BRCA1-associated protein 1 (BAP1) is a nuclear deubiquitinating enzyme that is associated with multiprotein complexes that regulate key cellular pathways, including cell cycle, cellular differentiation, cell death, and the DNA damage response. In this study, we found that the reduced expression of BAP1 pro6motes the survival of neuroblastoma cells, and restoring the levels of BAP1 in these cells facilitated a delay in S and G2/M phase of the cell cycle, as well as cell apoptosis. The mechanism that BAP1 induces cell death is mediated via an interaction with 14-3-3 protein. The association between BAP1 and 14-3-3 protein releases the apoptotic inducer protein Bax from 14-3-3 and promotes cell death through the intrinsic apoptosis pathway. Xenograft studies conﬁrmed that the expression of BAP1 reduces tumor growth and progression in vivo by lowering the levels of pro-survival factors such as Bcl-2, which in turn diminish the survival potential of the tumor cells. Patient data analyses conﬁrmed the ﬁnding that the high-BAP1 mRNA expression correlates with a better clinical outcome. In summary, our study uncovers a new mechanism for BAP1 in the regulation of cell apoptosis in neuroblastoma cells. Introduction addition, terminal differentiation therapy and immu- notherapy is used as a current standard therapy for high- Neuroblastoma originates from the sympathetic ner- vous system and is composed of undifferentiated and risk neuroblastomas in order to eliminate residual tumor poorly differentiated neuroblasts arising from the differ- cells that are resistant after chemotherapy and stem cell 3–5 ent stages of the sympathoadrenal lineage of neural crest transplantation . origin . The patient age, N-Myc ampliﬁcation, deletion of Tumorigenesis in neuroblastoma can be caused by the the chromosome, and metastatic spread are important upregulation of cell survival signaling and the lack of factors with regard to treatment decision and patient cellular apoptosis. Therefore, understanding the prognosis. Although, N-Myc has important prognostic mechanism that leads to cell survival pathways can pro- value, ampliﬁcation is only observed in about 25% of vide avenues for the development of novel therapeutics. neuroblastoma cases and other factors contributing to The hallmark of apoptosis is the activation of caspases high-risk neuroblastoma are not known . Surgery, radio- that coordinate cleavage of substrates leading to cell therapy, and intensive induction chemotherapy with death. Generally, apoptosis is divided into the extrinsic autologous stem cell transplantation are commonly used and intrinsic pathway. The extrinsic pathway is mediated as treatment therapy for neuroblastoma patients. In via cell surface death receptors; whereas, the intrinsic 6–9 pathway is mediated via the mitochondrial . DNA- damaged cells are eliminated by the intrinsic pathway in Correspondence: Ramin Massoumi (Ramin.Massoumi@med.lu.se) which the Bcl-2 family of proteins plays a critical role . Department of Laboratory Medicine, Translational Cancer Research, Lund University, Medicon Village, Lund, Sweden This family is divided into anti-apoptotic proteins Department of Pharmacology and Emory Chemical Biology Discovery Center, (including Bcl-2, Bcl-XL, and Mcl-1), and pro-apoptotic Emory University School of Medicine, Atlanta, USA proteins, which is further divided into multi-domain Full list of author information is available at the end of the article Edited by J Chipuk © The Author(s) 2018 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to theCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Ofﬁcial journal of the Cell Death Differentiation Association 1234567890():,; 1234567890():,; Sime et al. Cell Death and Disease (2018) 9:458 Page 2 of 16 proteins, such as Bax, Bak, and BH3-only proteins, CRL-2149) cells were cultured in RPMI 1640 Medium including Bad, Bim, and HRK/DP5. Tumor cells often (HyClone, Thermo Scientiﬁc, USA), supplemented with increase the expression of anti-apoptotic Bcl-2 members 10% FBS (Sigma-Aldrich, Sweden), and 0.1% penicillin/ to avoid cancer cells undergoing apoptosis. Indeed, in a streptomycin (Gibco, Life Technologies, UK). SK-N-Be2c large subset of neuroblastoma patients, an elevated level cells (ATCC, CRL-2271) were cultured in MEM 10,11 of Bcl-2 has been detected . Besides regulating cancer (HyClone, Thermo Scientiﬁc, USA) supplemented with 10% FBS (Sigma-Aldrich) and 0.1% penicillin/streptomy- cell survival, chemotherapy-induced apoptosis is blocked in neuroblastoma through the involvement of the Bcl-2 cin (Gibco). protein family . BRCA1-associated protein 1 (BAP1) is a deubiquiti- Animal model nating enzyme that was discovered through its interaction SK-N-Be2c cells (5.0 × 10 cells in 0.1 mL PBS) stably with the RING ﬁnger domain of tumor suppressor protein expressing BAP1 or control expression plasmid (10 in 13,14 BRCA1 . BAP1 is a tumor suppressor gene deleted or each group) and SK-N-FI cells were subcutaneously mutated in various human cancer types, including breast, injected into the dorsal left or right ﬂank of 4-week-old nu/nu lung, renal cell carcinoma, metastatic uveal melanomas, female NMRI-nu/NMRI-Foxn1 (Janvier Labs). The 13,15–18 and malignant pleural mesotheliomas . In mice, the animals were maintained under speciﬁc pathogen-free disruption of BAP1 leads to the development of myeloid (SPF) conditions. All experimental procedures were neoplasia ; whereas, the expression of BAP1 suppresses approved by the Malmo and Lund Animal Ethics Com- the growth of non-small cell lung carcinoma cells in nude mittee with the ethical number M129-15. The tumor mice . Another function of BAP1 is to prevent abnormal growth was followed by measuring the volume every ﬁfth mitotic spindle formation and genome instability via the day. The subcutaneous tumors were collected, weighed, deubiquitination of α-tubulin in human breast cancer and chosen for further study. cells . BAP1 can also interact with several proteins associated with chromatin and transcription regulation, Transient transfection such as sex combs-like ASXL1 and ASXL2, forkhead Transient transfection assays were carried out in six- transcription factors FOXK1 and FOXK2, lysine-speciﬁc well plates at 60% conﬂuence, using PolyFect Transfection demethylase 1B (KDM1B), O-linked N-acetylglucosamine Reagent (Qiagen) or Lipofectamine 2000 Reagent (Invi- 21–23 transferase (OGT), and host cell factor 1 (HCF-1) . trogen, life technologies) in accordance with the manu- facturer’s recommendations. Twenty-four hours prior to Previous studies have shown that BAP1 has a role in cell 15,19,22,24,25 cycle regulation and cell proliferation . Further, transfection, cells were seeded in medium containing 10% BAP1 can regulate the cell cycle by inﬂuencing the FBS. Cells were incubated at 37 °C and 5% CO for 24–48 expression of E2F1 target genes in uveal melanoma cells . h. The medium was removed and the cells washed with The regulation of DNA damage response by BAP1 is PBS, after which fresh serum-containing medium was mediated via rapid poly(ADP-ribose)-dependent recruit- added to the cells. DNA, 1.5–4.0 µg, was mixed with an ment of the polycomb repressive deubiquitination (PR- appropriate volume of OptiMEM (Sigma), followed by the 27,28 DUB) complex to sites of DNA damage . Phosphor- addition of 10–25 µl PolyFect transfection reagent. The ylation of BAP1 at S592 is an important regulatory samples were incubated at room temperature for 10 min, mechanism to dissociate BAP1 from chromatin and in order to allow the complex formation to be completed, regulate-speciﬁc genes during DNA replication and prior to being transferred to the cells. Cells were subse- repair . quently incubated for 24–48 h, collected and used for In this study, we investigated the role of BAP1 as a further analyses. tumor suppressor gene in neuroblastoma based on the Transient transfection of siRNA (10 µM) was performed 3p-chromosomal location of BAP1 and that alteration in in six-well plates at 60% conﬂuence, using Lipofectamine chromosome arms 3p is a common event in neuro- RNAiMAX Reagent (Invitrogen, life technologies) fol- blastoma. It was found that the pro-apoptotic function of lowing the manufacturer’s instructions. Forty-eight hours BAP1 is mediated via binding to 14-3-3 protein, which post transfection, cells were collected and used for further further facilitated cell death signaling in neuroblastoma. analyses. The siRNA sequences were listed as following: 14-3-3 zeta: 5′-AAAGTTCTTGATCCCCAATGC-3′ Materials and methods Crtl. GFP: 5′-GACCCGCGCCGAGGTGAAGGTTT-3′ Cell culture The human neuroblastoma cell lines were cultured for Drug treatment 5 days at 37 °C and 5% CO as follows: IMR32 (ATCC, The non-transfected SK-N-RA cells in six-well plates 30,31 CCL-127), SK-N-SH-RA , SK-N-FI (ATCC, CRL- were treated for 24 h with BV02 (SML0140, Sigma- 2142), SK-N-SH (ATCC, HTB-11), and SK-N-DZ (ATCC, Aldrich) at different concentrations (5, 20, and 100 µM). Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 3 of 16 The percentages of apoptotic cells were determined using according to the manufacturer’s instructions (Chemo- the Burkers chamber. Metec). To assess the cell cycle proﬁle following noco- dazole synchronization, ethanol-ﬁxed cells were washed Cell lysis and subcellular fractionation twice with PBS, resuspended in propidium iodide (PI) Cells were grown on 10 cm plates to 80% conﬂuence, staining solution (20 μg/ml PI, 200 μg/ml Rnase A in PBS collected in PBS, and lysed in buffer comprising 1 M Tris- containing 0.1% Triton X-100), and incubated at 37 °C for 30 min. For each condition, 10,000 cells were analyzed by HCl (pH 7.6), 5 M NaCl, 0.5 M EDTA, 1% Triton X-100, and Complete® protease inhibitor cocktail (Roche, Ger- a ﬂow cytometer (BD FACS Verse ) and the cell cycle many). The extraction of protein from tumor tissue was proﬁle was analyzed using the FlowJo v10 software. prepared after samples were homogenized and sonicated in cold RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM Western blot analysis NaCl, 1.0 mM EDTA, 0.1% SDS, 1.0% Triton X-100, 1.0% Western blotting was carried out, following standard sodium deoxycholate) freshly supplemented with protease protocol. The membrane was incubated with primary inhibitor cocktail (Roche, Germany). After centrifugation antibody in 5% BSA followed incubation with corre- at 10,000×g for 15 min at 4 °C, the protein in the super- sponding horseradish peroxidase (HRP)–conjugated sec- natant was measured using the Protein concentration and ondary antibody (DAKO, Denmark) for 1 h at RT. Bands was assessed using the Bradford method and sample were detected using immunochemical detection accord- concentrations were adjusted accordingly. Lysates were ing to the manufacturer’s instruction (Santa Cruz, USA). boiled for 5 minutes at 100 °C with a 2× sample buffer, For the chemiluminescent reaction, Supersignal Substrate containing 1 mM dithiothreitol (DTT). (Thermo Scientiﬁc) was used according to manufacturer’s For cell fractionation, cells were ﬁrst resuspended in ice- instructions. Chemiluminescence was detected with an cold buffer A (10 mM HEPES (pH 7.9), 10 mM KCl, 1.5 LAS-1000 charge-coupled device camera (Fujiﬁlm) and mM MgCl , 0.34 M sucrose, 10% glycerol, 1 mM DTT, images were processed using the ImageJ software. complete protease inhibitors EDTA-free (Roche, Ger- many) and 0.1 mM phenylmethylsulfonyl ﬂuoride), 0.1% Quantitative PCR Triton X-100 was added followed by incubation for 10 The cells were rinsed in cold PBS and total RNA min on ice, and centrifuged at 4 °C for 5 min at 1300×g. extracted using the Perfect Pure RNA tissue kit (5 PRIME, The supernatant containing cytosolic fraction was Germany) according to manufacturer’s instruction. The removed and the pellet was enriched in chromatin and purity of RNA was analyzed and quantiﬁed by a Nano- washed once in buffer A before re-suspension in protein Drop spectrophotometer (Saveen Werner, Sweden) and sample buffer (2% SDS, 312.5 mM Tris, pH 6.8, 5% gly- used for cDNA synthesis according to the manufacturer’s cerol, 0.003% BFB, and 50 mM DTT) and boiled in sample instruction (QPCR cDNA synthesis kit, Stratagene, USA). buffer for 10 min and analyzed by western blotting using PCR runs were performed in the QuantStudioTM 7 Flex anti-tubulin (Sigma-Aldrich, Sweden) and anti-LaminB System using SYBR®Green Reagent (Applied Biosystems), (Thermo Scientiﬁc, USA) as molecular markers for the with the following program; 2 min 50 °C, 10 min in 95 °C cytosolic and nuclear fractions, respectively. followed by 40 three-step cycles consisting of 95 °C for 20 s and 60 °C for 30 s and 72 °C for 1 min. The following Cell cycle synchronization primers were used: Cells were synchronized in G0/G1 or G2/M phase BAP1_S: 5′-GACCCAGGCCTCTTCACC-3′ either by serum starvation for 48 h or nocodazole treat- BAP1_A: 5′-AGTCCTTCATGCGACTCAGG-3′ ment for 18 h (100 ng/ml, sigma) respectively. The syn- 14-3-3 Zeta_S: 5′-CCTGCATGAAGTCTGTAACT- chronized cells were released from their respective GAG-3′ arrested phase by the addition of 10% serum containing 14-3-3 Zeta_A: 5′-GACCTACGGGCTCCTACAACA- RPMI (HyClone, Thermo scientiﬁc, USA), and samples 3′ were taken at various times. The cells were ﬁxed by GAPDH_S: 5′-TGCACCACCAACTGCTTAGC-3′ dripping ice-cold 70% ethanol while vortexing the cells GAPDH_A: 5′-GGCATGGACTGTGGTCATGAG-3′ vigorously and stored for 12–24 h at −20 °C. Ethanol- suspended cells were centrifuged and the ethanol carefully Retrovirus production and transduction decanted. The cells were washed once with PBS and then Retroviral vector alone or vector encoding wild-type resuspended with NucleoCounter Solution 3 (1 μg/ml FLAG and HA-tagged BAP1 (plasmid 22539) containing a 4′,6-diamidino-2-phenylindole, 0.1% Triton X-100 in puromycin resistance gene for selection of stably trans- PBS), followed by incubation for 5 min at 37 °C. Samples duced cells were obtained from Add gene. Retrovirus of 10 μl volume were loaded into a slide chamber (NC- production was performed by Research engineer, Vector slide A8), and the DNA fragmentation protocol was used Unit at Lund University, Sweden. Cells were seeded in six- Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 4 of 16 well plates with 40% conﬂuency in RPMI medium analysis was carried out to determine the fraction of (HyClone, Thermo-scientiﬁc, USA) supplemented with proliferating cells. To deﬁne EdU labeled cell fractions 10% FBS, 2 mM L-glutamine, 100 units/mL penicillin and (Click-iT EdU Alexa Flour 488 positive), EdU-unlabeled 100 µg/mL streptomycin (Sigma-Aldrich, Sweden). Next SK-N-SH cells were used as a negative control after day, the medium was changed with 1 ml medium con- simultaneously stained with click-iT reaction solution. taining hexadimethrine bromide (Sigma-Aldrich, Sweden; For each condition, 20,000 events were acquired by a ﬂow cytometer (BD FACS Verse™) and the percentage indi- ﬁnal concentration 8 mg/ml) to each well. A total of 10 µl of retroviral particles was added to appropriate wells and cated in the bar graph was analyzed using FlowJo cells were incubated 24 h at 37 °C in a humidiﬁed incu- v10 software. The results are the mean of two indepen- bator in an atmosphere of 5–7% CO . The medium was dent experiments. changed with 1 ml RPMI containing puromycin (Sigma- Aldrich, Sweden). The transduction efﬁciency was initially Annexin V-phycoerythrin staining and ﬂuorescence- evaluated using western blot and Immunoﬂuorescence activated cell sorting analysis staining. Cells were stained using Annexin V-phycoerythrin (PE) apoptosis detection kit (BD Pharmingen) according to the Apoptosis and cell proliferation manufacturer’s recommendations. Brieﬂy, the ﬂoating Cells were grown on six-well plates, collected at 24 or cells in the medium together with the respective adhered 48 h post transfection, and counted using the Burkers cells were collected and washed twice with ice-cold PBS chamber, or Countess® Automated Cell Counter (Invi- before suspended in 100 μl binding buffer (1 × 10 /ml). trogen). For apoptosis analyses, the cells were ﬁxed in PFA Next, Annexin V-PE (5 µl) and 7-AAD (5 µl) were added on coverslips, and stained with a Vindelöv solution, and the mixture was incubated for 15 min in the dark. containing propidium iodide. After washing, the cover- Finally, 400 µl binding buffer was added to the cells and slips were mounted onto glass slides and examined by analyzed using a ﬂow cytometer (BD FACS Verse ). The ﬂuorescence microscopy. Cells were scored for apoptosis, data analysis was performed using FlowJo v10 software. based on nuclear morphology. Apoptosis was furthermore evaluated, using NucleoCounter NC-3000 (Chemometec), Measurement of mitochondrial membrane potential in conformity with the DNA fragmentation assay. Brieﬂy, The mitochondrial membrane potential (ΔΨm) was transfected and non-transfected cells (controls) were evaluated by ﬂow cytometric analysis with JC-1 staining. grown on 6-well plates. The cells were collected by Stably transduced neuroblastoma cells both with empty trypsinization and pooled with the cells ﬂoating in the vector (control) and BAP1 were collected after trypsini- medium. After a short centrifugation, the supernatant was zation. For arsenic induction, cells were seeded in 100 mm removed and the precipitated cells were washed once with Petri dish (5 × 105 cells) for 48 h, and then treated with PBS. After a second centrifugation, the cells were resus- arsenic trioxide (6 µM) and collected after 24 h. The cells pended in a small volume of PBS, and the single-cell resuspended in PBS (1 × 10 ) following washing steps suspensions were added to 70% ethanol for ﬁxation. The were incubated with 5 µM JC-1 at 37 °C, 5% CO , for 30 samples were vortexed and stored for 12–24 h at −20 °C. min in the dark. For each condition, 10,000 events were The ethanol-suspended cells were centrifuged and the acquired by a ﬂow cytometer (BD FACS Verse™) and the ethanol carefully decanted. Cells were washed once with percentage indicated in the dot plot was analyzed using PBS and then resuspended with NucleoCounter Solution FlowJo v10 software. 