HDAC2 depletion promotes osteosarcoma’s stemness both in vitro and in vivo: a study on a putative new target for CSCs directed therapy

Saos2 and MG63 osteosarcoma cell lines were purchased from ATCC CELL BANK; cells were placed in DMEM culture medium, supplemented with 10% FBS, 100 mM 2P-ascorbic acid, 2 mM L-glutamine, 100 U/ml penicillin, 100 mg/ml streptomycin (all purchased from Invitrogen, San Giuliano Milanese, Milan, Italy) and placed in 25 ml flasks with filtered valves. Treatments with 5-aza-2′-deoxycytidine (DAC) (Sigma Aldrich, Milan, Italy) and valproic acid (VPA) (Sigma Aldrich), both diluted in H₂O, were administered at 3 μM and 0,5 mM, respectively, and in combination replenishing them every 24 h for 48 h.

MTT assay (3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide) was used to measure the inhibition of cell proliferation and to calculate the IC50. MTT was added in cells exposed to either DAC or VPA at the concentrations of 1,3,5, 10 and 100 μM at 48, 72 and 96 h for DAC and 0,5 and 1 mM at 48, 72 and 96 h for VPA. After identifying non-cytotoxic concentrations, we performed an MTT assay to evaluate the combined effect of both drugs at 48,72 and 96 h.

Four hours later, the formazan precipitate was dissolved in 100 μL dimethyl sulfoxide, and then the absorbance was measured in an ELISA reader (Thermo Molecular Devices Co., Union City, USA) at 550 nm.

Treated and untreated cells were plated at a density of 10 × 10⁴ cells/well in 6-well plates. Cells were harvested and re-suspended in PBS at 48, 72 and 96 h. An aliquot of cell suspension was diluted with 0.4% trypan blue (Sigma Aldrich), pipetted onto a haemocytometer and counted under a microscope at 200x magnification. Live cells excluded the dye, whereas dead cells admitted the dye and consequently stained intensely with trypan blue. The number of viable cells for each experimental condition was counted and represented on a linear graph.

Treated and untreated MG63 and Saos2 cells were plated at a density of 6 × 10⁴ cells/well in 6-well ultra-low-attachment plates (Corning, Corning, NY, USA) in DMEM/F12 cell medium supplemented with progesterone (10 nM), putrescine (50 nM), sodium selenite (15 nM), transferrin (13 ng/ml), insulin (10 ng/ml; Sigma, St. Louis, MO, USA), human EGF (10 ng/ml), and human bFGF (10 ng/ml; Sigma Aldrich). Fresh aliquots of EGF and bFGF were added every day. After 48–72 h from culture, sarcospheres were visible with an inverted phase-contrast microscope (TS 100; Nikon, Milan, Italy). The sarcosphere-forming efficiency (SSFE) was calculated as the number of sphere-like structures formed in 2 days divided by the original number of cells seeded and expressed as a percentage (mean ± SD).

Treated and untreated cells were detached by trypsin, counted and washed in 0.1% BSA in PBS. At least 500,000 cells were incubated with fluorescent-labeled monoclonal antibodies or respective isotype controls (1/10 diluted 4 °C for 30 min in the dark). After washing steps, the labeled cells were analyzed by flow cytometry using a BD FACS Aria III (Becton & Dickinson, Mountain View, CA, USA). The antibodies used were mouse anti-human CD133/2 PE (Miltenyi Biotec S.r.l. Calderara di Reno, Bologna, Italy), mouse anti-human CD29 PerCP-Cy5.5 (BD Pharmingen, Buccinasco, Milan, Italy), mouse anti-human CD44 FITC (BD Pharmingen, Buccinasco, Milan, Italy), and mouse anti-human CD324 PE (BD Pharmingen).

