Bimp1, a MAGUK family member linking protein kinase C activation to Bcl10-mediated NF-kappaB induction.

Bcl10 and MALT1, products of distinct chromosomal translocations in mucosa-associated lymphoid tissue lymphoma, cooperate in activating NF-kappaB. Mice lacking Bcl10 demonstrate severe immunodeficiency associated with failure of lymphocytes to activate nuclear factor kappaB (NF-kappaB) in response to antigen receptor stimulation and protein kinase C activation. We characterize Bimp1, a new signaling protein that binds Bcl10 and activates NF-kappaB. Bimp1-mediated NF-kappaB activation requires Bcl10 and IkappaB kinases, indicating that Bimp1 acts upstream of these mediators. Bimp1, Bcl10, and MALT1 form a ternary complex, with Bcl10 bridging the Bimp1/MALT1 interaction. A dominant negative Bimp1 mutant inhibits NF-kappaB activation by anti-CD3 ligation, phorbol ester, and protein kinase C expression. These results suggest that Bimp1 links surface receptor stimulation and protein kinase C activation to Bcl10/MALT1, thus leading to NF-kappaB induction.

Bcl10 and MALT1, products of distinct chromosomal translocations in mucosa-associated lymphoid tissue lymphoma, cooperate in activating NF-B. Mice lacking Bcl10 demonstrate severe immunodeficiency associated with failure of lymphocytes to activate nuclear factor B (NF-B) in response to antigen receptor stimulation and protein kinase C activation. We characterize Bimp1, a new signaling protein that binds Bcl10 and activates NF-B. Bimp1-mediated NF-B activation requires Bcl10 and IB kinases, indicating that Bimp1 acts upstream of these mediators. Bimp1, Bcl10, and MALT1 form a ternary complex, with Bcl10 bridging the Bimp1/ MALT1 interaction. A dominant negative Bimp1 mutant inhibits NF-B activation by anti-CD3 ligation, phorbol ester, and protein kinase C expression. These results suggest that Bimp1 links surface receptor stimulation and protein kinase C activation to Bcl10/MALT1, thus leading to NF-B induction.
B-cell lymphomas of mucosa-associated lymphoid tissue (MALT) 1 are the most common form of lymphoma arising in extranodal sites. MALT lymphoma most commonly occurs within the stomach where it is strongly associated with chronic Helicobacter pylori infection (1). Until recently, little has been known about the molecular events that lead to the development of MALT lymphoma. However, studies have now identified several genes that are involved in recurrent chromosomal translocations in MALT lymphoma. The t(1; 14)(p22:q32) results in the placement of the entire coding region of the Bcl10 gene adjacent to the strong immunoglobulin heavy chain locus transcriptional enhancer, leading to deregulated overexpression of Bcl10 (2,3). A second recurrent chromosomal translocation, t(11;18)(q21;q21), is seen in up to 50% of MALT lymphomas. This translocation results in the creation of a fusion gene that encodes a chimeric protein consisting of the N-terminal portion of the inhibitor of apoptosis, c-IAP-2 (also known as AP12) linked to the C terminus of a novel protein, MALT1 (4 -6).
Bcl10 (also known as CIPER, mE10, c-CARMEN, CLAP, and c-E10) is a caspase-recruitment domain (CARD)-containing protein that activates the transcription factor NF-B (2,(7)(8)(9)(10)(11). MALT1, the other target of chromosomal translocation in MALT lymphoma, is a novel protein composed of a death domain, two immunoglobulin-like domains, and a C-terminal caspase-like domain (5,6). Recent studies have shown that MALT1 binds specifically to Bcl10 and enhances the ability of Bcl10 to activate NF-B (12,13). Moreover, Bcl10 mediates the oligomerization of MALT1, an event that is sufficient for NF-B activation (13). In addition, the c-IAP-2-MALT1 fusion protein resulting from the t(11;18)(q21;q21) chromosomal translocation strongly activates NF-B (12,13). These observations suggest that MALT1 functions as a downstream effector in a Bcl10-MALT1 complex, and that Bcl10 and MALT1, although independent targets of chromosomal translocation, act in a common NF-B signaling pathway to promote MALT lymphoma.
Analysis of mutant mice lacking Bcl10 have shown that this protein plays a critical role in lymphoid activation and proliferation (14). In addition, Bcl10 is important for neural development, because a significant percentage of mice deficient in Bcl10 exhibit a defect in neural tube closure (14). Bcl10 Ϫ/Ϫ lymphocytes fail to activate NF-B in response to CD3, CD3/ CD28, and IgM ligation, indicating that Bcl10 is a signal transducer between lymphoid antigen receptors and NF-B (14). One class of signaling molecules that has been implicated in mediating both T-cell receptor and B-cell receptor signaling is the family of protein kinase C (PKC) enzymes. Mounting evidence suggests that the PMA-activated isoforms such as PKC and PKC␤ are important in CD3/CD28 and IgM receptor-mediated NF-B induction, respectively (15)(16)(17)(18). Bcl10 Ϫ/Ϫ lymphoid cells retain normal Ca 2ϩ mobilization but are defective in NF-B stimulation induced by PMA/Ca 2ϩ ionophore, suggesting that Bcl10 acts downstream of PKC in the antigen receptor-mediated signaling pathway (14).
