Caspase Recruitment Domain (Card-Dependent) Cytoplasmic Filaments Mediate Bcl10-Induced Nf-κb Activation

Mucosa-associated lymphoid tissue (MALT) lymphomas with t(1;14)(p22;q32) display a recurrent breakpoint upstream of bcl10 that juxtaposes the gene under the control of the Ig heavy chain enhancer (Willis et al. 1999; Zhang et al. 1999). As a consequence of this translocation, bcl10 cDNAs from MALT lymphomas are overexpressed and contain a variety of mutations, mostly consisting of either nucleotide deletion or insertion, which result in truncations within or distal to the CARD (Willis et al. 1999; Zhang et al. 1999). Thus, overexpression of truncated forms of bcl10, which retain constitutive NF-κB–inducing activity, transform primary rat embryonic fibroblasts (Willis et al. 1999). However, the mechanism by which deregulated expression of bcl10 leads to cellular oncogenesis is currently not known.

Here, we show that bcl10 protein assemblates in cytoplasmic filaments that serve as scaffold for recruitment of NF-κB–activating signal transduction molecules. Cytochalasin D, a potent inhibitor of actin filament function, disassembles bcl10 filaments and specifically inhibits bcl10-induced NF-κB activation. Thus, assemblage of bcl10 in filaments connected to the cytoskeleton network is essential for bcl10-mediated NF-κB induction.

HeLa, HEK 293, and Rat-2 cells were cultured in DME 10% FCS. HeLa and Rat-2 cells were transfected using lipofectamine (GIBCO BRL); 293 cells were transfected by calcium phosphate precipitation. To assess NF-κB activation, HeLa and Rat-2 cells were transfected with 2 μg of the indicated cDNAs, together with pNF-κB-luc in 12-well plates. Cells were then lysed and luciferase activity was determined with the Luciferase Assay System (Promega). A plasmid expressing β-galactosidase was added to the transfection mixture for normalization of the efficiency of transfection. Cytochalasin D was obtained from Sigma Chemical Co.

PCR-based random mutagenesis of bcl10 was carried out in the following reaction buffer: 10 mM Tris-HCl, pH 8, 50 mM KCl, 0.5 mM MnCl₂, 125 μM dNTPs. Reaction was performed using 2.5 U standard Taq polymerase, 200 ng of plasmidic template, and the oligos 5′-AAGAATTCCATGGAGCCCACCGCACC (forward) and 5′-AACTCGAGTCATGGAAAAGGTTCACAACTGCT (reverse). PCR products were purified and cloned in pcDNA3 expression vector (Invitrogen) provided with an HA epitope. The vectors pNF-κB-luc and pCMV-β-gal were from Clontech.

HeLa cells were grown and transfected in chamber slides. 16 h after transfection, cells were fixed in 4% paraformaldehyde for 15 min at room temperature and then permeabilized in PBS/0.1% Triton X-100. Primary antibodies were incubated for 30 min in 5% FCS–PBS, followed by several washes with 5% FCS–PBS, and then incubating for 30 min with secondary antibody in 5% FCS–PBS. All steps were done at room temperature. Sources of antibodies and reagents for immunofluorescence were: anti-HA (Roche Molecular Biochemicals); anti-FLAG and anti–α-actinin (Sigma Chemical Co.); anti-TRADD and anti-RIP (Santa Cruz).

The two-hybrid screening was conducted using the Matchmaker system (Clontech) according to the manufacturer's instructions. In brief, yeast strain HF7c, expressing GAL4-DR4 fused protein, was transformed with a human peripheral blood leukocyte cDNA library cloned in the pGAD10 vector (Clontech) by lithium acetate/PEG 4000 procedure. 2 × 10⁶ clones were analyzed. Transformed yeast were selected on SD/agar plates lacking leucine, tryptophane, and histidine for 5 d at 30°C. Selected colonies were blotted on filter paper, permeabilized in liquid nitrogen, and placed on another filter soaked in Z buffer (60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1mM MgSO₄, and 37.5 mM β-mercaptoethanol) containing 1 mM 5-bromo-4-chloro-3-indolyl-β-d-galactoside. Colonies that developed color were restreaked on selective plates to allow plasmid segregation and tested again for β-galactosidase activity. The liquid β-galactosidase assay was performed according to the manufacturer's instructions using CPRG as substrate.

Recombinant histidine-tagged proteins were made in Escherichia coli BL21 strain using the pET expression system (Novagen). Proteins were purified with Ni-NTA agarose beads (Quiagen) and mixed with recombinant α-actinin (Sigma Chemical Co.) in E1A buffer (50 mM Hepes, 500 nM NaCl, 0.1% NP-40, and 10% glycerol). Samples were incubated for 1–2 h at 4°C, extensively washed by pulse centrifugation with E1A buffer and resuspended in 40 μl sample buffer. 5 μl of the reaction was loaded for SDS-PAGE and Western blot analysis.

