Essential role of TAK1 in regulating mantle cell lymphoma survival

TGF-β–activated kinase 1 (TAK1) is a serine/threonine kinase that was identified in 1995 as a member of the MAPK kinase family (MAP3K7).¹  TAK1 is activated by TGF-β and by a variety of cytokines, including TNF, IL-1, CD40 ligand, toll-like receptors, and T- and B-cell receptors.^2-4  On receptor activation, TAK1 binds to adaptor proteins and subsequently activates key downstream kinases such as IκK, p38 MAPK, and c-jun N-terminal kinase. In turn, this step leads to activation of NF-κB and activator protein–1 transcription factors that modulate the expression of a variety of inflammatory cytokines.⁵  Collectively, these cellular events enable TAK1 to play a key role in regulating inflammation, immunity, and cell death in a variety of cell types.^6,7 

Within the hematopoietic system, TAK1 plays an important role in promoting T-cell development and B-cell maturation, function, and survival.^4,6,8,9  B cell–specific deletion of TAK1 has been shown to markedly decrease marginal zone B cells in mice and to be associated with impaired B-cell proliferation and survival.⁴  TAK1-deficient B cells also failed to activate NF-κB and c-jun N-terminal kinase in response to B-cell receptor stimulation.⁴  However, the expression and function of TAK1 in lymphoid malignancies, especially those known to aberrantly express activated NF-κB, remain unclear.¹⁰  Here, we show that TAK1 and its active phosphorylated form (pTAK1) are abundantly expressed in a variety of primary and cultured lymphoma cells. Furthermore, inhibiting TAK1 via the use of siRNA or the small-molecule inhibitor AZ-TAK1 inactivated NF-κB, down-regulated p38, and activated the intrinsic caspase pathway, resulting in profound induction of apoptosis. Our data demonstrate a pivotal role for TAK1 in regulating lymphoma cell survival and suggest that it may be a therapeutic target for lymphoma.

The human Hodgkin and Reed-Sternberg–derived cell lines HD-LM2, L-428, and KM-H2 were obtained from the German Collection of Microorganisms and Cell Cultures, Department of Human and Animal Cell Cultures. All cell lines were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (Gibco BRL), 1% l-glutamine, and penicillin/streptomycin in a humid environment of 5% CO₂ at 37°C. MCL lines (Jeko-1, Mino, and SP53) and anaplastic large-cell lymphoma cell lines (Karpas-299, SUP-M2, and SU-DHL-1) were cultured in a similar way. The phenotypes and genotypes of these cell lines have been previously published.^11-16  Peripheral blood samples were obtained from 5 patients with MCL. Patient samples had been deposited in The University of Texas MD Anderson Cancer Center Lymphoma Blood Bank at the time of diagnosis or relapse. All patients and healthy volunteers had previously provided consent for donation of peripheral blood samples in accordance with The University of Texas MD Anderson Cancer Center Institutional Review Board guidelines.

The TAK-1 inhibitor (AZ-TAK1) was obtained from AstraZeneca. For Western blot experiments, antibodies to pTAK1 (Thr187), p38, phosphorylated p38, caspase 8, caspase 9 and cleaved caspase 3, AKT, pAKT (Ser473), pAKT (Thr308), ERK, pERK, p65 NF-κB, IκB, pIκB, SMAC/Diablo, cytochrome c, TRAF, X-linked inhibitor of apoptosis (XIAP), Syk, pSyk, Btk, and pBtk were purchased from Cell Signaling Technology. Antibody to XIAP was purchased from Santa Cruz Biotechnology. Antibody to β-actin was from Sigma-Aldrich.

