SOX2OT variant 7 contributes to the synergistic interaction between EGCG and Doxorubicin to kill osteosarcoma via autophagy and stemness inhibition

The SaoS2 and U2OS osteosarcoma cell lines were purchased from the Cell Culture Center, Shanghai Institutes for Biological Sciences, the Chinese Academy of Sciences (Shanghai, People’s Republic of China). The cells were maintained in DMEM supplemented with 10% FBS, 25 mM hydroxyethyl piperazine ethanesulfonic acid buffer, 100 U/mL penicillin, and 100 μg/mL streptomycin in ahumidifed atmosphere of 5% carbon dioxide at 37 °C.

In order to form osteosarcoma spheres, 5 × 10³ cells /well suspended OS cells were cultured in low-adhesive six-well plates in serum-free DMEM-F12 (1:1) medium supplemented with 20 ng/ml fibroblast growth factor (FGF; R&D systems), 20 ng/ml epidermal growth factor (EGF; R&D systems), B27 (1:50) (Invitrogen), N2 (1:100) (R&D systems), and 10 ng/ml LIF (Millipore) for 10 days (that is sphere-forming culture medium). In addition, fresh DMEM-F12 (1:1) medium, EGF, and bFGF were added every other day. A single-cell suspension derived from primary spheres was obtained for the secondary sphere formation with the same method.

10 pairs of OS tissue samples and their relative adjacent tissues were collected from the second Xiangya hospital from Dec, 2015 to March, 2016. Unfortunately, RNA degradation was found in 4 pairs of tissues which could not be used in subsequent experiments. Use of these samples for all experiments was approved by the Ethics Committee of the 2nd Xiangya Hospital of Central South University.

SOX2OT transcript variants were amplified separately using RT-PCR according to previous literature reports. U-87 MG cell line was used as positive control [13, 15, 16]. The SOX2OT V7 sequence was synthesized by BGI (Beijing, China). The synthetic V7 DNA fragment and lentiviral vector PGMLV-6395 were enzyme digested with BamH I/EcoRI,Ligation was performed to construct V7 overexpressed lentivirus recombinant plasmid. The following lentivirus package was performed by Auragene biotech (Changsha, China).

Cells were plated at a density of 5000 per well of a 96-well plate, 24 hafter plating, cells were treated as the indicated concentrations. 20 ml MTT with a concentration of 5 mg/mlwas added to each well for an additional 4 hours. The blue MTT formazan precipitate was then dissolved in 150 μL of dimethyl sulfoxide per well with incubation for 10 min in a rotary platform at 37 °C. Cell proliferation inhibition ratio was calculated according to the absorbance at a wavelength of 570 nm (A value) in each well by ELISA analyzer (MK3, Themo, USA). Cell proliferation inhibition ratio (%) = (A value of control group–A value of treated group)/A value of control group× 100%.

According to previous study [3], MTT assay was used to detect the absorbance at the wavelength of 570 nm as the OD value. The coefficient of drug interaction (CDI) was calculated as follows: CDI = AB/(A × B). According to the absorbance of each group, AB is the ratio of the combination groups to control group; A or B is the ratio of the single agent group to control group. Thus, CDI value< 1, = 1 or > 1 indicates that the drugs are synergistic, additive or antagonistic, respectively.

Total RNA was extracted from cells using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions, and then the RNA was reverse transcribed using the PrimeScript RT Master Mix Perfect Real Time kit (TaKaRa, Dalian, China) to obtain the cDNA. Using the cDNA as the template, real-time PCR assay was performed using the pairs of primers listed in Table 1. The 20 μL real-time PCR reaction included 0.5 μL of cDNA template, 0.25 μL of Forward/Reverse primers respectively, 10 μL of RNase-free dH₂O, and 8 μL of 2.5× Real Master Mix (SYBR Green I). The reaction conditions included a pre-denaturation step at 94 °C for 10 s, and 40 cycles of 94 °C for 15 s and 58 °C for 60 s. After the reaction, the data were subjected to statistical analysis.

