Oncogenic CARD11 Mutations in Human Diffuse Large B Cell Lymphoma

See allHide authors and affiliations

Science  21 Mar 2008:
Vol. 319, Issue 5870, pp. 1676-1679
DOI: 10.1126/science.1153629


Diffuse large B cell lymphoma (DLBCL) is the most common form of non-Hodgkin's lymphoma. In the least curable (ABC) subtype of DLBCL, survival of the malignant cells is dependent on constitutive activation of the nuclear factor–κB (NF-κB) signaling pathway. In normal B cells, antigen receptor–induced NF-κB activation requires CARD11, a cytoplasmic scaffolding protein. To determine whether CARD11 contributes to tumorigenesis, we sequenced the CARD11 gene in human DLBCL tumors. We detected missense mutations in 7 of 73 ABC DLBCL biopsies (9.6%), all within exons encoding the coiled-coil domain. Experimental introduction of CARD11 coiled-coil domain mutants into lymphoma cell lines resulted in constitutive NF-κB activation and enhanced NF-κB activity upon antigen receptor stimulation. These results demonstrate that CARD11 is a bona fide oncogenein DLBCL, providing a genetic rationale for the development of pharmacological inhibitors of the CARD11 pathway for DLBCL therapy.

Diffuse large DLBCL is the most common type of non-Hodgkin's lymphoma, accounting for 30 to 40% of cases (1). DLBCL consists of three subtypes: germinal center B cell–like (GCB) DLBCL, activated B cell–like (ABC) DLBCL, and primary mediastinal B cell lymphoma (25). Although more than half of DLBCLs are curable (6), the ABC DLBCL subtype, which accounts for roughly one-third of the cases, has an inferior prognosis (3).

A characteristic feature of ABC DLBCLs is constitutive activation of the NF-κB pathway, which plays a pivotal role in proliferation, differentiation, and survival of normal lymphoid cells (7). It was previously shown that survival of ABC DLBCLs depends on NF-κB function; inhibition of the NF-κB pathway was toxic to ABC DLBCL but not to GCB DLBCL cell lines (7). In an RNA interference screen, small hairpin RNAs (shRNAs) targeting CARD11 (also known as CARMA1 or Bimp3) were toxic to ABC but not GCB DLBCL cell lines (8). During antigen stimulation of normal lymphocytes, CARD11 functions as a signaling scaffold protein to coordinate the activation of IκB kinase β (IKK), a positive regulator of the NF-κB pathway (9).

Although these studies showed that CARD11 is required for the constitutive activation of NF-κB in ABC DLBCL, they did not address the mechanism of CARD11 activation in these lymphomas. Normal B cells require “tonic” B cell receptor (BCR) signaling to the NF-κB pathway for survival (10, 11), and CARD11 is required for the differentiation and/or survival of discrete B cell subpopulations (1215). Thus, ABC DLBCLs might “inherit” their dependence on CARD11 from their normal cellular counterparts. Alternatively, addiction of ABC DLBCLs to CARD11 signaling might stem from oncogenic mutations in this pathway. To address this issue, we initially resequenced all coding exons of CARD11 in 16 ABC DLBCL biopsies and four ABC DLBCL cell lines; we discovered missense base substitutions in three biopsies and one cell line. All substitutions affected the coiled-coil domain, which mediates CARD11 oligomerization and NF-κB pathway activation (14, 16). We next resequenced the coiled-coil domain exons in 136 additional DLBCL biopsies. In all, we detected missense substitutions affecting amino acids in or immediately adjacent to the coiled-coil domain in 9.6% (7/73) of ABC DLBCL biopsies but in only 3.8% (3/79) of GCB DLBCL biopsies (Fig. 1A). As a control, we analyzed 19 mucosa-associated lymphoid tissue (MALT) lymphoma biopsies because MALT lymphomas can also have constitutive NF-κB activation; no coiled-coil domain substitutions were detected.

Fig. 1.

CARD11 mutations in DLBCL. (A) General domain structure of CARD11 and a schematic of the coiled-coil domains, showing the amino acids mutationally altered in DBLCL. Mutations from ABC DLBCLs (blue) and GCB DLBCLs (orange) are shown. Mutants 7 and 9 both cause two amino acid substitutions. Abbreviations for amino acid residues: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; Y, Tyr. (B) NF-κB target gene signature expression in ABC DLBCL and GCB DLBCL cases with either wild-type (WT) or mutant CARD11. Means and SEs are shown.

To test whether these base substitutions represent acquired tumor-specific mutations or rare CARD11 polymorphisms, we sequenced the germline CARD11 locus in the four cases for which constitutional DNA was available (cases 4, 5, 10, and 11). Each of these normal samples was wild type for the CARD11 coiled-coil region, and thus the base substitutions in the tumors were somatic mutations. In a fifth case, some alleles cloned from the tumor DNA had two separate base substitutions (9a and 9b) but other alleles had only one of these (9a). We conclude that base substitution 9b occurred as a somatic mutation that was acquired by the tumor after 9a. Together, these data make it likely that all of the coiled-coil domain substitutions represent somatic mutations—a view supported by their functional activity, as detailed below.

