A NOTCH1/LSD1/BMP2 co-regulatory network mediated by miR-137 negatively regulates osteogenesis of human adipose-derived stem cells

Background MicroRNAs have been recognized as critical regulators for the osteoblastic lineage differentiation of human adipose-derived stem cells (hASCs). Previously, we have displayed that silencing of miR-137 enhances the osteoblastic differentiation potential of hASCs partly through the coordination of lysine-specific histone demethylase 1 (LSD1), bone morphogenetic protein 2 (BMP2), and mothers against decapentaplegic homolog 4 (SMAD4). However, still numerous molecules involved in the osteogenic regulation of miR-137 remain unknown. This study aimed to further elucidate the epigenetic mechanisms of miR-137 on the osteogenic differentiation of hASCs. Methods Dual-luciferase reporter assay was performed to validate the binding to the 3′ untranslated region (3′ UTR) of NOTCH1 by miR-137. To further identify the role of NOTCH1 in miR-137-modulated osteogenesis, tangeretin (an inhibitor of NOTCH1) was applied to treat hASCs which were transfected with miR-137 knockdown lentiviruses, then together with negative control (NC), miR-137 overexpression and miR-137 knockdown groups, the osteogenic capacity and possible downstream signals were examined. Interrelationships between signaling pathways of NOTCH1-hairy and enhancer of split 1 (HES1), LSD1 and BMP2-SMADs were thoroughly investigated with separate knockdown of NOTCH1, LSD1, BMP2, and HES1. Results We confirmed that miR-137 directly targeted the 3′ UTR of NOTCH1 while positively regulated HES1. Tangeretin reversed the effects of miR-137 knockdown on osteogenic promotion and downstream genes expression. After knocking down NOTCH1 or BMP2 individually, we found that these two signals formed a positive feedback loop as well as activated LSD1 and HES1. In addition, LSD1 knockdown induced NOTCH1 expression while suppressed HES1. Conclusions Collectively, we proposed a NOTCH1/LSD1/BMP2 co-regulatory signaling network to elucidate the modulation of miR-137 on the osteoblastic differentiation of hASCs, thus providing mechanism-based rationale for miRNA-targeted therapy of bone defect. Supplementary Information The online version contains supplementary material available at 10.1186/s13287-021-02495-3.

that miR-137 regulates cell proliferation, differentiation [24][25][26][27][28], and neuronal maturation [29][30][31] in adult or mouse stem cells. Nevertheless, during the process of hASCs differentiating into osteoblastic lineage, the functions and epigenetic mechanisms of miR-137 have not been investigated except for our previous study [32], in which we disclose part of the mechanisms as the coordination between lysine-speci c histone demethylase 1 (LSD1) and BMP2-mothers against decapentaplegic homolog 4 (SMAD4) pathway. Considering that the relationships of osteogenesis-associated signals are complex and diverse and massive molecules participating in the LSD1/BMP2/SMADA4 network remain unascertained, we need to further clarify the regulatory mechanisms of miR-137 on the osteogenesis. NOTCH signal is a fundamental pathway in bone remodeling and skeletal homeostasis [33][34][35]. Hairy and enhancer of split 1 (HES1), a downstream molecule of NOTCH [36], is responsible for the actions of NOTCH in the skeleton, even though its osteogenic effects are cell type-speci c and context-dependent.
But HES1 binds to the osteocalcin (OCN) promoter and suppresses its transcription in osteoblastic cells [38]. HES1 inactivation not only increases the femoral length and trabecular number in the limb bud of transgenic mice, but also enhances mineral apposition rate and suppresses bone resorption in osteoblasts [39]. NOTCH1 has emerged as a target of miR-137 in human renal mesangial cells [40], retinal ganglion cells [41], neurons [42], non-small cell lung cancer cells [43] and breast cancer cells [44], but whether it is directly inhibited by miR-137 has not yet been identi ed in hASCs. In small cell lung cancer cells, NOTCH1 pathway is activated by LSD1 inhibitor and suppressed due to the binding of LSD1 [45]. Additionally, the induction of NOTCH signal impairs the activation of BMP pathway and the osteoblastic differentiation of dental follicle cells [46]. In contrast, NOTCH1 upregulates BMP2 expression in human aortic valve interstitial cells through the stimulation of NF-κB [47]. Our previous study con rmed that miR-137 knockdown induced BMP2-SMADA4 pathway through the downregulation of LSD1 dependently or independently [32], which coincides with the studies stating that LSD1 inhibition leads to increased BMP2 expression [48,49]. Accordingly, we postulate a signaling network entailing NOTCH1-HES1, LSD1 and BMP2-SMAD4 pathways to unveil the miR-137 modulation on the osteogenesis of hASCs.
This study identi ed the interactions of miR-137 and its downstream genes and revealed that the coregulatory signaling network of NOTCH1/LSD1/BMP2 mediated by miR-137 negatively modulated the osteogenesis of hASCs, suggesting that miR-137 might be applied as a promising therapeutic target for bone regeneration.

