We investigated whether targeting chromatin stability through a combination of the curaxin CBL0137 with the histone deacetylase (HDAC) inhibitor, panobinostat, constitutes an effective multimodal treatment for high-risk neuroblastoma.
The effects of the drug combination on cancer growth were examined in vitro and in animal models of MYCN-amplified neuroblastoma. The molecular mechanisms of action were analyzed by multiple techniques including whole transcriptome profiling, immune deconvolution analysis, immunofluorescence, flow cytometry, pulsed-field gel electrophoresis, assays to assess cell growth and apoptosis, and a range of cell-based reporter systems to examine histone eviction, heterochromatin transcription, and chromatin compaction.
The combination of CBL0137 and panobinostat enhanced nucleosome destabilization, induced an IFN response, inhibited DNA damage repair, and synergistically suppressed cancer cell growth. Similar synergistic effects were observed when combining CBL0137 with other HDAC inhibitors. The CBL0137/panobinostat combination significantly delayed cancer progression in xenograft models of poor outcome high-risk neuroblastoma. Complete tumor regression was achieved in the transgenic Th-MYCN neuroblastoma model which was accompanied by induction of a type I IFN and immune response. Tumor transplantation experiments further confirmed that the presence of a competent adaptive immune system component allowed the exploitation of the full potential of the drug combination.
The combination of CBL0137 and panobinostat is effective and well-tolerated in preclinical models of aggressive high-risk neuroblastoma, warranting further preclinical and clinical investigation in other pediatric cancers. On the basis of its potential to boost IFN and immune responses in cancer models, the drug combination holds promising potential for addition to immunotherapies.
Despite escalation of treatment intensity in recent years, high-risk neuroblastoma, frequently driven by MYCN amplification, has a survival rate below 50%. Furthermore, the long-term quality of life of survivors is often severely compromised by detrimental consequences of conventional treatments. In light of the high unmet need for more effective and safer treatments for high-risk neuroblastoma, this study proposes the combination of the curaxin CBL0137 with the HDAC inhibitor panobinostat as a novel treatment strategy targeting chromatin stability. Our data demonstrate that the drug combination induces enhanced chromatin destabilization, DNA damage accumulation, and synergistic inhibition of cancer cell growth in vitro and in vivo. In addition, the combination mediates part of its action through eliciting a type I IFN response and engaging the immune system. Thus, this provides opportunities for the use of this drug combination in the modulation of antitumor immunity.
Neuroblastoma is the most common extracranial solid tumor in children (1). It is a group of heterogeneous embryonal tumors that arise from organs with neural crest origin such as adrenal glands and sympathetic ganglia. On the basis of the patients' age, histology, and tumor staging, neuroblastoma is stratified into low, intermediate, and high-risk groups. High-risk neuroblastoma comprises half of newly diagnosed cases and is frequently characterized by MYCN amplification, which is a strong predictor of poor prognosis (1, 2). Current treatments for high-risk neuroblastoma include high doses of chemotherapy, surgery, radiotherapy, myeloablation and stem cell transplant, retinoid therapy, and immunotherapy. However, despite this intensive multimodal regimen, more than 50% of patients with high-risk neuroblastoma die within five years postdiagnosis. Even with achieving initial remission posttreatment, therapy resistance and early relapse are common in these high-risk patients (3). Moreover, conventional cytotoxic drugs, which still comprise the first-line treatment for neuroblastoma, have detrimental long-term health effects including treatment-induced secondary tumors, cardiac and neurological toxicity, and infertility (4–6). Development of more efficacious and safer drugs for high-risk neuroblastoma thereby remains a high unmet need.
CBL0137 is a nongenotoxic anticancer drug belonging to a family of small molecules called curaxins (7, 8). It is currently in phase I adult clinical trials for advanced melanoma or sarcoma (NCT03727789), and metastatic or unresectable advanced solid neoplasms (NCT01905228). Our group and others have demonstrated that CBL0137, either as a monotherapy or in combination with standard-of-care chemotherapeutic agents, significantly delays cancer progression in several high-risk childhood cancer models, including MYCN-amplified neuroblastoma, and aggressive forms of acute lymphoblastic leukemia (9–11).
CBL0137 exerts its anticancer effect through a novel mechanism of action. By intercalating into DNA and interfering with DNA–histone interactions, the drug induces a genome-wide nucleosome destabilization, or “chromatin damage,” without causing DNA damage such as nucleotide alterations or breaks (12–14). This chromatin damage promotes dysregulation of cellular transcriptional and replication programs, ultimately leading to cell death (14). Moreover, nucleosome destabilization induced by CBL0137 increases heterochromatin transcription, stimulating a double-stranded RNA-induced IFN response called Transcription of Repeats Activates INterferon (TRAIN) in tumors (10, 13). CBL0137 also acts as an indirect inhibitor of histone chaperone Facilitator of Chromatin Transcription (FACT), trapping FACT onto DNA (i.e., c-trapping; refs. 7, 9, 12, 15). This mediates an array of downstream effects, including p53 activation, NFκB suppression, as well as inhibition of DNA damage repair, all of which restrict cancer progression (7).
Inhibitors of histone deacetylases (HDAC) represent another group of compounds that affect chromatin stability (16). Panobinostat is a pan-HDAC inhibitor which, through increasing histone acetylation levels, loosens nucleosome structure and decondenses chromatin (16). In addition, the drug induces DNA damage and enhances antitumor immunity (17). Panobinostat is an FDA-approved HDAC inhibitor for multiple myeloma and has demonstrated efficacy in a wide range of preclinical pediatric cancer models including high-risk neuroblastoma (18–20). Currently, it is also in early-phase trials for hematologic (NCT00723203) and solid (NORTH Trial by ANZCHOG) childhood malignancies including neuroblastoma, with previous phase I trials showing good tolerability of the drug in pediatric patients (21, 22).
On the basis of their differing mechanisms of action in destabilizing chromatin, we hypothesized that CBL0137 could synergize with panobinostat. Herein, we observed enhanced efficacy of the combination of CBL0137 and panobinostat in animal models of high-risk neuroblastoma. Remarkably, this therapeutic enhancement is significantly more pronounced in an immunocompetent setting and associated with an induction of a type I IFN response and heightened immune responses.
