Cotargeting of BTK and MALT1 overcomes resistance to BTK inhibitors in mantle cell lymphoma

Bruton’s tyrosine kinase (BTK) is a proven target in mantle cell lymphoma (MCL), an aggressive subtype of non-Hodgkin lymphoma. However, resistance to BTK inhibitors is a major clinical challenge. We here report that MALT1 is one of the top overexpressed genes in ibrutinib-resistant MCL cells, while expression of CARD11, which is upstream of MALT1, is decreased. MALT1 genetic knockout or inhibition produced dramatic defects in MCL cell growth regardless of ibrutinib sensitivity. Conversely, CARD11-knockout cells showed antitumor effects only in ibrutinib-sensitive cells, suggesting that MALT1 overexpression could drive ibrutinib resistance via bypassing BTK/CARD11 signaling. Additionally, BTK knockdown and MALT1 knockout markedly impaired MCL tumor migration and dissemination, and MALT1 pharmacological inhibition decreased MCL cell viability, adhesion, and migration by suppressing NF-κB, PI3K/AKT/mTOR, and integrin signaling. Importantly, cotargeting MALT1 with safimaltib and BTK with pirtobrutinib induced potent anti-MCL activity in ibrutinib-resistant MCL cell lines and patient-derived xenografts. Therefore, we conclude that MALT1 overexpression associates with resistance to BTK inhibitors in MCL, targeting abnormal MALT1 activity could be a promising therapeutic strategy to overcome BTK inhibitor resistance, and cotargeting of MALT1 and BTK should improve MCL treatment efficacy and durability as well as patient outcomes.

In this study, we found that MALT1 was overexpressed in IBN-resistant (IBN-R) MCL cells. Genetic knockout (KO) or pharmacological inhibition revealed that MALT1, but not CARD11, drives BTKi resistance via bypassing upstream BTK/CARD11 signaling. Unbiased transcriptome and protein profiling revealed that MALT1 inhibition potently suppressed NF-κB, PI3K/AKT/mTOR, and integrin signaling, and diminished MCL cell proliferation and dissemination. Importantly, combinatorial treatment with Bruton's tyrosine kinase (BTK) is a proven target in mantle cell lymphoma (MCL), an aggressive subtype of non-Hodgkin lymphoma. However, resistance to BTK inhibitors is a major clinical challenge. We here report that MALT1 is one of the top overexpressed genes in ibrutinib-resistant MCL cells, while expression of CARD11, which is upstream of MALT1, is decreased. MALT1 genetic knockout or inhibition produced dramatic defects in MCL cell growth regardless of ibrutinib sensitivity. Conversely, CARD11-knockout cells showed antitumor effects only in ibrutinib-sensitive cells, suggesting that MALT1 overexpression could drive ibrutinib resistance via bypassing BTK/CARD11 signaling. Additionally, BTK knockdown and MALT1 knockout markedly impaired MCL tumor migration and dissemination, and MALT1 pharmacological inhibition decreased MCL cell viability, adhesion, and migration by suppressing NF-κB, PI3K/AKT/mTOR, and integrin signaling. Importantly, cotargeting MALT1 with safimaltib and BTK with pirtobrutinib induced potent anti-MCL activity in ibrutinib-resistant MCL cell lines and patient-derived xenografts. Therefore, we conclude that MALT1 overexpression associates with resistance to BTK inhibitors in MCL, targeting abnormal MALT1 activity could be a promising therapeutic strategy to overcome BTK inhibitor resistance, and cotargeting of MALT1 and BTK should improve MCL treatment efficacy and durability as well as patient outcomes.
Cotargeting of BTK and MALT1 overcomes resistance to BTK inhibitors in mantle cell lymphoma with acquired IBN resistance. These IBN-R cells expressed much higher levels of MALT1 protein, even as CARD11 protein was expressed at notably reduced levels ( Figure 2A). To determine whether MALT1 overexpression was correlated with increased paracaspase activity, we examined the cleavage of MALT1 substrates. Higher MALT1 expression indeed correlated positively with the increased cleavage of its substrates (Figure 2A). MALT1 endogenous cleavage activity in JeKo BTK KD-2 cells was further confirmed to be significantly higher than that observed in JeKo-1 cells (P < 0.0001) ( Figure 2B). MALT1 overexpression and its heightened paracaspase activity were further validated in MCL cell lines having primary IBN resistance ( Figure 2C) and in primary patient MCL cells (Supplemental Figure 2A We next established subcutaneous cell line-derived xenograft (CDX) models in immunodeficient NOD.Cg-Prkdc scid Il2rg tm1Wjl / SzJ (NSG) mice. Consistent with the in vitro cell proliferation assays ( Figure 2, H and I), MALT1 KO greatly suppressed tumor growth and serum levels of β-2-microglobulin (B2M), a systematic indicator of tumor load, in both JeKo-1 (P < 0.0001 and P < 0.05, respectively) and JeKo BTK KD-2 CDX (P < 0.001 and P < 0.0001, respectively) models, while CARD11 KO suppressed tumor growth and serum levels of B2M in only the JeKo-1 CDX model 1 (P < 0.0001 and P < 0.05, respectively) ( Figure 2, J-N). These data demonstrate that MALT1 is crucial in driving MCL tumorigenesis and IBN resistance via a compensatory mechanism that appears to bypass upstream BTK/CARD11 signaling.
