Targeted Therapy of TERT-Rearranged Neuroblastoma with BET Bromodomain Inhibitor and Proteasome Inhibitor Combination Therapy

High-risk neuroblastoma is a lethal pediatric cancer (1, 2). TERT gene rearrangements with transcriptional superenhancers results in chromatin remodeling, massive TERT gene overexpression, and neuroblastoma in approximately 24% of high-risk neuroblastomas. Patients with TERT-rearranged neuroblastoma show very poor prognosis (3, 4).

Transcriptional superenhancers are found at the loci of oncogenes, and are ideal targets for cancer therapy (5, 6). Superenhancers are occupied by master transcriptional regulators including the BET bromodomain protein BRD4. BRD4 binds superenhancers, activates superenhancer activity, and considerably upregulates the transcription and expression of superenhancer-associated critical oncogenes (5–7).

BET bromodomain inhibitors, such as JQ1, I-BET762, and OTX015, competitively bind and occupy the acetyl lysine recognition pocket of BRD4, disrupt BRD4 recruitment to superenhancers, suppress superenhancer-associated oncogene expression, and exert anticancer effects (6–9). Currently, several BET bromodomain inhibitors including I-BET762 and OTX015 are in multiple clinical trials in patients with solid tumors or hematologic malignancies (10, 11).

TERT is a component of telomerase which is essential for telomeric DNA maintenance. In addition, TERT regulates gene expression and protein synthesis in a telomerase-independent manner, resulting in cancer stemness and cell proliferation (12, 13). TERT overexpression thus induces tumorigenesis by telomerase activity–dependent and –independent mechanisms (14).

While telomerase inhibitors show anticancer effects in preclinical studies (15), treatment with Imetelstat (GRN163L), the only telomerase inhibitor which has been developed to clinical trials, does not improve the survival of patients with lung, breast, and brain cancer, and causes life-threatening side effects in patients with childhood brain cancer in clinical trials (16–18).

In this study, we examined the role of BRD4 in regulating TERT overexpression, cell-cycle progression, and proliferation in TERT-rearranged neuroblastoma cells. Carfilzomib was identified as the approved oncology drug exerting the best synergistic anticancer effects with OTX015 in TERT-rearranged neuroblastoma cells. OTX015 and carfilzomib synergistically blocked TERT protein expression, induced endoplasmic reticulum stress, induced TERT-rearranged neuroblastoma cell apoptosis, considerably suppressed TERT-rearranged neuroblastoma tumor progression, and improved mouse survival.

Human neuroblastoma GI-MEN (RRID:CVCL_1232), CHP134 (ECACC, catalog no. 06122002, RRID:CVCL_1124), and SHEP (RRID:CVCL_0524) cells were cultured in DMEM supplemented with 10% FCS. CLB-GA (RRID:CVCL_9529) cells were grown in RPMI medium with 10% FCS. WI38 fetal lung fibroblasts (RRID:CVCL_0579) were cultured in minimum essential medium with 10% FCS. GI-MEN, CLB-GA, and SHEP cells were provided by Jo Vandesompele (Center for Medical Genetics Ghent, Ghent, Belgium), Valérie Combaret (Centre Léon Bérard, Lyon, France), and Barbara Spengler (Fordham University, New York, NY), respectively, in 2016 and 20 years ago. CHP134 cells were obtained from the European Collection of Cell Cultures in 2010, and WI38 cells from the ATCC 20 years ago. Umbilical cord blood for research was provided by the Sydney Cord Blood Bank and experiments were approved by the Sydney Children's Hospital Human Research Ethics Committee. Mononuclear cells were purified from umbilical cord blood using Lymphoprep (Axis-Shield). CD34⁺ cells were isolated to greater than 95% purity and cultured in Stemline II serum-free medium (Sigma). All cell lines were confirmed to be Mycoplasma free by PCR and authenticated by short tandem repeat profiling by Garvan Institute of Medical Research (Darlinghurst, Australia) or Cellbank Australia between 2012 and 2020. Cell lines were used for experiments no more than 6 months after thawing. Patient-derived xenograft (PDX) CCI-NB07-RMT cells were extracted from mouse tumors and cultured in Iscove's modified Dulbecco's medium with 20% FCS and insulin-transferrin-sodium (Thermo Fisher Scientific).

Drug screening was performed with the previously described methods (19). CLB-GA neuroblastoma cells were treated with vehicle control, I-BET762 at inhibition of cell viability by 20% (IC₂₀, 0.5 μmol/L), the Approved Oncology Drugs (AOD) Set IV (the U.S. NCI) compounds at 1 μmol/L, or combination for 72 hours. Cell viability was determined by Alamar blue assays. Synergistic interaction between I-BET762 and the AODs was examined by R value using fractional product method (20).

