MAGUK-Controlled Ubiquitination of CARMA1 Modulates Lymphocyte NF-κB Activity

Abstract

The adaptor protein CARMA1 is required for antigen receptor-triggered activation of IKK and JNK in lymphocytes. Once activated, the events that subsequently turn off the CARMA1 signalosome are unknown. In this study, we found that antigen receptor-activated CARMA1 underwent lysine 48 (K48) polyubiquitination and proteasome-dependent degradation. The MAGUK region of CARMA1 was an essential player in this event; the SH3 and GUK domains contained the main ubiquitin acceptor sites, and deletion of a Hook domain (an important structure for maintaining inactive MAGUK proteins) between SH3 and GUK was sufficient to induce constitutive ubiquitination of CARMA1. A similar deletion promoted the ubiquitination of PSD-95 and Dlgh1, suggesting that a conserved mechanism may control the turnover of other MAGUK family protein complexes. Functionally, we demonstrated that elimination of MAGUK ubiquitination sites in CARMA1 resulted in elevated basal and inducible NF-κB and JNK activation as a result of defective K48 ubiquitination and increased persistence of this ubiquitination-deficient CARMA1 protein in activated lymphocytes. The coordination of degradation with the full activation of the CARMA1 molecule likely provides an intrinsic feedback control mechanism to balance lymphocyte activation upon antigenic stimulation.

The CARD-containing MAGUK protein 1 (CARMA1, or CARD11) is regarded as an orchestrator of both T-cell-dependent and T-cell-independent immune responses due to its requirement in the activation of IKK and JNK signaling pathways downstream of antigen receptor (AR) ligation in B and T cells (3, 6, 8, 17). CARMA1 overexpression and/or mutations have also been associated with lymphomagenesis, as it promotes sustained activation of NF-κB-dependent cell survival (10, 16, 18). Structurally, CARMA1 is a multidomain adaptor protein containing a caspase recruitment (CARD) and a coiled-coil (CC) domain linked upstream of a region that is related to the MAGUK family of proteins. This MAGUK region contains a postsynaptic density 95/disc large/zona occludens 1 (PDZ), a SRC homology 3 (SH3), and a guanylate kinase-like (GUK) domain (2, 20). In addition, CARMA1 contains a flexible serine/threonine-rich linker that bridges the CC and MAGUK domains. Phosphorylation of this linker by protein kinase Cβ (PKCβ) or PKCθ controls the activation status of CARMA1 (13, 29, 30); thus, this region has been designated as the PKC-regulated domain (PRD) (22). It is likely that PRD phosphorylation destabilizes an inhibitory conformation in CARMA1 that exposes the various interaction domains required to assemble its downstream signaling components. Consistent with this model, deletion of the PRD results in a constitutively active CARMA1, resulting in high basal NF-κB activation (14, 30).

Proximal downstream adaptors of CARMA1 include BCL10 and MALT1 (22). Genetic deletion of any of these proteins in cellular and animal models has revealed the importance of this pathway to immune cell function. While AR-induced activation of early signaling pathways, such as protein tyrosine phosphorylation, intracellular Ca²⁺ flux, and activation of extracellular signal-regulated kinase (ERK) and Akt are intact in CARMA1-, BCL10-, or MALT1-deficient lymphocytes, activation of NF-κB and JNK signaling pathways is markedly impaired (6, 24-26). This is manifested in defective lymphocyte proliferation and survival and in reduced immune responses.

While the events leading to the activation of the IKK signaling complex downstream of CARMA1 have been well characterized, the signals that down-modulate this pathway are less well understood. Studies of BCL10 turnover have yielded some possible mechanisms (34). After cell activation, BCL10 is posttranslationally modified by both phosphorylation (possibly via IKKβ or CaMKII) and polyubiquitination (polyUb) and undergoes degradation that results in the down-modulation of NF-κB activity. Several ubiquitin protein ligases (E3s), including Itch, NEDD4, cIAP2, and βTrCP, have been reported to drive BCL10 ubiquitination in lymphocytes and to promote its degradation through either lysosomal or proteasomal pathways (7, 12, 27, 37). Overexpression of such E3s downregulates BCL10-dependent pathways, including NF-κB and the production of interleukin-2. Interestingly, antigen receptor-induced phosphorylation and degradation of BCL10 is not observed in the absence of CARMA1 (12), indicating a role for CARMA1 in BCL10 turnover.

In this report, we demonstrate that endogenous CARMA1 is directly ubiquitinated and degraded by the proteasome in AR-activated lymphocytes. Structure-function analyses showed that the primary targets for ubiquitination within CARMA1 were localized within the MAGUK region. Mutation of all lysine residues (potential ubiquitin modification targets) to arginines within the MAGUK of CARMA1 produced a hyperactive molecule that promoted high NF-κB and JNK activation levels. Unlike the wild-type (WT) CARMA1 molecule, a lysine-to-arginine mutant CARMA1 was not modified by polyUb chains upon cell activation and had a resulting increase in protein stability. We identified a region between the SH3 and GUK domains that is highly similar to a region (termed the Hook region) that regulates the conformation of the MAGUK protein PSD-95. Notably, deletion of this Hook region was sufficient to trigger polyUb of CARMA1 as well as PSD-95 and Dlgh1, other MAGUK family proteins. These data suggest that activation of CARMA1 initiates a feedback mechanism controlled by the MAGUK domain that triggers ubiquitination and degradation of the CARMA1 signalosome, thereby limiting NF-κB and JNK signaling.

MATERIALS AND METHODS

Cells lines and reagents.

Ramos B, Jurkat T, and human embryonic kidney 293T cells were obtained from ATCC and were cultured in RPMI 1640 (Ramos and Jurkat) or Dulbecco's modified Eagle's medium (293T) with 10% fetal bovine serum. PKCβ^−/− DT40, CARMA1^−/− DT40, and CARMA1^−/− DT40 cells expressing Flag-CARMA1 were generated as described previously (29) and cultured in RPMI 1640 with 10% fetal bovine serum and 1% chicken serum. Primary murine lymphocytes were isolated from spleens using standard methods (31). Splenic B cells were purified by negative selection using anti-CD43 magnetic beads (Miltenyi Biotech). Cycloheximide, phorbol 12-myristate 13-acetate (PMA), ionomycin, chloroquine, ammonium chloride, protease inhibitor cocktail, and phosphatase inhibitor cocktails 1 and 2 were purchased from Sigma-Aldrich. Ro-318425 inhibitor, Z-Leu-Leu-Leu-al (MG132), and ALLN (calpain inhibitor 1) were purchased from EMD Biosciences. Anti-human CD3 (OKT3) was a gift from Steve Ziegler; anti-mouse CD3 (2C11) and anti-mouse and anti-human CD28 antibodies were purchased from BD Biosciences. Rat anti-mouse CD40 (IC10) was purchased from Southern Biotech. Antibodies against IKKγ (FL-419), IKKα/β (Η-470), ERK (C-16), pERK (E-4), hemagglutinin (HA; Y-11), and IκBα (C-15) were from Santa Cruz Biotechnology. Rabbit anti-pJNK was from Cell Signaling. Polyclonal anti-CARMA1 (AL220) was from Alexis Biochemical. Anti-actin, anti-myc (9E10), and anti-Flag (M2) antibodies were purchased from Sigma-Aldrich. Anti-ubiquitin (FK2) was from Biomol International. IRDye800 goat anti-rabbit IgG and Alexa Fluor 680 goat anti-mouse IgG were purchased from Rockland Immunochemicals and Invitrogen, respectively. cIAP antagonist (BV6) and anti-K48-specific antibodies were a gift from V. Dixit (Genentech).

