulation comparable between wt and nfkb1^−/− B cells (Figure 5E). This establishes that B-cell survival linked to ERK-dependent Bim phosphorylation can be independently regulated via distinct MAP3Ks activated by different signals.

Despite a transient ERK-dependent drop in Bim levels in wt B cells, Bim expression is restored within 6 hours of initiating LPS activation (Figure 3B). This indicated that mechanisms distinct from ERK-dependent Bim turnover prevented Bim-induced apoptosis during sustained LPS stimulation. Because Bim interaction with Bcl-2-like prosurvival proteins¹⁸  has also been shown to promote cell survival, these associations were examined in LPS-stimulated B cells.

To ensure that relevant times were chosen to assess Bim and pro-survival protein interactions, Bcl-2-like protein expression was re-examined at early times in LPS-activated B cells. In the absence of changes in Bcl-2 and Mcl-1 levels (Figure 1C), attention focused on A1 and Bcl-x_L. In wt and nfkb1^−/− B cells, A1 levels increased within 2 hours and were sustained for at least 18 hours (Figures 1C, 6A). Bcl-x_L induction was slower, with a small increase in expression first seen after 4 hours (Figure 6A). Like A1, LPS-induced Bcl-x_L expression was sustained. IP-Western blots were then performed to examine Bim interaction with A1, Bcl-x_L, and Bcl-2 in LPS-treated wt and nfkb1^−/− B cells (Figure 6B). Interactions between Bim and A1 or Bcl-x_L were undetectable in quiescent B cells but increased after LPS stimulation. Although Bim levels were reduced in wt but not nfkb1^−/− B cells after 2 hours (Figure 6B, compare lanes 2 and 6), paradoxically, A1 association with Bim in nfkb1^−/− B cells was less compared with wt B cells. After 6 hours, the abundance of the A1/Bim complex was equivalent in cells of both genotypes. Bcl-x_L/Bim association was detected in wt and nfkb1^−/− cells after 6 hours and increased by 12 hours. Unlike the A1/Bim complex, the pattern of Bcl-x_L and Bim association was equivalent in wt and nfkb1^−/− B cells over 12 hours. Bcl-2 association with Bim and the abundance of this complex did not alter significantly in either cell type during LPS stimulation.

Because c-Rel is required for the transcriptional induction of A1 and bcl-x_l in LPS-stimulated B cells, we determined whether the prosurvival role of c-Rel during TRL4 signaling involved neutralizing Bim proapoptotic activity. To this end, c-rel^−/−bim^−/− mice were generated and B-cell survival in response to LPS examined (Figure 6C). As expected,⁸  LPS-induced apoptosis was abnormally elevated in c-rel^−/− B cells but was blocked by the loss of Bim, demonstrating that c-Rel, like NF-κB1, promotes TLR4-activated B-cell survival by neutralizing Bim.

The findings that TLR4 signals trigger Bim degradation and increased binding of Bim to A1 and Bcl-x_L raised the question as to whether these 2 processes are part of one pathway or represent distinct pathways for neutralizing Bim. This was tested by comparing LPS-induced apoptosis in nfkb1^−/−, c-rel^−/−, and nfkb1^−/−c-rel^−/− B cells (Figure 6D). Although LPS-induced apoptosis was abnormally elevated in both nfkb1^−/− (lane 6) and c-rel^−/− (lane 7) B cells, tellingly death was even higher for nfkb1^−/−c-rel^−/− B cells (lane 8). bcl-2^T expression blocked the death of LPS-stimulated nfkb1^−/−c-rel^−/− B cells, confirming it involved the Bcl-2–regulated intrinsic pathway. This shows that NF-κB1 and c-Rel promote survival of LPS-stimulated B cells by nonredundant mechanisms that converge to neutralize Bim.