Compounds which reduced the number of viable CLB-GA cells by ≥90% on their own and compounds which reduced the number of viable CLB-GA cells by ≥80% when combined with I-BET762 with an R value of less than 0.7 were shortlisted for secondary drug screening. CLB-GA and GI-MEN neuroblastoma cells were treated with a range of doses of I-BET762 or OTX015 (0, 125, 250, 500 and 1,000 nmol/L), the shortlisted AODs (0, 3.90625, 7.8125, 15.625, 31.25, 62.5, 125, 250, 500, 1, 000 nmol/L), or combination. To determine whether the effects of I-BET762/OTX015 and the AOD were additive, bliss additivity was calculated with Bliss-additive formula (21). Synergy was further validated by the Chou–Talalay method (22) to calculate combination indexes (CI) for effective doses for 75% (ED75) and 90% (ED90) cell number reduction with CompuSyn software (Combosyn Inc, http://www.combosyn.com/).

ChIP sequencing was performed as we described previously (23, 24). PDX CCI-NB07-RMT neuroblastoma cells were treated with vehicle control or 2 μmol/L OTX015 for 48 hours, followed by ChIP with the ChIP Assay Kit (17-295, Millipore) and a rabbit anti-H3K27ac antibody (Abcam, catalog no. ab4729, RRID:AB_2118291), rabbit anti-BRD4 antibody (Bethyl, catalog no. A301-985A, RRID:AB_1576498) or control rabbit IgG (Thermo Fisher Scientific, catalog no. 10500C, RRID:AB_2532981). After immunoprecipitation, DNA was purified and subjected to sequencing with Illumina HiSeq 2000 at Ramaciotti Centre for Genomics (University of New South Wales, Sydney, New South Wales, Australia). The ChIP sequencing data have been deposited at the Gene Expression Omnibus (GEO) website(GSE147181, GEO, RRID:SCR_005012). In addition, published ChIP sequencing data from BE(2)-C and CHP134 cells with H3K27ac, H3K4me3, and BRD4 antibodies were downloaded from GEO websites (GSM2113517, GSM2113518, GSM2113520, and series GSE113139, GEO, RRID:SCR_005012) and also analyzed. For the CHP134 cell line, H3K4me3 data were available which was used to filter out potential promoter peaks. In all other cell lines, only H3K27ac was used to perform superenhancer detection. Peaks identified were used for superenhancer detection with ROSE (5) using parameters -s 12500 -t 1000. HOMER (HOMER, RRID:SCR_010881; ref. 25) and R Bioconductor packages (org.Hs.eg.db and TxDb.Hsapiens.UCSC.hg19.knownGene, Bioconductor, RRID:SCR_006442) were used for peak annotation. Genes whose transcription start site fell within 600 kbp flanking of detected superenhancers were considered candidate-associated genes.

Animal experiments were approved by the Animal Care and Ethics Committee of the University of New South Wales (Sydney, New South Wales, Australia). Female Balb/c nude mice (MGI, catalog no. 5652590, RRID:MGI:5652590) ages 5 to 6 weeks were injected subcutaneously into the right flank under anesthesia with GI-MEN, wild-type CLB-GA cells, or CLB-GA cells stably transfected with an empty vector or TERT open reading frame (ORF) expression construct per mouse (26). In addition, nonobese diabetic severe combined immune deficiency gamma (NSG) mice (IMSR, catalog no. ARC:NSG, RRID:IMSR_ARC:NSG) ages 5 to 6 weeks old were injected with 2 × 10⁶ PDX CCI-NB07-RMT cells. When the engrafted tumors reached 200 or 100 mm³, the mice were randomly divided into four groups and treated with vehicle control, OTX015 at 50 mg/kg body weight/day via oral gavage, carfilzomib at 6 mg/kg body weight/once every other day via intraperitoneal injection, or OTX015 plus carfilzomib. The treatments were continued until the mice were humanely culled when the tumors reached 1,000 mm³, and survival curves were plotted. Tumor tissues were collected, snap frozen or formalin fixed, and paraffin embedded for immunoblot, telomere length assays, telomere dysfunction-induced foci assay, or IHC analysis, respectively.

For statistical analysis, experiments were conducted three times. Data were analyzed with Prism 6 software (PRISM, RRID:SCR_005375 and GraphPad Prism, RRID:SCR_002798) and presented as mean ± SE. Differences were analyzed for significance using ANOVA among groups or unpaired Student t test for two groups. All statistical tests were two sided. A P value of less than 0.05 was considered statistically significant. Synergy or additivity was calculated by CI method for combinations of multiple doses of drugs, or by the fractional product (R) method for combinations of a single dose of drugs.