Plasmid constructs.

Vectors for retroviral expression of myc-tagged murine CARMA1 and myc-CARMA1-ΔPRD were described previously (30). For transient expression of full-length murine CARMA1 and CARMA1-ΔPRD and other MAGUK proteins, we first created a new expression vector. A double-stranded oligonucleotide encoding a myc eptitope with a strong Kozak translation start sequence was ligated into the 5′-most restriction site of the pcDNA3 (Invitrogen) multiple cloning site (MCS) by using standard molecular biology techniques. At the 3′ end of the MCS, an internal ribosomal entry site (IRES) followed by sequence encoding enhanced green fluorescent protein (GFP) was inserted. This construct allowed for direct cloning to achieve an N-terminal myc epitope tag and with bicistronic expression of GFP to assess transfection efficiency. Murine CARMA1, CARMA1-ΔPRD, rat DLG1 (a gift from Ronald Javier), and rat PSD-95 (obtained from Wei-dong Yao; Addgene plasmid 15463) were amplified from their parental vectors using a high-fidelity polymerase (Pfx50; Invitrogen) according to the manufacturer's protocol. The primers used in these amplifications contained chimeric 5′ restriction enzyme sites that allowed for direct digestion of PCR products and ligation into the desired vector. For transient expression of myc-tagged fragments of CARMA1, desired fragments were amplified as described above from the murine CARMA1 coding sequence. For these fragments of CARMA1, amplification primers with chimeric 5′ restriction digest sites were designed to achieve in-frame translation downstream of a 5′ 6×myc tag (23) cloned into pcDNA3. The entire coding sequences of all inserts generated by PCR amplification were verified by DNA sequencing. The amino acid sequence of the Hook domain was deleted from CARMA1, PSD-95, and Dlgh1 constructs by using a single site-directed mutagenesis reaction for each sequence (QuikChange XL; Stratagene). The coding sequence for CARMA1 KR was purchased from Mr. Gene and cloned into a mammalian expression vector by using standard molecular biology techniques.

Coding sequences of several E3 ligases were cloned into p3XFLAG-CMV 7.1 (Sigma-Aldrich). In each case, coding sequences were amplified as described above and allowed in-frame translation downstream of the 3×Flag epitope. Included in the cloning scheme were coding sequences for the following: human Cbl-b, amplified from parental plasmid pEF-Cbl-b (a gift from Yun-Cai Liu); human cIAP2, derived from an Incyte clone obtained from Open Biosystems (clone ID LIFESEQ8686507); murine Itch and murine NEDD4, amplified from plasmids obtained from Allan Weissman (Addgene plasmids 11427 and 11426, respectively). A plasmid expressing Flag-tagged human βTrCP was used as obtained from Peter Howley (Addgene plasmid 10865). To create 3×Flag cIAP2 with an inactivating point mutation within the RING domain (H574A), site-directed mutagenesis was performed as described above.

pCS 8×HA-Ub was a gift from Jim Roberts. Vectors for expression of HA-Ub with the various lysine-to-arginine substitutions (R48, K48, and R63) were gifts from Hui Xiao. The NF-κB reporter construct Igκ-IFN-luciferase was a gift from Joel Pomerantz; the Renilla luciferase transfection control pRL-TK was from Promega.

Generation of stable cell lines using retroviral transduction.

293T cells were transfected using the calcium phosphate method and 10 μg of a 10A1 packaging plasmid and 10 μg of mouse stem cell virus retroviral plasmid containing either myc-CARMA1-WT-IRES-GFP or the mutant myc-CARMA1-ΔPRD-IRES-GFP or myc-CARMA1-KR-IRES-GFP in the presence of 25 μM chloroquine. Cells were incubated with viral supernatants diluted 1:1 with fresh medium and 8 μg/ml of Polybrene and spinoculated for 20 min at 800 × g. Virally transduced cells were sorted based on GFP expression using a FACSAria cell sorter (BD Biosciences).

Transient transfections and IPs.

293T cells were transfected using FuGENE 6 (Roche) for 48 h with a 1:6 ratio of DNA and FuGENE 6. In most experiments, 25 μM proteasome inhibitor MG132 or ALLN was added to the cells 1 h before harvesting. For immunoprecipitations (IP), 293T cells or DT40 cells were harvested in radioimmunoprecipitation assay (RIPA) buffer (250 mM NaCl, 10 mM Tris-HCl, pH 7.4, 0.1% or 1% SDS, 0.5% sodium deoxycholate, 1%Triton X-100, 1 mM Na₃VO₄, 1 mM NaF, 15 mM n-ethylmaleimide, and protease and phosphatase inhibitors), and cells were vortexed thoroughly and diluted 10-fold with RIPA buffer without SDS and sodium deoxycholate. After incubation on ice for 15 min and centrifugation at 16,000 × g for 15 min, supernatants were collected and incubated with 1.5 μg of anti-myc antibody and protein G-Sepharose (GE Healthcare BioSciences) for 2 h at 4°C. Alternatively, supernatants received 25 μl of anti-Flag M2 affinity gel (Sigma-Aldrich) for 2 h at 4°C. Beads were extensively washed in detergent-free RIPA buffer, and proteins were eluted by boiling in 2× Laemmli sample buffer (Bio-Rad).

Inhibitor treatments, cell activation, and Western blotting (WB).

