The MYND domain-containing protein BRAM1 inhibits lymphotoxin beta receptor-mediated signaling through affecting receptor oligomerization

Cellular Signalling 23 (2011) 80–88

Contents lists available at ScienceDirect

Cellular Signalling j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c e l l s i g

The MYND domain-containing protein BRAM1 inhibits lymphotoxin beta receptor-mediated signaling through affecting receptor oligomerization Hao-Ping Liu a,1, Pei-Jung Chung a,⁎,1, Chih-Lung Liang b, Yu-Sun Chang a,c,⁎ a b c

Molecular Medical Research Center, Chang Gung University, No. 259, Wen-Hwa 1st Road, Kwei-Shan, Taoyuan 333, Taiwan Department of Microbiology and Immunology, Chung-Shan Medical University School of Medicine, No. 110, Sec. 1, Jianguo North Road, Taichung City 40201, Taiwan Graduate Institute of Basic Medical Science, Chang Gung University, No. 259, Wen-Hwa 1st Road, Kwei-Shan, Taoyuan 333, Taiwan

a r t i c l e

i n f o

Article history: Received 31 May 2010 Received in revised form 10 August 2010 Accepted 16 August 2010 Available online 21 August 2010 Keywords: LTβR MYND domain BRAM1 Apoptosis NF-κB JNK

a b s t r a c t MYND (myeloid-Nervy-DEAF-1) domains exist in a large number of proteins that are functionally important in development or associated with cancers. We have previously demonstrated that a MYND domaincontaining protein, the bone morphogenesis protein receptor-associated molecule 1 (BRAM1), is able to interact with Epstein-Barr virus-encoded latent membrane protein 1 (LMP1), which acts as a constitutively activated tumor necrosis factor receptor (TNFR). Herein we further demonstrated that BRAM1 additionally associates with the TNFR-superfamily member, the lymphotoxin beta receptor (LTβR), and hence inhibits LTβR-mediated function. Using the yeast two-hybrid assay, we demonstrated that BRAM1 interacts with LTβR mainly through the self-association domain of LTβR (aa 336–398). The co-immunoprecipitation experiment further revealed that BRAM1 as well as MYND domain-containing proteins, MTG8 and DEAF-1, interacts with LTβR via their MYND domains. The BRAM1-LTβR interaction impedes the self-association of LTβR and the recruitment of TNFR-associated factors 2 and 3 (TRAF2 and TRAF3), leading to abolishment of LTβR-induced NF-κB signaling, JNK activation, and caspase-dependent cell death. In sum, our data demonstrate that the MYND-containing protein BRAM1 abrogates LTβR function through a protein–protein interaction. These findings may provide a direction for the treatment of dysregulation of LTβR-mediated signaling. © 2010 Elsevier Inc. All rights reserved.

1. Introduction The evolutionarily conserved MYND domain is present in the AML1 chromosomal translocation partner ETO/MTG8 [1] or in several developmentally important proteins, such as Nervy [2], DEAF1 [3], and BS69 [4]. The MYND domain of ETO/MTG8 recruits a histone deacetylase complex and mediates gene repression [1,5]. Similarly, the MYND domain of BS69 has been suggested to be involved in gene repression through N-CoR recruitment [6]. BS69 also binds to adenovirus E1A and Epstein-Barr virus nuclear protein 2 (EBNA-2)

Abbreviations: BMP, Bone morphogenesis protein; BRAM1, BMP receptor-associated molecule 1; DEAF-1, Deformed epidermal autoregulatory factor-1; EBV, Epstein-Barr virus; JNK, c-Jun N-terminal kinase; LMP1, Latent membrane protein 1; LTβR, Lymphotoxin beta receptor; MYND domain, myeloid-Nervy-DEAF-1 domain; NF-κB, Nuclear factor kappa B; TGFβ, Tumor growth factor β; TNFR, Tumor necrosis factor receptor; TRAF, TNFR-associated factor. ⁎ Corresponding authors. Chung is to be contacted at Tel.: + 886 3 2118800x3557; fax: + 886 3 2118800x3533. Chang, Molecular Medical Research Center, Chang Gung University, No. 259, Wen-Hwa 1st Road, Kwei-Shan, Taoyuan 333, Taiwan. Tel./fax: + 886 3 2118683. E-mail addresses: [email protected] (H.-P. Liu), [email protected] (P.-J. Chung), [email protected] (C.-L. Liang), [email protected] (Y.-S. Chang). 1 These two authors have made equal contributions. 0898-6568/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.cellsig.2010.08.006

through the MYND domain, thereby inhibiting the trans-activation potential of these oncoviral proteins [7,8]. An alternatively spliced BS69 isoform, termed BRAM1, encompasses the MYND domain of BS69 and resides in cytosol [9]. BRAM1, as well as its orthologs in zebrafish and Caenorhabditis elegans, has been proposed to participate in BMP and TGFβ signaling pathways [9–11]. In addition, BRAM1 is able to interact with Epstein-Barr virus-encoded latent membrane protein 1 (LMP1) [12], which mimics a member of the TNFR superfamily by constitutively activating TNFR-associated factor (TRAF)-mediated signaling [13]. BRAM1 interaction with LMP1 requires the BRAM1 MYND domain, and negatively regulates LMP1induced NF-κB activation [12]. While the cytoplasmic domains of the TNFR-superfamily proteins exhibit little sequence homology, they all bind to a related set of signaling molecules, namely TRAFs. Thus, it is conceivable to speculate that BRAM1 may be capable of interacting with and functionally affecting other members in the TNFR superfamily, such as lymphotoxin beta receptor (LTβR). Notably, LTβR encompasses a PXLXP motif, which is also present in BS69-interacting partners with preference to interact with the BS69 MYND domain [8]. LTβR is prominent on the surface of most of the cell types, including epithelial cells [14–16]. The genetic ablation studies demonstrate that LTβR plays a central role in the development of lymphoid organs [17,18]. LTβR consists of 435 amino acids, with the