3 (1 µg/ml DAPI, 0.1% Triton X-100 in PBS), followed by incubation for 5 min at 37 °C. The samples of 10 µl LC–MS/MS analysis volume were loaded into a slide chamber (NC-slide A8), MS analyses were carried out on an Orbitrap Fusion and the DNA Fragmentation protocol was employed Tribrid MS system (Thermo Scientiﬁc) equipped with a according to manufacturer’s instructions (Chemometec). Proxeon Easy-nLC 1000 (Thermo Fisher). Injected pep- EdU incorporation: A thymidine analog, 5-ethynyl-2′- tides were trapped on an Acclaim PepMap C18 column deoxyuridine (EdU), incorporation was used to measure (3 µm particle size, 75 µm inner diameter × 20 mm cell proliferation using Click-iT® EdU Alexa Flour 488 cell length). After trapping, gradient elution of peptides was proliferation assay kit (Thermo Fisher Scientiﬁc). Brieﬂy, performed on an Acclaim PepMap C18 100 Å 3 μm, 150 BAP1-low and BAP1-overexpressing SK-N-SH cells were mm, 75 μm). The outlet of the analytical column was seeded in a six-well plate (250 × 10 cells/well) in DMEM coupled directly to the mass spectrometer using a Prox- with 10% FBS containing EdU (5 µM/well) and incubated eon nanospray source. The mobile phases for LC at 37 °C. After 24, 48, and 72 h, the cells were collected, separation were 0.1% (v/v) formic acid in LC–MS grade washed twice in PBS containing 1% BSA and click-iT water (solvent A) and 0.1% (v/v) formic acid in acetoni- reaction was performed on ﬁxed cells before FACS trile (solvent B). Peptides were ﬁrst loaded with a constant Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 5 of 16 pressure mode with a ﬂowrate of solvent A onto the GST antibody at 1:3000 (Sigma, A7340, USA) and anti-β- trapping column. Subsequently, peptides were eluted via Actin antibody at 1:5000 (Sigma, A5441, USA). the analytical column at a constant ﬂow of 300 nl/min. During the elution step, the percentage of solvent B Time-resolved ﬂuorescence resonance energy transfer (TR- increased in a linear fashion from 5 to 10% in 2 min, then FRET) assay increased to 25% in 85 min and ﬁnally to 60% in a further The TR-FRET assay was carried out with puriﬁed pro- 20 min. The peptides were introduced into the mass teins in 384-well black solid bottom plates (Corning, 3573, spectrometer via a Stainless steel emitter 40 mm (Thermo USA). His-BAP1 protein (Boston Biochem, E-345, USA), Fisher) and a spray voltage of 2.0 kV was applied. The control GST protein (Abcam, ab70456, USA) and puriﬁed capillary temperature was set at 275 °C. Data acquisition GST-14-3-3 proteins were mixed in a total volume of 30 was carried out using a data-dependent top N method µl FRET buffer (20 mM Tris (pH 7.0), 50 mM NaCl, 0.01% with cycle time of 2 s. The master scan was performed in NP-40). Anti-GST-Terbium antibody (Cisbio, the Orbitrap in the range of 380–1580 m/z at a resolution 61GSTTLB, USA) and anti-His-D2 antibody (Cisbio, of 60,000 FWHM. The ﬁlling time was set at maximum of 61HISDLA, USA) were added to each well at the ﬁnal 50 ms with limitation of 4 × 105 ions. Ion trap CID-MS2 dilution of 1:1000 and 1:500, respectively. The FRET was acquired using parallel mode, ﬁlling time 50 ms with signals were detected using Envision spectrophotometer limitation of 1.5 × 104 ions, a precursor ion isolation (Laser excitation at 337 nm, emission for the donor at 620 width of 0.7 m/z and resolution of 30,000 FWHM. Nor- nm and emission for acceptor at 665 nm). The data were malized collision energy was set to 35%. Only multiply presented as the ratio of (counts at 665/counts at 620 charged (2+ to 4+) precursor ions were selected for MS2. nm) × 10 with standard deviation calculated from tripli- The dynamic exclusion list was set to 30 s and a relative cate samples. mass window of 5 p.p.m. Immunoﬂuorescence and confocal microscopy Pull-down assay Cells were cultured on coverslips in six-well plates for Ni-NTA pull-down assay: SK-N-RA cells were trans- 24 h and then ﬁxed in paraformaldehyde (PFA) (4% for 15 fected with His6-ubiquitin and/or BAP1 expressing plas- min). Next, the cells were permeabilized with 0.3% Triton mids. Cells were lysed after 24 h in 8 M UREA, 100 mM X-100 solution, after which they were blocked with 1% Na HPO /NaH PO ; 10 mM Tris, pH 8.0, 10 mM imi- BSA and 5% goat serum for 30 min to prevent non- 2 4 2 4 speciﬁc antibody binding, and subsequently incubated for dazole and 5 mM DTT, and lysates (1 mg) were incubated with 40 μl Ni-NTA Superﬂow (Qiagen) overnight at room 1 h with primary antibodies against BAP1 (Clone C-4; sc- temperature. The resin was washed three times with lysis 28383, Santa Cruz, USA), α-tubulin (Sigma-Aldrich, buffer, once with 8 M UREA, 100 mM Na HPO / Sweden), and FLAG (Sigma-Aldrich, Sweden). After 2 4 NaH PO ; 10 mM Tris, pH 6.3; 10 mM imidazole; 5 mM washing coverslips, ﬂuorescent antibodies (Alexa 488 goat 2 4 DTT and once more with lysis buffer before eluting anti-mouse or Alexa 546 goat anti-rabbit (Invitrogen, proteins with lysis buffer containing 200 mM imidazole. USA)) were applied, then the coverslips were washed and Isolated proteins were subjected to immunoblotting using mounted in VECTASHIELD® with diamidino-2- antibodies against ubiquitin at 1:1000 (SC-8017, Santa phenylindole (DAPI) (Vector Laboratories). The images Cruz) or pan-14-3-3 at 1:500 (SC-629, Santa Cruz). were obtained using a Zeiss LSM710 confocal microscope. GST pull-down assay: Co-transfected cells were lysed in 1% NP-40 buffer (20 mM Tris (pH 8.0), 137 mM NaCl, 5% RT proﬁler PCR array analysis glycerol, 1% nonident P-40, Protease Inhibitor (Sigma, The apoptosis pathway-focused gene expression proﬁl- P8340, USA) and phosphatase inhibitor (Sigma, P5726, ing was performed using a 384-well human RT Proﬁler USA) and centrifuged for 10 min at 4 °C. Aliquots were PCR Array PAHS-012ZE (Qiagen). Following the total taken for input control and the remaining lysate were RNA extraction, cDNA was synthesized from SK-N-RA incubated with pre-washed (with lysis buffer) glutathione- cells transiently transfected with an empty vector used as conjugated sepharose beads (GE, 17075605, Sweden) for control (GFP) and full-length BAP1 based on the manu- 3 h at 4 °C. After washing three times with 1% NP-40 facturer’s instructions. In this array, 84 different genes buffer, the beads were resuspended in sample buffer (Bio- involved in programmed cell death were analyzed based Rad, 1610737, USA) and boiled at 95 °C for 5 min. Then on SYBR Green real-time PCR using the QuantStudio™ 7 the eluted samples and whole-cell lysate were subjected to Flex (Applied Biosystems™). Normalization was per- sodium dodecyl sulfate-polyacrylamide gel electrophor- formed using the ﬁve different housekeeping genes esis (SDS-PAGE) and analyzed by western blot with the included in the array and the fold-change was calculated anti-Flag antibody at 1:3000 (Sigma, A8592, USA), anti- using the RT Proﬁler PCR Array Data Analysis. Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 6 of 16 Gene expression data for survival synchronization after 48 h showed an accumulation of BAP1 and Bax normalized gene expression array data cells in the S, G2/M, and subG1 phases in BAP1 expres- (Affymetrix HG-U133A [GPL96] and HG-U133B sing cells compared to control cells (Fig. 1g, right panels). [GPL97] platforms; GSE3446) of 102 neuroblastoma patients were downloaded from the website R2: Genomics BAP1 promotes cell death in neuroblastoma cells Analysis and Visualization Platform (http://r2.amc.nl). To Next, we investigated whether the accumulation of test the association of gene expression levels with relapse- BAP1 expressing cells in subG1indicates cell death. The free survival, individual gene expression proﬁles were percentage of Annexin V positive cells (Fig. 2a) or cells dichotomized by median split into “high” or “low” having fragmented DNA (Fig. 2b) were signiﬁcantly expression groups, and Kaplan–Meier survival curves higher in the BAP1 expressing cells compared to BAP1- were plotted for each group. C91A or control cells (Fig. 2b). PI staining also displayed a higher number of BAP1 expressing cells undergoing cell Statistical analyses death compared to BAP1-C91A or control cells (Fig. 2c). Statistical analyses were performed using SigmaPlot In cell cycle synchronized neuroblastoma cells, we mon- software. The results are expressed as mean ± s.e.m. or as itored the commitment of BAP1 expressing cells to a percent. P-values <0.05 were deemed statistically sig- apoptosis after release. Poly (ADP-ribose) polymerase niﬁcant. Statistical comparisons were assessed by (PARP) cleavage was detected within 15–22 h post release ANOVA or by Student’s t-test (p < 0.05). in BAP1 but not in control expressing cells (Fig. 2d and Supplementary Fig. 6). Investigating the mitochondrial Results membrane potential (ΔΨm) in non-treated or arsenic- Reduced levels of BAP1 in neuroblastoma cells treated cells showed an elevated number of JC-1 dye In order to investigate whether BAP1 expression is positive cells (Fig. 2e, middle and right panels) and an altered in neuroblastoma, we studied the levels of BAP1 intense release of Mito Tracker into the cytosol (Supple- expression in different neuroblastoma cell lines. It was mentary Fig. 7) in BAP1 expressing cells compared to found that BAP1 was signiﬁcantly downregulated at the control cells. These results suggest that BAP1 interfere mRNA and protein levels in the SK-N-BE2c, SK-N-SH, with the intrinsic apoptosis pathway in neuroblastoma SK-N-RA, KcN-69n, and SH-S454 cells compared to cells. Furthermore, transient transfection of neuro- IMR32 and SK-N-F1 cells (Fig. 1a, b and Supplementary blastoma with increasing concentrations of the BAP1- Fig. 1). To investigate the role of BAP1 in neuroblastoma, cDNA showed a direct correlation between levels of BAP1 SK-N-SH, SK-N-RA, and SK-N-BE2c cells were virally expression and cell death (Fig. 2f and Supplementary infected with full-length BAP1 tagged with HA and FLAG Fig. 8). The comparison of growth rate between neuro- (Supplementary Fig. 2). Monitoring cell morphology of blastoma cells revealed that low-expressing BAP1 cells, neuroblastoma cells expressing full-length BAP1 showed SK-N-DZ cells (Supplementary Fig. 9A) showed an ele- that these cells but not the cells expressing a control vated proliferation rate and cell cycle progression, as well plasmid or the cells expressing a catalytically inactive as reduced cell death compared with BAP1 expressing SK- mutant of BAP1 (BAP1-C91A) facilitated neurite out- N-F1 cells (Supplementary Fig. 9B-D). In addition, low- growth 72–96 h post transfection (Fig. 1c, d and Supple- expressing BAP1 cells, SK-N-Be2c cells develop tumors mentary Fig. 3). Neurite outgrowth can indicate the non- (90% of animals) in xenograft experiments compared to proliferative stage of the cells and this can interfere with BAP1 expressing SK-N-FI cells (10% of animals) (Sup- the survival of neuroblastoma cells . To determine plementary Fig. 9E). Further, downregulation of BAP1 in whether BAP1 can affect cell growth, we analyzed the SK-N-FI cells by using small interfering RNA (siRNA) proliferation rate of BAP1, BAP1-C91A, and GFP control (Supplementary Fig. 10A) enhanced cell survival com- (Ctrl.) expression plasmid lentivirus infected cells. Rescue pared to siRNA treated control cells (Supplementary expression of BAP1, but not BAP1-C91A in two neuro- Fig. 10B-C). Treatment of BAP1-overexpressing cells with blastoma cells lines reduced the proliferation rate at 48 siRNA oligos against BAP1 reversed this effect and pro- and 72 h, compared with control cells (Fig. 1e and sup- moted survival of the cells (Fig. 2g), as well as reduced the plementary Fig. 4). The effect of BAP1 overexpression in subG1 population of the cells (Fig. 2h). altering cell cycle progression was demonstrated in non- To identify the global BAP-mediated pro- and anti- synchronized SK-N-Be2c cells where the frequency of apoptotic genes in neuroblastoma cells, we tested the cells in the S, G2/M, and subG1 phases were higher expression proﬁles of 84 well-known genes involved in compared to the control or BAP1-C91A expressing cells apoptosis and survival pathways. While few genes (Fig. 1f and Supplementary Fig. 5). To conﬁrm this result, encoding pro-apoptotic proteins (HRK, CD70) were we synchronized the cells in G2/M phase by treating the upregulated, other genes encoding anti-apoptotic proteins cells with nocodazole (Fig. 1g, left panels). The release of (BCL2L1, XIAP, BCL2L10, and BFAR) were Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 7 of 16 Fig. 1 Reduced expression of BAP1 in neuroblastoma cell lines. a Western blot analysis of BAP1 and actin expression in SK-N-Be2c, IMR32, SK-N- F1, SK-N-SH, and SK-N-RA cells (n = 5). b Fold changes in BAP1 mRNA expression, measured using real-time reverse transcription PCR (qRT-PCR) of cDNA from SK-N-Be2c, IMR32, SK-N-F1, SK-N-SH, and SK-N-RA cells. The IMR32 mRNA expression is set to 1 (mean ± s.e.m., n = 3, ***P < 0.001). c Phase-contrast microscopy and confocal images of SK-N-BE2c cells transiently transfected with empty plasmid (Ctrl.) or BAP1 expression plasmid for 72 h. The neurite outgrowth in BAP1 expressing cells is shown by staining with anti-α-tubulin antibody and DAPI stain nuclei (blue) (n = 5). d The morphological effects of the overexpression of BAP1 or empty plasmid (Ctrl.) in SK-N-BE2c cells by counting the number of cells with cell processes longer than the length of two cell bodies (mean ± s.e.m., n = 3, ***P < 0.001). e The cell proliferation was determined by cell counting after 24–96 h using lentivirus infected SK-N-SH (upper panel) and SK-N-RA (lower panel) with GFP (Ctrl.) or BAP1. The experiment was performed in triplicate. Data are presented as mean ± s.e.m. (n = 4, *P < 0.05). f, g Representative histogram plots (n = 3) displaying changes in cell cycle distribution as a result of BAP1 overexpression assessed using propidium iodide (PI) staining in SK-N-Be2c cells unsynchronized (f) or synchronized by nocodazole (100 ng/ml) for 18 h (g). The synchronized cells were released after 48 h and cell cycle proﬁle experiment was performed using BD FACS Verse™. The histogram plot and the percentage of cells in sub-G1, G1, S, and G2/M were determined using FlowJo v10 software (TreeStar) downregulated in BAP1 expressing cells compared to the different isoforms of 14-3-3 proteins including 14-3-3-σ, control cells (Fig. 2i). This result was conﬁrmed by -ζ,-ε, and -β (Fig. 3a, Supplementary Fig. 11, and Sup- showing that the transient overexpression of BAP1 plementary Table 1). Figure 3a shows the LC–MS/MS downregulates the total levels of Bcl-2, 24–48 h post analysis and one example of a tandem mass spectrum that transfection (Fig. 2j). unambiguously identiﬁes the sequence 191–224 of 14-3- 3-σ. Further, for the protein 14-3-3-σ, 23 peptides cor- BAP1 binding to 14-3-3 facilitate the apoptotic signaling in responding to 83% of sequence coverage were identiﬁed neuroblastoma (Supplementary Fig. 11 and Supplementary Table 1). Seeking an explanation for BAP-mediated apoptosis and Among the other previously identiﬁed BAP1 interacting the reduced expression of cell survival factors in neuro- partners (https://string-db.org/), we could conﬁrm bind- blastoma cells, we performed a LC–MS/MS analysis. ing to CBX3, IPO4, IPO5, and RBBP7 (Supplementary Immunoprecipitation (IP) of FLAG-tagged BAP1 in SK- Fig. 12). The interaction between BAP1 and 14-3-3 pro- N-SH followed by LC–MS/MS analysis identiﬁed tein was conﬁrmed by performing the co-IP of FLAG- Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 8 of 16 Fig. 2 (See legend on next page.) Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 9 of 16 (see ﬁgure on previous page) Fig. 2 BAP1 induces cell death and downregulates expression of survival genes. a SK-N-SH cells were transiently transfected with control or BAP1 expression plasmid for 48 h and analyzed for apoptosis by ﬂow cytometry following PE and 7-AAD staining. Quadrants are as follows: lower left: viable cells; lower right: early apoptotic; upper right: late apoptotic/necrotic. The numbers in quadrants represent the percentage of the viable cells (lower left), early apoptotic (lower right), and late apoptotic/necrotic (upper right). The table (right panel) shows the distribution of viable cells, early apoptotic and late apoptotic/necrotic from three independent experiments. b SK-N-RA or SK-N-SH cells were transfected with mock or FLAG-BAP1 for 24 h. Cells were ﬁxed and treated for DNA fragmentation assay. Data represent the percentage of cells with fragmented DNA. Data are presented as mean ± s.e.m. (n = 3, *P < 0.05). c Cell death visualized by propidium iodide (PI) staining in SK-N-RA and SK-N-SH cells infected with control (GFP) or FLAG-tagged full-length BAP1. Results are mean ± s.e.m. of percentage cells (*P < 0.05) with PI staining from three independent experiments counting 200 cells in triplicate. d Extracts from synchronized and released SK-N-RA cells transfected with FLAG-tagged full-length BAP1 were examined by western blotting (WB) in relation to post-release time points with an anti-PARP and anti-laminB. Arrowheads indicate FL-PARP: full- length PARP and CL-PARP: cleaved PARP. (n = 3). e Effect of BAP1 overexpression in altering mitochondrial membrane potential in SK-N-Be2c cells was measured by ﬂow cytometry following JC-1 staining. Cells without JC-1 staining were used to demarcate the gates prior to measure the ratio of red (mitochondria with a non-depolarized ΔΨ) and green (depolarized ΔΨ) ﬂuorescence emissions after JC-1 staining (5 μM). Mitochondrial membrane potential was shown to be altered in BAP1-overexpressing cells alone (35.9%) or following arsenic treatment (6 μM, 47.4%) (n = 3). f SK-N- BE2c cells were non-transfected (0 μg/μl) or transiently transfected with different concentration of full-length FLAG-tagged BAP1 expression plasmid (0.5 μg/μl, 1.5 μg/μl, 3.5 μg/μl, and 7.0 μg/μl) for 48 h and transfected with siRNA oligos against BAP1 for another 48 h as indicated in the ﬁgure. Results are mean ± s.e.m. of percentage cells (*P < 0.05) with PI staining from three independent experiments counting 200 cells in triplicate. g SK-N- SH cells were transiently transfected with full-length Flag-tagged BAP1 expression plasmid (7.0 μg/μl) for 48 h and transfected with siRNA oligos against BAP1 for 48 h. The cells were treated with Annexin V-PE and 7-AAD before ﬂow cytometry analysis. Representative experiment shows numbers in quadrants represent the percentage of the viable cells (lower left), early apoptotic (lower right), and late apoptotic/necrotic (upper right) (n = 3). h SK-N-SH cells were transiently transfected with full-length FLAG-tagged BAP1 expression plasmid (7.0 μg/μl) for 48 h and transfected with siRNA oligo against BAP1 for 48 h. Cells were ﬁxed and treated with DAPI. Representative experiment shows alterations in DNA content and cell cycle proﬁles using Nucleocounter NC-3000 (n = 3). i The Human Apoptosis RT² Proﬁler PCR Array revealed up- and downregulated genes involved in programmed cell death in SK-N-RA cells transiently transfected with full-length FLAG-tagged BAP1 compared to control (empty vector). The level of relative expression for each particular gene in the two samples are plotted in log–log scatter plot after normalization against ﬁve housekeeping genes. The line in the middle indicates relative fold changes. Red color indicates upregulated and green color indicates downregulated genes (n = 2). j Western blot analysis of full-length BAP1 expression in SK-N-RA cells analyzed by blotting with an anti-Bcl-2, anti-BAP1, or anti-actin antibody (n = 3) tagged BAP1 and pull-down of endogenous 14-3-3 in the cells expressing low levels of BAP1 (Fig. 4a) or cells that have been transfected with full-length BAP1 or neuroblastoma cells (Fig. 3b). The reciprocal IP using 14- 3-3 antibodies also indicated the formation of the com- BAP1-C91A (Fig. 4b). To analyze, whether 14-3-3 can plex between BAP1 and 14-3-3 (Fig. 3c). In addition, undergo ubiquitination, the overall levels of ubiquitin- ﬂuorescence resonance energy transfer (FRET) assay associated 14-3-3 in BAP1- or empty plasmid-transfected demonstrated a direct interaction between BAP1 and 14- neuroblastoma was investigated. We could not 3-3 (Fig. 3d). GST pull-down experiments demonstrated observe any differences in the levels of 14-3-3 ubiquiti- that BAP1 could associate with all of the 14-3-3 isoforms nation in the absence or presence of BAP1 expression in the cell (Fig. 3e). To deﬁne which domain(s) in BAP1 is (Supplementary Fig. 14). The overexpression of BAP1- required for interaction with 14-3-3-σ, a series of FLAG- C91A did not show any differences in the 14-3-3 ubi- tagged BAP1 deletion mutants (Fig. 3f) were transiently quitination compared to the BAP1 expressing cells transfected together with full-length GST-tagged 14-3-3- (Fig. 4b and Supplementary Fig. 14). Next, we tested the σ into HEK293T cells. GST pull-down assays demon- hypothesis that the BAP1 interaction with 14-3-3 releases strated that the deletion mutants containing the BAP1 the binding between 14-3-3 and Bax, which further domains NORS (non-regular secondary structure), and facilitates cell death. Indeed, an IP experiment, ULD (ubiquitin-like domain) interact with 14-3-3-σ showed that the interaction between 14-3-3 and Bax is (Fig. 3g and Supplementary Fig. 13A). Furthermore, abrogated in the presence but not in the absence of GST pull-down assays using 14-3-3-σ deletion mutants BAP1 (Fig. 4c). The reciprocal IP using Bax antibodies (Fig. 3g), showed an interaction between truncation 163- conﬁrmed the release of the interaction between 248 and full-length BAP1 (Fig. 3h and Supplementary Bax and 14-3-3 in the presence of BAP1 (Fig. 4d). The Fig. 13B). We could also conﬁrm that the short fragment treatment of neuroblastoma cells with BV02, which is an of BAP1-ULD domain and 14-3-3-alpha helices 7-9 inhibitor of 14-3-3 and prevents the scaffolding function regions facilitate interaction between these two proteins of 14-3-3 promoted cell death in a concentration- (Fig. 3i), while the interaction between BAP1 and 14-3-3- dependent manner (Fig. 4e). The percentage of apopto- σ inactive mutants were reduced compared to the full- tic cells treated with 20 µM BV02 was similar to the length 14-3-3-σ (Fig. 3j). We could not observe any dif- cells transfected with BAP1 (Fig. 4f). In addition, siRNA ferences in the total levels of 14-3-3 protein expression in oligos protected neuroblastoma cells against 14-3-3-ζ Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 10 of 16 Fig. 3 Identiﬁcation of 14-3-3 protein as a novel binding partner to BAP1. a Representative LC–MS/MS chromatogram of in-gel digested sample depicted in red, an example of the peptide from 1433S protein eluted at 70.8 min. The Orbitrap mass spectra of the quadruply charged 1433S peptide sequence 191–224 (m/z 836.9045) enabled mass determination of the peptide (isotopic distribution depicted in gray). The assigned fragmentation pattern of sequence 191–224 peptide is given in black. b, c SK-N-RA cells were transfected with FLAG-tagged full-length BAP1 or Flag- empty vector (Ctrl) and lysates were immunoprecipitated with antibodies against FLAG (b) or 14-3-3 (c) and probed with FLAG to detect BAP1 or 14- 3-3 as indicated in the ﬁgure (n = 3). d TR-FRET assay using FRET buffer (Ctrl), puriﬁed control GST (negative control), GST-14-3-3 protein, and His- BAP1 mixed in FRET buffer, followed by addition of anti-GST-Terbium and anti-His-D2 antibodies. The FRET signals were detected using an Envision spectrophotometer. The data were presented as the ratio of counts at 665/counts at 620 nm × 10 with standard deviation calculated from triplicate samples (mean ± s.e.m., n = 3, *P < 0.05). e GST pull-down experiment using co-transfection of cells with GST-tagged 14-3-3 isoforms and full-length BAP1 followed by addition of GST-conjugated sepharose beads. The eluted whole-cell lysate was subjected to SDS-PAGE and analyzed by western blot using anti- FLAG, anti-GST, or anti-actin antibodies (n = 3). f Schematic representation of 14-3-3σ (9 alpha-helix) and BAP1 domains consisting of UCH (ubiquitin C-terminal hydrolases), NORS (non-regular secondary structure), ULD (ubiquitin like domain), and NLS (nuclear localization sequence). Residues numbers refer to amino acid positions. g GST pull-down experiment using co-transfection of cells with GST-tagged 14-3-3σ and FLAG- tagged full-length or deletion mutants of BAP1 followed by addition of GST-conjugated sepharose beads. The eluted whole-cell lysate was subjected to SDS-PAGE and analyzed by western blot using anti-FLAG, anti-GST, or anti-actin antibodies (n = 3). h GST pull-down experiment using co- transfection of cells with GST-tagged full-length or deletion mutant of 14-3-3σ and FLAG-tagged full-length BAP1 followed by addition of GST- conjugated sepharose beads. The eluted whole-cell lysate was subjected to SDS-PAGE and analyzed by western blot using anti-FLAG, anti-GST, or anti-actin antibodies (n = 3). i GST pull-down experiment using co-transfection of cells with GST-tagged full-length or deletion mutant of 14-3-3σ and full-length or FLAG-tagged deletion mutant of BAP1 (aa 