For analysis of protein degradation, total splenocytes or cell lines were preincubated for 30 min with or without 50 μM cycloheximide (CHX). Cells stably expressing myc-CARMA1-ΔPRD were preincubated with the lysosome inhibitors ammonium chloride (50 mM) and chloroquine (100 μM) or with the proteasome inhibitors MG132 and ALLN (50 μM each) in complete culture medium for 20 min, followed by incubation with CHX or the vehicle (dimethyl sulfoxide [DMSO]) for 3 h. Cells were lysed in RIPA buffer.

myc-CARMA1-ΔPRD degradation was analyzed in Ramos cells after addition of CHX to fresh medium for the times indicated. For analysis of endogenous CARMA1 degradation in activated cells, primary lymphocytes isolated from mouse spleens were serum starved for 1 h, pretreated with CHX, and then stimulated with PMA (200 nM) and ionomycin (1 μg/ml) or 10 μg/ml of AR-specific antibodies for various time intervals. In all cases, cells were lysed with RIPA buffer and proteins were resolved in 8 or 10% SDS-PAGE gels. For analysis of endogenous ubiquitination of CARMA1, CARMA1^−/− DT40 cells stably expressing myc- or Flag-CARMA1 (4 × 10⁷/condition) were pretreated with 25 μM proteasomal inhibitor ALLN for 30 min at room temperature, and the cells were left unstimulated or stimulated with PMA (1 μM) and ionomycin (1 μg/ml) or 5 μg of anti-chicken IgM (M4) at 37°C. Whole-cell lysates (WCL) were prepared, CARMA1 was immunoprecipitated overnight using 3 μg of anti-myc, and complexes were collected with protein G beads for 3 h or with anti-Flag M2 affinity gel (Sigma-Aldrich) for 2 h at 4°C, respectively (45 μl). Proteins were transferred to polyvinylidene difluoride and blocked with Odyssey blocking buffer (Li-Cor Biosciences). Primary antibodies were incubated overnight. Secondary antibodies were incubated for 1.5 h at room temperature. Proteins were visualized and quantified by fluorescence detection using the Odyssey infrared imaging system from Li-Cor Biosciences.

NF-κB luciferase activity.

293T cells (5 × 10⁵ per sample) were transfected for 24 h using FuGENE 6 (Roche Applied Science, Indianapolis, IN) with 0.5 μg of both Igκ2-IFN-luciferase (NF-κB reporter) and pRL-TK (transfection control), 0.5 μg of myc-CARMA1-WT, and 1 μg of Flag-cIAP2, Flag-NEDD4, or Flag-Cbl-b. Five hours before harvesting the cells were either left unstimulated or stimulated with PMA (1 μM) and ionomycin (1 μg/ml). Jurkat T cells (10⁶ cells/sample) were cotransfected using Superfect (Qiagen, Valencia, CA) for 48 h with 3.2 μg of Igκ2-IFN-luciferase and 266 ng of pRL-TK, 1 μg of myc-CARMA1-WT, and 2 μg of myc-CARMA1-KR. Other series of experiments included 300 ng of cIAP2-WT or the cIAP2-RING mutant. Six hours before harvesting, cells were stimulated in plates coated with anti-CD3ɛ (10 μg/ml) and soluble anti-CD28 (1 μg/ml). NF-κB activation was analyzed by measuring the firefly and Renilla luciferase activities according to the dual luciferase reporter assay protocol from Promega (Madison, WI).

RESULTS

An activated CARMA1 mutant is rapidly degraded in lymphocytes.

We previously demonstrated that deletion of the PRD of CARMA1 (CARMA1-ΔPRD) produces a constitutively active protein which assembles the CARMA1 signalosome and triggers high levels of NF-κB activation in lymphoid cells (30). However, we noted that stably expressed CARMA1-ΔPRD consistently produced ∼5- fold-lower protein levels than the stably expressed WT-CARMA1 (Fig. 1A). This difference occurred despite equal levels of transgene mRNA expression, as measured by levels of the cis-linked GFP marker (Fig. 1A, bottom). These findings suggested that a posttranscriptional event(s) altered the stability CARMA1-ΔPRD. We therefore compared the turnover rates of WT-CARMA1 and CARMA1-ΔPRD proteins in unstimulated lymphocytes. Ramos B cells stably expressing these constructs were treated with CHX followed by anti-myc immunoblotting to detect CARMA1 (Fig. 1B and C). The protein level of WT-CARMA1 was partially reduced in the presence of CHX (around 30% at 2 h). However, CARMA1-ΔPRD levels were reduced ∼50% within 1 h of cycloheximide addition, reaching more than 80% degradation 2 h post-CHX addition, indicating that this active mutant is strongly and constitutively degraded. The levels of IKKα/β (data not shown) and NEMO (IKKγ [Fig. 1B]) were unaffected under the same experimental conditions, indicating the specificity of this degradation. These experiments clearly show that the ΔPRD mutation that generates a constitutively activated CARMA1 also promotes a marked decrease in subsequent protein stability.

FIG. 1.

Degradation of stably expressed, constitutively active CARMA1 in human Ramos B cells. (A) Expression of myc epitope (CARMA1) and IKK-α/β in human Ramos B cells stably transduced with retroviral vectors expressing myc-CARMA1 or myc-CARMA1-ΔPRD and cis-linked IRES GFP. Expression of bicistronic GFP was analyzed by fluorescence-activated cell sorting (bottom). Solid histograms, untransduced cells; open histograms, myc-CARMA1-WT- or -ΔPRD-expressing cells. Numbers indicate the GFP median fluorescence intensity. (B) CARMA1-transduced Ramos cells were incubated with 50 μM CHX from 0 to 2 h, and WB analysis was carried out using antibodies for myc (CARMA1), IKKγ, and actin. (C) Densitometry of CARMA1 turnover in cells expressing myc-CARMA1-WT versus myc-CARMA1-ΔPRD (medians ± standard errors of the means of four independent experiments). Ratios of myc-CARMA1 versus actin levels quantified by fluorescence detection were used for normalization, and relative fold differences were determined using the zero time point set as 1. n.s., not significant; **, P ≤ 0.01; *, P ≤ 0.05 according to Student's t test.

Endogenous CARMA1 is degraded in AR- and phorbol ester-stimulated lymphocytes.

We reasoned that degradation of the constitutively active CARMA1-ΔPRD mutant might result from cellular systems designed to remove misfolded proteins or, alternatively, result from a negative feedback signal downstream of the activated protein. To determine whether degradation of WT-CARMA1 also occurred following activation, we stimulated primary murine splenocytes (Fig. 2A and B) or purified splenic B cells (Fig. 2C) with or without chemical signals mimicking a strong AR signal (PMA and ionomycin [P/I]) or via AR-specific antibodies in the presence of the protein synthesis inhibitor CHX. Cell lysates were harvested using RIPA buffer to ensure solubilization of activated lipid raft-associated or cytoskeletal-associated CARMA1 (5, 30), and levels of endogenous CARMA1 were detected by immunoblotting. Without stimulation, the levels of CARMA1 diminished only slightly. When stimulated, the expression of CARMA1 was reduced as early as 1 h poststimulation and was 70% depleted 3 h after stimulation with either P/I (Fig. 2A) or anti-CD3/CD28 (Fig. 2B). A similar result was observed in anti-IgM-stimulated splenic B lymphocytes, in which CARMA1 degradation was clearly observed 3 h poststimulation (Fig. 2C). As a negative control, CD40 stimulation of splenic B cells (which does not signal through CARMA1) did not substantially alter the levels of CARMA1 protein, despite inducing strong IκBα degradation. These data indicate that P/I or AR stimulation shortens the half-life of endogenous CARMA1.

