The Mastermind-like 1 (MAML1) Co-activator Regulates Constitutive NF-κB Signaling and Cell Survival

Abstract

Nuclear factor-κB (NF-κB)-based signaling regulates diverse biological processes, and its deregulation is associated with various disorders including autoimmune diseases and cancer. Identification of novel factors that modulate NF-κB function is therefore of significant importance. The Mastermind-like 1 (MAML1) transcriptional co-activator regulates transcriptional activity in the Notch pathway and is emerging as a co-activator of other pathways. In this study, we found that MAML1 regulates NF-κB signaling via two mechanisms. First, MAML1 co-activates the NF-κB subunit RelA (p65) in NF-κB-dependent transcription. Second, MAML1 causes degradation of the inhibitor of NF-κB (IκBα). Maml1-deficient mouse embryonic fibroblasts showed impaired tumor necrosis factor-α (TNFα)-induced NF-κB responses. Moreover, MAML1 expression level directly influences cellular sensitivity to TNFα-induced cytotoxicity. In vivo, mice deficient in the Maml1 gene exhibited spontaneous cell death in the liver, with a large increase in the number of apoptotic hepatic cells. These findings indicate that MAML1 is a novel modulator for NF-κB signaling and regulates cellular survival.

Introduction

Nuclear factor-κB (NF-κB)² signaling regulates diverse biological responses, including cell proliferation, survival, inflammation, and immunity (for review, see 1–4). Deregulated NF-κB signaling is associated with many disease states such as AIDS, asthma, arthritis, cancer, diabetes, muscular dystrophy, stroke, and viral infection.

NF-κB consists of homo- or heterodimers of members of the Rel family: RelA (p65), RelB, c-Rel, p105 and its processing product p50, and p100 and its processing product p52. These proteins contain a Rel homology domain, a conserved 300-amino acid domain within their N termini that is responsible for DNA binding and homo- or heterodimerization. The common active forms of NF-κB are RelA/p50 or RelA/p52 heterodimers. In its inactive state, NF-κB remains sequestered in the cytoplasm by members of the inhibitor IκB family. Although the IκB family consists of IκBα, β, γ (p105), δ (p100), ϵ and Bcl-3, the best studied and major IκB protein is IκBα. NF-κB-based signaling results from a variety of stimuli, including T cell receptor signals, cytokines, and viral and bacterial products. In the canonical pathway, an IκB kinase (IKK) complex is activated upon response to these stimuli, and two kinases in this complex, IKKα and IKKβ, phosphorylate IκB. Phosphorylation triggers IκB for ubiquitination by the Skp/Cullin/F-box-containing ubiquitin ligase complex, leading to the degradation of IκB by the 26 S proteasome. NF-κB subsequently becomes liberated from its interaction with IκB, rapidly translocates to the nucleus, and binds to its cognate DNA-binding site in the promoter or enhancer regions of specific NF-κB target genes. Thus, the result of NF-κB activation triggered from a myriad of cellular activators is highly regulated gene expression.

The biological and pathogenic importance of NF-κB signaling emphasizes the need to control its action tightly, both physiologically and therapeutically. Indeed, research in recent years has produced significant insights into regulation of the NF-κB signaling pathway. These studies have revealed that NF-κB regulation occurs at multiple levels, including signal-induced kinase cascades leading to IκB degradation, regulation of NF-κB nuclear translocation, and interaction with other signaling pathways that modulate transcriptional activation of NF-κB target genes. The interaction of NF-κB with other signaling pathways is particularly interesting and complex. For example, Notch receptor-mediated signaling is a critical developmental signaling pathway and has complicated cross-communications with NF-κB (for review, see 5, 6).

In this study, we reveal a novel function for Mastermind-like 1 (MAML1) in regulating the NF-κB signaling pathway based on cell culture-based studies and the analysis of Maml1-knock-out (ko) mice. MAML1 belongs to a family of three MAML transcriptional co-activators (for review, see 7), which were originally identified as essential co-activators for Notch receptors (8–10). Excitingly, recent studies have indicated that MAML1 has Notch-independent activities (for review, see 11), co-activating other transcription factors, including the muscle transcriptional factor MEF2C (12), p53 (13), and β-catenin (14). Here, we found that MAML1 interacts with nuclear RelA (p65) to promote NF-κB-dependent transcription events. MAML1 also interacts with NF-κB inhibitor IκBα and causes its degradation. Maml1-deficient mouse embryonic fibroblasts (MEFs) showed impaired TNFα-induced NF-κB responses and enhanced TNFα-mediated cellular cytotoxicity. In vivo, Maml1-ko mice exhibited ongoing hepatocyte cell death, and hepatocytes from Maml1-ko mice were hypersensitive to TNFα-mediated cell death. Our combined data indicate that MAML1 is a novel modulator for NF-κB signaling and regulates cellular survival.