H.-P. Liu et al. / Cellular Signalling 23 (2011) 80–88

C-terminal 194 amino acid tail presented toward the intracellular space [19]. As with most TNFR-superfamily members, LTβR signaling is initiated by receptor aggregation resulting from ligand binding, and sequentially activates multiple signaling pathways, including those involving NF-κB activation, JNK signaling, and cell death [20–22]. LTβR mediates most of its functions by interacting with TRAF2, 3, 4 and 5, but not TRAF6 [23,24]. The sub-region of C-terminal LTβR (aa 336– 398) that is essential for the receptor self-association encompasses a minimal TRAF3 binding region (389PEEGDPG395) [19] and mediates cell death [19,22,25]. This sub-region also overlaps with the region responsible for NF-κB activation [19], which is TRAF2 or TRAF5 recruitment essential [23]. Subtle differences exist in the structure–function relationship among LTβR and other TNFR-superfamily members. The cytoplasmic domains of TNFRI and Fas self-associate via their death domains, thereby prompting signaling events in the TNFRI and Fas pathways [26,27]. In contrast, the cytoplasmic domains of LTβR and TNFRII do not contain death domains [27,28]. Instead, LTβR self-associates through a non-death domain, leading to cell apoptosis in the absence of ligand binding [22]. In this study, we aimed to assess whether BRAM1 with the MYND domain binds to LTβR and affects LTβR-mediated function. Using the yeast two-hybrid and co-immunoprecipitation assays, we verified the interaction between LTβR and BRAM1, which primarily depends on the self-association domain of LTβR rather than the PXLXP motif as expected. The functional studies further revealed that BRAM1–LTβR interaction results in inhibition of LTβR-induced signaling and cell death. These findings advance understanding of the structural constraints for the protein–protein interaction involving MYND domain, and may provide a direction for the therapeutic intervention of LTβR signaling dysregulation. 2. Materials and methods 2.1. Plasmids The plasmid pCMV-LTβR encoding the full-length LTβR has been described previously [29]. The Myc-tagged and HA-tagged LTβR constructs were generated by subcloning the LTβR coding region from pCMV-LTβR into the pCMV-Myc and pCMV-HA (Clontech), respectively. For generation of pGFP-LTβR, a cDNA fragment encoding the full-length GFP was amplified by polymerase chain reaction (PCR) using the pEGFPC3 (Clontech) as template, and primers 5′-AACTAGTTAGTTATTAATAGTAATCAAT-3′ (sense) and 5′-AACTAGTAAAAAA TTTCGTTCATTTT-3′ (antisense). PCR was carried out as follows: 35 cycles of 95 °C, 30 s; 55 °C, 1 min; and 72 °C, 1 min. The 1.5-kb GFP PCR product was subcloned into the SpeI site of pCMV-LTβR to produce pGFP-LTβR. The HA-tagged and Flag-tagged BRAM1 constructs were generated by insertion of the cDNA fragments for BRAM1 into the pCMV-HA and pFLAG-CMV-2 (Sigma), respectively. The pFLAG-cMTG8, encoding the C-terminal 173 amino acids of MTG8 (aa 405–577), and the pFLAGcDEAF-1, encoding the C-terminal 162 amino acids of DEAF-1 (aa 404– 565), were also generated using pFLAG-CMV-2. The shRNA for BRAM1 (shBRAM1) was generated as previously described [30]. In brief, a double-strand oligonucleotide encompassing a sequence derived from the open reading frame of human BRAM1 (nucleotides 109–127) was designed in forward and reverse orientation and separated by a 9-base-pair spacer region (ttcaagaga) to allow formation of the hairpin structure in the expressed oligo-RNA: sense strand, 5′-gatccccAAGAAGTTAAGTGCCTCTTttcaagagaAAGAGGCACTTAACTTCTTtttttggaaa; antisense strand, 5′-agcttttccaaaaaAAGAAGTTAAGTGCCTCTTtctcttgaaAAGAGGCACTTAACTTCTTggg. The resulting double-stranded oligonucleotide was constructed into the Bgl II and Hind III sites of the pSUPER vector for expression under the control of the H1 RNA promoter. A scrambled sequence derived from the nucleotides 109–127 was used as a control shRNA (shCtrl).

81

2.2. Cell culture and transfection HeLa cells (ATCC No. CCL2), HEK293 cells (ATCC No. CRL-1573), and 293T cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 2 mM glutamine, 100 units/ml streptomycin, and 100 units/ml penicillin. All transfection experiments were carried out with 70–80% confluent cells. Transfection of HeLa cells was performed by electroporation at 960 μF and 0.23 kV, using an electrophoresis power supply and an electroporation chamber (Bio-Rad). Transient transfection in HEK293 cells and 293T cells was carried out with the calcium phosphate precipitation or the lipofectamine method. 2.3. Yeast two-hybrid assay Yeast transformation and β-galactosidase detection in S. cerevisiae Y190 was performed according to the manufacturer's instructions (PT1113-1; Clontech). To examine the interaction between LTβR and BRAM1, we fused the cDNA fragments encoding the full length or the C-terminal 116 amino acids of BRAM1 (C116) to the activation domain of the Gal4 protein (denoted as Gal4-AD) in the pGAD10 vector (Clontech). cDNA fragments encoding the LTβR cytoplasmic domain were fused to the DNA-binding domain of the Gal4 protein (denoted as Gal4-BD) in pAS2.1 vector (Clontech). A series of cDNA fragments encoding the cytoplasmic domains of LTβR was subcloned into pAS2.1, producing plasmids designated LTβR aa 249–435, aa 249– 255, aa 249–335, aa 336–397 and aa 398–435, respectively. The Ura+ Trp+ His+ Leu+ transformants were further selected and confirmed via 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) assays. 2.4. Preparation of cell lysates and immunoblotting Cells were washed thrice with PBS; cell pellets were resuspended in 5 volumes of lysis buffer (10 mM Tris–HCl, pH 7.1, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1% Triton X-100, 1 μg/ml leupeptin, 1 μg/ml pepstatin, and 1 μg/ml aprotinin) and incubated in ice for 20 min and followed by sonication. Immunoblotting was carried out as previously described [12]. Briefly, cell extracts were resolved by 10% SDS-PAGE and electrophoretically transferred onto PVDF membranes (Millipore). Membranes were probed with the following antibodies, respectively: anti-FLAG antibody (M2, Kodak), anti-Myc polyclonal antibody (sc-789, Santa Cruz Biotechnology), anti-HA monoclonal antibody (sc-40, Santa Cruz Biotechnology), and anti-TRAF3 polyclonal antibody (sc-948, Santa Cruz Biotechnology). Western blot analysis of endogenous I-κBα molecules was carried out by incubation of cell lysates with monoclonal anti-phospho antibody (B-9; Santa Cruz Biotechnology), and polyclonal anti-I-κBα antibody (C-21; Santa Cruz Biotechnology). Polyclonal anti-BRAM1 antibody was generated by immunizing rabbits with a synthetic peptide corresponding to the last 15 amino acids of BRAM1, and purified by using an affinity column. Immunoreactive protein bands were detected using an enhanced chemiluminescence detection kit (Santa Cruz Biotechnology). 2.5. Co-immunoprecipitation assay Immunoprecipitation assays were performed using HEK293 or 293T cells. For the anti-FLAG immunoprecipitation assay, cells were lysed in 200 μl of lysis buffer (25 mM Tris–HCl, pH 7.0, 300 mM NaCl, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 1 μg/ml leupeptin, 1 μg/ml pepstatin, and 1 μg/ml aprotinin). For the anti-HA and the anti-Myc immunoprecipitation assays, cells were lysed in lysis buffer containing 20 mM Tris–HCl, pH 8.0, 137 mM NaCl, 1% NP-40, 1 mM phenylmethylsulfonyl fluoride, 1 μg/ml leupeptin, 1 μg/ml pepstatin, and 1 μg/ml aprotinin. Cleared lysates were incubated at 4 °C for 2 h with anti-FLAG M2 affinity gel (Kodak), anti-HA affinity