Cells were cultured in 12-well plates at a concentration of 0.5 × 10⁶ cells/mL. Cell viability was assessed by use of the nonradioactive cell proliferation MTS [3-(4,5 dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl-2-(4-sulfophenyl)-2H-tetrazolium] assay with CellTiter96AQ_ueous One Solution Reagent (Promega), as previously published.¹⁷  To summarize, 80 μL of cell suspension with 20 μL of CellTiter96AQ_ueous One Solution Reagent was incubated in 96-well plates for 1 hour at 37°C and 5% CO₂, and formazan absorbance was measured at 490 nm on a μQuant plate reader equipped with Gen5 software (Biotek Instruments). Each measurement was made in triplicate, and the mean value was determined. Results represent the mean value of 3 independent experiments.

Total cellular proteins were extracted by incubation in lysis buffer (Cell Signaling Technology) for 40 minutes on ice and then centrifuged to remove cellular debris. The protein in the resulting supernatant was quantified. Then the protein was diluted 1:2 in protein sodium dodecyl sulfate loading buffer (Cell Signaling Technology) and heated to 95°C for 5 minutes. A total of 30 μg of protein was loaded onto 12% tris-HCl sodium dodecyl sulfate polyacrylamide electrophoresis Ready Gels (Bio-Rad), transferred to a nitrocellulose transfer membrane (Bio-Rad), and detected by the use of SuperSignalWest Dura Extended Duration Substrate (Pierce Chemical), as previously described.^18,19 

Proteins from cytoplasmic and nuclear cell fractions were extracted with NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific). To summarize, cells were incubated for 10 minutes on ice with the cytoplasmic extraction reagent I-II and then centrifuged. The proteins in the resulting supernatant represent the cytoplasmic extract. Nuclear extract was obtained by incubating cell pellets with Nuclear Extraction Reagent for 40 minutes. Cell suspensions were placed on ice and vortexed every 10 minutes for a total of 40 minutes. After centrifugation, the supernatants represent the nuclear extract. The proteins in the resulting cytoplasmic/nuclear extracts were quantified by with the bicinchoninic acid method (Pierce Chemical) according to the manufacturer's instructions.

Mitochondrial and cytosol fractions were isolated with the Mitochondria/Cytosol Fractioning Kit (BioVision). Cells were washed with ice-cold phosphate-buffered saline, resuspended in 1 mL of Cytosol Extraction Buffer Mix containing dithiothreitol and protease inhibitors, and incubated on ice for 10 minutes. Cells were homogenized by 30 passes with a Dounce tissue grinder. Cell homogenates were incubated on ice for 10 minutes and then centrifuged at 700g for 10 minutes at 4°C. Protein concentration of the resulting postnuclear fraction (supernatant) was collected, and cells were further centrifuged at 10 000g for 40 minutes at 4°C. The resulting supernatant was used as the cytosolic fraction and the pellet as the membrane fraction. The membrane fraction was resuspended and homogenized in 100 μL of Mitochondrial Extraction Buffer Mix to obtain the mitochondrial protein lysates.

Apoptosis was determined by Annexin V–FITC Apoptosis Detection Kit, performed according to the manufacturer's instructions (BD Pharmingen). Data were acquired a FACSCalibur flow cytometer (BD Biosciences). Results were obtained by analyzing data with FlowJo Version 7.6.1 software (TreeStar) Results represent the mean value of 3 independent experiments.

Mitochondrial membrane potential (ΔΨ) was evaluated by using the mitochondrial specific fluorescent probe 5,5′, 6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolycarbocyanine iodide (JC-1), performed according to the manufacturer's instructions. (Trevigen). Data were acquired by FACSCalibur flow cytometer (BD Biosciences). Results were obtained by analyzing data with FlowJo Version 7.6.1 software (TreeStar). Results represent the mean value of 3 independent experiments.

Three sets of double-stranded siRNA for TAK-1 and of silencer-negative siRNA (as a control) were purchased from Invitrogen. The sequences of the siRNAs are available on request. The MCL cell lines (Jeko-1, Mino, and SP53) were plated at 1 × 10⁶ cells/mL in 12-well plates. Double-stranded siRNAs (2μM) were transfected with the Nucleofector Kit (Lonza). For MCL cell line Jeko-1, we used Nucleofector Solution V and the transfection program X-001, for Mino and SP-53 we used Nucleofector Solution T and the transfection program T-001. Cells were harvested after 48 hours and subjected to Western blot analysis. This protocol gave a transfection efficiency of 70%-80%.

