Extrinsic Apoptosis Is Impeded by Direct Binding of the APL Fusion Protein NPM-RAR to TRADD Free

Acute promyelocytic leukemia (APL) is a malignant proliferation of differentiation-competent myeloblasts and promyelocytes (1). In the vast majority of cases, APL is characterized by t(15;17)(q22;q21), which introduces the gene for the retinoic acid receptor α (RARA) into the locus encoding the PML protein. The resultant PML-RAR fusion encodes the N-terminal protein-interaction and leucine-zipper domains of PML fused to the DNA-binding zinc finger, leucine-rich dimerization, and C-terminal ligand-binding and coactivator/corepressor domains of RARA (1). Forced expression of PML-RAR in mice results in an APL-like phenotype (2–4).

The molecular basis by which PML-RAR disrupts normal myeloid development is complex (1). PML-RAR, having greater affinity for corepressors than RARA, is capable of binding to retinoic acid responsive promoters, and suppressing transcription of retinoic acid target genes. PML-RAR also has unique DNA-binding properties, and may act as a rogue transcriptional activator or repressor. PML-RAR may affect transcription pathways indirectly, through its ability to bind with and sequester RXR, a key binding partner for many members of the nuclear hormone receptor family. PML itself localizes to nuclear structures known as PML oncogenic domains, and within these nuclear bodies, PML interacts with a diverse series of proteins, including DAXX, p53, Rb, CREB-binding protein, SKI, MYB, MDM2, and SUMO. By virtue of its ability to delocalize PML and disrupt the structure of PML-containing nuclear bodies, PML-RAR may affect a wide variety of cellular functions contributing to apoptosis, cellular senescence, and cell-cycle regulation. In addition, PML-RAR, through recruitment of corepressor containing histone deacetylase activity to PML-containing complexes, may also affect the acetylation and function of proteins that bind to PML, as has been shown for p53 (5).

We have been investigating the rare cases of APL that do not express the PML-RAR fusion. These leukemias manifest a similar phenotype, but a different genotype, and as such represent “experiments of nature” with which to test mechanistic hypotheses (6). Nine variant translocations have been characterized on a molecular basis, and all express fusion proteins containing the same C-terminal sequences of RARA as are expressed in PML-RAR: t(11;17)q(23;q21), which fuses the PLZF transcriptional repressor to RARA (7); t(5;17)(q35;q21), which joins nucleophosmin (NPM) to RARA (8); t(11;17)(q13;q21), which fuses the nuclear matrix protein NUMA to RARA (9); der17, which fuses STAT5b with RARA (10); fusion of RARA with the regulatory subunit of the cyclic adenosine monophosphate-dependent protein kinase PRKAR1A on 17q24 (11); t(4;17), which fuses FIP1L1 to RARA (12); t(X;17)(p11;q21), which fuses the BCL-6 corepressor protein BCOR to RARA (13); der (2)t(2;17)(q32;q21) which fuses the singe-stranded DNA binding protein OBFC2A to RARA (14); and t(3;17)(q26;q21) which joins the F-box/WD-40 protein TBLR1 to RARA (15). We have focused our studies on t(5;17), which, after PLZF-RAR, is the second most common of the variants and manifests a similar phenotype to t(15;17) APL, including the ability of t(5;17) blasts to differentiate in the presence of all-trans retinoic acid (16).

The t(5;17) translocation fuses the same C-terminal sequences of RARA expressed in PML-RAR to the N-terminal 117 amino acids of NPM (8). We have shown in both in vitro and in vivo models that, like PML-RAR, ectopic expression of NPM-RAR induces an APL-like phenotype (17, 18). We have previously shown that NPM-RAR localizes throughout the nucleoplasm (19) and interacts with coactivator and corepressor molecules (20). NPM-RAR binds to DNA both as homodimers and heterodimers with RXR, and similar to PML-RAR, its activity as a transcriptional activator varies by cell type and target sequence (20).

NPM (also known as B23, numatrin, and NO38) is a molecular chaperone originally identified through its ability to bind and transport nascent ribosomal units in the nucleolus (21). NPM shuttles between nucleolus, nucleus, and cytoplasm, and participates in a myriad of cellular functions. NPM binds to centrosomes: phosphorylation by CDK2/cyclin E dissociates NPM from centrosomes and initiates centrosomal duplication. NPM is redistributed from the nucleolus in response to cytotoxic drugs and genotoxic stress. NPM binds several cellular and viral proteins, including human immunodeficiency virus Rev and Tat proteins, hepatitis delta virus delta antigens, retinoblastoma protein, p53, MDM2, ARF, YY1, and IRF-1 (for review, see Colombo and colleagues; ref. 21).