Materials and Methods
Chemicals and reagents
CBL0137 was provided by Incuron, Inc. Panobinostat, entinostat (MS-275), and SAHA (vorinostat) were purchased from Sapphire Bioscience Pty Ltd., Selleck Chem, and Cayman Chemical, respectively.
Cell lines, cell culture, and cell-based assays
All cell lines used in this study are Mycoplasma free (Mycoplasma testing every 6 months) and have been authenticated using STR profiling. The patient-derived xenograft (PDX) model COG-N-424x was obtained from the Childhood Cancer Repository, Texas Tech University Health Sciences Center in Lubbock, Texas. The culturing of neuroblastoma cell lines SK-N-BE(2)-C (referred to as BE(2)-C), KELLY, SH-SY5Y, and NH02A cells, and cell-based assays were performed as described previously (23–26). Caspase-3/7 activity was measured using the Invitrogen CellEvent Caspase-3/7 Green Flow Cytometry Assay Kit according to the manufacturer's recommendations. Etoposide-treated cells were included as a positive control. Samples were run on a BD FACSCanto II and analyzed by FlowJo software. Data are represented as percentages of cells positive for caspase 3/7 activity.
Histone eviction, micrococcal nuclease, and heterochromatin transcription reporter cell assays
Histone eviction, Micrococcal Nuclease (MNase), and heterochromatin transcription reporter-cell assays were performed as described previously (13, 27, 28). Generation of HT1080 cells with nuclei tagged with mCherry-H1.5 histone and HeLa-TI cells carrying an integrated avian sarcoma genome with silent GFP gene was reported previously (13, 27, 28).
DNA damage repair assays
Assays to assess DNA damage repair after panobinostat treatment (immunoblotting for γH2AX and RPA, immunofluorescence for γH2AX/53BP1-positive foci, and pulsed-field gel electrophoresis for detection of double stranded DNA breaks) were performed as described previously (9).
HT1080 6TG cells were a gift from Eric Stanbridge (University of California, Irvine, CA). HT1080 6TG H2B-mCherry and HT1080 6TG BAX/BAK DKO H2B-mCherry cells were derived as described previously (24). Imaging was performed on a Zeiss Cell Observer SD spinning disk confocal microscope. HT1080 6TG H2B-mCherry and HT1080 6TG BAX/BAK DKO cell lines were imaged using Differential Interference Contrast (DIC) microscopy combined with fluorescent imaging (10% light source intensity of 561 nm laser, 1 × 1 binning, EM gain of 600 and 50% light source intensity of TL LED, 1 × 1 binning, EM gain of 150) using appropriate filter sets, a 40x/0.95 air objective, 16-bit depth, at 37° C, 10% CO2 and 3% oxygen. A total of 10 z-stacks (11.25 μm) were captured in an image scaled to 170.67 × 170.67 pixels. Images were captured with Zen software using an Evolve Delta (Photometrics) camera, every 6 minutes for up to 3 days. Image analysis and processing were done using Zen software. Two sets of independent experiments were performed. Quantification of nuclear blebbing was done by manual counting of nuclei in a representative image from each treatment group for the same timepoint.
Western blotting experiments were performed as described previously (9, 23). The following antibodies were used: anti-γH2AX antibody [3F2] (phospho S139, ab22551, Abcam, Melbourne, VIC, Australia), anti-H3K27ac antibody (06–599, Millipore, North Ryde, NSW, Australia or ab4729, Abcam), anti-pRPA (ser4/8) antibody (A300–245A, Bethyl Laboratories), anti-cleaved PARP antibody (95415, Cell Signaling Technology, Genesearch), anti-total PARP antibody (9532, Cell Signaling Technology), and anti-actin (A2066, Sigma) and anti-GAPDH antibody (Sapphire Bioscience) as loading controls.
Colony formation and cytotoxicity assays
Colony and cytotoxicity assays of neuroblastoma cell lines were performed according to previously published methods (9, 23). For colony assays, cells were plated at 500 cells/well in 6-well plates, treated with drugs for 72 hours and after an incubation of 10 to 14 days, colonies were stained and counted. For cytotoxic assays, cells were plated at 5,000 cells/well in 96-well plates and treated for 72 hours before addition of resazurin-based dye and reading of fluorescence on a plate reader. Each experiment was performed in triplicate and drug synergy was determined by Chou and Talalay's median effect equations using CalcuSyn software (Biosoft) as described previously (9, 23).
All animal experimental procedures were approved by the University of New South Wales Animal Care and Ethics Committee according to the Animal Research Act, 1985 (New South Wales, Australia) and the Australian Code for the Care and Use of Animals for Scientific Purposes (2013).
The Th-MYCN transgenic mouse model of neuroblastoma has been previously described (9, 29, 30). All experiments utilized only Th-MYCN+/+ mice with 10 mice per treatment group. Treatment was commenced once the tumor reached 5 mm in diameter by palpation and mice were euthanized when the tumor reached 10 mm in diameter, when signs of a thoracic tumor manifested or in the absence of tumor relapse, at 20 weeks of age.
For neuroblastoma xenograft models, female BALB/c nude mice (nu/nu) (BALB/c-Foxn1nu/Arc) at 4 to 5 weeks of age were purchased from Australian Resources Centre and allowed to acclimatize for one week before subcutaneous inoculation of 5 × 106 BE(2)-C cells in suspension or 1 × 106 COG-N-424x cells in RPMI (Life Technologies, Mulgrave, Victoria, Australia) and an equal volume of growth factor–reduced Matrigel (Corning Inc.) into the dorsal flank. Each group consisted of six mice.
For immunocompromised versus immunocompetent models, neuroblastoma tumors of approximately 1,000 mm3 in size were harvested from Th-MYCN+/+ mice, digested, and then pooled to obtain single-cell suspensions. These cells were subsequently subcutaneously transplanted into either BALB/c nudes or the wild-type 129/SvJ littermates of Th-MYCN+/+ mice, with 4.8 × 106 tumor cell suspension/mouse in the dorsal flank.