635–694) followed by addition of GST-conjugated sepharose beads. The eluted whole-cell lysate was subjected to SDS-PAGE and analyzed by western blot using anti-FLAG, anti-GST, or anti-actin antibodies (n = 3). j GST pull-down experiment using co-transfection of cells with GST-tagged 14-3-3σ, 14-3-3σ mutant (σ-K49E), and full-length BAP1 followed by addition of GST- conjugated sepharose beads. The eluted whole-cell lysate was subjected to SDS-PAGE and analyzed by western blot using anti-FLAG, anti-GST, or anti-actin antibodies (n = 3) Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 11 of 16 Fig. 4 BAP1 binding to 14-3-3 releases Bax for inducing apoptosis. a Western blot analysis of 14-3-3 and actin expression in SK-N-Be2c, SK-N-SH, and SK-N-RA cells (n = 5). b Western blot analysis of 14-3-3 and actin expression in SK-N-Be2c transiently transfected with empty vector, full-length BAP1, or catalytically inactive mutant of BAP1 (BAP1-C91A) (n = 4). c, d SK-N-RA cells were transfected with FLAG-tagged full-length BAP1 or FLAG- empty vector. The lysates were immunoprecipitated with antibodies against 14-3-3 (c) or Bax (d) and probed with antibodies against 14-3-3 or Bax as indicated in the ﬁgure (n = 3). e Titration of BV02 in SK-N-RA cells for 24 h using three different concentrations (5, 20, and 100 μM). The percentage of apoptotic cells is presented as four independent experiments in triplicate (*p < 0.05). f SK-N-RA cells were transfected with control expression plasmid (white), transfected with full-length BAP1 expression plasmid (blue), or transfected with control expression plasmid and treated with 20 μM BV02 for 24 h (green). The percentage of apoptotic cells is presented as three independent experiments in triplicate (*p < 0.05). g Relative 14-3-3-ζ mRNA expression levels as fold changes measured using real-time reverse transcription PCR (qRT-PCR) of cDNA from transfected SK-N-RA cells with siRNA against 14-3-3-ζ or siRNA-GFP-Ctrl. The data are represented as three independent experiments (*p < 0.05). h The percentage of apoptotic cells in SK- N-RA cells transfected with siRNA against 14-3-3-ζ or siRNA-GFP-Ctrl in four independent experiments (*p < 0.05). i Model of BAP1-induced regulation of cell survival and cell death in neuroblastoma cells. BAP1 induces cell death via an interaction with 14-3-3 protein. The association between BAP1 and 14-3-3 releases the apoptotic inducer protein Bax from 14-3-3 and promotes cell death through the intrinsic (mitochondria) apoptosis pathway facilitated cell death (Fig. 4g, h). These results suggest that neuroblastoma cell growth in vivo. As expected, BAP1 BAP1 interaction with 14-3-3 prevents cell survival sig- expression was elevated in tumor cells infected with BAP1 naling, which is essential for growth of the neuroblastoma compared with control cells using IHC and western cells (Fig. 4i). blotting (Fig. 5e, g). In addition, the Bcl-2 expression in Next, we subcutaneously implanted 5.0 × 10 of control BAP1 tumor cells isolated from mice were reduced and BAP1 expressing neuroblastoma cells in nude mice compared to the control cells in line with the elevated and followed the growth of the tumor for 25 days. It was number of cleaved caspase-3 cells observed in BAP1 evident that the weight and mass of BAP1 stably expres- expressing tumor cells (Fig. 5e–g). Investigating the levels sing tumors were markedly reduced compared to control of cytochrome C in the tumor tissues isolated from ani- mice (Fig. 5a–d), conﬁrming the effects of BAP1 on mals revealed the downregulation of this protein in the Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 12 of 16 Fig. 5 BAP1 reduces tumor growth in vivo. a–d Nude mice were randomly divided into 2 groups of 10 mice after each being injected with 5 × 10 SK-N-BE2c cells stably expressing empty expression vector (Ctrl) or FLAG-tagged full-length BAP1. Tumor sizes were measured at the indicated time 3 2 points. Tumor volumes were estimated based on the length and width of the tumors: [mm ] = (length [mm]) × (width [mm]) × 0.5. (V = L × W2 × 0.5). The number of animals with certain tumor volume is presented in d. The mean tumor volume for each treatment group was graphed plus/ minus standard errors (*p < 0.05). e Immunohistochemistry analysis of tumor biopsies isolated from control (Ctrl) or BAP1 expressing neuroblastoma cells using H&E or antibodies against BAP1 (1:500), Bcl-2 (1:1000), and cleaved caspase-3 (Asp175, 1:500). Arrowheads indicate cleaved caspase-3 positive cells. f Quantiﬁcation of the number of cleaved caspase-3 (Asp175) positive cells in percentage visualized in the sections isolated from control or BAP1 expressing neuroblastoma cells (p < 0.05*). g Relative BAP1 expression and its effect on the expression of Bcl-2 and cytochrome C was determined by western blot from tumor tissue. Actin was used as the internal controls for western blotting control cells compared with BAP1 expressing tumors deposited at GEO database with accession numbers (Fig. 5g). GSM2394365, GSM2394387, GSM2394364, GSM2394375, GSM2394372, and GSM2394383, respec- BAP1 expression and copy number alterations in tively (GEO series: GSE89968). We used available log neuroblastoma patients and cell lines ratios and B allele frequencies to call copy number The analysis of gene expression array data of 102 neu- alterations with the R-Bioconductor package “copy roblastomas (Stage 4 and non-MYCN ampliﬁed) showed number” . The program was used to detect and remove that there is a trend toward the association between high outliers, impute missing values and to perform allele BAP1 and Bax expression and better relapse-free survival speciﬁc segmentation of SNP-array data. Log ratio of the patients (Fig. 6a, b). Furthermore, we evaluated thresholds for calling segments as gains or losses were set BAP1 gene expression levels in two data sets of NB at 0.2 and −0.16, respectively. Figure 6d shows the whole- patients provided with COG risk category annotation. It genome copy number alteration proﬁle for the afore- was found lower levels of BAP1 expression in high- mentioned cell lines. This analysis allowed us to conﬁrm risk individuals of both data sets compared to low-risk the well-known genomic features of the reported cell patients (Intermediate risk patients were excluded, lines, including the loss-of-chromosome 1p in IMR32 and Fig. 6c). Copy number alterations proﬁles for SK- the ampliﬁcation of MYCN in SK-N-DZ and IMR32. N-SH, SK-N-F1, IMR32, SK-N-DZ, SK-N-Be2, and SK- Figure 6e depicts a zoom-in of the copy number altera- N-Be2c cell lines were obtained from recently tions proﬁle for chromosome 3. We found that the cell 33,34 published genotyping proﬁles by using Illumina line SK-N-Be2 had a large deletion (17,639,767 base pairs HumanOmniExpress-24 v1.0 BeadChip (GPL21168). [bp], represented by 3870 probes on the chip) at Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 13 of 16 Fig. 6 BAP1 expression and copy number alterations in neuroblastoma patients and cell lines. a, b BAP1 (P = 0.14) and Bax (P = 0.229) expression is associated with good outcome. Kaplan–Meier analysis using published array data from 102 patients with stage 4, non-MYCN ampliﬁed neuroblastoma. c Box plots showing the Log2 transformed expression proﬁles of BAP1 NB cohorts divided by COG Risk category (Intermediate risk patients were excluded). Left panel: BAP1 expression in the TARGET-NBL data set. Right panel: BAP1 expression in GSE73517 data. Gene expression data were downloaded from R2: Genomics Analysis and Visualization Platform (http://r2.amc.nl) and plotted using the R environment. Statistical signiﬁcance was assessed by ANOVA test. d Whole-genome aberration plot of the six cell lines from the GEO series GSE89968 (in blue) are reported chromosomal losses and (in red) chromosomal gains, as well as the diploid regions in white. The black box highlights chromosome 3. e Zoomed-in view of chromosome 3 indicating altered regions. f Copy number alteration proﬁle for chromosome 3 of SK-N-Be2c, SK-N-Be2c, and SK-N-DZ. LogR: log ratio values, BAF: B allele frequencies. The plots in d, e show deletion of BAP1 locus chromosome 3p including BAP1 (Fig. 6f, upper panel) and 1019 samples, we selected those for which mutation shared with its descendant cell line SK-N-Be2c shown in data, copy number proﬁles, and gene expression data Fig. 6f (middle panel). Further, we detected a homozygous were available (thus, restricting the analysis to 877 deletion (Fig. 6f, lower panel), including chromosome samples). Subjects that showed an alteration of BAP1 3p21.1 of SK-N-DZ cell line as reported in Cancer Cell locus or its expression level numbered 148/877 