We first examined the effect of TERT siRNAs on TERT-rearranged neuroblastoma cell proliferation and cell-cycle progression. RT-PCR and immunoblot analyses confirmed that two independent TERT siRNAs, TERT siRNA-1 and TERT siRNA-2, effectively knocked down TERT mRNA and protein expression in TERT-rearranged GI-MEN and CLB-GA neuroblastoma cells (Supplementary Fig. S1A and S1B; ref. 3). Telomerase activity and Alamar blue assays showed that transfection with TERT siRNAs significantly reduced telomerase activity (Supplementary Fig. S1C) and the numbers of GI-MEN and CLB-GA cells (Supplementary Fig. S1D). Cell-cycle analysis revealed that TERT knockdown for 72 hours significantly increased the percentage of GI-MEN and CLB-GA cells at the G₁ phase and decreased the percentage of the cells at the S phase (Supplementary Fig. S1E and S1F).

Next, immunoblot analysis confirmed that transfection with wild-type or D712A-mutant telomerase activity-deficient TERT ORF expression construct (27) led to considerable TERT protein overexpression in GI-MEN and CLB-GA cells (Supplementary Fig. S1G). Cell-cycle analysis showed that both the wild-type and the D712A-mutant TERT expression construct rescued the TERT-rearranged neuroblastoma cells from G₁ cell-cycle arrest and growth inhibition due to TERT siRNA-2 which targeted the 3′-untranslated region of TERT mRNA (Supplementary Fig. S1H and S1I). As TERT is known to induce cancer cell proliferation by enhancing global protein synthesis (13), our puromycin incorporation assays showed that TERT knockdown with TERT siRNAs significantly reduced global protein synthesis (Supplementary Fig. S1J). The data suggest that TERT promotes global protein synthesis and thus induces TERT-rearranged neuroblastoma cell proliferation and cell-cycle progression.

We examined whether BRD4 regulated TERT oncogene expression in TERT-rearranged neuroblastoma cells. RT-PCR and immunoblot analyses showed that transfection with two independent BRD4 siRNAs effectively knocked down BRD4 mRNA and protein expression and reduced TERT mRNA and protein expression 48 hours after siRNA transfection (Supplementary Fig. S2A and S2B). In comparison, transfection with BRD4 siRNAs for 48 hours reduced N-Myc but not TERT mRNA and protein expression in MYCN-amplified CHP134 neuroblastoma cells (Supplementary Fig. S2C and S2D).

GI-MEN and CLB-GA cells stably expressing doxycycline-inducible control short hairpin RNA (shRNA) or one of two independent BRD4 shRNAs (shRNA-1 and shRNA-2) were established using the FH1tUTG construct (28, 29). The BRD4 siRNAs and shRNAs targeted different BRD4 mRNA sequences. RT-PCR and immunoblot analyses showed that treatment with doxycycline considerably reduced BRD4 and TERT mRNA and protein expression in doxycycline-inducible BRD4 shRNA, but not control shRNA, GI-MEN, and CLB-GA cells (Fig. 1A and B; Supplementary Fig. S2E). Furthermore, BRD4 knockdown by doxycycline significantly reduced telomerase activity in doxycycline-inducible BRD4 shRNA-1 and BRD4 shRNA-2 GI-MEN and CLB-GA cells (Supplementary Fig. S2F). The data demonstrate that BRD4 is required for TERT expression and telomerase activity in TERT-rearranged neuroblastoma cells.

We next examined whether BRD4 is required for TERT-rearranged neuroblastoma cell proliferation and cell-cycle progression. Alamar blue assays showed that knocking down BRD4 with siRNAs reduced the number of TERT-rearranged neuroblastoma cells (Supplementary Fig. S2G). We then established GI-MEN and CLB-GA cells stably transfected with an empty vector or a TERT ORF expression construct. The exogenous TERT ORF expression was controlled by a viral promoter and was therefore not expected to respond to BRD4. Immunoblot analysis confirmed that BRD4 siRNAs downregulated TERT protein expression in empty vector–expressing cells, but did not have an effect on TERT protein expression in TERT ORF construct expressing cells (Fig. 1C). Alamar blue assays demonstrated that exogenous TERT ORF expression induced neuroblastoma cell proliferation in both GI-MEN and CLB-GA cells, and largely reversed cell growth inhibition mediated by BRD4 siRNAs (Fig. 1D).

In addition, Alamar blue assays showed that treatment with doxycycline significantly reduced the number of doxycycline-inducible BRD4 shRNA, but not control shRNA, GI-MEN, and CLB-GA cells (Supplementary Fig. S2H). Cell-cycle analysis revealed that BRD4 knockdown increased the proportion of doxycycline-inducible BRD4 shRNA GI-MEN and CLB-GA cells at the G₁ phase and decreased the proportion of the cells at the S phase (Fig. 1E), and clonogenic assays showed that BRD4 knockdown considerably reduced the numbers of colonies formed by GI-MEN and CLB-GA cells (Fig. 1F and G).

Taken together, these data demonstrate that TERT promotes TERT-rearranged neuroblastoma cell proliferation and cell-cycle progression through telomerase-independent mechanism including enhancing protein synthesis, and that BRD4 promotes TERT-rearranged neuroblastoma cell proliferation and cell-cycle progression through modulating TERT expression.