Until now, proteins operating upstream of the Bcl10-MALT1 signaling complex have not been definitively identified. MALT1, a proposed downstream effector of Bcl10, does not contain a CARD, and therefore cannot participate in a CARD/ CARD interaction with Bcl10. Recently, the protein CARD9 was identified as a CARD-containing protein that interacts selectively with the CARD of Bcl10, but the physiological significance of this binding has not yet been explored (19). Furthermore, the mechanisms by which surface receptors and PKC activation are linked to Bcl10 are unknown. Here we characterize a new Bcl10-binding molecule, Bimp1 (Bcl10-interacting MAGUK protein), a CARD-containing protein with homology to CARD9. Unlike CARD9, Bimp1 contains Src homology (SH3), Z0 -1 homologous (PDZ), and guanylate kinase (GUK)-like domains. Bimp1 is a member of the MAGUK (membrane-associated guanylate kinase) family, a class of proteins that function as molecular scaffolds in assembling multiprotein complexes at the plasma membrane (20,21). We show that Bimp1 is a component of a novel NF-B signaling pathway that links surface receptor signaling and PKC activation to Bcl10, MALT1, and the IB kinases (IKKs).

EXPERIMENTAL PROCEDURES
Identification of Bimp1 and Bimp2 cDNAs-The nucleotide sequence of cDNAs encoding peptides with homology to the CARD of CARD9 (GenBank TM accession number AF311287) were found in the EST data bases of GenBank TM using the TBLASTN program. cDNAs encoding a full-length coding region corresponding to Bimp1 and Bimp2 were obtained from Incyte Genomics Inc. (St. Louis, MO). The authencity of the cDNAs was verified by DNA sequencing.
Northern Blot Analysis-A 430-base pair fragment of Bimp1 cDNA was obtained by digestion with BlnI and PmlI. The fragment was radiolabeled by random priming using a commercial kit (Roche Molecular Biochemicals) and then used to probe a mouse poly(A) ϩ multitissue RNA blot (CLONTECH Laboratories).
Transfections and NF-B Activation Assays-2 ϫ 10 5 HEK293T cells were plated in standard 12-well culture dishes and transfected with the reporter constructs pEF1-BOS-␤-gal (30 ng) and pBVIx-Luc (3.0 ng) plus the indicated expression plasmids using the calcium phosphate method. The total amount of transfected plasmid DNA was adjusted with pcDNA3 vector such that it was constant within each individual experiment. NF-B activation was assessed by measuring luciferase activity (normalized for ␤-galactosidase expression) in cell extracts 24 h after transfection (8). 33 ng of pcDNA3-p35 was also transfected in all cases to prevent cell death. For transfections of Rat-1 and 5R fibroblast cell lines, 0.5 ϫ 10 5 cells were plated in 12-well plates and transfected with indicated expression plasmids as described (13). The cells were harvested 24 h after transfection, and the lysates were assayed for NF-B activation as described above. In experiments using the T-cell hybridoma cell line 68-41, 1 ϫ 10 6 cells were transfected with 20 ng of pBVIx-Luc and 20 ng of pRL-TK (Promega), 10 ng of pcDNA3-p35 plus 50 ng of indicated expression plasmids using the DEAE-dextran method, and NF-B induction was assessed using the dual luciferase kit according to the manufacturer's instructions (Promega). 10 ng/ml of TNF␣ was added to the medium 4 h prior to harvest when indicated. For cells treated with PMA/Ca 2ϩ ionophore, 50 ng/ml of PMA and 0.7 g/ml of Ca 2ϩ ionophore A23187 were added to the medium 4 h prior to harvest. 68-41 cells were stimulated with 5 g/ml anti-CD3 2C11 coated onto 12-well plates 4 h prior to harvest when indicated.
Immunoprecipitations and Western Blotting-HEK293T cells were harvested 24 h following transfection and lysed in 0.2% Nonidet P-40 lysis buffer (29). FLAG-tagged Bimp1 and its binding partners were immunoprecipitated by combining 25 l of a 50% suspension of anti-FLAG M2 affinity gel (Sigma) with 1 ml of soluble cell lysate and incubated for 3 h at 4°C. Immunoblotting was performed using mouse monoclonal anti-FLAG (Sigma), rabbit polyclonal anti-Myc (Santa Cruz), mouse monoclonal anti-HA 12CA5 (Roche Molecular Biochemicals), or rabbit polyclonal anti-HA (Santa Cruz) antibodies.