EXPERIMENTAL PROCEDURES

Mice and Histology

Maml1-ko mice were generated and genotyped as described (12). Experiments were performed according to a protocol approved by the IACUC committee of the University of Florida. For routine histological analysis, tissue samples were fixed in Bouin's solution and paraffin-embedded. Tissue sections were then stained with hematoxylin and eosin. For immunofluorescence staining, tissue samples were fixed in 4% paraformaldehyde in phosphate-buffered saline and OCT-embedded. The terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assays were performed on frozen tissue sections using an In Situ Cell Death Detection kit, TMR red (Roche Applied Science).

Antibodies

Mouse anti-FLAG antibody (clone M2) and anti-β-actin antibody were from Sigma. Hemagglutinin (HA) (clone HA.11) was from Babco. IκBα (9242), phospho-IκBα Ser³² (9241), MAML1 (4608), NF-κB p65 (sc-109), GFP (sc-9996), anti-Myc (sc-40), goat anti-mouse IgG-horseradish peroxidase (sc-2302), and goat anti-rabbit IgG-horseradish peroxidase (sc-2301) were from Santa Cruz Biotechnology. Alexa Fluor 594 donkey anti-rabbit IgG (A-21207) and anti-mouse IgG (A-21203) were from Invitrogen.

Plasmids

GFP-p65, pNF-κB-luc, and pEF-RL were gifts from Dr. Warner C. Greene and Dr. Lin-feng Chen (15). DsRed-p50 and EGFP-c_Rel were gifts from Dr. Barbara A. Osborne (16). HA-tagged IκBα, both wild-type and superrepressor (SR) S32A/S36A mutant, under the control of β-actin promoter in pRC vector, were described previously (17). FLAG-tagged MAML1, MAML1(1–302), MAML1(Δ71–301)nls in pFLAG-CMV2 vector, and FLAG-tagged MAML1 in the retroviral pLNCX vector were described previously (12). pSG5-luc (Promega) is a firefly luciferase reporter plasmid that contains five copies of the Gal4-binding sites upstream of a minimal TATA box. GFP-IκBα was made by PCR-amplifying and subsequent cloning of IκBα cDNA into pEGFP-C1 vector. The Gal4 DB-RelA fusion was constructed by cloning RelA cDNA into the pFA-CMV2-DB vector. The shMAML1 sequence was cloned in pSuperRetro vector (Oligoengine), and the target sequence is 5′-GAGGAATCTTGACAGCGCC-3′.

Cell Culture, Retroviral Transduction, Transfection, Reporter Assays, and Immunofluorescence Staining

HeLa, 293T, and U20S cells and MEFs were cultured in Dulbecco's modified Eagle's medium with 10% fetal calf serum. Retroviral production and infection were described previously (12). Superfect and Effectene transfection reagents (Qiagen) were used. Immunofluorescence staining and luciferase-based reporter assays were performed as described previously (9).

Western Blotting and Immunoprecipitation

Cells were serum-starved for 18 h and then treated with various concentrations of human TNFα (Sigma) as indicated in the figure legends. A nuclear extract kit (Active Motif) was used to prepare both the cytoplasmic and nuclear fractions. The procedures for whole cell protein extract preparation, Western blotting and immunoprecipitation were described previously (9).

Electrophoretic Mobility Shift Assays

For each assay, 5 μg of nuclear extracts were incubated with double-stranded oligonucleotides labeled with [γ-³²P]ATP (10 μCi/ml) in binding buffer (10 mm Tris-HCl (pH 7.6), 50 mm NaCl, 1 mm EDTA, 1 mm dithiothreitol, 0.1% Triton X-100, 12.5% glycerol, and 0.1 μg/μl poly(dI-dC)). The oligonucleotide probe contains a classical NF-κB binding sequence (5′-AGTTGAGGGGACTTTCCCAGGC-3′) from the κ light chain enhancer. For competition experiments, a 100-fold excess of cold oligonucleotide was used. For the supershift assay, the nuclear extract was incubated with anti-p65 antibody for 30 min at room temperature before the addition of the labeled probe. After incubation with the probe for 30 min, the reaction mixture was analyzed on a 5% nondenaturing acrylamide gel. The gel was then dried and exposed for autoradiography.

TNFα-induced Cytotoxicity Assay

Primary hepatocytes, primary MEFs and transduced HeLa cells, were serum-starved for 18 h and then treated with various concentrations of TNFα in Dulbecco's modified Eagle's medium for 24 h. Trypan blue exclusion assays were used to quantify cell viability, and then the ratio of the number of nonviable cells to that of total cells was calculated.

RESULTS

Maml1-deficient Mice Exhibit Increased Cell Death in the Liver

We previously showed that the Maml1-ko mice fail to thrive and die within 10 days after birth (12). We determined that these mice exhibit a muscular dystrophy-like defect (12) and are unable to generate a type of mature B cells, marginal zone B cells (18). To characterize other potential defects that account for growth retardation and early death in the Maml1-ko mice, we performed further histological analyses and observed multiple regions of cell death in the livers of Maml1-ko mice (Fig. 1A). In striking contrast, regions of cell death were not observed in livers from wild-type (WT) littermates. The lesions became more apparent after neonatal day 3 (P3), appearing to grow in severity until death of the mice, with an increase in size and the number of necrotic regions. These data suggested that liver failure could be a cause of death for these mice.

