Abstract
Purpose: Ewing sarcoma (ES) is a rare and highly malignant cancer that occurs in the bone and surrounding tissue of children and adolescents. The EWS/ETS fusion transcription factor that drives ES pathobiology was previously demonstrated to modulate cyclin D1 expression. In this study, we evaluated abemaciclib, a small-molecule CDK4 and CDK6 (CDK4 and 6) inhibitor currently under clinical investigation in pediatric solid tumors, in preclinical models of ES.
Experimental Design: Using Western blot, high-content imaging, flow cytometry, ELISA, RNA sequencing, and CpG methylation assays, we characterized the in vitro response of ES cell lines to abemaciclib. We then evaluated abemaciclib in vivo in cell line–derived xenograft (CDX) and patient-derived xenograft (PDX) mouse models of ES as either a monotherapy or in combination with chemotherapy.
Results: Abemaciclib induced quiescence in ES cell lines via a G1 cell-cycle block, characterized by decreased proliferation and reduction of Ki-67 and FOXM1 expression and retinoblastoma protein (RB) phosphorylation. In addition, abemaciclib reduced DNMT1 expression and promoted an inflammatory immune response as measured by cytokine secretion, antigen presentation, and interferon pathway upregulation. Single-agent abemaciclib reduced ES tumor volume in preclinical mouse models and, when given in combination with doxorubicin or temozolomide plus irinotecan, durable disease control was observed.
Conclusions: Collectively, our data demonstrate that the antitumor effects of abemaciclib in preclinical ES models are multifaceted and include cell-cycle inhibition, DNA demethylation, and immunogenic changes.
Translational Relevance
Abemaciclib is a highly selective small-molecule inhibitor of CDK4 and CDK6 recently approved for the treatment of metastatic breast cancer and currently under clinical evaluation for multiple indications including pediatric solid tumors (NCT02644460). Recently, it was reported that cancers with both D-cyclin–activating mutations and functional RB are sensitive to abemaciclib. Ewing sarcoma (ES) is driven by a chimeric oncogenic transcription factor implicated in activating the CDK4-cyclin D-RB axis. In this study, we demonstrate that abemaciclib reduced tumor size in preclinical ES models; combination with cytotoxic chemotherapies enhanced this antitumor activity resulting in durable responses. Furthermore, in addition to the well-characterized and expected abemaciclib-induced cell-cycle block, we report newly discovered abemaciclib-related effects, including regulation of DNA methylation and interferon pathway signaling, which may enhance abemaciclib activity.
Introduction
Ewing sarcoma (ES) is a rare and aggressive tumor occurring in the bone and/or surrounding soft tissue of children and adolescents (1–3). Treatment regimens composed of surgery, radiation, and high-dose chemotherapy have improved outcomes for patients with localized disease, with a 5-year overall survival of 50% to 70% (4, 5). However, ES continues to present a clinical challenge in the recurrent or metastatic disease setting, where the 5-year overall survival rate drops to <30%. Furthermore, ES survivors often suffer from chronic conditions attributed to intensive therapy (6, 7). Therefore, it is critical to find new agents to improve patient outcomes and limit long-term side effects related to treatment.
The pathogenic biology of ES is not well understood and thus complicates the identification of potential therapeutic targets. ES is characterized by specific chromosomal translocations resulting in a chimeric EWS/ETS transcription factor, the most common of which is EWS/FLI1 generated by the t(11;22)(q24;q12) translocation (8, 9). Although the oncogenic EWS/FLI1 protein is thought to contribute to tumorigenesis through cell-cycle modulation (10, 11), its expression alone is insufficient to promote tumor formation; furthermore, EWS/FLI1 is historically difficult to drug and therapeutic pursuit of its downstream targets is the subject of much debate (12). Interestingly, ES is reported to have a high degree of DNA methylation compared with normal human mesenchymal stem cells, which can lead to epigenetic silencing of key tumor suppressor genes (13, 14). There is some evidence that EWS/FLI1 may play a role in this transcriptional epigenetic modification (15) and, therefore, contribute to disease progression through the resulting deregulation of methylation patterns (13). Furthermore, ES, like other pediatric tumor types, has a low mutational burden compared with adult cancers (16); however, there are several genetic aberrations frequently reported in ES which contribute to cell-cycle dysregulation. For example, homozygous or hemizygous deletion of CDKN2A (which encodes p16INK4a, a tumor suppressor protein that binds to CDK4 and CDK6) is reported in 10% to 30% of primary ES patient tumors and up to 50% of ES cell lines and is a significant predictor of a poor prognosis for ES patients (8, 10, 17, 18). The combination of EWS/ETS fusion protein activities, genetic mutations, and epigenetic mechanisms likely all contribute to the complexity of ES pathogenesis.
Abemaciclib (LY2835219; Verzenio™) is a small-molecule inhibitor of cyclin-dependent kinase 4 (CDK4) and CDK6 recently approved for the treatment of HR+/HER2− advanced breast cancer. In addition, abemaciclib is currently in clinical development for non–small cell lung and other adult solid cancers (19–21) as well as newly diagnosed diffuse intrinsic pontine glioma and pediatric recurrent or refractory solid tumors (NCT02644460). Cell-based studies in preclinical cancer models confirmed that the G1 cell-cycle arrest resulting from abemaciclib-mediated CDK4/6 inhibition requires a functional RB protein (RB+) and leads to sustained antitumor effects through the induction of senescence, apoptosis, and altered cellular metabolism (22–24). Interestingly, preclinical models of human adult and pediatric cancers with D-type cyclin-activating features (DCAF, genomic aberrations known to elevate D-cyclin levels) were recently shown to be highly sensitive to abemaciclib (25). Furthermore, previous studies revealed that cyclin D1 is a direct transcriptional target of EWS/FLI1 (26, 27), and ES cell lines require both CDK4 and cyclin D1 for survival (28). In this report, we demonstrate that abemaciclib has single-agent activity in preclinical ES models through the expected G1 cell-cycle arrest and can be combined with chemotherapy to promote durable in vivo responses; in addition, our data implicate abemaciclib in the reversal of inhibitory epigenetic modifications and activation of interferon pathway signaling.
Materials and Methods
Cell culture
Human ES cell lines A673, SK-ES-1, RD-ES, and SK-N-MC were purchased from ATCC (cat#CRL-1598, HTB-86, HTB-166, and HTB-10) and maintained in suggested medium. Cado-ES-1 and MHH-ES-1 were obtained from The German Collection of Microorganisms and Cell Culture (DSMZ, cat#255 and 167) and maintained in RPMI 1640 with 10% heat-inactivated FBS. ES-1, ES-2, ES-3, ES-4, ES-6, ES-7, ES-8, EW8, TC71, CHLA-258, RH18, RH30, RH36, RH41, RD, JR1, and SMS-CTR were kindly supplied by Dr. Peter Houghton (Greehey Children's Cancer Research Institute; ref. 29) and maintained in RPMI 1640 with 10% heat-inactivated FBS and 2 mmol/L glutamine. All cells were maintained at 37°C and 5% CO2 in tissue culture-treated flasks.
Test compound
Abemaciclib (LY2835219, Eli Lilly and Company) was dissolved in DMSO at a stock concentration of 10 mmol/L for in vitro use and prepared in 1% HEC in 25 mmol/L phosphate buffer, pH 2, for in vivo experiments.
Cell proliferation
Assessment of cell proliferation following 96 hours of abemaciclib treatment was conducted using CellTiter-Glo Luminescent Cell Viability Assay (Promega, cat#G7571), alamarBlue assay (Thermo Fisher Scientific, cat#DAL1025) as previously described (30), or by quantification of viable cells on a Beckman Coulter ViCell XR. Absolute EC50 values were calculated using GraphPad Prism 7 (GraphPad Software, Inc.) or by XLFit software (Microsoft) and presented as the average of triplicate experiments ± standard error of the mean (SEM).
Protein analysis
Generation of whole-cell lysates, SDS–PAGE, and immunoblot procedures were conducted as previously described (31). Tumor lysates were generated by homogenizing excised tumors in 1% SDS, boiling for 5 minutes, and clarifying by centrifugation. Antibodies and conditions are listed in Supplementary Table S1.
Conditioned media were collected from abemaciclib-treated cells at indicated time points and frozen at −80°C for downstream analysis. Relative expression levels of secreted cytokines from conditioned media were determined using the Human XL Cytokine Array Kit according to the manufacturer's protocol (R&D Systems, cat#ARY022). Quantitation of IP-10, MCP-1, and IL8 from conditioned media was performed using Duoset ELISA Kits from R&D Systems (cat#DY266, DY279, and DM3A00). Absorbance was measured at 450 nm on a Spectromax250 (Molecular Devices). Presence of IFNα, β, γ, and λ was analyzed using the U-PLEX Interferon Combo (hu) kit (Meso Scale Diagnostics, cat#K15094K-1), Human IL29/IL28B Duoset ELISA (R&D Systems, cat#DY1598B), Human IFNγ High Sensitivity ELISA (eBioscience, cat#BMS228HS), and VeriKine-HS Human IFN Beta serum ELISA Kit (PBL Assay Science, cat#41415) according to the manufacturer's protocol.
High-content imaging
High-content imaging and cell immunofluorescence (IF) analysis were performed as previously described (32, 33). Briefly, cells were treated with abemaciclib in flasks for long-term treatments then seeded in poly-D-lysine coated 96-well black/clear plates (Corning, cat#354640). For experiments with ≤4 days of abemaciclib treatment, cells were seeded directly into plates. Following treatment, cells were fixed in 3.7% formaldehyde (Sigma-Aldrich, cat#F-1268) in Dulbecco's PBS (D-PBS), permeabilized with 0.1% Triton X-100 in D-PBS, and blocked with 1% BSA in D-PBS. Cells were incubated with primary antibodies overnight at 4°C, followed by 3 washes and incubation with secondary antibodies for 1 hour at room temperature. Antibody details are in Supplementary Table S1. DNA was stained with Hoechst 33342. Click-iTTM EdU Alexa FluorTM 488 HCS assay was performed per the manufacturer's protocol following a 30-minute incubation with EdU (Molecular Probes, cat#C10350). Cells were imaged using a CellInsight NXT platform and analyzed by the TargetActivation V.4 Bioapplication (Thermo Fisher Scientific). Gates for percent responders (percent positive for desired marker) were set based on the DMSO-treated group for each cell line.
Evaluation of DNA methylation
Genomic DNA was isolated according to the manufacturer's protocol using Wizard Genomic DNA Purification Kit (Promega, cat#A1120) and evaluated using the MethylFlash Methylated DNA Quantification Kit per the manufacturer's instructions (EpiGentek, cat#p-1034). Individual promoter methylation analysis by pyrosequencing was conducted at the University of Texas Health Science Center San Antonio Bioanalytics and Single-Cell Core. Genomic DNA (300 ng/sample) was subjected to bisulfite conversion using the EZ DNA Methylation Kit (Zymo Research, cat#D5001). PyroMark CpG Methylation Assays were used to detect methylation at predicted CpG sites (Qiagen, cat#978746). Methylated CpG sites were detected by the PyroMark Q96 MD System. Specific assays used were Hs-CDK6_01_PM00029239, Hs-CDK6_02_PM00020246, Hs-RASSF1_01_PM00013293, Hs-RASSF1_02_PM00013300, and Hs-RASSF1_03_PM00013307. The LINE-1_a5M assay was custom designed with the following primers: Forward, AGATTATATTTTATATTTGGTTTAGAGG; reverse, ACCCCCTTACAATTTAATCTCAAACTACTA; sequencing, ATTTGGTTTAGAGGGTT. Methylated CpG sites were detected by the PyroMark Q96 MD System. Incomplete bisulfite conversion checkpoint was set as 5%. The methylation percentage of each interrogated CpG site was calculated and visualized using the MultiExperiment Viewer v4.8 (Dana-Farber Cancer Institute, Boston, MA) and GraphPad Prism 7.
Flow cytometry
C12FDG assays were performed as previously described (34), and cells were analyzed on a Beckman Coulter F500 using WinList6.0 software. For BrdU analysis, cells were pulsed with 10 μmol/L BrdU for 1 hour prior to harvest, then fixed in ice-cold 70% ethanol. Cells were washed and incubated in 2 mol/L HCl for 20 minutes to denature DNA. After washing, HCl was neutralized with 0.1 mol/L sodium borate, pH 8.5. Cells were stained with anti-BrdU FITC-conjugated antibody for 30 minutes, washed, and stained with propidium iodide for 30 minutes. Samples were run on a Beckman Coulter FC500. Data were analyzed using FlowJo v10 and ModFit LT (Verity House Software).
Reverse transcription quantitative PCR and RNA sequencing (RNA-seq)
RNA was extracted from cells grown in continual presence of abemaciclib using the RNeasy Mini Kit (Qiagen, cat#74104). cDNA prepared using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, cat#43-688-14) was added to TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific, cat#4444557) along with the appropriate TaqMan assay and run using the ViiA 7 Real-Time PCR System (Applied Biosystems). TaqMan assays (Thermo Fisher Scientific) used were as follows: Hs0098418_m1 (IDO1), Hs00818803_g1 (HLA-B), Hs01013996_m1 (STAT1), Hs00292922_m1 (ERVH48-1), Hs01060665_g1 (ACTB), Hs00171042_m1 (CXCL10), Hs00971960_m1 (IRF1), and Hs02758991_g1 (GAPDH). Fold changes were calculated using the ΔΔCt method. RNA-seq was performed at the UCLA Technology Center for Genomics and Bioinformatics on a Illumina HiSeq3000 using the TruSeq Stranded mRNA Library Prep Kit with paired end sequencing, read length of 150 bp, and targeted read depth of 100M reads/sample. Reads were quality trimmed by sickle v1.33 with default parameters. Sequencing reads were mapped to human (GRCh37.p13) reference genome using GSNAP (v.2013-11-27). Read counts were generated against exons annotated in NCBI gene models (genome builds downloaded in December 2013), then summed at the gene level to provide a single number per gene per sample using a custom Perl script. RNA-seq data were further subjected to a QC pipeline developed at Eli Lilly and Company. Briefly, base quality/base composition, heterologous organism contamination, adapter content, mapping rate/mapped read counts, 3′ bias, template length, sample identity, and rRNA/mito content were checked.
In vivo evaluation
In vivo studies were approved by the Eli Lilly and Company Animal Care and Use Committee. To evaluate abemaciclib in cell line–derived xenografts (CDX), 5 × 106 ES cells were resuspended in Hank's Balanced Salt Solution and injected subcutaneously into female nude mice. After randomization into treatment groups based on tumor volume and body weight, mice were given vehicle (1% HEC in 25 mmol/L phosphate buffer, pH 2) or 50 mg/kg abemaciclib orally once daily for 28 days (QD×28). Tolerability of combination treatments (abemaciclib + chemotherapy) was considered acceptable if mice did not lose >20% body weight during treatment. For combination studies, animals were given abemaciclib (50 mg/kg, QD × 28, p.o.) plus doxorubicin [3 mg/kg, once weekly for 4 weeks (Q7D × 4), i.v.], cyclophosphamide (100 mg/kg, Q7D × 4, i.p.), temozolomide [TMZ, 66mg/kg, once daily for 5 days then rested for 16 days for two cycles (QD × 5, rest ×16) × 2, p.o.], and/or irinotecan [2.5 mg/kg, (QD × 5, rest ×16) × 2, i.p.]. For abemaciclib/TMZ combinations, administration was offset by 12 hours on the same dosing day. For in vivo target inhibition experiments, mice were sacrificed after 10 days of treatment. Half of each excised tumor was frozen at −80°C for future protein analysis by Western blot and half was formalin-fixed and paraffin embedded for immunohistochemistry (IHC). Fluorescence-based IHC was performed and analyzed according to previously described methods (35). Fourteen patient-derived ES xenograft (PDX) models were treated as described above and evaluated at Champions Oncology. For analysis of combination effects, the BLISS independence method was used to define a statistically significant effect. Combinations were defined as additive if the combination arm was statistically different from both of the single-agent arms.
Results
High cyclin D1 expression in ES cell lines correlates to abemaciclib sensitivity
Cancer cell lines most sensitive to abemaciclib were previously shown to have DCAF (25); the EWS/FLI oncogenic transcription factor may also be considered a DCAF as it directly promotes expression of cyclin D1 (26, 27). In a large cell line panel consisting of both adult and pediatric tumor cells, 50% of ES cell lines (3/6) and 11/27 (40.7%) of other tumor cell lines with DCAF were highly sensitive to abemaciclib at clinically achievable concentrations (EC50 ≤ 1 μmol/L), whereas only 4.2% of cell lines with mutant RB1 (2/48) had submicromolar EC50 values (Fig. 1A). SK-N-MC, which has compromised RB function (RB−) due to the RB1-R698M/S homozygous mutation, was determined to be the least sensitive ES cell line to abemaciclib and was, therefore, used as an abemaciclib-resistant control cell line. Evaluation of abemaciclib in a focused panel of 17 ES cell lines validated this observation as the majority of cell lines (16/17; 94.1%) were determined to have antiproliferative absolute EC50 values below 1 μmol/L (Fig. 1B). As DCAFs elevate D-cyclin expression, cyclin D1 protein levels were evaluated across a panel of pediatric cell lines (Fig. 1C). Strong cyclin D1 signal was readily apparent in ES cells but not in cancer cell lines less sensitive to abemaciclib; however, the absolute expression level of cyclin D1 did not correlate with the rank order of abemaciclib sensitivity across the ES cell lines. Interestingly, most ES cell lines expressed low levels of CDK4 and lacked CDK6 expression when compared with non-ES cell lines.
ES and other cell lines with DCAF are highly sensitive to abemaciclib. A, Box plot graphical representation of abemaciclib sensitivity in a 560-cell line panel. Data are divided into categories based on those cells with an EWS/FLI1 fusion (EWS.FLI), cells with DCAF, cells with RB mutation (RB), or all other cells (other). +, geometric mean; black diamond, SK-N-MC ES cell line which has both the EWS/FLI fusion protein and an RB1 mutation; gray diamond, NCI-H1048 small cell lung cancer cell has both DCAF and an RB1 mutation. Ordinary one-way ANOVA of log EC50 values with Sidak multiple comparison test. *, P < 0.05; ****, P < 0.0001. For a complete list of cell lines tested, see Supplementary Table S1 of ref. 25. B, Cell lines were treated with abemaciclib for 96 hours and proliferation determined using either cell counts or alamarBlue assay. Absolute EC50 values are the average of three biological replicates. Black bars, ES cell lines; gray bars, rhabdomyosarcoma (RMS) cell lines; error bars, SEM; dotted line, clinically achievable sustained concentration. C, Whole-cell lysates from a panel of pediatric cell lines from multiple histologies: ES (black bars), neuroblastoma (CHP, KELLY, IMR-32), RMS (SJCRH30, RH41, RD), and osteosarcoma (MG-63, HuO9) were analyzed by Western blot for the indicated proteins. Quantitation below Western blot is cyclin D1 normalized to GAPDH.
Abemaciclib induces a G1 cell-cycle arrest in RB+ ES cell lines
In RB+ cell lines (ie, those with functional RB), CDK4/6 inhibition results in RB hypophosphorylation leading to a cell-cycle block at the G1–S transition and decreased E2F-driven transcription (23). Following abemaciclib treatment, RB phosphorylation at both serine 807 and the CDK4/6 specific serine 780 (22) decreased, protein expression of both the E2F target topoisomerase IIα and the cell-cycle transcription factor FOXM1 decreased, and cyclin E increased in RB+ cell lines, all consistent with a G1–S cell-cycle block (Fig. 2A). Cyclin D1 slightly increased with abemaciclib treatment in the RB+ cell lines but was unchanged in SK-N-MC RB− cells. Evaluation of DNA content in abemaciclib-treated ES cells confirmed a G1 cell-cycle arrest in RB+ cells (Fig. 2B); furthermore, a reduction in DNA synthesis, as evidenced by reduced BrdU staining, was observed in all RB+ cells (Fig. 2C).
Abemaciclib induces a cell-cycle block in ES cell lines with intact RB. A, After treatment with abemaciclib for 48 hours, cells were lysed and the indicated total and phosphorylated proteins were assessed by Western blot analysis. B and C, After 96 hours of treatment with 500 nmol/L abemaciclib, cells were pulsed with BrdU for 1 hour and then fixed. Cells were costained with anti-BrdU and PI and analyzed by flow cytometry. B, Cell cycle was based on PI staining and modeled with ModFit software. Cell lines with RB proficiency are encompassed by the gray box. Percentage of cells in S-phase was determined by BrdU analysis and is indicated by the boxed-in portion of the graph.
Abemaciclib promotes a quiescent phenotype and resistance to chemotherapy in ES cell lines
Preclinical studies have previously demonstrated that cell-cycle arrest caused by CDK4/6 inhibition can lead to senescence in multiple tumor types (23, 24, 36, 37). Senescence is a complex phenotype ultimately defined by an irreversible cell-cycle exit (38); therefore, we evaluated abemaciclib-treated ES cell lines for senescence by measuring multiple endpoints. An increase in C12FDG, the substrate for β-galactosidase, was observed in A673 cells following 14-day treatment with abemaciclib (Supplementary Fig. S1A). In general, abemaciclib also reduced Ki-67 expression in RB+ cell lines indicating decreased proliferation and promoted a senescence-like phenotype as evidenced by lower FOXM1 expression, increased p21 expression (Supplementary Fig. S1B), and increased cell size (Supplementary Fig. S1C). We also assessed the senescence-associated secretory phenotype (SASP; ref. 36) through measurement of secreted cytokines from conditioned media (CM) of abemaciclib-treated ES cells. Although secretion of some SASP proteins (namely, CCL2/MCP-1 and IL8) was noted, most cytokine increases in the CM were associated with an acute immune response (Supplementary Fig. S1D). Although abemaciclib-treated RB+ cell lines exhibited multiple characteristics of cellular senescence, none of the cell lines were positive for all senescence-related endpoints.
Abemaciclib-treated RB+ ES cells developed resistance to chemotherapy and targeted agents alike, which is characteristic of senescence; however, unlike true senescence, this acquired resistance was reversible following removal of abemaciclib from culture medium (Fig. 3A). Furthermore, A673 cells were able to proliferate following abemaciclib washout as measured by an increase in cells undergoing DNA synthesis and phosphorylation of RB protein (Fig. 3B–D). Therefore, although abemaciclib-treated ES cells displayed various features of senescence, they were ultimately able to reenter the cell cycle and regain sensitivity to other therapeutic agents, suggesting that abemaciclib promotes quiescence rather than senescence in ES.
Abemaciclib-treated ES cells are resistant to chemotherapy agents during simultaneous treatment but regain sensitivity upon abemaciclib washout. A, ES cell lines were treated with 1 μmol/L abemaciclib or vehicle control continuously for 14 days. Cells were then cotreated with the indicated drug or abemaciclib was washed out for 7 days prior to drug treatment. Cell proliferation was measured 72 hours postdrug treatment and EC50 values for chemotherapy agents calculated based on a 10-point serial dilution curve. B–D, A673 cells were treated continually for 14 days with abemaciclib or abemaciclib was washed out prior to fixation (Day post-WO) and replaced with normal growth media. On day 14, cells were fixed, stained, and imaged for active DNA synthesis by EdU expression (B) and RB phosphorylation (C). Data are presented as the mean of four technical replicates ± SEM measuring % positive cells. D, Representative images from each time point taken at 20X magnification. Blue, nuclei; green, EdU; red, pRB.
Abemaciclib diminished DNMT1 protein and increased interferon signaling in ES cell lines
Selective CDK4/6 inhibitors including abemaciclib can promote an immune response through suppression of the RB-E2F-DNMT1 axis, resulting in global upregulation of interferon (IFN) signaling in breast cancer models (39). Similarly, we observed a sustained decrease in DNMT1 protein and mRNA levels in RB+ ES cells after treatment over a range of time points from 24 hours to 16 days (Fig. 4A and B). Abemaciclib reduced global DNA methylation levels in RB+ cell lines, but without concomitant decrease in DNA promoter methylation of CDK6, LINE-1, or RASSF1 (Supplementary Fig. S2B and S2C). In addition to the reduction of cell-cycle–related E2F transcriptional targets, such as CCNA2, CCNB1, E2F2, FOXM1, AURKA, and AURKB in all RB+ ES cell lines (Fig. 4B), abemaciclib elicited a variable immune response as measured by elevation in secreted proinflammatory cytokines (Supplementary Fig. S1D); marked upregulation of the canonical IFN signaling pathway (Fig. 4A–C; Supplementary Fig. S2A and S2E); and increased expression of the antigen presentation protein MHC-1 which was maintained at both the transcript and protein level after abemaciclib washout (Fig. 4A and D; Supplementary Fig. S2E).
Abemaciclib treatment induces a loss of DNMT-1, active interferon pathway signaling, and durable antigen presentation response in vitro. ES cell lines were treated continually with 500 nmol/L abemaciclib for up to 16 days. A, Whole-cell lysates were analyzed by Western blot for expression of the indicated proteins. B, Day 14 biological triplicate samples were split for RNA and protein analysis, RNA-seq results are presented as a heat map. ANOVA models were used to eliminate machine, flow-cell, and lane effects. Gene count data were quantile normalized and log2 transformed. For each gene, the difference in expression between the control and abemaciclib-treated in each cell line was estimated using linear regression model. The Bonferroni correction P value of 2.43E−06 was used to determine differential expression. Genes presented are split into four groups depicted by colored bar, Blue, E2F target genes showing cell-cycle changes; orange, DNMT group; purple, interferon-stimulated genes; green, HLA genes. C, Interferon α, β, γ, and λ levels were assessed in the conditioned media of ES cells treated for 14 days with 500 nmol/L or 1 μmol/L abemaciclib. Changes in IFNλ (left) and IFNβ (29) levels are shown. IFNα and IFNγ were not detected in any samples. Dotted line represents lowest IFN concentration of standard curve. n = 3, two-way ANOVA analysis, ***, P < 0.001; ****, P < 0.0001. D, A673 cells were treated continually for 14 days with abemaciclib or abemaciclib was washed out 4, 3, 2, and 1 day prior to fixation. Top left, %POSITIVE MHC-1 is mean of 4 technical replicates ± SEM. Top right, Images show representative fields taken at 20× magnification for the indicated timepoint. Post, days postwashout. Blue, nuclei; yellow, MHC-1. Bottom, ES cell lines were treated continually for 14 days with abemaciclib or abemaciclib was washed out 96, 72 48, or 24 hours prior to harvest and replaced with growth media. On day 14, cells were lysed and indicated proteins were analyzed by immunoblot.
Based on the apparent increase in IFN/JAK/STAT1 pathway activity, we measured secreted IFNs from CM of abemaciclib-treated cells. Surprisingly, IFNα, -β, -γ, and -λ were largely undetected in ES cell CM, with the exception of IFNβ and -λ in CM from RD-ES (Fig. 4C). In addition, MHC-1 upregulation only occurred when ES cell lines were stimulated with IFN concentrations detectable by ELISA (Supplementary Table S2). These data suggest that selective inhibition of CDK4/6 by abemaciclib may elicit a ligand-independent IFN response in ES cells.
Combination of abemaciclib with chemotherapy promotes durable in vivo responses in preclinical ES models
Abemaciclib was initially evaluated in the RD-ES CDX mouse model where stable disease was achieved with single-agent treatment (Fig. 5A); however, tumor growth resumed after the end of the dosing interval. Immunofluorescent analysis of abemaciclib-treated tumors confirmed a G1–S cell-cycle block by diminished Ki-67, decreased nuclear DNA content of cells, and reduced TUNEL staining when compared with control tumors (Fig. 5B). Areas of control tumors with high TUNEL staining corresponded to the necrotic core of rapidly growing tumors. Inhibition of tumor growth positively correlated with a statistically significant reduction of RB phosphorylation and topoisomerase IIα protein expression, as expected following prolonged CDK4 inhibition (Fig. 5C).
Abemaciclib has single-agent activity in the xenograft model RD-ES giving a stable disease response for duration of treatment. Mice bearing RD-ES xenografts were treated orally with 50 mg/kg abemaciclib (
) or vehicle (•) daily. A, Tumor growth over time. Dotted lines indicate dosing interval. Mean ± SEM is displayed, n = 3/group. B and C, Xenografts were harvested after 10 days of treatment. B, IHC analysis for cell death (TUNEL), cell proliferation (Ki-67), and cell number (DNA content). Bar graphs under each set of representative images quantify staining for each endpoint, n = 8 fields/group. One-way ANOVA with Dunnett comparison. C, Analysis of protein expression by immunoblotting. Protein levels were quantitated using the Chemidoc XRS software (Bio-Rad). Reported values are normalized to GPDH (n = 8/group). Error bars, SEM. Paired Student t test; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.
As stable disease in mice bearing RD-ES tumors during abemaciclib monotherapy was not maintained beyond the end of treatment, studies combining abemaciclib with chemotherapeutic agents were conducted in 14 PDX models of ES using an “n of 1” study design. Abemaciclib plus doxorubicin increased response durability as measured by tumor doubling time, with combination treatment demonstrating superiority to either single agent in eight of 14 PDX models, while no added benefit was observed with abemaciclib/cyclophosphamide combinations (Supplementary Figs. S3 and S4). Further analysis in three of these PDX models using more animals (n = 3/arm) confirmed that abemaciclib plus doxorubicin promoted durable antitumor responses (stable disease or partial tumor regression; Fig. 6A and B; Supplementary Table S3; Supplementary Figs. S3 and S4). Abemaciclib-induced inhibition of tumor growth positively correlated with a reduction in RB phosphorylation, topoisomerase IIα, and DNMT1 expression (Fig. 6A; Supplementary Figs. S3C, S3E, and S4).
Abemaciclib plus chemotherapy demonstrates additive effects of stable disease and tumor regression in preclinical models of ES. A, Animals bearing CTG-0142 PDX were given vehicle, abemaciclib (daily), doxorubicin (once weekly), or the combination for 28 days (n = 3/group). Dotted line marks end of dosing schedule. Additional mice (3/arm) were added to each cohort and xenografts were harvested after 10 days of treatment (x) and interrogated for expression of DNMT1 and topoisomerase IIα and phosphorylation of RB. B, Waterfall plot of CTG-0142 PDX study at day 28. C, Mice with A673 tumors were treated with vehicle, abemaciclib (daily), temozolomide (TMZ) and irinotecan (once daily for 5 days, rest for 16 days), or the combination (n = 5/group). Dotted line represents end of dosing schedule. D, Waterfall plot of A673 CDX study at day 31. Error bars, SEM. Blue bars, progressive disease (PD, ≥10%); red bars, stable disease (SD, −50% to 10%); green bars, partial response (PR, ≤−50% and >14 mm3).
Patients with relapsed or recurrent ES who receive treatment regimens including irinotecan and temozolomide (TMZ) have response rates of up to 68%, with a 1-year overall survival rate of 65% (40, 41); therefore, we also tested this combination plus abemaciclib in A673 and RD-ES CDX models and 9 “n of 1” PDX models (Fig. 6C and D; Supplementary Table S3; Supplementary Figs. S3 and S4). In the A673 model, the triple combination resulted in five animals with stable disease and 1 animal with a partial regression which was maintained beyond the end of the dosing interval (Fig. 6C and D; Supplementary Table S3 and Supplementary Fig. S4A). In the RD-ES model, the triple combination resulted in partial regression (Supplementary Table S3). In one PDX model, the triple combination resulted in a complete regression, while in the remaining majority of PDX models evaluated (8/9), the response to abemaciclib/irinotecan/TMZ was either stable disease or progressive disease (Supplementary Table S3). Combinations were well tolerated in mice based on body weight measurements (Supplementary Figure S4).
Discussion
ES is a devastating malignancy predominantly occurring in adolescents and young adults. Despite aggressive treatment regimens, patient prognosis is often poor and marked by tumor recurrence and/or metastasis. Furthermore, systemic therapy may result in chronic illness or development of secondary cancers; therefore, it is imperative to identify and evaluate new therapeutic options. Abemaciclib is a CDK4/6 inhibitor approved for the treatment of HR+/HER2− advanced breast cancer currently under clinical evaluation for both adult and pediatric solid malignancies. In this report, we evaluated abemaciclib in preclinical models of pediatric cancers and determined that ES, with functional RB tumor suppressor protein, was sensitive at clinically achievable concentrations. Abemaciclib displayed a three-pronged mechanism of action in vitro: inhibition of the cell cycle, deregulation of DNA methylation, and activation of the interferon pathway. Furthermore, tumor growth was inhibited by abemaciclib, either alone or in combination with chemotherapy, in multiple ES xenograft models.
Recently, our group reported that cancer cell-line sensitivity to abemaciclib was associated with genetic aberrations, which result in activated cyclin D and a dependence on D-cyclin genes for proliferation (25). As the EWS/FLI1 fusion oncoprotein can promote the expression of cyclin D1 (26, 27), it is reasonable to consider this chimeric transcription factor as an additional DCAF. Indeed, ES cell lines have higher relative cyclin D1 protein expression levels than non-DCAF, abemaciclib-resistant rhabdomyosarcoma (RMS) cell lines, and a super-enhancer screen identified both cyclin D1 and CDK4 as necessary for ES cell survival and anchorage-independent growth (28).
Senescence and apoptosis following inhibition of CDK4/6 have been well documented in preclinical models of breast cancer and neuroblastoma (23, 24). Although abemaciclib promoted various features of cellular senescence in ES cells, the absence of apoptosis and the observation that cells ultimately resumed growth upon abemaciclib withdrawal indicate induction of quiescence rather than senescence. Another unexpected function of abemaciclib in ES is its ability to negatively regulate DNA methylation, an aspect of the compound that has not been reported in pediatric cancer. High levels of DNA methylation reported in ES tumors and cell lines putatively contribute to a gene-silencing mechanism that inhibits tumor suppressor expression (13, 14). Consistent with this idea, reports have linked aberrant DNA methylation with the pathogenesis of many other pediatric tumor types (42). Abemaciclib rapidly decreased expression of DNMT1, the enzyme responsible for maintenance DNA methylation, in ES cells, thus reducing global DNA methylation; however, this did not lead to appreciable changes in DNA promoter methylation of 3 key genes. These findings are supported by prior observations that DNMT1 knockdown leads to limited genomic demethylation (43). Consistent with previous data generated in breast cancer models (39), abemaciclib did not affect DNMT3A expression. Both DNMT3A and 3B, which are responsible for de novo methylation, can functionally overlap with DNMT1 (43), and it is, therefore, possible that DNMT3A could compensate for the absence of DNMT1 and assume maintenance methylation activity. However, questions remain whether the abemaciclib-mediated loss of DNMT1 in ES leads to reexpression of other epigenetically silenced genes (e.g., tumor suppressors) or if DNMT3A plays a role in maintenance methylation.
Mechanistically, loss of DNA methylation may lead to elevated expression of endogenous retrovirus (ERV) genes, thereby activating cytosolic double-stranded RNA (dsRNA)-sensing pathways and inducing an IFN response (44). Following abemaciclib treatment, RB+ ES cell lines exhibited a prototypical IFN response, as evidenced by increased gene transcripts for STAT-1, IRF-1, CXCL-10, IDO-1, and HLA-B. STAT1 pathway activation was confirmed at the protein level by increased pSTATY701, IRF9, and MHC-1. However, this response seems to be ligand-independent as we were unable to detect expression of any interferon ligands in the CM of all but one ES cell line. In RD-ES CM, low levels of IFNβ and IFNλ (also known as type I and type III IFN, respectively) were detected after abemaciclib treatment. Types I and III IFN both induce strong antiviral responses but differ in the activating ligands (type I, IFNα/β; type III, IFNλ) and receptor distribution (type I, global; type III, restricted to mucosal epithelial cells; refs. 45, 46). The epithelial-restricted expression of type III IFN was recently illustrated by the sole induction of IFNλ and its associated downstream immune response observed in breast cancer cell lines following abemaciclib treatment (39). As ES is by definition a mesenchymal tumor, the abemaciclib-induced IFNλ increase detected in RD-ES cells is an interesting exception to the current understanding of type III IFN receptor distribution, as is the apparent ligand-independent IFN pathway activation in other ES cell lines. Although it is possible that trace amounts of IFN may have been secreted at levels below detection limits, it is unlikely given that MHC-1 was not expressed in ES cells following stimulation with IFN concentrations well within ELISA detection limits. This observation suggests that the abemaciclib-induced MHC-1 response could be a ligand-independent intrinsic cellular response with a yet undefined mechanism.
The innate immune response to virus includes IFN secretion, which stimulates release of inflammatory chemokines such as CXCL10/IP-10 and IL8. These chemokines act as chemoattractants for activated T cells, neutrophils, and basophils, and result in upregulation of MHC class I cell-surface antigen proteins and mounting of an adaptive immune response. Recently, abemaciclib was shown to induce a T-cell inflammatory signature and upregulation of antigen presentation in tumors of syngeneic murine carcinoma models (47) and CDK4/CDK6 inhibition in conjunction with immune-checkpoint blockade is currently being tested in the clinic in breast and other adult solid tumors (NCT02791334, NCT03294694). Similarly, abemaciclib triggered a heightened immune phenotype analogous to an innate/adaptive response in ES cells; therefore, the potential clinical implications of this observation in pediatric oncology should be the subject of future studies and merits the development of immune-competent ES models. Combination of abemaciclib with chemotherapy improved the durability of disease control in preclinical ES models, though benefit was dependent on both the specific chemotherapy and the model. The additive effect of abemaciclib plus chemotherapy is a surprising observation given the in vitro resistance data; however, such an effect is reported with other chemotherapies in other contexts including temozolomide in glioblastoma (48) and antimitotics in breast cancer (49), suggesting that our current assumptions concerning the combinability of CDK4/6 inhibitors with chemotherapy may be overly simplistic. Abemaciclib plus doxorubicin resulted in additive, durable responses ranging from stable disease to partial regression in multiple ES mouse models. Although the exact mechanism of action for this combination is currently undefined, ES cells treated with abemaciclib in vitro retained partial sensitivity to doxorubicin despite acquiring resistance to other chemotherapeutics, indicating that doxorubicin may act in a cell-cycle–independent manner. In addition, DNA intercalators, such as doxorubicin, can inhibit DNMT1 (50) and could potentially enhance the demethylating effects of abemaciclib. The combination of abemaciclib/TMZ resulted in an enhanced response significantly different from TMZ alone in mice bearing A673 xenografts. Interestingly, while TMZ/irinotecan controlled A673 tumor growth, the addition of abemaciclib prevented tumor regrowth during dosing holidays. In contrast to the superior responses observed with abemaciclib/doxorubicin or abemaciclib/TMZ/irinotecan combinations, the combination of abemaciclib plus cyclophosphamide was not superior to either single agent. It is possible that the sequencing of chemotherapy with abemaciclib may provide greater benefit than simultaneous administration and further studies are required to fully explore various regimens to identify an optimal treatment strategy.
Overall, abemaciclib exhibits a unique mechanism of action encompassing cell-cycle blockade, DNA demethylation, and activation of the adaptive immune response in ES models. Although our data strongly support evaluation of abemaciclib in immune-competent ES models, development of such models has proved elusive. Therefore, further studies of the multifaceted CDK4/6 inhibitor abemaciclib should be pursued in additional preclinical models of adult and pediatric malignancies.
Disclosure of Potential Conflicts of Interest
R. Flack holds ownership interest (including patents) in Eli Lilly. P. J. Ebert holds ownership interest (including patents) in Eli Lilly. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: M.S. Dowless, J.R. Stephens, S.G. Buchanan, P.J. Houghton, L.F. Stancato
Development of methodology: M.S. Dowless, J.R. Stephens, J.B. Olsen, P.J. Houghton
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.S. Dowless, T.J. Shackleford, M. Renschler, J.R. Stephens, R. Flack, W. Blosser, J. Stewart, J. Dempsey, A.B. VanWye, P.J. Ebert, J.B. Olsen, P.J. Houghton
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.S. Dowless, M. Renschler, W. Blosser, S. Gupta, J. Stewart, Y. Webster, J. Dempsey, A.B. VanWye, P.J. Ebert, P. Iversen, J.B. Olsen, X. Gong, S.G. Buchanan
Writing, review, and/or revision of the manuscript: M.S. Dowless, C.D. Lowery, T.J. Shackleford, R. Flack, A.B. VanWye, P.J. Ebert, P. Iversen, J.B. Olsen, S.G. Buchanan, P.J. Houghton, L.F. Stancato
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.S. Dowless, C.D. Lowery, M. Renschler, R. Flack, W. Blosser, A.B. VanWye, P. Iversen
Study supervision: M.S. Dowless
Acknowledgments
The authors would like to acknowledge the Cancer Prevention Research Institute of Texas for its support of the Bioanalytics and Single-Cell Core at the University of Texas Health Science Center San Antonio by grant RP150600. T. Shackleford received funding from grant CA165995. This study was funded by Eli Lilly and Company, Lilly Corporate Center, Indianapolis, Indiana.
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.
Footnotes
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
- Received April 25, 2018.
- Revision received June 19, 2018.
- Accepted August 16, 2018.
- Published first August 21, 2018.
- ©2018 American Association for Cancer Research.
References
Volume 24, Issue 23