A gene expression signature of NF-κB pathway activity was highly expressed in all ABC DLBCL biopsies, irrespective of CARD11 mutational status, as expected (7) (Fig. 1B). Notably, this signature was higher in GCB DLBCLs with mutant CARD11 than in those with wild-type CARD11, indicating activation of the NF-κB pathway by these mutations (P = 0.046).

To explore the functional consequences of the CARD11 mutations, we expressed mutant or wild-type CARD11 proteins in a variant of the Jurkat human T cell line that lacks endogenous CARD11 expression and harbors an NF-κB–driven enhanced green fluorescent protein (EGFP) reporter (17). Six of 11 CARD11 mutants (numbers 2, 3, 5, 7, 9a, 9ab, and 10) produced strong NF-κB activity without any exogenous stimulation, which was not observed with wild-type CARD11 even though wild-type and mutant CARD11 proteins accumulated to comparable levels (Fig. 2A and fig. S1, A and B). The five remaining mutants (numbers 4, 6, 8, 9b, and 11) produced a more modest degree of constitutive NF-κB activation, which was nonetheless greater than that of wild-type CARD11. In response to T cell receptor activation, cells expressing wild-type CARD11 showed a moderate increase in NF-κB activity, but the induction was more pronounced in cells bearing the CARD11 mutants (Fig. 2A). In the BJAB human B cell line, these mutants induced expression of NF-κB target genes, including CD83, to a much greater degree than wild-type CARD11, both with or without BCR stimulation (fig. S1, C to E). We expressed the CARD11 isoforms as fusion proteins with EGFP and monitored CD83 expression as a function of CARD11 expression by flow cytometry. The CD83 response to increasing expression of the CARD11 mutants was much steeper than observed with wild-type CARD11 (Fig. 2B and fig. S1F). Thus, at comparable protein levels, the CARD11 mutants were consistently more effective than wild-type CARD11 in activating NF-κB.

Fig. 2.

CARD11 coiled-coil mutants constitutively activate the NF-κB pathway and enhance antigen receptor signaling to NF-κB. (A) Expression of wild-type CARD11 or CARD11 mutants in the Jurkat human T cell line JPM50.6, which lacks endogenous CARD11 and harbors an NF-κB–driven expression EGFP reporter. Shown are EGFP levels in unstimulated cells (left) or in cells stimulated by cross-linking with antibodies to CD3 and CD28 (right). (B) Linear regression analysis of EGFP-CARD11 levels (in decile bins) and CD83 expression in the GCB DLBCL cell line BJAB expressing either exogenous wild-type or mutant EGFP-CARD11 isoforms. (C) Cell-based assay of IKK activation by exogenous CARD11 isoforms. The GCB DLBCL cell line OCI-Ly19 was engineered to express an IκBα–Photinus luciferase fusion protein, a control Renilla luciferase, and doxycycline-inducible constructs for CARD11 isoforms. Values represent IκBα– Photinus luciferase/Renilla luciferase ratios in doxycycline induced cells, normalized to ratios in uninduced cells. A decrease in the ratio indicates increased degradation of the IκBα–Photinus luciferase reporter, attributable to increased IKK activity. (D) Aggregate formation by CARD11 coiled-coil domain mutants. Composite fluorescence photomicroscopy of BJAB cells expressing EGFP-tagged wild-type CARD11 or mutant 3, with 4′,6′-diamidino-2-phenylindole nuclear counterstain (blue); scale bar, 10 μm. (E) Correlation between CARD11 aggregate formation [as measured in (C)] and IKK activation, showing percent decrease in the IκBα-luciferase reporter in OCI-Ly19 cells after 9 hours of doxycycline induction of wild-type or mutant CARD11. A regression line representing the inverse correlation of the variables is shown (Pearson correlation coefficient = –0.694; P = 0.0059). Ctrl., EGFP control.

To determine whether the CARD11 mutants functioned upstream of IKK in the NF-κB pathway, we measured IKK activity with the use of a B cell line expressing a fusion protein between Photinus luciferase and the IKK substrate IκBα, as described (8, 18). Whereas wild-type CARD11 did not alter the IκBα-luciferase level, all CARD11 mutants produced a time-dependent decrease in the reporter, indicating IKK activation (Fig. 2C). Mutants 3, 9a, 9ab, and 10 produced the strongest IKK activation, in accord with their greater ability to induce NF-κB in other cell lines.

We next examined the subcellular localization patterns of the EGFP-CARD11 isoforms expressed in BJAB cells by fluorescence microscopy. In unstimulated cells, wild-type CARD11 was localized diffusely in the cytoplasm, whereas CARD11 mutants were concentrated in cytosolic aggregates (Fig. 2D). To quantify the aggregation of CARD11 mutants, we used a multispectral imaging flow cytometer that obtains a microscopic image of each flow-sorted cell (fig. S2, A and B) (18, 19). Fewer than 5% of cells expressing wild-type CARD11 had aggregates, whereas 38 to 88% of the cells expressing CARD11 mutants showed aggregate formation (fig. S2C). The degree of aggregation by each mutant correlated with its ability to spontaneously activate the NF-κB pathway, as judged by degradation of the IκBα-luciferase reporter (P = 0.0059; Fig. 2E). The CARD11 mutant aggregates colocalized with total and phosphorylated IKK as well as MALT1, a signaling protein required for IKK activation by CARD11 (20), but in cells with wild-type CARD11, IKK and MALT1 were diffusely cytoplasmic (fig. S2D). We conclude that the CARD11 mutations confer a gain-of-function phenotype, which is characterized by constitutive IKK activity, enhanced NF-κB response to exogenous stimulation, and a propensity to form cytoplasmic structures that colocalize with NF-κB signaling components.