FIGURE 1.

Maml1-deficient mice exhibit increased cell death in the liver. A, regions of cell death are present specifically in the Maml1-null livers. Frozen liver sections from Maml1-ko and their wild-type (wt) control littermates at neonatal day 6 (P6) were stained with hematoxylin and eosin. The upper panels are low magnifications, and lower ones are high magnifications. More than 10 ko mice and their control littermates at various ages were examined. B, the number of apoptotic cells increases in the Maml1-null livers. TUNEL assays were performed on the frozen tissue sections from Maml1-ko and their WT control littermates at neonatal day 1 (P1). The sections were counterstained with DAPI to label nuclei. C, the number of apoptotic cells increases due to Maml1 deficiency in both neonatal day 1 (P1) and day 3 (P3). The average numbers of apoptotic cells are presented based on triplicate staining sections from liver samples of at least two mice/group in two independent experiments. Error bars, S.E.

Certain hepatocytes within the lesions clearly exhibited enlarged cytoplasms in the hematoxylin and eosin-stained section, suggesting that they were necrotic. To investigate whether Maml1-ko hepatocytes might also be more susceptible to apoptosis, we performed TUNEL assays and found that there was a significant increase in the number of apoptotic (TUNEL-positive) cells in the Maml1-ko liver sections compared with that in the controls in both neonatal day 1 and day 3 (Fig. 1, B and C). The difference is not as dramatic at day 3 compared with the earlier day 1, suggesting that more apoptosis occurs at the earlier stage. These data indicate that Maml1 deficiency results in an increase in cell death in the liver, in part, due to an increase in apoptotic cells.

The liver phenotypes of the Maml1-ko mice are reminiscent of several knock-out models that are defective in the NF-κB pathway. For instance, knock-out of the components of the NF-κB pathway including RelA (p65) (19) and IKKβ (20) caused severe cell death in the liver. The death of hepatocytes is believed to be caused by impaired NF-κB responses of these cells to endogenous production of TNF because the mice that lack both RelA (p65) and TNF genes are viable and have normal livers (21). Therefore, normal NF-κB activity is required to prevent TNFα-mediated apoptosis in hepatocytes. Based on the resemblance of the liver phenotypes of our Maml1-ko and the knockouts of NF-κB components, we thus investigated whether Maml1 plays any role in modulating NF-κB signaling and whether the ablation of the Maml1 gene results in defective NF-κB signaling and sensitizes the cells to TNFα-mediated cytotoxicity.

NF-κB-dependent Transcription Is Regulated by MAML1 Co-activation

To determine whether MAML1 regulates NF-κB signaling, we first investigated whether MAML1 affects NF-κB target gene expression by monitoring the activities of a luciferase reporter containing six copies of artificial NF-κB-responsive elements as readouts (Fig. 2A). We found that in U20S cells, the NF-κB-responsive reporter was dramatically activated by MAML1 (M1) expression without further stimulation with cytokines, and the activation increased in a dose-dependent manner (Fig. 2B). To determine the specific domain(s) of the MAML1 co-activator required for NF-κB activation, we tested two MAML1-truncated mutants (Fig. 2, B and C). One is M1(Δ71–301)nls, lacking residues 71–301 but containing an added nuclear localization sequence. This mutant was unable to activate the NF-κB-responsive reporter. Because this mutant contains a p300/CREB-binding protein-binding site (8), MAML1-mediated p300 binding may be critical for MAML1 activation of NF-κB signaling. The other mutant was M1(1–302), which contains the N-terminal basic domain and p300 binding domain but not the C-terminal transcriptional domain (TAD). M1(1–302) also was unable to activate the NF-κB reporter. These data indicate that MAML1 co-activates NF-κB-dependent transcription, and it is likely that MAML1-mediated activation of NF-κB signaling may require both the p300 binding and TAD domains of the MAML1 protein.

FIGURE 2.

MAML1 enhances NF-κB-dependent transcription. A, a diagram for a NF-κB luciferase reporter containing six copies of responsive NF-κB binding sites (pNF-κB-luc) is shown. B, MAML1 expression activates a NF-κB-responsive reporter in a dose-dependent manner, and the MAML1 mutants that lack either the p300 binding domain or the C-terminal TAD fail to activate NF-κB transcription. U20S cells were transfected with 5 ng of Renilla luciferase plasmid as internal control, 0.2 μg of pNF-κB-luc luciferase reporter, and increasing amounts of expression plasmids encoding FLAG-tagged full-length (FL) or truncated MAML1. Cell lysates were prepared at 44–48 h after transfection for luciferase assays, and the pNF-κB-luc reporter activities are expressed as fold activation relative to cells not expressing MAML1. The data presented were pooled from three independent experiments. Error bars, S.E. C, a diagram of full-length and truncated MAML1 mutants is shown.