82

H.-P. Liu et al. / Cellular Signalling 23 (2011) 80–88

resin (Roche), or anti-Myc antibody (Santa Cruz Biotechnology) immobilized to protein G-Sepharose beads (Amersham Pharmacia Biotech). Samples were then fractionated by SDS-PAGE, and proteins were immunoblotted with specific antibodies as described previously. 2.6. Immunocomplex kinase assays For assaying JNK activity, 293 cells were co-transfected with pCMVLTβR and pFLAG-BRAM1. Twenty-four hours after post-transfection, cells were lysed in solution B (20 mM Tris–HCl at pH 7.0, 0.5 mM dithiothreitol and 1 mM phenylmethylsulfonyl fluoride). For immunoprecipitation of endogenous JNK1, 500 μg of cleared cell extracts was incubated with anti-JNK1 antibody (C17; Santa Cruz Biotechnology) at 4 °C for 2 h and immobilized on protein G-Sepharose beads. The beads were washed thrice with 1 ml of solution B containing 0.5 M NaCl, and once with kinase reaction buffer (20 mM Tris–HCl, pH 7.5, 20 mM MgCl2, 0.5 μM dithiothreitol and 0.2 μM ATP). The activity of JNK1 was measured by kinase assays wherein kinase reaction buffer was mixed with 10 μCi [γ-32P] ATP and 2 μg of GST-c-Jun (aa 1–79) recombinant protein as a substrate. The reaction was stopped by the addition of 4× Laemmeli sample buffer and boiling for 5 min. Samples were separated by 10% SDS-PAGE and phosphorylated GST-c-Jun was identified by autoradiography. 2.7. Assay for NF-κB activity The NF-κB activity was determined by a luciferase (Luc)-reporter assay. HEK293 cells at 70% confluency were transfected with 3 μg of BRAM1 shRNA (shBRAM1), control shRNA, or pSUPER vector. After a 48-hour interval, cells were re-transfected with 5 μg of FLAG-LTβR or control vector, combining 0.5 μg of pNF-κB-Luc reporter, which contains four copies of consensus NF-κB-binding sites upstream of the luciferase cDNA. The mNF-κB-Luc reporter, containing mutated NF-κB-binding sites, was used as a negative control for NF-κBdependent transcription. Cells were harvested at 24 h post-2nd transfection with luciferase lysis buffer (25 mM Tris–phosphate pH 7.8, 2 mM DTT, 2 mM 1,2-diaminocyclohexane-N, N, N′, N′-tetraacetic acid, 10% glycerol, 1% Triton X-100), and the luciferase activity was measured by using a luminometer (LB593, Berhold). 2.8. Cell death assay Cell death induced by over-expression of LTβR was determined by cell morphology and flow cytometry. HeLa cells were co-transfected with pGFP-LTβR and other expression vectors as indicated elsewhere. Forty-eight hours after post-transfection, cells were examined under a microscope and then harvested and stained with Propidium Iodide (Sigma). The percentage of apoptotic cells were determined by measuring the sub-G1 population of GFP-positive cells via flow cytometry (FASCSTar cytofluoremeter, Becton Dickinson). 2.9. Assay for caspase-3 activation The basis of the assay is in the design of the ApoAlert™ pCaspase3Sensor Vector (Clontech) containing the Caspase-3 Sensor protein sequence. This protein sequence codes for a dominant N-terminal nuclear export signal (NES), an optimized caspase-3 cleavage site, enhanced yellow fluorescent protein (EYFP) and a C-terminal nuclear localization signal (NLS). When caspase-3 is activated, it cleaves off the NES sequence, which then allows the NLS to translocate EYFP to the nucleus; otherwise the dominant NES sequence targets the EYFP to the cytosol. HeLa cells grown on coverslips were co-transfected with 0.5 μg of pCaspase3-Sensor, 3 μg of BRAM1 shRNA or pSUPER vector, in combination with 1 μg of pFLAG-BRAM1 or control vector. At 6 h post-transfection, cells were incubated with anti-LTβR mAb 31G4D8 (4 μg/ml) to stimulate LTβR-mediated cell death [22] for

