Lnc‐NTF3‐5 promotes osteogenic differentiation of maxillary sinus membrane stem cells via sponging miR‐93‐3p

Abstract Background The function and the mechanism of long non‐coding RNAs (lncRNAs) on the osteogenic differentiation of maxillary sinus membrane stem cells (MSMSCs) remain largely unknown. Materials and Methods The expression of lnc‐NTF3‐5 and Runt‐related transcription factor 2 (RUNX2), Osterix (OSX), and Alkaline Phosphatase (ALP) was examined by quantitative real‐time PCR (qRT‐PCR) in MSMSCs during the process osteogenic differentiation. Then the function of lnc‐NTF3‐5 was evaluated by loss‐ and gain‐of‐function techniques, as well as qRT‐PCR, western blot, and Alizarin Red staining. In addition, the microRNAs (miRNAs) sponge potential of lnc‐NTF3‐5 was assessed through RNA immunoprecipitation, dual luciferase reporter assay, and in vivo ectopic bone formation. Results Lnc‐NTF3‐5, RUNX2, OSX, and ALP increased alone with the differentiation. Inhibition of lnc‐NTF3‐5 decreased the expression of RUNX2, OSX, and ALP both at mRNA and protein levels. Alizarin red staining showed similar trend. In contrast, overexpression of lnc‐NTF3‐5 presented totally opposite effects. Besides, overexpression of lnc‐NTF3‐5 could decrease the expression of microRNA‐93‐3p (miR‐93‐3p). Enhance miR‐93‐3p could also inhibit the expression level of lnc‐NTF3‐5. RNA immunoprecipitation demonstrated that lnc‐NTF3‐5 is directly bound to miR‐93‐3p and dual luciferase reporter assay proved that miR‐93‐3p targets 3′ UTR of RUNX2 to regulate its expression. Ultimately, in vivo bone formation study showed that lnc‐NTF3‐5 and miR‐93‐3p inhibitor co‐transfection group displayed the strongest bone formation. Conclusions The novel pathway lnc‐NTF3‐5/miR‐93‐3p/RUNX2 could regulate osteogenic differentiation of MSMSCs and might serve as a therapeutic target for bone regeneration in the posterior maxilla.

inadequate bone height in the posterior maxilla. 4 Clinical studies have shown that maxillary sinus augmentation can be reached by elevating the MSM with or without grafting, 5 which suggested that maxillary sinus membrane possesses osteogenic activity. Further studies have isolated and characterized maxillary sinus membrane stem cells (MSMSCs) in maxillary sinus membrane, which displayed stem cell properties similar to bone marrow mesenchymal stem cells (BMMSCs). 6 Our previous study demonstrated that MSMSCs could express MSC markers such as STRO-1, CD146, CD29. and CD44. 6 These cells can form colonies and differentiate into adipocytes, osteoblasts, and chondrocytes. 6 . Evidence further showed that MSMSCs can differentiate into osteoblasts both in vitro and in vivo, 7-10 which represented a useful cell source of MSCs for bone tissue engineering.
Despite several researches on MSMSCs differentiation, the precise molecular mechanisms of osteogenic differentiation remain unclear.
Osteogenic differentiation of MSCs is controlled by a variety of signaling pathways. 11 Recently, long non-coding RNAs (lncRNAs) have emerged as important players in controlling stem cells fate and behavior. 12 For instance, a preliminary research found that a number of lncRNAs were significantly altered during the osteogenic differentiation of C3H10T1/2 cells. 13 Anti-differentiation ncRNA (ANCR) functions as a suppressor of osteogenic differentiation through a pathway suggested to involve enhancer of zeste homolog 2 (EZH2) and RUNX2. 14 Huang et al. revealed a H19/miR-675/TGF-b1/Smad3/HDAC pathway in regulating osteogenic differentiation of human MSCs. 15 In addition, lncRNA-POIR 16 and MEG3 17 were reported to regulate osteogenic differentiation through different mechanisms, whereas knockdown of HoxA-AS3 expression enhanced osteogenic differentiation and osteogenesis markers expression in MSCs. 18 These studies implied that lncRNAs both positively and negatively control the differentiation state of stem cells. Thus, the role of lncRNAs in MSMSCs differentiation is a great worthy studying question.
In our previous research, differential lncRNAs expression profiles in MSMSCs during osteogenic differentiation were established via microarray, and several upregulated lncRNAs between undifferentiated and osteogenic differentiated MSMSCs have been identified. 19 LncRNA-MODR/miR-454/RUNX2 pathway was suggested to regulate osteogenic differentiation of MSMSCs. 19 Meanwhile, lnc-NTF3-5 (ENST0000537192.1) displayed a significantly increase after 7 d of induction. Nevertheless, the function and the underlying molecular mechanism of lnc-NTF3-5 regulates the osteogenic differentiation of MSMSCs are still unknown.
To address these questions, this study aims to explore the role of lnc-NTF3-5 in osteogenic differentiation of MSMSCs and how it regulates the process. This study may provide a new molecular mechanism for osteogenic differentiation of MSMSCs and a new therapeutic target for bone regeneration in the posterior maxilla.