Immunohistochemical methods were performed by the use of routinely processed paraffin-embedded tissue specimens from tumors and cytospin smears from mantle cell lines (Mino, P53, and Jeko) as previously described.²⁰  The slides were incubated for 1.5 hours at room temperature with antitotal-TAK1 antibody (dilution 1:150; Abcam). All tissue sections underwent heat-induced epitope retrieval in pH 6.0 citrate buffer (Dako) followed by cooling-down period of 20 minutes. Endogenous peroxidase reaction was blocked by 3% H₂O₂ solution for 10 minutes. To avoid nonspecific binding of primary antibodies, serum-free blocking solution (Dako) was applied for 40 minutes at room temperature.

Detection was performed with the LSAB plus streptavidin HRP system (Dako). Immunofluorescence labeling was also performed with cytospin smears from mantle cell lymphoma (MCL) cell lines as well as paraffin-embedded tissue section specimens of malignant lymphomas as previously described.²¹  For the cytospin specimens, the slides were incubated for 1 hour at room temperature with an antiphospho (p)TAK1 antibody (dilution 1:200; Abcam). For the paraffin-embedded tissue sections, the slides were incubated overnight at 4°C using the same antibody (dilution 1:50). The tissue sections underwent heat-induced epitope retrieval in pH 6.0 citrate buffer (Dako) followed by a cooling-down period of 20 minutes. Alexa Fluor 594 (red) goat antibody IgG conjugate (Molecular Probes) was used as secondary antibody. The nuclei were counterstained with 4′-6′ diamidino-2-phenylindole.

Sections from an infiltrating ductal breast carcinoma and from a diffuse large B-cell lymphoma cell line, SUDHL2, were used as positive and negative controls, respectively. The Jurkatt cell line was also used as a positive control for pTAK. In addition, the baseline levels of expression of total and pTAK1 were assessed in a reactive lymph node. Protein expression levels for both, total-TAK1, and pTAK1 were scored as negative, weakly positive (+), and strongly positive (++), depending on the staining signal intensity in comparison with the positive control; if greater or equal, then expression was considered high and if lower, then expression was considered low.

An in vitro kinase assay was used to measure the ability of AZ-TAK1 to inhibit the kinase activity of TAK1 toward a recombinant protein substrate. The TAK1 kinase domain and TAB protein were coexpressed in baculovirus and purified by FLAG tag. The purified protein complex and biotinylated full-length kinase dead substrate (BT-MKK6kd) were then used in a 384-well automated ALPHA (Amplified Luminescent Proximity Homogeneous Assay) screen assay. Specifically, BT-MKK6kd substrate, TAKTAB enzyme complex, and reaction buffer were mixed and aliquoted into a 384-well plate containing a dilution of AZ-TAK1 and appropriate controls. ATP buffer was then added to initiate the reaction. After 1 hour, an alphascreen detection mixture, containing alphascreen acceptor and donor beads, antiphospho-MKK6 antibody, and EDTA, was added to stop the reaction. This allowed the streptavidin-coated donor beads to bind to the biotin-labeled MKK6kd substrate and the protein A coated acceptor beads to bind to the antiphospho-MKK6 antibody. When the antiphospho-MKK6 antibody bound to the phosphorylated MKK6kd substrate, the 2 beads were brought in close proximity of each other, resulting in the generation of a luminescent signal that was quantified using a Perkin-Elmer Envision plate reader. Data were analyzed to determine the mean IC₅s for AZ-TAK1.