Identification of Bimp1 cDNA and Analysis of Amino Acid
Sequence-In an effort to identify potential binding partners that could participate in a CARD/CARD interaction with Bcl10, we searched public data bases of ESTs for clones with homology to the CARD of CARD9 (19). Sequence analysis of one of the ESTs obtained from our search revealed a mouse cDNA that encodes a 1021-amino acid protein with a predicted molecular mass of 114 kDa. This protein was designated Bimp1 (Fig. 1A). A search for a human orthologue revealed several overlapping partial cDNAs that could be constructed to form a single open reading frame encoding a 1032-amino acid protein with 90% identity to the mouse sequence. The corresponding human gene mapped to chromosome 22q13 (GenBank TM accession numbers AL022315 and AL049851). In our search, we also identified a second mouse cDNA encoding a protein with significant homology to Bimp1 that we have designated Bimp2 (Fig. 1A). Bimp2 is 1000 amino acids in length with a predicted molecular mass of 113 kDa.
Analysis of the amino acid sequence of Bimp1 revealed that its structure is comprised of five putative functional domains: an N-terminal CARD, a central coiled-coil domain, and a Cterminal structure comprised of SH3, PDZ, and GUK-like domains (Fig. 1B). The presence of the PDZ, SH3, and GUK domains identifies Bimp1 and Bimp2 as members of the growing MAGUK family of proteins (20,21). These proteins serve as molecular scaffolds for signaling pathway components by utilizing multiple protein/protein interaction domains to cluster receptors, adaptor proteins, and cytosolic signaling proteins at the cell membrane (20,21). The central coiled-coil structure in Bimp1 is predicted to participate in protein oligomerization (30). The CARD of Bimp1 does share some degree of sequence homology with the CARD of a variety of cell death pathway components (Fig. 1C), but it is most strikingly similar to the CARD of CARD9 (47% identity) and Bimp2 (47% identity) (19) (Fig. 1C).
Bimp1 Is Expressed in Multiple Tissues-We performed Northern blot analysis to determine the distribution of Bimp1 transcripts in various mouse tissues. A ϳ6.8-kilobase Bimp1 transcript was detected in all tissues examined including heart, brain, spleen, lung, liver, skeletal muscle, kidney, and testes (Fig. 1D). The transcript was most abundant in heart and kidney. A smaller ϳ3.8-kilobase Bimp1 transcript was also detected in brain but was not detected in any of the other tissues sampled (Fig. 1D) The significance of this alternative form of Bimp1 mRNA is not currently known.
Bimp1 Activates NF-B-Several CARD-containing proteins, including Bcl10, have been shown to induce NF-B activation (8,22). To test whether Bimp1 activates NF-B, a Bimp1 expression plasmid was co-transfected with an NF-B luciferase reporter plasmid. HA-tagged Bimp1 induced NF-B activity in a dose-dependent manner to a maximum of ϳ250-fold induction as compared with empty vector ( Fig. 2A). Transfection of untagged Bimp-1 expression plasmid also resulted in NF-B activation, showing that this activity cannot be attributed to the epitope tag (data not shown).
The Bimp1 CARD Is Essential but Not Sufficient for NF-B Activation-To test the importance of the CARD of Bimp1 in NF-B activation, we utilized two mutants of Bimp1, one that lacked the entire CARD (Bimp1⌬CARD) and one that contained only the CARD (Bimp1CARD) (Fig. 2B). Deletion of the CARD completely abolished NF-B activation, indicating that the CARD is essential for Bimp1-mediated NF-B signaling (Fig. 2B). However, the Bimp1CARD mutant was also inactive when expressed by itself, suggesting that Bimp1 domains other than the CARD are also required for NF-B-inducing activity. Immunoblotting confirmed that the mutant forms of Bimp1 were expressed as effectively as wild-type Bimp1 (Fig. 2B, inset). Bimp1 and Bimp2 Act through the IKK Complex to Activate NF-B-In response to certain stiumuli, IB is phosphorylated by the IKK complex, which includes two IB kinases, IKK␣ and ␤, and one regulatory subunit, IKK␥. The phosphorylation of IB leads to its ubiquitination and degradation, thus allowing NF-B to translocate to the nucleus where it activates transcription of target genes (31). To investigate the role of the catalytic subunits of the IKK complex in Bimp1 signaling, we tested the ability of Bimp1 to induce NF-B activation in wildtype MEFs and in MEFs deficient in IKK␣ or IKK␤ subunit or both (28). Bimp1 induced comparable levels of NF-B activation in wild-type MEFs and MEFs deficient in IKK␤ but reduced levels in cells deficient in IKK␣ (Fig. 3A). No detectable Bimp1-mediated NF-B induction was observed in MEFs deficient in both IKK␣ and IKK␤ subunits (Fig. 3A). Transfection of Nod1, a molecule known to activate NF-B through the IKK complex yielded similar results (Fig. 3A) (22). In control experiments, expression of IKK␤ effectively induced NF-B activation in cells lacking IKK␣/IKK␤, showing that NF-B signaling machinery downstream of the IKK complex is intact in these MEFs (data not shown). To determine whether the IKK␥ regulatory subunit was important for Bimp1-mediated NF-B activation, we co-expressed Bimp1 with a dominant negative form of IKK␥ that lacks amino acids 1-134 and is therefore unable to bind IKK␣ and ␤ (23). Co-expression of the dominant negative IKK␥ resulted in effective inhibition of Bimp1-medi- ated NF-B activation (Fig. 3B). The IKK␥ dominant negative mutant did not block NF-B activation mediated by IKKi, a kinase that is known to activate NF-B independently of IKK␥ (27) (Fig. 3B). As a positive control, a dominant negative form of IB␣ (IB␣-S32A/S36A), which cannot be phosphorylated and therefore cannot be degraded (26), blocked both Bimp1 and IKKi-mediated NF-B activation (Fig. 3B). A dominant negative form of MyD88, a key factor in NF-B signaling from the interleukin-1/Toll receptors (25), did not block either Bimp1 or IKKi-mediated NF-B activation (Fig. 3B).