For all subcutaneously implanted tumors, mice were randomized to the treatment groups and treatment started when the tumor reached 100 mm3 in volume, except for the BE(2)-C tumor model for which treatment started at 50 mm3. Tumor size was measured every other day and calculated as 1/2 (length × width × depth).
Mice were given CBL0137 intravenously (40 mg/kg, except for xenograft models which were treated with 60 mg/kg, dissolved in 50 mg/mL Captisol) every 4 days for eight doses, panobinostat intraperitoneally (5 mg/kg/day, dissolved in 5% PEG400 and 5% Tween80 saline) for 7 days, or the combination. Mice were euthanized when the tumor reached 1,000 mm3, or until the end of the holding period if tumor relapse did not occur. Two-sided log-rank tests were performed to determine statistical significance of survival differences between treatment groups. P values below 0.05 were considered significant.
Whole transcriptome sequencing and analysis
Total RNA-seq libraries were prepared using Illumina's cBot cluster generation system with TruSeq PE Cluster Generation Kits (catalog no. PE-401–3001) according to the manufacturer's protocol. Libraries were pooled and sequencing performed in paired-end mode using Illumina NextSeq 500, with approximately 33M reads per sample. Whole transcriptome sequencing and analysis were performed as described previously (31). Single-sample gene-set enrichment analysis (ssGSEA) was performed on the gene expression data generated from the whole transcriptome sequencing using the R graphical user interface ssgsea2.0 (https://github.com/broadinstitute/ssGSEA2.0). All RNA-sequencing data are available through GEO Series accession number GSE151689 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE151689).
Immune cell deconvolution analysis
Immune deconvolution analysis was performed using CIBERSORTx to impute cell fractions from whole transcriptome expression profiles against the previously developed LM22 immune cell signature matrix on control mice or mice treated with either CBL0137, panobinostat, or the combination (32, 33).
RNA isolation and qRT-PCR
Total RNA was isolated using a Qiagen RNeasy kit and cDNA synthesized using the iScript cDNA Synthesis Kit (Bio-Rad). SYBR green quantitative PCR (qRT-PCR) was run and analyzed on Bio-Rad's CFX96 Real-Time PCR Detection System with primers described in Supplementary Table S1.
Slides of 4-μm thickness were prepared and hematoxylin & eosin stained as described previously (9).
Immunofluorescence and whole-tumor scanning
Slides of 4-μm thickness were prepared and stained as described previously (34). Slides were incubated with primary antibodies against IFIT3 (Millipore ABF1048, 1:100) and Ki67 (Ab16667, Abcam, 1:60), and Alexa Fluor–labeled secondary antibodies (Life Technologies). Whole tumor images were scanned using the Aperio FL (Leica Biosystems). Images were analyzed using ImageScope software (Leica Biosystems) and ImageJ. Three tumors were imaged and quantified for each treatment group. For each tumor image, three fixed-size areas were randomly selected, measured, and averaged. Fixed threshold values of each fluorescence channel were used to highlight and quantify area size. DAPI measurement was used for normalization. All the images were quantified using the same set of parameters, and were adjusted for contrast and brightness in the exact same way.
Flow cytometry analysis of immune cells
Tumors were harvested and digested with 20 μg/mL type II DNase I (Sigma, D4527) and 1 mg/mL collagenase IV (Worthington LS004186) to obtain single-cell suspensions. Spleen and blood samples were processed with red blood cell lysis buffer (1 mmol/L ammonium bicarbonate and 114 mmol/L ammonium chloride). Antibodies (1:100 ratio) and Zombie Aqua live/dead dye (1:1000 ratio; BioLegend, 423102) were added to cell suspensions in PBS and incubated for 20 minutes on ice before flow analysis on BD FACSCanto. Data were analyzed on FlowJo and represented as percentages
of positive cells. Antibodies: FITC rat anti-mouse CD45 (BioLegend, 103108), APC rat anti-mouse CD3 (BioLegend, 100235), PE rat anti-mouse CD4 (BioLegend, 116005), and PE/Cyanine7 rat anti-mouse CD8a (BioLegend, 100721).
LC/MS-MS determination of CBL0137 and panobinostat concentration in plasma and tumor
Neuroblastoma tumors and plasma were collected from Th-MYCN mice and snap frozen for storage. All samples were prepared at the same time for analysis using a methanol extraction procedure. Approximately 100-mg sections of tumor tissue were supplemented by methanol 1:10 (wt:vol) containing 0.1% formic acid and homogenized using a hand-held homogenizer. To complete the compound extraction, the sample solutions were rocked overnight at 2°C to 8°C and then centrifuged for 10 minutes at 14,000 rpm and 2°C to 8°C to precipitate the homogenized tissue pellet. The supernatant was then separated and diluted 1:4 (vol/vol) with mobile phase A (MPA) and analyzed by LC/MS-MS versus a reagent-matched calibration curve. Plasma samples were prepared by 1:4 (vol/vol) methanol extraction followed by vortexing for 5 minutes. The samples were then centrifuged for 10 minutes at 14,000 rpm and 2°C to 8°C to precipitate the proteins and the supernatant was collected. A detailed description of the LC/MS-MS method is provided in Supplementary Methods.
GraphPad Prism8 was used for all statistical analyses. Event-free survival curves were compared by two-sided log-rank tests. All data are represented as mean ± SEM. Comparisons of variables between groups were performed by ANOVA followed by Tukey multiple comparisons, or unpaired two-tailed t tests as indicated in figure legends. P values < 0.05 were considered significant.