Cells were lysed in cell lysate, and then centrifuged at 12,000×g for 20 min at 4 °C. The supernatant was collected and denatured. Proteins were separated in 10% SDS-PAGE and blotted onto polyvinylidene difluoride membrane (PVDF). The PVDF membrane was treated with TBST containing 50 g/L skimmed milk at room temperature for 4 h, followed by incubation with the primary antibodies against the LC3B (1:3000, ab51520, Abcam), Atg5 (1:1000, ab108327, Abcam), Atg7 (1:100,000, ab524721,Abcam), Beclin1 (1 μg/ml, ab62557, Abcam), P62 (1:500, 18,470–1-AP, Proteintech, China), OCT-4 (1:1000, Proteintech), ABCG2 (1:50, ab24115, Abcam), c-Myc (1:10,000, ab32072, Abcam), Sox2 (1 μg/ml, ab97959, Abcam) and anti-β-actin (1:1000, Cell signaling) respectively, at 37 °C for 1 h. Membranes were rinsed and incubated for 1 h with the correspondent peroxidase-conjugated secondary antibodies. Chemiluminescent detection was performed with the ECL kit (Pierce Chemical, Rockford, IL, USA). The amount of the protein of interest, expressed as arbitrary densitometric units, was normalized to the densitometric units of ß-actin.

CD133/CD44 immunomagnetic double screening were improved and performed according to the methods reported in previous literature [17]. A single-cell suspension of Saos-2 and U2OS cells was incubated with magnetic microbeads-conjugated with the mouse anti-human CD133 monoclonal antibody (Miltenyi Biotec, USA) for 30 min. After washing, the CD133+ cells were separated using the magnetic cell sorting system (autoMACS; Miltenyi Biotec, USA). The purified CD133+ cells were expanded for 14 days by culturing and then harvested as a single-cell suspension to be incubated with magnetic microbeads-conjugated mouse anti-human CD44 monoclonal antibody (Miltenyi Biotec, USA) for 30 min. After washing, the CD44+ cells were separated using the magnetic cell sorting system as described above. This two-step isolation enabled us to obtain sufficient number of CD133+/CD44+ CSCs for the following experiment.

To verify the purity of the isolated CD133+/CD44+ CSCs, cells were stained according to the supplied antibody protocols. Mouse anti-Human CD133/1 (Clone: AC133)-PE and mouse anti-human CD44 (Clone: DB105)-FITC (Miltenyi Biotec, USA) were used. Then flow cytometry analysis was performed using a FACSCalibur instrument (Becton Dickinson).

To evaluate the colony-forming ability of different cells, self-renewal capacity assay was performed. We seeded OS cells or CSCs into 96-well plates at a density of 1 × 10³ cells / well using sphere-forming culture medium with TGF-beta (20 ng/ml) and the formation of colonies was quantifed at 14 days after inoculation. Colonies with over 50 cells were counted under an Olympus microscope.

In this study, autophagy levels were detected using immunofluorescence staining and MDC staining beside autophagy marker detection. Cells were cultured on cover glasses coated with 0.1% gelatin in PBS in 6-well tissue culture plates with DMEM. The cells were incubated overnight then washed with PBS, and fixed in 4% paraformaldehyde for 15 min. They were then permeabilized in 0.2% Triton X-100 for 10 min prior to blocking in 6% bovine serum albumin (BSA) for 30 min. For immunofluorescence staining, the cells were incubated overnight with anti-LC3B (1:200, 12741S, Cell Signaling, USA) at 4 °C, followed by incubation with goat anti-rabbit IgG (H + L)-Cy3 (1:200, SA012, Auragene) for 1 h at room temperature in dark. Nuclei were counterstained with DAPI (10 μg/ml, C0065, Solarbio, China). Finally, cells were analyzed using a confocal fluorescence microscope (BX50; Olympus, Japan) and the percentage of cells with LC3 puncta was calculated by taking advantage of Image-Pro Plus 6.0. When autophagic vacuoles were labeled with monodansylcadaverine (MDC), cells were incubated with 0.05 mM MDC in RPMI1640 at 37 °C for 10 min. After incubation, cells were washed three times with PBS and immediately analyzed with a fluorescence microscopeequipped with a filter system (V-2A excitation filter:380/420 nm, barrier filter:450 nm). Images were captured and imported into Adobe Photoshop.

Animal experiments were performed in strict accordance with the Guide for the Care and Use of Laboratory Animals of the second Xiangya hospital of Central South University. The protocol was approved by the Committee on the Ethics of Animal Experiments of the second Xiangya hospital of Central South University. NOD/SCID mice at age of 3–5 weeks, male, were maintained in pathogen-free conditions at animal facility. The OSCs with or without SOX2OT V7 overexpression were resuspended in serum-free medium and mixed with Matrigel at the ratio of 1:1. NOD/SCID mice were randomly divided into 4 groups (n = 3 per group). 1 × 10⁵ indicated cells were inoculated subcutaneously into the inguinal folds of NOD/SCID mice. 7 days after inoculation, tumor formation was evaluated by palpation of injection sites., thenthe mice that developed palpable tumors were intraperitoneally injected EGCG (30 mg.kg-1) in combination with or without Notch-3 knockdown lentivirus (Lv-Notch3-) at the injection site every 3 days for 3 times. At the end of experiment (21 days), the mice were sacrificed under deep anesthesia with pentobarbital. The tumors were then dissected and captured.

