Stretch-Induced Lncrna SNHG8 Inhibits Osteogenic Differentiation by Regulating EZH2 in Hpdlscs

Background: Periodontal ligament stem cells (PDLSCs) are important for the remodeling of the alveolar bone while tooth moving. However, the effect of long non-coding RNA (lncRNA) on osteogenic differentiation of PDLSCs under mechanical force remains unclear. Methods: In this study, we compared stretched and non-stretched PDLSCs by high-throughput sequencing. The verication and selection of lncRNAs were achieved by quantitative reverse transcription polymerase chain reaction (qRT-PCR). PDLSCs osteogenic differentiation potentials were assessed by alkaline phosphatase (ALP) staining, Alizarin Red staining, qRT-PCR, and western blot. The application of mechanical force used Flexcell-FX-6000-Tension System in vitro, and constructing rats’ tooth movement model in vivo. To verify the osteogenic regulation ability of small nucleolar RNA host gene 8 (SNHG8), PDLSCs were stretched or applied osteogenic induction after been infected by lentivirus. RNA uorescence in situ hybridization, isolation of nuclear and cytoplasmic RNA, qRT-PCR and western blot were performed to locate SNHG8. Western blot and qRT-PCR to nd the relationship between enhancer of zeste homolog 2 (EZH2) and SNHG8. Results: Our results demonstrated that among lncRNAs altered screened by high-throughput sequencing, the expression level of SNHG8 steadily decreased after being stretched. Analysis of mRNA expression and protein levels revealed an upregulation of ALP and RUNX2, ALP and Alizarin Red staining showed more obvious alkaline phosphatase and more mineralized nodules in SNHG8 knockdown PDLSCs. In vivo experiments showed lower expression of the homologous gene of SNHG8 after tooth movement, and better ability of ectopic osteogenesis after knockdown SNHG8. The verication of SNHG8’s nuclear location led us to infer that SNHG8 may interact with EZH2. The qRT-PCR and western blot results disclosed EZH2 expression reduced along with the knockdown of SNHG8. Furthermore, knockdown of EZH2 lead to PDLSCs’ osteogenic differentiation ability increasing under osteogenic induction according to the mRNA level of ALP and RUNX2 accompanied by ALP and Alizarin Red staining results. Conclusion: In general, our study conrmed that mechanically sensitive lncRNA SNHG8 can inuence the osteogenic differentiation of PDLSCs through epigenetic pathways without directly encoding protein, which provides solid evidence for the regulation by non-coding genes.