As BET bromodomain inhibitors I-BET762 and OTX015 block BRD4 function and are promising anticancer agents in clinical trials (10, 11), we examined whether BET bromodomain inhibitors showed anticancer effects against TERT-rearranged neuroblastoma cells. Alamar blue assays showed that treatment with the BET bromodomain inhibitors JQ1, I-BET762 (GSK525762), or OTX015 reduced the number of viable GI-MEN and CLB-GA cells; however, the anticancer effect plateaued at higher doses (Supplementary Fig. S3A).

To identify the anticancer agents that synergize with I-BET762, we undertook a chemical library screen using the AODs Set IV from the U.S. NCI. CLB-GA cells were treated with 101 AODs at 1 μmol/L either alone or in combination with 0.5 μmol/L I-BET762 which reduced the number of CLB-GA cells by 20% on its own. As single agents, 11 AODs at 1 μmol/L reduced CLB-GA cell viability by ≥ 90% (Supplementary Fig. S3B; Supplementary Table S1) and were therefore subjected to multiple-dosage secondary drug screening. When combined with I-BET762, 5 of the remaining 90 AODs showed synergistic anticancer effects with R < 0.7 (fractional product method), and the combination therapies reduced the number of CLB-GA cells by more than 80% (Supplementary Fig. S3C; Supplementary Table S1).

The 11 compounds which reduced cell viability by ≥90% on their own and the five compounds which synergized with I-BET762 were subjected to a secondary drug screen with multiple dosages of I-BET762, AODs, or combination in GI-MEN and CLB-GA cells. Three classes of AODs were found to exert strong synergy with I-BET762 with CIs of <0.7: proteasome inhibitors including carfilzomib and bortezomib, DNA topoisomerase II inhibiting and DNA intercalating chemotherapy agents, and tubulin inhibitor ixabepilone which was less effective (Fig. 2A). However, Alamar blue assays showed that combination therapy with I-BET762 and the chemotherapy agents at the same dosages as used in neuroblastoma cells induced cytotoxicity in embryonic fibroblast WI38 cells (Supplementary Fig. S4A). In contrast, combination therapy with I-BET762 and the proteasome inhibitors carfilzomib or bortezomib did not show toxicity to the normal cells (Supplementary Fig. S4B).

The synergistic anticancer effects between the proteasome inhibitors carfilzomib and bortezomib and the BET bromodomain inhibitors I-BET762 and OTX015 were further examined. GI-MEN and CLB-GA cells were treated with vehicle control, various dosages of carfilzomib or bortezomib, various dosages of I-BET762 or OTX015, or combinations, followed by Alamar blue assays. Synergy/additivity analysis using the Bliss-additivity model and the CI method showed the strongest synergy by carfilzomib and OTX015 combination therapy over other combinations (Fig. 2B and C; Supplementary Fig. S5).

GI-MEN and CLB-GA cells were treated with vehicle, OTX015, carfilzomib, or combination, followed by staining with Annexin-V and flow cytometry analysis of apoptosis. While treatment with OTX015 or carfilzomib alone slightly increased the percentage of cells positively stained with Annexin-V, OTX015 and carfilzomib significantly and synergistically induced apoptosis (R = 0.28 for GI-MEN cells and R = 0.56 for CLB-GA cells, fractional product method; Fig. 2D). Importantly, treatment of the normal embryonic fibroblast WI38 and human CD34⁺ stem/progenitor cells with vehicle control, OTX015, carfilzomib, or combination did not induce apoptosis (Fig. 2D).

We next examined whether BRD4 siRNA, like the BET bromodomain BRD4 inhibitors, also synergized with carfilzomib. Flow cytometry analysis showed that BRD4 siRNA and carfilzomib also synergistically induced GI-MEN and CLB-GA cell apoptosis (Fig. 2E).

Taken together, the data confirm that the proteasome inhibitor carfilzomib is the AOD exerting considerable synergistic anticancer effects with BET bromodomain inhibitors against TERT-rearranged neuroblastoma cells, and that OTX015 and carfilzomib synergistically induce TERT-rearranged neuroblastoma cell apoptosis with little toxicity against normal cells.

BET bromodomain inhibitors are well known to suppress oncogene expression (5, 6, 8), and proteasome inhibitors regulate protein expression (30). We next examined whether OTX015 and carfilzomib synergistically regulated TERT expression, telomerase activity, and telomere length in TERT-rearranged neuroblastoma cells. RT-PCR and immunoblot analyses showed that OTX015, but not carfilzomib, reduced TERT mRNA and protein expression in both cell lines, that OTX015 and carfilzomib did not co-operatively reduce TERT mRNA expression in CLB-GA cells, and that OTX015 and carfilzomib co-operatively reduced TERT protein expression in both of the cell lines (Fig. 3A and B). Telomerase activity and telomere length assays showed that OTX015 or carfilzomib alone reduced telomerase activity, and that OTX015 and carfilzomib significantly and synergistically decreased telomerase activity (R = 0.78 for GI-MEN cells and R = 0.49 for CLB-GA cells, fractional product method), but showed no effect on telomere length (Fig. 3C and D).

