Targeting Myddosome Signaling in Waldenström's Macroglobulinemia with the Interleukin-1 Receptor-Associated Kinase 1/4 Inhibitor R191

Waldenström's macroglobulinemia is diagnosed in patients who have at least 10% bone marrow involvement with lymphoplasmacytic lymphoma and a serum monoclonal IgM paraprotein (1, 2). Although some patients have asymptomatic disease which can be followed without treatment, others present with, or progress to symptomatic Waldenström's due to bulky disease, cytopenias, constitutional symptoms, or hyperviscosity, all of which require treatment. Chemotherapy for this disease has traditionally been based on the use of various combinations of drugs, including alkylating agents, anti-CD20 monoclonal antibodies, corticosteroids, nucleoside analogues, and proteasome inhibitors (1, 2). These regimens typically are effective at reducing disease burden and provide good long-term outcomes, but complete responses are rare and Waldenström's is still considered incurable.

Understanding of the molecular pathobiology of Waldenström's took a huge leap forward with the finding that the vast majority of patients harbor an L265P mutation in the myeloid differentiation primary response gene 88 (MYD88; refs. 3, 4). MYD88 is recruited to activated Toll-like receptors (TLR) and interleukin-1 receptor (IL-1R; refs. 5, 6), resulting in formation of MYD88 homodimers, which then interact with interleukin-1 receptor-associated kinase 4 (IRAK4) and IRAK2 to form the so-called Myddosome (7). Subsequent recruitment to, and phosphorylation of IRAK1 by the Myddosome is required for binding of IRAK1 to tumor necrosis factor receptor–associated factor 6 (TRAF6), an E3 ubiquitin ligase that mediates activation of the inhibitor of NF-κB kinase (IKK). Once IKK is activated, it phosphorylates the inhibitor of NF-κB (IκB), prompting its ubiquitination and proteasome-mediated degradation, freeing NF-κB to translocate to the nucleus and activate its gene expression program (reviewed in ref. 7) to promote B-cell survival and proliferation (8–15). The L265P mutation within the MYD88 Toll/interleukin-1 receptor (TIR) homology domain changes its structure and induces receptor activation-independent Myddosome formation, leading to constitutive NF-κB activation (16).

In Waldenström's models, MYD88^L265P forms complexes with and activates Bruton's tyrosine kinase (BTK), providing a rationale for the use of tool compound BTK inhibitors, which showed activity against preclinical models (3). These findings set the stage for testing of the BTK inhibitor ibrutinib, whose remarkable clinical activity against relapsed and/or refractory Waldenström's (17) prompted its regulatory approval. Because IRAK1/4 remains an interesting target for this disease, we set out to evaluate the activity of a novel set of potent and specific inhibitors of these kinases (18). As reported herein, these agents reduced viability and activated apoptosis in Waldenström's models, inhibited NF-κB and protein kinase B/Akt signaling, showed efficacy against primary Waldenström's cells and in vivo, and acted synergistically with other drugs. Thus, these studies provide a good rationale for moving these IRAK1/4 inhibitors to the clinic for patients with Waldenström's macroglobulinemia.

The IRAK1/4 inhibitors R191, R568, and R630 were provided by Rigel Pharmaceuticals, Inc. (18). R191 was profiled in a Panlabs Hit Profiling Screen (30 targets), and gave significant (>50%) inhibition of three targets at 10 μmol/L other than IRAK1/4: Adenosine A1 (51%), cannabinoid CB1 (97%), and serotonin (5-hydroxltryptamine) 5-HT_2B (96%). In follow-up cellular assays at Panlabs, R191 did not show any significant effect on cannabinoid CB1-induced GTPγS binding in CHO-K1 cells, but did antagonize serotonin 5-HT_2B-induced inositol monophosphate (IP1) release. R191 blocked IP1 release with an EC₅₀ of 1.93 μmol/L, which may be due in part to cell cytotoxicity, as R191 caused cytotoxicity in CHO-K1 cells with an EC₅₀ of 8.46 μmol/L, but R191 did not significantly inhibit a panel of related serotonin receptors. The effect of R191 was also evaluated on the human potassium channel using human embryonic kidney (HEK) 293 cells transfected with a human Ether-a-go-go-related gene (hERG), and the maximal inhibition by R191 of the potassium-selective IKr current was 36.6% at 10 μmol/L. With regard to selectivity, R191 was profiled against a panel of 76 kinases representative of the different subfamilies of the full kinome. In addition to IRAK1 and IRAK4 (IC₅₀ 3 nmol/L), R191 inhibited the kinase activity of 11 kinases by >80% at 250 nmol/L and two kinases at 50 nmol/L. The primary off-target activity of R191 was against Fms-related tyrosine kinase (FLT)-3, which was confirmed in cellular assays [Flt3 ligand-induced extracellular signal-regulated kinase-1/2 phosphorylation and MV411 (Flt3-internal tandem duplication) proliferation]. In addition, R191 inhibited the cellular activity of additional targets (TAK1, AXL receptor tyrosine kinase, TANK binding kinase 1, and Bruton's tyrosine kinase) at concentrations of ∼400 nmol/L or above, at least 20-fold over its activity on IRAK1/4 signaling.