FIG. 2.

Degradation of endogenous CARMA1 in AR- or P/I-stimulated primary lymphocytes. Total splenocytes or splenic B cells were purified from adult C57/B6 mice, serum starved, and pretreated with 50 μM CHX. (A) Splenocytes were left unstimulated (DMSO) or stimulated with P/I for the indicated times. (Lower graph) Quantification of CARMA1 expression showing P/I-stimulated versus unstimulated cells. The graph shows means ± standard errors of the means from four independent experiments. (B) Splenocytes were left unstimulated (−) or stimulated with 10 μg/ml of both anti-CD3 and anti-CD28 for the indicated times. (C) Purified splenic B cells were stimulated with 10 μg/ml of anti-IgM or anti-CD40 (IC10) for the indicated times. In all experiments total CARMA1, IκB-α, and ERK expression levels were evaluated by immunoblotting (WB). Ratios of CARMA1 versus ERK levels were used for normalization, and fold differences were determined using the zero time point set as 1. n.s., no significant; **, P ≤ 0.01 according to Student's t test.

CARMA1 degradation is controlled by the proteasome.

Protein degradation occurs through two main cellular routes: the ubiquitin-proteasome and the autophagy-lysosome pathways. As an initial approach to determine which pathway mediates degradation of activated CARMA1, Ramos B cells stably expressing CARMA1-ΔPRD were preincubated with proteasome (ALLN or MG132) or lysosome (ammonium chloride or chloroquine) inhibitors and then subsequently treated with CHX for 3 h (Fig. 3A). Inhibition of the proteasome, but not the lysosome, blocked degradation of CARMA1-ΔPRD and IκBα (a positive control that is constitutively targeted for proteasomal degradation downstream of CARMA1-ΔPRD). To verify that endogenous CARMA1 in primary lymphocytes was also degraded by the proteasome, purified splenocytes were isolated and preincubated with proteasome inhibitors and CHX and then stimulated with P/I (Fig. 3B). While degradation of CARMA1 was observed without proteasome inhibitors, its turnover was significantly halted in their presence, revealing that CARMA1 degradation is regulated by the proteasome. Note that residual degradation was observed with proteasome inhibitors for both CARMA1 and the IκBα positive control, indicating that these inhibitors incompletely suppress the proteasome, a finding previously reported by others (4, 19).

FIG. 3.

Constitutively active and endogenous CARMA1 are processed by the proteasome. (Α) Ramos B cells expressing myc-CARMA1-ΔPRD were either untreated (PBS or DMSO) or preincubated with lysosome inhibitors (NH₄Cl or chloroquine) or proteasome inhibitors (MG132 or ALLN) for 20 min and subsequently treated with CHX or vehicle (DMSO) for 3 h. myc-CARMA1-ΔPRD, IκB-α, and actin were detected by WB. (B) Splenocytes were preincubated with DMSO or 100 μM proteasomal inhibitor ALLN or MG132 and then stimulated with P/I for the indicated times. Endogenous CARMA1, IκB-α, and ERK were detected by WB. Ratios of myc or endogenous CARMA1 versus actin or ERK levels were used for normalization, and relative fold differences were determined using the zero time point set as 1.

Cell activation targets CARMA1 for K48-linked polyubiquitination.

Proteins are usually targeted for proteasomal degradation by posttranslational modification with ubiquitin chains (polyUb) that are linked through conjugation at K48 of the ubiquitin molecule. To validate that activated CARMA1 is targeted for proteasomal degradation, we assessed whether activation could trigger polyUb modification of CARMA1. Due to the difficulty of efficiently immunoprecipitating endogenous CARMA1 with available antibodies, we used tagged, full-length CARMA1 proteins overexpressed in cell lines, along with constructs expressing HA-tagged ubiquitin (HA-Ub). Cells were lysed using moderately stringent lysis conditions (RIPA buffer) (21, 36), and CARMA1 was immunoprecipitated from cell extracts using anti-myc antibodies and protein G beads (myc-IP). Subsequent Western blotting allowed the detection of polyUb using anti-HA antibodies.

Consistent with its constitutive proteasome-dependent turnover, a high-molecular-weight smear of polyUb proteins was observed above CARMA1-ΔPRD (a characteristic of polyUb conjugation) (Fig. 4A; see also Fig. 6D, below). Similarly, a basal polyUb smear could also be detected in WT-CARMA1 (Fig. 4C, left panel; see also Fig. 6D). Notably, anti-myc antibodies also detected these high-molecular-weight smears in WT, ΔPRD, or subdomains of CARMA1 in multiple experiments (Fig. 4A and D; see also Fig. 5A and 6A, bottom panels), suggesting that CARMA1, rather than a coassociated protein, was directly modified by polyUb attachment. As expected, detection of polyUb-conjugated proteins was less sensitive when using the myc antibody (one tag per protein) than with the HA antibody (many polyUb tags per protein). To support the idea that CARMA1, and not a copurifying protein, is directly conjugated by polyUb, we also evaluated CARMA1 polyUb under more stringent cell lysis conditions (RIPA buffer with 1% SDS) in order to ensure the disruption of interactions with potential copurifying proteins (Fig. 4B and C). Similar to our previous results, both CARMA1-ΔPRD (Fig. 4B) and WT-CARMA1 (Fig. 4C; see also Fig. 6F) were consistently polyUb after 1% SDS solubilization. These combined data indicate that CARMA1, rather than an associated protein, is a direct target of polyubiquitination. Finally, we also demonstrated increased polyUb attachment to WT-CARMA1, within 5 min of P/I-induced stimulation in 293T cells (Fig. 4D), implying that this polyUb is a consequence of CARMA1 activation.

FIG. 4.