Using a different approach, we confirmed the requirement for IKK␥ in Bimp1 signaling by testing NF-B induction in the IKK␥-defective 5R cell line, a derivative of the Rat-1 fibroblast line (32). While in the parental Rat-1 cells Bimp1 effectively induced NF-B, we observed no Bimp1-mediated NF-B activation in the IKK␥-deficient 5R cells (Fig. 3C). As a control, treatment of cells with TNF␣ was also unable to activate NF-B in IKK␥-deficient cells. In contrast, transfection with IKK␤, which is downstream of IKK␥ in this signaling pathway, activated NF-B in both the parental Rat-1 cells as well as the IKK␥-deficient 5R cells, showing that the NF-B transcriptional machinery remained intact in the 5R cells (Fig. 3C). Like Bimp1, Bimp2 was unable to activate NF-B in IKK␥-deficient cells (Fig. 3C). Taken together, these results indicate that Bimp1 and Bimp2, like Bcl10, participate in a pathway of NF-B activation that depends on the IKK complex and suggest that IKK␣ and IKK␤ are functionally redundant in the Bimp1-mediated NF-B activation pathway.
Bimp1 and Bimp2 Act Upstream of Bcl10 to Activate NF-B-To test whether Bimp1 functions upstream of Bcl10, we assessed the ability of Bimp1 to activate NF-B in Bcl10deficient cells (14). In Bcl10 ϩ/Ϫ MEFs, expression of Bimp1 resulted in 10 -15-fold induction of NF-B compared with empty vector, whereas in Bcl10 Ϫ/Ϫ MEFs, expression of Bimp1 did not result in appreciable NF-B activation (Fig. 4A). The same results were obtained using two independent Bcl10 Ϫ/Ϫ cell lines. Expression of the Bimp1 homologue, Bimp2, yielded similar results, with 10-fold induction of NF-B activity in Bcl10 Ϯ MEFs, but no induction in the Bcl10 Ϫ/Ϫ cell lines (Fig. 4A). As a positive control, expression of Nod1, a known activator of NF-B (22), resulted in 15-25-fold induction of NF-B activity in both Bcl10 ϩ/Ϫ and Bcl10 Ϫ/Ϫ MEFs, illustrating that the NF-B signaling machinery remained intact in the Bcl10 Ϫ/Ϫ cells (Fig. 4A). Taken together, these results provide strong evidence that Bimp1 and Bimp2 operate upstream of Bcl10 in a pathway leading to NF-B activation.
Bimp1 Binds to Bcl10 through a CARD/CARD Interaction-The CARD is an ␣-helical domain that links signaling components through homophylic CARD/CARD interactions (33). To determine whether Bimp1 binds to the CARD of Bcl10, Bimp1 was co-expressed with Bcl10, cell extracts were prepared, and Bimp1 was immunoprecipitated with anti-FLAG antibody. Immunoblotting revealed that Bcl10 co-immunoprecipitated with Bimp1 (Fig. 4B). The binding of Bimp1 to Bcl10 was dependent on the N-terminal CARD of Bcl10, because Bimp1 failed to co-precipitate a variant of Bcl10 containing a point mutation (L41Q) that disrupts CARD/CARD interactions (8) (Fig. 4B). The association between Bimp1 and Bcl10 was specific in that RICK, another CARD-containing protein (23), did not co-precipitate with Bimp1 (Fig. 4B). In reciprocal experiments, Bcl10 co-immunoprecipitated with wild-type and mutant Bimp1 containing the CARD alone but not with a mutant of Bimp1 lacking the CARD (Fig. 4C). These results indicate that Bimp1 and Bcl10 associate through a CARD/CARD homotypic interaction.