another 48 h. Cells treated with 0.4% DMSO served as a positive control. Afterwards cells were fixed with 3.7% formaldehyde in PBS for 15 min. The photomicrographs were acquired using a LEICA Confocal Spectral Microscope. The percentage of cells with nuclear translocation of EYFP were scored among 50–80 cells expressing Caspase3Sensor in each experiment. 2.10. Statistical analysis All statistical analyses were performed using the SPSS 13.0 software (Chicago, IL, USA). Unless indicated elsewhere, the data were analyzed with the Student's t-test. Differences were considered significant at a level of P 0.05. 3. Results 3.1. LTβR interacts with BRAM1 primarily through the self-association domain of LTβR and the MYND domain of BRAM1 We first performed a yeast two-hybrid assay to examine whether BRAM1 is able to interact with LTβR. As measured by the βgalactosidase activity, a strong interaction was observed between the Gal4-AD-fused BRAM1 and the Gal4-BD-fused LTβR cytoplasmic tail (aa 249–435) as shown in Table 1. This interaction (156 Gal units) was much stronger than the LTβR self-association (22 β-Gal units) as well as the interactions between BRAM1 and each of two LMP1 variants (7 and 6 β-Gal units, respectively), indicating that BRAM1 has a potency to interact with LTβR. We next dissected the structural elements required for LTβR interaction with BRAM1 by using LTβR deletion constructs and MYND domain-containing C-terminal BRAM1 (C116) in yeast two-hybrid assays. The data showed that LTβR mainly relies on its self-association domain (aa 336–397) to interact with BRAM1 (Fig. 1A). Deletion of the PXLXP motif-containing region (aa 249–335) of LTβR showed no hindrance on its interaction with BRAM1, although this region or the most distant C-terminus of LTβR (aa 398–435) alone retained a moderate interaction with BRAM1. The results demonstrated that the self-association domain of LTβR is the major element responsible for interacting with BRAM1. We then performed a co-immunoprecipitation assay to confirm the interaction between LTβR and BRAM1 in HEK293 cells. As shown in Fig. 1B, FLAG-tagged BRAM1 was co-precipitated with HA-tagged LTβR, indicating an interaction between these two proteins. To further investigate the role of MYND domain in this event, we examined whether LTβR interacts with other MYND domain-containing proteins, such as MTG8 and DEAF-1. Similar to BRAM1, the C-terminus of MTG8 or DEAF-1, encompassing a MYND domain, was co-precipitated

Table 1 Protein–protein interactions in the yeast-two-hybrid assays. Transformant Gal4-BD

Gal4-AD

LTβR (aa 249–435) LTβR (aa 249–435) B95.8-LMP1 (aa 187–386) NPC-LMP1 (aa 186–381) pTD1 pAS2.1

LTβR (aa 249–435) BRAM1 BRAM1 BRAM1 pVA3 pGAD10

Arbitrary units of β-gal activity 22 156 7 6 187 b1

The yeast strain Y190 was transformed with the indicated plasmids, and transformants were selected on appropriate selection media. Isolated colonies were grown on SD media containing 25 mM 3-AT and assayed for β-galactosidase activity. Three individual transformants were assayed in each case, and the average values of β-Gal units are shown. The interactions between pTD1–pVA3 and pAS2.1–pGAD10 were used as positive and negative controls, respectively.

H.-P. Liu et al. / Cellular Signalling 23 (2011) 80–88

83

Fig. 2. Prevention of LTβR dimerization by co-expression of BRAM1 as demonstrated by immunoprecipitation assay. 293T cells were co-transfected with FLAG-tagged and Myctagged LTβR expression vectors with or without the different amounts (1 or 3 μg) of HA-tagged BRAM1 expression vector. After 24 h, cell lysates were prepared and immunoprecipitated with anti-FLAG M2 affinity gels. The resulting precipitated complexes were then immunoblotted with anti-Myc antibody and anti-BRAM1 antibody, to detect the LTβR that was dimerized with FLAG-tagged LTβR and the BRAM1 that associated with the FLAG-tagged LTβR, respectively. The equivalent presence of FLAG-tagged LTβR in the immunocomplexes was detected using anti-FLAG antibody.

anti-HA antibodies. As shown in Fig. 2, FLAG-tagged LTβR and Myctagged LTβR were capable of forming complexes, as evidenced by the presence of Myc-tagged LTβR in immunocomplexes precipitated with

Fig. 1. Interaction between LTβR and MYND-containing proteins. (A) Interaction between the cytoplasmic region of LTβR and MYND domain-containing C-terminal BRAM1 (C116) in the yeast two-hybrid system. Constructs for the G4BD-fused cytoplasmic region of LTβR and its various deletion forms were indicated in the schematic. The interaction between every two proteins was assessed by monitoring the growth of transformants after plating on selective medium combined further blue colorselection. The β-galactosidase activity of each transformant was assayed in liquid cultures. Three individual colonies were examined in each case. The fold increase shown on the right was determined by dividing the β-galactosidase activity of transformants co-expressing various LTβR constructs and BRAM1 (C116), by the β-galactosidase activity of transformants expressing the control vectors. (B) Co-immunoprecipitation of MYND-containing proteins with LTβR. HEK293 were co-transfected with 2 μg of plasmids for HA-tagged LTβR and 1 μg of plasmids for FLAG-tagged MYND domaincontaining protein BRAM1, C-terminal MTG8 (cMTG8), or C-terminal DEAF-1 (cDEAF1). FLAG-tagged BAP was used as a non-relevant control. Cell extract was subjected to co-immunoprecipitation using anti-HA affinity resins. The resulting protein complexes were then analyzed by immunoblotting with specific anti-LTβR and anti-FLAG antibodies.

with HA-tagged LTβR (Fig. 1B). In contrast, the MYND domain-less protein, BAP, was not detected in the co-precipitated protein complexes. These data demonstrated that MYND domain enables BRAM1, as well as MTG8 and DEAF-1, to interact with LTβR. 3.2. BRAM1 interferes with LTβR self-association via a protein–protein interaction The activation of LTβR signaling is mediated by self-association or the ligand binding, which induces receptor aggregation and the subsequent activation of multiple signaling pathways. Accordingly, we investigated whether the interaction between LTβR and BRAM1 interferes with LTβR aggregation. In order to differentiate between the two LTβR molecules in a self-associated pair, we co-expressed FLAG-tagged and Myc-tagged LTβR molecules, along with increasing amounts of HA-tagged BRAM1 in 293T cells. Forty-eight hours after post-transfection, the cell extracts were immunoprecipitated with anti-FLAG beads, followed by immunoblotting with anti-Myc and