Here we show that NF-κB1 enhances the survival of TLR4-stimulated B cells by regulating the MAP3K Tpl2 required for ERK-dependent Bim phosphorylation and degradation. Coupled with the c-Rel–dependent induction of A1 and bcl-xl that result in the formation of A1/Bim and Bcl-x_L/Bim complexes after LPS stimulation, these distinct NF-κB1 and c-Rel–regulated survival mechanisms synergize to neutralize Bim proapoptotic activity in TLR4-activated B cells.

Loss of p105NF-κB1 increases spontaneous plus TLR4 activation induced B-cell apoptosis.⁸  Bcl-2 transgene expression prevents sustained spontaneous nfkb1^−/− B-cell death; loss of Bim, although blocking the initial apoptosis, affords only partial protection thereafter. Similarly, constitutive ERK activation only inhibits spontaneous wt and nfkb1^−/− B-cell death during the first 24 hours. Therefore, the initial phase of spontaneous nfkb1^−/− B-cell death is Bim- and ERK-dependent, whereas subsequent death involves other BH3-only proteins that are not ERK regulated. In contrast, elevated TLR4-induced nfkb1^−/− B-cell apoptosis is blocked by loss of Bim or sustained ERK signaling. However, death in LPS-stimulated nfkb1^−/−bim^−/− B-cell cultures, although equivalent to that of bim^−/− and wt cells (Figure 2), remains higher than either wt or nfkb1^−/− B cells expressing Bcl-2^T (Figure 1A) and wt cells expressing activated Raf/ER (Figure 4C). Collectively, these findings are best reconciled in a model where multiple BH3-only proteins contribute to TLR4-induced B-cell death, with their apoptotic function neutralized by Bcl-2 or constitutive ERK signaling. Transgenic Bcl-2, unlike endogenous Bcl-2, would presumably be expressed at levels sufficient to neutralize multiple LPS-activated BH3-only proteins. Whereas Bcl-2 transgene expression is equivalent in wt and nfkb1^−/− B cells (Figure S1), a direct comparison of endogenous mouse and transgenic human Bcl-2 levels in these cells is precluded by an inability of the antibodies used here to recognize both forms of Bcl-2. Why Bim appears to be more important in the LPS-induced death of nfkb1^−/− B cells than wt cells is unclear but may indicate that the hierarchical importance of particular BH3-only proteins in cell death varies on different genetic backgrounds.

Here we demonstrate that Tpl2/ERK signaling in B cells initiated through TLR4 leads to transient Bim phosphorylation and degradation, with disruption of this pathway resulting in the enhanced death of LPS-activated B cells. This represents a cell-type–specific mechanism for protecting cells from TLR4-induced apoptosis, given that the ERK defect in LPS-activated nfkb1^−/− macrophages^11,12  does not result in increased apoptosis (Figure S5). Transient ERK-dependent Bim degradation in B cells over a 3- to 4-hour period, initiated within approximately 60 minutes of LPS stimulation (Figure S5), points to this being a critical window for preventing Bim-mediated apoptosis. This conclusion is supported by Raf/ER activation of ERK in LPS-treated nfkb1^−/− B cells needing to occur within this time frame to protect these cells from enhanced death. This points to Bim activation at this early juncture in LPS stimulation triggering subsequent cell death via a “hit-and-run” mechanism (Figure 4D). Although ERK signaling prevents apoptosis in TLR4-activated nfkb1^−/− B cells by counteracting Bim, our data suggest that constitutive ERK signaling inhibits this apoptosis by acting on additional targets. One candidate, Bad, another BH3-only proapoptotic protein controlled by ERK,⁵⁰  was excluded from having a role in the spontaneous or TLR4 stimulation-induced death of wt or nfkb1^−/− B cells (Figure S6A,B). Furthermore, despite compelling biochemical and genetic data presented here for the importance of Tpl2/ERK in preventing B-cell death, we cannot exclude a role for p50NF-κB1 transcription in protecting cells against TLR4 proapoptotic signals.