In order to determine whether EGCG could produce synergistic effects with Dox on osteosarcoma cell growth inhibition, MTT assays were applied and the coefficient of drug interaction (CDI) was calculated. The results showed either for U2OS or SaoS2 cell lines, when Dox was used in combination with EGCG, the growth inhibition was significantly up-regulated (Fig. 1a). As expected, according to the MTT assay and the coefficient of drug interaction (CDI, shown in Table 2), EGCG and DOX yielded synergistic interactions (CDI<1). As reported previously, autophagy contributed to the synergistic antitumor effects of EGCG and DOX [3], autophagy levels were detected in various ways. Firstly, immunofluorescence staining was performed to detect the LC3 translocation in U2OS cells. The results showed that in control group, LC3 was uniformly distributed in the cytoplasm, while LC3 expression occurred predominantly in punctuate dot-like structures when treated with Dox which wasconsistent with autophagy induction. Whereas the LC3 puncta decreased obviously in U2OS cells treated with EGCG combined with DOX (Fig. 1b). Secondly, the expression of autophagy markers were detected using qRT-PCR and western blot. It was found that, the relative mRNA levels of Atg5 and Beclin-1 were decreased significantly when OS cells were treated with Dox and EGCG together but there was no obvious expression alteration of Atg7 (Fig. 1c). The decrease in LC3II/I ratio and the simultaneous increase of autophagy substrate P62 suggested that autophagy levels were attenuated to some exent in the process of combining Dox with EGCG in both OS cell lines (Fig. 1d). In order to further define the synergistic growth inhibition between EGCG and DOX on OS cells was caused by autophagy mediation, cells were pre-treaed with 3-MA or rapamycin (Rapa) respectivly in different experimental groups. The MTT results showed that when pre-treated those cells with 3-MA, a representative autophagic antagonist, the DOX-induced proliferative inhibition effects was dramatically aggravated, while when pre-treated those cells with Rapa, a representative autophagic agonist, the synergistic growth inhibition effect of concurrent administration of EGCG andDox was significantly decresed (Fig. 1e). The experimental results illustrated that EGCG-induced autophagy inhibition contributes to the synergistic interaction between EGCG and Doxorubicin to kill the osteosarcoma cells.

Using qRT-PCR, we initially examined the expression pattern of newly identified human SOX2OT variants (transcripts 1/4/6/7/8) in 6 OS tumor tissue mixture, with specific primers for each variant [13]. A glioblastoma multiform cell line U-87 was utilized as a positive control to compare the expression pattern of the variants.. As it is shown in Fig. 2a, no or very low expression was observed for SOX2OT transcripts 6 and 8. However, variant 1 showed a moderately high and variant 4 and 7 presented significantly high expression levels in OS tumor tissues. The expression pattern was different from that of U-87 which was used as a positive control.

The relative expression of the two most abundantly expressed SOX2OT variants (transcripts 1 and 7) were measured in OS tumor tissues and their responding adjacent tissues (n = 6) by qRT-PCR. The data was normalized to samples from adjacent tissues (A1) obtained from patient NO.1 (n = 6). The data revealed that SOX2OT variants 7 (V7) increased in 5/6 tumor tissues compared with their relative adjacent tissues (Fig. 2b). Besides, the relative mRNA levels of V7 were significantly in MG63, U2OS and SaoS2 cells when compared with primary osteoblast (Fig. 2c). These results indicated that V7 was overexpressed in osteosarcoma to some extent. It is noteworthy that the relative mRNA levels of V7 up-regulated obviously in OS cells treated with DOX but down-regulated in EGCG treated OS cells compared to control group. Meanwhile, the EGCG could decrease the V7 expression induced by DOX treatment significantly (Fig. 2d). These results suggested that V7 may be one of the important targets for EGCG to increase efficiency of DOX.