Fig. 3. Prevention of LTβR-TRAF association by BRAM1. (A) 293T cells were cotransfected with Myc-LTβR (5 μg), FLAG-TRAF2 (0.1 μg), and HA-BRAM1 (1, 2 or 3 μg) expression vectors. The proteins that were associated with Myc-tagged LTβR were immunoprecipitated with anti-Myc antibody. The LTβR-associated TRAF2 was detected by immunoblotting with anti-FLAG antibody. (B) 293T cells were co-transfected with FLAG-LTβR and HA-BRAM1 expression vectors (1 and 3 μg, respectively). Cells extract was harvested at 24 h post-transfection, and then applied to co-immunoprecipitation assay using anti-FLAG affinity resins. The resulting precipitated protein complexes were immunoblotted with anti-TRAF3 antibody to detect the endogenous TRAF3 that was recruited to FLAG-tagged LTβR. The trace of TRAF3 was quantified by using a densitometer. The loss of TRAF3 in the precipitated complexes in the presence of BRAM1 is shown as the percentile of that in the absence of BRAM1 (100%). The percentile mean was calculated from three independent experiments and P value was analyzed using Kruskal–Wallis's test (P = 0.024).

84

H.-P. Liu et al. / Cellular Signalling 23 (2011) 80–88

Fig. 5. The inhibitory effect of BRAM1 on LTβR-induced JNK activation. LTβR-mediated JNK activation was determined by in vitro kinase assay. Endogenous JNK1 were immunoprecipitated from 293T cells that had been co-transfected with LTβR and FLAGBRAM1 expression vectors. The kinase activity of JNK1 was assayed by using GST-c-Jun as a substrate and indicated by autoradiogram of GST-c-Jun. UV irradiation served as a positive control of JNK1 activation.

LTβR-initiated signaling that involves NF-κB or JNK activation. Our observation that over-expression of BRAM1 inhibits LTβR aggregation suggests that the interaction between LTβR and BRAM1 may block the recruitment of TRAFs to the receptor. To verify this possibility, we performed co-immunoprecipitation assays in 293T cells. As shown in Fig. 3A, FLAG-tagged TRAF2 was detected in the immunocomplexes precipitated with Myc-tagged LTβR, indicating recruitment of TRAF2 to the receptor. This recruitment was efficiently inhibited by increasing amounts of BRAM1. Similarly, co-expression of BRAM1 inhibited the recruitment of endogenous TRAF3 to LTβR (Fig. 3B). There was a significant inverse correlation between the levels of BRAM1 and TRAF3 in the precipitated complexes (Spearman's test; r = − 0.96; P = 2.58 × 10− 5). The data collectively demonstrated that BRAM1 competes with TRAF2 and TRAF3 for association with LTβR. Fig. 4. The inhibitory effect of BRAM1 on LTβR-mediated NF-κB activation. (A) Attenuation of LTβR-induced I-κBα phosphorylation by over-expression of BRAM1 (FLAG-BRAM1). Cells were co-transfected with 0, 1, 3, 5, or 10 μg of LTβR expression vectors, together with 0, 1, 2, or 3 μg of FLAG-BRAM1. Twenty-four hours after post-transfection, cell lysates were subjected to immunoblot analysis to determine the endogenous levels of phosphorylation of I-κBα in response to LTβR expression. The top two panels monitored the amount of LTβR and BRAM1. The lower two panels showed the phosphorylated and total forms of I-κBα as examined by immunoblotting with anti-phospho-I-κBα and anti-I-κBα antibodies, respectively. (B) Augmentation of LTβR-induced NF-κB activation by shRNA-mediated knockdown of BRAM1. LTβR-induced NF-κB-activated transcription was assayed in HEK293 cells, which had been transfected with 3 μg of BRAM1 shRNA, control shRNA, or pSUPER vector for 48 h, followed by second transfection of NF-κB-Luc reporter (0.5 μg) together with FLAG-LTβR or control vector (5 μg) for another 24 h. The luciferase (Luc) activity was assayed as detailed in the Materials and methods section. In each case, the basal activity in the absence of LTβR expression was set as 1-fold. The fold of induction of Luc activities by LTβR is shown on the Y axis. The knockdown efficacy of BRAM1 shRNA was assessed in cells co-expressing FLAG-BRAM1 (1 μg) and BRAM1 shRNA (3 μg), followed by immunoblotting with anti-BRAM1 antibody (the inset); α-tubulin was used as an internal control. *, P b 0.01.

anti-FLAG antibody. However, the addition of BRAM1 severely impacted the self-association of LTβR; Myc-tagged LTβR was no longer detectable in the complexes, instead, BRAM1 was coprecipitated with FLAG-tagged LTβR. Therefore, these results suggest that the interaction between BRAM1 and LTβR interferes with the receptor self-association.

3.3. BRAM1 impedes recruitment of TRAF2 and TRAF3 to LTβR Following receptor aggregation, TRAFs are reportedly associated with LTβR via the LTβR self-association domain [15,23,25]. These TRAF proteins, such as TRAF2 and TRAF3, are functionally important in

3.4. BRAM1 inhibits LTβR-mediated NF-κB activation We then sought to investigate the effect of BRAM1 on LTβRinduced signaling pathways. We first examined whether overexpression of BRAM1 may abrogate LTβR-induced phosphorylation of I-κBα, the events occurring upstream of NF-κB activation. As shown in Fig. 4A, ectopically expressed LTβR induced phosphorylation and subsequent degradation of I-κBα, which was significantly inhibited by co-expression of BRAM1 in a dose-dependent manner. To further evaluate the inhibitory effect of BRAM1 on LTβRinduced NF-κB activation, we used a NF-κB-driven Luc-reporter assay to measure the activation of NF-κB-dependent transcription in HEK293 cells; these cells had been transfected with control or BRAM1 shRNA for 48 h, followed by co-transfection of pNF-κB-Luc reporter with pFLAG-LTβR or control vector for another 24 h. The data in Fig. 4B revealed that the activation of NF-κB by LTβR was significantly enhanced in cells expressing BRAM1 shRNA (shBRAM1), compared with those pre-expressing pSUPER vector or control shRNA (shCtrl). The results further evidenced the inhibitory effect of BRAM1 on LTβR-mediated NF-κB activation. 3.5. BRAM1 attenuates LTβR-mediated JNK activation To determine whether BRAM1 also affects LTβR-induced JNK activation, we performed the immunocomplex kinase assay to measure the endogenous JNK1 activity in the cells co-expressing LTβR with BRAM1. As shown in Fig. 5, the phosphorylation of c-Jun (phospho-c-Jun) in the LTβR-expressing cells was 5.3-fold increased compared to the vector-expressing control, indicating the activation