Introduction
Human periodontal ligament stem cells (hPDLSCs) are stem cells derived from periodontal tissues, which have multiple ability to differentiate (1)(2)(3). PDLSCs are essential for the remodeling of the alveolar bone during orthodontic procedure (4,5). According to classic pressure-tension theory (6), force-induced bone remodeling is caused by cell differentiation in the presence of osteogenic-related chemical messengers, cytokines such as hydrogen sul de, transcription factors such as RUNX2, and the Wnt/β-catenin pathway (7)(8)(9).
Mechanical force is a common stimulation in physiological and pathological activities. With the continuous in-depth research such as cell behavior and tumor formation, scholars have discovered that mechanical force not only can guide the differentiation and proliferation of cells in embryonic development, but also participates in the development of force-related organs and tissues. Furthermore, it also plays an important role in determining the fate of stem cells (10). Since orthodontics provide a cellular environment closely related to mechanical forces, our previous studies have focused on mechanical force-related genes and their ability to affect osteogenic differentiation. We found the express pattern and the function of coding and non-coding genes induced by mechanical force and their downstream pathway (11,12). Then a series of microRNAs (miRNAs) analysis experiments proved miR-21 as a key miRNA related to mechanical force induced osteogenic differentiation (13). We also con rmed that long non-coding RNAs (lncRNAs) and mRNAs can competitively interact with the same miRNA (14). The close relationship between miRNAs and lncRNAs is one of the bases to deduce the cellular physiology function of lncRNAs. However, it remains to be determined if there are mechanical force-sensitive lncRNAs, and whether these lncRNAs can regulate the osteogenic differentiation of hPDLSCs under mechanical force.
LncRNAs are a type of non-coding RNA with a length of more than 200 nucleotides. They were previously regarded as genetic "noise" because they do not encode proteins. In recent decades, the development of high-throughput sequencing has allowed researchers to further study this eld (15,16). LncRNAs can bind to and target chromatin regulators, act as RNA enhancers (17), interact with miRNAs (18,19), and mediate related signaling pathways, thereby regulating physiological and pathological processes (20,21) at the transcription and post-transcription levels (22,23). The mechanically sensitive lncRNA small nucleolar RNA host gene 8 (SNHG8), which was screened through high-throughput sequencing, was showed to interact with enhancer of zeste homolog 2 (EZH2) based on the RNA immunoprecipitation assay (24). SNHG8 is a member of the SNHG family of genes, which is closely related to the prognosis and progression of a variety of cancers, and plays a key role in regulating tumor progression and cell proliferation (25)(26)(27). EZH2 is one of the core subunits of polycomb repressive complex 2 (PRC2), and the correlation between EZH2 and mesenchymal stem cell (MSC) osteogenic differentiation has been con rmed (28)(29)(30).
In this study, the high-throughput sequencing results were screened for lncRNAs with high Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) scores, which found considerable lncRNAs showed difference in expression level after being stretched. We selected several lncRNAs and veri ed their changes in expression using quantitative reverse transcription polymerase chain reaction (qRT-PCR). Then we knocked down SNHG8 to determine whether lncRNA silencing could alter osteogenic differentiation under stretching or osteogenic induction in a series of experiments. The in vivo experiments achieved consistent results. We also con rmed the interaction between SNHG8 and EZH2, and veri ed the close relationship between the expression level of EZH2 and the osteogenic differentiation of hPDLSCs. of 2.0×10 6 cells per well. After the density reached ~80% con uence, the cells were serum deprived (2% serum) for 24 h before stretching. We imposed 10% stretch at 0.5 Hz. The control group was cultured in same silicone bottomed plates and the same culture environment without stretching.

High-throughput sequencing
Total RNA was extracted from the non-stretched and stretched groups of cells using RNAiso TM Plus (Takara, Shiga, Japan) according to the manufacturer's protocols. Strand-speci c cDNA libraries were constructed following a previously described protocol (34) and were sequenced on the Illumina HiSeq 2000/2500 sequencer (LC Biotech, Guangzhou, China). Sequencing was done according to the HiSeq 2000 User Guide with paired-end program.
RNA extraction and quantitative reverse transcription polymerase chain reaction (qRT-PCR) Total RNA was isolated from PDLSCs using RNAiso TM Plus (Takara) according to the manufacturer's protocol. The extracted total RNA was reverse-transcribed using the Prime Script RT Reagent Kit with gDNA Eraser (Takara). Relative RNA level was detected using the LightCycler-480 system (Roche Diagnostics GmbH, Mannheim, Germany) and TB Green Premix Ex Taq II (Takara). GAPDH and U6 were used as internal controls to quantify and normalize the results. The PCR reaction conditions were as follows: 95 ℃ for 30 s, then 55 cycles of 95 ℃ for 10s, 60 ℃ for 30 s. The 2-ΔΔCT value was used for comparative quantitation. The sequences of primers are shown in Table 1. All PCR processes were performed in triplicate.
Regarding SNHG8, two lentiviral constructs designated sh-SNHG8-1# and sh-SNHG8-2# were generated based on different regions of the human SNHG8 sequence (NCBI Gene ID: 100093630). For EZH2 knockdown, we constructed sh-EZH2 based on previous researches (35). The negative control containing a nonspeci c RNA oligonucleotide was constructed as previously described (36,37). Cells were observed under a uorescence microscope and an inverted phase contrast microscope (TH4-200; Olympus, Tokyo, Japan).
Construction of tooth movement model in vivo Twenty ve 6-week-old male wistar rats (Charles River, Beijing, China) were used for the construction of tooth movement model in vivo. All rats were fostered 12/12 h day/night cycle to simulate common environment, and were adaptively fed for 3 days before experiment. The maxillary left rst molar and the upper incisors were ligatured by 0.25 mm stainless steel with a nickel-titanium closed-coil spring (TOMY, Japan) in between. The nickel-titanium spring provide a force of approximate 20 g. In order to x the structure, we drilled grooves at the left upper incisor tooth cervix, and xed with light-curing resin. To avoid the individual differences of rats, each rat was performed tooth movement operation on dentition of the left maxillary, and the right side was not operated as a self-control. After retained the structure for 3 days, 7 days, 14 days, and 21 days, the periodontal tissues (include alveolar bone and periodontal ligament) were isolated for qRT-PCR.