Afuresertib, bortezomib, and ibrutinib were purchased from Selleck Chemicals. Other reagents were from Sigma-Aldrich Corporation, unless otherwise indicated below.

The Waldenström's cell line BCWM.1 was a kind gift from Dr. Steven Treon (Dana Farber Cancer Institute, Boston, MA). These cells are CD5⁻/CD10⁻/CD19⁺/CD20⁺/CD23⁺/CD27⁻/CD38⁺/CD138⁺/CD40⁺/CD52⁺/CD70⁺/CD117⁺/cIgM⁺/cIgG⁻/cIgA⁻/ckappa⁻, and clambda⁺, harbor the MYD88 L265P mutation, and express low levels of wild-type CXCR4 (19). MWCL-1 Waldenström's cells were generously provided by Dr. Stephen Ansell (The Mayo Clinic, Rochester, MN), and are CD3⁻/CD19⁺/CD20⁺/CD27⁺/CD38⁺/CD49D⁺/CD138⁺/cIgM⁺ and ckappa⁺. They also harbor the MYD88 L265P mutation and a wild-type CXCR4, but they have a TP53 missense mutation at exon 5 (V143A, GTG>GCG; ref. 19). Cells were propagated in RPMI 1640 media (Thermo Fisher/Life Technologies) supplemented with 10% FCS (GE Healthcare/HyClone), 2 mmol/L L-glutamine (Thermo Fisher/Invitrogen), 100 U/mL of penicillin, and 100 μg/mL of streptomycin (both from Mediatech). Cell line authentication was performed by the MD Anderson Characterized Cell Line Core Facility using short tandem repeat DNA fingerprinting with the AmpFlSTR Kit (Thermo Fisher/Applied Biosystems).

Three primary samples were obtained from patients undergoing bone marrow aspiration after they had provided informed consent in compliance with the Declaration of Helsinki and according to an Institutional Review Board-approved protocol. CD20⁺ primary cells were purified by positive selection using magnetic-activated cell sorting with CD20⁺ microbeads (Miltenyi Biotec).

Peripheral blood mononuclear cells (PBMCs) were isolated from the buffy coats of healthy donors after standard Ficoll-Paque gradient separation, and all the experiments were performed in RPMI 1640 medium.

The cells were seeded at 20,000 cells/well on 96-well plates and treated with vehicle alone, or vehicle with increasing concentrations of R191 for 24 to 72 hours. Cell viability was tested using the Premix WST-1 Cell Proliferation Assay according to the manufacturer's instructions (Clontech Laboratories). Conversion of WST-1 to the formazan dye was measured at 650 nm using a Victor 3V plate-reader (Perkin Elmer). Data provided are representative of at least three independent experiments expressed as the mean ± SD for samples assayed in triplicate.

Cell-cycle analysis based on DNA content was performed using propidium iodide (PI) staining. Briefly, cells were collected and fixed in 70% cold ethanol in PBS for 2 hours at 4°C. After two subsequent washes with cold PBS, cells were incubated with PI staining solution (1× PBS, 100 μg/mL RNAse A, and 40 μg/mL PI) for 30 minutes. For analysis of apoptosis, cells were stained with the Annexin V Apoptosis Detection Kit (BD Biosciences) according to the manufacturer's protocol. Once harvested, cells were resuspended in a binding buffer containing Annexin V-FITC and PI or TO-PRO-3 and incubated for 15 minutes in the dark. Stained cells were analyzed by flow (FACSCalibur; BD Biosciences) using FlowJo software (Tree Star).