The combination of CBL0137 and panobinostat induces enhanced chromatin destabilization
As CBL0137 and panobinostat both promote an open chromatin conformation (1), we postulated that combining these two drugs would induce enhanced nucleosome and chromatin destabilization. Our previous studies showed that eviction of linker histone H1 from the nucleosomes is a highly sensitive measure of nucleosome destabilization (13). We therefore assessed the combined effect of CBL0137 and panobinostat on histone H1 eviction by evaluating the subnuclear localization of H1 in nuclei of HeLa cells overexpressing mCherry-histone H1 (Fig. 1A). H1 lost from chromatin accumulates in nucleoli, which allows easy monitoring of this process in live cells (13). While panobinostat and CBL0137 as single agents induced nucleolar accumulation of histone H1 (mCherry-histone H1, red) in less than 20% of cells at highest applied doses, the combination of both drugs markedly enhanced this effect (Fig. 1A).
Our previous data also demonstrated that nucleosome destabilization induced by CBL0137 results in an increased transcription of silenced heterochromatin regions (13). To determine whether concurrent exposure to panobinostat could further enhance this effect, we used HeLa cells carrying an integrated avian sarcoma virus genome containing a green fluorescent protein (GFP) gene of which expression is silenced under basal conditions of chromatin condensation (13, 28). Indeed, while CBL0137 dose-dependently increased GFP expression, the combination of CBL0137 and panobinostat doubled the percentage of cells with GFP expression in comparison with cells treated with the single agents (Fig. 1B), confirming that this combination promotes enhanced chromatin destabilization and heterochromatin transcription.
A highly organized chromatin state is essential to maintain structural integrity of the nucleus. Chromatin decompaction and destabilization can lead to morphologic aberrations such as nuclear blebbing (35). We assessed whether the combination of CBL0137 and panobinostat impacted nuclear structural integrity by live-cell imaging using HT1080 6TG cells expressing H2B-mCherry, which facilitates nuclear visualization (24). While panobinostat caused nuclear blebbing in a small proportion of cells, when treated with the combination, approximately 60% of the cells displayed nuclear protrusions (Fig. 1C; Supplementary Videos S1–S4). As the HT1080 6TG cells are a p53-mutant line, the observed effect is independent of p53 activation (24). Moreover, a similar percentage of cells with nuclear blebbing was seen in apoptosis-deficient BAX−/−/BAK−/− cells, confirming that the combination of CBL0137 and panobinostat negatively impacted nuclear structural integrity and that this observation was not a direct downstream effect of apoptosis (Supplementary Fig. S1A and S1B).
The combination of CBL0137 and panobinostat induces chromatin destabilization, activates an IFN response, and enhances DNA damage in neuroblastoma cells
To confirm that the combination of CBL0137 and panobinostat enhanced chromatin destabilization in neuroblastoma cells, we performed MNase assays to assess the relative condensation of chromatin in cells after treatment with CBL0137, panobinostat, or the combination. As MNase preferentially digests protein-free DNA, the pattern of DNA fragmentation resulting from treatment of cell nuclei is an indicator of relative chromatin condensation. Compared with untreated nuclei, DNA isolated from neuroblastoma cell nuclei treated with either CBL037 or panobinostat as a single agent was digested into smaller fragments more rapidly (Fig. 1D). This effect was even more pronounced upon combined treatment with CBL0137 and panobinostat, with the isolated DNA appearing as a smear without clear laddering. This confirms enhanced chromatin destabilization in neuroblastoma cells treated with the combination of CBL0137 and panobinostat.
Downstream of chromatin destabilization, CBL0137 induces an IFN response in tumor cells (10, 13). Panobinostat was also recently shown to activate a type I IFN response in dendritic cells through epigenetic reprogramming (36). To determine whether adding panobinostat to CBL0137 augmented the IFN response in tumor cells, we assessed the effect of the drug combination on expression levels of IFIT3. The expression of this IFN-induced gene was previously shown to be significantly upregulated in cancer cells in vitro and in vivo in response to CBL0137 treatment (10). Adding panobinostat to CBL0137 significantly increased IFIT3 gene expression in neuroblastoma cells compared with treatment with vehicle or CBL0137 alone (Fig. 1E), confirming that the combination enhanced IFN signaling within the tumor cells.
Furthermore, panobinostat treatment has been shown to result in DNA damage accumulation in cancer cells (37), while CBL0137 inhibits repair of DNA damage caused by genotoxic chemotherapy, sensitizing tumor cells to chemotherapy (9). To assess whether CBL0137 potentiates the effects of panobinostat via inhibition of the repair of panobinostat-induced DNA damage in high-risk neuroblastoma cells, we examined the extent of DNA repair in a MYCN-amplified, multidrug-resistant neuroblastoma cell line BE(2)-C, post panobinostat-treatment in the presence or absence of CBL0137. Following 20 hours of panobinostat pretreatment and a 20-hour recovery, BE(2)-C neuroblastoma cells were able to spontaneously repair the panobinostat-induced DNA damage as evidenced by a reversal of the panobinostat-induced increase in number of treated BE(2)-C cells with γH2AX and 53BP1 double-positive foci in the nuclei (Fig. 2A). However, upon addition of CBL0137 during the recovery phase, the resolution of γH2AX and 53BP1 double-positive foci was significantly attenuated indicating that the repair of panobinostat-induced DNA damage was prevented by the drug (Fig. 2A). Similarly, phosphorylated RPA and γH2AX levels, as markers of DNA damage, in panobinostat-pretreated BE(2)-C and KELLY neuroblastoma cells returned to baseline levels upon recovery in the absence of CBL0137 (Fig. 2B; Supplementary Fig. S2A and S2B). This effect was diminished when cells recovered in the presence of CBL0137, while H3K27 acetylation levels remained unchanged (Fig. 2B; Supplementary Fig. S2A and S2B). The inhibitory effects of CBL0137 on the repair of panobinostat-induced DNA damage were further confirmed in a pulsed-field gel electrophoresis assay in BE(2)-C cells demonstrating that the presence of CBL0137 during recovery after panobinostat administration attenuated repair of DNA double-strand breaks (Fig. 2C; Supplementary Fig. S2C).