Constitutively active CARMA1-ΔPRD and activated CARMA1-WT are conjugated with K48-linked ubiquitin in 293T cells and B lymphocytes. (A) 293T cells were cotransfected with combinations of myc-CARMA1-ΔPRD and HA-Ub for 48 h. Cells were lysed with RIPA buffer (0.1% SDS, 0.5% Na-deoxycholate, 1% Triton X-100, and 250 mM NaCl) and myc-CARMA1-ΔPRD was immunoprecipitated with anti-myc antibodies and protein G beads (myc-IP). Ubiquitination was detected by anti-HA (Ub; brackets), and loading was detected by anti-myc WB. (B) 293T cells were transfected and processed as for panel A, with the exception that cells were lysed with RIPA containing 1% SDS and were diluted 10-fold to allow myc-CARMA1 immunoprecipitation (myc-IP) as described in Materials and Methods. (C) 293T cells were cotransfected for 48 h with combinations of myc-CARMA1 and HA-Ub and lysed with RIPA (0.1% SDS) or RIPA containing 1% SDS. myc-IP and WB were carried out as described for panel A. (D) 293T cells were cotransfected with myc-CARMA1 and HA-Ub. Cell were harvested and stimulated with P/I for the indicated times. Cells were processed and analyzed as for panel A. (E and F) Human Ramos B cells (25 × 10⁶/condition) retrovirally transduced to stably express myc-CARMA1 (E) and CARMA1^−/− DT40 B cells (4 × 10⁷/condition) with or without (-) Flag-CARMA1 expressed from the β-actin locus (F) were stimulated with anti-IgM for the indicated times, and then cells were lysed with RIPA. myc-IP or anti-Flag IP (flag-IP) was carried out, and CARMA1 ubiquitination in myc- and flag-IPs was detected using an anti-ubiquitin antibody (FK2). WB assays with anti-myc and anti-Flag antibodies were used as loading controls. (G) 293T cells were transfected with myc-CARMA1-ΔPRD and specific HA-Ub mutants (R48, Lys-48 mutated to Arg; K48, all Lys mutated to Arg except Lys-48; R63, Lys-63 mutated to Arg) in order to determine the type of ubiquitin chains formed. Cells were processed as for panel A. (H) CARMA1^−/− DT40 cells retrovirally transduced to express myc-CARMA1 were stimulated with P/I for the indicated times, and endogenous K48-ubiquitination was detected using a specific anti-K48-linked ubiquitin antibody in myc-IPs. In all experiments the cells were preincubated for 1 h with 25 μM ALLN. (D, E, and H) Cell activation was analyzed by anti-pERK/ERK WB in WCL.

FIG. 6.

The Hook domain in the MAGUK region suppresses CARMA1 ubiquitination at the GUK and SH3 domains. (A, upper panel) Diagram of the structural domains included in the 6×myc-tagged CARMA1 fragments (labeled A to G). 293T cells were cotransfected with HA-Ub and candidate 6×myc fragments. (B) Amino acid sequence alignment of the MAGUK regions from murine CARMA1, PSD-95, and Dlgh1. Amino acid identities are highlighted in black, and similarities are outlined in boxes. Bars beneath the sequences show the locations of the various domains. Blue, tripartite SH3 domain (the complete SH3 domain is formed spatially through protein folding rather than as a linear sequence); red, Hook segment; black, GUK domain. (C) 293T cells were cotransfected with HA-Ub and candidate 6×myc-tagged CARMA1 fragments. (D) 293T cells were cotransfected with HA-Ub and either myc-CARMA1-WT, -ΔPRD, or -ΔHook. (E). 293T cells were cotransfected with HA-Ub and either WT or ΔHook forms of myc-Dlgh1. (F) 293T cells were cotransfected with HA-Ub and either WT or ΔHook forms of myc-CARMA1 or myc-PSD-95. Cells were lysed with RIPA buffer (A to E) or with RIPA containing 1% SDS and then lysates were diluted 10-fold (F). myc-IPs were performed and ubiquitination and myc-tagged proteins were detected using anti-HA and anti-myc antibodies, respectively.

FIG. 5.

PKCβ controls CARMA1 ubiquitination and degradation in B cells. (A) CARMA1^−/− DT40 cells retrovirally transduced to express myc-CARMA1 were preincubated with vehicle (DMSO) or 5 μM Ro-318425 (PKC inhibitor) for 30 min before being stimulated with P/I for the indicated times in the presence of 25 μM ALLN. Ubiquitination was observed in myc-IPs using anti-Ub, and loading was determined by using anti-myc antibody. The anti-myc WB was overexposed to show the anti-myc laddering. (B) CARMA1^−/−, PKCβ^−/−, or WT DT40 cells were left stimulated for the indicated times with P/I in the presence of CHX (right panel). (C) WT and PKCβ^−/− splenic B cells were left unstimulated or stimulated with P/I for the indicated times in the presence of CHX. Cells were lysed with RIPA, and WCLs were tested for CARMA1, ERK, and PKCβ expression. For all panels, JNK and ERK phosphorylation levels were evaluated as specific activation markers. Asterisk, cross-reactivity of pJNK antibody with the pERK signal (A). Bar graphs represent CARMA1 expression relative to the loading control, ERK; results at the zero time points were set as 1.

We next carried out experiments to evaluate whether endogenous ubiquitination of activated CARMA1 occurred upon AR or P/I stimulation in B lymphocytes (Fig. 4E and F). Ramos B cells stably expressing myc-tagged WT-CARMA1 were stimulated with anti-IgM for specific time periods, and polyUb was detected in myc-IPs by using a monoclonal anti-Ub antibody (FK2) (Fig. 4E). As observed with the HA-tagged Ub, a clear anti-Ub smear above CARMA1 migration was detected that peaked in intensity between 5 and 10 min after AR stimulation. Likewise, CARMA1^−/− DT40 cells with a Flag-CARMA1 expression cassette knocked into the chicken β-actin locus also showed an anti-Ub smear above the size of the tagged CARMA1 following stimulation via the B-cell receptor (BCR) and selective immunoprecipitation using anti-Flag antibodies (Fig. 4F).

There are seven potential lysine residues associated with the Ub moiety, and the lysine residue linkage used to form the polyUb chain has distinct functional consequences. Proteins modified by K48-linked polyUb chains are targeted to the proteasome for degradation, whereas K63-linked polyUb chains function in signal transduction pathways (11). To characterize which type of polyUb chain modifies activated CARMA1, we tested myc-CARMA1-ΔPRD ubiquitination in 293T cells by using three specific HA-Ub mutants: R48 (K48 mutated to Arg), R63 (K63 mutated to Arg), and K48 (all Lys residues except K48 mutated to Arg). While ubiquitination was unaltered using R63 or K48 mutants, it was abolished when the R48 mutant was used, demonstrating that CARMA1-ΔPRD is mainly modified by K48 polyUb chains (Fig. 4G). CARMA1 K48-specific ubiquitination in myc-IPs was also assessed in myc-CARMA1-reconstituted CARMA1^−/− DT40 B cells by using specific anti-K48 ubiquitin antibodies (Fig. 4H). In agreement with our data using specific HA-Ub mutants, we detected an endogenous K48 polyUb signal that peaked at ∼5 min poststimulation with P/I, with a drop in signal at subsequent time points. Of note, the low level of endogenous CARMA1 ubiquitination observed in unstimulated CARMA1-reconstituted cells is a finding consistent with previous observations, demonstrating constitutive NF-κB activation in this and other transformed B-cell lines (1).

PKCβ controls CARMA1 ubiquitination and degradation in DT40 B cells.