Cells were harvested, washed in PBS, and lysed with buffer (Cell Signaling Technology) containing protease and phosphatase inhibitors (Roche Diagnostics). Lysates were sonicated and clarified by centrifugation, and protein concentrations were determined using the DC Assay Kit (Bio-Rad). Fifty micrograms of protein was boiled after addition of loading buffer (Cell Signaling Technology), separated by gel electrophoresis, transferred onto nitrocellulose, and blocked with Tris-buffered saline with 0.01% Tween 20 and 5% nonfat milk prior to incubation with primary and secondary antibodies. Proteins were visualized using the Enhanced Chemiluminescence Kit (Thermo Fisher/Pierce) according to the manufacturer's instructions and exposed on Hyperfilm-ECL (GE Healthcare Biosciences). Protein band intensity was analyzed by densitometry using the ChemiDoc Touch Imaging System (Bio-Rad) and normalized to β-actin as a control.

Primary antibodies used included those detecting p44/42 MAPK (#9102) and phospho-Thr202/Tyr204 p44/42 MAPK (#4370), Akt (#9272) and phospho-Ser473 Akt (#4058), NF-κB p65 (#8242) and phospho-Ser536 NF-κB p65 (93H1, #3033), phospho-Ser32 IκBα (#2859), IRAK1 (#4359S), IRAK4 (#4363) and phospho-Thr345/Ser346 IRAK4 (#11927), cyclin-dependent kinase (CDK)-4 (#12790) and -6 (#13331), phospho-Ser235/236 S6 kinase (#2211), and phospho-Ser21/9 glycogen synthase kinase (GSK)-3α/β (#9327S), all from Cell Signaling Technology. Primary antibodies detecting TRAF6 (ab33915), V-myc avian myelocytomatosis viral oncogene homolog (c-Myc; ab178457), and TATA binding protein (TBP; ab63766) were from Abcam. Primary antibodies detecting phospho-Thr209 IRAK1 (#SAB4504246), tumor protein p53 inducible nuclear protein-1 (TP53INP1; #SAB4503641), 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 4 (PFKFB4; #SAB1300132), and β-actin (A1978) were from Sigma-Aldrich.

Total RNA extracted using Trizol (Thermo Fisher/Life Technologies) underwent first-strand cDNA synthesis with the High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems). TaqMan Gene Expression Master Mix and probes used to perform quantitative PCR (qPCR) were from Applied Biosystems, with GAPDH used as internal controls. Reactions were carried out in triplicate in the ABI PRISM 7900 HT Sequence Detection System (Thermo Fisher/Life Technologies), and relative amounts of products were determined by the comparative Ct method.

Luciferase was excised from pGL3-basic (Promega) and ligated into pCDH-CMV-MCS-EF1-copGFP (Systems Bioscences). Constitutively-active (Akt-DD, T308D/S473D; ref. 20) and dominant-negative Akt (Akt-AAA, K179A/T308A/S473A; ref. 21) mutants from Addgene were cloned into pCDH-CMV-MCS-EF1-puro (Systems Bioscences). Trono vectors (Addgene; ref. 22) expressing shRNA targeting the human c-myc sequences TGAGACAGATCAGCAACAA (shRNA 1-1) and AACGTTAGCTTCACCAACA (shRNA 2-5) were used to knockdown cellular c-myc. The Lentiviral vectors were then used for production of recombinant Lentiviruses by transient transfection of 293T cells. Briefly, subconfluent 293T cells were cotransfected with 20 μg of an expression vector (with or without inserted sequences), 15 μg of pAX2, and 5 μg of pMD2G-VSVG by calcium phosphate precipitation. Medium was changed after 16 hours, and recombinant lentiviruses were harvested 24 and 48 hours later. Raw virus supernatants were concentrated by poly-ethyleneglycol precipitation, and target cells were transduced in growth medium containing 6 μg/mL polybrene. Transduced cells were subjected to drug selection, or selected based on GFP expression by flow cytometry at 5 days posttransduction.