Mice
The animal experiments were conducted in strict conformity with the guidelines of Animal Welfare Committee of Health Science Center in Peking University (LA2019019). Male, 5-week-old BALB/c-nu/nu nude mice (Charles River, Wilmington, MA, USA) were randomly assigned to 3 groups (n = 6 per group) and maintained with speci c pathogen-free conditions.

Cell lines
The hASCs isolated from three separate donors were purchased in ScienCell Research Laboratories (Carlsbad, CA, USA). For each donor, the in vitro cell experiments were performed at least three times individually. For proliferation culture, cells were maintained in proliferation medium (PM), containing Dulbecco's modi ed Eagle medium (Thermo Fisher Scienti c, Rockford, IL, USA), 1% (v/v) penicillin/streptomycin (Thermo Fisher Scienti c) and 10% (v/v) fetal bovine serum (ExCell Bio, Shanghai, China). When the cells reached 70-80% con uence, osteoinduction was performed by adding osteogenic medium (OM), which contained the above culture medium for promoting proliferation, 100 nM dexamethasone (Sigma-Aldrich), 0.2 mM L-ascorbic acid (Sigma-Aldrich) and 10 mM β-glycerophosphate (Sigma-Aldrich). The cell culture conditions were 37°C with 5% CO 2 and 100% relative humidity.
Alkaline phosphatase (ALP) staining and quanti cation After 7 days of culture in PM or OM, the hASCs were used for ALP staining and activity test according to the published protocol [10]. ALP staining was operated following the BCIP/NBT staining kit (Beyotime, Shanghai, China) instructions. For the quantitative tests of ALP activities, cells were washed with phosphate buffer saline (PBS) and 1% Triton X-100 (Solarbio, Beijing, China), then scraped in Milli-Q water and subjected to three cycles of freezing and thawing. By employing the BCA method and the pierce BCA protein assay kit (Thermo Fisher Scienti c), total protein was read at 562 nm and computed with a bovine serum albumin standard curve according to the manufacturer's protocol. Afterwards, ALP activity was detected at 520 nm applying an alkaline phosphatase assay kit (Jiancheng, Nanjing, Jiangsu, China) and nally normalized to the total protein concentrations of cells.
Alizarin red S (ARS) staining and quanti cation After 14 days of culture in PM or OM, the hASCs were applied to detect the matrix mineralization.
Following being xed with 95% ethanol for 30 min, cells were soaked in 1% ARS staining solution (pH 4.2; Sigma-Aldrich, St. Louis, MO, USA) for 20 min at room temperature. To assess the degree of mineralization, stained areas of each well were separately dissolved in 100 mM cetylpyridinium chloride (Sigma-Aldrich) for 1 h and the absorbances were detected at 562 nm. Finally, the relative ARS intensity was normalized to the total protein concentrations of cells.
RNA extraction, reverse transcription, and quantitative real-time polymerase chain reaction (qRT-PCR) After 3, 7, and 14 days of culture in PM or OM respectively, total RNA of cells was isolated with TRIzol (Invitrogen, Carlsbad, CA, USA) and synthesized into the rst-strand cDNA using a reverse transcription system (Takara, Tokyo, Japan). All the transcripts were quanti ed using the FastStart universal SYBR green master (ROX) (Roche, Indianapolis, IN, USA) and a 7500 real-time PCR detection system (Applied Biosystems, Foster City, CA, USA). Relative expression levels of mRNA and miRNA were normalized to GAPDH mRNA and U6 snRNA, respectively. The sequences of the primers employed were listed in Additional le 4: Table S2.