MI-2 blocks MALT1's paracaspase activity and MCL cell proliferation by suppressing NF-κB signaling. To assess the potential role of MALT1's paracaspase activity, we treated MCL cells with MI-2 (31). A 6-hour MI-2 treatment suppressed endogenous MALT1 cleavage activity in a dose-dependent manner in JeKo-1 (P < 0.0001), JeKo BTK KD-2 (P < 0.0001), and Z138 (P < 0.0001) cells ( Figure 3A and Supplemental Figure 3A). A 72-hour treatment with MI-2 produced a potent loss of viability in JeKo-1-and JeKo-derived resistant cells as well as other cell lines with half-maximal inhibitory concentrations (IC 50 ) in the nanomolar range ( Figure 3B and Supplemental Figure 3, B and C). After 24 hours of treatment, MI-2 also effectively inhibited cell viability of primary MCL patient cells MALT1 inhibitors (e.g., MI-2 and safimaltib) and BTKis (e.g., IBN and PBN) produced potent anti-MCL activity in both IBN-R MCL cell lines and patient-derived xenograft (PDX) models. Thus, we envision that cotargeting of MALT1 and BTK could be a promising therapeutic strategy to overcome BTKi resistance in MCL.

Results
MALT1 is overexpressed in IBN-R MCL cells. BTKis like ibrutinib and BCL2 inhibitors like venetoclax (VEN) have been demonstrated to be highly efficacious in treating patients with MCL. However, the development of single or dual resistance to targeted therapy is common. To investigate the potential mechanism underlying this resistance, we performed whole-transcriptome profiling of 9 MCL cell lines that had varying degrees of targeted agent sensitivity (see Methods) (Supplemental Figure 1A; supplemental material available online with this article; https:// doi.org/10.1172/JCI165694DS1). MALT1 was among the top differentially expressed genes (DEGs) across the genome-wide transcriptome and among all NF-κB signaling genes, and it was more highly expressed in IBN-R than IBN-sensitive (IBN-S) cells (P < 0.001) ( Figure 1, A-C, and Supplemental Figure 1, B and C). MALT1 expression was also significantly higher in cells resistant to both IBN and VEN (Dual-R cells) than in cells sensitive to both drugs (Dual-S cells) (P < 0.001) ( Figure 1C). Interestingly, expression of CARD11, which encodes the upstream binding partner of MALT1, was lower in IBN-R MCL cells (Supplemental Figure 1, B and C).
Elevated MALT1 expression was also observed in IBN-R MCL cells (P < 2.22 × 10 -16 ) compared with IBN-S MCL cells in primary patient samples, based on our single-cell RNA sequencing analysis ( Figure 1D). Furthermore, quantitative PCR revealed that MALT1 expression was also higher in primary MCL cells compared with healthy PBMCs (P < 0.05) (Supplemental Figure 1D). Consistent with the whole-transcriptomic analysis, IBN-R and Dual-R MCL patient samples expressed significantly higher levels of MALT1 compared with IBN-S MCL patient samples (P < 0.001 and P < 0.01, respectively) ( Figure  1E). High MALT1 expression was associated with poor clinical outcomes (P < 0.05) ( Figure 1F) as it also was for 2 similarly defined patient cohorts described by others (P < 0.05) (Supplemental Figure 1, E and F) (28,29).
MALT1 is crucial in mediating NF-κB and other signaling pathways (8)(9)(10)19). Gene set enrichment analysis (GSEA) revealed that multiple cancer signaling hallmarks, including MYC targets, NF-κB signaling, and G 2 /M checkpoint, were associated with IBN-R and Dual-R MCL cells (Supplemental Figure  1G). Importantly, deeper analysis showed that both canonical and noncanonical NF-κB signaling were upregulated in IBN-R and Dual-R cells at the transcriptional level ( Figure 1G). This was further confirmed by the activation of NF-κB family members in IBN-R cells compared with IBN-S cells (Supplemental Figure  1H), suggesting upregulated NF-κB signaling associated with poor MCL patient survival.