TERT protein is targeted for ubiquitination and degradation by DYRK2-mediated EDD-DDB1-VprBP E3 ligase complex activation (31). Our immunoblot data showed that OTX015 and carfilzomib synergistically upregulated DYRK2 protein expression in GI-MEN and CLB-GA cells (Fig. 3E). Importantly, immunoblot analysis showed that DYRK2 knockdown blocked OTX015 and carfilzomib combination therapy–mediated TERT protein reduction (Fig. 3F).

To examine whether OTX015 and carfilzomib exerted synergistic anticancer effects through reducing TERT protein expression, we treated GI-MEN and CLB-GA cells stably transfected with an empty vector or a TERT ORF expression construct with vehicle control, OTX015, carfilzomib, or combination. The exogenous TERT ORF expression was controlled by a viral promoter without superenhancers. Treatment with OTX015 and carfilzomib showed no effect on TERT protein expression in GI-MEN and CLB-GA cells transfected with the TERT ORF expression construct, but synergistically reduced TERT protein expression in empty vector cells (Fig. 3G). Flow cytometry analysis showed that OTX015 and carfilzomib significantly increased the proportion of empty vector cells undergoing apoptosis, and that forced overexpression of TERT ORF significantly reduced this effect (Fig. 3H).

Taken together, the data demonstrate that OTX015 and carfilzomib synergistically induce TERT-rearranged neuroblastoma cell apoptosis partly through synergistically reducing TERT protein expression.

Treatment with the proteasome inhibitor bortezomib results in upregulation of Nrf2 protein which suppresses endoplasmic reticulum stress, leading to cancer cell resistance (32, 33). We have reported that treatment with the BET bromodomain inhibitor JQ1 blocks Nrf2 pathway activation and enhances cancer cell death (34). We therefore examined whether OTX015 and carfilzomib synergistically induced TERT-rearranged neuroblastoma cell death partly by blocking Nrf2 expression and synergistically inducing endoplasmic reticulum stress.

Immunoblot analysis showed that carfilzomib considerably upregulated the expression of Nrf2 and its canonical targets HMOX1 and NQO1, and that OTX015 blocked carfilzomib-induced Nrf2, HMOX1, and NQO1 upregulation in GI-MEN and CLB-GA cells (Fig. 4A). In addition, OTX015 alone or carfilzomib alone did not have an effect, but OTX015 and carfilzomib combination therapy considerably activated the expression of CHOP, ATF4, and phosphorylated eIF2a proteins (Fig. 4A), the markers for endoplasmic reticulum stress (35), and activated oxidative stress (Supplementary Fig. S6A), demonstrating that OTX015 and carfilzomib synergistically induce endoplasmic reticulum stress and oxidative stress.

Next, our immunoblot analysis confirmed that Nrf2 siRNAs blocked carfilzomib-induced upregulation of Nrf2, its targets HMOX1 and NQO1, and similar to OTX015, considerably activated the expression of the endoplasmic reticulum stress marker proteins CHOP, ATF4, and phosphorylated eIF2a (Fig. 4B). Alamar blue assays showed that combination of carfilzomib and Nrf2 siRNAs, similar to OTX015, synergistically reduced the number of viable GI-MEN and CLB-GA cells (Fig. 4C), and that combination of CHOP siRNAs and ATF4 siRNAs partly rescued cells from OTX015 and carfilzomib combination-induced cytotoxicity (Supplementary Fig. S6B).

GI-MEN and CLB-GA cells were then transfected with a construct expressing empty vector or Nrf2 (34), followed by treatment with vehicle control, OTX015, carfilzomib, or combination. Alamar blue assays showed that Nrf2 overexpression significantly suppressed the anticancer effect of OTX015 and carfilzomib combination therapy (Fig. 4D). In addition, Alamar blue assays showed that treatment with the endoplasmic reticulum stress inhibitor salubrinal or the oxidative stress inhibitor N-acetyl-l-cysteine (36, 37) significantly suppressed the anticancer effect of OTX015 and carfilzomib combination therapy (Fig. 4E; Supplementary Fig. S6C).

Taken together, the data suggest that treatment with carfilzomib leads to Nrf2 protein overexpression, Nrf2 pathway activation and resistance to endoplasmic reticulum stress, that OTX015 blocks the effects, and that OTX015 and carfilzomib exert synergistic anticancer effects partly by synergistically inducing endoplasmic reticulum stress.