The combination of CBL0137 and HDAC inhibitors synergistically inhibits neuroblastoma cell growth
As chromatin stability, suppression of the IFN pathway and repair of DNA damage are critical for cancer cell proliferation, we assessed the downstream effects of the combination of CBL0137 and panobinostat on neuroblastoma cell growth and survival. CBL0137 and panobinostat synergistically reduced colony formation of human and mouse MYCN-amplified neuroblastoma cell lines (Fig. 2D; Supplementary Fig. S3A and S3B). This attenuated tumor cell growth was accompanied by increased caspase-mediated apoptosis induction as indicated by elevated levels of caspase-3/7 activity (Fig. 2E) and enhanced PARP cleavage (Fig. 2F; Supplementary Fig. S3C and S3D), which was preventable by preincubation with the pan-caspase inhibitor QVD (Supplementary Fig. S3E).
To confirm that the mechanistic basis of the observed synergy between CBL0137 and panobinostat was dependent on the inhibitory action of panobinostat on HDACs, we tested whether similar synergy was also observed between CBL0137 and other HDAC inhibitors. We observed that CBL0137 synergized with the pan-HDAC inhibitor SAHA as well as with entinostat, an inhibitor of class I HDACs, in colony formation assays with neuroblastoma cells (Supplementary Fig. S4A and S4B). Similarly, CBL0137 combined with entinostat and SAHA induced caspase-mediated apoptosis evidenced by increased levels of caspase-3/7 activity (Supplementary Fig. S4C) and PARP cleavage (Supplementary Fig. S4D). CBL0137 thus synergizes with other HDAC inhibitors in addition to panobinostat, providing further evidence that inhibition of histone deacetylation lies at the mechanistic basis of the detected synergy between CBL0137 and panobinostat.
As one of the important anti-cancer actions of CBL0137 involves inhibition of histone chaperone FACT, we subsequently investigated whether loss of FACT would influence the observed synergy between CBL0137 and panobinostat. We therefore assessed the effect of silencing the expression of SSRP1, the major protein subunit of FACT, on sensitivity of neuroblastoma cells to panobinostat. We observed that panobinostat decreased colony growth to a similar extent in BE(2)-C neuroblastoma cells, irrespective of the presence or absence of SSRP1 silencing, indicating that silencing of FACT alone does not synergize with HDAC inhibition in neuroblastoma cells (Supplementary Fig. S5).
We previously reported on a positive feedback loop between FACT and MYCN signaling in MYCN-amplified neuroblastoma and on the particular potency of CBL0137 against neuroblastoma cell lines with high MYCN expression. We also demonstrated that CBL0137 limits MYCN-amplified neuroblastoma tumor growth by decreasing expression levels of the MYCN oncogene, thereby limiting N-MYC signaling, an important oncogenic driver in MYCN-amplified neuroblastoma. Similarly, panobinostat was shown to decrease MYCN expression in neuroblastoma models (18). We found that combining CBL0137 and panobinostat further reduced N-MYC protein levels in neuroblastoma cells compared with either agent alone (Supplementary Fig. S6A). As N-MYC is instrumental for the proliferation and survival of MYCN-amplified neuroblastoma cells, the decreased expression of N-MYC is likely to negatively impact neuroblastoma cell proliferation.
To assess whether the combination of CBL0137 and panobinostat also exerted synergistic cytotoxicity against non-MYCN–amplified neuroblastoma cells, we performed synergy assays in non-MYCN amplified human neuroblastoma SH-SY5Y cells. Synergy assays confirmed that CBL0137 and panobinostat were highly synergistic in SH-SY5Y cells (Supplementary Fig. S6B) indicating that synergy between CBL0137 and panobinostat is independent of MYCN gene amplification status. These data are consistent with our finding that the drug combination induces several anticancer mechanisms that are independent of N-MYC signaling (Figs. 1 and 2).
Overall, our in vitro data demonstrate that the combination of CBL0137 and panobinostat limited several cancer cell survival pathways and synergistically suppressed cell growth in high-risk neuroblastoma cell lines.
The combination of CBL0137 and panobinostat suppresses tumor progression in high-risk neuroblastoma mouse models
We have previously shown that CBL0137 as a single agent significantly extends survival in preclinical models of MYCN-amplified neuroblastoma (9), and others have demonstrated preclinical efficacy of panobinostat in these models (18). On the basis of our in vitro data demonstrating synergistic inhibitory effects of the combination of CBL0137 and panobinostat on neuroblastoma cell growth and survival, we hypothesized that in vivo therapeutic enhancement would occur when combining both drugs in preclinical neuroblastoma models.
Athymic nude mice were subcutaneously engrafted with either the drug-refractory neuroblastoma cell line BE(2)-C, or a highly aggressive neuroblastoma patient-derived xenograft line (COG-N-424x), both of which carry MYCN gene amplification (9, 38). The combination treatment extended the median survival from 12.4 to 23.7 days in the BE(2)-C model and from 20.6 to 29.9 days in mice bearing COG-N-424x neuroblastoma tumors, corresponding to an extension of survival of 91% and 45% compared with vehicle-treated mice, respectively (Fig. 3A and B; Supplementary Table S2). The survival extension induced by the combination of CBL0137 and panobinostat was significantly longer than that achieved for either single agent, demonstrating therapeutic enhancement of both drugs in combination (Fig. 3A and B; Supplementary Table S2).
There is clear evidence for an activation of an IFN response in tumors treated with CBL0137 as demonstrated by increased expression of IFIT3 (Supplementary Fig. S7). Moreover, endpoint tumors from mice treated with CBL0137, panobinostat or the combination showed increased presence of necrotic lesions compared with vehicle controls, consistent with these drugs inducing neuroblastoma cell death in vitro (Supplementary Fig. S8).
The combination of CBL0137 and panobinostat induces complete tumor regression in Th-MYCN transgenic neuroblastoma mice
It was recently shown that one of the anticancer mechanisms of CBL0137 constitutes the stimulation of antitumor immunity downstream of chromatin destabilization and induction of an IFN response (10, 13, 39). To determine whether the therapeutic enhancement of CBL0137 by addition of panobinostat could be further potentiated in the presence of a fully functioning immune system, we next determined the efficacy of the combination in an immunocompetent transgenic mouse model of MYCN-amplified neuroblastoma.