MALT1 acts as an oncogenic driver of IBN resistance in MCL cells. To investigate how MALT1 overexpression may confer IBN resistance, we first assessed MALT1 protein levels in JeKo BTK KD-1 and -2 cells with intrinsic IBN resistance and in JeKo-R cells , and IBN-S (n = 13) cells. Statistical significance was determined based on the adjusted P values using Dunnett's approach. **P < 0.01; ***P < 0.001. (F) High MALT1 mRNA expression correlated with progress-free survival in MCL patients. The log-rank test was used to assess the statistical significance of progression-free survival. (G) GSEA identifies NF-κB signaling pathways as top cancer hallmarks that were upregulated in IBN-R cells compared with IBN-S cells. FDRs were generated using the Benjamini-Hochberg method. Box-and-whisker plots in C and E show the median ± 1 quartile, with whiskers extending from the hinge to the smallest and largest values within 1.5 × (interquartile range) from the box boundaries. The values beyond the ends of the whiskers are outliers. All other data represent the mean ± SD. cells in the spleen (P < 0.0001), liver (P < 0.0001), BM (P < 0.0001), and PB (P < 0.01) ( Figure 5, B-I). BTK KD in the JeKo-1 cells showed comparably decreased tumor cell presence in the spleen (P < 0.001) and BM (P < 0.001), which was also seen with MALT1-KO cells ( Figure 5, C-F). However, BTK KD did not result in dramatic suppression in tumor cell presence in PB or liver ( Figure 5, B and G-I).
CARD11 KO in JeKo-1 cells showed modest effects on tumor cell presence in PB (P < 0.05), and it had no obvious impact on tumor cell incidence in the spleen, liver, or BM ( Figure 5, B-G). MALT1 KO in JeKo BTK KD-2 cells did not appear to have an effect on the tumor burden in the spleen or liver, since these cells were already very limited in their quantity at these sites. However, MALT1 KO, but not CARD11 KO, showed an additional significant decrease in the tumor cell incidence in BM (P < 0.0001) and PB (P < 0.01).
Next, we performed an in vivo short-term homing assay using a patient apheresis sample that contained 95% or greater MCL cells. The cells were pretreated with MI-2 or DMSO for 30 minutes, and then washed before injecting intravenously into NSG mice. The percentage of MCL cells in PB was determined to be similar for both groups at 1 hour after injection (Supplemental Figure 4A), but at 4 days after drug pretreatment it was significantly reduced in the MI-2 group compared with control in PB (P < 0.05), spleen (P < 0.05), and BM (P < 0.0001) (Supplemental Figure 4, B-D).
To assess the long-term effect of MALT1 inhibition on MCL tumor dissemination, we established PDX models by intravenous injection of IBN-R PDX cells. Treatment with MI-2, but not IBN, significantly decreased tumor burden in the spleen (P < 0.05), BM (P < 0.05), and PB (P < 0.05) ( Figure 5, J and K). These data indicate that MALT1 inhibition is potentially useful in suppressing dissemination in IBN-R MCL tumors.
MALT1 inhibition potently suppresses cell PI3K/AKT/mTOR signaling, adhesion, and migration in vitro. We employed unbiased reverse-phase protein array (RPPA) analysis on cultured MCL cells to further understand the possible mechanisms of MALT1-driven dissemination. IBN-S and -R cells were treated with MI-2 for 6 hours and subjected to RPPA profiling. Many molecules involved in PI3K/AKT/mTOR signaling, including phosphorylated mTOR, rictor, AKT, P70-S6K, and S6, were dramatically downregulated in all of these cells following MI-2 treatment ( Figure 6A).
Activation of PI3K/AKT/mTOR and integrin-β1 signaling is reportedly important for TME-driven IBN resistance (27) and was confirmed in IBN-R cells. Therefore, we hypothesized that MALT1 modulates MCL dissemination via regulating PI3K/AKT/mTOR and integrin-β1 signaling. Phosphorylation of AKT, S6, and p90RSK was upregulated in JeKo-1-derived IBN-R cell lines ( Figure 6B). A 1-hour MI-2 pretreatment effectively blocked phosphorylation and activation of PLCγ2, BTK, and AKT when JeKo-1 cells were stimulated with anti-IgM for 5 minutes to induce BCR-triggered PI3K/AKT/mTOR signaling ( Figure 6C). Consequently, MI-2 treatment also led to a dose-dependent reduction in intracellular ATP levels in all MCL cell lines tested ( Figure 6D).