Ectopic osteogenesis in vivo
The osteogenic differentiation potential of PDLSCs with different SNHG8 expression levels were tested by in vivo ectopic bone formation analysis. Brie y, untransfected hPDLSCs (Control group), hPDLSCs transfected with empty plasmids (sh-NC group), and hPDLSCs transfected with effective lentivirus (sh-SNHG8 group) were transferred subcutaneously to 5-week-old nude mice (Charles River) with osteoinductive calcium phosphate bioceramic material (TH/P 1020, Sichuan University, China). After 10 weeks of fostering, the nude mice were executed and the ectopic bone formation under the skin was harvested.

Isolation of nuclear and cytoplasmic RNA
The nucleus and cytoplasm of PDLSCs were separated using the Ambion® PARIS™ Kit (Life Technologies, Frederick, MD, USA) according to the manufacturer's instructions. We lysed approximately 5.0×10 6 cells in ice-cold cell fractionation buffer, and separated the cytoplasmic fraction from the nuclear fraction by low-speed centrifugation. Then we lysed the nuclear fraction in cell disruption buffer. Two kinds of fraction were mixed with lysis/binding solution separately, washed with washing solution, and eluted with preheated elution solution. For qRT-PCR, GAPDH was used as the control for the nuclear fractions and U6 was the control for the cytoplasmic fractions.

RNA uorescence in situ hybridization
The uorescence in situ hybridization (FISH) assay was performed using a Fluorescence In Situ Hybridization Kit (Ribobio, Guangzhou, China) according to the manufacturer's instructions. After xed in 4% paraformaldehyde, PDLSCs were washed with phosphate-buffered saline containing 0.5% Trition X-100 to increase cell permeability. We observed PDLSCs with an inverted phase contrast microscope (DMi8; Leica, Germany) after incubating PDLSCs overnight at 37 ℃ with hybridization solution containing the SNHG8, U6 and 18S probes. The excitation wavelengths were 405 nm for DAPI and 488 nm for the probes.
Statistical analysis.
All statistical calculations were performed using SPSS19.0 (SPSS Inc., Chicago, IL, USA). All data are normally distributed and presented as the mean±standard deviation of three to ve independent samples. Differences between the results obtained from various experimental groups were analyzed by the Student's t-test or one-way analysis of variance. P < 0.05 was considered statistically signi cant.

Results
Cell culture and identi cation of biological characteristics The cultured cells had the typical spindle-shaped structure of hPDLSCs (Fig. 1A). According to the negative results for CD31 and CD45, positive results for CD146 and Stro-1 by ow cytometry, we con rmed that hPDLSCs were isolated (Fig. 1B). To verify the osteogenic and adipogenic differentiation of PDLSCs, we induced osteogenic and adipogenic differentiation induction in vitro. Then we performed ALP staining and Alizarin Red S staining to identify the osteogenic differentiation level, and Oil Red O staining to evaluate the adipogenic differentiation level. There were more calcium nodules in the osteogenic induction group were more than in the non-induction group (Fig. 1C, D). And the Oil Red O staining shown that the adipogenic induction group had successfully induced differentiation of lipidladen fat cells compared with the non-induction group (Fig. 1E).
Identi cation of mechanical force-sensitive lncRNA In order to select lncRNAs that showed different expression levels, we applied tension force to the hPDLSCs ( Fig. 2A) and subjectived non-stretched and stretched groups of hPDLSCs to high-throughput sequencing. The detailed sequencing results have been published (14). The results showed that 14 704 lncRNAs had different expression levels after being stretched, of which 7 526 were known lncRNAs and 7 178 were unknown lncRNAs. To narrow the range, lncRNAs were selected with more than three times the expression difference after the application of mechanical force (107 were known lncRNAs, 67 of which showed a downward trend, and 40 of which showed an upward trend. 1 252 were novel lncRNAs, 496 of which showed a downward trend, and 756 of which showed an upward trend). Twelve lncRNAs (6 known, 6 unknown) with large differences in the expression levels and a sum of fpkm greater than 5 were selected for qRT-PCR veri cation of expression in the two groups of hPDLSCs. The expression trend of most of the chosen lncRNAs are consistent with the high-throughput sequencing results (Fig. 2B). Among them, the decrease of SNHG8 after the mechanical force showed excellent stability. These data indicate the potential relationship between SNHG8 and mechanical force in hPDLSCs.