All of the APL variant fusion proteins, including NPM-RAR, contain dimerization domains within the sequences of the N-terminal fusion partner (6), and it has been proposed that dimerization of RARA alone might be sufficient for development of APL (22–24). However, Sternsdorf and colleagues (25) have shown that transgenic mice expressing a construct containing artificial dimerization domains fused to RARA fail to develop the full APL phenotype, suggesting that the N-terminal fusion partners provide an important gain of function, in addition to their ability to induce dimerization.

Limited information is available about the functionality of the N-terminal 117 amino acids of NPM expressed in NPM-RAR. This region contains two nuclear export signals, as well as elements important for dimerization (21). In this manuscript, we test the hypothesis that the N-terminal sequences of NPM-RAR express a protein-interaction domain that contributes to NPM-RAR-leukemogenesis. Through a proteomic analysis of NPM-RAR–interacting proteins, we identified the TNF type I–associated DEATH domain protein TRADD as specifically binding to NPM-RAR.

HEK293, HeLa, and U937 cells were obtained from the American Type Culture Collection. These cells were not further authenticated in our laboratory. HEK293 and HeLa cells were maintained in a humidified 5% CO₂ atmosphere in DMEM (Mediatech) supplemented with 10% FBS (GIBCO), 2 mmol/L glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin. U937 cells were grown in RPMI-1640 (Mediatech) containing the same supplements. HEK293 or HeLa cells were transfected by CaPO4 coprecipitation, as described (20). Pools of stably transfected cells were selected in medium containing 1 mg/mL G418 (Gibco). The generation and characterization of the stably transfected U937 cells have been previously described (26).

Rabbit polyclonal anti-RARA and anti-TRADD antibodies were purchased from Santa Cruz Biotechnology. Mouse monoclonal anti-NPM antibody was obtained from Invitrogen. Anti-PARP was obtained from Cell Signaling Technology, and anti–caspase-3 was from Assay Designs. Anti–β-actin and anti-FLAG mouse monoclonal antibodies were purchased from Sigma.

pCDNA vector containing coding regions for the calmodulin-binding peptide and the IgG-binding domains of protein A was a gift of R. Steinman (University of Pittsburgh). NPM-RAR or RARA was subloned in the appropriate reading frame to generate C-terminal TAP-tag fusions. Pools of clones of HEK293 cells stably expressing the TAP-tagged NPM-RAR, RAR proteins, and control were washed two times with ice-cold PBS and then lysed in 10 mL of ice cold NP-40 lysis buffer containing 20 mmol/L Tris, pH 8.0, 137 mmol/L NaCl, 10% Glycerol, 1% (v/v) NP-40, 10 mmol/L NaF, 5 mmol/L NaVO₄, and protease inhibitor cocktail (Sigma). After incubation on ice for 30 minutes, the lysates were centrifuged at 14,000 rpm for 15 minutes at 4°C. The resulting supernatants were incubated with 100 μL of packed IgG-Sepharose (GE Healthcare) for 4 hours at 4°C. After extensively washing with ice-cold lysis buffer containing 500 mmol/L KCl, proteins were eluted with PBS containing 1% SDS. The eluate was concentrated and desalted using Amicon Microcon (Millipore) centrifugal filter devices (3,000 MWCO), and the proteins were resolved by 1D SDS-PAGE. Bands were visualized by Coomassie staining and the entire lane was excised into ten gel bands and subjected to in-gel trypsin digestion as described previously (27). Tryptic digests were analyzed in duplicate by nanoflow reversed-phase LC/MS-MS using a nanoflow LC (Dionex Ultimate 3000, Dionex Corporation) coupled online to a linear ion trap MS (LTQ-XL, ThermoFisher Scientific), and peptide/protein identifications were derived as described previously (27).

The coding region of TRADD or NPM-RAR was subcloned into pcDNA3/V5-His (Invitrogen) in the appropriate reading frame to generate fusion proteins bearing a V5-His tag at the C terminus. Fidelity was confirmed by DNA sequencing and Western blotting.

To test for binding between the V5-His-tagged TRADD and NPM-RAR, transiently transfected HEK293 cells were lysed with NP-40 lysis buffer containing 10 mmol/L imidazole. Complexes were captured on Ni-NTA beads (Qiagen). The beads were washed with NP-40 buffer containing 20 mmol/L imidazole before elution with 2× SDS sample buffer. Samples were analyzed by immunoblotting with anti-RARA or anti-TRADD antibodies. To determine association between FLAG-tagged NPM-RAR and TRADD, transiently transfected cells were lysed in NP40 lysis bufer and incubated with the M2 anti-FLAG antibody. Complexes were captured on protein A sepharose beads, and separated on SDS-PAGE. Immunoblot analyses were prepared and probed as described (20).