Our whole-transcriptome profiling revealed that apical junction regulation was one of the top cancer hallmarks notably resistant to IBN (Supplemental Figure 3D). This activity was lost when MALT1 expression was depleted by shRNA in JeKo-1 and JeKo BTK KD-2 cells ( Figure 3C). MI-2 also induced robust apoptosis in MCL cells regardless of their IBN sensitivity (P < 0.001) (Supplemental Figure 3E), which was accompanied by cleavage of poly(ADP-ribose) polymerase (PARP) and caspase 3 in JeKo BTK KD-2 and primary MCL patient cells ( Figure 3D). To further assess the potency of MI-2 against MCL in vivo, we engineered JeKo BTK KD-2 cells to stably express the luciferase reporter gene and then established JeKo BTK KD-2 CDX models. MI-2 treatment (25 mg/ kg, daily) significantly inhibited tumor growth in these IBN-R CDX models (P < 0.001) (Figure 3, E and F).
Both canonical and noncanonical NF-κB signaling pathways were upregulated in IBN-R cells ( Figure 1G), suggesting that MALT1 overexpression may be the key factor conferring their constitutive activation. To address this, we employed unbiased whole-transcriptome GSEA profiling of JeKo-1 and JeKo BTK KD-2 cells treated with MI-2 or vehicle (dimethyl sulfoxide, DMSO) for 6 hours. We identified NF-κB signaling and related inflammatory responses as the top downregulated pathways upon MALT1 inhibition (Supplemental Figure 3F). Our analysis also revealed that MALT1 inhibition suppressed both canonical and noncanonical NF-κB signaling in these cells ( Figure 4A). This was further validated by assays assessing p65, p50, c-Rel, RelB, and p52 activity in JeKo-1 (P < 0.05) and JeKo BTK KD-2 (P < 0.0001) cells (Figure 4, B and C). In the IBN-S cells, canonical NF-κB family members were suppressed to a greater extent upon MALT1 inhibition compared with the noncanonical family members. Conversely, in the JeKo BTK KD-2 cells, both the canonical and noncanonical NF-κB family members were markedly inhibited.
Excess ROS production can trigger apoptosis (32), and NF-κB activation attenuates ROS production to promote cell survival (33). MALT1 inhibition triggered upregulation of the ROS pathway (P < 0.05) and ROS production (P < 0.05) and loss of mitochondrial membrane potential (ΔΨ m ) (P < 0.0001) (Supplemental Figure 3, G and H), suggesting a role of mitochondrial dysfunction in the observed MI-2-induced ROS production. Together, these data suggest that targeting MALT1 paracaspase activity with MI-2 is effective in overcoming IBN resistance in MCL via modulating NF-κB activity and the ROS pathway.
MALT1 and BTK are required for MCL cell dissemination in vivo. Once disseminated, MCL can involve one or more lymph nodes, peripheral blood (PB), bone marrow (BM), spleen, liver, gastrointestinal tract, and even the central nervous system (34), as can be recapitulated using CDX or PDX xenografts (35). Further, MALT1-dependent cleavage of BCL10 reportedly controls integrin-dependent adhesion of T cells and MALT lymphomas (16,36). Therefore, we speculated that MALT1 may mediate cell adhesion, and potentially MCL dissemination, via interaction between MCL cells and the TME.
We established disseminated CDX models by intravenous injection using JeKo-1 and JeKo BTK KD-2 cells with or without MALT1 or CARD11 KO ( Figure 5A). In CDX models using JeKo-1 cells, the tumor cells accumulated markedly in the spleen, liver, BM, and even in PB ( Figure 5, B-I), demonstrating MCL dissemination to these tissues. In JeKo-1 cells with MALT1 KO, we observed a significant decrease in the frequency of tumor  L and M). At the end of experiments, subcutaneous tumors were dissected, weighed, and imaged (N). Two-way ANOVA was used in E and F and H-M to assess the effect of 2 factors (i.e., time and cell line) and 1-way ANOVA was used in N. Statistical significance was determined based on the adjusted P values using Šídák's method. Data represent mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Figure 3F). In addition to apical junction regulation, further analysis revealed that MALT1 inhibition suppressed multiple pathways involved in MCL dissemination, including cell adhesion molecules, focal adhesion complexes, and adherens junction proteins (Supplemental Figure 5A). DEG analysis showed that multiple integrin signaling molecules and cell adhesion molecules were downregulated upon MALT1 inhibition in both JeKo-1 and JeKo BTK KD-2 cells (Figure 6, E and F). Most of these proteins can directly or indirectly regulate integrin-mediated signaling and cell adhesion (37)(38)(39).