Treated and untreated cells were detached by trypsin, counted and washed in 0.1% BSA in PBS. At least 500,000 cells were stained using Caltag Fix & Perm Kit (Invitrogen, Milan, Italy) following the manufacturer’s guidelines with some modifications. Briefly, the cells were fixed with medium A for 15 min at room temperature, washed with PBS and permeabilized with 0,5% triton X 100 for one hour at room temperature. After washing steps, the cells were incubated with medium B plus antibodies or respective isotype controls for 1 h at room temperature. The primary antibodies used were: H3trimethyl k9, H3acetyl k9, H3trimethyl k27, H3trimethyl k4, and H3dimethyl k4, and HDAC2 antibodies, all purchased by Abcam. Total methylated DNA was detected by immunolabeling with anti-5-methylcytidine antibody (Abcam). This antibody recognizes 5-methylcytidine, a modified base found in the DNA of plants and vertebrates. It is specific to the methylated base and has minimal reactivity to non-methylated cytidine or cytosine. For staining, cells were fixed with cold methanol for 10 min. After PBS washing, cells were treated with 2 N HCL for 30 min at 37 °C, followed by an incubation of 5 min with 0,1 M borate buffer at pH 8,5. After washing steps, the cells were incubated with medium B plus antibody or respective isotype controls for 1 h at room temperature.

Detection was performed by incubation with a FITC-conjugated, affinity purified F(ab0)2 fragment of goat anti-mouse or anti-rabbit IgG (Abcam, Cambridge, UK) for 1 h at room temperature in the dark. The labeled cells were analyzed by flow cytometry using a BD FACS Aria III (Becton & Dickinson, Mountain View, CA, USA). Histone modification and global DNA methylation associated to the cells was measured using FITC fluorochrome and calculated as mean fluorescence intensity (MFI) for each sample. The analyses are performed considering 100% untreated samples. 1 × 10⁴ events were acquired and histograms showed the mean ± SEM of the MFI values obtained from three independent experiments.

For Western blot analyses, cells were lysed in RIPA buffer (50 mM Tris-HCl pH 7.2, 150 mM NaCl, 1% NP40, 0.1% SDS, 0.5% DOC, 1 mM PMSF, 25 mM MgCl2, and supplemented with a phosphatase inhibitor cocktail). Protein concentration was determined by the BCA assay (Bio-Rad Laboratories, Hercules, CA). Equivalent amounts of protein (50 μg) were electrophoresed on 10% SDS–polyacrylamide gels. Precision Plus Protein™ Dual Color Standards (Bio-Rad) was used to determine molecular weight. Gel was electroblotted onto nitrocellulose membrane by using a Trans Blot Turbo system (Bio-Rad) following the manufacturer’s instructions. The membrane was blocked with 5% milk in TBS-0.1% Tween (TTBS) for 1 h at RT and washed with TTBS. Membrane was then incubated with specific primary anti-human antibodies against HDAC1 (1:5000; Abcam), HDAC2 (1:1000; Abcam), DNMT1 (1:250; Abcam), DNMT3a (1:500; Abcam) or GAPDH (1:5000; Millipore) over night at 4 °C. Total GAPDH antibody was used for assessing loading. Membrane was then washed with TTBS and incubated with the appropriate HRP-conjugated secondary antibody diluted 1:5000 in 3% milk in TTBS for 1 h at RT. Membrane was then washed three times with TTBS. Immunoreactive protein bands were visualized by the Pierce™ ECL Western Blotting Substrate (Thermo Scientific) according to the manufacturer’s instructions. The densitometric analyses were performed using Image J software and were expressed as ratio of GAPDH densitometry versus tested protein densitometry.

Treated and untreated cells were grown on coverslips and fixed in methanol. Cells were permeabilized with 1% TRITON X 100 in PBS and blocked using 1% BSA. Cells were stained with monoclonal anti-AcH3K9 (Abcam) and corresponding IgG negative controls. Cells were subsequently stained with FITC rabbit anti mouse secondary antibodies, counterstained with DAPI. Visualization was performed using an EVOS microscope. Fluorescence intensities from images of six randomly selected microscopic fields of cells were semi-quantitatively analyzed by densitometry using Image J software NIH Image.

The wound-healing assay was performed to measure two-dimensional movement. All cell fractions (untreated or treated) were cultured in 24-well plates at a density of 50.000 cells/well until confluence. A wound was created in the center of the cell monolayers by a sterile pipette tip. The phase contrast images were captured after 0, 4, 24 and 48 h. The analyses are performed considering 100% wound size at the time of culture and analyzed by Image J software. All experiments were performed in triplicates.