After treatment with cyclohexamide (Sigma) and human recombinant TNFα (R&D Systems), cells were washed with ice-cold PBS and resuspended in NP-40 lysis buffer. After 30-minute incubation on ice, lysates were cleared by centrifugation. 50 μg of lysate were resolved by 8% or 12% SDS-PAGE and transferred onto Nitrocellulose membranes (Bio-Rad). After blocking with 5% non-fat milk, membranes were incubated with primary antibodies as recommended by the supplier. The immune complexes were detected by the ECL detection system (Invitrogen).

Cells were exposed to 10 ng/mL TNFα (R&D System) and 1 μg/mL of cycloheximide (Sigma) for one hour. Annexin V staining was determined using an FITC-conjugated Annexin V antibody and propidium iodide (PI) according to the instructions provided by the manufacturer (BD Bioscience), using an Accuri C6 flow cytometer (Accuri Cytometers), gating on the Annexin V postive/PI negative population. The population of apoptotic cells with sub-G₀/G₁ DNA content was determined by PI staining of cells fixed in ice-cold 70% ethanol before staining with 1 mg/mL PI and analysis by flow cytometry (Accuri C6 flow cytometer, Accuri Cytometers).

U937 and U937-transfected vector only or NPM-RAR–expressing cells were seeded into 96-well plates at a density of 1.5 × 10⁴ cells per well. After 24 hours, cells were treated with 10 ng/mL TNFα and 1 μg/mL cycloheximide. Activity of caspase-8 was measured using caspase-Glo 8 detection assay kit (Promega). Luminescence was measured using the 1420 VICTOR3 micro-plate reader (PerkinElmer).

Hela cells were grown on glass cover-slips and transiently transfected with FLAG-NPM-RAR and V5/His-TRADD. Cells were fixed with 2% para-formaldehyde in PBS for 15 minutes at room temperature, permeabilized with 0.1% Triton X-100 in PBS for 15 minutes, and nonspecific binding sites were blocked with PBS containing 2% BSA for 45 minutes. After blocking, cells were incubated with a 1:200 dilution of mouse monoclonal anti-V5 antibody (Invitrogen) and/or a 1:500 dilution of rabbit polyclonal anti-FLAG antibody (Sigma). After washing with PBS containing 0.5% BSA, cells were incubated for 1 hour with a 1:1,000 dilution of Alexa 488–conjugated anti-rabbit antibody and 1:500 dilution of Alexa 546–conjugated anti-mouse antibody (Invitrogen). Nuclei were stained with 1:5,000 dilution of DRAQ5 (Cell Signaling Technology). Confocal images were captured using a Leica TCS SL microscope (Leica).

To identify binding partners of NPM-RAR, we engineered constructs expressing NPM-RAR or RARA fused at the C-terminus to sequences encoding the calmodulin-binding domain and protein A. These were introduced into HEK293 cells, and transfected cells were lysed. Interacting proteins were captured on Ig-coated Sepharose beads. The beads were washed extensively with high stringency buffers (500 mmol/L KCl) to minimize nonspecific protein interactions. The eluted proteins were size separated on one-dimensional polyacrylamide gels. Bands were excised from the gel, eluted, and subjected to tryptic digestion and LC/MS-MS. Amongst proteins uniquely interacting with NPM-RAR, but not RARA, we identified TRADD. The specificity of the interaction was validated by coprecipitation of transiently transfected HEK cell lysates, using either TRADD as the target and NPM-RAR as the bait (Fig. 1A), or alternatively NPM-RAR as the target and TRADD as the bait (Fig. 1B).

To confirm that TRADD and NPM-RAR interact in cells, we transiently cotransfected HeLa cells with epitope-tagged expression plasmids for NPM-RAR and TRADD. We found that NPM-RAR localizes primarily to the nucleoplasm with a small component expressed in the cytoplasm. TRADD colocalized with NPM-RAR in the cytoplasm (Fig. 2).

NPM-RAR contains N-terminal sequences implicated in NPM dimerization. We have previously determined that NPM-RAR can form NPM-RAR/NPM-RAR homodimers as well as NPM-RAR/NPM heterodimers (20). TRADD has not been previously reported as interacting with NPM. To test whether NPM was present in the NPM-RAR/TRADD complex, we probed immunoblot analyses of TRADD-precipitated proteins with an antibody that recognizes the C-terminal sequences of NPM. We did not find signal for wild-type NPM in the complex (Fig. 1), indicating that the ability to bind NPM is a novel attribute of the NPM-RAR fusion protein.