Two IRAK4 loss-of-function mutants (23) were a kind gift from Dr. Andrei Medvedev (University of Connecticut Health Center, Farmington, CT). The first contains a C-to-T substitution at nucleotide 877 (877–879), creating a premature stop codon at amino acid 293, whereas the second has an AC deletion at nucleotides 620–621. They were introduced into the WT pRK7-IRAK4 vector by site-directed mutagenesis and verified by sequencing. These constructs were cloned into Lentiviral vectors and used for production of BCWM.1-derived stable cell lines.

Gene expression profiling (GEP) and gene set enrichment analyses (GSEAs) were performed as described previously (24). For reverse phase protein array (RPPA) analysis, two replicates of BCWM.1 and MWCL-1 cells were treated with vehicle or 1.25 μmol/L R191 for 24 hours. Cells were lysed in RIPA buffer and lysates were centrifuged at 14,000 rpm for 10 minutes at 4°C. The supernatants were mixed with 4× SDS sample buffer [250 mmol/L Tris, pH 7.4, 2% (w/v) SDS, 25% (v/v) glycerol, and 10% (v/v) 2-mercaptoethanol] and boiled for 5 minutes. The RPPA analysis was performed by the MD Anderson RPPA Core Facility (http://www.mdanderson.org/education-and-research/resources-for-professionals/scientific-resources/core-facilities-and-services/functional-proteomics-rppa-core/index.html) as described previously (25).

Six-week-old nonobese diabetic mice with severe combined immunodeficiency (NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ) were from Jackson Laboratories. BCWM.1 GFP⁺/luc⁺ cells (1 × 10⁷ cells) with Matrigel (BD Biosciences) were injected subcutaneously. On day 7 post-inoculation, mice were divided into treatment groups (N = 5 each) to receive intraperitoneal PBS or 20 mg/kg of R191 on days 1 to 5 of three consecutive 7-day cycles. Tumor size/burden was evaluated by three methods: measured by calipers bi weekly, with tumor volume calculated using the formula for an ellipsoid sphere [(length × width²)/2]; by bioluminescence imaging [after IP injection of 200 μL d-luciferin (15 mg/mL; Caliper Life Sciences) in PBS] using the IVIS Imaging System (Caliper Life Sciences) and Living Image 4.4.SP2 software (Caliper Life Sciences); and by ELISA (Bethyl Laboratories) measuring human IgM levels in samples from tail veins on treatment days 14 and 21. Surviving mice were euthanized and tumor tissue was extracted for further analysis.

To develop a model of systemic Waldenström's, BCWM.1 GFP⁺/luc⁺ cells (1 × 10⁶) were injected intravenously into the tail veins of mice. On day 3 post-inoculation, mice were divided into treatment groups (N = 5 each) and treated as above. Tumor size/burden was evaluated by bioluminescence imaging, and blood samples were collected for human IgM testing. Animals were euthanized and bone marrows were harvested for further analysis at the end of the study (day 21 post-inoculation, day 14 of treatment). All in vivo modeling was performed under protocols approved by the MD Anderson Institutional Animal Care and Use Committee.

Combination indices (CIs) from drug response data were calculated using CalcuSyn software (Biosoft). Statistical analyses were performed with unpaired t tests in GraphPad, and P values <0.05 were considered statistically significant. The t test, ANOVA, or their corresponding nonparametric methods (Wilcoxon rank-sum test or Kruskal–Wallis test) were used to detect differences for continuous variables between groups.

We evaluated the potency of a panel of novel IRAK1/4 inhibitors which had been identified originally through a cell-based screen (18). R191, R568, and R630 all produced a dose-dependent reduction in viability of BCWM.1 and MWCL-1 Waldenström's cells after a 24-hour exposure (Supplementary Fig. S1) based on data from tetrazolium dye-based assays. As the median IC₅₀ was lowest for R191, we selected this agent for further evaluation. Over the course of a 72-hour exposure, R191 produced a reduction in viability (Fig. 1A), and the time-dependence of its action was demonstrated in studies evaluating 24-, 48-, and 72-hour time points (Fig. 1B). To confirm that R191 inhibited IRAK1 and IRAK4, we analyzed the expression of the activated/phosphorylated kinases and total IRAK1/4 by Western blotting. Notably, R191 significantly decreased total and phospho-Thr209 IRAK1, and of total and phospho-Thr345/Ser346 IRAK4, in both cell models in a concentration-dependent manner (Fig. 1C). Finally, R191 decreased levels of the IRAK1-interacting partner TRAF6 in a dose-dependent manner (Fig. 1D).