To measure in vitro tumorigenicity due to drug treatments, treated and untreated cells at a density of 500 cells per well in 24-well plates were plated in soft agar, in triplicates. The test was performed using 0.8 and 0.3% agar in DMEM as the base and top layers, respectively. Cells were incubated for 21 days at 37 °C in a humidified atmosphere at 5% CO₂ in air, and 50 ml of DMEM culture medium was added twice a week. At the end of the incubation period, colonies were stained with nitroblue tetrazolium (NBT, Sigma) at a concentration of 1 mg/2 ml in PBS and counted using an inverted microscope (Nikon TS 100, Milan, Italy). The colony efficiency was calculated as proportion of colonies per total number of seeded cells. The data were analyzed by Image Pro Plus software. All experiments were performed in triplicates.

Total RNA was extracted using TRIzol Reagent (Invitrogen, Milan, Italy) according to the manufacturer’s protocol. RNA concentration and purity were determined by A260 and A260/A280 ratios, respectively. The integrity of total RNA was assessed on standard 1% agarose/formaldehyde gels. The RNA samples were treated with DNase I to remove residual traces of DNA. Expression levels of genes involved in angiogenesis were analyzed using real-time quantitative PCR. All real-time PCR reactions were performed using StepOne Thermocycler (Applied Biosystems, Monza, Italy) and the amplifications were done using the SYBR Green PCR Master Mix (Applied Biosystems, Monza, Italy). The thermal cycling conditions were: 50 °C for 2 min followed by an initial denaturation step at 95 °C for 2 min, 40 cycles at 95 °C for 30s, 60 °C for 30s and 72 °C for 30s. Real time q-PCR results were normalized to those of the housekeeping gene glyceraldehyde-3-phospate dehydrogenase (GAPDH). The primer sequences were the following (Additional file 1):

Mock MG63 and HDAC2 depleted-MG63 cells were injected subcutaneously into each flank of locally bred NOD/SCID IL2R-gamma-0 (NSG) mice [25, 26].

For this purpose, cells were enzymatically dissociated, diluted in PBS, mixed with Matrigel, and injected subcutaneously in mice. Mice were monitored every 5 days for the appearance of subcutaneous tumors. After 30 days, mice were sacrificed, and the tumor volume was calculated by the formula (l x w²)/2. The injection experiments were made in triplicate. All mouse experiments were performed according to the Institutional Animal Care and Use Committee procedures and guidelines of University of Campania.

We investigate the possible epigenetic mechanism occurring in osteosarcoma by treating Saos2 and MG63 cells with VPA and DAC, alone or in combination of both. Therefore, we determined non-cytotoxic concentrations of both drugs using MTT and growth curve evaluation. MTT analyses performed 48 h post-treatment showed no significant differences in cell growth, while after 72 h (and using different concentrations of both drugs) there was high cytotoxicity. Furthermore, at 100 μM of DAC, cells were dead independently by treatment length. The use of VPA concentration of 0.5 mM up to 48 h did not induce differences in cell growth, whereas 1 mM VPA reduced cell growth in a time-dependent manner (Additional file 2). Cell growth curves analyses were also performed, using the above concentrations, without showing any statistically significant differences in cell proliferation at 48 h confirming data of MTT analyses (Additional file 2). Moreover, the calculation of IC50 showed that, both for Saos2 and MG63 cells, IC50 values were 10, 5 and 3 μM for 48,72 and 96 h, respectively for DAC treatment, and 0,7, 0,6 and 0,1 mM for 48,72 and 96 h respectively for VPA treatment. Being our aim that to evaluate the epigenetic alterations occurring in osteosarcoma, we selected the concentrations of 3 μM of DAC and 0,5 mM of VPA and their combination for 48 h for all further experiments.