TRADD is a death-domain containing adapter molecule that interacts with the TNFR1. TRADD plays an integral part in activation of the extrinsic apoptotic pathway. To determine whether NPM-RAR-expression affected TNF-induced apoptosis, we tested the TNF sensitivity of U937 cells stably transfected with NPM-RAR. The engineering and characterization of these clones are described elsewhere (26). Independent U937-NPM-RAR clones, or as controls parental U937 cells or clones stably transfected with vector alone, were treated with TNF and assayed by PI-exclusion and Annexin V staining. This assay was performed three times with similar results: averaging all three experiments, 30.8% (±5.3%) of parental U937 cells and 48.6% (±3.9%) of a vector-control clone, but only 18.1% (±7.6%) and 22% (±6.1%) of two independent NPM-RAR–expressing clones became Annexin V positive at 60 minutes, indicating a muted response of NPM-RAR–expressing cells to TNF-induced apoptosis (Fig. 3A). To confirm these results, we also analyzed cells for fragmentation of nuclei, as assessed by the percentage of cells staining with PI with sub-G₀/G₁ DNA content (Fig. 3B). Flow cytometry of PI stained cells also indicated a muted apoptotic response of the NPM-RAR–expressing clones, with apoptosis occurring in 18.7% of parental U937 cells 120 minutes after exposure to TNFα; 35.9% in the vector-transfected clone, but only 8.6% and 8.5% in the two NPM-RAR–expressing U937 clones. Repetition of the assay generated similar results (see Supplementary Fig. S1).

TNF activation of the TNFR leads to TRADD-dependent cleavage and activation of the initiator procapsase-8. To assess the effect of NPM-RAR on this pathway, we treated NPM-RAR or control U937 cells with TNF, and harvested cell lysates for immunoblot analyses. Figure 4 indicates retention of non-cleaved procaspase-8 in NPM-RAR–expressing cells, compared with parental U937 cells or a control U937 clone.

Activation of the initiator caspase-8 leads to cleavage and activation of the effector caspase-3. If NPM-RAR inhibits activation of caspase-8, we would also predict decreased activation of caspase-3 upon exposure to TNF in NPM-RAR–expressing cells. Figure 5 indicates lack of activation of procaspase-3 upon stimulation with TNF in the NPM-RAR–expressing cells.

Activation of the effector caspase-3 initiates a series of late apoptotic events, including cleavage and inactivation of PARP. If indeed NPM-RAR inhibited TNF-mediated apoptosis, we would expect retention of the full-length form of PARP in NPM-RAR–expressing cells upon stimulation with TNF. Figure 5 indicates the refractoriness of NPM-RAR cells to TNF-induced cleavage of PARP.

Though the molecular pathway has not been fully characterized, it has been observed that one of the events associated with TNF-mediated apoptosis is accumulation of p53 protein (28). To assess the effects of NPM-RAR on TNF-induced accumulation of p53, we prepared immunoblot analyses of lysates of the U937 clones stimulated with TNF, and probed them with an anti-p53 antibody. Figure 6 shows increased accumulation of p53 in control clones treated with TNF, consistent with prior reports. NPM-RAR–expressing clones failed to demonstrate TNF-induced upregulation of p53.

We performed proteomic analysis to identify proteins that bind to NPM-RAR. We identified TRADD as a binding partner of NPM-RAR. This interaction was validated by reciprocal coprecipitation and immunofluorescent colocalization. TRADD is an adapter protein that interacts with the TNFR1 receptor, and mediates the activation of programmed cell death and NF-κB activation. We found that cells that expressed NPM-RAR were less sensitive to TNF induction of apoptosis.

The N-terminal 117 amino acids of NPM that are represented in NPM-RAR contain a putative protein interaction domain. Within the N-terminal 117 amino acids of NPM are two nuclear export signals that interact with the nuclear exportin CRM (21). Indeed, the N-terminal region of NPM has been implicated in oligomerization (29), and we have shown that NPM-RAR binds un-rearranged NPM (30). Our data suggest direct association between NPM-RAR and TRADD. We neither found evidence for an interaction between TRADD and RARA in our proteomic screen of RARA-interacting proteins, nor evidence for direct interaction between wild-type NPM and TRADD. This leads us to propose that TRADD interaction is unique to the fusion protein, possibly due to a modified configuration of the N-terminal NPM domain that occurs as a result of fusion with RARA.