In order to explore the biological function of SOX2OT V7 in osteosarcoma, V7 gain-of-function OS cell models were established by V7 overexpressed lentivirus infection. qRT-PCR was performed to verify the efficiency of V7 overexpression. The results showed that in V7 gain-of-function cell models, mRNA level of SOX2OT-V7 enhanced significantly in both cell lines (Fig. 3a). At the same time, it was found that EGCG treatment could significantly reduce the endogenous expression level of V7 in both parental and V7 overexpressed OS cells. Therefore, V7 is one of the targets of EGCG in osteosarcoma cells. To investigate whether V7 is an autophagy regulator in OS cells, relative mRNA levels of autophagy associated genes Atg5, Atg7 and Beclin1 were detected using qRT-PCR. The results demonstrated that the relative mRNA levels of Atg5, Atg7 and Beclin1 were all up-regulated significantly in SOX2OT V7 overexpressed U2OS cells compared with control cells regardless of EGCG treatment (Fig. 3c). In order to explore whether the autophagy regulation of V7 is associated with the synergistic effects of EGCG and DOX, western blots, immunofluorescence staining and MDC staining were performed in SOX2OT V7 overexpressed OS cells under DOX treatment with or without EGCG. The results showed that, when V7 was overexpressed in OS cells, the ratio of LC3II/I increased obviously and the autophagy substrate P62 decreased significantly which explained that SOX2OT V7 could induce autophagy in OS cells again. On the other hand, it was interesting that the inhibitory effect of EGCG on DOX-induced autophagy became faint in V7 overexpressed OS cells, the difference of the ratio of LC3II/I and the expression of P62 was not obvious between V7 overexpressed and control cells treated with DOX in combination with EGCG (Fig. 3d). Autophagy induction is associated with LC3, conjugated LC3 moves into autophagosomes and tightly binds to the autophagosome membrane. Thus, LC3 translocation is a reliable biomarker of autophagy [18]. Immunofluorescence staining was performed to detect the LC3 translocation. Meanwhile, MDC staining was another commonly used means of autophagy detection. The results showed that in control cells (Lv-Con), EGCG treatment could reduce the LC3 puncta and the MDC fluorescence signals obviously compared with DOX single treatment groups. While when V7 was overexpressed in OS cells, both the LC3 puncta and the MDC fluorescence signals enhanced obviously compared with the control group (Lv-Con). While, the difference of LC3 puncta and the MDC fluorescence signals were not obvious between V7 overexpressed (Lv-SOX2OT-V7) and control cells (Lv-Con) treated with DOX in combination with EGCG (Fig. 3e).The above results basically prove that EGCG inhibited DOX-induced autophagy by targeting SOX2OT V7 partially.

Cancer stem cells (CSCs) is a small subset of cancer cells characterized by self-renewal capability and pluripotent differentiation potential. Therefore, CSCs are believed to be of great importance in carcinogenesis, metastasis, drug-resistance and cancer recurrence. The existence of CSCs in osteosarcoma that are capable of forming suspended spherical, clonal colonies, preferentially express CSC key marker genes has been demonstrated [19]. EGCG could inhibit SOX2OT V7 which is closely related to pluripotency regulator SOX2 of CSCs, therefore, we speculated that EGCG would exert certain inhibitory effect on osteosarcoma stem cells (OSCs). CD133 is considered to be the most common CSC marker, and has been used repeatedly for the successful isolation of osteosarcoma stem cells [19]. A growing number of studies have found that using immunomagnetic double positive screening could obtain CSCs with higher purity and stemness [20, 21]. In this study, stem cell like cells were obtained from osteosarcoma cells by CD133+/CD44+ immunomagnetic beads double positive screening method and the isolated CD133+/CD44+ CSCs were identified by flow cytometry analysis. The results showed that after immunomagnetic double positive screening, the percentage of CD133+/CD44+ positive cells reached over 90% in U2OS (Fig. 4a) parental cells, which was significantly higher than that before screening. In order to further verify the relative stemness of cells after immunomagnetic beads screening, qRT-PCR was used to detect the relative mRNA levels of conventional OSC markers. The results showed the mRNA levels of Nanog, OCT4 and Sox2 were all upregulated significantly after CD133+ screening compared with their parental cells. It is worth mentioning that, after further CD44+ screening of CD133+ cells, further significant upregulation of Nanog, OCT4 and Sox2 was observed in CD133+/CD44+ cell compared with CD133+ cells (Fig. 4b). The results also illustrated to some extent that CD133+/CD44+ double positive screening was more effective than CD133+ single screening. On the other hand, mRNA level of SOX2OT V7 was also detected in immunomagnetic bead-screened cells and was found with similar alteration to those of OSC markers.(Fig. 4c). Subsequently, the screened CD133+/CD44+ cells were termed as OSCs. To determine whether EGCG could inhibit osteosarcoma stem cells in vitro, qRT-PCR and western blot were performed to detect the expression of OSC markers. The results showed the Nanog, OCT4, Sox2, c-Myc and ABCG2 were all upregulated significantly in OSCs compared with parental cells (Con), but their expression reduced obviously after EGCG addition compared with OSCs without EGCG treatment (Fig. 4d and e).