No targeted therapy against TERT-rearranged neuroblastoma has been tested in clinical trials in patients. We investigated the anticancer efficacy of OTX015 and carfilzomib combination therapy in two xenograft models of TERT-rearranged neuroblastoma. GI-MEN and CLB-GA cells were each engrafted into 48 mice, which were randomized into four treatment groups (vehicle, OTX015, carfilzomib, or combination of OTX015 and carfilzomib) when tumors reached 0.2 cm³. While 44 of 48 mice in the CLB-GA group developed tumors 3 weeks after xenografting, only 15 of 48 mice in the GI-MEN group developed tumors 6 months after xenografting. Treatment with OTX015 or carfilzomib alone inhibited neuroblastoma growth, compared with vehicle control, in mice xenografted with CLB-GA or GI-MEN cells. In combination, OTX015 and carfilzomib caused considerable tumor growth inhibition and initial tumor regression (Fig. 5A and B). Immunoblot analysis of mouse tumor tissues revealed that carfilzomib did not alter TERT protein expression, while OTX015 modestly reduced TERT protein expression. In combination, OTX015 and carfilzomib abolished TERT protein expression in both GI-MEN and CLB-GA cell xenograft tumors (Fig. 5C). In addition, OTX015 and carfilzomib combination therapy synergistically and considerably induced apoptosis in tumors, as characterized by significantly increased proportion of tumor cells positively stained by TUNEL (R = 0.42 for CLB-GA xenografts and R = 0.33 for GI-MEN xenografts, fractional product method; Fig. 5D and E).

Taken together, the data demonstrate that OTX015 and carfilzomib synergistically block TERT protein expression, induce tumor cell apoptosis, suppress tumor progression, and improve survival in mice with established TERT-rearranged neuroblastoma tumors.

To examine whether OTX015 and carfilzomib exert synergistic anticancer effects against TERT-rearranged neuroblastoma in vivo through blocking TERT protein expression, we xenografted CLB-GA cells stably transfected with an empty vector or TERT ORF expression construct into nude mice. In mice xenografted with CLB-GA cells stably transfected with an empty vector, OTX015 and carfilzomib considerably and synergistically reduced tumor progression and improved mouse survival (Fig. 6A). In contrast, in mice xenografted with CLB-GA cells stably transfected with a TERT ORF expression construct, the anticancer effect of OTX015 and carfilzomib combination therapy was substantially reversed (Fig. 6B). Immunoblot analysis showed that OTX015 and carfilzomib synergistically blocked TERT protein expression in mice xenografted with empty vector CLB-GA cells, but not TERT-ORF CLB-GA cells (Fig. 6C). TUNEL assays demonstrated that OTX015 and carfilzomib synergistically and considerably induced apoptosis in tumor tissues from the mice xenografted with empty vector CLB-GA cells, and that the effect was significantly reduced in tumor tissues from the mice xenografted with TERT-ORF CLB-GA cells (Fig. 6D). In addition, in tumors from the mice xenografted with empty vector CLB-GA cells, OTX015 and carfilzomib did not show co-operative effect on telomere length (Fig. 6E), but synergistically induced DNA damage response at telomere, as revealed by positive γ-H2AX staining (Fig. 6F).

Taken together, the data demonstrate that OTX015 and carfilzomib exert considerable anticancer effects against TERT-rearranged neuroblastoma in vivo through TERT-dependent but telomerase activity-independent mechanisms including DNA damage.

We have established PDX TERT-rearranged neuroblastoma cells, CCI-NB07-RMT cells, with the massive deletion of chromosome 5:50791668 – 5:1295647 (20). RT-PCR confirmed that TERT mRNA expression was considerably higher in PDX CCI-NB07-RMT cells than TERT-rearranged or MYCN-amplified neuroblastoma cell lines, while N-Myc was hardly detectable (Supplementary Fig. S7A). ChIP sequencing showed an enhancer region rearranged to the TERT gene due to chromosome 5:50791668 – 5:1295647 deletion, and BRD4 protein binding to the enhancer in PDX CCI-NB07-RMT cells (Supplementary Fig. S7B). In comparison, the enhancer existed at the same locus in TERT-nonrearranged/MYCN-amplified BE(2)-C but not CHP134 cells, and the enhancer was >49 Mb away from the TERT gene in BE(2)-C cells (Supplementary Fig. S7B). In addition, after PDX CCI-NB07-RMT cells were treated with OTX015, ChIP PCR showed that BRD4 protein binding at the enhancer region was dramatically reduced (Supplementary Fig. S7C).

PDX CCI-NB07-RMT cells extracted from tumors from nonobese diabetic severe combined immune deficiency gamma (NSG) mice were xenografted into new groups of NSG mice, and treated with vehicle control, OTX015, carfilzomib, or combination. As shown in Supplementary Fig. S7D, OTX015 and carfilzomib synergistically suppressed tumor progression and improved mouse overall survival. Immunoblot analysis of tumor tissues from the mice showed that OTX015 and carfilzomib synergistically and significantly diminished TERT protein expression (Supplementary Fig. S7E). The data further demonstrate OTX015 and carfilzomib combination as an effective therapeutic strategy for TERT-rearranged neuroblastoma.