Western blotting
The hASCs were rinsed with ice PBS three times and immersed in RIPA buffer (HuaxingBio, Beijing, China) mixed with protease inhibitor cocktail (HuaxingBio). The pierce BCA protein assay kit (Thermo Fisher Scienti c) was used to determine the protein concentration. A 25 μg sample of protein was added and separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, then followed by transfer to the polyvinylidene di uoride membranes (Millipore, Bedford, MA, USA). Strips on the membranes were blocked with 5% nonfat dry milk (BioRuler, Danbury, CT, USA) for 1 h at room temperature, incubated overnight at 4°C with primary antibodies at a dilution of 1:1000, and then for 1 h at room temperature with goat anti-rabbit secondary antibodies labeled with horseradish peroxidase ( Dual-luciferase reporter assay The 3' untranslated region (3' UTR) alignments of the target regions in NOTCH1 were predicted by TargetScan and RNA22. Reporter vectors were constructed based on the previous method [32]. The 3' UTR sequences of NOTCH1, which contained the possible binding sites of miR-137, were PCR ampli ed and then inserted into pEZX-MT06 vectors (GeneCopoeia, Rockville, MD, USA) to create NOTCH1-WT (wild-type NOTCH1) luciferase reporter plasmids. Mutated forms were generated by site-directed mutagenesis (GeneCopoeia) and named NOTCH1-MT (mutant-type NOTCH1) luciferase reporter plasmids. For luciferase assay, the hASCs were planted on 24-well culture plates with a density of 5 × 10 4 /well and co-transfected with 1 µg NOTCH1-WT or -MT plasmids, 100 nM NC or miR-137 mimics, and lipofectamine 3000 (Invitrogen). The luciferase activities were examined by a dual-luciferase reporter assay system (Promega, Madison, WI, USA) 48 h later, and standardized to renilla luciferase activity for each transfected well.

Heterotopic osteogenesis examinations in vivo
After the transfection with NC, miR-137 and anti-miR-137 lentiviruses, hASCs of the third passage were maintained in PM for 1 week, collected and incubated with auto-setting calcium phosphate cement (ACPC; Rebone, Shanghai, China) for 1 h at 37°C. Then the hASC-ACPC mixtures were transplanted subcutaneously to the dorsal regions of nude mice (n = 6 per group) for the analyses of heterotopic bone formation in vivo 8 weeks later. After being collected and xed in 4% paraformaldehyde, the samples were photographed with soft X-ray. The radiograph was obtained by applying a Senograph 2000D molybdenum-rhodium twin target X-ray apparatus (GE, Fair eld, CT, USA). The radiation distance is 20 cm and the radiographing conditions were 22.0 kV, 35.0 mA. For histological evaluation, the specimens were decalci ed in 10% ethylene diamine tetraacetic acid solution (pH 7.4) for 14 d, embedded into para n, then sliced into 5 μm-thick sections before the subsequent hematoxylin and eosin (HE) staining and Masson trichrome staining. The rabbit anti-OCN primary antibodies diluted to 1:100 (Servicebio, Wuhan, Hubei, China; GB11233) were used for immunohistochemical (IHC) staining.

Statistical analysis
Data and statistical analyses were conducted with SPSS Statistics 20.0 software (IBM, Armonk, NY, USA). All data were shown as mean ± standard deviation (SD) of three individual experiments. For the comparison of two or multiple groups, Student's t-test or one-way analysis of variance (ANOVA) combined with Tukey's test were applied, respectively. A two-tailed test with p value < .05 was indicated as statistically signi cant.