downregulated following MALT1 inhibition (Supplemental
To address this further, we next screened extracellular matrix (ECM) components. Among 8 ECM components, we identified fibronectin and laminin as the dominant ECM factors involved in cell adhesion using MCL cell lines; interestingly, IBN-R cells (JeKo-R, JeKo BTK KD-1, and JeKo BTK KD-2) showed significantly higher cell adhesion to fibronectin (P < 0.0001), laminin (P < 0.0001), and bovine serum albumin (P < 0.0001) compared with IBN-S JeKo-1 cells (Supplemental Figure 5B). MI-2 treatment diminished MCL cell adhesion to fibronectin (P < 0.001), laminin (P < 0.01), and fetal bovine serum (FBS) (P < 0.0001), especially for IBN-R MCL cells (Figure 6, G and H, and Supplemental Figure 5C). Furthermore, MI-2 treatment diminished the capacity of MCL cells to migrate through Transwell inserts to a preseeded HS-5 stromal cell monolayer (P < 0.0001) ( Figure  6I). MCL cells showed greater capacity in cell migration through Transwell inserts to stromal cells than to the culture supernatants harvested from HS-5 stromal cell cultures (P < 0.0001) (E and F) NSG mice bearing luciferase-expressing JeKo BTK KD-2-derived subcutaneous xenografts were treated with vehicle (n = 5) or MI-2 (n = 5) at 25 mg/kg daily via intraperitoneal injection for 24 days. Tumor growth was monitored by live animal luminescence imaging (E) and the luciferase flux was plotted (F). Two-way ANOVA was used in C and F, and statistical significance was determined based on the adjusted P values using Šídák's method. *P < 0.05; ***P < 0.001; ****P < 0.0001.
( Figure 6J), suggesting that this HS-5-induced cell migration requires MCL-stromal cell contact. Together, these data suggest that MALT1 plays important roles in mediating cell adhesion, migration, and dissemination.
Cotargeting of MALT1 and BTK overcomes IBN resistance in vitro and in vivo. To screen for combinational therapies that have the potential to overcome IBN resistance, we treated MCL cells with MI-2 in combination with more than 10 drugs either FDA approved or under investigation. Among these, MI-2 in combination with IBN showed the greatest anti-MCL efficacy against 2 IBN-R patient samples and 1 IBN-R PDX sample ( Figure 7A and Supplemental Figure 6A). This was further validated using MCL cell lines and additional patient and PDX samples ( Figure  7B and Supplemental Figure 6B). Consistent with this, the MI-2 and IBN combination significantly induced apoptosis that was higher than with either single agent in JeKo-1 and JeKo BTK KD-1 and -2 cells (Supplemental Figure 6C).
Unbiased RPPA profiling showed that the MI-2 and IBN combination induced a synergistic effect on protein profiles of both JeKo-1 and JeKo BTK KD-2 cells ( Figure 7C). The top cancer hallmarks that were suppressed by the combination were PI3K/AKT/ mTOR signaling, apical junction proteins, G 2 /M checkpoint proteins, and E2F target proteins ( Figure 7D), while apoptosis and hypoxia were among the top cancer hallmarks that were upregulated ( Figure 7D). PI3K/AKT/mTOR and NF-κB signaling were further confirmed to be dramatically downregulated via Western blotting in JeKo-1 and JeKo BTK KD-2 cells treated with the MI-2 and IBN combination, compared with either single agent or vehicle ( Figure 7E). MI-2 in combination with PBN also showed stronger antitumor activity than either single agent in IBN-R cells (P < 0.0001) ( Figure 7F). Of note, JeKo-R and JeKo BTK KD-2 were demonstrated to be resistant to PBN (Supplemental Figure 6D).
MI-2 is commonly used as a chemical tool to inhibit MALT1 paracaspase activity in mechanistic and functional studies. For rational therapeutic development, we next tested another MALT1 inhibitor named safimaltib. Safimaltib was recently developed as a specific MALT1 paracaspase inhibitor (40), and it is currently under early clinical investigation in patients with non-Hodgkin lymphoma and chronic lymphocytic leukemia (ClinicalTrials.gov NCT03900598). Safimaltib showed effective anti-MCL activity not only in IBN-S cells, but also in both IBN-R and PBN-R cells (Supplemental Figure 6E). Additionally, safimaltib in combination with IBN or PBN was highly synergistic against JeKo-R cells in vitro ( Figure 7G). Furthermore, safimaltib at 50 mg/kg daily in combination with PBN at 30 mg/kg twice daily dramatically inhibited tumor growth of an IBN-R PDX model (P < 0.01), and prolonged mouse survival (P < 0.05) beyond that observed by either single agent treatment (Figure 7, H and I). No effects on body weight were observed for either single agent or the combination treatment during this experiment (Supplemental Figure 6F). Together, these data indicate that cotargeting of MALT1 and BTK is promising to overcome resistance to BTK inhibitors in MCL. C). Two-way ANOVA was used in B and C, and statistical significance was determined based on the adjusted P values using Šídák's method. Data represent mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ondary sites like liver or PB. BTK can regulate integrin activation and cell adhesion via PI3K-dependent signaling (41). BTK inhibition by IBN has been shown to cause lymphocytosis, a process of compartment shifts of tumor cells from lymphoid tissues to the periphery, in patients with chronic lymphocytic leukemia (42) or MCL (43). Our unbiased transcriptomic profiling revealed that many genes involved in integrin signaling and cell adhesion were downregulated upon MALT1 inhibition in MCL cells. Furthermore, PI3KCD and AKT1, which are involved in PI3K/AKT/mTOR signaling, were also found to be downregulated upon MALT1 inhibition. Therefore, it is likely that, in addition to promoting cell survival through NF-κB signaling, MALT1 overexpression may enhance the capacity and scope of cell adhesion to prevent BTKiinduced lymphocytosis and tumor cell killing and thus confer BTKi resistance and immune evasion. Conversely, CARD11 seems to be dispensable in modulating MCL cell adhesion and dissemination. Our data suggest that an MCL-TME-mediated compensatory mechanism is critical for conferring BTKi resistance by integrin

Discussion
We demonstrate that both MALT1 and CARD11 are crucial for MCL malignancy in IBN-S cells, and MALT1 becomes aberrantly expressed in IBN-R MCL cells and correlates negatively with CARD11 expression in these cells. Both MALT1 and CARD11 are crucial for the proliferation of IBN-S MCL cells in vitro and in vivo. However, only MALT1 is critical for the cell proliferation of IBN-R MCL cells and for MCL cell dissemination. These data suggest what we believe is a novel function for MALT1 in driving IBN resistance and in modulating MCL cell dissemination, both in a CARD11-independent manner.
BTK KD in JeKo-1 also demonstrated a defect in MCL cell dissemination to primary MCL sites like spleen and BM, but not to sec- Freshly isolated primary PDX cells were injected intravenously into NSG mice to establish disseminated PDX models (n = 5 per group). At 6 weeks after injection, the mice were treated with vehicle, ibrutinib (50 mg/ kg), or MI-2 (25 mg/kg) daily for 4 weeks. At the end of the experiment, mouse spleens were weighed (J). The CD5 + CD20 + MCL cells were measured in spleens (K, left panel), BM (K, middle panel), and PB (K, right panel). Oneway ANOVA was used in B-E, H, J, and K, where statistical significance was determined based on the adjusted P values using Šídák's method. Data represent mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. opportunities for the development of therapeutic resistance in MCL. Therefore, we envision that this can ultimately prevent therapy relapse and promote disease-free survival for MCL patients.
Safimaltib is a first-in-class, oral, specific MALT1 inhibitor currently under multiple clinical investigations to evaluate its clinical efficacy as single agent (ClinicalTrials.gov NCT03900598) or in combination with covalent BTKis IBN and JNJ-64264681 (ClinicalTrials.gov NCT04876092 and NCT04657224). Different from covalent BTK inhibitors, PBN is a noncovalent BTKi with high selectivity and potency. The BRUIN trial demonstrated that PBN can overcome resistance to a covalent BTKi in half of patients with prior failure to covalent BTKis (24). Therefore, we expect to see better efficacy of safimaltib in combination with PBN than with covalent BTKis. Our data provide evidence to support the potential clinical translation of safimaltib and PBN combination in patients with BTKi resistance in MCL and CLL.
In addition to MCL in this study, abnormal MALT1 activity is critical for driving MALT lymphoma. MALT lymphoma is the most common extranodal subtype of non-Hodgkin lymphoma. The chromosomal translocation t(11;18) (q21:q21) is characteristics of MALT lymphoma, which results in expression of an API2-MALT1 fusion oncoprotein. API2-MALT1 protein contains the N-terminus of API2 and the C-terminus of MALT1 (immunoglobulin-like domains and caspase-like domain) (45). API2-MALT1 can auto-oligomerize and thus has constitutive paracaspase activity, leading to potent and constitutive activation of NF-κB signaling (45). In such a way, it does not require any upstream signaling, including activation of CARD11 and assembly of the CARD11-BCL10-MALT1 complex, to mediate NF-κB signaling as the wild-type MALT1 does. In our study, we showed evidence that MALT1 is overexpressed and constitutively active in IBN-R cells. MALT1 KO, but not CARD11 KO, was able to dramatically diminish the tumor growth and dissemination in vitro and in vivo in the resistant MCL cells, supporting our notion that MALT1 hyperactivity critically contributes to IBN resistance via bypassing BTK/CARD11 upstream signaling. It has been shown that t(11;18)-positive lymphomas are associated with treatment resistance and a higher tendency to disseminate. Therefore, it is likely that MALT lymphoma is resistant to IBN treatment, which requires further investigation.