MAML1 Cooperates with RelA (p65) to Activate NF-κB-dependent Transcription

To determine the mechanisms underlying the transcriptional activation of NF-κB signaling by MAML1, we first determined the possibility of in vivo interactions of MAML1 and NF-κB transcription factors by examining potential co-localization of MAML1 with NF-κB transcription factors p65, p50, and c-Rel. We found that when co-expressed with MAML1, only RelA (p65) dramatically changed its subcellular localization from the cytoplasm to the nuclear dots (Fig. 3A) and co-localized with MAML1 (supplemental Fig. 1). The specific MAML1 and RelA (p65) interaction was further supported by immunoprecipitation studies showing that MAML1 and RelA (p65) co-immunoprecipitate at endogenous and exogenous levels (Fig. 3, B and C). We further showed that the MAML1 C-terminal TAD domain 303–1016 amino acids might be required for binding to RelA (p65) because the MAML1(1–302) mutant that lacks this domain was unable to bind to RelA (p65) (Fig. 3D and supplemental Fig. 2). Importantly, MAML1 cooperates with RelA (p65) in the activation of the NF-κB responsive reporter (Fig. 3E). In contrast, the mutant MAML1(1–302), a mutant lacking RelA (p65) binding but retaining p300 binding, acted as a dominant negative to inhibit RelA (p65)-induced NF-κB promoter (Fig. 3F), possibly competing away a crucial transcriptional co-activator for NF-κB signaling, the p300 transcriptional co-activator (22). Moreover, MAML1 was able to promote the ability of RelA (p65) when fused to the Gal4 DB to activate a luciferase reporter that contains Gal4-binding sites in the promoter (Fig. 3G), indicating a co-activator role of MAML1 for RelA (p65). All of these data together indicate that MAML1 interacts with RelA (p65) and co-activates RelA (p65)-mediated transcription, providing one molecular mechanism for MAML1-mediated NF-κB transcriptional activation.

FIGURE 3.

MAML1 interacts with RelA (p65) and cooperates with RelA (p65) to activate NF-κB-responsive transcription. A, MAML1 causes redistribution of RelA (p65) from the cytoplasm to the nucleus. 293T cells were cultured on poly-d-lysine-treated coverslips and transfected with GFP-tagged RelA (p65) with or without the co-transfection of FLAG-tagged MAML1. Cells were then photographed at 24-h after transfection. B, endogenous RelA (p65) and MAML1 co-immunoprecipitate. HeLa cell nuclear extracts were used to immunoprecipitate (IP) with anti-MAML1 antibodies or control IgG (as a negative control). The nuclear extract input and immunoprecipitates were analyzed by Western blotting for RelA (p65) and MAML1. C, exogenous MAML1 and RelA (p65) co-immunoprecipitate. 293T cells were co-transfected with various combinations of expression constructs expressing GFP-tagged RelA (p65) and/or FLAG-tagged MAML1 as indicated. For each transfection, the total plasmid DNA was kept constant with the backbone vectors. Cell lysates were prepared for immunoprecipitation with M2 beads (anti-FLAG). Both whole cell lysates (WCL) and immunoprecipitates were separated on SDS-polyacrylamide gels and analyzed by Western blotting for GFP-tagged RelA and FLAG-tagged MAML1. D, the C-terminal region (amino acids 303–1016) of MAML1 is required for RelA (p65) binding. 293T cells were co-transfected with GFP-tagged p65 and full-length or truncated MAML1, and then various forms of MAML1 proteins were immunoprecipitated using anti-FLAG antibodies (M2) and blotted for the presence of GFP-RelA. E, MAML1 cooperates with RelA (p65) to activate the NF-κB-responsive promoter. Assays were performed as described in Fig. 2B, except that various amounts of RelA(p65) expression vectors were co-transfected with MAML1. F, MAML1(1–302) functions as a dominant negative mutant to inhibit RelA (p65)-mediated NF-κB signaling. Assays were performed as described in Fig. 2B, except that various amount of p65 expression vectors were co-transfected with MAML1(1–302). G, MAML1 enhances the ability of Gal4 DB-RelA (p65) fusion to activate a luciferase reporter that contains multiple Gal4 binding sites. Assays were performed as described in Fig. 2B, except that various amounts of Gal 4 DB-RelA(p65) expression vectors were co-transfected with MAML1, and the Gal4-responsive reporter was used. E–G, error bars, S.E.