One ABC DLBCL cell line, OCI-Ly3, had a single mutated CARD11 allele (fig. S3), but three others (HBL-1, U2932, and OCI-Ly10) had wild-type CARD11 alleles, all of which nonetheless depend on CARD11 for survival (8). We devised a complementation experiment to test whether OCI-Ly3 cells require mutated CARD11 for survival, or whether wild-type CARD11 would suffice. OCI-Ly3 cells were first engineered to express the coding region for wild-type CARD11 or CARD11 mutant 2 (derived from OCI-Ly3). These cells were then transduced with a shRNA targeting the CARD11 3′-untranslated region to knock down the expression of the endogenous CARD11 gene but spare the exogenously introduced CARD11 coding regions. Exogenous CARD11 mutant 2 protected OCI-Ly3 cells from CARD11 shRNA toxicity, but exogenous wild-type CARD11 did not (Fig. 3A). Other CARD11 mutants that strongly activated NF-κB were able to rescue OCI-Ly3 from CARD11 shRNA toxicity, but mutants with moderate ability to constitutively activate NF-κB only partially rescued the cells. Conceivably, these less active CARD11 mutants may have been selected to enhance the NF-κB response of the malignant clone to stimuli from the microenvironment or to other endogenous stimuli. Unlike the results seen for OCI-Ly3 cells, wild-type CARD11 was able to rescue OCI-Ly10 cells from the toxicity of CARD11 shRNA, as could mutant CARD11 (Fig. 3B). These findings demonstrate that some ABC DLBCLs become addicted to the action of mutant CARD11, whereas others have alternative mechanisms to activate wild-type CARD11 that are not present (or needed) in ABC DLBCLs with mutated CARD11.

Fig. 3.

CARD11 mutation is required for survival of the ABC DLBCL cell line OCI-Ly3. (A) Mutant but not wild-type CARD11 rescues OCI-Ly3 cells from CARD11 shRNA toxicity (see text for details). (B) Both wild-type and mutant CARD11 rescue the ABC DLBCL cell line OCI-Ly10 from CARD11 shRNA toxicity. (C) Expression of an isolated CARD11 coiled-coil domain kills ABC DLBCL cell lines with mutant CARD11 (OCI-Ly3) or wild-type CARD11 (OCI-Ly10), but not a GCB DLBCL cell line (BJAB).

Our results firmly establish CARD11 as a bona fide oncogene in DLBCL. Previous reports described the dependence of ABC DLBCLs on CARD11 signaling to NF-κB (7, 8), but they did not address whether this represents tonic BCR signaling (10, 11) or is a feature of the stage of B cell differentiation from which ABC DLBCLs derive. We have now shown that a subset of ABC DLBCLs acquire activating CARD11 mutations that induce NF-κB constitutively, providing genetic evidence that addiction to the CARD11 pathway is a phenotype that is positively selected during evolution of these lymphomas. CARD11 mutations may allow the lymphoma cells to engage the NF-κB pathway in the absence of antigen receptor signals. In this regard, it is notable that CARD11 coiled-coil domain mutants were constitutively active in OCI-Ly19 and OCI-Ly3 cells, both of which lack detectable PKC-β protein (fig. S4) (21), which phosphorylates and activates CARD11 during normal BCR signaling (22, 23).

The fact that the CARD11 mutations in ABC DLBCL are confined to the coiled-coil domain highlights the central importance of this domain in CARD11 function. That the mutations are activating suggests that the coiled-coil domain keeps CARD11 in a latent state, which can be disrupted physiologically by antigen receptor signaling or pathologically by mutation. Several of the CARD11 mutations that we uncovered introduce helix-breaking prolines, whereas others affect charged amino acids, which can regulate the oligomerization of coiled-coil domains as part of “trigger sites” and can contribute to the stability of coiled-coil interactions (24). These observations suggest that coiled-coil interactions may actually be a feature of inactive CARD11.

How might CARD11 be attacked therapeutically? Expression of the isolated CARD11 coiled-coil domain was toxic to both OCI-Ly3 and OCI-Ly10 cells, but not to the GCB DLBCL cell line BJAB (Fig. 3C). This result shows that DLBCLs with activating mutations in CARD11 are just as vulnerable to interference with coiled-coil function as those with wild-type CARD11. Together, our results provide a compelling genetic and functional rationale for the development of molecular inhibitors aimed at the coiled-coil domain of CARD11, which could have activity in ABC DLBCL, the subtype of DLBCL that is least curable by our current therapies (25).

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article