SNHG8 has a negative effect on osteogenic differentiation of PDLSCs under mechanical force
To determine the effect of SNHG8 on hPDLSCs under stretched, we performed a series of experiments. First, to identify the time point at which there is the most signi cant change in SNHG8 expression under mechanical force, we evaluated the expression of SNHG8 in hPDLSCs after 6, 12, and 24 hours of stretching. The results showed that SNHG8 had the most signi cant decline after 6 hours of mechanical force (Fig. 3A). We use the lentivirus for the SNHG8 knockdown. To nd out a suitable MOI value, we used sh-SNHG8-1# and sh-SNHG8-2# to conduct a preliminary experiment. Although the MOI of 50 showed the higher transfection e ciency, the cells morphology changed from a long spindle shape to an irregular multi-synaptic shape. We decided that MOI of 30 was the best MOI value and transfection time of 24 hours that have a high transfection e ciency with no effect on cell activity (Fig. 3B). The most e cacious target was sh-SNHG8-2# (Fig. 3C). Fluorescence observation and qRT-PCR veri cation were used to verify the transfection e ciency. Mechanical force was separately applied to untransfected hPDLSCs (Control group), hPDLSCs transfected with empty plasmids (sh-NC group), and hPDLSCs transfected with effective lentivirus (sh-SNHG8 group). The qRT-PCR and western blot results showed that the expression of ALP was increased in the sh-SNHG8 group after being stretched, and RUNX2 in the sh-SNHG8 group expressed higher after stretching (Fig. 3D-F). These data con rmed that SNHG8 has a negative effect on the expression osteogenic differentiation relative genes of hPDLSCs under mechanical force.

SNHG8 has a negative effect on the osteogenic-induced osteogenic differentiation of PDLSCs
To investigate the function of SNHG8 in the osteogenic induced osteogenic differentiation of PDLSCs, we detected the expression level of SNHG8 after osteogenic induction. Interestingly, with the osteogenic induction of hPDLSCs, except for the uctuation in the seventh day of induction, the expression level of SNHG8 showed a signi cant and stable decrease (Fig. 4A). Which may suggest the negative in uence of SNHG8 during osteogenic differentiation of PDLSCs. To further con rm the effect of SNHG8 on the osteogenic differentiation of hPDLSCs, we cultured the control group, sh-NC group, and sh-SNHG8 group by mineralization induction medium. The qRT-PCR and western blot results showed signi cantly higher ALP expression in the sh-SNHG8 group after induction. RUNX2 expression, however, was more signi cantly upregulated in the non-induced group (Fig. 4B-D). We considered it may be because RUNX2 usually changes in the early stage of osteogenesis, and this kind of change is not so obvious when the induction time endured long. ALP staining (Fig. 4E) and Alizarin Red S staining (Fig. 4F, G) showed that more mineralized nodules were produced in the SNHG8 knockdown groups. These results show that downregulated of SNHG8 can signi cantly increase osteogenic differentiation in hPDLSCs.