Methods
Cell samples. MCL specimens and healthy PBMC samples were acquired from patients and healthy donors, respectively, after obtaining written informed consent. The MCL patient characteristics are presented in Supplemental Table 1.
Reagents and antibodies. IBN, PBN, MI-2, and VEN were purchased from Selleck Chemicals. Safimaltib was generated by custom synthesis through a contract research organization service. Supplemental Table 2 lists antibodies used and their sources. DMSO was purchased from Sigma-Aldrich.
MALT1 promotes T cell receptor-dependent activation of mTOR signaling and increases metabolism in CD4 + T cells (44), indicating crosstalk between NF-κB and mTOR signaling in regulating cellular metabolism. Our RPPA analysis showed that MI-2 treatment blocked AKT/mTOR signaling, which confers TME-driven IBN resistance in MCL cell lines (27). MI-2 treatment of MCL cells led to ROS production, loss of ΔΨ m , and thus abolished ATP production. These findings suggest that MALT1 plays a crucial role in regulating MCL cell survival and associated cellular energy metabolism via NF-κB and PI3K/AKT/mTOR signaling.
KO or pharmacological inhibition of MALT1 diminished MCL cell proliferation and survival irrespective of IBN sensitivity, suggesting that MALT1 is intimately associated with MCL viability. Recent clinical data on PBN (24) demonstrated that BTK is still targetable in many covalent BTKi-resistant patients. Therefore, dual targeting of MALT1 and BTK may greatly improve clinical outcomes in patients whether they are naive to BTKi therapy or have prior failure to it. Indeed, our preclinical data on dual targeting of MALT1 and BTK using combinations of the respective inhibitors demonstrates that this treatment strategy promotes potent anti-MCL activity in MCL cells with resistance to BTKis. In BTKi-sensitive cells, the dual targeting shuts down NF-κB and PI3K/AKT/ mTOR signaling required for MCL cell survival, proliferation, adhesion, and dissemination. Conversely, in BTKi-resistant cells, MALT1 becomes aberrantly expressed and thus promotes cell survival and proliferation while conferring TME-mediated BTKi resistance as an alternative compensatory mechanism. Thus, dual targeting of BTK and MALT1 blocks multiple adaptive Error bars were generated from 3 independent replicates (G-J). Two-way ANOVA was used in D and G-J, and statistical significance was determined based on the adjusted P values using Šídák's method. Data represent mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.  replicates (B, F, and G). (H and I) Freshly isolated primary PDX cells were injected subcutaneously into NSG mice to establish PDX models (n = 6 per group). When the subcutaneous tumors became palpable, the mice were treated with vehicle, pirtobrutinib (30 mg/kg twice daily), or safimaltib (50 mg/kg daily), alone or in combination. Tumor growth (H) and mouse survival (I) were monitored and plotted. Data represent mean ± SD. One-way ANOVA was used in B and F, 2-way ANOVA was used in H, and the pairwise log-rank test was used in I. Statistical significance was determined based on the adjusted P values using Šídák's method. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.
Immunoblotting. The immunoblotting assay was performed as described previously (46). Briefly, 5 × 10 6 to 10 × 10 6 cells were seeded and treated as indicated. The cells were lysed in lysis buffer containing 50 mM HEPES (pH 7.4), 250 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 1 mM Na 3 VO 4 , 1 mM PMSF, 1 mM NaF, and a protease inhibitor mixture (all purchased from Roche Diagnostics). Protein concentration in cell lysates was measured using a Quick Start Bradford Protein Assay Kit (Bio-Rad) and the lysates were analyzed by SDS-PAGE and Western blotting. Primary antibodies used to detect total proteins, phosphorylated proteins, and their sources are presented in Supplemental Table 2.
MALT1 ELISA. The method for the MALT1 ELISA measuring MALT1 endogenous cleavage activity has been described elsewhere (51). Briefly, 0.2 × 10 6 to 1 × 10 6 MCL cells/mL were seeded on a 6-well plate and treated with MI-2 for 6 hours. The cells were harvested and analyzed by MALT1 ELISA to measure MALT1 endogenous cleavage activity. The experiments were repeated at least 3 times.
NF-κB DNA-binding assay. Nuclear fractions were extracted via the Nuclear Extract Kit (40410, ActiveMotif) and the DNA-binding activity of NF-κB family members was measured using a TransAM NF-κB Family Kit (43296, ActiveMotif) according to the manufacturer's manual.