To determine the cellular localization of oncogenic bcl10, we transfected HeLa cells with an HA-tagged vector encoding the full-length bcl10, and the expressed protein was detected using a monoclonal anti-HA antibody. The results in Fig. 1 show that bcl10 exhibits a clear pattern of discrete and interconnecting cytoplasmic filaments resembling the death-effector filaments (Perez and White 1998; Siegel et al. 1998). Similar filaments were observed when bcl10 was visualized using specific affinity-purified rabbit antisera raised against a bcl10-derived polypeptide (Fig. 1 D). The filaments were not an artifact of the staining procedure, since competition with the immunogenic peptide abolished the immunofluorescence (data not shown). Double-staining experiments indicated that bcl10 filaments partially overlap other cellular filaments, such as tubulin, keratin, and actin (data not shown).

The NH₂-terminal region of bcl10 contains a CARD motif (Costanzo et al. 1999; Koseki et al. 1999; Srinivasula et al. 1999; Thome et al. 1999; Willis et al. 1999; Yan et al. 1999; Zhang et al. 1999), which is both necessary and sufficient for NF-κB induction (Costanzo et al. 1999; Koseki et al. 1999; Willis et al. 1999; Zhang et al. 1999). To verify whether the CARD was involved in determining the filamentous arrangement of bcl10, we tested deleted versions of bcl10 for filament formation. A truncated bcl10 polypeptide (bcl10^128-233) that lacks the CARD region showed a nonfilamentous diffuse distribution, whereas the CARD alone (bcl10^1-127) formed filaments identical to the full-length protein (Fig. 2 A). Since the CARD of bcl10 retains the NF-κB–inducing activity, these experiments suggested a correlation between filaments formation and bcl10-mediated NF-κB activation. To further explore this possibility, we generated the point mutants bcl10(L41Q) and bcl10(G78R), because mutations of corresponding residues in the CARD of the cell death-inducing proteins, CED-3 and RAIDD, abrogate their functional activity (Shaham and Horvitz 1996; Duan and Dixit 1997). Although expressed at levels comparable to the wild-type protein (data not shown), both mutants, bcl10(L41Q) and bcl10(G78R), did not form filaments and failed in activating NF-κB (Fig. 2 B).

bcl10 filament formation is not an effect of NF-κB activation because when bcl10 was cotransfected with dominant negative forms of IκB and NIK, which abrogate bcl10-induced NF-κB activation (Costanzo et al. 1999; Koseki et al. 1999; Srinivasula et al. 1999), bcl10 filaments were formed (Fig. 2 C).

To determine whether bcl10 filaments are unequivocally linked to NF-κB activation, we generated a panel of random mutations in the CARD region of the protein. Microscopy and functional analysis of these mutants indeed revealed that only bcl10 mutants still able to form filaments could activate NF-κB (Fig. 3).

Previously, we and others have shown that bcl10 activates NF-κB through a molecular pathway involved in tumor necrosis factor receptor 1 (TNF-R1) signaling (Costanzo et al. 1999; Koseki et al. 1999; Srinivasula et al. 1999). To determine whether bcl10 filaments served as scaffolds for recruitment of signal transduction molecules, we performed double-staining experiments of bcl10 together with antibodies against proteins involved in TNF-RI signaling. Fig. 4 A shows that TRADD, an adapter molecule that is recruited to TNF-RI complex and signals NF-κB activation (Hsu et al. 1995), localizes on bcl10 filaments. To determine whether TRADD could be actively recruited on bcl10 filaments, HeLa cells were left untreated or transfected with bcl10 and localization of endogenous TRADD was detected with anti-TRADD antibody by confocal microscopy. The results in Fig. 4B and Fig. C, show that whereas in untransfected cells, endogenous TRADD displays a diffuse cytoplasmic and nuclear distribution, expression of bcl10 results in translocation of cytoplasmic TRADD to bcl10 filaments. This observation is consistent with coprecipitation of TRADD with bcl10 (Costanzo et al. 1999), and implies that only filamentous organization of bcl10 allows recruitment of TRADD. The death domain-containing serine/threonine kinase RIP is essential for TNF-induced NF-κB signaling (Kelliher et al. 1998) and is recruited to TNF-R1 complex via interaction with TRADD (Hsu et al. 1996). Similarly to TRADD, whereas in untransfected cells endogenous RIP displays a diffuse cytoplasmic localization (data not shown), expression of bcl10 results in translocation of RIP to bcl10 filaments (Fig. 4 D; and data not shown).