To better understand the mechanisms by which R191 exerted its effects, we performed cell-cycle analysis of Waldenström's cells treated with R191 for 24 hours. R191 increased the fraction of cells in G₀–G₁ phase, and decreased the fraction in S phase in both cell lines in a concentration-dependent manner (Fig. 2A; Supplementary Fig. S2A). This apparent G₀–G₁ cell-cycle arrest suggested that R191 could be modulating levels of CDK4 and CDK6, and Western blot analysis indeed showed decreased levels of both in BCWM.1 and MWCL-1 cells (Fig. 2B). We also noticed an increased proportion of cells in the sub-G₀–G₁ fractions in R191-treated cells during cell-cycle analysis, suggesting the activation of programmed cell death. BCWM.1 and MWCL-1 cells were therefore analyzed by flow cytometry after Annexin V-staining, and this assay confirmed that R191 induced a dose-dependent increase in apoptosis of these cells (Fig. 2C, left; Supplementary Fig. S2B), and also in a primary patient sample (Fig. 2C, right). In order to examine the impact of R191 on normal cells, we exposed PBMCs to increasing R191 concentrations for 24 hours, and then evaluated the levels of apoptosis. Programmed cell death was activated by R191 in <20% of PBMCs even at 2.5 μmol/L (Supplementary Fig. S3). In contrast, BCWM.1 and MWCL-1 cells were more sensitive, since up to 50% to 60% of cells became apoptotic at this concentration (Fig. 2C, left), and the same was true for primary cells, of which 76% were apoptotic at 1.25 μmol/L (Fig. 2C, right). The ability of R191 to activate apoptosis was confirmed by Western blot analysis showing an increased abundance of cleaved and activated caspases-3, -8, and -9, and of cleaved PARP (Fig. 2D, left). Finally, increased levels of cleaved caspase-3 were seen in primary patient samples exposed to R191 (Fig. 2D, right).

Given the role of Myddosome and IRAK signaling in NF-κB induction, we next tested NF-κB activity by evaluating the phosphorylation status of IκBα and the NF-κB subunit p65, as well as nuclear p65 translocation. Western blotting showed decreased phosphorylation of both IκBα and p65 in BCWM.1 and MWCL-1 cells (Fig. 3A). Fractionation studies showed that R191 reduced nuclear p65 levels, and enhanced its abundance in the cytoplasm (Fig. 3B), consistent with inhibition of NF-κB signaling. To further confirm the inhibitory effect of R191, we used lentiviral transduction to express the degradation-resistant IκBα-super-repressor in BCWM.1 cells and, as expected, this reduced downstream phospho-p65 levels (Fig. 3C). These super-repressor expressing cells were more sensitive to R191 treatment (Fig. 3D), confirming the important role of NF-κB signaling as a target for Waldenström's therapy.

Activation of Akt signaling is important to survival and homing of Waldenström's cells (26), and because it can be a downstream target for NF-κB, we next evaluated the impact of R191 on protein kinase B/Akt. Judging by the level of activated, phospho-Ser473 Akt by Western blotting, R191 exposure decreased Akt activation in BCWM.1 (Fig. 4A) and MWCL-1 (Fig. 4B) cells. This was associated with decreased activation of downstream GSK3α/β, as judged by the levels of phospho-Ser21/9 GSK3 (Fig. 4A and B). Similarly, other Akt signaling intermediates showed decreased activation, including PDK1, mTOR, and S6 (Fig. 4A and B), consistent with an inhibitory effect on the entire pathway. Moreover, the use of RPPAs to provide data about the broader proteomic effects of R191 confirmed the reduction of activation of several Akt pathway intermediates (Supplementary Fig. S4A and S4B). To further substantiate the effect of R191 on Akt activation, we next compared BCWM.1 cells expressing either a constitutively-active Akt mutant (Akt-DD; T308D/S473D) or a dominant-negative mutant (Akt-AAA; K179A/T308A/S473A). Expression of the Akt-DD mutant activated downstream signaling as indicated by an increase in phospho-GSK3 and phospho-S6 levels, among others, whereas Akt-AAA reduced these intermediates (Fig. 3C). As expected, BCWM.1 cells expressing the constitutively active Akt mutant were more resistant to R191 (Fig. 4D, top), whereas cells expressing dominant negative Akt showed slightly increased sensitivity. Qualitatively similar findings were noted in MWCL-1 cells expressing the Akt-DD mutant (Fig. 4D, bottom).