Tak1 crystal structure was solved in house (Proteros Biostructures GmbH, unpublished work, August 10, 2008), and the structure was prepared by use of the Maestro 8.5 protein preparation wizard (Schrödinger); waters were deleted, bond orders assigned, and hydrogens added. The ligand was prepared with the use of default settings in LigPrep Version 2.5 (Schrödinger), and docking studies were performed with the program Glide.²² 

In summary, HeLa cells were plated in 96-well plates and after pretreatment for 1 hour with AZ-TAK1, cells were stimulated for 20 minutes with 2 ng/mL IL-1. The cells were then fixed with paraformaldehyde, thereafter permeabilized, and stained with the antibodies. Fluorescent immunolabeling was used to monitor changes in the nuclear translocation of NF-κB p65, phospho-cJun, and phospho-p38. NF-κB p65 was obtained from Santa Cruz Biotechnology, and phospho-p38 Ab, and phospho-cJun were obtained from Cell Signaling technologies. Alexa Fluor dye conjugated secondary Abs were obtained from Molecular Probes. Cellular screening, data acquisition, and data interpretation were conducted on the Array Scan HCS Reader (Cellomics).

MCL cell lines were incubated in RMPI with or without FBS for 6-24-48 hours, before supernatants were collected and examined for TNF-α production by ELISA (R&D Systems), according to the manufacturer's instructions and as previously published.¹⁷  Each experiment was performed in triplicate, and results represent the mean value from 3 different experiments.

A mass spectroscopy-based approach evaluating single nucleotide polymorphisms was used to detect the KRAS mutation. PCR and extension primers for KRAS were designed with Sequenom. Assay design. Chips were run in duplicate on a Sequenom Mass-Array MALDI-TOF MassArray system. Sequenom Typer Software and visual inspection were used to inspect mass spectra.²³ 

The 2-tailed Student t test was used to estimate the statistical significance of differences in results among the 3 experiments. Significance was set at P < .05.

First, we examined the pattern of TAK1 expression in a panel of lymphoid cell lines derived from Hodgkin lymphoma (HD-LM2, L428, and KM-H2), MCL (Jeko-1, Mino, and SP53), and anaplastic large-cell lymphoma (Karpas-299, SU-DHL1, and SUP-M2) and compared the results with those of PBMCs from healthy persons (Figure 1A). Cellular stress-induced activation of TAK1 involves phosphorylation of the activation loop when TAK1 is in a signaling complex with TAB1 and TAB2.²⁴  Notably, transautophosphorylation of Thr187 appears to be correlated with endogenous TAK1 kinase activity.²⁵ 

We therefore sought to examine the level of TAK1 phosphorylation by using an antibody directed against pThr187 (hereafter referred to as pTAK1). PBMCs from 3 healthy donors weakly expressed TAK1 but did not express the active phosphorylated form of pTAK1 (Thr187). In contrast, TAK1, pTAK1, and its binding partner (TRAF6) were expressed in all MCL and Hodgkin lymphoma cell lines but expressed to a lesser extent in the anaplastic large-cell lymphoma cells. The expression pattern and cellular localization of TAK1 and pTAK1 in the MCL cells were then examined by the use of fluorescence microscopy. Both TAK1 and pTAK1 were predominantly expressed in the cytoplasm and were observed in ∼ 90% of cells, but was also weakly expressed in some of the nuclei (Figure 1B).

To investigate the mechanism of TAK1 activation in MCL cell lines, we examined the levels of TAK1 and pTAK1 in normal culture system and under starved culture conditions. For these experiments, we incubated Jeko-1, Mino, and SP53 for 3-6-24 hours in RPMI with and without FBS 20%. Western blot results showed no difference in TAK1 and pTAK1 expression between starved and not starved cells. Both Jeko-1 and Mino cells secreted TNF-α in the supernatants, which may have contributed to the aberrant activation of TAK1 (supplemental Figure 3, available on the Blood Web site; see the Supplemental Materials link at the top of the online article).