BET bromodomain inhibitors exert anticancer effects by displacing BRD4 from superenhancers, leading to transcriptional silencing of target oncogenes, such as MYCN, MYC, and BCL2 (5, 6, 8). In this study, we have found that BRD4 knockdown or BET bromodomain inhibitor treatment significantly reduces TERT mRNA and protein expression as well as telomerase activity in TERT-rearranged neuroblastoma cells, leading to TERT-rearranged neuroblastoma cell-cycle arrest at the G₁ phase and growth inhibition.

BET bromodomain inhibitor monotherapy does not cause tumor regression in mice or patients (9–11). We have previously identified antimicrotubule drugs as the AODs exerting the best synergistic anticancer effects with BET bromodomain inhibitors against MYCN-amplified neuroblastoma cells (34). In this study, we have identified the proteasome inhibitors bortezomib and carfilzomib as the AODs exerting the best synergistic anticancer effects with BET bromodomain inhibitors against TERT-rearranged neuroblastoma cells with little toxicity to normal cells, and demonstrated that OTX015 and carfilzomib exert more effective anticancer effects than I-BET762 and bortezomib combinations. This is consistent with our observation that OTX015 is slightly more effective than I-BET762 as monotherapy, and consistent with the literature that the second-generation proteasome inhibitor carfilzomib shows better anticancer efficacy than the first-generation proteasome inhibitor bortezomib (38, 39). While BET bromodomain inhibitors mainly suppress BRD4 but also show weak effect on other BRD proteins (40, 41), we have confirmed that BRD4 siRNAs, similar to OTX015, exert synergistic and significant anticancer effects with carfilzomib against TERT-rearranged neuroblastoma cells. The data suggest that BET bromodomain inhibitors exert synergistic anticancer effects with carfilzomib mainly through blocking BRD4 function.

Cancer cells develop resistance due to proteasome inhibitor-mediated Nrf2 upregulation (33). In this study, we have found that carfilzomib activates the Nrf2 pathway and suppresses endoplasmic reticulum stress, that OTX015 blocks carfilzomib-mediated Nrf2 pathway activation, and that combination therapy with OTX015 and carfilzomib synergistically induces endoplasmic reticulum stress and cell death in TERT-rearranged neuroblastoma cells. In addition, cell death due to OTX015 and carfilzomib combination therapy is partly blocked by Nrf2 overexpression or treatment with an endoplasmic reticulum stress inhibitor. These data suggest that OTX015 and carfilzomib induce TERT-rearranged neuroblastoma cell death partly by regulating Nrf2 pathway and causing endoplasmic reticulum stress.

TERT is targeted for ubiquitination and degradation through DYRK2-mediated EDD-DDB1-VprBP E3 ligase activation (31). In this study, we have found that OTX015 and carfilzomib synergistically block TERT protein but not mRNA expression and upregulate DRYK2 expression, and that DYRK2 knockdown blocks OTX015 and carfilzomib-modulated TERT protein reduction. The data suggest that OTX015 and carfilzomib synergistically reduce TERT protein by upregulating DYRK2, and that upregulating DYRK2 is a novel therapeutic approach for TERT-rearranged neuroblastoma. We have also demonstrated that forced TERT overexpression largely blocks OTX015 and carfilzomib-induced TERT-rearranged neuroblastoma cell apoptosis in vitro. Importantly, OTX015 and carfilzomib combination therapy considerably blocks TERT protein expression, induces tumor cell apoptosis, suppresses TERT-rearranged neuroblastoma tumor progression, and improves survival in mice xenografted with TERT-rearranged neuroblastoma GI-MEN, CLB-GA cell lines, or PDX cells, and forced TERT overexpression in neuroblastoma cells largely blocks the anticancer effects of OTX015 and carfilzomib combination therapy in mice. In addition, OTX015 and carfilzomib combination therapy is more effective in mice xenografted with GI-MEN cells, presumably because GI-MEN cell tumors progress slower than CLB-GA and PDX cell tumors. Taken together, the data demonstrate that OTX015 and carfilzomib exert synergistic anticancer effects against TERT-rearranged neuroblastoma partly by reducing TERT protein expression.

In addition to telomere maintenance, TERT promotes tumorigenesis by telomere-independent mechanisms, including enhancing global protein synthesis (13). In this study, we have found that TERT knockdown leads to a reduction in global protein synthesis in TERT-rearranged neuroblastoma cells, and that forced overexpression of wild-type or telomerase-deficient TERT both blocks TERT or BRD4 knockdown-induced neuroblastoma cell growth inhibition and cell-cycle arrest. In addition, while OTX015 and carfilzomib synergistically blocks TERT protein expression and suppresses telomerase activity, forced overexpression of wild-type or telomerase-deficient TERT both block the anticancer effects of the combination therapy. These data indicate a noncanonical function of TERT in TERT-rearranged neuroblastoma.