To understand the impact of R191 on gene expression, we next performed GEP and GSEA, which confirmed inhibition of Akt/mTOR signaling (Supplementary Table S1; Supplementary Fig. S5A). Additional pathway changes of note included inhibition of the Wingless-related integration site (Wnt)/β-catenin axis, and upregulation of CD40 signaling and the endoplasmic reticulum (ER)-stress response (not shown). Interestingly, R191 impacted the expression of a c-myc-regulated gene-set (Supplementary Fig. S5B), and of genes important for cellular metabolism, including glycolysis. The effect of R191 on c-myc and PFKFB4 expression was confirmed at the mRNA level by qRT-PCR (Fig. 5A; Supplementary Fig. S6A, respectively), and at the protein level by Western Blotting (Fig. 5B; Supplementary Fig. S6B, respectively). To examine if c-myc suppression was important to the effects of R191, we compared the viability of control BCWM.1 cells and BCWM.1 cells in which c-myc was downregulated by shRNAs (Fig. 5C). Downregulation of c-MYC sensitized BCWM.1 cells to the effects of R191 (Fig. 5D), consistent with this possibility.

As R191 may have off-target effects, we suppressed expression of first IRAK1 and then IRAK4 individually in BCWM.1 cells. Decreasing IRAK1 or IRAK4 sensitized these cells to R191 (Supplementary Fig. S7A and S7B), as would be expected if IRAK inhibition is a major mechanism of R191′s action since lower target levels would require less drug to give a similar impact. Also, we expressed two previously described IRAK4 loss-of-function mutants (23), and studied their effect. Expression of both mutants sensitized cells to R191 compared with control cells (Fig. 5E) and, at low R191 concentrations, wild-type IRAK4 overexpression protected cells. Also, the loss of function mutants contributed to inhibition of downstream signaling (Fig. 5F). For example, NF-κB activation as measured by phospho-p65 levels was reduced to a greater extent by R191 in cells with these mutants than in control or wild-type cells. The same was true for phospho-Akt and c-Myc expression, which were decreased to a greater extent by R191 with the dominant-negative constructs than with either approach alone, consistent with the concept that R191 and the mutants had a similar target.

GSEA analysis described above had suggested that R191 induced an upregulation of the ER stress response, and we therefore sought to confirm if this was the case. By using qRT-PCR, R191 was indeed found to induce increased expression of the 94 kDa glucose-regulated protein (GRP94), activating transcription factor 4 (ATF4), and the spliced variant of X-box binding protein 1 (XBP1; Supplementary Fig. S8A, upper panels). Moreover, message for CCAAT-enhancer-binding protein homologous protein (CHOP) and ER degradation enhancing alpha-mannosidase like protein 1 (EDEM) were enhanced as well (Supplementary Fig. S8A, lower panels). Because eukaryotic cells often induce autophagy to overcome ER stress (27), we also evaluated the impact of R191 on the autophagic machinery. Increased levels of microtubule-associated protein 1A/1B-light chain 3 form II were seen in BCWM.1, and to some extent in MWCL-1 cells after exposure to R191 (Supplementary Fig. S8B). In addition, levels of p62/Sequestosome 1, which can be degraded upon autophagy processing (28), were decreased by R191 (Supplementary Fig. S8B).