MAML1 Interacts with IκBα and Causes the Degradation of IκBα

Signal-stimulated phosphorylation and ubiquitination of the inhibitor IκBα is a key process for the liberation and nuclear translocation of NF-κB, leading to subsequent target gene transcription. Also, IκBα has a nuclear function as a repressor of NF-κB proteins (23–25). We showed that MAML1 is predominantly a nuclear protein (supplemental Fig. 4). Here, we tested whether MAML1 functionally interacts with IκBα in the nucleus. We transfected GFP-tagged IκBα into 293T cells and treated cells with the vehicle control DMSO or leptomycin B, an inhibitor of nuclear export. We found that IκBα normally is localized to the cytoplasm but is retained in the nucleus when treated with leptomycin B (Fig. 4A), indicating that IκBα is a protein shuttling between the cytoplasm and the nucleus. When co-expressed with MAML1, IκBα exhibited nuclear localization and co-localized with MAML1 in the nuclear speckles even without the leptomycin B treatment (Fig. 4A), suggesting that MAML1 might interact with nuclear IκBα and helps retain IκBα in the nucleus through their interaction. Indeed, the interaction of MAML1 and IκBα appears to be direct, as shown by GST pulldown assay (Fig. 4B).

FIGURE 4.

MAML1 interacts with IκBα and causes IκBα degradation. A, MAML1 and IκBα co-localize in the nucleus. 293T cells were transfected with GFP-tagged IκBα with or without the co-transfection of FLAG-tagged MAML1 (M1). At 24 h after transfection, cells were treated with either DMSO or leptomycin B (LMB) at 10 ng/ml overnight. Cells were finally fixed and photographed. B, MAML1 interacts with IκBα in a GST pulldown assay. GST-IκBα or GST was used to pull down the in vitro translated ³⁵S-MAML1. C, MAML1 overexpression results in reduced expression levels of endogenous IκBα. Stable HeLa cells transduced with MAML1 retroviruses (MAML1: HeLa_pLNCX_MAML1) and the control cells (Vec: HeLa_pLNCX vector) were subjected to Western blot analyses for endogenous IκBα, MAML1, and β-actin expression. Here, MAML1 antibody recognizes both endogenous and exogenous MAML1. D, MAML1 causes IκBα degradation, and both the p300 binding domain and the TAD of MAML1 are required for this activity. 293T cells were transfected with HA-tagged IκBα and FLAG-tagged full-length (FL) or truncated MAML1, and then the expression levels of IκBα and MAML1 were detected by Western blot analyses with IκBα and anti-FLAG antibodies. β-Actin expression was used as a loading control. E, MAML1 fails to cause the degradation of IκBα SR, and MAML1-induced IκBα degradation is blocked by a proteasome inhibitor. 293T cells were transfected with either WT or the SR form of HA-tagged IκBα in the presence or absence of FLAG-tagged full-length MAML1. Cells were split into two groups on the second day and treated with MG132 or the vehicle control DMSO for 8 h. IκBα, MAML1, and β-actin expression was determined by Western blot analyses using anti-HA, anti-FLAG, and anti-β-actin antibodies. The relative band intensities of IκBα were quantitated relative to β-actin expression levels. F, MAML1 promotes IκBα ubiquitination. HeLa cells were co-transfected with HA-tagged IκBα, FLAG-tagged MAML1, and Myc-tagged ubiquitin (Ub) and treated with MG132 on the second day for 8 h before cell lysates were harvested. IκBα was subsequently immunoprecipitated (IP) with HA antibodies and blotted for anti-Myc antibodies to detect polyubiquitinated IκBα species. The total lysates were also blotted for anti-HA, anti-FLAG (M2), and β-actin antibodies.

Because IκBα is able to repress the function of NF-κB proteins in the nucleus, we next determined whether the interaction of MAML1 and IκBα altered IκBα expression levels. To evaluate the effect of MAML1 on endogenous IκBα protein levels, we established a stably transduced HeLa cell line with MAML1 retroviruses to achieve MAML1 overexpression (Fig. 4C). We found that increased MAML1 expression results in reduced level of endogenous IκBα (Fig. 4C). To determine the effect of MAML1 expression on exogenous IκBα expression levels, we co-transfected 293T cells with HA-tagged IκBα and various forms of FLAG-tagged MAML1 and analyzed IκBα protein levels by Western blotting. We found that the protein levels of transfected and endogenous IκBα were reduced significantly in the presence of MAML1 co-expression (compare lane 2 with lane 1 in Fig. 4D), indicating that MAML1 causes IκBα degradation. The MAML1 mutants that have a deletion of the p300 binding domain (Δ71–301)nls or of the C-terminal TAD domain (MAML1(1–302)) did not alter IκBα expression effectively (lanes 3 and 4 in Fig. 4D); therefore, both domains are required for MAML1 to regulate IκBα expression. These data indicate that MAML1 mediates the reduced expression levels of both endogenous and exogenous IκBα.