Th-MYCN transgenic mice spontaneously develop neuroblastomas at 3 to 5 weeks of age, and the tumors rapidly progress to maximum tumor burden within five to seven days without therapeutic intervention. In agreement with our findings in vitro, treatment with the CBL0137 and panobinostat combination significantly increased IFIT3 protein levels in vivo (Supplementary Fig. S9A and S9B). Panobinostat as a single agent and combined with CBL0137 augmented histone H3 acetylation levels (Supplementary Fig. S9C), confirming target engagement by both drugs at applied doses in vivo.
While treatment of the Th-MYCN transgenic mice with CBL0137 and panobinostat as single agents significantly extended event-free survival, the combination of CBL0137 and panobinostat produced complete ablation of established neuroblastoma in 100% of the Th-MYCN transgenic mice (Fig. 4A; Supplementary Table S3). None of the mice treated by the combination displayed tumor relapse at the termination of the experiment, whereas tumors progressed to the maximum burden within 60 days posttreatment initiation in all the mice treated with either vehicle, CBL0137, or panobinostat alone, demonstrating the powerful antitumor effect of the combination in this model (Fig. 4A; Supplementary Table S3). These results indicate that efficacy of the CBL0137 and panobinostat combination may be augmented by the presence of an intact immune system.
The combination of CBL0137 and panobinostat induces an IFN response in Th-MYCN transgenic neuroblastoma mice
To confirm that the combination of CBL0137 and panobinostat affected immune pathways in the tumor, whole transcriptome expression profiling (RNA-seq) was performed on tumors from mice treated with a single dose of CBL0137, panobinostat, or the combination. The combination of CBL0137 and panobinostat induced a strong alteration in gene expression compared with the control or single-agent treatment groups (Fig. 4B). We observed a highly significant upregulation of multiple pathways associated with immune responses, and downregulation of cell cycle, nucleosome assembly, and cell division pathways (Fig. 4C). In particular, gene sets associated with immune responses and type I IFN were substantially enriched in tumors treated with the combination (Fig. 4D). This was confirmed by ssGSEA analysis showing uniform enrichment of these pathways in individual samples from each treatment group, with a higher level of enrichment in the combination treatment group compared with vehicle and single-agent treatment (Supplementary Fig. S10A). Many genes encoding for cytokines and chemokines are differentially expressed in mice following treatment with the CBL0137 and panobinostat combination by comparison with vehicle-treated mice (Supplementary Table S4). The genes with the most significant and highest fold change in expression (i.e., Ccl12, Ifnb1, Cxcl10, Cxcl11, Isg15, Cxcl9, Ccl5, Ccl7, etc) are all IFN-induced, which is consistent with our finding that the combination induces a robust IFN response. We further validated the upregulation of IFN-regulated genes, including Ifit3, Ifit3b, Ifit1b1, Ccl12, Ccl8, and Slfn4 by qRT-PCR in the neuroblastoma tumors (Fig. 4E). These data together demonstrated a robust interferon response and immune-stimulating effect of the combination compared with either single agent. Conversely, gene sets related to chromatin function were negatively impacted as shown by GSEA and ssGSEA analyses (Supplementary Fig. S10B and S10C). Genes involved in cell proliferation, cell cycle, and chromatin assembly, such as Mki67, Ccnb1, and Cdk4, were significantly downregulated by the combination as demonstrated by qRT-PCR and immunofluorescence of neuroblastoma tumors (Supplementary Fig. S11A–S11C). Similar changes in expression of genes involved in cell cycle and division pathways were observed in human neuroblastoma cells treated with the combination in vitro (Supplementary Fig. S11D).
Importantly, the observed changes in gene expression upon CBL0137 and panobinostat treatment were confined to the tumor, and were not observed in other nontumor tissues (liver, lung, and kidney) of treated mice (Supplementary Fig. S12). Moreover, induction of the IFN response genes Ifit3 and Ccl12 in liver, lung, and kidney tissue by either CBL0137 or the combination was at least five to 10 times lower than that in the tumor, while Ki67 was not affected by the drugs in these normal tissues, suggesting the synergy of the combination is tumor-specific.
In line with this, and consistent with our previous findings that CBL0137 accumulates in tumors compared with normal tissues (9), we observed that CBL0137 levels in the Th-MYCN tumors were more than 100 times higher than in plasma and were not affected by addition of panobinostat (Supplementary Fig. S13A). The panobinostat concentration was lower in tumors treated with the combination than in those treated with panobinostat alone, possibly due to decreased uptake by the compromised tumor cells (Supplementary Fig. S13B).
The combination of CBL0137 and panobinostat induces an immune response in Th-MYCN transgenic neuroblastoma mice
To confirm that the combination of CBL0137 and panobinostat activated an immune response, and to study the effects of the drug combination on the immune response within neuroblastoma tumors, we next performed CIBERSORT, a computational method designed to quantify cell fractions from bulk tissue gene expression profiles (RNA-seq data). This approach has been used to characterize tumor-infiltrating lymphocytes and accurately estimate the immune composition of tumors (40). CIBERSORT was applied to the RNA-seq data from murine neuroblastoma tumors from Th-MYCN mice treated with vehicle, CBL0137, panobinostat, or the combination for 16 hours. On the basis of this analysis, the combination of CBL0137 and panobinostat significantly increased (fourfold) tumor infiltration by CD45+ lymphocytes, while a trend was observed for higher levels of immune infiltration by treatment with either agent alone compared with the vehicle control (Fig. 5A). The drug combination also significantly increased the number of CD4+ T cells, both naïve and memory CD4+ T cells, by comparison with the vehicle control and single-agent treatment arms (Fig. 5A), while a trend for higher levels of total CD3+ T cells was also noted for the combination relative to vehicle controls (Supplementary Fig. S14). Although no significant changes were observed in amounts of CD8+ T cells, NK cells, B cells, or monocytes within the tumor, intratumoral levels of macrophages and dendritic cells were elevated after combination treatment compared with treatment with the vehicle or either agent alone (Supplementary Fig. S14).