Bimp1, Bcl10, and MALT1 Can Form a Ternary Protein Complex-Bcl10 has been shown to bind to MALT1 and to mediate MALT1 oligomerization, an event that is sufficient for the activation of NF-B (12,13). Because Bimp1 binds Bcl10 and operates upstream of Bcl10 in an NF-B signaling pathway, we tested whether Bimp1 can associate with MALT1 through Bcl10. Because wild-type MALT1 is expressed at very low levels in our transfection system, we use a truncated mutant, MALT1-(1-330), which retains full binding to Bcl10 and can be expressed at higher levels than the wild-type MALT1 (13). When MALT1 was co-expressed with Bimp1, in the absence of Bcl10, the MALT1 protein did not co-precipitate with Bimp1 (Fig. 5A). However, MALT-1 did co-precipitate with Bimp1 when Bcl10 was also included in the transfection (Fig.  5A). These results support the notion that Bimp1, Bcl10, and MALT1 can form a multimeric protein complex, with Bcl10 forming a bridge between Bimp1 and MALT1. In an effort to FIG. 2. Bimp1 activates NF-B in a CARD-dependent manner. A, concentration-dependent activation of NF-B by Bimp1. 293T cells were transfected with the indicated amounts of pcDNA3-Bimp1-HA expression plasmid in the presence of pBVIx-Luc (an NF-B-responsive luciferase reporter) and pEF1-BOS-␤-gal (used for normalization of transfection efficiency) as described under "Experimental Procedures." 24 h after transfection, cell extracts were prepared, and NF-B activation was assessed by measuring the luciferase:␤-galactosidase ratio as described under "Experimental Procedures." B, the Bimp1 CARD is necessary but not sufficient for NF-B activation. 293T cells were transfected with 50 ng of either pcDNA3-Bimp1-FLAG, pcDNA3-Bimp1-CARD-FLAG, or pcDNA3-Bimp1⌬CARD-FLAG, along with pEF1-BOS-␤-gal and pBVIx-Luc. The cells were harvested 24 h after transfection, and NF-B activity was determined as above. The lysates were also subjected to 8 -15% SDS-PAGE and immunoblotting with monoclonal antibody to FLAG. Wt or WT, wild type. further confirm that Bcl10 can mediate the interaction between Bimp1 and MALT1, we utilized mutant versions of Bcl10, which disrupt either Bimp1/Bcl10 or Bcl10-MALT1 binding. Bcl10 (L41Q) contains a point mutation in the CARD, which prevents binding to Bimp1 (Fig. 4B). When Bcl10(L41Q) replaced wild-type Bcl10 in our co-immunoprecipitation experiments, MALT1 did not co-precipitate with Bimp1, indicating that disruption of the Bimp1/Bcl10 interaction destroyed the ability of Bimp1 to recruit MALT1 to a binding complex (Fig.  5A). The Bcl10(⌬107-119) mutant is deficient in MALT1 binding and in NF-B activation (13). This Bcl10 mutant bound to Bimp1, suggesting that its overall structural integrity is maintained and that the CARD is intact (Fig. 5A). However, MALT1 did not co-precipitate with Bimp1 when Bcl10(⌬107-119) was included in the transfection in place of wild-type Bcl10 (Fig.  5A). These results suggest that disruption of Bcl10-MALT1 binding abolishes the ability of Bimp1 to recruit MALT1 to a binding complex. Taken together, our results indicate that Bimp1, Bcl10, and MALT1 can form a multimeric signaling complex in which Bcl10 forms a molecular bridge between Bimp1 and MALT1.
Bimp1 Cooperates with Bcl10 and MALT1 in NF-B Induction-MALT1 specifically enhances Bcl10-mediated NF-B activation, suggesting that Bcl10 and MALT1 cooperate in a common signaling pathway leading to NF-B activation (13).
To determine whether Bimp1 cooperates with Bcl10 and MALT1 in this same signaling pathway, we tested whether Bcl10 and MALT1 could enhance Bimp1-mediated NF-B activation. Because overexpression of Bcl10 induces potent NF-B activation, we transfected cells with low amounts of Bcl10 expression plasmid and assessed the effect of Bcl10, in the absence or the presence of MALT1, on Bimp-1-mediated NF-B induction. Transfection of 2.5 ng of Bcl10, with or without MALT1, resulted in no appreciable activation of NF-B (Fig. 5B). However, co-transfection of 2.5 ng of Bcl10 with Bimp1 resulted in an approximately 2-fold enhancement of NF-B activation compared with Bimp1 alone (Fig. 5B). Furthermore, co-transfection of Bimp1 with the combination of Bcl10 plus MALT1 resulted in an even greater enhancement (3.5-fold) of NF-B activation compared with Bimp1 alone (Fig.  5B). The cooperative effects of Bcl10 and MALT1 were specific for Bimp1, because co-transfection with Bcl10, with or without MALT1, had no effect on NF-B activation by RICK (Fig. 5B). Immunoblotting confirmed that neither Bimp1 nor RICK levels were affected by co-expression of Bcl10 and MALT1, indicating that Bcl10 and MALT1 were specifically influencing the function of the Bimp1 signaling pathway (Fig. 5B). These results provide evidence that Bcl10 and MALT1 can cooperate with Bimp1 in a signaling pathway leading to NF-B activation. To further verify the specificity of the functional cooperation of these proteins, we utilized a Bcl10 variant, L41Q, which contains a single point mutation in the CARD that abolishes binding to Bimp1 as well as the ability of Bcl10 to activate NF-B (8) (Fig. 4B). When Bcl10 (L41Q) was co-transfected with Bimp1, either in the absence or the presence of MALT1, absolutely no enhancement of NF-B activation was seen (Fig.  5C). Again, immunoblotting showed that the levels of Bimp1 expression were unaffected by co-transfection with Bcl10 (L41Q) and MALT1. These results support the hypothesis that Bimp1 is able to cooperate in a signaling pathway with Bcl10 and MALT1 through a CARD-mediated mechanism.