To test the activity of R191 against Waldenström's in vivo, immunodeficient mice were inoculated with BCWM.1 cells subcutaneously to provide a local model, or intravenously to provide a systemic model. Mice bearing subcutaneous xenografts received three R191 cycles starting on day 3 post-inoculation, whereas systemic xenograft-bearing mice received two cycles starting on day 7 post-inoculation. R191 significantly delayed growth of subcutaneous Waldenström's cells based on caliper measures of tumor size (Fig. 6A, left), and prolonged survival of these mice (Fig. 6A, right). Using whole-animal in vivo imaging to measure disease burden, R191 significantly delayed tumor growth in both the subcutaneous (Fig. 6B, left; Supplementary Fig. S9A) and systemic models (Fig. 6B, right; Supplementary Fig. S9B). This reduction was paralleled by a reduced level of circulating human IgM in the serum of mice with subcutaneous (Fig. 6C, left) and systemic disease (Fig. 6C, right). Notably, there were no significant changes in the weights of R191-treated mice throughout the in vivo experiments (Supplementary Fig. S8C). As had been the case in cell line studies, R191 reduced expression of CDK6, and enhanced the levels of cleaved caspase 3 in vivo (Supplementary Fig. S10). Finally, analysis of bone marrows from mice with systemic disease showed decreased levels of GFP⁺/CD20⁺ tumor cells in the R191-treated mice compared with vehicle-treated mice (Fig. 6D).

Earlier mechanistic insights into the activity of R191 suggested that combining it with other targeted therapies could in the future provide attractive options for the clinic. In that the expression of a dominant-negative Akt sensitized cells to R191, we evaluated the Akt inhibitor afuresertib (29). R191 and afuresertib together enhanced the reduction of viability in both Waldenström's cell lines (Supplementary Table S2), with combination indexes well below 1.0, indicating strong synergy. Because BTK is another target in the Myddosome signaling pathway, we tested the ability of ibrutinib to add to the efficacy of R191, and strongly synergistic effects were seen between the two in BCMW.1 cells (Supplementary Table S2). Because ER stress was induced by R191, we next combined it with bortezomib, another agent that works in part through inducing a stress response (30), and synergy was again seen in BCMW.1 cells (Supplementary Table S2). Finally, the alkylating agent bendamustine is one of the standards of care for initial therapy as part of a combination with rituximab for Waldenström's patients. We therefore combined bendamustine with R191, and found synergistic effects at lower bendamustine concentrations, though at higher ones there was an additive effect, with combination indices of approximately 1.0.

Genomic studies of Waldenström's macroglobulinemia have identified a number of commonly recurring mutations in this disease, which clinically shares some features of a low-grade B-cell lymphoma and a plasma cell dyscrasia. Among these are mutations of CXCR4, which are found in 30% to 40% of patients, of AT-rich interaction domain 1A (ARID1A), which occur in 17%, and of CD79B, which can be detected in 8% to 15% of patients (31). However, the most common, and essentially pathognomonic lesion is in MYD88, which is seen in 95% to 97% of patients. Moreover, the downstream effects of this mutation include formation of MYD88 homodimers and, in combination with IRAKs, the so-called Myddosome. This produces constitutive NF-κB activation, and provides downstream proliferative and survival signaling to these cells. Validation of the importance of this pathway's role in the pathobiology of this disease was provided by the strong clinical activity of the BTK inhibitor ibrutinib (17), which remains the only agent that has achieved regulatory approval for Waldenström's. Indeed, the overall response rate of 90.5% was so impressive that this drug was approved for both newly diagnosed and more advanced patients even though the study enrolled only those with relapsed or refractory disease. However, complete responses are rare, and treatment-emergent toxicities are not infrequent, including neutropenia in >20%, thrombocytopenia in almost 15%, and epistaxis or atrial fibrillation in up to 5%. These findings suggest that other therapeutic targets remain of relevance in this disease, and because MYD88^L265P activates NF-κB through both BTK and IRAK-1/4 (32), the latter could clearly be considered in this category.

Based on the rationale above, we evaluated the therapeutic potential of R191, a newly developed IRAK1/4 inhibitor, in preclinical Waldenström's models. R191 suppressed the viability of, and induced apoptosis in patient-derived cell lines and patient-derived primary samples while reducing activation of IRAKs 1 and 4 (Figs. 1 and 2). Moreover, R191 decreased levels of TRAF6, a crucial component of MYD88^L265P–induced IRAK-mediated signaling (33), and subsequent NF-κB activation (34) in cell lines (Figs. 2 and 3). Beyond that, Akt signaling is critical to the survival and homing of Waldenström's cells (26), and R191 suppressed activation of this pathway (Fig. 4), providing further support for its rational use against this disease.