RPPA analysis. RPPA analysis was conducted as described previously (48). Briefly, RPPA analysis was conducted by the MD Anderson RPPA Core Facility. A total of 54 samples, representing 6 cell lines (JeKo-1, JeKo-R, JeKo BTK KD-2, Maver-1, Mino, and Z-138) and 3 doses of MI-2 (0, 1, 2 μM), tested in triplicate, were analyzed by RPPA. The slide images were quantified using MicroVigene 4.0 (Vigene-Tech). Spot-level raw data were processed using the R package SuperCurve, developed in-house. This package returns the estimated protein concentration (raw concentration), as well as a quality control score. Raw concentration data were normalized via median-centering each sample across all of the proteins to correct loading bias. In total, 307 antibodies and secondary antibody negative controls were analyzed. NormLog2_ MedianCentered values were selected for heatmap generation. The differences in values of proteins in MI-2-treated samples were normalized to untreated samples, and differences less than -0.3 and greater than 0.3 in 3 or more cell lines were selected for the heatmap. The heatmap was generated using Cluster 3.0 software (https://cluster2.software. informer.com/3.0/) and visualized in Treeview (https://www.treeview. co.uk/). The results were presented in a high-resolution bitmap format. Each treatment was performed in triplicate.
Screen for ECM, cell adhesion, and migration assay. These assays were performed as described previously (52). Briefly, the ECM Cell Adhesion Array Kit (ECM540, Millipore) was used to screen for MCL cell adhesion to the ECM according to the manufacturer's manual. For specific cell adhesion assay, the plates were first coated with 10% FBS, fibronectin, or laminin, seeded with MCL cells pretreated with MI-2 at 0.5 μM for 30 minutes, and incubated for an additional 4 hours. The cells in suspension were thoroughly washed with PBS, and the resulting cells were (Supplemental Figure 1A). JeKo BTK KD-1 and -2 cells derived from JeKo-1 cells have intrinsic IBN resistance due to BTK depletion (47), and JeKo-R cells have acquired resistance to IBN (47). Mino-VR, Rec-VR, and Granta-519-VR cells have acquired resistance to VEN (48). We grouped these cells into 3 pairings based on their drug sensitivity: (a) IBN-R versus IBN-S cells, (b) VEN-R versus VEN-S cells, and (c) Dual-R versus Dual-S cells (Supplemental Figure 1A). In addition, Maver-1 and Z138 are primarily resistant to IBN, while SP-49 is sensitive to IBN. Cell lines were authenticated by single-nucleotide polymorphism profile fingerprinting.
Generation of MCL cells with stable gene KD via shRNA or with inducible expression of MALT1. The method for generating MCL cells with stable MALT1 KD has been described elsewhere (30). Quantitative real-time PCR was used to quantify the KD efficiency; 70% or more of mRNA expression was knocked down in all cells with stable MALT1 KD.
Establishment of JeKo BTK KD GFP-Luc cells. JeKo BTK KD-2 cells were infected with pseudolentiviruses packaged with pGF1-CMV-GFP-Luc and then sorted for GFP-positive cells. These cells were further expanded for a second sorting for GFP-positive cells and maintained in the same conditions as the other MCL cell lines.
Bulk RNA sequencing and analysis. Bulk RNA sequencing and analysis were performed as described previously (50). The sequencing data set has been deposited in the European Genome-Phenome Archive (EGA) database with the accession number EGAD00001009771.
cDNA synthesis and quantitative real-time PCR. The method for cDNA synthesis and quantitative real-time PCR has been described elsewhere (30). Briefly, first-strand cDNA was synthesized via the iScript cDNA Synthesis Kit (1725037, Bio-Rad) according to the manufacturer's manual. Quantitative real-time PCR was conducted using SsoAdvanced Universal SYBR Green Supermix (1725271, Bio-Rad) according to the manufacturer's manual. KiCqStart SYBR Green predesigned primers for MALT1 and GAPDH (Sigma-Aldrich) were used in real-timer PCR experiments. MALT1 mRNA expression was normalized to GAPDH and further normalized to one of the PBMCs samples from the healthy donors.
Cell viability and apoptosis assays. These assays were performed as described previously (46). Briefly, cells from MCL cell lines were seeded at 10,000 cells per well, and PDX tumor cells or primary patient tumor cells were seeded at 125,000 cells per well in a 96-well, white, flat-bottomed plate (3558, Corning). Cells were treated in triplicate with various doses of compounds. The compounds were prepared in DMSO to make a stock solution, from which 2-fold serial dilutions were prepared. Cell viability was determined at 72 hours (cell lines) or 24 hours (primary tumor cells) after seeding. For viability testing, cells to lysed with Cell Titer-Glo Luminescent Cell Viability Assay Reagent (Promega). Further, luminescence was measured via a BioTek Synergy HTX Multi-Mode microplate reader. For cell apoptosis assay, cells treated with or without MI-2 for 24 hours were stained with annexin V and propidium iodide (Abcam), followed by flow cytometry in a NovoCyte Flow Cytometer (ACEA Biosciences) to quantify apoptosis. The experiments were repeated at least 3 times.