Fig. 6. Attenuation of LTβR-mediated cytotoxicity by BRAM1. (A) Morphology of cells expressing GFP-LTβR or GFP-LTβR together with FLAG-BRAM1. At 24 h and 48 h posttransfection, the morphology of cells from each experiment was observed under the Axioplan 2 microscope (Zeiss, Germany). The images are shown at a 200× magnification. (B) Cell cycle analysis. HeLa cells were co-transfected with plasmids for GFP-LTβR and FLAG-BRAM1. Forty-eight hours after transfection, cells were stained with Propidium Iodide and analyzed by flow cytometry to determine their DNA contents. The presented cytometric profiles display DNA content (X axis) versus cell number (Y axis). The percentage (%) of apoptotic cells, as assessed by measuring the number of GFP-positive cells with sub-G1 DNA contents, are marked for each profile (scale bar). The cells treated with CrmA, a caspase inhibitor preventing cells from apoptosis, and the cells treated with DMSO at a cytotoxic dose (0.4%), were used as a negative and a positive control of cell death, respectively.

H.-P. Liu et al. / Cellular Signalling 23 (2011) 80–88

of JNK1 by LTβR. Co-expression of FLAG-tagged BRAM1 was shown to attenuate the LTβR-mediated JNK activation from 5.3- to 2.7-fold, demonstrating an inhibitory effect of BRAM1 on such an event.

85

Together, our data revealed that BRAM1, via its interaction with LTβR, impedes LTβR self-association and TRAFs recruitment, leading to attenuation of LTβR-mediated NF-κB and JNK signaling pathways.

86

H.-P. Liu et al. / Cellular Signalling 23 (2011) 80–88

Fig. 7. The inhibitory effect of BRAM1 on LTβR-induced caspase-3 activation. (A) Nuclear translocation of EYFP-tagged Caspase3-Sensor. HeLa cells, co-expressing Caspase3-Sensor with BRAM1 shRNA or pSUPER vector, were stimulated with or without anti-LTβR mAb (31G4D8) for 48 h. The knockdown efficacy of BRAM1 shRNA (the inset) was assessed as described in Fig. 4B. The untreated cells and DMSO-treated cells presented in the top panel were used as a negative and a positive control of cell death, respectively. The cells in each experiment were then fixed and applied for image acquisition using a LEICA Confocal Spectral Microscope. Scale bar, 8–20 μm. (B) The percentage of cells exhibiting caspase-3 activation. The number of cells with nuclear translocation of EYFP in each experiment shown in (A) was scored among 50–80 cells expressing Caspase3-Sensor, and the results were represented as the mean ± S.E.M. from three independent experiments. *, P b 0.01; ** and ***, P 0.001.

3.6. BRAM1 inhibits LTβR-induced cell death Over-expression of LTβR, which results in recruitment of TRAFs, is able to induce cell death [31]. Since expression of BRAM1 impedes the association between LTβR and TRAFs (Fig. 3), we next examined the influence of BRAM1 on LTβR-mediated cell death in HeLa cells. The morphology of cells expressing GFPLTβR for 48 h showed a rounded-up cell shape, indicating that these cells undergo cell death (Fig. 6A). In contrast, the cells coexpressing BRAM1 with LTβR exhibited a flattened cell shape, similar to that shown by the untreated cells or the cells treated with a caspase inhibitor, CrmA, which is known to inhibit cell death. Using the flow cytometry-based cell cycle analysis, we measured the population of GFP-positive cells with the sub-G1 DNA contents at 48 h post-transfection. As shown in Fig. 6B, expression of GFP-LTβR caused 34.7% of cells arresting at sub-G1 phase; co-expression of BRAM significantly reduced the percentage of apoptotic cells from 34.7% to 11.2%, confirming that BRAM1 inhibits LTβR-induced cell death.

To better elucidate the inhibitory role of BRAM1 on endogenous LTβR-induced apoptosis, we examined caspase-3 activation in HeLa cells, which had been transfected with pFLAG-BRAM1, BRAM1 shRNA, or in combination, followed by stimulation with or without anti-LTβR mAb, 31G4D8. Apoptotic cells were determined by nuclear translocation of EYFP-tagged Caspase3-Sensor under observation with a LEICA Confocal Microscope. As shown in Fig. 7A, the cells in each experiment showed a cytoplasmic distribution of Caspase3-Sensor in the absence of anti-LTβR mAb stimulation (middle panel). Upon mAb stimulation, the cells expressing control vector (pSUPER) exhibited nuclear translocation of Caspase3-Sensor (bottom panel); however, co-expression with FLAG-tagged BRAM1 (F-BRAM1 + pSUPER) efficiently abrogated this event (Fig. 7B). Moreover, shRNA-mediated knockdown of FLAG-BRAM1 (F-BRAM1 + shBRAM1) enabled the cells to sense mAb-stimulated LTβR signaling again, leading to nuclear translocation of Caspase3-Sensor (Fig. 7A and B). In consistence, knockdown of endogenous BRAM1 (shBRAM1) was shown to amplify the apoptosis induced by anti-LTβR mAb (Fig. 7B). Taken together, these results clearly demonstrate that BRAM1 likely functions as a negative regulator in LTβR-induced apoptosis.