| Informed consent
Studies were conducted in accordance with the guidelines of the Medical Ethics Committee of Sun Yat-Sen University and approved by the Sun Yat-Sen University Joint Institutional Review Board (IACUC-DB-16-1212). All donors provided informed written consent prior to specimen collection.

| Cell culture
HEK-293T cells were purchased from the American Type Culture Collection (ATCC). MSMSCs were isolated from normal human MSM according to our previously published methods. 6 Briefly, tissues were minced and digested with 3 mg/mL collagenase type I (Sigma, St. Louis, Missouri) and 4 mg/mL dispase (Roche, Mannheim, Germany) for 1 h at 378C. Then the samples were passed through a 70 lm strainer (Falcon, BD Labware, Franklin Lakes, New Jersey). Cells were supplemented into 75-cm 2 culture flasks (Costar, Cambridge, Massachusetts) with alpha modification of Eagle's medium (GIBCO BRL, Grand Island, New York), and then incubated in 5% CO 2 and 378C.

| Osteogenic differentiation
Osteogenic differentiation was conducted according to the instructions from the manufacturer (Invitrogen, Carlsbad, California). MSMSCs were plated into 6-well plates (Corning Life Sciences, Tewksbury, Massachusetts) at a density of 1 3 10 4 cells/well. BMP2 (100 ng/mL) was used to induce the differentiation when cells were reached 80% confluence.

| Quantitative real-time PCR
RNA was extracted from cells via TRIzol reagent (Invitrogen, Carlsbad, California), then cDNA was synthesized with SuperScript III reverse transcriptase kit (Qiagen, Valencia, California) according to the instruction from the manufacturer. The qRT-PCR was performed with Taq-Man Universal PCR master mix (Applied Biosystems, Foster City, California) as described previously. 6 Relative mRNA and lncRNA expression were calculated with normalization to GAPDH, and U6 was used to normalize expression of miRNA. The primers used for amplification are listed in Table 1.  Meditec, Inc, Dublin, California). Alizarin Red staining quantification was performed according to our previously published methods. 9 One milliliter of 10% cetylpyridinium chloride (Sigma, St. Louis, Missouri) was added to each well for 45 min. The absorbance was read at 562 nm.

| RNA immunoprecipitation (RIP)
RIP assay was performed with Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, Billerica, Massachusetts) in accordance with the instructions from the manufacturer. Briefly, magnetic beads were incubated with antibody against Ago2 (Millipore, Billerica, Massachusetts) and mouse IgG control. RNA was extracted and relative expression levels of lnc-NTF3-5 and miR-93-3p in extract were detected by qRT-PCR as described above.

| Statistical analyses
All results were presented as the mean 6 standard deviations (SD) from at least in triplicate and were analyzed using two-tailed unpaired Student's t-test. The P values were adjusted using the Bonferroni correction for post hoc analysis for multiple group analysis. A P value < .05 was considered statistically significant. Alkaline Phosphatase (ALP) were also upregulated from day 0 to day 7 and declined on day 14 ( Figure 1C). As indicated by RNA-FISH assay, lnc-NTF3-5 was found to be distributed in cytoplasm of MSMSCs, which was similar to 18S expression. In contrast, U6 was located in the nucleus ( Figure 1D). RNA-FISH results suggested that lnc-NTF3-5 exert its function in cytoplasm.