It has been previously demonstrated that the direct effectors of CARMA1 activation in lymphocytes are PKCβ and PKCθ isoforms (13, 29, 30). To test if CARMA1 polyUb is sensitive to PKC activity, ubiquitination of CARMA1 was assessed in activated cells pretreated with the pan-PKC inhibitor Ro-318425 (Fig. 5A). CARMA1^−/− DT40 B cells reconstituted with myc-CARMA1 were activated with P/I and probed for endogenous polyUb in myc-IPs. Similar to our results in Fig. 4, we observed a clear polyUb signal that peaked at ∼5 min poststimulation and receded at subsequent time points (Fig. 5A, vehicle-treated cells). P/I-inducible CARMA1 ubiquitination was greatly diminished in PKC inhibitor-treated cells. The loss of CARMA1 ubiquitination specifically correlated with CARMA1 activation, as JNK but not ERK phosphorylation was decreased by PKC inhibition (Fig. 5A, lower panel). To determine whether CARMA1 degradation in B cells specifically requires upstream PKCβ signals, PKCβ^−/− DT40 B cells, as well as primary splenic B cells from WT or PKCβ knockout mice, were analyzed for endogenous CARMA1 degradation after P/I stimulation (Fig. 5B and C). In WT cells, CARMA1 rapidly degraded when stimulated with P/I; however, CARMA1 levels in PKCβ^−/− cells remained stable, even after P/I treatment. Cell lysates were also evaluated for PKCβ expression and defective pJNK but normal pERK signals. Taken together, these observations link the upstream requirement for CARMA1 phosphorylation with its subsequent ubiquitination and indicate that CARMA1 is a direct target of endogenous K48-linked polyUb upon lymphocyte stimulation.

The SH3-GUK Hook controls ubiquitination of CARMA1 and other MAGUK proteins.

To determine the specific domains of CARMA1 that are targets of ubiquitination, constructs expressing 6×myc-tagged fragments of CARMA1 were generated and individually transfected into HEK 293T cells in association with HA-Ub (Fig. 6A). myc-IPs were performed, followed by immunoblotting for HA and myc. This analysis revealed strong polyUb of the GUK domain (fragment F). In addition, fragments containing the SH3 domain were also ubiquitinated, but with reduced intensity (fragments C and G). Of note, the fragments containing the CC (fragments B and E) had a discrete higher-molecular-weight band when probed with the HA antibody, perhaps revealing a monoubiquitination event. This possibility is currently under investigation and will not be discussed further in this report.

Interestingly, fragment A, comprised of the full MAGUK region along with the PRD, showed only weak ubiquitination (Fig. 6A). This finding was surprising, as this fragment contained both the SH3 and GUK domains, which were strongly ubiquitinated when expressed individually. Others have shown by crystallography that another MAGUK protein, PSD-95, is folded in a closed/inactive conformation mediated by intramolecular interactions between the SH3 domain and regions proximal to the GUK domain. Elimination of a flexible region located between the SH3 and GUK domains, termed the Hook domain, removed the structural constraint of this auto-inhibitory, intramolecular interaction, allowing PSD-95 to open and oligomerize (a property of activated MAGUKs) (15, 32, 33). Thus, we hypothesized that the weaker ubiquitination observed using the full MAGUK domain of CARMA1 (Fig. 6A), in comparison to the GUK region alone, might reflect a similar intramolecular closed configuration of the CARMA1 MAGUK. Through alignment of the CARMA1 MAGUK region with that of PSD-95 and Dlgh1, we identified a putative Hook region in CARMA1 (Fig. 6B). We first compared ubiquitination between constructs expressing fragment A with the same fragment lacking the Hook domain. As shown in Fig. 6C, when the Hook domain from fragment A was deleted, the levels of polyUb greatly increased. Fragment E was used as a negative control in this experiment. We next analyzed polyUb using full-length myc-CARMA1 with and without deletion of the Hook region (Fig. 6D). Deletion of the Hook was sufficient to drive ubiquitination of CARMA1 to levels similar to that of the constitutively active CARMA1-ΔPRD molecule.

Our data showing that deletion of the Hook triggers CARMA1 polyUb led us to hypothesize that other active, open structured MAGUK proteins might also be targets for polyUb ligation. We therefore tested whether deletion of the Hook region similarly influenced the polyUb status of two other candidate MAGUK proteins: Dlgh1, with known expression in lymphocytes, and PSD-95, in which the Hook region was previously characterized (Fig. 6E and F). 293T cells were cotransfected with HA-Ub and either WT or ΔHook versions of CARMA1, Dlgh1, or PSD-95, solubilized with either 0.1% or 1% SDS, and then polyUb was analyzed after myc-IP. High-level, constitutive polyUb was observed in Dlgh1 proteins lacking the Hook domain, while WT forms showed only weak ubiquitination (Fig. 6E). Similarly, ΔHook versions of both CARMA1 and PSD-95 showed higher polyUb smears than their WT counterparts using 1% SDS for cell lysis, a finding consistent with direct polyUb (Fig. 6F). Overall, these results strongly suggest that degradation of CARMA1 as well as other MAGUK family proteins may occur in response to a similar conformational change that exposes the Ub acceptor sites within the MAGUK region.

Lysines within the MAGUK region of CARMA1 are main targets of K48 polyUb.

To further test whether the MAGUK region is the primary polyUb target within the full-length CARMA1 protein, we generated a mutant protein in which all lysines within the MAGUK region (29 lysines) were replaced with arginine (CARMA1-KR). myc-tagged WT-CARMA1 or CARMA1-KR was cotransfected with HA-Ub in 293T cells. Following cell stimulation, CARMA1 ubiquitination was detected by HA and K48 polyUb immunoblotting (Fig. 7A). While WT-CARMA1 ubiquitination was detected using either anti-HA or anti-K48 antibodies, peaking at 5 to 15 min poststimulation, only traces of ubiquitination were observed using the CARMA1-KR mutant, despite equivalent cell activation levels (as assessed using inducible ERK phosphorylation).

FIG. 7.

Elimination of ubiquitination of the MAGUK region reduces CARMA1 degradation and results in increased JNK and NF-κB activation. (A) 293T cells were cotransfected with HA-Ub and either myc-CARMA1-WT or a myc-CARMA1 mutant in which the lysine residues in the MAGUK region were replaced by arginines (myc-CARMA1-KR). Cells were harvested and stimulated with P/I for the indicated times. CARMA1 ubiquitination was detected as described previously using anti-HA and anti-K48 specific antibodies. Activation was detected using anti-pERK WB and WCL. (B) Retrovirally transduced CARMA1^−/− DT40 cells were stimulated with P/I for the indicated times, and myc-CARMA1-WT (upper panels) or myc-CARMA1-KR (lower panels) degradation was analyzed by anti-myc WB in WCL. Cell activation and loading were assessed by anti-pERK/ERK WB. (C) WCLs from P/I-stimulated 293T cells transfected with WT- or KR-CARMA1 were generated and analyzed for JNK phosphorylation using polyclonal pJNK-specific antibodies. The bar graph represents the levels of pJNK relative to myc-CARMA1 expression. Zero time points from myc-CARMA1-transfected cells were set as 1. Data show averages ± standard errors of the means from three independent experiments. (D) Jurkat cells were cotransfected with reporter vector Igκ2-IFN-luciferase (NF-κB-Luc reporter), pRL-TK (transfection control), and DNA concentrations for myc-CARMA1-WT, myc-CARMA1-KR, or empty vector that resulted in similar expression of CARMA1 molecules. Cells were stimulated without or with CD3- and CD28-specific antibodies for 6 h. NF-κB-dependent luciferase activity normalized to the transfection control is shown (means ± standard errors of the means of three independent experiments). **, P ≤ 0.01; *, P ≤ 0.05 according to Student's t test. (Inset) myc and ERK WB for analysis of expression of myc-CARMA1-WT and myc-CARMA1-KR in 293T cells.