or 17%. Line Encyclopedia data available at cBioportal website Copy number alteration proﬁle for the SK-N-SH (http://www.cbioportal.org). Here we used the Affymetrix cell line treated with retinoic acid (SK-N-SH-RA) was Genome-Wide SNP 6.0 Array to estimate genomic copy available at the GEO repository (GSM999325, series: number proﬁles and Affymetrix HG U133 Plus 2.0 plat- GSE40698) and at the UCSC Genome Browser as part of form to measure gene expression levels. Of the HAIB Genotype track set from ENCODE/HudsonAlpha Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 14 of 16 (http://genome.ucsc.edu/cgi-bin/hgFileUi?db=hg19&g= tested, BAP1 expression promoted the downregulation of wgEncodeHaibGenotype). No genomic alterations were Bcl-2. The Bcl-2 family consists of anti-apoptotic, pro- found for the SK-N-SH cell line and for the derived SK-N- apoptotic multi-domain, and pro-apoptotic BH3-only SH-RA cell line in chromosome 3p. proteins. Bcl-2 and Bcl-XL are well known anti-apoptotic family members; whereas, Bax and Bak are among the Discussion pro-apoptotic multi-domain proteins . In neuro- BAP1 is a tumor suppressor gene that functions as a blastoma, a large subset of patients harbor elevated levels deubiquitinating enzyme. The human BAP1 locus is of Bcl-2 compared to normal tissues and the silencing or located on chromosome 3p21.3. This region is commonly inhibition of Bcl-2 in neuroblastoma cell lines results in deleted or rearranged in many types of human cancer, apoptosis . Further, siRNA targeting of Bcl-2 in neuro- such as lung, breast, melanoma, and renal cell carci- blastoma causes a high level of cell apoptosis and a sig- 13,37 43 noma . The role of BAP1 in neuroblastoma is niﬁcant suppression of tumor growth . Previous studies unknown. In this study, we found that BAP1 was down- also demonstrated that high-risk neuroblastoma cell lines regulated both at the RNA and protein level in a subset of derived from human tumors with the poorest prognosis neuroblastoma cell lines. Large scale hemizygous loss-of- are dependent on Bcl-2 for survival . In addition, ele- chromosome 3p is a common event in neuroblastoma, vated Bcl-2 family protein expression was correlated to 38,39 12 and it is associated with tumors from older children . chemotherapy resistance . In correlation with these Recurrent deletions have been found at locus 3p21-p22 in previous ﬁndings, our data suggest that the expression of neuroblastoma and other common tumors, including BAP1 leads to reduced expression of pro-survival factors breast and lung cancer, indicating that the same tumor such as Bcl-2, which in turn reduces the survival potential suppressors are involved in tumorigenesis in these dif- of these cells. Indeed, in vivo xenograft studies using ferent tumors . Deletion of chromosome 3p was also BAP1 expressing cells showed reduced tumor growth non-randomly associated with deletion of chromosome compared to the control cells. Further, the tumors iso- 11q in neuroblastoma . Furthermore, loss-of- lated from BAP1 expressing cells harbored an elevated chromosomal material in chromosome arms 3p was dis- number of cleaved caspase-3 and increased the expression covered as a new high-risk subgroup of neuroblastoma of cytochrome C. In correlation to this result, the analyses stage 4 disease . To investigate the role of BAP1, we of neuroblastoma patient tumor materials revealed that overexpressed full-length BAP1 or BAP1-C91A in differ- the high BAP1 and Bax mRNA expression shows a trend ent neuroblastoma cell lines. The restoration of full- toward a better clinical outcome. Recent whole-genome length BAP1 but not BAP1-C91A expression in these cells sequencing studies of neuroblatoma did not ﬁnd genomic inhibited cell proliferation and promoted cell death sug- aberrations of the BAP1 locus . However, 3p deletion in gesting that deubiquitin activity of BAP1 is necessary for neuroblastoma has been reported in previous studies . the observed phenotype. More precisely, BAP1 increased Therefore, additional studies including treatment proto- the subG1 subpopulation of cells. Previous studies showed col data, protein expression, and larger sample sizes are that BAP1 is known as a key regulator of cell cycle pro- necessary to investigate the potential role of the genomic 22,25 gression through G1/S transition and that BAP1 aberrations of the BAP1 locus in driving the carcinogen- depletion in double-thymidine arrested HeLa cells leads esis of neuroblatoma. to delayed entry into the S phase, potentially as a result of LC–MS/MS analysis identiﬁed 14-3-3 proteins as novel an inability of cells to respond to DNA damage during this binding partners for BAP1. The intracellular localization phase . Furthermore, in NCI-H226, which are BAP1-null and scaffolding potential of 14-3-3 through direct inter- cells, overexpression of wild-type BAP1 promotes G1 action with their target protein is vital for regulating phase exit and subsequently speeds the transition to S intracellular signaling pathways . The 14-3-3 proteins phase . can prevent cell apoptosis via direct interaction with Bcl-2 Two different strategies were applied to analyze the family members including Bad and Bax. In contrast, mechanism of BAP1 inducing cell death in neuro- apoptotic stimulation, releases Bad or Bax from 14-3-3, blastoma. First, we investigated the expression of apop- and induces cytochrome C-mediated cell death . The tosis and survival target genes in cells expressing BAP1 isoform speciﬁcity in our mass spectrometry analysis compared to control cells. Second, we performed identiﬁed 14-3-3-σ,-ζ,-ε, and -β. Pull-down experiments LC–MS/MS analysis to identify binding partners for and confocal microscopy demonstrated direct interaction BAP1 that could propagate cell death signaling. Gene and co-localization in the nucleus. Testing the total levels expression proﬁling identiﬁed multiple genes in the cell of 14-3-3 in the neuroblastoma cells, we found no changes survival pathway that were downregulated in BAP1 in the cell lines harboring reduced levels of BAP1. In expressing cells. One of these target genes was the addition, we found no changes in the levels of 14-3-3 Bcl-2 protein. Independent of the neuroblastoma cell line ubiquitination in the absence or presence of full-length Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 15 of 16 Received: 5 December 2017 Revised: 15 March 2018 Accepted: 19 March BAP1 and BAP1-C91A. Instead, we found that BAP1 interaction with 14-3-3 releases the interaction between 14-3-3 and Bax, which in turn promotes cell death (Fig. 4i). The siRNA targeting of 14-3-3-ζ, or treating the cells with BV02, an inhibitor of 14-3-3 scaffolding protein References docking sites, mimicked the effect of BAP1-mediated 1. Maris, J. M. Recent advances in neuroblastoma. N. Engl. J. Med. 362,2202–2211 (2010). apoptosis. In general, the elevated levels of 14-3-3-ζ have 2. Brodeur, G. M. Neuroblastoma: biological insights into a clinical enigma. Nat. been shown to be associated with different types of can- Rev. Cancer 3,203–216 (2003). cer, including lung, breast, prostate, myeloma, glioma, 3. Matthay, K. K. et al. Neuroblastoma. Nat. Rev. Dis. Prim. 2, 16078 (2016). 4. Tomolonis, J. A.,Agarwal,S.&Shohet,J.M.Neuroblastoma pathogenesis: esophageal, head and neck, oral, pancreatic, ovarian, and deregulation of embryonic neural crest development. Cell Tissue Res. (2017). skin. In most of these studies, the expression of 14-3-3-ζ https://doi.org/10.1007/s00441-017-2747-0. was correlated with poor prognosis and chemoresis- 5. Tsubota, S. & Kadomatsu, K. Origin and initiation mechanisms of neuro- 46,48 blastoma. Cell Tissue Res. (2018). https://doi.org/10.1007/s00441-018-2796-z. tance . In neuroblastoma BH3 peptides containing the 6. Lee, G.-S. et al. Anti-cancer activity of the ethylacetate fraction from Orostachys domains of the pro-apoptotic BH3-only proteins Bid and japonicus for modulation of the signaling pathway in HepG2 human hepa- Bad, -induced apoptosis and demonstrated antitumor toma cells. Food Sci. Biotechnol. 23,269–275 (2014). 7. Ghobrial,I. M., Witzig,T.E.& Adjei, A. A. Targeting apoptosis pathways in efﬁcacy in an in vivo model . cancer therapy. Cancer J. Clin. 55,178–194 (2005). In summary, our present ﬁndings suggest that the 8. Gupta, S. Molecular signaling in death receptor and mitochondrial pathways reduced expression of BAP1 triggers neuroblastoma cells of apoptosis (Review). Int. J. Oncol. 22,15–20 (2003). 9. Elmore, S. Apoptosis: a review of programmed cell death. Toxicol. Pathol. 