Because MAML1 has been implicated in post-translational modification of its interacting proteins and IκBα undergoes phosphorylation for subsequent ubiquitination and degradation in response to signals, we hypothesize that MAML1 might cause IκBα post-translational modification. First, to investigate whether MAML1-mediated IκBα degradation is dependent on the phosphorylation status of IκBα, we used a SR IκBα that carries mutations on two serine residues that are critical for its phosphorylation by IKKs and the subsequent ubiquitination and degradation (S32A and S36A). We transfected 293T cells with various combinations of MAML1 and WT or SR IκBα for 24 h and then treated cells with MG132, a proteasome inhibitor (or the vehicle control DMSO) for 8 h. We found that MAML1 significantly reduced the expression of WT IκBα (compare lane 3 with lane 2 in Fig. 4E) but not SR IκBα (compare lane 5 with lane 4 in Fig. 4E). Therefore, Ser³² and Ser³⁶ sites of IκBα are critical for MAML1-mediated IκBα degradation, suggesting a possible role for MAML1 in inducing IκBα phosphorylation. When cells were treated with MG132, IκBα expression was not significantly reduced by MAML1 compared with DMSO treatment (compare lane 8 with lane 3 in Fig. 4E), indicating that MAML1-induced IκBα degradation is through a proteasome-mediated pathway.

Finally, we tested whether MAML1 promotes IκBα ubiquitination. Indeed, when HeLa cells were co-transfected with HA-tagged IκBα, FLAG-tagged MAML1, and Myc-tagged ubiquitin and then treated with MG132, polyubiquitinated IκBα species were readily detected after IκBα immunoprecipitation (Fig. 4F, lane 3), indicating that MAML1 promotes IκBα ubiquitination. The above data combined indicate that MAML1 enhances IκBα ubiquitination and degradation, which would lead to enhanced activities of nuclear NF-κB.

Maml1 Deficiency Results in Impaired NF-κB Responses

To determine whether Maml1 deficiency results in defective NF-κB responses, we first introduced the NF-κB-responsive promoter reporter into WT and Maml1-ko MEFs and treated cells with TNFα. We found that although WT MEFs showed dose-dependent activation of the NF-κB-responsive promoter in response to TNFα treatment, Maml1-ko MEFs had significantly decreased activation (Fig. 5A), indicating that there is impaired activation of NF-κB signaling. We further showed that the expression of human MAML1 in Maml1-ko MEFs enhanced the overall NF-κB signaling but not up to the level that exhibited by WT MEFs (Fig. 5B), indicating that reintroduction of human MAML1 expression only resulted in partial rescue in this assay.

FIGURE 5.

Maml1-deficient MEFs display defective TNFα-induced NF-κB responses. A, Maml1 deficiency resulted in reduced activation of the NF-κB-responsive transcription in response to TNFα treatment. Both WT and Maml1-deficient MEFs in the 10-cm dishes were transfected with 2.5 μg of a NF-κB-responsive luciferase reporter along with 50 ng of a Renilla luciferase plasmid. Each group of cells was then split into the 12-well plates on the second day and grown for additional 8 h before being subjected to serum starvation for 18 h. Cells were then stimulated with various concentrations of TNFα for 8 h, and cell lysates were harvested for luciferase assays. The reporter activity of Maml1-ko MEFs without TNFα treatment was defined as 1.0. B, expression of human MAML1 enhanced overall TNFα-induced NF-κB-responsive transcription, but only partially rescued the loss of Maml1 function in MEFs. Assays were performed as described in A, except that either MAML1 expression vectors or empty vectors were co-transfected in the Maml1-ko MEFs. The reporter activity of Maml1-ko MEFs without TNFα treatment and with vector was defined as 1.0. A and B, error bars, S.D. C, Maml1 deficiency results in changes in the levels of pIκBα and total IκBα in response to TNFα treatment. Both WT and ko MEFs were serum-starved for 18 h and then treated with 10 ng/ml TNFα for various times. Cell lysates were collected for Western blot analyses for p-IκBα, total IκBα, and β-actin levels. D, Maml1 deficiency results in reduced DNA-binding RelA (p65). Electrophoretic mobility shift assays were performed with the nuclear extracts prepared from WT and Maml1-ko MEFs at various time points after TNFα treatment (10 ng/ml). Arrows mark the positions of p65-specific DNA complexes, as the band at this position can be further supershifted by the addition of anti-p65 antibodies (supplemental Fig. 3).

To examine the biochemical changes in the NF-κB pathway directly, we examined the levels of p-IκBα and total IκBα at various time points of TNFα-induced NF-κB responses. By Western blot analysis, we found that in WT MEFs, p-IκBα levels significantly increased at the 5-min time point and then reduced over the 2-h period, which correlates with the degradation of the total IκBα at this time point and then gradual restoration of normal IκBα level within 60 min. On the other hand, Maml1 deficiency led to an increase in basal endogenous IκBα levels (compare lane 7 with lane 1 in Fig. 5C) and a significant reduced level of p-IκBα at the 5-min time point and incomplete degradation of IκBα at the 5-, 15-, and 30-min time points. These data indicate that Maml1 affects dynamic changes in the levels of IκBα phosphorylation and protein levels in response to TNFα treatment.