To investigate the effect of DAC and VPA on osteosarcoma stem cells, we analyzed different markers of stemness by RT-PCR and flow cytometry. We performed a gene expression analysis for OCT4, SOX2, NANOG and CD133. Treatment of Saos2 cells with DAC induced an increase of OCT4, SOX2 and CD133mRNA levels, but not of NANOG mRNA level. On the contrary, treatment with VPA induced an increase of NANOG, OCT4 and CD133 mRNA levels but not of SOX2 mRNA levels. The combination of both drugs induced a strong increase of OCT4 and CD133 mRNA levels. VPA and DAC treatment on MG63 cells, another osteosarcoma cell line, induced an increase of all stemness genes when compared to those of untreated cells. Interestingly, drug combination led to a strong increase of NANOG mRNA levels (Fig. 1a). Flow cytometry analyses demonstrated that VPA and DAC induced an increase of SOX2, OCT4 and NANOG proteins, in both cell lines (Fig. 1b and Additional file 3). Remarkably, CD133 expression was increased after treatments both in Saos2 and MG63 cell lines. In particular, both VPA and DAC induced a considerable increase of CD133 expression and particularly in MG63 cells (Fig. 1c). The combination of the two drugs resulted in almost 3-fold increase of CD133 expression, when compared to untreated cells. Treatment with VPA or DAC induced a two-fold increase of CD133 expression and when compared to untreated cells (Additional file 4). In conclusion, DAC and VPA induced an increase of stemness as reflected by increased mRNA and protein levels of CD133, OCT4, Sox2 and NANOG.

VPA and DAC have been previously shown to induce epithelial to mesenchymal transition (EMT) [27, 28]. It has been shown that those transcription factors that induce EMT in epithelial cells are involved, for example, in drug resistance and other aggressiveness features of osteosarcoma [29]. Therefore, we determined the expression of SLUG, TWIST, vimentin, and e-cadherin, osteocalcin, CD29 and CD44 by real-time PCR and flow cytometry (Additional file 5).

VPA, DAC and their combination did not affect mRNA expression levels of vimentin, whereas VPA alone induced a strong increase of TWIST and e-cadherin mRNA levels in both cell lines when compared to untreated cells. DAC treatment led to a slightly increase of TWIST and a strong up-regulation of e-cadherin mRNA levels.

Treatment combining VPA and DAC induced only a strong increase of e-cadherin. Regarding to osteocalcin, treatments induced a decrease of mRNA levels compared to those of untreated cells (Additional file 5).

These results were partially confirmed by flow cytometry. Although it was possible to observe an increase of vimentin expression, this was not statistically significant in MG63 cells, whereas no change in its expression was detectable for Saos2 cells (Additional file 5) which confirms our observations on mRNA levels. Moreover, flow cytometry analyses showed that e-cadherin was undetectable in MG63 and Saos2 cells (Additional file 5). Osteocalcin, CD29 and CD44 expressions were not affected by treatments (almost 90% of expression). In summary and apart from inducing a slight increase of TWIST, DAC and VPA did not affect EMT. Similarly, no effect was observed for vimentin, osteocalcin, CD29 and CD44 expression following treatments.

Tumor stemness is also characterized by the ability to form spheres and colonies in soft agar. Therefore, we calculated the efficiency by which VPA and DAC induced sarcospheres and colonies formation. Both drugs led to an increase of sarcospheres formation rate, and when in combination, the rate was 3-fold increased (Fig. 2a). One method of analyzing the tumorigenic potential is the soft agar assay, which measures anchorage-independent growth. Assessment of growth kinetics revealed increased efficiency in colony-forming ability of treated cells and when compared to the untreated ones (Additional file 6). In particular, the combination induced a strong effect with a mean 3 fold-increase in colony-forming efficiency. Another characteristic that correlates with stemness is migration ability. Wound healing experiments to measure cell invasion revealed that both VPA and DAC enhanced motility of Saos2 and MG63 cells when compared to untreated controls and for 24 h. In this context, VPA induced a higher motility in MG63 compared to untreated and DAC treated cells. On the other hand, in Saos2 cells, drug combination increased cell migration, when compared to untreated and single agent VPA or DAC treated cells (Fig. 2b). Therefore, VPA, DAC and their combination enhanced the in vitro migration of investigated osteosarcoma cell lines.