A Mutant of Bimp1 Lacking the CARD Blocks NF-B Activation Induced by CD3 Ligation, PMA/Ionomycin, and PKC-Unlike normal T-cells, Bcl10 deficient T-cells are unable to activate NF-B in response to CD3/CD28 (14). Bcl10-deficient lymphoid cells also fail to activate NF-B in response to PMA/ ionomycin treatment (14), suggesting that Bcl10 functions downstream of PKC in a pathway leading to NF-B activation. To investigate the role of Bimp1 in this pathway, we utilized the mutant version of Bimp1, which lacks the CARD (Bimp1⌬CARD) and therefore cannot bind to Bcl10 (Fig. 4C). We tested whether this mutant could function as a dominant negative by inhibiting anti-CD3, PMA/Ca 2ϩ ionophore, and PKC-mediated NF-B activation. First, we evaluated the effect of expression of Bimp1⌬CARD in 293T cells and found that Bimp1⌬CARD expression resulted in effective inhibition of PMA/Ca 2ϩ ionophore A23187-induced NF-B activation (Fig.  6A). In contrast, the mutant of Bimp1 that consists of the CARD alone (Bimp1CARD) had no inhibitory effect on NF-B induction by PMA/Ca 2ϩ ionophore (Fig. 6A). In control experiments, a dominant negative form of MyD88, a key factor in a different NF-B signaling pathway that is triggered by Toll/ FIG. 4. Bimp1 activates NF-B through a Bcl10-dependent mechanism and binds to Bcl10 through CARD/CARD interaction. A, Bimp1-mediated activation of NF-B is dependent on Bcl10. Bcl10 ϩ/Ϫ or Bcl10 Ϫ/Ϫ MEFs were transfected with 900 ng of the indicated expression plasmid: pcDNA3-Bimp1-HA, pcDNA3-Bimp2-HA, or pcDNA3-Nod1-HA. 48 h after transfection, the cells were lysed, and NF-B induction was assayed as described under "Experimental Procedures." The data shown are representative of four separate experiments. Transfections and assays were done in triplicate, and the error bars represent the standard deviations. B, Bimp1 binds specifically to the CARD of Bcl10. 293T cells were transfected with indicated combinations of expression plasmids encoding Myc-tagged Bcl10 (pcDNA3-Bcl10-Myc, 2 g) or Bcl10 mutant (pcDNA3-Bcl10 (L41Q)-Myc, 2 g), FLAG-tagged Bimp1 (pcDNA3-Bimp1-FLAG, 4 g), and Myc-tagged RICK (pcDNA3-RICK-Myc, 3 g). 24 h post transfection, the cells were lysed, and FLAG-tagged proteins were immunoprecipitated as described under "Experimental Procedures." The products were resolved by 10% SDS-PAGE and visualized by Western blot with indicated antibodies. C, the Bimp1 CARD binds to Bcl10. 293T cells were transfected with indicated expression plasmids: pcDNA3-Bimp1-FLAG, 4 g; pcDNA3-Bimp1⌬CARD-FLAG, 4 g; pcDNA3-BimpCARD-FLAG, 4 g; and pcDNA-Bcl10-Myc, 2 g. 24 h after transfection, the cells were lysed, and FLAG-tagged proteins were immunoprecipitated as described. The products were resolved by 8 -15% SDS-PAGE and visualized by Western blot with indicated antibodies. WB, Western blot; *, phosphorylated form of Bcl10.

FIG. 5. Bimp1 recruits MALT1 to a multimeric binding complex and cooperates with Bcl10-MALT1 in NF-B activation.