Beyond showing activity against cells grown in culture, R191 delayed the growth of, and induced apoptosis in murine Waldenström's models in which tumor cells were grown either as a local, subcutaneous mass, or after systemic, intravenous injection (Fig. 6). Moreover, as would be hoped for from a drug with potential for clinical application, R191 reduced human IgM levels in the sera of tumor-bearing mice, and bone marrow involvement with Waldenström's cells (Fig. 6). In addition to its ability to induce apoptosis and reduce anti-apoptotic Akt signaling, this activity may have been in part due to R191-mediated reduction of c-MYC expression (Fig. 5), which correlated with cell-cycle arrest and decreased CDK4 and CDK6 levels (Fig. 1). Finally, R191 showed promising activity in combination with agents currently in use against Waldenström's (Supplementary Table S2), including afuresertib, bendamustine, and bortezomib, the latter possibly due to an impact on stress pathways (Supplementary Fig. S8). Calcium dysregulation, oxidative damage, and perturbations in posttranslational protein modifications can lead to accumulation of misfolded/unfolded proteins and ER stress. An unfolded protein response (UPR) is triggered to expand the processing capacity of the overwhelmed ER (35). Production of large quantities of IgM by Waldenström's cells puts a large burden on the protein processing machinery, and makes them susceptible to ER stressors. Indeed, proteasome inhibitors were shown previously to inhibit proliferation and to induce apoptosis in Waldenström's models (36), and are therefore used against this disease in patients. Interestingly, there was also strong synergy, as measured by combination indices of 0.251 to 0.496, with ibrutinib itself, supporting the possible use of dual BTK and IRAK1/4 inhibitor regimens. Members of the latter category, such as PF-06650833 (37), are currently in phase I and II trials for patients with inflammatory conditions such as rheumatoid arthritis (NCT02996500), suggesting they could be applied to Waldenström's as well.

Outside of Waldenström's macroglobulinemia, IRAK1/4 inhibitors could have other applications. For example, a Phe196Ser mutation of IRAK1 is a common, essential driver for Kaposi sarcoma herpesvirus lymphoma (38), and the MYD88^L265P mutation occurs in almost 30% of activated B-cell like diffuse large B-cell lymphomas (39). Also, IRAK1 activation and overexpression have been reported in myelodysplastic syndrome and acute myeloid leukemia (40), and may play a role in melanoma as well (41). Overall, as our data suggest that R191 effectively inhibited the BTK-IRAK1/4-NF-κB signaling axis in Waldenström's cells in vitro and in vivo, further studies may be warranted of this agent in these other malignancies as well.

R.Z. Orlowski is a consultant/advisory board member for BioTheryX, Bristol-Myers Squibb, Celgene, GSK Biologicals, Janssen Pharmaceutica, Juno Therapeutics, Kite Pharma, Legend Biotech USA, Molecular Partners, and Takeda Pharmaceuticals North America Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: H. Ni, F. Shirazi, H. Lin, Y. Hitoshi, S.P. Treon, R.Z. Orlowski

Development of methodology: H. Ni, H. Wang,

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Kuiatse, S.M. Ansell, S.K. Thomas, Z. Wang, R.E. Davis, R.Z. Orlowski

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Ni, V. Baladandayuthapani, H. Lin, R.J. Jones, Z. Wang, R.E. Davis, R.Z. Orlowski

Writing, review, and/or revision of the manuscript: F. Shirazi, H. Lin, I. Kuiatse, R.J. Jones, Z. Berkova, R.Z. Orlowski

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): I. Kuiatse, Z. Berkova, Y. Hitoshi, S.P. Treon

Study supervision: R.Z. Orlowski

Other (designed and performed most of the initial research): H. Ni

Other (consented patients for primary sample collection and provided helpful suggestions): H.C. Lee

R.Z. Orlowski, the Florence Maude Thomas Cancer Research Professor, would like to acknowledge support from the National Cancer Institute (R01 CA194264, R01 CA184464, P50 CA142509, U10 CA032102), and the Leukemia & Lymphoma Society (SCOR-12206-17). Core facilities that were key to these studies, including the Characterized Cell Line Core and the Flow Cytometry and Cellular Imaging Core, were supported by the MD Anderson Cancer Center Support Grant (P30 CA016672).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.