TAK1 is essential for p38 and NF-κB activation and therefore plays a key role in regulating inflammation, immunity, and survival, but its function in lymphoid malignancies remains unknown.^3-7  We examined the molecular and functional consequences of TAK1 gene silencing in MCL cells, which are known to rely on activated NF-κB pathway to promote their survival. Using 3 different sets of double-stranded siRNAs,^1-3  we found that the use of siRNAs 2 and 3 effectively silenced the expression of TAK1 in all cell lines, and therefore we used siRNA3 in subsequent experiments (Figure 2A). TAK1 gene silencing resulted in inactivation of p38 and NF-κB, as indicated by the dephosphorylation of p38 and IκB. In addition, TAK1 silencing slightly reduced ERK phosphorylation, especially in the Jeko-1 cells (Figure 2B). These molecular changes were associated with induction of apoptosis in up to 71% of the lymphoma cells within 48 hours (Figure 2C-D). Collectively, these data demonstrate that TAK1 plays a critical role in regulating lymphoma cell survival and suggest that TAK1 may serve as a therapeutic target in lymphoma.

AZ-TAK1 was discovered after a small molecule-based lead identification program at AstraZeneca. Modeling studies of AZ-TAK1 suggest that the compound binds to the hinge region of the ATP binding site of Tak1 via correlated hydrogen bonds with the hinge Ala81. A crystal structure of similar compounds in Tak1 (Proteros Biostructures GmbH, unpublished work, August 10, 2008) confirm the predicted binding mode of AZ-TAK1 (Figure 3A). The inhibitory activity of AZ-TAK1 against TAK1 (MAP3K7) was evaluated by the use of a complex of full-length recombinant human TAK1 and its activating partner TAB1 and full-length kinase dead MKK6 as a substrate. Using this assay at Km ATP (10μM), we found that the IC₅₀ for AZ-TAK1 was 8.0 ± 0.05 nmol/L (Figure 3B). Enzymatic reactions performed with 2000μM of ATP resulted in an IC₅₀ increase to 177 ± 0.05 nmol/L, indicating that AZ-TAK1 binds competitively with ATP (supplemental Figure 1). The selectivity of AZ-TAK1 was evaluated in a panel of 30 kinases representing the major kinase families (Figure 3C). The most sensitive kinases in addition to TAK1 were CMGC-family kinases HIPK2, CDK9, and GSK3β, with IC₅₀ values estimated as 3 nmol/L, 9 nmol/L, and 19 nmol/L, respectively (Figure 3C).

Using an IL-1–induced HeLa cell line model, we found that AZ-TAK1 inhibited the phosphorylation of p38 and c-Jun and prevented the nuclear translocation of p65, a hallmark of NF-κB activation, within 1 hour of incubation (supplemental Figure 1A-B). Moreover, we performed additional experiments to evaluate the effect of AZ-TAK1 and TAK-1 silencing SiRNA on SYK and BTK (supplemental Figure 2A-B). AZ-TAK1 decreased the phosphorylation of Syk and BTK in the 3 MCL cell lines (supplemental Figure 2A). Furthermore, TAK1 silencing by siRNA also was associated with a decrease in Syk phosphorylation and BTK phosphorylation, which were more prominent in the SP53 cells (supplemental Figure 2B). These data may indeed suggest that greater concentrations of AZ-TAK1 may also inhibit Syk, which may have contributed to the efficacy of the small molecule. However, siRNA silencing of TAK1 also was associated with a decrease in pSyk and pBTK, which may suggest that TAK1 inhibition may lead to inhibition of BCR signaling by modulating the expression of cytokines and growth factors that can induce Syk and BTK phosphorylation.

We examined the biologic activity of AZ-TAK1 in a panel of 9 lymphoma cell lines and PBMCs isolated from 3 healthy donors. AZ-TAK1 potently inhibited lymphoma cell line proliferation and induced apoptosis in a dose- and time-dependent manner. In 8 of the 9 cell lines, the IC₅₀ of AZ-TAK1 was < 0.25μM. In contrast, AZ-TAK1 had no significant effect on PBMC survival (Figure 4A-B). The antiproliferative effect of AZ-TAK1 was mainly a result of the induction of apoptosis, as determined by dual staining with Annexin V and propidium iodide and FACS analysis (Figure 4C-D). After 48 hours of incubation, 0.1μM AZ-TAK1 induced apoptosis in 28%, 24%, and 74% of Mino, SP53, and Jeko cells, respectively. The induction of apoptosis increased to 42%, 46%, and 86% when cells were incubated with 0.5μM concentration. Moreover, the antiproliferative activity of AZ-TAK1 was not associated with a significant effect on cell-cycle fractions (Figure 4E).