H.-P. Liu et al. / Cellular Signalling 23 (2011) 80–88

4. Discussion The MYND domain is present in a large group of proteins, including ETO/MTG8, Bop, BLU, RP-8, Nervy, and numerous proteins predicted in species from yeast to mammals [3,11,32,33]. The MYND domain consists of a cluster of cysteine and histidine residues, arranged with an invariant spacing to form a potential zinc-binding motif [3]. Mutating the conserved cysteine residues in the DEAF-1 MYND domain does not abolish DNA binding, revealing that the MYND domain should be involved in protein–protein interactions [3]. Indeed, the current evidence suggests that the MYND domain constitutes a protein–protein interaction domain that functions as a co-repressor–recruiting interface [1,3,9,10]. In agreement with this, the MYND domain of ETO/MTG8 interacts directly with the N-CoR and SMRT co-repressors [5,34]. A divergent MYND domain present in BS69 is also shown to interact with N-CoR and mediate transcriptional repression [6]. BS69 has been shown to interact with proteins bearing a PXLXP sequence motif [5]. In addition to BS69, m-Bop also interacts with a PXLXP motif-containing protein, skNAC [35], revealing the interaction tendency between MYND domain and PXLXP motif. Supposedly, this tendency applies to the interaction between LTβR, which contains a PXLXP motif, and BRAM1, which shares the MYND domain with BS69. Unexpectedly, herein we demonstrated that the PXLXP motif of LTβR is not sufficient for its interaction with BRAM1 (Fig. 1A); instead, the interaction mainly relies on the receptor self-association domain. Similar discrepancy is seen in BRAM1 interaction with LMP1, the CTAR2 domain of which crucial for BRAM1 binding actually lacks a PXLXP motif [12]. These findings collectively suggest that BRAM1 has alternative propensities to interact with TNFR receptor proteins. The structural constraints for such interactions may lie in the receptor confirmation instead of the sequence motif. In agreement with this, the C-terminal MTG8 and DEAF-1, with divergent MYND domains as compared to BRAM1, are capable of binding to LTβR as well (Fig. 1B). The role of BRAM1 in LTβR signaling is similar to that of the silencer of death domain (SODD) in TNFRI signaling. Since receptor selfassociation often results in activation of the downstream effectors in the absence of ligand stimulation, over-expression of TNFRI or LTβR causes the receptors to self-aggregate and signal in a ligand-independent fashion. In TNFRI signaling, SODD associates with the death domain of TNFRI to prevent constitutive TNFRI signaling [36]. SODD and TNFRI are pre-associated, and TNF-induced aggregation of TNFRI leads to the disruption of the SODD-TNFRI complex. In contrast, the TNFRI signaling transducers, TRADD and TRAF2, are not constitutively associated with TNFRI, but rather are recruited to TNFRI following its association with TNF [29]. SODD inhibits the interaction between TNFRI and TRADD; analogously, BRAM1 interact with death domain-lacking receptors, such as LMP1 [12] and LTβR (this study). BRAM1 prevents self-aggregation of LTβR (Fig. 2), and impedes the recruitment of TRAF2 and TRAF3 (Fig. 3). Functionally, this interaction can inhibit the signaling and the apoptosis induced b.y over-expression of LTβR or stimulation with anti-LTβR mAb (Figs. 4–7). Therefore, BRAM1 appears to function as a negative regulator in the signal cascade of LTβR, reminiscent of the role of SODD in TNFR signaling. Although BRAM1 transcripts have been detected in various human tissues [9] (and data not shown), the human BRAM1 protein has not yet been successfully detected. This obstacle limits the in-depth understanding of functional aspects of BRAM1. In this study, we seek to make more sense of BRAM1 biological function by using a shRNAbased knockdown approach. Indeed, we found that the LTβRmediated NF-κB activation and cell death are augmented in BRAM1 shRNA-expressing cells (Figs. 4B and 7). These findings reinforce the inhibitory role of BRAM1 in LTβR signaling. Accordingly, the availability of BRAM1 in cells is one of the factors determining the LTβR-mediated function. Further investigations are needed for elucidating whether the expression of BRAM1 is sensitive to LTβR

87

signal, and how this event occurs under certain physiological situations. In sum, our study demonstrated that BRAM1 acts as a negative signal regulator located at the very proximal end of LTβR complex assembly. Our findings may provide a direction for abrogating the dysregulation of LTβR signaling as a therapeutic intervention of diseases, by usage of the high-affinity binding between BRAM1 and LTβR (Table 1). For instance, the self-association domain of LTβR is also shown to associate with the Hepatitis C virus (HCV) core protein [37,38]. The direct association between LTβR and the HCV core protein modulates the NF-κB and cytolytic pathways of LTβR/LT-α1β2 [37,39]. Thus, it will be interesting to test whether BRAM1 interferes with the association between LTβR and the HCV core protein, thereby blocking the downstream signaling and protecting HCV-infected cells from cell lysis. 5. Conclusion In this study, we demonstrated that the MYND domain-containing protein, BRAM1, is able to interact with a TNFR-superfamily member, LTβR. The interaction between LTβR and BRAM1 primarily depends on the self-association domain of LTβR rather than a PXLXP motif. This interaction interferes with LTβR self-association and TRAF recruitment, leading to inhibition of NF-κB activation, JNK signaling and caspase-dependent cell death induced by LTβR. These findings advance understanding of the structural propensity for the protein– protein interaction involving MYND domains, and may provide a direction for the treatment of dysregulation of LTβR-mediated signaling. Role of the funding source This work was supported by National Health Research Institute (Taiwan) Grant NHRI-EX92-9214BI; Chang-Gung University and Chang-Gung Memorial Hospital Grant CMRPD32006; Chang Gung Molecular Medicine Research Center grant CMRPD140041; and MOE Program for Promoting Academic Excellency in Universities (Grant number 89-B-FA04-1-4). The funding sources had not involved in study design; in the collection, analysis, and interpretation of data; in the writing of the report; in the decision to submit the paper for publication. Conflict of interest The authors declare that they have no competing interests. Authors' contributions HPL and PJC carried out most of the experiments and wrote the manuscript. CLL participated in the immunoprecipitation experiments. YSC coordinated the study and finalized the manuscript. All authors read and approved the final manuscript. Acknowledgements We thank Dr. Shie-Liang Hsieh (Institute of Microbiology and Immunology, National Yang-Ming University, Taiwan) for providing the LTβR expression vector and the LTβR-stimulating mAb, 31G4D8. We also thank Ms. Hsiao-Chien Chu (Institute of Microbiology and Immunology, National Yang-Ming University, Taiwan) for construction of the pGFP-LTβR expression vector. References [1] J. Wang, T. Hoshino, R.L. Redner, S. Kajigaya, J.M. Liu, Proc. Nat. Acad. Sci. U.S.A. 95 (1998) 10860.