To explore the landscape of histone modifications and DNA methylation that could be linked to our tumor stemness model and in response to VPA and DAC treatments, we performed an analysis of histone modifications that were previously shown to be involved in embryonic stemness, such as the repressive histone marks H3-trymethyl k9 (H3K9me3), H3-trymethyl k27 (H3K27me3), active histone marks H3-trymethyl k4 (H3K4me) and H3-dymethyl k4 (H3K4me2) [30]. Moreover, we investigated global epigenetic profiles of H3K9 acetylation and DNA methylation (Fig. 3). Flow cytometry analyses showed that VPA and DAC treatments, alone and in combination, induced a decrease of H3K9me3 and H3K27me3, and an increase of H3K4me2 and H3K4me3 (Fig. 3a). This indicates a correlation between those histone modifications and cancer stemness. These data were further confirmed by H3K9 acetylation and DNA methylation (Fig. 3). In fact, we found increased levels of H3K9 acetylation (p < 0.001) in treated cells, when compared with untreated cells (Fig. 3a). We also observed a substantial increase of acetylation for both cell lines when co-treated with VPA and DAC. These results were confirmed by immunofluorescence (Fig. 3b) semi-quantitatively analyzed by densitometry (additional file 7) where we observed increased H3K9 acetylation in treated cells as reflected by elevated fluorescence using the AcH3K9 antibody and when compared to untreated counterparts. Moreover, the levels of 5′ Methylcytidine, analyzed by flow cytometry, showed a significant reduction of global DNA methylation (p < 0.001) in treated cells and when compared with untreated cells (Fig. 3c). To better investigate this event, we performed a western blotting for DNA methyltransferases 1 and 3a (DNMT1, DNMT3a) that are the main enzymes involved in global DNA methylation [31] (Fig. 4a). In Saos2 and MG63 cell lines, DNMT1 expression did not change after drug treatment, whereas DNMT3a expression severely decreased (Fig. 4a). This is a very interesting finding, as it highlights that DNMT3a could be a key factor involved in cancer stemness in osteosarcoma carcinogenesis. VPA has been suggested to inhibit the activity of histone deacetylases 1 and 2 (HDAC1, 2) [32]. Therefore, we investigated the expression of the two enzymes after treatment in order to investigate their possible role in cancer stem cells. Western blotting analyses showed that in Saos2 and MG63 cell lines HDAC1 expression did not change after treatment, therefore HDAC1 does not appear to be involved in acetylation processes. However, HDAC2 levels decreased in both cell lines after treatment, mainly following VPA treatment and drug combination (Fig. 4b).

To confirm our results regarding the involvement of HDAC2 in promoting cancer stemness, we performed a silencing of HDAC2 using HDAC2 shRNA both in MG63 (Fig. 5a) and Saos2 cells (Additional file 8). HDAC2 depleted-MG63 and Saos2 cells showed an increase of stemness markers CD133, OCT4, SOX2 and NANOG when compared to control cells (Fig. 5b and Additional file 8). This result was comparable to that observed following VPA and DAC treatments. Moreover, we found that HDAC2 depleted-MG63 and Saos2 cells had an increased H3K9 acetylation (Fig. 5c and Additional file 8) and a decreased global DNA methylation (Fig. 5d and Additional file 8). This observation also confirms the results obtained following VPA and DAC treatments. Interestingly, an increase in sarcospheres (Fig. 5e and Additional file 8) and colonies formation rates (Fig. 5f and Additional file 8) was also detectable for HDAC2 depleted-MG63 and Saos2 cells and compared to mock cells, which further supports the idea that HDAC2 Knockdown was effective in promoting stemness of osteosarcoma cells (Fig. 5 and Additional file 8).

In addition, to investigate the in vivo effects of HDAC2 downregulation on osteosarcoma tumorigenesis, we injected knocked down HDAC2-MG63 cells in mice. We found that the HDAC2 downregulation significantly promoted tumor growth in nude mice compared to mock MG63 cells (Fig.6a) with a tumor size of 2,9 ± 0,5 fold greater than those detected in mock MG63 cells (Fig. 6b). These results indicate that HDAC2 Knockdown may promote the osteosarcoma tumorigenesis.