The observation that Tpl2/ERK signaling only induces a transient drop in Bim levels during sustained LPS activation indicates that this mechanism alone probably does not fully protect B cells from Bim-induced apoptosis. The finding that exaggerated levels of death for LPS-stimulated c-rel^−/− B lymphocytes were blocked by loss of Bim, despite ERK activation and Bim phosphorylation being normal in c-rel^−/− B cells (A.B., unpublished data, July 2007), indicates that c-Rel promotes survival by inhibiting Bim via a different mechanism. Consistent with NF-κB1 and c-Rel neutralizing Bim in TLR4-activated B lymphocytes through nonredundant mechanisms, the death of LPS-stimulated nfkb1^−/−c-rel^−/− B cells was greater than that of nfkb1^−/− or c-rel^−/− B cells. The increased death of LPS-activated c-rel^−/− B cells coincides with a failure to up-regulate A1 and Bcl-x_L expression, which in wt cells leads to the generation of Bim/A1 and Bim/Bcl-x_L complexes. The binding of Bim to A1 precedes its binding to Bcl-x_L, consistent with the kinetics of A1 and Bcl-x_L induction. That A1 and Bcl-x_L bind Bim after LPS stimulation indicates that both proteins probably contribute to the inhibition of Bim. This is consistent with models in which the neutralization of multiple prosurvival Bcl-2 family members by BH3-only proteins is required to trigger apoptosis.^18,22  Surprisingly, less A1 bound Bim at the 2-hour time point in LPS-activated nfkb1^−/− B cells, despite Bim not being targeted for degradation in these cells. This unexpected finding may indicate that Bim phosphorylation enhances binding to A1; this remains to be determined.

Results presented here support an integrated signaling model in which ERK- and c-Rel–dependent pathways coordinated through IKK promote optimal survival of TLR4-activated B cells (Figure 7). IKK2 activation leads to the nuclear translocation of c-Rel dimers required for A1 and bcl-xl transcription, with elevated expression of these Bcl-2 prosurvival proteins leading to the formation of Bim/A1 and Bim/Bcl-x_L complexes. IKK2 also induces Tpl2-dependent ERK activation, resulting in Bim phosphorylation and turnover. Why both of these mechanisms are necessary to neutralize Bim is unclear. Furthermore, our results establish how Bim is inactivated in TLR4-stimulated B cells but do not shed light on how Bim activates Bax or Bak.

For example, ERK phosphorylation of Bim can inhibit Bim/Bax interaction,⁵¹  a finding consistent with a model whereby excess Bim in LPS-treated nfkb1^−/− B cells promotes cell death by directly binding and activating Bax. However, increased binding of A1 and Bcl-x_L to Bim in LPS-activated B cells is compatible with models in which Bim only binds to prosurvival proteins. Although we have failed to detect Bim and Bax interactions in B cells (R. Grumont, unpublished data, February 2007), it is unclear whether this result is real or instead reflects technical issues, such as the low abundance of these complexes in primary cells or differences in their stability in particular detergents.^52,53  Future work is required to resolve this issue and to ascertain the relationship between Bim phosphorylation and binding to Bcl-2–like proteins and whether multiple Bcl-2–like proteins are necessary to protect B cells from TLR4-induced apoptosis.

The authors thank Dr Bouillet for the provision of the bim^−/− mice and Dr L. O'Reilly for antibodies.

This work was supported by the Leukemia & Lymphoma Society (United States) and the National Health and Medical Research Council of Australia.

Contribution: A.B. and S.G. designed research; A.B. and R. Grumont performed most of the experiments; R. Gugasyan did the FACS analysis on splenic B cells; C.W. did the macrophage survival experiment; and A.B., A.S., and S.G. wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Steve Gerondakis, The Macfarlane Burnet Institute of Medical Research and Public Health, 85 Commercial Road, Prahran, Victoria 3004 Australia; e-mail: gerondakis@burnet.edu.au.