35, to survive and grow faster compared to the BAP1 495–516 (2007). expressing cells. The restoration of BAP1 expression in 10. Goldsmith,K.C.et al. MitochondrialBcl-2 family dynamics deﬁne neuroblastoma cells facilitated cell death mediated by therapy response and resistance in neuroblastoma. Cancer Res. 72,2565–2577 (2012). direct interaction with 14-3-3 and releasing Bax from this 11. Lamers, F. et al. Targeted BCL2 inhibition effectively inhibits neuroblastoma complex. Our ﬁndings might have important implications tumour growth. Eur. J. Cancer 48,3093–3103 (2012). for BAP1 to play a key role in the regulation of cell death 12. Goldsmith, K. C. & Hogarty, M. D. Targeting programmed cell death pathways with experimental therapeutics: opportunities in high-risk neuroblastoma. in cancer cells, and the expression of BAP1 could have Cancer Lett. 228, 133–141 (2005). prognostic implications in predicting a good prognosis in 13. Jensen, D. E. et al. BAP1: a novel ubiquitin hydrolase which binds to the BRCA1 cancer patients. RING ﬁnger and enhances BRCA1-mediated cell growth suppression. Onco- gene 16,1097–1112 (1998). 14. Nishikawa, H. et al. BRCA1-associated protein 1 interferes with BRCA1/BARD1 Funding RING heterodimer activity. Cancer Res. 69,111–119 (2009). This work was supported by the Swedish Cancer Foundation, the Magnus 15. Ventii, K. H. et al. BRCA1-associated protein-1 is a tumor suppressor that Bergvall Foundation, the Lindhes advokatbyrå, the Gunnar Nilsson Cancer requires deubiquitinating activity and nuclear localization. Cancer Res. 68, Foundation, the Gyllenstiernska Krapperup Foundation, Associazione Italiana 6953–6962 (2008). per la Ricerca sul Cancro (Grant no. 19255), and by funding from the European 16. Abdel-Rahman, M. H. et al. Germline BAP1 mutation predisposes to uveal Research Council (ERC), under the European Union’s Seventh Framework melanoma, lung adenocarcinoma, meningioma, and other cancers. J. Med. Programme for Research and Technology Development, grant agreement no. Genet. 48,856–859 (2011). , awarded to R.M. 17. Testa, J. R. et al. Germline BAP1 mutations predispose to malignant meso- thelioma. Nat. Genet. 43,1022–1025 (2011). 18. Murali, R., Wiesner, T. & Scolyer, R. A. Tumours associated with BAP1 mutations. Pathology 45,116–126 (2013). Author details 19. Dey, A. et al. Loss of the tumor suppressor BAP1 causes myeloid transfor- Department of Laboratory Medicine, Translational Cancer Research, Lund mation. Science 337,1541–1546 (2012). University, Medicon Village, Lund, Sweden. Department of Pharmacology and 20. Zarrizi, R., Menard, J. A., Belting, M. & Massoumi, R. Deubiquitination of γ- Emory Chemical Biology Discovery Center, Emory University School of tubulin by BAP1 prevents chromosome instability in breast cancer cells. Medicine, Atlanta, USA. Biotech Research and Innovation Centre, University of Cancer Res. 74, 6499–6508 (2014). Copenhagen, 2200 Copenhagen, Denmark. Department of Clinical Sciences 21. Yu, H. et al. The ubiquitin carboxyl hydrolase BAP1 forms a ternary complex Lund, Neurology, Lund University, Faculty of Medicine, Lund, Sweden. with YY1and HCF-1and is acritical regulator of gene expression. Mol. Cell Biol. Università degli Studi di Napoli Federico II, Dipartimento di Medicina 30,5071–5085 (2010). Molecolare e Biotecnologie Mediche, via Sergio Pansini 5, 80131 Naples, Italy. 22. Misaghi, S. et al. Association of C-terminal ubiquitin hydrolase BRCA1- CEINGE Biotecnologie Avanzate, Via G Salvatore 486, 80145 Naples, Italy. associated protein 1 with cell cycle regulator host cell factor 1. Mol. Cell Biol. IRCCS SDN, Istituto di Ricerca Diagnostica e Nucleare, Via Gianturco 113, 80143 29,2181–2192 (2009). Naples, Italy 23. Okino, Y., Machida, Y., Frankland-Searby, S. & Machida, Y. J. BRCA1-associated protein 1 (BAP1) deubiquitinase antagonizes the ubiquitin-mediated activa- Conﬂict of interest tion of FoxK2 target genes. J. Biol. Chem. 290, 1580–1591 (2015). The authors declare that they have no conﬂict of interest. 24. Doxsey, S. Re-evaluating centrosome function. Nat. Rev. Mol. Cell Biol. 2, 688–698 (2001). 25. Eletr,Z.M.& Wilkinson, K. D. An emerging modelfor BAP1’sroleinregulating Publisher's note cell cycle progression. Cell Biochem. Biophys. 60,3–11 (2011). Springer Nature remains neutral with regard to jurisdictional claims in 26. Pan, H. et al. BAP1 regulates cell cycle progression through E2F1 target genes published maps and institutional afﬁliations. and mediates transcriptional silencing via H2A monoubiquitination in uveal melanoma cells. Int. J. Biochem. Cell Biol. 60,176–184 (2015). Supplementary Information accompanies this paper at https://doi.org/ 27. Ismail, I. H. et al. Germline mutations in BAP1 impair its function in DNA 10.1038/s41419-018-0500-6. double-strand break repair. Cancer Res. 74,4282–4294 (2014). Ofﬁcial journal of the Cell Death Differentiation Association Sime et al. Cell Death and Disease (2018) 9:458 Page 16 of 16 28. Yu, H. et al. Tumor suppressor and deubiquitinase BAP1 promotes DNA 39. Vandesompele, J. et al. Unequivocal delineation of clinicogenetic subgroups double-strand break repair. Proc. Natl Acad. Sci. USA 111,285–290 (2014). and development of a new model for improved outcome prediction in 29. Eletr, Z. M., Yin, L. & Wilkinson, K. D. BAP1 is phosphorylated at serine 592 in S- neuroblastoma. J. Clin. Oncol. 23, 2280–2299 (2005). phase following DNA damage. FEBS Lett. 587, 3906–3911 (2013). 40. Hoebeeck, J. et al. High resolution tiling-path BAC array deletion 30. Pahlman, S., Ruusala, A. I., Abrahamsson, L., Mattsson, M. E. & Esscher, T. mapping suggests commonly involved 3p21-p22 tumor suppressor Retinoic acid-induced differentiation of cultured human neuroblastoma cells: genes in neuroblastoma and more frequent tumors. Int. J. Cancer 120, acomparisonwithphorbolester-induced differentiation. Cell Differ. 14, 533–538 (2007). 135–144 (1984). 41. Breen, C., O’Meara, A.,McDermott, M.,Mullarkey,M. & Stallings, R. Coordinate 31. Encinas, M. et al. Sequential treatment of SH-SY5Y cells with retinoic deletion of chromosome 3p and 11q in neuroblastoma detected by acid and brain-derived neurotrophic factor gives rise to fully differentiated, comparative genomic hybridization. Cancer Genet. Cytogenet. 120,44–49 neurotrophic factor-dependent, human neuron-like cells. J. Neurochem. 75, (2000). 991–1003 (2000). 42. Tomek, M., Akiyama, T. & Dass, C. R. Role of Bcl-2 in tumour cell survival and 32. Kobayashi, T., Masoumi, K. C. & Massoumi, R. Deubiquitinating activity of CYLD implications for pharmacotherapy. J. Pharm. Pharmacol. 64,1695–1702 (2012). is impaired by SUMOylation in neuroblastoma cells. Oncogene 34,2251–2260 43. Shen, M. et al. An MRI-visible non-viral vector for targeted Bcl-2 siRNA delivery (2015). to neuroblastoma. Int. J. Nanomed. 7,3319–3332 (2012). 33. Chresta,C.M., Masters, J. R. & Hickman, J. A. Hypersensitivity of human testi- 44. Molenaar, J. J. et al. Sequencing of neuroblastoma identiﬁes chromothripsis cular tumors to etoposide-induced apoptosis is associated with functional p53 and defects in neuritogenesis genes. Nature 483,589–593 (2012). and a high Bax:Bcl-2 ratio. Cancer Res. 56, 1834–1841 (1996). 45. Peifer, M. et al. Telomerase activation by genomic rearrangements in high-risk 34. Harenza, J. L. et al. Transcriptomic proﬁling of 39 commonly-used neuro- neuroblastoma. Nature 526,700–704 (2015). blastoma cell lines. Sci. Data 4, 170033 (2017). 46. Aghazadeh, Y. & Papadopoulos, V. The role of the 14-3-3 protein family in 35. Nilsen, G. et al. Copynumber: efﬁcient algorithms for single- and multi-track health,disease,and drug development. Drug Discov. Today 21,278–287 copy number segmentation. BMC Genom. 13, 591 (2012). (2016). 36. Barretina, J. et al. The Cancer Cell Line Encyclopedia enables predictive 47. Gardino, A. K. &Yaffe,M.B.14-3-3 proteins as signaling integration points for modelling of anticancer drug sensitivity. Nature 483, 603–607 (2012). cell cycle control and apoptosis. Semin. Cell Dev. Biol. 22, 688–695 (2011). 37. Harbour, J.W.etal. Frequent mutation of BAP1 in metastasizing uveal mel- 48. Tzivion, G., Gupta, V. S., Kaplun, L. & Balan, V. 14-3-3 proteins as potential anomas. Science 330,1410–1413 (2010). oncogenes. Semin. Cancer Biol. 16,203–213 (2006). 38. Spitz,R., Hero,B., Ernestus, K.& Berthold,F. Deletions in chromosome arms 3p 49. Goldsmith, K. C. et al. BH3 peptidomimetics potently activate apoptosis and and 11q are new prognostic markers in localized and 4s neuroblastoma. Clin. demonstrate single agent efﬁcacy in neuroblastoma. Oncogene 25,4525–4533 Cancer Res. 9,52–58 (2003). (2006). Ofﬁcial journal of the Cell Death Differentiation Association
Cell Death & Disease – Springer Journals
Published: Apr 24, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera