Abstract
The identification of new therapeutic strategies against osteosarcoma, the most common primary bone tumor, continues to be a primary goal to improve the outcomes of patients refractory to conventional chemotherapy. Osteosarcoma originates from the transformation of mesenchymal stem cells (MSC) and/or osteoblast progenitors, and the loss of differentiation is a common biological osteosarcoma feature, which has strong significance in predicting tumor aggressiveness. Thus, restoring differentiation through epigenetic reprogramming is potentially exploitable for therapeutic benefits. Here, we demonstrated that the novel nonnucleoside DNMT inhibitor (DNMTi) MC3343 affected tumor proliferation by blocking osteosarcoma cells in G1 or G2–M phases and induced osteoblastic differentiation through the specific reexpression of genes regulating this physiologic process. Although MC3343 has a similar antiproliferative effect as 5azadC, the conventional FDA-approved nucleoside inhibitor of DNA methylation, its effects on cell differentiation are distinct. Induction of the mature osteoblast phenotype coupled with a sustained cytostatic response was also confirmed in vivo when MC3343 was used against a patient-derived xenograft (PDX). In addition, MC3343 displayed synergistic effects with doxorubicin and cisplatin (CDDP), two major chemotherapeutic agents used to treat osteosarcoma. Specifically, MC3343 increased stable doxorubicin bonds to DNA, and combined treatment resulted in sustained DNA damage and increased cell death. Overall, this nonnucleoside DNMTi is an effective novel agent and is thus a potential therapeutic option for patients with osteosarcoma who respond poorly to preadjuvant chemotherapy. Mol Cancer Ther; 17(9); 1881–92. ©2018 AACR.
Introduction
Osteosarcoma, the most common primary bone tumor, occurs mainly in children and adolescents and therefore has an important social impact despite its rarity (1). To date, the most effective therapeutic strategy against osteosarcoma is surgical resection combined with chemotherapy (2, 3). This multidisciplinary approach has increased the survival of patients with localized tumors over the past few decades, achieving a 5-year survival rate of up to 70% (4, 5). However, outcomes for patients with metastasis at diagnosis or for those who fail first-line treatment remain unsatisfactory (6). The multiple and complex genetic aberrations that characterize osteosarcoma have prevented the identification of specific, common oncogenic drivers (7) and the successful application of targeted therapies, thwarting any improvement in survival (8). The adoption of immunotherapy with checkpoint inhibitors may also be relevant in the treatment of patients with sarcomas, as an increasing number of successes in other tumors have been reported. Such impressive results, however, have not been replicated in osteosarcoma, and novel therapeutic approaches are therefore urgently needed. Osteosarcoma is thought to originate from the transformation of mesenchymal stem cells (MSC) and/or osteoblast progenitors in any phase of differentiation (9). Transformation generates a block in normal development coupled with abnormal proliferation processes, and differentiation loss is a widespread biological feature of these tumors and has a strong prognostic significance (10). Thus, restoring differentiation by unlocking the stops driven by genetic and epigenetic alterations may be exploited as a therapeutic opportunity. Several lines of evidence have indicated that chromatin modifications in cancer versus normal MSCs include increased gene-specific DNA methylation and increased histone methylation, while the loss of DNA methylation occurs in normal processes of MSC differentiation (for a review, see ref. 11). Here, we analyzed the effects of a novel nonnucleoside DNA methyltransferase inhibitor (DNMTi), the N-(3-(2-amino-6-methylpyrimidin-4-ylamino)phenyl)-3-(quinolin-4-ylamino)benzamide MC3343, on cell proliferation, survival, and differentiation in a representative panel of osteosarcoma patient-derived cell lines and xenografts. DNA hypermethylation is a common feature in tumor development. In humans, DNA methylation occurs at the C-5 position of cytosines, mainly in a CpG dinucleotide context, and is catalyzed by members of the DNMT family of enzymes that transfer a methyl group from S-adenosyl-methionine (AdoMet) to the DNA cytosine C5 (12, 13). Catalytically active DNMTs (DNMT1, DNMT3a, and DNMT3b) share common features, especially a regulatory N-terminal domain and a catalytic C-terminal domain, which are conserved among different species. DNMT1, the first purified and characterized and most abundant DNMT in somatic cells, has a greater affinity for hemimethylated DNA rather than unmethylated DNA. DNMT1 acts mainly after DNA replication to methylate the newly synthesized strand and is typically known as a maintenance methyltransferase (13). DNMT1 is essential during developmental stages and in adult somatic cells to ensure cell proliferation and survival (14). In cancer cells, disruption of DNMT1, either genetically or pharmacologically through the use of DNMTis, is sufficient to stop the growth of tumors (12), reverse the undifferentiated state, and inhibit the invasiveness of cells, thus providing a very interesting therapeutic target (15). DNMTis are currently classified as nucleoside analogues, such as azacitidine (5azaC, Vidaza) and decitabine (5azadC, Dacogen), or nonnucleoside obtained by repositioning old drugs (e.g., hydralazine, procainamide, curcumin, and epigallocatechin gallate) or by identifying new drugs (e.g., RG-108, nitrophenyl-flavones, Δ2-isoxazolines, and SGI-1027; refs. 12, 16). The first category of drugs has been approved for the treatment of myelodysplasia and several solid tumors including some cases of neuroblastoma, despite their poor bioavailability, chemical instability, and toxic side effects (17–19). To date, DNMTis have been scarcely investigated in osteosarcoma; most existing studies involve two approved nucleoside inhibitors, 5azaC and 5azadC (4, 20–25), and scarce data are available regarding nonnucleoside DNMTis. To improve the efficacy and DNMT selectivity of nonnucleoside analogues, we chemically manipulated the SGI-1027 structure (26) by preparing all of its regioisomers, as well as some truncated and symmetric bis-quinoline and bis-pyrimidine analogues to test against DNMT1 and DNMT3a, and identified MC3343 as the best inhibitor (27). MC3343 is a strong DNA binder, showing preferential binding to DNA in CG-rich regions, and destabilizes the DNMT1–AdoMet DNA complex (28). Here, MC3343 efficiently inhibited osteosarcoma cell proliferation, induced the specific reexpression of osteoblastogenesis genes, and acted synergistically with doxorubicin and cisplatin (CDDP). We found that MC3343 improved the accessibility of doxorubicin to DNA and significantly increased doxorubicin-induced DNA damage and cell death.
Materials and Methods
Chemicals
MC3343 was prepared as reported previously (27, 28). 5azadC was purchased from Sigma-Aldrich. Stock solutions were prepared in DMSO and stored at −20°C.
DNMT1 and DNMT3a assays in a cell-free context
DNMT1 and DNMT3a enzyme inhibition assays were performed according to Valente and colleagues (27). A detailed description is provided in the Supplementary Materials and Methods.
DNMT activity assay in a cell context
DNMT activity was measured using an EpiQuik DNA Methyltransferase Activity/Inhibition Kit (EpiGentek). Cells were treated with MC3343 (1–10 μmol/L) or 5azadC (0.1–10 μmol/L) for 48 hours, and 5 μg of nuclear extracts isolated using the EpiQuik Nuclear Extraction Kit (EpiGentek) was added to each reaction well according to the manufacturer's protocol. Absorbance was determined using a microplate spectrophotometer at 450 nm (GloMax luminometer, Promega). The data are expressed as the mean ± SE of three independent experiments.
Cell lines
The patient-derived human osteosarcoma cell lines Saos-2, U-2OS, and MG-63 were obtained from the ATCC (1992). The patient-derived osteosarcoma cell lines IOR/OS9, IOR/OS10, IOR/OS14, IOR/OS20, and IOR/SARG were established and previously characterized at the Laboratory of Experimental Oncology of Rizzoli Orthopaedic Institute (Bologna, Italy; ref. 29). Cell lines were profiled for the DNA copy number changes (30) and DNA methylation status at approximately 27,000 CpG sites (31). The cell lines PDX-OS#1-C4 and PDX-OS#2-C2 were obtained from osteosarcoma patient-derived xenografts (PDX) after four or two passages in the animal, respectively (2015). All cell lines were tested for Mycoplasma contamination (Mycoalert Mycoplasma Detection Kit, Lonza) before use and authenticated by short tandem repeat PCR analysis; detailed information is provided in the Supplementary Materials and Methods. All cell lines were immediately amplified to constitute liquid nitrogen stocks and were never passaged for more than 1 month upon thawing. Cells were maintained in Iscove's Modified Dulbecco's Medium (IMDM) supplemented with 10% heat-inactivated FBS (Lonza), penicillin (20 U/mL), and streptomycin (100 μg/mL; Sigma-Aldrich) in a 37°C humidified environment at 5% CO2.
Cell culture experiments
To perform cell culture experiments, 2 × 105 cells per well were plated, and DNMTi MC3343 (1–30 μmol/L) or 5azadC (0.1–30 μmol/L) was added after 24 hours. Cells were exposed for up to 96 hours before being counted by Trypan blue vital cell count (Sigma-Aldrich). In parallel, cells were treated with DMSO-containing medium as a control. The highest final concentration of DMSO in the medium was ≤0.005%, and DMSO had no effect on cell growth inhibition. In combination experiments, cells were treated for 96 hours with MC3343 alone (control) or in combination with a fixed ratio of doxorubicin (Pfizer), CDDP (Sandoz), and methotrexate (Sigma-Aldrich). For cell-cycle analysis after 24 to 48 hours of treatment, 5-bromo-2′-deoxyuridine (BrdU, Sigma-Aldrich) labeling was used to assess cell proliferation. Then, the cells were processed for indirect immunofluorescence staining using anti‐BrdU mAb (1:8; BD Biosciences, #347580) as a primary antibody, anti‐mouse FITC as a secondary antibody (1:100; Thermo Fisher Scientific, #31569) and 20 μg/mL propidium iodide for the analysis of DNA content before flow cytometry analysis (Sysmex, Partec). To measure the intranuclear uptake of doxorubicin, living cells were exposed to doxorubicin (1–10 μmol/L for 30–60 minutes at 37°C; ref. 32) with or without MC3343 (10 μmol/L) or 5azadC (1 μmol/L). γ-H2AX expression and quantification was assessed by flow cytometry after staining with an anti–γ-H2AX antibody (Ser139, 1:50, Cell Signaling Technology, #9718) and anti-rabbit FITC as a secondary antibody (1:80; Dako, #F0205).
siRNA-mediated knockdown of DNMT1 and DNMT3a was performed with siRNAs from Thermo Fisher Scientific Dharmacon; SMARTPOOL siGENOME_siRNA (FE5M004605010020 for DNMT1 and FE5M006672030020 for DNMT3a). siGENOME_non targeting_siRNA was used as control (FE5D0012061320). siRNAs (25 and 50 nmol/L) were transfected into Saos-2 and PDX-OS#1-C4 cells using the transfection agent TransIT-x2 System (Mirus, #MIR6000) according to the manufacturer's instructions. For details, please refer to Supplementary Materials and Methods.
Osteosarcoma differentiation toward osteoblasts
Four days after seeding, osteosarcoma cells were exposed to specific osteogenic medium (IMDM supplemented with 2% FBS, 5 mmol/L β-glycerophosphate, and 50 μg/mL ascorbic acid, Sigma-Aldrich) without (control) or with MC3343 (1–5 μmol/L) or 5azadC (0.1–0.5 μmol/L) and maintained in differentiating conditions for up to 21 days. The medium was renewed every 4 days, and cells were harvested at various time points to collect total RNA and verify bone mineralization. Specific osteoblastic markers were evaluated by qRT-PCR as described previously (33). The primers or assays (Applied Biosystems) used are as follows: GAPDH forward: 5′-GAAGGTGAAGGTCGGAGTC-3′, reverse: 5′-GAAGATGGTGATGGGATTTC-3′ and probe: 5′-CAAGCTTCCCGTTCTCAGCC-3′; OCN, forward: 5′-GGGCTCCCAGCCATTGAT-3′ and reverse: 5′-CAAAGCCTTTGTGTCCAAGCA-3′; ALPL (Hs01029144_m1); and COL1A2 (Hs01028970_m1). Amplification reactions were performed using a ViiA7 Real-Time PCR System (Life Technologies). Osteoblastic differentiation was assessed after 7 to 14 days by alkaline phosphatase (ALPL) staining using a Leukocyte Alkaline Phosphatase Kit (Sigma-Aldrich) according to the manufacturer's instructions. To visualize bone mineralization, the plates were stained with 40 mmol/L Alizarin red stain (ARS, Sigma-Aldrich). ARS was extracted from the monolayer and transferred to a 96‐well plate for reading at 550 nm (GloMax luminometer, Promega). For a more general view of the osteogenic differentiation of osteosarcoma cells, the Human Osteogenesis RT2 Profiler PCR Array (Qiagen‐SA Biosciences) was used.
Immunofluorescence staining
Cells were seeded on coverslips and treated with increasing concentrations of MC3343 for up to 72 hours, fixed with 4% paraformaldehyde, and permeabilized with 0.15% Triton X‐-100/PBS. The samples were incubated with anti-DNMT1 (sc-271729), anti-DNMT3a (sc-20703), anti-DNMT3b (sc-130740) antibodies (1:50, Santa Cruz Biotechnology), or with MIB-1 antibody (1:50; Calbiochem-Novabiochem, NA21-2) and visualized with an ECLIPSE 90i microscope (Nikon) equipped with Plan Apo VC 60× oil NA 1.4. Images were captured under identical conditions using a digital color camera (Nikon DS5 MC) and NIS‐-Elements AR3.10 software (Nikon). The percentage of Ki-67–positive cells in samples of at least 500 cells was quantified and expressed as the Ki-67 labeling index.
Western blotting
Subconfluent cells were treated as described above and processed for Western blotting following standard procedures. The following primary antibodies were used: anti-DNMT1 (1:1,000, Bethyl Laboratories, A300-041A-M); anti-DNMT3a (1:1,000, sc-20703), anti-DNMT3b (1:1,000, sc-130740), anti-Cyclin D1 (1:2,000, sc-753), and anti-GAPDH (1:5,000, sc-25778; all purchased from Santa Cruz Biotechnology); anti-pRBs (S807/811, S795, 1:1,000, Rb Sampler Kit, #9969); anti-RB1 (1:1,000, #9309; all purchased from Cell Signaling Technology); and anti-ACTIN (1:1,000, Merck Millipore, MAB1501). Anti-rabbit or anti-mouse secondary antibodies conjugated to horseradish peroxidase (1:1,000–1:20,000, GE Healthcare) were used, and bands were visualized with enhanced chemiluminescence (ECL) Western blotting detection reagents (EuroClone).
PDXs from primary human osteosarcoma and in vivo treatments
Rag 2−/−; Il2rg−/− breeders were kindly provided by Drs. T. Nomura and M. Ito of the Central Institute for Experimental Animals (Kawasaki, Japan; ref. 34). Mice were then bred in our animal facilities under sterile conditions. Male mice ages 5 to 11 weeks were used to generate PDXs by subcutaneous xenografts of osteosarcoma fragments at the level of interscapular brown fat. PDX-OS#1 was derived from a primary high-grade osteoblastic osteosarcoma. Pharmacologic treatments began when the tumor reached a volume of 0.01 cm3. Tumor size was measured using a caliper, and tumor volume was calculated according to the following formula: tumor volume = π[√(a × b)]3/6, where a = maximal tumor diameter and b = tumor diameter perpendicular to a. The animals were randomized to receive intraperitoneal injections of 0.8 mg/kg 5azadC, peritumoral injections of 20 mg/kg MC3343, or the vehicle control (PBS:DMSO 1:1) 5 days a week. Doxorubicin was administered intraperitoneally for two consecutive days (50 μg/injection). Mouse body weights and tumor volumes were measured at least once a week.
IHC and von Kossa staining
An avidin–biotin–peroxidase procedure was used to immunostain formalin-fixed, paraffin-embedded PDX cells. After antigen retrieval with a citrate buffer solution (pH 6.0), the slides were stained with MIB-1 antibody (1:100; Calbiochem-Novabiochem, NA21-2) or anti-DNMT1 (1:50; Santa Cruz Biotechnology, sc-271729).
To visualize calcium deposits, von Kossa staining was performed. Tissue sections were treated with a 2% silver nitrate solution (Sigma-Aldrich) according to the manufacturer's protocol and counterstained with Nuclear Fast Red (Sigma-Aldrich).
Statistical analyses
The data are presented as the mean ± SE for in vitro and in vivo studies. Correlations between two variables were obtained by Pearson test, and IC50 (concentration required to inhibit cell proliferation by 50%) values were calculated from the linear transformation of dose–response curves. To define drug–drug interactions [synergism: combination index (CI) < 0.9; additivity: 0.90 < CI < 1.10; antagonism: CI > 1.10; 35], CalcuSyn software (Biosoft) was used. Differences among means were analyzed by Student t tests or χ2 tests.
Ethics statement
Animal experiments were performed following the European directive 2010/63/UE and Italian law DL 26/2014. Experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee (“Comitato per il Benessere Animale”) of the University of Bologna and the Italian Ministry of Health with letter 782/2015-PR (Responsible Researcher Prof. Patrizia Nanni). The ethics committee of the Rizzoli Institute approved the studies (0019012/2016 and 0009323/2016) and patient-informed consent forms were obtained for the establishment of PDX models.
Results
The nonnucleoside DNMTi MC3343 efficiently inhibits osteosarcoma cell proliferation and induces osteoblastic differentiation
In cell-free assays, MC3343 displayed increased potency against DNMT1 compared with RG-108 and SGI-1027, with an EC50 (effective concentration needed to inhibit 50% of the enzyme effect) of 5.7 μmol/L, which is 68-fold more potent than RG-108 and 2-fold more effective than SGI-1027. Against DNMT3a, the potency (EC50 = 1.7 μmol/L) of MC3343 was 290-fold higher than that of RG-108 but 2-fold less than that of SGI-1027 (Supplementary Table S1; ref. 27). In the representative Saos-2 osteosarcoma cells, MC3343 induced dose- and time-dependent decreases in DNMT1, DNMT3a, and DNMT3b protein levels (Fig. 1A and B) as well as enzyme activity (Fig. 1C). The effects were similar to those observed after treatments with the well-known nucleoside analog demethylating agent 5azadC (Fig. 1). Ten human osteosarcoma cell lines, 8 derived directly from patients and 2 from PDX-OS mouse models, were exposed to MC3343, and the results were compared with those of 5azadC (Table 1). The IC50 values ranged from 5 to 15 μmol/L for MC3343 and from 0.5 to >30 μmol/L for 5azadC. Sensitivity to MC3343 inversely correlated with the expression of DNMT1 and DNMT3a (r = −0.640 or −0.637, respectively, P < 0.05, Pearson correlation test), and this inverse relationship was not observed for 5azadC. To sustain the specific effects of DNMT inhibition on osteosarcoma cell growth, we also silenced DNMT1, DNMT3a, or both and verified the impact on Saos-2 growth under standard conditions (Supplementary Fig. S1A and S1B).
MC3343 efficiently inhibits the activity and expression of DNMTs in osteosarcoma cells. A and B, Western blot analysis (A) and immunofluorescence staining (B) of DNMT1, DNMT3a, and DNMT3b in untreated (CTRL), MC3343-treated (3–10 μmol/L), and 5azadC-treated (1–10 μmol/L) Saos-2 cells. Equal loading was monitored by anti-ACTIN blotting. Digital images were acquired under identical conditions at the same time using image analysis software (NIS Elements, Nikon). Scale bars, 20 μm. C, Effects on DNMT activity measured by the EpiQuik DNA Methyltransferase Activity assay (*, P < 0.05; **, P < 0.01, Student t test).
In vitro efficacy of MC3343 and 5azadC in relation to the basal level of DNMTs expression in osteosarcoma cell lines
Analysis of the cell cycle and apoptosis indicated that both MC3343 and 5azadC slowed cell proliferation (Ki-67; Fig. 2A) and, accordingly, induced cyclin D1 downregulation and decreased the phosphorylation of the retinoblastoma gene product (pRB1; Fig. 2B). Cell-cycle analysis indicated that MC3343 increased the percentage of cells in G0–G1 or G2–M phases of the cell cycle (Fig. 2C), confirming its antiproliferative effect. In contrast, no induction of apoptosis was observed for either MC3343 or 5azadC. Along with the loss of expression of many genes encoding cell-cycle regulators, epigenetic modifications also have pronounced effects on regulating cell fate and differentiation. To investigate the effects of MC3343 on osteosarcoma differentiation, we took advantage of the osteoblastic-like Saos-2 osteosarcoma cell line, which can differentiate toward an osteogenic lineage when maintained in low-serum medium supplemented with ascorbic acid (50 μg/mL) and β-glycerophosphate (5 mmol/L; ref. 36). Osteoblasts produce and secrete proteins that constitute bone matrix, such as collagens and osteocalcin (OCN), and are essential for osteoid extracellular matrix (ECM) mineralization, a process mediated by alkaline phosphatase (ALPL). Accordingly, we evaluated the effects of MC3343 on the expression of osteoblast differentiation genes, such as collagen type I alpha 2 chain (COL1A2), ALPL and OCN, by qRT-PCR; these data were analyzed together with the amount of mineralized ECM evaluated by ARS staining. In total, 1 μmol/L of MC3343 (corresponding to the IC30 values) inhibited cell proliferation and significantly increased ECM mineralization (Fig. 3A and B). Moreover, we demonstrated decreased expression of COL1A2, an early marker of osteoblast differentiation, and upregulation of ALPL and OCN, markers of intermediate and terminal differentiation, respectively (Fig. 3C and D). Notably, 5azadC had a similar effect on cell proliferation but was unable to drive osteosarcoma cells toward a specific differentiation profile. After exposure to 5azadC, Saos-2 cells did not mineralize in vitro (Fig. 3B), and ALPL and other osteoblastic markers were unmodified (Fig. 3C and D). To further confirm that the effects of MC3343 are indeed due to specific inhibition of DNMT, we silenced DNMTs using siRNAs and verified that DNMT deprivation under differentiating conditions inhibited cell proliferation (Supplementary Fig. S1C) and induced osteoblastic differentiation, as shown by increased ALPL activity and matrix mineralization (Supplementary Fig S1D and S1E). The different effects of MC3343 and 5azadC on differentiation were confirmed in a cell line derived from a PDX-OS (PDX-OS#1-C4) (Fig. 3C and D) and were even clearer with the Human Osteogenesis RT² Profiler PCR Array, which allows for a quick and reliable gene expression analysis of 84 genes related to osteogenic differentiation. Under differentiating conditions on day 7, exposing Saos-2 cells to MC3343 induced the expression (>1.4-fold) of several genes involved in ossification. In particular, activation of several osteoblastic differentiating mediators was evident, including augmented bone morphogenetic proteins (BMPs; increased BMP1, BMP2, and BMP5) and receptors (BMPR1A, 1B); upregulated growth factors involved in the early induction of osteoblastogenesis, such as basic FGF2 (FGF2), receptor activator of NF-κB ligand (TNFSF11), and insulin-like growth factor 1 (IGF1); increased transcriptional factors (SOX9, TWIST1); and bone γ-carboxyglutamate protein (BGLAP, also known as OCN), which encodes a highly abundant bone protein secreted by osteoblasts that regulates bone remodeling (Supplementary Table S2). In addition, MC3343 repressed chordin (CHRD), which encodes a secreted BMP antagonist protein, resulting in enhanced bone regeneration. Gene repression was also observed for matrix metallopeptidase 10 (MMP10), growth differentiation factor 10 (GDF10), and VEGFR (FLT1), which are all involved in tumor malignancy. Globally, 30 genes were found to be upregulated and 8 genes were repressed after MC3343 exposure (Supplementary Table S2). In contrast, 5azadC induced generalized but rather undefined reexpression of the genes (76/84, 90% induced; 0/84, 0% repressed). Differences between the two agents were more evident on day 14, where MC3343 repressed 19 of 84 genes and induced only 8 of 84 genes, while 5azadC continued to induce generalized re-expression of the genes (65/84, 77%). Specifically, repression induced by MC3343 involved genes expressed in chondrocytes or osteoclasts, such as those encoding cartilage oligomeric matrix protein (COMP), type I collagens, and calcitonin receptor (CALCR), indicating that the compound specifically drives osteosarcoma cells toward the osteoblast fate. In contrast, all these genes were upregulated by 5azadC, supporting its undefined action (Supplementary Table S2).
MC3343 inhibits osteosarcoma cell proliferation. A, Percentage of Ki-67–positive cells among untreated (white bars), MC3343 (3–10 μmol/L; light and dark gray, respectively), or 5azadC-treated (1–10 μmol/L for Saos-2 and IOR/OS9 cells or 0.1 to 1 μmol/L for MG-63 cells; light and dark gray, respectively) osteosarcoma cells (*, P < 0,05; **, P < 0.01; ***, P < 0.001, Student t test). B, Western blot analysis of DNMT1, cyclin D1, pRB (S795, S807/811), and total RB1 after 48 hours of exposure to MC3343 or 5azadC. Equal loading was monitored by anti-ACTIN blotting. Blots are representative of at least two independent experiments. C, Cell-cycle analysis of three representative osteosarcoma cell lines after 48 hours of exposure to MC3343 as determined by flow cytometry.
MC3343 drives osteosarcoma cells toward an osteoblast-differentiated phenotype. A, Evaluation of Saos-2 cell growth after exposure to MC3343 or 5azadC in differentiating medium by vital counting with Trypan blue. Data are presented as the mean ± SE of three separate experiments (***, P < 0.001, Student t test). B, Mineralized matrix formation by Saos-2 cells after exposure to MC3343 or 5azadC in differentiation medium as measured by ARS quantification. Data are normalized to cell number and are presented as the mean ± SE of three separate experiments (***, P < 0.001, Student t test). C, qRT-PCR expression of representative osteoblast differentiation markers in Saos-2 or PDX-OS#1-C4 cells after treatment with MC3343 or 5azadC in differentiating conditions. PDX-OS#1-C4 cell line is derived from the fourth in vivo passage of PDX-OS#1. COL1A2, ALPL, and OCN are displayed to indicate the early, intermediate, and late stages of the osteoblastic differentiation process, respectively. Expression levels of target genes (RQ) were normalized to GAPDH expression level, and the results are expressed as: 2−ΔΔCt, where ΔCt = Ct target gene – Ct GAPDH and ΔΔCt = ΔCt sample − ΔCt calibrator (untreated cells at each time point). All samples were run in triplicate for each experiment. Data are presented as the mean ± SE of three separate experiments (*, P < 0.05; **, P < 0.01; ***, P < 0.001, Student t test). D, Evaluation of ALPL activity by histochemical analysis after 14 days of exposure to MC3343 or 5azadC in differentiating conditions. Representative images are shown (NIS Elements, Nikon). Scale bars, 200 μm.
MC3343 reduces PDX osteosarcoma growth and induces differentiation
Given that PDXs preserve the tumor genomic profiles of patients and cancer biology better than patient-derived cell lines (37), the therapeutic potential of MC3343 was investigated against PDX-OS#1 xenografts. Mice were treated with 20 mg/kg MC3343 or 0.8 mg/kg 5azadC when a tumor volume of 0.01 cm3 was reached. Peritumoral MC3343 administration was chosen as the first attempt to achieve a higher local concentration and test the direct activity of only the drug and not its metabolites. MC3343 was well tolerated, with no significant changes in mouse weights (Supplementary Fig. S2A), and significantly reduced tumor growth, which reproduced the effects observed with 5azadC (Fig. 4A, P < 0.05 Student t test). The significant inhibition of cell proliferation was confirmed by Ki-67 evaluation (Fig. 4B). Consistent with the in vitro data, only MC3343, and not 5azadC, was able to drive osteoblast differentiation, despite their similar effects on cell growth. Osteoblast differentiation was proven by von Kossa staining for calcium deposits and increased OCN positivity after MC3343 treatment (Fig. 4C). Mineralized matrix induction by MC3343 may have limited the shrinkage of the tumors, which resulted in their having dimensions similar to those treated with 5azadC, masking the additional value of the compound.
MC3343 inhibits tumor growth and induces osteoblastic differentiation in osteosarcoma PDXs. A, Inhibition of PDX-OS#1 tumor growth in vivo. Drugs were administered as follows: 20 mg/kg MC3343 (n = 7) or 0.8 mg/kg 5azadC (n = 4) daily 5 times/week for 14 days. Student t test: *, P < 0.05; **, P < 0.01; compared with the controls (vehicle n = 8). Data, mean tumor volume ± SE. B, Percentage of Ki-67–positive cells in untreated and treated tumors (**, P < 0.01; ***, P < 0.001, Student t test). C, Representative IHC images of DNMT1, Ki-67, OCN, and von Kossa staining in untreated and treated tumors are shown. Scale bars, 50 μm.
MC3343 synergizes with doxorubicin and CDDP in osteosarcoma cells
In vitro, MC3343 was found to synergize with doxorubicin and CDDP (CI, 0.580 ± 0.02 for doxorubicin; CI, 0.784 ± 0.01 for CDDP) but not methotrexate (CI, 1.317 ± 0.08) in PDX-OS#1-C4, 5azadC induced subadditive effects when combined with doxorubicin (CI: 1.500 ± 0.25), while silencing of DNMT1 or DNMT3a potentiated the cytotoxic effects of doxorubicin (IC50 values: 18.05 ± 4.27 ng/mL when cells were transfected with SCR vs. 2.40 ± 0.42 ng/mL or 1.79 ± 0.58 ng/mL when cells were silenced for DNMT1 or DNMT3a, respectively) again mirroring the effects of MC3343 (Supplementary Fig. S1G). The advantage of using combined treatments was confirmed in vivo, when MC3343 was administered in combination with doxorubicin (Fig. 5A). Tumor growth was significantly inhibited when both drugs were administered together, with improved efficacy being observed compared with the single agents. Reduced body weights were observed in mice receiving the combined treatment (Supplementary Fig. S2B).
MC3343 potentiates osteosarcoma cell sensitivity to cytotoxic agents acting by targeting DNA. A, In vivo inhibition of PDX-OS#1 tumor growth by the MC3343 and doxorubicin combination. Drugs were administered as follows: doxorubicin (50 μg/injection for 2 days, n = 5), MC3343 (20 mg/kg for 14 days, n = 5), doxorubicin + MC3343 (n = 8). Student t test: *, P < 0.05; **, P < 0.01; ***, P < 0.001 compared with untreated mice (n = 9). B, Mean fluorescence intensity (MFI) of doxorubicin within nuclei of PDX-OS#1-C4 cells treated with doxorubicin (10 μmol/L) alone or in combination with MC3343 (10 μmol/L) or 5azadC (1 μmol/L) quantified by NIS-Elements AR 3.10 (Nikon) software. Plots are presented as the mean ± SE after 1 hour of drug exposure; ***, P < 0.001, Student t test. Representative images are shown. Scale bars, 20 μm. C, Percentage of sub-G0 PDX-OS#1-C4 cells upon treatment with doxorubicin (1 μmol/L) for 3 hours and subsequent washout with IMDM (CTRL), MC3343 (10 μmol/L), or 5azadC (1 μmol/L) for 24 hours. D, γ-H2AX immunofluorescence staining in PDX-OS#1-C4 cells after 1 hour of exposure to 1 μmol/L doxorubicin and washout with complete medium with or without MC3343 (10 μmol/L) or 5azadC (1 μmol/L) for 6 to 48 hours. The percentage of positive cells was obtained by flow cytometry analysis (*, P < 0.05, Student t test). Representative images are shown. Scale bars, 20 μm.
As recently observed (38, 39), epigenetic drugs are capable of increasing chromatin accessibility to conventional chemotherapeutic agents, significantly enhancing their effects. Taking advantage of doxorubicin autofluorescence, we measured the fluorescence signals in cell nuclei with or without MC3343 or 5azadC or siRNAs against DNMT1 or DNMT3a. Given that the fluorescence signal of doxorubicin is extinguished once the drug forms a stable bond with DNA (40), stabilized nuclear fluorescence associated with a significant decrease in the signal at 580 nm, as determined by flow cytometry (32), was observed when cells were simultaneously exposed to doxorubicin and MC3343 compared with doxorubicin alone but not when cells were exposed to doxorubicin and 5azadC (Fig. 5B). Similar results were obtained when cells were simultaneously exposed to doxorubicin and siRNA against DNMT1 or DNMT3a compared with SCR (Supplementary Fig. S1F). Accordingly, the percentage of sub-G0 cells was increased when doxorubicin was given in association with MC3343 but not when the drug was administered with 5azadC (Fig. 5C). To further demonstrate that MC3343, but not 5azadC, sustains doxorubicin activity, we analyzed doxorubicin-induced DNA damage represented by phosphorylated γ-H2AX foci (Fig. 5D). γ-H2AX foci were clearly evident after 1 hour of doxorubicin treatment and were maintained up to 48 hours after doxorubicin washout when cells were exposed to MC3343, while the kinetics of γ-H2AX foci was not modified when cells were exposed to 5azadC (Fig. 5D). This result indicates that inhibition of DNMTs by MC3343 limited the efficacy of the DNA damage repair process and accordingly, the percentage of growth inhibition shifted from 32% to 65% when PDX-OS#1-C4 cells were exposed to 1 μmol/L doxorubicin alone or doxorubicin and MC3343 in combination for 24 hours.
Discussion
Strategies based on the use of agents promoting cell differentiation have been successfully applied to only hematologic cancers thus far (41–43). However, differentiation-based therapies would be highly desirable for solid tumors, especially for pediatric cancers or tumors resistant to traditional treatments (44). Several epigenetic regulators have emerged as valuable drivers of malignancy reversion and cell differentiation, and consequently, their pharmacologic inhibitors have entered clinical evaluation (45). Among multiple mechanisms participating in the epigenetic regulation of gene expression in cancer, aberrant DNA methylation is recognized as being highly involved in the development and progression of human malignancies. Thus, DNMTs appear to be promising therapeutic targets. Here, we present the antiproliferative and prodifferentiation effects of the novel nonnucleoside DNMTi MC3343, an analogue of the quinoline-based inhibitor SGI-1027 (26), on osteosarcoma. Tested in patient-derived cell lines and xenografts, the compound severely limited cell proliferation while increasing the expression of genes specifically related to osteoblastogenesis and inducing matrix mineralization. Compared with 5azadC, MC3343 had similar antiproliferative effects but was more potent in promoting differentiation. Analysis of genes involved in osteoblastogenesis after treatment indeed demonstrated the gene specificity of MC3343, which appeared to reactivate the cascade of regulatory genes associated with osteoblastic differentiation that is aberrantly silenced in osteosarcoma. This specific effect was not observed for 5azadC, which induced only general, rather unspecific reexpression of genes. The specific induction of osteoblastic differentiation is indeed very relevant for osteosarcoma, which should be considered a differentiation disease (46). Transformation is thought to occur in MSCs and/or osteoprogenitors in any phase of differentiation (9, 47) and blocks the highly structured process that leads to the normal terminal differentiation of osteoblasts (46). Osteosarcoma molecular features suggest that both genetic and epigenetic disruptions of osteoblast differentiation pathways may occur during tumor development. In particular, osteosarcoma cells showed more hypermethylated genes than hypomethylated genes (31, 48), thus supporting the use of a DNMTi in this disease. Here, we demonstrated that differentiation was stimulated in vitro and in vivo by MC3343, which restored the balance between proliferation and differentiation, promoted osteosarcoma cells toward the mature osteoblast phenotype, and limited tumor growth. Although conventional osteosarcomas are high grade, poorly differentiated lesions, well-differentiated, low-grade osteosarcomas exist and display more favorable clinical outcomes (49). As in leukemia, prodifferentiating agents may thus be exploited to transform severe cancers into more amenable afflictions (17). In addition to inhibiting the growth of osteosarcoma PDXs, DNMT blockade with MC3343 also increased the chemosensitivity of osteosarcoma cells to doxorubicin and CDDP, two major drugs used in the treatment of patients with osteosarcoma that require access to DNA for activity, but not to methotrexate, a dihydrofolate reductase inhibitor that kills tumor cells by affecting folate homeostasis. The lack of synergism between MC3343 and methotrexate further supported the unique mechanism of action of MC3343. Pharmacologic manipulation of DNMTs may potentiate the efficacy of cytotoxic agents that act by targeting DNA either by inducing chromatin decondensation and increasing drug associations to DNA (38, 39) or by reverting the epigenetic silencing of genes involved in cell-cycle regulation, DNA repair, and apoptosis (50), which are known to largely affect tumor cell sensitivity to agents causing DNA damage. We showed that MC3343, as well as siRNA against DNMTs, was capable of increasing stable doxorubicin bonds with DNA, and the combination of MC3343 and doxorubicin consequently enhanced doxorubicin-induced DNA damage and cell death and had increased efficacy against tumor growth. Generally, it is certainly true that differentiated cells are usually less sensitive to cytotoxic therapies due to their limited proliferative ability. However, MC3343, while acting as a purely cytostatic agent when used alone, increased the apoptotic activity of agents affecting DNA. Notably, none of these effects have been observed with 5azadC, further supporting the use of the novel DNMT inhibitor MC3343 in the treatment of osteosarcoma.
Overall, these results demonstrate at a preclinical level that the novel DNMT inhibitor MC3343 is a very promising option for osteosarcoma differentiation therapy and may be ideally considered an adjuvant agent in the treatment of patients with a poor response to preadjuvant chemotherapy, evaluated as a percentage of necrosis at the time of surgery (51, 52), and with unfavorable prognoses. For effective translation to preclinical studies, it will be important to fine-tune the drug dosing and timing strategies to improve outcomes and reduce toxicity. The demonstration that MC3343 treatment increases the efficacy of doxorubicin and CDDP in killing osteosarcoma cells offers the possibility of reducing the toxicity and side effects of these agents, valuable extra benefits for teenage and young adult patients with osteosarcoma. Nonetheless, the identification of key features able to predict which patients are likely to respond or become resistant to these drugs is highly desirable. Although our preliminary data on a panel of 10 lines indicated a direct relationship between MC3343 sensitivity and DNMT1/DNMT3a expression, this relationship requires validation. In addition, considering that differentiated cells may be less sensitive to cytotoxic drugs, a careful evaluation of drug schedule and treatment timing is necessary before any possible clinical application.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: M.C. Manara, S. Valente, A. Mai, K. Scotlandi
Development of methodology: M.C. Manara, S. Valente, C. Cristalli, C. Zwergel, R. Mazzone, G. Stazi, P.B. Arimondo, M. Pasello, C. Guerzoni, A. Mai
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Nicoletti, L. Landuzzi, P. Picci, P. Nanni, P.-L. Lollini, K. Scotlandi
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.C. Manara, S. Valente, C. Cristalli, G. Nicoletti, L. Landuzzi, M. Pasello, C. Guerzoni, P. Nanni, P.-L. Lollini, A. Mai, K. Scotlandi
Writing, review, and/or revision of the manuscript: M.C. Manara, S. Valente, C. Cristalli, G. Nicoletti, L. Landuzzi, C. Zwergel, R. Mazzone, G. Stazi, P.B. Arimondo, M. Pasello, P. Nanni, P.-L. Lollini, A. Mai, K. Scotlandi
Study supervision: M.C. Manara, A. Mai, K. Scotlandi
Acknowledgments
This work was funded by AIRC (IG 18451 to K. Scotlandi; IG 19162 to A. Mai), the Italian Ministry of Health (TRANSCAN_ project TORPEDO_ER-2015-23604059 to K. Scotlandi), Progetto Ateneo Sapienza (to A. Mai), PRIN 2016 (prot 20152TE5PK to A. Mai), and the NIH (R01GM114306 to A. Mai). The authors thank Cristina Ghinelli for editing the manuscript.
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 Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
- Received August 25, 2017.
- Revision received December 4, 2017.
- Accepted June 21, 2018.
- Published first June 29, 2018.
- ©2018 American Association for Cancer Research.