Because the DNA-bound RelA (p65) levels are often used as an indicator of active NF-κB signal, we analyzed DNA-bound forms of RelA (p65) by performing electrophoretic mobility shift assays. We found that Maml1 deficiency resulted in overall reduced levels of DNA-bound RelA (p65) compared with the WT controls after TNFα treatment (Fig. 5D), further indicating that MAML1 deficiency caused reduced NF-κB activities in response to TNFα.

Maml1 Expression Levels Are Inversely Correlated with Cellular Sensitivity to TNFα-induced Cytotoxicity

Previous research indicates that normal NF-κB response is required to protect cells from TNF-induced cytotoxicity (19, 21). To determine whether there is any biological activity associated with the function of MAML1 in the NF-κB pathway, we examined the role of MAML1 in regulating TNFα-induced cell death by analyzing the effects of Maml1 deficiency or gain-of-function of MAML1 on TNFα-induced cell death.

Because Maml1-deficient mice exhibited a large degree of cell death in the liver, which was correlated with the higher levels of apoptotic cells (Fig. 1), we wanted to determine whether hepatocytes are more sensitive to TNFα treatment due to Maml1 deficiency. We isolated primary hepatocytes from Maml1-ko and their WT control littermates at neonatal day 1 and determined the sensitivity of these cells to TNFα-induced cytotoxicity by treating cells with various concentrations of TNFα and then quantifying the live cells using trypan blue exclusion assays. We found that Maml1-null cells were more prone to TNFα-induced cell death as indicated by a significant increase in dead cells in the Maml1-ko samples (Fig. 6A). Similar responses were also found in the Maml1-ko MEFs (Fig. 6B) and HeLa cells with retroviral-based short hairpin RNA-mediated MAML1 knockdown (Fig. 6C). These data indicate that the loss or reduced MAML1 expression results in defective NF-κB signaling and sensitizes the cells for TNFα-mediated cytotoxicity. On the other hand, we found that MAML1-overexpressing HeLa cells after transduced with retroviral-based MAML1 showed significant reduced cell death compared with cells transduced with an empty vector (Fig. 6D). These data indicate that increased MAML1 expression has protective activity against TNFα-induced cell death.

FIGURE 6.

MAML1 expression levels are inversely correlated with cellular sensitivity to TNFα-induced cytotoxicity. A, primary Maml1-ko hepatocytes are more sensitive to TNFα-mediated cytotoxicity. Primary hepatocytes were isolated, serum-starved for 18 h, and treated with 0, 20, or 40 ng/ml TNFα for 24 h. Nonviable cells were determined using trypan blue exclusion assays and are presented in the graph. B, Maml1-ko MEFs are more sensitive to TNFα-mediated cytotoxicity. Assays are same as described above. C, HeLa cells with Maml1 knockdown are more sensitive to TNFα-mediated cytotoxicity. Stably transduced HeLa MAML1 knockdown cells (infected with pSR_shRNA MAML1 retroviral viruses and selected with puromycin) along with the control HeLa pSR vec cells (infected with empty retrovirus) were serum-starved for 18 h and then subjected to TNFα treatment for 24 h with 0 or 10 ng/ml TNFα. D, HeLa cells with Maml1 overexpression are more resistant to TNFα-mediated cytotoxicity. Stably transduced MAML1 HeLa cells were compared with the control infected with empty vector. Assays are same as described above. Error bars, S.E.

DISCUSSION

The present study demonstrates that MAML1 is a novel regulator for constitutive NF-κB signaling. We found that MAML1 effectively enhances NF-κB transcription via two mechanisms. First, MAML1 co-activates RelA (p65)-mediated transcription. Second, MAML1 enhances the degradation of IκBα. The regulatory function of MAML1 in NF-κB signaling is supported by its ability to regulate cell survival because Maml1 deficiency results in enhanced cell death in the liver, which is correlated with defective NF-κB responses and enhanced sensitivity to TNFα-induced cytotoxicity in primary hepatocytes and MEFs.

Previously, it was shown that NF-κB is required for hepatocyte survival under stress-inducing conditions. For instance, a normal NF-κB response is necessary to protect hepatocytes from endogenous TNF injury in vivo. This is supported by the evidence that loss of the RelA (p65) subunit results in massive hepatocyte apoptosis with embryonic lethality between embryonic days 14 and 15 (19). TNFα treatment triggers apoptosis and/or necrosis of hepatocytes in vivo (26–28). However, mice that are deficient for both RelA (p65) and TNF genes are viable and have normal livers (21), indicating that RelA(p65)-mediated antiapoptotic signals prevent cell death from TNF injury in vivo. Also, the induction of NF-κB is required for liver regeneration (29, 30) because loss of NF-κB activities after partial hepatectomy resulted in massive hepatocyte apoptosis. We showed in this study that Maml1 deficiency resulted in increased cell death in the livers and isolated primary Maml1-null hepatocytes and MEFs have impaired NF-κB response and are sensitive to TNFα cytotoxicity. Moreover, using cell culture models where MAML1 has been either overexpressed or knocked down, we found that MAML1 expression levels are inversely related to cell death induced by TNFα treatment. However, the exact downstream mechanisms by which MAML1 regulates TNFα-induced cell death are unclear at the present. In light of the role of MAML1 in TNFα-induced cell death and its ability to regulate NF-κB signaling, it will be important in the future to dissect the detailed molecular mechanisms and assess the contribution of the NF-κB pathway in mediating MAML1 function in cell death pathway. Nonetheless, our data indicate that MAML1 is a novel regulator for constitutive NF-κB signaling events, and the impaired NF-κB response due to the Maml1 deficiency may provide direct explanations for the enhanced cell death observed in the livers of the Maml1-ko mice.