After confirming that EGCG did inhibit OSCs, it was our next concern that the inhibition was partly targeting SOX2OT V7. OSCs screened from U2OS and SaoS2 were infected with V7 overexpressed lentivirus (Lv-V7) to generate gain-of-function OSC models of SOX2OT V7. qRT-PCR results showed that SOX2OT was successfully overexpressed in OSCs from both U2OS and SaoS2 (Fig. 5a). The sphere clone formation assay was performed to evaluate the self-renewal ability of OSCs. The results showed EGCG treatment could abate the tumor sphere formation ability of OSCs and V7 overexpression could enhance the spheroid formation ability in OSCs obviouly not only in sphere number but also in spheroid size. While, when the V7 overexpressed OSCs were treated by EGCG with the same condition, he alteration of sphere formation ability was no longer obvious compared with V7 overexpressed OSCs treated without EGCG (Fig. 5b and C). Then, MTT assay was performed to study the inhibitory effect of EGCG and DOX, alone or in combination on OSC growth. It was demonstrated that EGCG could inhibit the growth of OSCs which seems slightly stronger than that of DOX. V7 over-expression in OSCs could reduce the growth inhibitory effect of both DOX and EGCG when they were single used. At the same time, the OSC growth inhibition effect of EGCG combined with DOX was also significantly decreased after overexpression of V7 in OSCs (Fig. 5d). These results suggested that EGCG inhibited the stemness of OSCs, and SOX2OT V7 was one of the important targets for the OSC inhibition of EGCG.

Notch signaling pathway plays an essential role in the development of cancer. Besides, Notch signaling pathway is also one of the classic signaling pathways of cancer stem cells. It was reported that EGCG could down-regulate the expression of Notch signaling genes and target genes [22‐24]. To illustrate whether the OSC inhibitory effect of EGCG targeting SOX2OT V7 is partly achieved by mediating of Notch signaling, relative mRNA levels of target genes, receptors and ligands of Notch signaling were detected using qRT-PCR in V7 overexpressed OSCs with or without EGCG treatment. The results showed that the mRNA level of Notch target genes Hey1 and Hes1 enhanced significantly in OSCs compared with their parental cells. Besides, Hes1 and Hey1 expression further increased in V7 overexpressed OSCs compared with Lv-con OSCs. The results suggested V7 overexpression could further activated Notch signaling which has been active in OSCs. On the other hand, EGCG treatment could depresse the mRNA levels of Hes1 and Hey1 in OSC control groups which indicated that EGCG inhibited the Notch signaling pathway in OSCs. However, the inhibitory effects of EGCG on Hes1 and Hey1 were not significant in V7 overexpressed OSCs which showed that overexpression of V7 was associated with a significant reduction in the inhibitory effect of EGCG on Notch signaling. To further understand the factors involved in this process, Notch receptors (Notch1/2/3) and ligands (Jagged1/2, Dll1/3/4) were invested by qRT-PCR. The results showed only relative expression of Notch3 enhanced obviously compared with Lv-con group when V7 was overexpressed in OSCs. Meanwhile,EGCG treatment reduced Notch 3 expression in Lv-con group cells which was almost abolished when V7 was overexpressed in OSCs. In terms of ligand, only DLL3 showed similar expression changes with Notch3 (Fig. 6a). Based on the above experimental results, we speculated preliminarily that Notch3 / DLL3 signaling plays a role in OSC inhibitory effect of EGCG targeting SOX2OT V7.

Then we used nude mouse tumorigenicity assay to further verify the above results (Table 3). It was demonstrated that, when being inoculated the same number of OSCs, V7 overexpressed OSCs induced tumors with larger volume compared with Lv-con group cells. In the course of tumorigenesis, EGCG was injected into nude mice inoculated with Lv-V7 OSCs which did not reduce the tumor size obviously. It is worth noting that the tumor size was much smaller when EGCG was injected into nude mice inoculated with Lv-V7 OSCs in combination of Notch3 knockdown lentivirus (Lv-Notch3-). This result was proved again that OSC inhibitory effect of EGCG targeting SOX2OT V7 was partly achieved by restrain of Notch3 to some extent (Fig. 6b and C).