As would be expected from a mutant that loses K48 polyUb targets, the CARMA1-KR protein exhibited a reduced turnover rate when stably expressed in P/I-stimulated CARMA1^−/− DT40 cells (Fig. 7B). CARMA1-WT degradation was clearly observed in stimulated cells preferentially at 180 min poststimulation; in contrast, there was limited loss of the CARMA1-KR mutant protein during this time, supporting the idea that K48 polyUb of CARMA1 drives its degradation.

Importantly, the CARMA1-KR mutant also exhibited enhanced basal and inducible downstream signaling to activate NF-κB and JNK (Fig. 7C and D). Transfection of WT and KR forms of CARMA1 and P/I-induced activation of 293T cells revealed that CARMA1-KR induced high basal and inducible pJNK levels (Fig. 7C). Similarly, cotransfection of myc-CARMA1-KR (versus myc-CARMA1-WT) and an NF-κB-luciferase reporter into Jurkat T cells revealed a 2- to 4-fold increase in basal, and also an AR-inducible, NF-κB activity (Fig. 7D). These data support the idea that the reduced K48 polyUb signal in the CARMA1-KR mutant (Fig. 7A, right) reflects loss of bona fide in vivo ubiquitin target sites rather than a nonspecific effect due to altered downstream signaling activity. Taken together, these data support the conclusion that CARMA1 is targeted by K48-specific ubiquitination within the MAGUK region following cell activation and that these events coordinately downregulate CARMA1-dependent JNK and NF-κB activation.

cIAP2 can target CARMA1 for ubiquitination but is not essential for CARMA1 degradation in vivo.

Several E3 enzymes, including cIAP2, Itch, NEDD4, and βTrCP, have been implicated in BCL10 ubiquitination and degradation in lymphocytes (7, 12, 27, 37). Because CARMA1 interacts with BCL10 upon AR stimulation, we postulated that one or more of these E3s may also be involved in CARMA1 ubiquitination. Cotransfection experiments using HA-Ub, myc-CARMA1-ΔPRD, and Flag-tagged β-TrCP, cIAP2, cIAP2 inactive mutant (35), Itch, NEDD4, or Cbl-b revealed that only cIAP2 and NEDD4 increased myc-CARMA1-ΔPRD polyUb (Fig. 8A). As anticipated, the catalytic activity of cIAP2 was required for CARMA1 polyUb (Fig. 8B). Functionally, only overexpression of cIAP2 modestly but significantly reduced the CARMA1-dependent NF-κB activation in 293T cells (∼30%) (Fig. 8C). In transfected Jurkat cells, overexpression of cIAP2, but not its inactive mutant, reduced the CD3/CD28-driven, CARMA1-dependent NF-κB activation by ∼50% (Fig. 8D). These results suggested that in vivo cIAP2 might modulate CARMA1-dependent NF-κB activation by inducing CARMA1 degradation. However, when we tested the effect of a cIAP antagonist that depletes the endogenous pool of cIAP1/2, P/I-induced CARMA1 degradation in primary lymphocytes was not affected (Fig. 8E).

FIG. 8.

E3s target CARMA1 and downregulate CARMA1-dependent NF-κB activation but not its degradation. (A) 293T cells were cotransfected with myc-CARMA1-ΔPRD, HA-Ub, and candidate flag-tagged (f) E3s. HA (Ub) and myc (CARMA1) WB were assessed in myc-IPs. Expression of E3s was assessed in an anti-Flag WB assay in WCL. (B) 293T cells were cotransfected with myc-CARMA1-ΔPRD, f-cIAP2-WT, or the f-cIAP2-RING mutant. Ub and CARMA1 were detected in myc-IPs by anti-HA and anti-myc WB. WCL were analyzed to assess flag (cIAP2) expression level (bottom panel). (C) 293T cells were cotransfected with myc-CARMA1-WT and 2×Flag (f)-cIAP2, f-NEDD4, or f-Cbl-b and NF-κB and transfection control reporters. NF-κB-dependent luciferase activity was normalized to the transfection control. Data are means ± standard errors of the means of five independent experiments. For statistical analysis, Student's t test was used to compare cells expressing only CARMA1 versus those expressing CARMA1 plus specific E3s. **, P ≤ 0.05. Expression of CARMA1 and E3s was confirmed by WB of WCL using tag-specific antibodies. (D) Jurkat T cells were cotransfected for 48 h with NF-κB and control reporter genes and with various doses of myc-CARMA1-WT (266, 133, 66, and 33 ng) and f-cIAP2-wt, f-cIAP2-RING mutant, or empty vector. Unstimulated or anti-CD3ɛ/CD28-stimulated cells were assessed for NF-κB reporter activity. (E) Total splenocytes were pretreated with cIAP antagonist (BV6) for 1 h in order to promote cIAP1/2 turnover, and then stimulated with P/I for the indicated times in the presence of CHX and BV6. Endogenous CARMA1, cIAP1/2, and actin expression levels were analyzed by WB. CARMA1 fold differences were obtained from the CARMA1 versus actin ratios and normalized by setting the zero time points as 1.

DISCUSSION

Although it is known that CARMA1 activation promotes its interaction with downstream effector molecules that activate the NF-κB and JNK pathways (22), we show here for the first time that a secondary consequence of this activation is the targeting of CARMA1 for polyUb and proteasome-mediated degradation. Our data clearly demonstrate degradation of overexpressed WT (or constitutively active) CARMA1 and of endogenous CARMA1, in cell lines and primary B and T cells, respectively, in response to antigen receptor activation signals (Fig. 1 and 2). Upon cell activation, CARMA1 exhibited high-molecular-weight smearing on Western blots probed for either endogenous or HA-tagged ubiquitin. That these ubiquitin signals specifically detected polyUb-CARMA1, and not a coassociated protein, was confirmed by detection of similar high-molecular-weight smearing via Western blotting using an antibody that recognized epitope-tagged CARMA1 and by solubilizing cells using high-stringency buffers containing 1% SDS (Fig. 4 to 6). Using tagged ubiquitin mutants and K48-ubiquitin-specific antibodies, we found that both WT-CARMA1 and the constitutively active CARMA1-ΔPRD were modified by K48-linked polyUb (Fig. 4, 5, and 7). Together with the observation that chemical inhibitors of the proteasome abolished degradation of activated CARMA1 (Fig. 3), our findings provide strong evidence that activated CARMA1 is targeted for ubiquitination and subsequent degradation by the proteasome. We observed that the time course of CARMA1 polyUb (peaking 5 to 15 min poststimulation) is more rapid than the detectable decrease in CARMA1 protein due to degradation (∼1 h poststimulation). Note that after stimulation, only a small fraction of total cellular CARMA1 is recruited to lipid rafts, and this fraction most likely represents the activated protein (30). Therefore, polyUb presumably targets only a small pool of CARMA1 at a given time. We hypothesize that detection of a loss of CARMA1 is less sensitive and thus requires a longer period to integrate sufficient CARMA1 degradation for detection.

