BCL10 Mutants: Architects of Oncogenic Signaling Provide a Blueprint for Precision Medicine

Dissection of oncogenic signaling promises to usher in a new era of precision medicine in which the constellation of mutations in each tumor reveals the active pathways maintaining survival, laying bare the Achilles’ heel of a patient's malignancy. Armed with the knowledge of essential functional variants for each subtype of cancer, an oncologist need only review the list of mutations in a patient's chart to both correctly diagnose the disease and pluck the properly targeted agent from their quiver to quell the beast. The work of Xia and colleagues reveals that this model may be too simplistic. The authors discovered that two common mutant variants of the same gene, BCL10, influence oncogenic NF-κB signaling and sensitivity to BTK and MALT1 inhibitors in distinct ways in lymphoma (1). This highlights a challenge for precision medicine and demonstrates why deep functional and mechanistic studies are needed to successfully implement targeted therapies into clinical practice.

In this sense, dissection of oncogenic signaling and defining the genomic landscape in diffuse large B-cell lymphoma (DLBCL) were the first important steps toward successful precision medicine approaches. The discovery of their dependence on unique forms of oncogenic signaling initiated by the B-cell receptor (BCR) led to the development of a bevy of targeted agents inhibiting this pathway (2), which are currently in various stages of clinical development. DLBCL, as the most common type of non-Hodgkin lymphomas, is a genetically complex disease with heterogeneous clinical outcomes. Two dominant subtypes, activated B cell–like (ABC) and germinal center B cell–like (GCB) DLBCL, were identified by gene expression profiling and correlated with response to standard immunochemotherapy. The ABC subtype of DLBCL is defined by, and is critically dependent upon, NF-κB signaling initiated downstream of autoreactive BCRs that engage a kinase cascade of Src family kinases, BTK and PRCKB, to induce the assembly of a complex of CARD11, BCL10, and MALT1 (CBM; ref. 2). Upon CBM assembly, BCL10 filaments begin to oligomerize, forming long helical filaments that bind additional molecules of MALT1 (3). BCL10 filaments serve as a multiprotein signaling scaffold to recruit downstream effectors, resulting in NF-κB activation (Fig. 1). Inhibition of chronic active BCR signaling with BTK inhibitors has shown great promise clinically in ABC-DLBCL (4, 5).

Yet not all patients with ABC-DLBCL respond to BTK inhibitors, begging the question of further disease heterogeneity within this subtype. More recent genomic studies aiming to resolve this question have coalesced around four major genetic subsets of ABC (MCD/C5, BN2/C1, N1, and A53/C2) defined by the co-occurrence of separate genetic features (MYD88^L265P/CD79B, BCL6/NOTCH2, NOTCH1, and TP53 with aneuploidy). These findings were supported by functional proteogenomic screens that identified a multiprotein complex, driving NF-κB activity, that consists of mutant MYD88, TLR9, and the BCR (MY-T-BCR). The MY-T-BCR was enriched in MCD cell lines and ibrutinib-responsive patient samples (6). Thus, evidence is mounting that each genetic subset may utilize variable forms of oncogenic BCR signaling, although individual tumors likely achieve the requisite survival threshold of NF-κB through distinct mechanisms. For instance, MCD and BN2 tumors require all three CBM components for survival, yet BCL10 mutations are enriched (∼40% of patients) only in the BN2 subtype, potentially highlighting another such distinction (6, 7).

The Melnick and Wu laboratories focused their deep mechanistic study on BCL10 mutations noticing that the alterations fall into two distinct domains of the BCL10 protein—missense mutations within the N-terminal CARD domain and distal truncation mutations within the serine/threonine-rich domain. Using two recurrent exemplar variants, the authors investigated the effects on NF-κB activity and BCL10 filament formation in vitro. Whereas both types of BCL10 mutations induced NF-κB reporter activity in DLBCL cells, truncating mutations in the C-terminal region of BCL10 were demonstrably stronger. The authors found in a series of elegant experiments that the E140X truncating mutant formed BCL10 filaments faster and at lower concentrations compared with the R58Q CARD domain mutant. Interestingly, they found that truncating mutations achieve this through the deletion of a novel MALT1 interaction motif. Although MALT1 potentiates NF-κB signaling, the novel domain in BCL10 appears to specifically bind monomeric MALT1 that inhibits BCL10 polymerization. In contrast, electron microscopy and thermal melt assays of R58Q BCL10 filaments demonstrated more stable structures, potentially due to increased hydrogen bonding of the glutamine (Q) substitution with the neighboring T59 residue. The authors proposed it was the formation of this glutamine ladder that stabilized R58Q BCL10 filaments.