In summary, BRD4 is required for TERT oncogene overexpression, cell-cycle progression, and proliferation in TERT-rearranged neuroblastoma cells. The proteasome inhibitor carfilzomib is the agent exerting the best synergistic anticancer effects with the BET bromodomain inhibitor OTX015 against TERT-rearranged neuroblastoma cells. OTX015 and carfilzomib synergistically induce TERT-rearranged neuroblastoma cell apoptosis by blocking TERT protein expression and inducing endoplasmic reticulum stress, and substantially suppress tumor progression and improve survival in mice xenografted with TERT-rearranged neuroblastoma cell lines or PDX cells (Supplementary Fig. S7F). Although OTX015 itself has not been tested in children, OTX015 is currently in phase II clinical trials in adult patients with cancer and the BET bromodomain inhibitor BMS-986158 is now in a phase I clinical trial in pediatric patients with cancer (ClinicalTrials.gov Identifier: NCT03936465). Carfilzomib is an approved oncology drug and is currently in phase I clinical trials in pediatric patients with cancer (ClinicalTrials.gov Identifiers: NCT02303821 and NCT02512926). OTX015 and carfilzomib combination treatment is likely to be translated into the first clinical trial of targeted therapy against TERT-rearranged neuroblastoma in patients.

J. Chen reports grants from Cancer Council NSW and Tour de Cure Foundation during the conduct of the study. P. Polly reports grants from Cancer Council NSW and Tour de Cure Foundation during the conduct of the study. R.R. Reddel reports personal fees from Tessellate Bio and other from Komipharm International Ltd outside the submitted work; in addition, R.R. Reddel has a patent for C-circle assay licensed. M. Fischer reports personal fees from Bayer, Novartis, BMS, Janssen-Cilag, and Roche outside the submitted work. H.A. Pickett reports other from Tessellate Bio outside the submitted work. T. Liu reports grants from Cancer Council NSW and Tour de Cure Foundation during the conduct of the study. No disclosures were reported by the other authors.

J. Chen: Data curation, formal analysis, investigation, visualization, writing-original draft, writing-review and editing. C. Nelson: Data curation, formal analysis, investigation, visualization, writing-original draft, writing-review and editing. M. Wong: Data curation, formal analysis, investigation. A.E. Tee: Data curation, formal analysis, investigation. P.Y. Liu: Data curation, supervision, investigation. T. La: Formal analysis, investigation. J.I. Fletcher: Resources, validation, methodology. A. Kamili: Resources, validation, investigation, methodology. C. Mayoh: Data curation, software, formal analysis. C. Bartenhagen: Data curation, software, formal analysis. T.N. Trahair: Resources, methodology, writing-review and editing. N. Xu: Resources, investigation, methodology. N. Jayatilleke: Data curation, software, formal analysis. M. Wong: Data curation, software, formal analysis. H. Peng: Data curation, software, formal analysis. B. Atmadibrata: Investigation. B.B. Cheung: Resources, methodology. Q. Lan: Resources, methodology. T.M. Bryan: Resources, methodology, writing-review and editing. P. Mestdagh: Resources, methodology. J. Vandesompele: Resources, methodology. V. Combaret: Resources, methodology. V. Boeva: Resources, methodology. J.Y. Wang: Resources, methodology. I. Janoueix-Lerosey: Resources, methodology. M.J. Cowley: Data curation, software, supervision. K.L. MacKenzie: Resources, methodology. A. Dolnikov: Resources, methodology, writing-review and editing. J. Li: Data curation, software, supervision. P. Polly: Supervision, visualization, writing-original draft, writing-review and editing. G.M. Marshall: Resources, methodology, writing-review and editing. R.R. Reddel: Resources, methodology. M.D. Norris: Resources, methodology, writing-review and editing. M. Haber: Resources, methodology, writing-review and editing. M. Fischer: Data curation, supervision. X.D. Zhang: Conceptualization, resources, formal analysis, methodology, writing-review and editing. H.A. Pickett: Conceptualization, resources, formal analysis, supervision, validation, visualization, methodology, writing-original draft, writing-review and editing. T. Liu: Conceptualization, resources, formal analysis, supervision, funding acquisition, validation, visualization, writing-original draft, project administration, writing-review and editing.

We thank Marco Herold at Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia, for providing the FH1tUTG construct. The authors were supported by Cancer Council NSW (to T. Liu), National Health & Medical Research Council Australia, and Tour de Cure Foundation (to T. Liu). P.Y. Liu is a research fellow of Cancer Institute NSW. V. Boeva was supported by the ATIP-Avenir Program, the ARC Foundation (RAC16002KSA-R15093KS), the “Who Am I?” Laboratory of Excellence (ANR-11-LABX-0071), and the French Government (ANR-11-IDEX-0005-02). Children' Cancer Institute Australia is affiliated with UNSW Australia and Sydney Children's Hospitals Network.

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.