MiR-137 reversely regulates hASC differentiation along osteoblastic lineage in vitro
Our previous study displayed an overall downward expression trend of miR-137 in hASCs during the osteoblastic induction and identi ed its negative role in this biological process [32]. Since lacking research on the osteogenic function of miR-137, rstly, we re-veri ed the reliability of our previous results. After transfecting hASCs with lentiviruses of NC, miR-137 overexpression, and miR-137 knockdown (Additional le 1: Fig. S1a), we evaluated the transfection rate was over 90% by computing the percentage of GFP-tagged cells (Additional le 1: Fig. S1b). Meanwhile, the transfection effects were quantitatively determined on 3 d, 7 d and 14 d by qRT-PCR analysis (Additional le 1: Fig. S1c).
ALP staining and activity assays displayed that miR-137 overexpression reduced ALP activity of hASCs under proliferation condition or osteogenic induction, but miR-137 knockdown resulted in adverse consequences (Fig. 1a, b). ARS staining and quanti cation were applied to test the calcium deposits of extracellular matrix. More mineralized nodules were presented in miR-137 knockdown group while less in miR-137 overexpression group when both were compared with NC group (Fig. 1c, d). Besides, the characteristic genes expressed in different stages of osteogenesis, including RUNX2, ALP, and OCN, were examined by qRT-PCR and presented signi cant decreases in miR-137 overexpression group but dramatically increased in miR-137 knockdown group (Fig. 1e). According to these data, we substantiated that miR-137 inhibits in vitro osteoblastic activity of hASCs.
MiR-137 reversely regulates hASC differentiation along osteoblastic lineage in vivo In order to validate in vivo osteogenic effects of miR-137, hASCs were transfected with NC, miR-137, and anti-miR-137 lentiviruses and separately mixed with ACPC, and then the compounds were subcutaneously implanted into the dorsum of nude mice (Fig. 2a). After eight weeks, the total volume and mean density of the harvested samples were assessed and manifested an apparent enhancement in miR-137 knockdown group but remarkable reduction in miR-137 overexpression group (Fig. 2b, c).
Histological analyses of bone formation were performed by staining of HE, Masson trichrome and IHC staining for OCN. HE staining showed more new bone formation in miR-137 knockdown group when comparing with NC group, which displayed only a very small amount of osteoid, but we could hardly observe any new bone or osteoid in miR-137 overexpression group. Similarly, thicker and more compact blue-green stained collagen ber bundles were detected in miR-137 knockdown group than in another two groups, but overexpression of miR-137 led to the thinnest collagen deposition. Moreover, we found that dark-brown stained OCN granules were the most widespread in the cells of miR-137 knockdown group, fewer in NC group, and none could be discerned in miR-137 overexpression group (Fig. 2d). Consequently, in accordance to the results of in vitro experiments, miR-137 subdues the osteoblastic activity of hASCs in vivo.
MiR-137 regulates NOTCH1-HES1 pathway by directly targeting NOTCH1 To ascertain the in uences of miR-137 on NOTCH1 pathway, we rst examined the expression of NOTCH1 and its downstream signal HES1 with miR-137 overexpression or knockdown. When compared with NC group, the mRNA and protein levels of NOTCH1 showed obvious increase in miR-137 knockdown group while marked reduction in miR-137 overexpression group. Contrary to NOTCH1, the expression tendency of HES1 accorded with the changes of miR-137 (Fig. 3a-c). These ndings suggested that NOTCH1 is negatively regulated while HES1 is positively regulated by miR-137.
To further identify whether miR-137 could directly bind to NOTCH1 in hASCs as it does in other cell lines [40][41][42][43][44], dual-luciferase reporter assays were carried out. The presumed targeting sites of miR-137 in the 3' UTR of NOTCH1 were forecasted by two prediction softwares (TargetScan and RNA22). Then the luciferase reporter vectors carrying the 3' UTR of NOTCH1-WT or NOTCH1-MT (Fig. 3d, e) were constructed and the relative luciferase activities were detected. MiR-137 mimics signi cantly repressed the luciferase activity in NOTCH1-WT group while had no signi cant in uences in NOTCH1-MT group when both groups were compared with their respective NC groups (Fig. 3f). These results validated that miR-137 directly binds to the 3' UTR of NOTCH1 and induces the expression of HES1 in hASCs.