88

H.-P. Liu et al. / Cellular Signalling 23 (2011) 80–88

[2] J. Wildonger, R.S. Mann, Dev. Biol. 286 (2005) 507. [3] C.T. Gross, W. McGinnis, EMBO J. 15 (1996) 1961. [4] H. Ogata-Kawata, K. Yamada, M. Uesaka-Yoshino, N. Kagawa, K. Miyamoto, Front. Biosci. 12 (2007) 1911. [5] B. Lutterbach, D. Sun, J. Schuetz, S.W. Hiebert, Mol. Cell. Biol. 18 (1998) 3604. [6] H. Masselink, R. Bernards, Oncogene 19 (2000) 1538. [7] G. Hateboer, A. Gennissen, Y.F. Ramos, R.M. Kerkhoven, V. Sonntag-Buck, H.G. Stunnenberg, R. Bernards, EMBO J. 14 (1995) 3159. [8] S. Ansieau, A. Leutz, J. Biol. Chem. 277 (2002) 4906. [9] K. Kurozumi, M. Nishita, K. Yamaguchi, T. Fujita, N. Ueno, H. Shibuya, Genes Cells 3 (1998) 257. [10] K. Morita, M. Shimizu, H. Shibuya, N. Ueno, Proc. Nat. Acad. Sci. U.S.A. 98 (2001) 6284. [11] K.M. Wu, C.J. Huang, S.P. Hwang, Y.S. Chang, J. Biomed. Sci. 13 (2006) 345. [12] P.J. Chung, Y.S. Chang, C.L. Liang, C.L. Meng, J. Biol. Chem. 277 (2002) 39850. [13] H.P. Liu, C.C. Wu, Y.S. Chang, EMBO J. 25 (2006) 4120. [14] P.D. Crowe, T.L. VanArsdale, B.N. Walter, C.F. Ware, C. Hession, B. Ehrenfels, J.L. Browning, W.S. Din, R.G. Goodwin, C.A. Smith, Science 264 (1994) 707. [15] W.R. Force, B.N. Walter, C. Hession, R. Tizard, C.A. Kozak, J.L. Browning, C.F. Ware, J. Immunol. 155 (1995) 5280. [16] C.F. Ware, T.L. VanArsdale, P.D. Crowe, J.L. Browning, Curr. Top. Microbiol. Immunol. 198 (1995) 175. [17] D.L. Drayton, S. Liao, R.H. Mounzer, N.H. Ruddle, Nat. Immunol. 7 (2006) 344. [18] A. Futterer, K. Mink, A. Luz, M.H. Kosco-Vilbois, K. Pfeffer, Immunity 9 (1998) 59. [19] W.R. Force, A.A. Glass, C.A. Benedict, T.C. Cheung, J. Lama, C.F. Ware, J. Biol. Chem. 275 (2000) 11121. [20] J.L. Browning, K. Miatkowski, I. Sizing, D. Griffiths, M. Zafari, C.D. Benjamin, W. Meier, F. Mackay, J. Exp. Med. 183 (1996) 867. [21] F. Mackay, G.R. Majeau, P.S. Hochman, J.L. Browning, J. Biol. Chem. 271 (1996) 24934.

[22] M.Y. Wu, P.Y. Wang, S.H. Han, S.L. Hsieh, J. Biol. Chem. 274 (1999) 11868. [23] H. Nakano, H. Oshima, W. Chung, L. Williams-Abbott, C.F. Ware, H. Yagita, K. Okumura, J. Biol. Chem. 271 (1996) 14661. [24] T.L. VanArsdale, S.L. VanArsdale, W.R. Force, B.N. Walter, G. Mosialos, E. Kieff, J.C. Reed, C.F. Ware, Proc. Nat. Acad. Sci. U.S.A. 94 (1997) 2460. [25] W.R. Force, T.C. Cheung, C.F. Ware, J. Biol. Chem. 272 (1997) 30835. [26] H.Y. Song, J.D. Dunbar, D.B. Donner, J. Biol. Chem. 269 (1994) 22492. [27] M.P. Boldin, I.L. Mett, E.E. Varfolomeev, I. Chumakov, Y. Shemer-Avni, J.H. Camonis, D. Wallach, J. Biol. Chem. 270 (1995) 387. [28] V. Haridas, B.G. Darnay, K. Natarajan, R. Heller, B.B. Aggarwal, J. Immunol. 160 (1998) 3152. [29] C.M. Chen, L.R. You, L.H. Hwang, Y.H. Lee, J. Virol. 71 (1997) 9417. [30] T.R. Brummelkamp, R. Bernards, R. Agami, Science 296 (2002) 550. [31] I.A. Rooney, K.D. Butrovich, A.A. Glass, S. Borboroglu, C.A. Benedict, J.C. Whitbeck, G.H. Cohen, R.J. Eisenberg, C.F. Ware, J. Biol. Chem. 275 (2000) 14307. [32] P.G. Feinstein, K. Kornfeld, D.S. Hogness, R.S. Mann, Genetics 140 (1995) 573. [33] G.P. Owens, W.E. Hahn, J.J. Cohen, Mol. Cell. Biol. 11 (1991) 4177. [34] B. Lutterbach, J.J. Westendorf, B. Linggi, A. Patten, M. Moniwa, J.R. Davie, K.D. Huynh, V.J. Bardwell, R.M. Lavinsky, M.G. Rosenfeld, C. Glass, E. Seto, S.W. Hiebert, Mol. Cell. Biol. 18 (1998) 7176. [35] R.J. Sims 3rd, E.K. Weihe, L. Zhu, S. O'Malley, J.V. Harriss, P.D. Gottlieb, J. Biol. Chem. 277 (2002) 26524. [36] Y. Jiang, J.D. Woronicz, W. Liu, D.V. Goeddel, Science 283 (1999) 543. [37] H. Hsu, H.B. Shu, M.G. Pan, D.V. Goeddel, Cell 84 (1996) 299. [38] M. Matsumoto, T.Y. Hsieh, N. Zhu, T. VanArsdale, S.B. Hwang, K.S. Jeng, A.E. Gorbalenya, S.Y. Lo, J.H. Ou, C.F. Ware, M.M. Lai, J. Virol. 71 (1997) 1301. [39] L.R. You, C.M. Chen, Y.H. Lee, J. Virol. 73 (1999) 1672.