MALT1 is the catalytically active subunit of the CBM functioning as a paracaspase to cleave itself, BCL10, and negative regulators of NF-κB, including CYLD, TNFAIP3, and Rel-B (8). Paracaspase activity is induced upon MALT1 binding to BCL10, so it was unclear what effect BCL10 truncating mutants would have on MALT1 activity. Again, the E140X truncating BCL10 variant demonstrated increased MALT1 activity over R58Q, which was more potent than BCL10 wild-type (WT). Moreover, truncating mutations of BCL10 reduced but did not relieve the requirement for CARD11 for survival, even though MALT1 activity was maintained upon CARD11 knockdown. This suggested that BCL10 filaments formed with truncating BCL10 mutations could oligomerize in the absence of a strong BCR signal. Consistent with this function, BCL10 E140X mutants were less sensitive to BTK inhibitors.

How then should BCL10-mutant lymphomas be treated? Given the greater MALT1 activity observed (and building upon the authors’ previous work with MALT1 inhibitors; ref. 9), three different MALT1 inhibitors were tested, and each was found to quench MALT1 activity equivalently. Surprisingly, the potent E140X variants demonstrated a greater reduction in NF-κB activity, reduced IC₅₀in vitro, and reduced in vivo tumor growth compared with BCL10 WT alleles when treated with either C3, a potent MALT1 active site inhibitor, or with allosteric inhibitors of MALT1 (JNJ-6790246 and MLT-748). Strikingly though, the same phenotype was not observed in cells bearing the BCL10 R58Q variant. Despite similar levels of MALT1 enzymatic inhibition, MALT1 inhibitors did decrease cell growth and R58Q was able to maintain NF-κB levels above BCL10 WT after inhibitor treatment (Fig. 1). Although the mechanism for this mysterious discrepancy remains elusive, the authors speculate that the more stable filaments formed by BCL10 R58Q may have additional gain-of-function abilities to recruit downstream NF-κB signal transducers such as TRAF6 or the linear ubiquitylation chain assembly complex (LUBAC).

One of the key questions raised by this study is whether BCL10 mutations will guide clinical decision-making in the future. Currently, DLBCL treatment is revolutionized by innovative therapeutics including CAR T cells, antibody–drug conjugates, bispecific antibodies, and targeted agents such as BTK and BCL2 inhibitors. All have shown clinical efficacy and are being tested in various combinatorial regimens. A recent clinical trial (PHOENIX) tested the BTK inhibitor, ibrutinib, combined with R-CHOP (rituximab/cyclophosphamide/doxorubicin/prednisone/vincristine) in patients with newly diagnosed non-GCB-DLBCL. An increase in survival with ibrutinib was observed in younger ABC patients (10). Retrospective genetic analyses demonstrated a remarkable 3-year event-free survival of 100% in ibrutinib-treated MCD and N1 patients. BN2 patients, however, showed less benefit from the addition of ibrutinib to R-CHOP (5). The work from Xia and colleagues suggests this may be due to the high prevalence of BCL10 mutations among the BN2 subtype (∼40% of patients). Yet, the highly BTK inhibitor–responsive genetic subtypes N1 and MCD do have BCL10 amplifications, missense CARD domain mutations, and truncations in BCL10, albeit at lower frequencies than BN2 (7). Thus, it will be of great interest to determine which, if any, BCL10-mutant MCD and N1 patients responded to BTK inhibitors.

The data from Xia and colleagues also suggest that truncating BCL10 mutations increase the sensitivity of DLBCL cells to MALT1 inhibitors. This could be of potential relevance for future therapeutic approaches, especially in light of ongoing clinical trials such as VIPOR (venetoclax, ibrutinib, prednisone, obinutuzumab, revlimid), a multitargeted agent approach, that is showing promising clinical efficacy in heavily pretreated, therapy-resistant, DLBCL. It could well be that MALT1 inhibitors may enrich our current cocktails of targeted agents, especially for those lymphomas carrying truncating BCL10 mutations. The most effective treatment regimen for each genetic subtype of DLBCL remains to be determined. Clearly, DLBCL is not a single beast but rather a multiheaded hydra that will require the deployment of a battery of targeted agents, most likely in combination with other therapeutic modalities, before all patients, including those with BCL10 mutations, benefit.

T. Oellerich reports grants from the German Research Foundation (SFB1530) during the conduct of the study, as well as grants and personal fees from Merck KGaA, grants from GILEAD, and personal fees from Kronos Bio and Roche outside the submitted work. No disclosures were reported by the other author.