At the molecular level, AZ-TAK1 induced similar changes to those observed with TAK1 silencing experiments. As shown in Figure 5A, AZ-TAK1 inhibited the phosphorylation of TAK1, p38, and IκBα (also shown in supplemental Figure 4) in addition to decreasing the nuclear content of p65, indicative of NF-κB inactivation. AZ-TAK1 activated the intrinsic caspase pathway by down-regulating XIAP protein expression and cleavage of caspases 9 and 3 (Figure 5B). Furthermore, the caspase 9 inhibitor Z-LEHD-FMK either partially (Jeko-1, Mino) or completely (SP53) inhibited AZ-TAK1–induced cell death (Figure 5C). Moreover, AZ-TAK1–mediated activation of caspase 9 was associated with the release of cytochrome c from the mitochondria into the cytosol, which was augmented by the release of SMAC/Diablo from the mitochondria (Figure 5D). Finally, we examined the effect of AZ-TAK1 on mitochondria membrane potential by using the fluorescent membrane-permeant JC-1 dye staining method and flow cytometric analysis (Figure 5D). AZ-TAK1 increased mitochondrial depolarization, indicated by a decrease in the red/green fluorescence intensity ratio. Cells with depolarized mitochondria accounted for 75% of total cell number when treated with AZ-TAK1 compared with control (8%; Figure 5D). Collectively, these data demonstrate that AZ-TAK1 induces cell death in lymphoma cells by releasing cytochrome c and SMAC/Diablo from the mitochondria, down-regulating XIAP, and activation of caspase 9 and 3.

To determine the biologic and molecular effects of AZ-TAK1 in primary lymphoma cells, we examined its activity in 5 MCL specimens isolated from patients with high numbers of circulating lymphoma cells in the peripheral blood. TAK1 and pTAK1 were found to be abundantly expressed by Western blot in primary lymphoma cells (Figure 6A). Using immunohistochemistry methods, we found that TAK1 expression was highly expressed in 6 of 13 (46%) lymph node sections involved with MCL, in 12 of 12 (100%) Hodgkin lymphoma lymph node sections, in 15 of 16 (90%) of diffuse large B-cell lymphoma lymph nodes sections, and in 8 of 8 (100%) of ALK-positive anaplastic large cell lymphoma (Figure 6B-C). Moreover, TAK1 was weakly expressed in 6 additional cases. Similar to what we have observed in the cell line studies, AZ-TAK1 (300nM) inhibited TAK1 phosphorylation, down-regulated XIAP, and cleaved caspase 3 (Figure 6A). These molecular changes were associated induction of cell death in more than 50% of primary lymphoma cells within 24 hours of incubation (Figure 6D)

In this study, we demonstrated that TAK1 and its active phosphorylated protein are abundantly expressed in a variety of primary and cultured lymphoma cells. We also demonstrated that silencing of TAK1 expression or inhibiting its function in lymphoma cells resulted in inactivation of NF-κB and p38 signaling, in addition to activation of the intrinsic caspase pathway, resulting in lymphoma cell death. Our data suggest that TAK1 may serve as a therapeutic target in lymphoma. Although TAK1 activation mutations have been observed at a very low frequency in lymphoma,²⁶  the underlying mechanisms for TAK1 activation in the majority of lymphomas remain unknown. In these cases, it is possible that TAK1 activation is related to dysregulated cytokine and growth factor stimulation in the lymphoma microenvironment.