We subsequently analyzed the composition of the tumor lymphocyte infiltrate (CD45+ cells) by performing flow cytometry on digested Th-MYCN tumors treated with CBL0137, panobinostat or the combination for 16 hours. Consistent with the CIBERSORT analysis, we confirmed that the lymphocyte infiltrate of tumors treated with the combination contained a significantly increased percentage of CD4+ T cells (Fig. 5B). The blood and spleens of animals treated with the combination also contained significantly increased percentages of CD4+ T cells in these tissues compared with vehicle-treated mice, confirming the immune-stimulating effect of the combination (Fig. 5B). Consistent with the CIBERSORT data, flow analysis showed that there was no increase in intratumoral CD8+ cells after treatment with the combination or with either drug alone (Supplementary Fig. S15). However, a significant increase in CD8+ T cells was seen in the spleen, but not the blood, after treatment with the combination, suggesting the mounting of a cytotoxic T-cell response against the tumor within the spleen at this time point after treatment (Supplementary Fig. S15). No significant changes were observed in the proportions of other immune cell subsets within the tumor lymphocyte infiltrate (data not shown).
Taken together, these data indicate that the combination of CBL0137 and panobinostat activates a T-cell response in an immunocompetent neuroblastoma model and that this response is associated with complete tumor regression.
Presence of an intact immune system with an adaptive immune component promotes durable tumor regression by the combination of CBL0137 and panobinostat
To provide further evidence that the activation of a T-cell response by the combination of CBL0137 and panobinostat contributes to its anticancer effects and therapeutic enhancement by comparison with the effect of either treatment alone, we compared the efficacy of the drug combination against neuroblastoma tumors xenografted in immunocompetent versus immunocompromised mouse strains. We subcutaneously transplanted dissociated Th-MYCN tumor cells into either immunodeficient BALB/c nude mice that lack T cells and have a dysfunctional adaptive immune system, or into the immunocompetent 129/SvJ mice that have a fully functioning immune system, and then treated with CBL0137, panobinostat, or the combination (Fig. 5C; Supplementary Fig. S16). Panobinostat treatment displayed a similar efficacy in the immunodeficient and immunocompetent mice. However, CBL0137 and particularly the combination was more effective against tumors engrafted in the immunocompetent mouse strain (Fig. 5C; Supplementary Fig. S16; Supplementary Table S5). All the immunocompetent mice (100%) treated with the combination achieved complete tumor regressions, compared with 50% of the immunodeficient mice in the same treatment group. The immunocompetent mice treated with the combination exhibited complete and durable tumor regression without relapse for up to a year at which point the experiment was terminated. These findings confirm that the combination of CBL0137 and panobinostat has reduced efficacy in an immunodeficient model that lacks T cells and that the presence of a fully functional immune system with an adaptive immune component allows the drug combination to exert its full anticancer potential.
Taken together, our study has modeled an effective, previously unexplored, multiagent cancer treatment strategy targeting chromatin stability. Our data demonstrate that upon combining CBL0137 with panobinostat, multiple anticancer mechanisms are activated and channeled into two major pathways, culminating in an enhanced direct and indirect inhibition of cancer progression (Fig. 6). More importantly, presence of an intact immune system is shown to be critical in achieving a durable therapy response, indicating that the combination of CBL0137 and panobinostat may regulate key factors in antitumor immunity.
The identification of safer and cancer-selective multimodal treatment options for high-risk neuroblastoma is urgent. Herein we report on a novel combination of anticancer drugs, the curaxin CBL0137, and the HDAC inhibitor panobinostat, which was well-tolerated in several mouse strains and displays enhanced preclinical efficacy in multiple animal models of high-risk neuroblastoma. Mechanistic studies reveal that CBL0137 and panobinostat amplify each other's effects to accelerate nucleosome disassembly and histone eviction, and this dual targeting of chromatin stability may thus constitute an alternative therapeutic modality for high-risk neuroblastoma.
We propose a model whereby CBL0137 and panobinostat work in tandem to destabilize chromatin, enhance heterochromatin transcription, activate an IFN response and promote DNA damage accumulation in cancer cells, resulting in direct inhibition of cancer cell growth and survival (Fig. 6). A parallel pathway boosts a T-cell immune response against the tumors, indirectly limiting cancer progression. The resulting two-pronged attack on cancer cells by the combination of CBL0137 and panobinostat induces complete tumor regressions in an immunocompetent model of MYCN-amplified neuroblastoma, which is dependent on the presence of an adaptive immune component.
The combination of CBL0137 and panobinostat elicits a tumor-specific IFN response and growth suppression that is absent in normal tissues, suggesting a highly targeted antitumor effect. The tendency of CBL0137 to accumulate in tumor cells could further mitigate adverse effects even with prolonged treatment. These are ideal properties of a drug candidate for pediatric cancers where severe debilitating side effects from standard multimodal chemotherapies are a major concern. The combination of CBL0137 and panobinostat thereby presents a clinically viable option for the treatment of high-risk neuroblastoma, providing support for further investigations into targeting chromatin stability as a treatment strategy for other childhood cancers.
Recently, immunotherapies have achieved remarkable results in halting progression of relapsed and refractory cancers that were once considered incurable (41). On the basis of this initial success, currently available immunotherapeutic regimens are rapidly moving into first-line therapy. However, the efficacy of immunotherapy is hampered by immune-evasion of the tumor and the development of drug resistance. The identification of compounds that can further potentiate current immunotherapies through reprogramming the tumor microenvironment and/or enhancing antitumor immunity is therefore highly desirable.
One of the well-documented immune evasion mechanisms in solid tumors is the suppression of IFN pathways (42). As CBL0137 has been shown to induce a robust IFN response in tumors, the drug could reactivate innate immune sensing. This is supported by our findings of increased immune pathway engagement upon CBL0137 treatment in our transgenic neuroblastoma model. Combining CBL0137 with the known immunomodulator panobinostat further augmented this effect, and produced durable and complete tumor regression in immunocompetent tumor-bearing mice. Corroborating our findings, other studies have uncovered a strong link between the host immune response and tumor cell chromatin structure and modifications, initiating a wave of clinical and animal studies, in which epigenetic targeting agents are combined with immunotherapies to improve efficacy (43).