To assess whether cytoskeleton dynamics influence bcl10 activity, HeLa cells expressing bcl10 were treated with cytochalasin D, a potent inhibitor of actin filament function, and bcl10 localization was determined. As shown in Fig. 5, cytochalasin D disassembles bcl10 filaments and specifically inhibits bcl10-induced NF-κB activation, whereas NF-κB activation induced by TNF stimulation or NIK overexpression is not affected. Thus, assemblage of bcl10 in filaments is essential for bcl10-mediated NF-κB induction.

bcl10 filament formation could either reflect an intrinsic property of that protein or be ordered via interactions with preexisting filamentous proteins. The observation that cytochalasin D disassembles bcl10 filaments prompted us to explore the possibility these filaments might be organized by CARD-mediated association with cytoskeletal components. To test for this, we performed a two-hybrid screen fusing the CARD of bcl10 to the DNA binding domain of GAL4 and searched a plasmid library of fusion between the GAL4 transcription activation domain and cDNAs from peripheral blood leukocytes. 13 independent clones were isolated that activated the β-galactosidase reporter gene when ∼2 × 10⁶ transformants were analyzed. Restriction mapping and partial sequencing of these 13 cDNAs revealed that six positive clones had the same ∼2-kb cDNA insert encoding for the polypeptide Pro³⁵⁵–Leu⁸⁹² of α-actinin, a member of the actin-binding proteins superfamily that is thought to cross-link actin filaments. As summarized in Table, a library plasmid encoding for Pro³⁵⁵–Leu⁸⁹² of α-actinin did not activate β-galactosidase by itself, when coexpressed with the empty GAL4BD vector, or with unrelated control plasmids. Conversely, it strongly interacted with the CARD of bcl10. However, both wild-type and mutant forms of bcl10 bind similarly to α-actinin in yeast and in vitro (Table and Fig. 6 A). Together with cytochalasin D experiments, the two-hybrid data suggest that interaction of bcl10 with cytoskeleton components is necessary, but not sufficient for filament formation.

Next, we tested whether filament formation could result from bcl10 multimerization. Immunoprecipitation analysis shown in Fig. 6 B revealed that wild-type bcl10 self-associates via a CARD-mediated homophilic interaction, however, both mutants bcl10(L41Q) and bcl10(G78R) did not dimerize, suggesting that bcl10 filaments result from self-assembly. Thus, both dimerization of bcl10 and binding to α-actinin are necessary for filament formation and NF-κB activation.

bcl10 is expressed in many normal tissues and overexpression of the gene causes cellular transformation (Willis et al. 1999; Zhang et al. 1999). Therefore, we tested whether filamentous organization of bcl10 and deregulated NF-κB activation both derive from disregulated level of expression of the protein. Indeed, in Rat-2 cells, endogenous bcl10 displays a diffuse cytosolic distribution, whereas overexpression of the gene induces filament formation and NF-κB activation (Fig. 7). These results, however, do not exclude the possibility that local micro-assembling of the protein may physiologically occur in normal conditions or after specific stimulation.

Independent evidence suggests that the transforming capability of bcl10 is the result of a nonreceptor-mediated NF-κB activation arising from deregulated expression of the gene (Willis et al. 1999; Zhang et al. 1999). In the present study, we have shown that organization of bcl10 into ordered structures is essential for NF-κB–inducing activity of the protein. Similarly to scaffolding and anchoring proteins that determine selective activation of mitogen-activated protein kinases by sequestration and localization of signaling complexes, filamentous arrangement of bcl10 is required for recruitment of TRADD, and may be essential for local concentration and activation of other downstream signaling proteins. Hence, as recruitment of TRADD to TNF-RI complex initiates a signal transduction pathway leading to NF-κB activation, the same signaling pathway can be activated upon TRADD recruitment to different cellular compartments.

Cytoskeletal-like shape of bcl10 filaments and direct interaction of bcl10 with cytoskeletal proteins suggest that cytoskeleton dynamics may regulate bcl10 signaling. Indeed, activation of the NF-κB transcription factors is directly influenced by changes in the cytoskeleton network (Rosette and Karin 1995). bcl10 filaments resemble the death-effector filaments, cytoplasmic filaments that sequester and activate procaspase zymogens (Perez and White 1998; Siegel et al. 1998), and similar structures are formed when the CARD of caspase-2 is overexpressed (Colussi et al. 1998). Generation of death-effector filaments is essential for activation of a signaling cascade leading to cellular apoptosis, and inhibition of filament formation prevents cell death. Thus, formation of ordered assembly of protein may represent a general mechanism for receptor-independent activation of specific signal transduction pathways.

We thank Drs. J. Andersson, J. Pieters and L. D'Adamio for comments and critical review of the manuscript.

The Basel Institute for Immunology was founded and is supported by F. Hoffmann-La Roche and Co. Ltd., CH-4005 Basel, Switzerland.