The finding that CARMA1 is selectively degraded after activation identifies a previously unappreciated pathway for regulating CARMA1, and thus lymphocyte activation. As an initial characterization of this degradation pathway, we tested two alternate, not mutually exclusive, hypotheses to explain how activated CARMA1 is marked for degradation by the ubiquitin-proteasome pathway: (i) CARMA1 is posttranslationally modified by an upstream or downstream effector of the AR pathway; or (ii) activated CARMA1 assumes a structural configuration that allows downstream E3 ligases access to its polyUb target(s). We found that activation of the key upstream BCR effector, PKCβ, is essential for triggering ubiquitination and degradation of full-length CARMA1 (Fig. 5). However, because deletion of the PRD was also sufficient to drive this ubiquitination/degradation cascade, the role for upstream AR-dependent signals in triggering K48 polyUb appears to be primarily to activate CARMA1, and the ability to direct K48 polyUb is likely intrinsic to the activated molecule. Whether this degradation cascade is driven by a signal event downstream of activated CARMA1 is yet unknown. However, we have determined that genetic deletion of two kinases activated downstream of CARMA1, TAK1 and IKKβ, has no effect on degradation of CARMA1-ΔPRD in DT40 B cells (data not shown).

The predominant domains within CARMA1 targeted for polyUb were the GUK and the SH3 domains. Protein fragments that contained these domains underwent significant polyUb only when not expressed within the context of the full-length MAGUK. Previous structural studies have shown that the SH3 and GUK domains of PSD-95 interact to create a closed conformation that limits ligand-induced oligomerization (15, 32). Elimination of the Hook located between the SH3 and GUK disrupts this intramolecular interaction, releasing SH3 and GUK for oligomerization (15). We identified a domain similar to the PSD-95 Hook within CARMA1 and Dlgh1. Deletion of this region strongly promoted MAGUK polyUb, likely due to conformational changes that facilitated E3 accessibility to the SH3 and GUK. We predict that this mechanism will prove to be of importance to other MAGUK family members, as we found that ubiquitination was induced upon deletion of the Hook domain from two other MAGUK family members, PSD-95 and Dlgh1 (Fig. 6).

Interestingly, our data showed that CARMA1-ΔPRD is targeted for ubiquitination despite having an intact Hook region. Additional studies will be required to determine if the activated conformation of CARMA1-ΔPRD leads to a destabilization of the SH3-GUK interactions, or, alternatively, whether another signal downstream of activated CARMA1 is required to initiate these events. Our data adds to previous work that highlights the importance of the SH3-GUK linker to CARMA1 function. While early characterizations of CARMA1 showed that both the SH3 and GUK were required for CARMA1 activity (20), later, deletion of the amino acid sequence linking the SH3 and GUK (113 amino acids that included the 64-amino-acid Hook) revealed that this region was essential for CARMA1 activation, too (14). Even more striking, a single point mutation within the Hook domain of chicken CARMA1 (S909A; a putative consensus target site for PKCs) also abolished CARMA1 activation (28). A full comprehension of CARMA1 activation will thus require further investigation to determine how the Hook domain regulates both CARMA1 activation and subsequent stability.

CARMA1 ubiquitination was specifically promoted in vitro by at least two (cIAP2 and NEDD4) of five E3 enzymes reported to have a role down-modulating AR signals. cIAP2 and NEDD4 belong to distinct families of E3 ligases, and have no significant structural similarity, leading us to suspect that the overexpression assays may not be relevant to CARMA1 polyUb in vivo. Consistent with this, we observed only partial inhibition of CARMA1-dependent NF-κB activation following overexpression of cIAP2 in 293T cells or lymphocytes (Fig. 8C). Further, addition of a cIAP antagonist failed to modulate CARMA1 degradation in splenocyes (Fig. 8). Overall, these findings suggest that other yet-unidentified E3s may play a more critical role in these events, or that multiple E3s may function redundantly to control CARMA1 turnover. Of note, it has recently been reported that CARMA1 is monoubiquitinated when coexpressed with Cbl-b in NK T cells (9). From our evaluation of a panel of E3s, we observed that Cbl-b had no effect on the ubiquitination status of CARMA1-ΔPRD. Although the reason for our discordant findings is unclear, further research will be required to determine the physiological role for candidate E3s in CARMA1 ubiquitination and turnover.

Based upon the data presented in this work, we propose a model for the events that downregulate activated CARMA1 (Fig. 9). Prior to cell stimulation, CARMA1 likely maintains a closed conformation mediated by intramolecular interactions limiting the accessibility of both the CARD (preventing constitutive NF-κB/JNK activation) and the SH3 and GUK domains (preventing CARMA1 oligomerization, polyUb, and degradation). Upon AR engagement, PKCβ/θ phosphorylate specific serine residues within the CARMA1 PRD, releasing CARD inhibition and allowing recruitment and activation of downstream effectors required for NF-κB/JNK activation (13, 22, 28, 29). Activation of CARMA1 also disrupts SH3-GUK intramolecular interactions, allowing CARMA1 oligomerization through the GUK (33). It remains unknown whether interaction with a specific ligand or a posttranslational modification (e.g., phosphorylation) triggers these predicted conformation changes. Regardless of the initiating events, our data imply that the MAGUK region within activated CARMA1 efficiently promotes recruitment of E3 enzymes that drive ubiquitination and proteasomal degradation of CARMA1. Additionally, previous studies (34) and our own unpublished data show that activated CARMA1 also promotes BCL10 degradation. These coordinated events are likely essential for appropriately balancing lymphocyte activation and immune responses following antigenic stimulation.

FIG. 9.

Working model of coordinate CARMA1 degradation. CARMA1 exists predominantly in a double closed conformation. Upon cell activation, PKCβ/θ phosphorylate the PRD, promoting conformational changes and protein interactions that facilitate NF-κB and JNK activation. In parallel, conformational changes in the MAGUK region promote recruitment of one or various E3s to CARMA1, leading to K48-specific ubiquitination of the exposed GUK and SH3 domains followed by proteasome-dependent degradation of CARMA1. These combined events function coordinately to attenuate CARMA1-dependent NF-κB activation. The key E3s required for this process and the mechanisms that control Hook-dependent conformational changes are still unknown.