In this study, we demonstrated that both cultured and primary MCL cells were sensitive to TAK1 inhibition. Inhibition of TAK1 was associated with the release of cytochrome c from the mitochondria, an important initial step for activating caspase 9. This process was further amplified by the release of SMAC/Diablo and the down-regulation of XIAP in both MCL cell lines and in primary MCL cells.²⁷  XIAP, one of the most potent cellular inhibitors of apoptosis, is known to be overexpressed in a variety of lymphomas.^28-30  XIAP activity is inhibited by SMAC/Diablo, which is released from the mitochondria.³¹  Because inhibiting XIAP function or down-regulating its expression can sensitize cancer cells to chemotherapy, the development of pharmacologic XIAP inhibitors, such as Smac mimetics, is currently under investigation.³²  However, Smac mimetics are not sufficient to induce cell death because they work best in combination with other apoptosis-inducing agents.³³  In this study, AZ-TAK1 induced apoptosis primarily by releasing cytochrome c, which was amplified by the down-regulation of XIAP and inhibition of its function by releasing SMAC/Diablo. Although, like many other ATP-competitive kinase inhibitors, it cannot be ruled out that AZ-TAK1 exerts some of these effects through inhibition of other kinases, the combined evidence from the siRNA and small molecule data are compelling. Indeed, this triple effect suggests that AZ-TAK1 has potent single-agent activity against lymphoma and may not require the addition of other anticancer agents.

The ability of TAK1 to regulate several survival mechanisms—such as p38, NF-κB, and XIAP—makes it an appealing target for cancer therapy. Nonetheless, pharmacologic inhibition of TAK1 should be explored with caution, given that targeting TAK1 may also result in excessive toxicity to normal tissues. Although we found that AZ-TAK1 was less toxic to PBMCs than to lymphoma cell lines, knockout experiments of TAK1 resulted in massive hepatic and BM toxicity in mice.³⁴  This observation may not preclude the development of TAK1-targeted therapy. For example, although knockout experiments of heat shock protein 90 (HSP90) also proved to be lethal in mice, the authors of recent clinical trials using pharmacologic inhibitors of HSP90 have demonstrated the safety and promising clinical activity of targeting HSP90 in humans.^35-37  Furthermore, the potential toxicity to normal tissue can be reduced by exploring novel targeted delivery approaches of TAK1 inhibitors to tumor cells with the use of nanoparticles or antibody-drug conjugates.

In conclusion, our data demonstrate a critical role for TAK1 in promoting MCL cell survival. Future studies should aim at examining the expression pattern and function of TAK1 in other types of lymphoma and should identify factors that are responsible for TAK1 overexpression and activation of TAK1 in lymphoma. These studies can facilitate the development of TAK1-targeted therapy for the treatment of lymphoma.

The authors thank Virginia M. Mohlere (Department of Quantitive Science Administration, The University of Texas MD Anderson Cancer Center, Houston) for editorial assistance.

This work was supported in part by The University of Texas MD Anderson Cancer Center Support Grant CA016672, the Living Legend Fund, and The Robert and Vicki Smith Fund (to A.Y.).

Contribution: D.B. conceived the study, performed experiments, analyzed data, prepared the figures, and wrote the paper; S.P., K.B., and S.S.N. provided valuable reagents, shared data, evaluated the results, and approved the final manuscript; F.V. performed experiments and analyzed data; D.T. and J.S. designed and made the AZ-TAK1 compound and modeled binding mode in Tak1 ATP binding site; and A.Y. conceived the study, helped design experiments, evaluated the results, and approved the final manuscript.

Conflict-of-interest disclosure: S.P., K.B., D.T., and J.S. work for Astra-Zeneca. The remaining authors declare no competing financial interests.

Correspondence: Anas Younes, MD, Department of Lymphoma and Myeloma, Unit 0429, The University of Texas MD Anderson Cancer Center,1515 Holcombe Blvd, Houston, TX 77030; e-mail: ayounes@mdanderson.org.