Together, our results suggest that the combination of CBL0137 and panobinostat holds potential as an immunotherapeutic or immunomodulatory approach. Additional work defining the entire spectrum of the pathways of this immune enhancement by the combination offers great promise for designing more effective therapies and identifying biomarkers for patient stratification. Moreover, to further elucidate the clinical potential of the immunomodulatory actions of CBL0137, future combination studies with current immunotherapies such as immune checkpoint inhibitors and CAR T cells are warranted.
K. Somers reports grants from Kids Cancer Alliance, Tenix Foundation, and Anthony Rothe Memorial Trust during the conduct of the study. M. Karsa reports grants from Tenix Foundation and grants from Anthony Rothe Memorial Trust during the conduct of the study. E. Ronca reports grants from Cancer Institute NSW during the conduct of the study. A. Bongers reports grants from Tenix Foundation and grants from Anthony Rothe Memorial Trust during the conduct of the study. L. Zhai reports grants from Cancer Australia and The Kids' Cancer Project during the conduct of the study. D.R. Carter reports grants from Cancer Australia and The Kids' Cancer Project during the conduct of the study. R.W. Johnstone reports personal fees from MecrX; grants from MecrX, BMS, and Roche; grants from Novartis during the conduct of the study; and grants from AstraZenica outside the submitted work. A.J. Cesare reports grants from Australian National Health and Medical Research Council during the conduct of the study; other support from Goodridge Foundation, Stanford Brown, Inc., Neil and Norma Hill Foundation, and other support from Sydney West Radiation Oncology Network outside the submitted work. D.S. Ziegler reports personal fees from Bayer, Amgen, and personal fees from Day One outside the submitted work. A.V. Gudkov reports a patent 9,169,207 issued to Incuron, LLC, a patent 8,486,697 issued to Incuron, LLC, a patent 8,765,738 issued to Incuron, LLC, a patent 9,108,916 issued to Incuron, LLC, a patent 9,566,265 issued to Incuron, LLC, and a patent 10,137,109 issued to Incuron, LLC. K.V. Gurova reports a patent for Carbazol compounds and therapeutic uses of the compounds issued. No disclosures were reported by the other authors.
L. Xiao: Conceptualization, data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration. K. Somers: Conceptualization, data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration. J. Murray: Formal analysis, validation, investigation, methodology, writing–review and editing. R. Pandher: Formal analysis, validation, investigation, methodology, writing–review and editing. M. Karsa: Validation, investigation. E. Ronca: Formal analysis, validation, investigation, writing–review and editing. A. Bongers: Validation, investigation. R. Terry: Formal analysis, validation, investigation. A. Ehteda: Resources, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. L.D. Gamble: Resources, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. N. Issaeva: Resources, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. K.I. Leonova: Resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. A. O'Connor: Resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. C. Mayoh: Data curation, software, formal analysis, validation, investigation, methodology, writing–review and editing. P. Venkat: Resources, data curation, software, formal analysis, visualization, methodology. H. Quek: Resources, validation, investigation, methodology. J. Brand: Validation, investigation. F.K. Kusuma: Validation, investigation. J.A. Pettitt: Validation, investigation. E. Mosmann: Validation, investigation. A. Kearns: Validation, investigation. G. Eden: Validation, investigation. S. Alfred: Resources, validation, investigation. S. Allan: Validation, investigation, writing–review and editing. L. Zhai: Conceptualization, investigation. A. Kamili: Conceptualization, resources, writing–review and editing. A.J. Gifford: Conceptualization, resources, validation, investigation, writing–review and editing. D.R. Carter: Conceptualization, writing–review and editing. M.J. Henderson: Conceptualization, resources, investigation, methodology, writing–review and editing. J.I. Fletcher: Conceptualization, resources, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. G. Marshall: Conceptualization, writing–review and editing. R.W. Johnstone: Conceptualization, resources, formal analysis, investigation, methodology, writing–review and editing. A.J. Cesare: Conceptualization, resources, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. D.S. Ziegler: Conceptualization, formal analysis, supervision, funding acquisition, methodology, project administration, writing–review and editing. A.V. Gudkov: Conceptualization, resources, formal analysis, supervision, funding acquisition, methodology, writing–review and editing. K.V. Gurova: Conceptualization, resources, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. M.D. Norris: Conceptualization, formal analysis, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing. M. Haber: Conceptualization, formal analysis, supervision, funding acquisition, investigation, methodology, writing–review and editing.
We would like to thank QIMR Histology Staining Facility, Garvan Histopathology Facility, the Animal Facility at Children's Cancer Institute, and Ramaciotti Centre for Genomics at UNSW, for providing technical assistance for this project. We thank Scott Page and the Australian Cancer Research Foundation Telomere Analysis Centre at the Children's Medical Research Institute (CMRI) for microscopy infrastructure. L. Xiao and E. Ronca are supported by a grant from Cancer Institute NSW (ECF171127) and philanthropy from Neuroblastoma Australia. K. Somers is supported by funding from the Kids Cancer Alliance (KCA). K. Somers, M. Karsa, A. Bongers, and M. Henderson are supported by funding from Tenix Foundation and Anthony Rothe Memorial Trust. D. Carter and L. Zhai are supported by grant (1123235) awarded through the Priority-driven Collaborative Cancer Research Scheme and cofunded by Cancer Australia and The Kids' Cancer Project. A. O'Connor and A. Cesare is supported by grants from the National Health and Medical Research Council (NHMRC; 1053195, 1106241, 1104461), and philanthropy from the Goodridge Foundation and Stanford Brown, Inc (Sydney, Australia). M. Haber and M. Norris are supported by grants from NHMRC (APP1132608 and APP1085411), Cancer Institute NSW (14/TPG/1–13), Cancer Council NSW (PG 16–01), Tour de Cure, and Neuroblastoma Australia.
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.