Mechanistically, we found that MAML1 enhances NF-κB-dependent transcription via functional interactions with IκBα and RelA (p65). In unstimulated cells, the NF-κB complex is inhibited by IκBα proteins, which inactivate NF-κB by trapping it in the cytoplasm. Phosphorylation of serine residues on the IκB proteins by IκB kinases marks them for destruction via the ubiquitination pathway, thereby allowing activation and nuclear translocation of the NF-κB complex. Also, IκBα was previously shown to have a nuclear function as a repressor of NF-κB proteins (23–25). Here, we found that MAML1 interacts with IκBα in the nucleus, leading to IκBα ubiquitination and degradation. Moreover, SR IκBα, which is a phosphorylation-defective mutant and can cause cells to be unresponsive to stimuli, is resistant to MAML1-induced degradation. All of these data support that a role for MAML1 in IκBα phosphorylation and subsequent ubiquitination.

Currently, the mechanism underlying MAML1 regulation of IκBα is unclear; however, MAML1 was shown to cause phosphorylation of its interacting partners including Notch, p300, and MEF2C, although the responsible kinases are not yet defined (8, 12). We hypothesize that MAML1 might interact with certain IκBα kinase(s) in the nucleus which are capable of phosphorylating IκBα and affecting IκBα stability. MAML1-induced IκBα phosphorylation and degradation could lead to enhanced NF-κB signaling. Therefore, it will be important in the future to determine whether the MAML1-containing complex has such IκBα kinase activities and if so, what specific kinase(s) contribute to IκBα phosphorylation and degradation.

Moreover, we identified a second mechanism that accounts for MAML1-mediated promoted activities of NF-κB signaling. MAML1 also regulates NF-κB at the transcriptional level and is able to activate NF-κB-dependent transcription via its interaction with RelA (p65). This was further demonstrated by the ability of MAML1 to promote Gal4 DB-RelA fusion in activating a Gal4-responsive promoter. Several co-activators have been identified for NF-κB signaling, and the most well studied among them is p300/CREB-binding protein (1–4). We found that MAML1 co-activator activities appear to be greater than p300 in the reporter assays (supplemental Fig. 5). One interesting possibility is that p300 recruitment might be required for MAML1-enhanced NF-κB activities because deletion of the p300 binding domain from the MAML1 co-activator results in its inability to activate NF-κB. Also, MAML1(1–302) containing the p300 binding domain can function as a dominant negative to block RelA (p65)-induced NF-κB-dependent transcription, and one potential mechanism could be by competing away p300. It remains to be determined regarding the relationship of the MAML1 co-activator and other co-activators including p300/CREB-binding protein in NF-κB-mediated transcription.

MAML1 belongs to a family of defined transcriptional co-activators for the Notch pathway that enhance Notch signaling through interactions with Notch and CSL. More recently, its co-activator activities for other transcription factors were also revealed, including MEF2C, p53, β-catenin, as well as NF-κB in this study. How the MAML1 exerts such diverse activities is currently unknown. A growing body of evidence supports the idea of complicated cross-talks between Notch and NF-κB pathways (for review see 5, 6). The complex interactions between two pathways could result in either synergistic or antagonistic effects of these two pathway activities (16, 31–33). It was shown that Notch signaling regulates the transcription of the components in the NF-κB pathway, including a member of the NF-κB family of transcription factors, NF-κB2 (p100) (34) and IκBα (35). Conversely, NF-κB affects Notch signaling by regulating the expression of Notch ligand Jagged1 as well as Notch targets, showing a synergistic interaction of two pathways during marginal zone B cell development and T cell receptor activation (16, 33, 36). The mechanistic interactions of these two pathways and the functional outcomes require further elucidation in defined cellular contexts. Because MAML1 has roles in both the Notch and NF-κB pathways, it potentially represents another layer of regulation for cross-talks between these two pathways.

In summary, we identified a novel role for MAML1 in the regulation of the NF-κB pathway, and this regulatory activity is important in cellular survival. Mechanistically, MAML1 promotes NF-κB-dependent transcription via mediating IκBα degradation and co-activating RelA (p65). The exact biological roles for MAML1 regulation of NF-κB in other tissues besides the liver or other processes in vivo are not yet clear, but this study reveals important functional implications for MAML1 in light of the critical role of NF-κB in human malignancies and innate immune response.