NPM-ALK in t(2;5) anaplastic lymphoma is derived from the same breakpoint in the NPM locus, and expresses the same domains of NPM as in NPM-RAR (31). The N-terminus of NPM is also contained in the NPM-MLF1 fusion of t(3;5)(q25;q34) that is expressed in rare cases of myelodysplastic syndrome and acute myeloid leukemia (32); in t(3;5) the breakpoint involves a different intron, to generate expression of the 175 N-terminal amino acids of NPM. The same N-terminal sequences are also expressed in NPMc mutations, found in a majority of cases of acute myelogenous leukemia (AML) with normal karyotype, in which a variety of frameshift mutations lead to an altered C-terminus (33). It is intriguing to speculate whether these other oncoproteins also bind to TRADD.

Several mechanisms may explain the observation that NPM-RAR binding to TRADD inhibits downstream proapoptotic actions of TRADD. TRADD is localized both in the cytoplasm, where it can interact with the TNFR and other cytoplasmic proteins through which it activates the extrinsic apoptotic pathway, as well as in the nucleus, where it apparently mediates apoptosis through a poorly defined pathway that does not depend on activation of caspase-8 (34, 35). We have previously reported that NPM-RAR localizes primarily throughout the nucleoplasm, with a small component in the cytoplasm (19). PML-RAR has also been observed to localize to both the nucleus and cytoplasm (36). Our immunofluorescence data indicate that NPM-RAR and TRADD colocalize primarily within the cytoplasm. Thus, we propose that NPM-RAR interaction in the cytoplasm blocks TRADD from interacting with FADD and caspase-8 to participate in TNF-mediated apoptosis. Our model is reminiscent of that proposed for the tumorigenic properties of the Epstein-Barr Virus latent membrane protein 1 (LMP1), which similarly binds to TRADD to inhibit apoptotic signaling (37, 38).

The intertwining of the extrinsic and intrinsic apoptotic pathways is complex. Several reports have indicated that TNF activation, through the TNFR1 receptor and the adapter protein TRADD, leads to increased expression of p53, possibly through modulation of p19ARF (28). We found that NPM-RAR blocked TNF-mediated P53 expression as well. We have previously shown that NPM-RAR modulates the activity of p53 in a direct fashion (manuscript in preparation); the mechanism by which NPM-RAR affects apoptotic pathways could thus be diverse.

The significance of the TNF pathway in AML has not been studied in depth. Nevertheless, Sawanobori has reported that TNFR1 is expressed on the majority of hematopoietic precursors, and Rushworth has suggested that altered heme-oxidase–mediated pathways in leukemic cells could quell the proapoptotic effects of TNFR1 (39, 40). The recent finding that TRADD has tumor suppressor functions (28) leads us to speculate that it may have an important role in leukemogenesis. NPM-RAR and LMP1 (37, 38) are the only oncogenic proteins yet identified as directly interacting with and inhibiting the function of TRADD. Our work suggests a novel pathway through which NPM-RAR modulates TNF signaling in the t(5;17) variant.

The majority of APL cases express PML-RAR, not NPM-RAR. Testa and colleagues have shown that PML-RAR inhibits TNF-induced apoptosis, though through a different mechanism of downregulation of expression of the TNFR (41). Witcher and colleagues have shown that TNF enhances ATRA-induced differentiation in APL cells (42, 43). Nuclear TRADD colocalizes to PML-nuclear bodies (44), and PML is necessary for nuclear TRADD-mediated apoptosis. PML-nuclear bodies have been shown to regulate the availability of the apoptotic adapter protein FLASH (FLICE-associated Huge Protein; ref. 45), as well as the proapoptotic protein DAXX (46, 47). Inhibition of the function of wild-type PML by PML-RAR, and the disruption of PML-nuclear bodies, most likely affect these mediators of the extrinsic pathway. Thus, interference with the extrinsic apoptotic pathway may be a common feature by which APL fusion proteins contribute to leukemogenesis.

No potential conflicts of interest were disclosed.

Conception and design: R.L. Redner

Development of methodology: A. Chattopadhyay, R.L. Redner

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Chattopadhyay, B.L. Hood, T.P. Conrads

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Chattopadhyay, B.L. Hood, T.P. Conrads, R.L. Redner

Writing, review, and/or revision of the manuscript: A. Chattopadhyay, T.P. Conrads, R.L. Redner

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Chattopadhyay, R.L. Redner

Study supervision: R.L. Redner

The authors thank Richard Steinman and Daniel Johnson for valuable discussions and sharing of reagents.

This work was supported by NIH R01 CA67346 and P30 CA047904.

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.