NOTCH1 knockdown impairs osteogenesis by inducing HES1 and LSD1 while inhibiting BMP2-SMAD4 pathway
To determine the in uences of NOTCH1 knockdown on the osteoblastic potential of hASCs, we applied ALP and ARS staining combined with quantitative analysis and found that NOTCH1 knockdown attenuated ALP activity and extracellular mineralization (Fig. 4a-d). As the downstream molecules of NOTCH1, HES1 and RUNX2 were further detected at mRNA and protein levels in hASCs transfected with NOTCH1 knockdown lentiviruses. Coincident with the impacts of miR-137 on NOTCH1-HES1 pathway, NOTCH1 knockdown induced the expression of HES1 while repressed RUNX2 (Fig. 4e-g). Our results indicated that NOTCH1 knockdown impedes the osteogenic potential of hASCs by the stimulation of HES1, corroborating the former conclusions that miR-137 inhibits osteogenesis by the downregulation of NOTCH1 and upregulation of HES1.
Since our previous study has demonstrated that miR-137 upregulated LSD1 while downregulated BMP2 and SMAD4 in hASCs [32], we further investigated the in uences of NOTCH1 knockdown on LSD1 and BMP2-SMAD4 pathway. As predicted, both the mRNA and protein expression of LSD1 signi cantly increased after knocking down NOTCH1, whereas BMP2 and SMAD4 decreased apparently (Fig. 4h-j). Therefore, we deduced that the osteogenic inhibition of NOTCH1 knockdown is also dependent on the activation of LSD1 and suppression of BMP2-SMAD4 pathway. Collectively, NOTCH1 knockdown attenuates osteoblastic differentiation by inducing HES1 and LSD1 while inhibiting BMP2-SMAD4 pathway in hASCs.
LSD1 knockdown regulates NOTCH1-HES1 pathway As above, we have a rmed that NOTCH1 acts as a negative regulator in LSD1 expression. But considering the complex interplay of signaling molecules, we tried to clarify whether LSD1 had feedback effects on NOTCH1-HES1 pathway. Notably, we found that LSD1 knockdown led to a higher level of NOTCH1 while lowered the expression of HES1 when compared with NC group (Fig. 5a-c), thus prompting a reciprocal negative relationship between NOTCH1 and LSD1.