The reduction of SNHG8 has a positive effect on osteogenic differentiation in vivo
We then decided to test whether the change of SNHG8 expression after receiving mechanical force in vivo had a similar reaction with cells experiments in vitro. According to previews studys (38), analogous genes may have similar functions. Therefore, we detected the SNHG8's homologous gene in rats, Smim4, after tooth movement of wistar rats (Fig. 5A, B). The qRT-PCR results showed that the Smim4 level in the periodontal tissue of the mechanical force side decreased steadily from 3 day to 14 days during rats' tooth movement, and returned to the normal level by 21 days (Fig. 5C). It proved that the expression of the Smim4 gene has decreased during the early stage of tooth movement in vivo. To gure out whether the change of SNHG8 expression can also alter the ability of osteogenic differentiation, we implanted control group, sh-NC group, and sh-SNHG8 group PDLSCs under the back of nude mice. The results of Masson's trichrome staining and SafraninO-staining (Fig. 5D) showed that the sh-SNHG8 group formed more collagen bers in nude mice, which is the necessary matrix for bone formation. In addition, the sh-SNHG8 group also had obvious red stained area of glycoprotein, which suggested cartilage formation. The formation of cartilage also suggests that the sh-SNHG8 group has better bone formation potential.

Expression level of SNHG8 correlates with EZH2
Compering to the 18S probe which located in the cytoplasm and the U6 probe which located in the nucleus, the SNHG8 probe of FISH assay indicated that SNHG8 located in both cytoplasm and nucleus, but most in nucleus (Fig. 6A). The separately detection of SNHG8 from cytoplasm and nucleus also supported these results (Fig. 6B). Since several researches have shown that lncRNAs in the nucleus can regulate physiological or pathological process by interacting with PRC2 (39-41), we considered that SNHG8 may also regulate the osteogenic differentiation of hPDLSCs through this pathway. We detected the RNA and protein levels of the main subunits of the PRC2 complex in the control, sh-NC and sh-SNHG8 groups. The qRT-PCR and western blot results showed that the RNA and protein levels of EZH2 and SUZ12 in the knockdown group were signi cantly reduced (Fig. 6C-E). In hPDLSCs with EZH2, EED and SUZ12 knockdown (Fig. 6F), the expression level of SNHG8 was also signi cantly decreased (Fig. 6G). These data suggest that SNHG8 can interact with PRC2, especially EZH2.

SNHG8 regulates osteogenic differentiation under mechanical force through EZH2
To con rm the osteogenic relevance of PRC2, we tested the expression of the three main subunits of PRC2 in cells under mineralized induction at different time points. The RNA expression of EZH2 decreased steadily with the increase of mineralized induction time, and the expression level of SUZ12 decreased during the osteogenic induction period, but uctuated during the process. However, the expression level of EED did not decrease signi cantly during the process of osteogenic induction (Fig.   7A). To further con rm whether EZH2, SUZ12, and EED have effects on osteogenic differentiation, we knocked down these three major subunits with small interfering RNA (siRNA) and short hairpin RNA (shRNA), and then performed osteogenic induction of control, negative control and PRC2 knockdown groups. The mRNA expression of ALP and RUNX2 in EZH2-knockdown group was signi cantly increased, the expression of ALP and RUNX2 in the SUZ12 knockdown group had distinct increase after 7 days of induction, and there is no apparent change of ALP and RUNX2 in the EED knockdown group (Fig. 7B). ALP staining (Fig. 7C) and Alizarin Red S staining (Fig. 7D, E) also showed that more mineralized nodules were produced in the EZH2 and SUZ12 knockdown groups. In general, PRC2 did have a negative effect on the osteogenic differentiation of hPDLSCs, which may be the reason why SNHG8 negatively regulates osteogenic differentiation.

Discussion
Tooth movement based on the remodeling of alveolar bone is the basis of orthodontics. A comprehensive and systematic understanding of cellular changes promotes a better understanding of physiological responses in clinical orthodontics. Although some studies have focused on lncRNAs related to osteogenic differentiation (30), mechanically sensitive lncRNAs and their in uence on the osteogenic differentiation of periodontal-derived stem cells are still unclear.
In our research, the lncRNA was chosen for further research due to the signi cant change after being stretched, SNHG8, has been reported to regulate the progression of non-small-cell lung cancer by sponging miR-542-3p (42). Coincidentally, miR-542-3p was proved to negatively regulates the osteogenic differentiation of vascular muscle smooth cells by targeting bone morphogenetic protein 7 (43). Accordingly, since the interacting between SNHG8 and miR-543-3p has been con rmed, SNHG8 may have the same regulatory ability on negatively regulating osteogenic differentiation. In this paper, our cell experiments in vitro and animal experiments in vivo both provided credible results. Interestingly, according to the results of animal experiments, the SNHG8 knockdown group not only showed greater collagen bers, but also formed more obvious cartilage compared with the non-transfected group. The effect of SNHG8 on chondrogenic differentiation is not discussed in this paper, but the signi cantly enhanced ability of chondrogenic differentiation after knockdown SNHG8 may be able to indicate the regulation to differentiation of stem cells by SNHG8.
According to the conclusions of previews studies, we can infer that most lncRNAs present in the nucleus can direct chromatin modi ers to speci c genomic sites (17,22), for example, PRC2 and histone H3 lysine 9 methyltransferase-mediated DNA trimethylation (44). Because SNHG8 is mainly located in the nucleus and lncRNAs may interact with PRC2 to regulate gene expression at the transcription or posttranscription level (45)(46)(47), we considered its effect on osteogenic differentiation under mechanical force may be related to epigenetic genes. PRC2 is an epigenetic inhibitor necessary for development in vivo and differentiation of embryonic stem cells in vitro (48), and mainly acts as a transcriptional inhibitor through the trimethylation of H3K27 histones, which is the core feature of achieving chromatin silencing (49). A recent study con rmed the direct interaction between SNHG8 and EZH2 (24), the main subunit of PRC2 and the key regulatory factor in epigenetics. Therefore, we were interested in determining if it has a regulatory in osteogenic differentiation and whether this role contributes to the process and mechanism of alveolar bone reconstruction in tooth movement, which prompted us to conduct a series of studies on the effect of PRC2 on osteogenic differentiation. As an epigenetic regulator, EZH2 is related to MSC osteogenic differentiation, and have been reported to regulate osteogenesis-related genes such as RUNX2. For example, not only osteogenic differentiation can lead to decreased binding of EZH2, SUZ12, and H3K27 trimethylation at the RUNX2 promoter (28), EZH2 can also inhibit RUNX2 expression and subsequent osteoblast differentiation (50). Combined with the fact that EZH2 can interact with SNHG8, and SNHG8 has been con rmed by us to have a negative regulatory effect on the osteogenic differentiation of PDLSCs, we preliminarily inferred and veri eded the hypothesis that EZH2 may also negatively regulate PDLSCs. After con rming the close relationship between SNHG8 and EZH2, we veri ed the important role of EZH2 on the osteogenic differentiation of PDLSCs, which is consistent with the conclusion of the above studys, and further suggested the possible relationship between EZH2 and the applying of mechanical force. In our experiments, despite SNHG8 cannot encode a protein, which means it cannot regulate physiological or pathological processes by forming a form of a protein with speci c functions. However, after knocking down SNHG8 or EZH2, the expression of the other decreased, which indicated that there is an interaction between these two genes. Their interaction is a solid basis for SNHG8 to perform regulatory effects without directly encoding proteins. We believe that this process not only reveals the regulation of non-coding RNA, but may also reveal the epigenetic regulation of mechanical forces acting on cells and tissues.
Our study prospectively predicted and con rmed the effect of mechanically sensitive lncRNA SNHG8 on the osteogenic differentiation of PDLSCs. We found that SNHG8 has a negative effect on osteogenic differentiation from epigenetics aspect, veri ed the connection between SNHG8 and EZH2, and con rmed the role of EZH2 and other PRC2 subunits in the osteogenic differentiation of PDLSCs. The undeniable limitation of this study is that no speci c changes of downstream of EZH2 (such as the perform of H3K27) have been studied, and it has not veri ed whether EZH2 can directly respond to mechanical stimulation. However, our study stimulates new ideas for future research into the effect of genes related to cell differentiation under mechanical force and the mechanistic relationship between lncRNAs and epigenetic regulators.

Conclusion
Our results provide a solid evidence for the mechanical force-sensitive lncRNA SNHG8 to regulate the osteogenic differentiation PDLSCs through EZH2. It provides a new idea for non-coding RNA to regulate cell differentiation through coding RNA. In addition, it can act as a theoretical basis for further clinical application of genetic engineering.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.