A, Bimp1 recruits MALT1 to a multimeric binding complex in which Bcl10 functions as a molecular bridge. 293T cells were transfected with indicated combinations of expression plasmids: pcDNA3-Bimp1-FLAG, 4 g; pcDNA3-Bcl10-Myc, 2 g; pcDNA3-Bcl10(L41Q)-Myc, 2 g; pcDNA3-Bcl10(⌬107-119)-Myc, 6 g; and pcDNA MALT1-(1-330)-HA, 6 g. 24 h after transfection, the cell extracts were prepared, and FLAG-tagged proteins were immunoprecipitated as described under "Experimental Procedures." The products were resolved by 10% SDS-PAGE and visualized by Western blot with indicated antibodies. B, co-transfection of Bcl0 and MALT1 with Bimp1 leads to enhanced NF-B induction. 293T cells were transfected with the indicated expression plasmids: (pcDNA3-Bcl10-Myc, 2.5 ng; pcDNA3-MALT1-Myc, 500 ng; pcDNA3-Bimp1-HA, 125 ng; and pcDNA3-RICK-Myc, 125 ng). NF-B induction was measured as described under "Experimental Procedures." The data shown are representative of four separate experiments. The experiments were done in triplicate, and the error bars represent the standard deviations. The cell extracts were subjected to immunoblotting with mouse monoclonal anti-HA antibody to assess the level of expression of Bimp1 or with rabbit polyclonal anti-Myc antibody to assess the level of RICK expression. C, the synergy between Bimp1, Bcl10, and MALT1 is dependent on an intact CARD in Bcl10. The expression plasmid pcDNA-Bcl10 (L41Q)-Myc, which encodes a mutant Bcl10 with a defective CARD, was utilized in place of the wild-type plasmid pcDNA3-Bcl10-Myc. Immunoblotting was performed as described above.
interleukin-1 receptors, did not block PMA-induced NF-B activation, whereas a dominant negative form of IKK␤, which operates downstream of Bimp1 and Bcl10, did inhibit activation (Fig. 6A). The effect of the Bimp1 dominant negative was specific in that cells treated with TNF␣ or transfected with an expression plasmid encoding a constitutively active mutant Toll-like receptor 2 (34) displayed full activation of NF-B in the presence of Bimp1⌬CARD (Fig. 6A). MALT1, a proposed downstream effector of Bcl10, can activate NF-B if its caspaselike domain is artificially dimerized via linkage to FKBP (13). NF-B induction by the MALT1-FKBP expression plasmid was not inhibited by co-expression of Bimp1⌬CARD, which is in agreement with the notion that Bimp1 operates upstream of Bcl10, whereas MALT1 operates downstream of Bcl10 in this signaling pathway (Fig. 6A).
We next examined the ability of the Bimp1⌬CARD to function as a dominant negative mutant in inhibiting NF-B activation triggered by anti-CD3 in the T-cell hybridoma cell line 68-41 (35). In this system, Bimp1⌬CARD expression blocked both anti-CD3 antibody and PMA/Ca 2ϩ ionophore-induced NF-B activation but had no effect on NF-B induction by TNF␣ (Fig. 6B). Again, expression of the Bimp1CARD mutant did not inhibit NF-B induction triggered by CD3 ligation or PMA (Fig. 6B). As expected, dominant negative MyD88 mutant did not block NF-B induction mediated by anti-CD3 or PMA, whereas dominant negative IKK␤ effectively blocked this pathway (Fig. 6B). To further verify that Bimp1 operated downstream of PKC, we next tested the ability of the Bimp1⌬CARD mutant to block PKC-induced NF-B activation. Bimp1⌬CARD blocked both PKC␣ and PKC⑀-mediated NF-B activation but did not block TNF␣ and Toll-like receptor 2-mediated activation (Fig. 6C). In control experiments, expression of a dominant negative form of DAP-kinase, a mediator of IFN-␥-induced programmed cell death, had no effect on NF-B activity (36), whereas dominant negative IKK␤ effectively blocked NF-B induction by PKC␣ and ⑀, TNF␣, and TLR2. Based on these results, we propose that Bimp1, through its interaction with Bcl10, functions to couple surface receptor stimulation and PKC activation to the induction of NF-B. DISCUSSION We have identified a novel protein, Bimp1, that binds Bcl10 and activates NF-B. Bimp1 is homologous to Bimp2, another protein that also activates NF-B in a Bcl10-dependent manner. The domain structures of Bimp1 and Bimp2 identify these proteins as members of the MAGUK family, a class of proteins that are composed of multiple discreet modules, including the SH3, PDZ, and GUK-like domains, which mediate proteinprotein interaction. We propose that Bimp1 and Bimp2 represent a subfamily of MAGUK proteins that are distinguished from other subfamilies by the presence of an N-terminal CARD. As this manuscript was completed, Bertin et al. (37) published a report identifying two CARD-containing MAGUK proteins that they named CARD11 and CARD14. CARD14 may represent the human orthologue of Bimp2. These authors showed that CARD11 and CARD14 bind to Bcl10 and activate NF-B. However, this report does not establish whether CARD11 and CARD14 operate upstream of Bcl10, nor does it address the relationship of CARD11, CARD14, and Bcl10 to T-cell receptor triggering and PKC activation (37).