BMP2 knockdown inhibits HOTCH1 while induces LSD1
To gain further insights into the relationships between NOTCH1, LSD1 and BMP2 signals, we then examined the expression of NOTCH1 and LSD1 at mRNA and protein levels after knocking down BMP2. Importantly, the expression of NOTCH1 dramatically decreased while LSD1 increased with BMP2 knockdown (Fig. 5d-f). Combining with the above results that NOTCH1 knockdown inhibited BMP2-SMADA4 pathway, we veri ed a positive feedback loop between NOTCH1 and BMP2. Furthermore, our previous study revealed that silencing of LSD1 promoted the osteoblastic potential of hASCs by stimulating BMP2-SMAD4 signaling pathway [32]. Here again, we observed upward tendencies in the expression of BMP2, SMAD4, RUNX2 and ALP with knockdown of LSD1 (Additional le 2: Fig. S2). Therefore, a negative interplay between LSD1 and BMP2 was also con rmed by us.
In conclusion, these data revealed that depending on the reciprocal negative regulation between NOTCH1 and LSD1, LSD1 and BMP2, as well as the synergistic function between NOTCH1 and BMP2, miR-137 negatively regulates osteogenesis of hASCs through the NOTCH1/LSD1/BMP2 co-regulatory signaling network (Fig. 6).
However, little is known concerning its functions and regulatory mechanisms on the osteoblastic differentiation, especially in mesenchymal stem cells. Silencing of miR-137-3p is found to facilitate the osteogenesis of bone marrow-derived mesenchymal stem cells by targeting RUNX2 [50]. Previously, we demonstrated that silencing of miR-137 promoted the osteoblastic activity in hASCs and revealed its modulation on the signaling network of LSD1/BMP2/SMAD4 as part of the mechanisms [32].
Interestingly, our former research con rmed a positive role of miR-137 in LSD1 expression, which was contrary to several studies reporting that miR-137 directly binds to LSD1 [23,26,51,52]. The contradicted outcomes might be associated with the various biological features of different cell types, and we deduced that there probably exist intermediary regulators working between miR-137 and LSD1 during the osteogenesis of hASCs. Strikingly, this study identi ed that NOTCH1 was a direct target of miR-137 in hASCs and NOTCH1-HES1 pathway was engaged in the crosstalk between LSD1 and BMP2-SMAD4 pathway. In this way, NOTCH1 signal mediated the control of miR-137 on LSD1/BMP2/SMAD4 network. Moreover, the interrelations of the above signals were validated comprehensively and a NOTCH1/LSD1/BMP2 co-regulatory network was established, further elucidating the epigenetic mechanisms of miR-137 during the process of hASCs differentiating into osteoblastic lineage.
After recon rming the inhibitory impacts of miR-137 on the osteoblastic activity of hASCs both in vitro and in vivo, we demonstrated that miR-137 negatively regulated the expression of NOTCH1 while positively regulated HES1. NOTCH signaling pathway in uences tumorigenesis as well as embryonic development [53] because of its crucial role in cell fate determination, proliferation, differentiation, and apoptosis [54]. Though it is still debatable whether NOTCH signal plays a positive or negative role in osteogenesis [55], our data displayed impaired osteogenic capacity of hASCs after knocking down NOTCH1. More strikingly, NOTCH1 was validated as a direct target gene of miR-137 in hASCs, the same as in other cell lines [40][41][42][43][44]. HES1 is known as a potential downstream target of NOTCH1 in many studies, but it is not affected in NOTCH1 knockout mice while the expression of HES5, HES-related repressor protein (HERP)1, -2, and − 3 are greatly diminished [56][57][58]. As a transcriptional regulator in the NOTCH signaling pathway, recombination signal binding protein (RBPJ) gene disruption in homozygous mice exhibits reduced HES5 expression, but not for HES1 [56]. Given the various effects on OCN, osteopontin and RUNX2 [37,38], HES1 in uences the osteogenesis inconsistently depending on the different cellular environments. This study showed upregulated HES1 with NOTCH1 knockdown, indicating that NOTCH1 plays as a negative regulator in the expression of HES1. Collectively, we brought insight into how the NOTCH1-HES1 pathway was regulated by miR-137. LSD1 has been linked to the repression of NOTCH1 pathway in various cell types [45,[59][60][61][62][63], though one study states that it functions as a corepressor when associated with RBPJ-repressor complex and as a NOTCH1 coactivator upon NOTCH activation [64]. Nevertheless, few studies have reported the interplay between NOTCH1 and LSD1 during the osteogenesis of hASCs and whether this interaction contributes to the osteogenic regulation of miR-137 is still unknown. Coincident with the in uences of miR-137 on NOTCH1 and LSD1, we uncovered a negative interaction between NOTCH1 and LSD1 with separate knockdown of them. More intriguingly, despite inducing NOTCH1, LSD1 knockdown reduced the expression of HES1. Thus, the opposite expression trends of NOTCH1 and HES1 caused by LSD1 knockdown might reinforce the downregulation of HES1 by NOTCH1 alone. These results veri ed the crosstalk between NOTCH1-HES1 pathway and LSD1, through which miR-137 regulates the osteogenic differentiation of hASCs.
BMP signal is a canonical pathway in skeleton and BMP2-SMAD4 pathway has been shown to participate in the osteogenic regulation of miR-137 by us [32]. After knocking down NOTCH1 and BMP2 individually, we observed suppressed expression of NOTCH1, BMP2 and SMAD4 simultaneously, indicating a positive interrelationship between NOTCH1 and BMP2-SMAD4 pathway. Considering our previous results that LSD1 knockdown activated the BMP2-SMADA4 pathway [32], in turn, we investigated the impacts of BMP2 on LSD1. As expected, LSD1 was signi cantly upregulated with BMP2 knockdown. Thus, these ndings suggested a negative feedback loop between LSD1 and BMP2-SMADA4 pathway.

Conclusions
In summary, our study provided a relatively comprehensive rationale for the negative control of miR-137 during the hASC differentiation towards osteoblastic lineage and established a co-regulatory network of NOTCH1/LSD1/BMP2 to elucidate the underlying mechanisms, which is of substantial importance for potential targeted therapy of bone-related diseases.

Consent for publication
Not applicable.

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

Competing interests
The authors declare that they have no competing interests.

Funding
This study was nancially granted by grants from the National Natural Science Foundation of China (No. Authors' contributions CF conceived the project, provided nancial supports, and wrote the manuscript. XM performed the main experiments, analyzed data and prepared gures. YW, LL, YZ, and HL helped with the animal work. YL assisted with the data analysis and offered insightful ideas.      A diagrammatic view of NOTCH1/LSD1/BMP2 network in miR-137-mediated differentiation of hASCs towards osteoblastic lineage. In this complex network model, miR-137 controls NOTCH1-HES1, LSD1, and BMP2-SMAD4 pathways simultaneously. Depending on the positive feedback loop between NOTCH1 and BMP2, as well as the negative reciprocal relationship between LSD1 and NOTCH1 or BMP2, the impacts of miR-137 on the above three signaling pathways are strengthened and the NOTCH1/LSD1/BMP2