| Lnc-NTF3-5 upregulation promoted osteogenic differentiation of MSMSCs
Combined with previous observations, we speculated that lnc-NTF3-5 can stimulate osteogenic differentiation of MSMSCs induced with BMP-2. To prove this hypothesis, lentiviral vector was constructed to overexpress lnc-NTF3-5 in MSMSCs. The results justified our conjecture. The expression level of lnc-NTF3-5 was increased to more than 15-fold higher than the vector group (Figure 2A). After cultured in

| MiR-93-3p suppressed the expression of RUNX2
To further explore the mechanism of lnc-NTF3-5 promote osteogenic differentiation, Target Scan and miRDB were used to analysis the bind- in response to miR-93-3p were analyzed. As shown in Figure 5C and D, RUNX2 protein levels were significantly decreased by the treatment with the miR-93-3p mimics and increased by the treatment with the miR-93-3p inhibitor. Taken together, these results indicated that miR-93-3p inhibits RUNX2 expression through direct binding to the target site in the RUNX2-3 0 UTR.
To further explore how lnc-NTF3-5 exert its function, we next trans-

| Lnc-NTF3-5 increased in vivo ectopic bone formation in human MSMSCs
The results described above demonstrated that lnc-NTF3-5 may function as miR-93-3p sponges to regulate osteogenic differentiation of MSMSCs in vitro, we next explored whether the regulation of the lnc-

| D I SCUSSION
LncRNAs are generally defined as transcripts of longer than 200 nucleotides without evident of coding proteins and can regulate diverse cellular processes. 21 They are expressed at significantly lower levels but more cell type specific than protein-coding RNAs. 22 Our recent study found that lncRNAs can affect the osteogenic differentiation of MSMSCs, the top 5 lncRNAs were validated by qRT-PCR. 19 LncRNA-MODR plays a positive role in osteogenic differentiation of MSMSCs. 19 Similarly, researchers found that lncRNA-MEG can promote osteogenic differentiation as well. 17 This osteogenic promotion effect was demonstrated in human adipose-derived mesenchymal stem cells (ADMSCs) recently. 23 . The current study found that lnc-NTF3-5 was able to enhance RUNX2, OSX, and ALP expression and further promoted the osteogenic differentiation of MSMSCs, which suggested that lnc- There is now evidence that lncRNAs regulate gene expression and function at both transcriptional and post-transcriptional levels. 29 Four different mechanisms have been reported in lncRNAs and miRNAs interaction: miRNAs can bind to lncRNAs and reduce its stability; lncRNAs can act as sponges of miRNAs; lncRNAs can compete with miRNAs for binding to target mRNAs; and lncRNAs can generate miRNAs. 30 The interaction between lncRNAs and miR-NAs represents a new aspect of physiological and pathological Many studies suggested that miRNAs can directly target the osteogenic master regulator RUNX2. 33 To elucidate the role of lnc-NTF3-5 in regulating osteogenic differentiation, bioinformatics analysis was performed to predict whether miR-93-3p can bind to the key transcription factor for osteogenesis RUNX2. Results found that miR-93-3p can directly bind to 2310-2316 nt of RUNX2 3 0 UTR, which suggested that miR-93-3p may regulate RUNX2 expression. Previous studies revealed that a plenty of miRNAs were involved in suppressing osteogenesis by targeting RUNX2, such as miR-103a, 34 miR-133a, 35 miR-135a, 33 miR-204/211, 36 and miR-217. 37 miR-93-3p was reported to be upregulated in HIV-associated neurocognitive-disordered patients, 38 while downregulated in radioresistant nasopharyngeal carcinoma cells. 39 Meanwhile, miR-93-3p was suggested to be a potential biomarker of acute kidney injury (AKI) in intensive care units (ICU), and cardiac surgery (CS) patients since it was significantly downregulated during the days prior to AKI diagnosis. 40 . However, the function of miR-93-3p in osteogenic differentiation is still unknown. This study not only first demonstrated that miR-93-3p acts as a negative regulator of osteogenic differentiation but also extended the RUNX2-targeting miR-NAs family by showing that miR-93-3p strongly inhibit RUNX2 expression at both mRNA and protein levels.
Since lnc-NTF3-5 functions as miR-93-3p sponges, we next explored whether the regulation of the lnc-NTF3-5 and miR-93-3p in MSMSCs also exerts an effect on bone formation in vivo. Results suggested that lnc-NTF3-5 overexpression led to significant new bone formation in mice. Moreover, co-transfection of miR-93-3p inhibitor led to an even enhanced bone formation in mice. These data suggest that lnc-NTF3-5 promoted bone formation through competitively binding to miR-93-3p in vivo.
In conclusion, this study first characterized the osteogenic promoting function of lnc-NTF3-5 in MSMSCs and demonstrated that lncRNA-microRNA crosstalk may be the underlying mechanism. Lnc-NTF3-5 competitively binding to miR-93-3p promotes the expression of RUNX2. This study indicated lnc-NTF3-5 might be a novel biomarker or therapeutic target for promoting bone regeneration in the posterior maxilla.