To our knowledge, Bimp1 is the first molecule to be definitively identified as operating upstream of Bcl10 and as linking PKC activation to Bcl10-mediated NF-B induction. Based on our data, we propose a model for a Bimp-mediated signaling pathway leading to NF-B activation (Fig. 6D). In this proposed pathway, surface receptor stimulation leads to activation of PKC, and a Bimp protein functions as a critical link between PKC and Bcl10. Because the Bimp proteins are members of the MAGUK family, it is likely that Bimps function as molecular scaffolds in recruiting molecules such as plasma membrane receptors, phospholipase C, PKC, Bcl10, MALT1, and other as yet unidentified molecules to a large signaling complex, thereby holding together the signaling components that allow NF-B activation to occur (Fig. 6D). The Bimps may, like many MAGUK proteins, anchor to the cytoplasmic tail of surface receptors through PDZ domain interaction (20). Our data suggest that Bimp1 operates immediately upstream of Bcl10 with the CARD/CARD interaction between Bcl10 and Bimp1 being essential for NF-B signaling. Bimp1, Bcl10, and MALT1 can form a ternary complex whereby Bcl10 links Bimp1 to MALT1. The formation of this multiprotein complex has functional significance because Bimp1, Bcl10, and MALT1 can cooperate to activate NF-B. Recent studies have shown that MALT1 binds to Bcl10 through its immunoglobulin domains (12,13). Furthermore, Bcl10 mediates the oligomerization of MALT1, an event that is sufficient for NF-B activation, thus suggesting that MALT1 is a downstream effector of Bcl10 (13). It is not known whether MALT1 is the only downstream effector of Bcl10 or whether other molecules also function downstream of Bcl10. Bimp1/Bcl10-mediated activation of NF-B occurs through a mechanism that requires the regulatory subunit IKK␥ and the kinase subunits IKK␣ or IKK␤ of the IKK complex. The molecular events that link oligomerization and activation of the MALT1 caspase-like domain to activation of the IKK complex remain unknown.
Recent knockout studies revealed that the Bcl10-mediated NF-B activation pathway plays an essential role in antigen receptor signaling in B-and T-cells (14). In addition, the diacylglycerol/PMA-sensitive PKC isoforms have also been implicated as critical components of NF-B signaling in lymphoid cells (15)(16)(17)(18). For example, antigen receptor stimulation in T-cells activates PKC, a step that is required for NF-B induction (18). Likewise, PKC␤ has been shown to be essential for antigen receptor signaling in B-cells (15). We have now identified a molecule, Bimp1, which appears to link PKC activation to the Bcl10-dependent pathway of NF-B induction. Importantly, the role of Bimp1 in lymphocytes is specifically demonstrated by our studies of CD3 signaling in T-cell hybridomas. In these cells, activation of NF-B by anti-CD3 is completely blocked by a dominant negative mutant of Bimp1. The notion that PKC, Bimp1, and Bcl10 participate in a common NF-B signaling pathway is supported by the fact that mice deficient in PKC and PKC␤ demonstrate defects in lymphoid activation that are similar to those observed in mice lacking Bcl10 (15,18).
Although our results implicate Bimp1 in linking T-cell re-ceptor stimulation and PKC activation to Bcl10, it is possible that Bimp1 and Bimp2 may also be involved in PKC-dependent NF-B signaling pathways in a wide array of other physiologic systems including the developing neural system. Embryos from mice lacking Bcl10 as well as embryos lacking both IKK␣ and IKK␤ exhibit defects in neural tube closure associated with enhanced apoptosis in the hindbrain neuroepithelium (14,28). These observations suggest that Bcl10-mediated NF-B activation acts in a survival pathway to protect developing neurons. It will be important to determine whether PKC and Bimp proteins act upstream of Bcl10 in this neuronal survival pathway.
The importance of Bimps in linking PKC activation to NF-B signaling may extend to cellular processes other than lymphocyte activation and central nervous development. Recent studies suggest that PKC isoforms other than PKC and PKC␤ may mediate the induction of NF-B in a variety of physiologic systems. For example, PKC⑀ activation is a critical step in the induction of NF-B by platelet-derived growth factor (38). Also, the induction of NF-B that occurs in protective ischemic preconditioning in cardiomyocytes is mediating by PKC⑀ (39). PKC␣ has been implicated as an upstream activator of NF-B in both human osteosarcoma and lung epithelial cell lines (40,41). The finding that dominant negative Bimp1 inhibits NF-B activation induced by both PKC␣ and PKC⑀ suggests that Bimp1 may be involved in linking multiple classical and novel PKC isoforms to NF-B activation in a variety of cells. These events could be mediated by either Bimp1 and/or Bimp2, because these CARD/MAGUK proteins are highly homologous and are likely to participate in similar signaling pathways. Future studies will be aimed at exploring the potential role of the PKC/Bimp/Bcl10-MALT1 pathway in a variety of physiologic systems, identifying the upstream signals that trigger activation of PKC and characterizing the mechanism(s) by which PKC activation is linked to Bimp/Bcl10-MALT1 proteins in this novel NF-B signaling pathway.