Molecular control of nitric oxide synthesis through eNOS and caveolin-1 interaction regulates osteogenic differentiation of adipose-derived stem cells by modulation of Wnt/β-catenin signaling

Background Nitric oxide (NO) plays a role in a number of physiological processes including stem cell differentiation and osteogenesis. Endothelial nitric oxide synthase (eNOS), one of three NO-producing enzymes, is located in a close conformation with the caveolin-1 (CAV-1WT) membrane protein which is inhibitory to NO production. Modification of this interaction through mutation of the caveolin scaffold domain can increase NO release. In this study, we genetically modified equine adipose-derived stem cells (eASCs) with eNOS, CAV-1WT, and a CAV-1F92A (CAV-1WT mutant) and assessed NO-mediated osteogenic differentiation and the relationship with the Wnt signaling pathway. Methods NO production was enhanced by lentiviral vector co-delivery of eNOS and CAV-1F92A to eASCs, and osteogenesis and Wnt signaling was assessed by gene expression analysis and activity of a novel Runx2-GFP reporter. Cells were also exposed to a NO donor (NONOate) and the eNOS inhibitor, l-NAME. Results NO production as measured by nitrite was significantly increased in eNOS and CAV-1F92A transduced eASCs +(5.59 ± 0.22 μM) compared to eNOS alone (4.81 ± 0.59 μM) and un-transduced control cells (0.91 ± 0.23 μM) (p < 0.05). During osteogenic differentiation, higher NO correlated with increased calcium deposition, Runx2, and alkaline phosphatase (ALP) gene expression and the activity of a Runx2-eGFP reporter. Co-expression of eNOS and CAV-1WT transgenes resulted in lower NO production. Canonical Wnt signaling pathway-associated Wnt3a and Wnt8a gene expressions were increased in eNOS-CAV-1F92A cells undergoing osteogenesis whilst non-canonical Wnt5a was decreased and similar results were seen with NONOate treatment. Treatment of osteogenic cultures with 2 mM l-NAME resulted in reduced Runx2, ALP, and Wnt3a expressions, whilst Wnt5a expression was increased in eNOS-delivered cells. Co-transduction of eASCs with a Wnt pathway responsive lenti-TCF/LEF-dGFP reporter only showed activity in osteogenic cultures co-transduced with a doxycycline inducible eNOS. Lentiviral vector expression of canonical Wnt3a and non-canonical Wnt5a in eASCs was associated with induced and suppressed osteogenic differentiation, respectively, whilst treatment of eNOS-osteogenic cells with the Wnt inhibitor Dkk-1 significantly reduced expressions of Runx2 and ALP. Conclusions This study identifies NO as a regulator of canonical Wnt/β-catenin signaling to promote osteogenesis in eASCs which may contribute to novel bone regeneration strategies. Electronic supplementary material The online version of this article (doi:10.1186/s13287-016-0442-9) contains supplementary material, which is available to authorized users.

Nitric oxide (NO) is a signaling molecule with a short half-life [19,20]. It can react within the cell where it is produced or penetrate cell membranes to affect adjacent cells [21]. NO exerts a variety of physiological effects such as regulating blood pressure via smooth muscle relaxation [22], mediating immune responses [23], controlling cell proliferation [24], modulating apoptosis [20], promoting growth factor-induced angiogenesis [4,21], accelerating wound healing [4,25], and functioning as a neurotransmitter [26]. These responses can be mediated through activating the primary NO effector soluble guanylyl cyclase to produce cGMP [27] by NO-based chemical modifications of proteins through S-nitrosylation [28] or through epigenetic modification [29]. NO is known to play an important role in bone homeostasis. It is generated by many cell types present in the bone environment, most notably the osteoblast [30].
NO is synthesized from L-arginine by three isozymes of nitric oxide synthase (NOS), including neuronal NOS (nNOS), endothelial NOS (eNOS), and cytokine-inducible NOS (iNOS) [31]. Both iNOS [32,33] and eNOS [34] have been shown to play a role in osteoblast differentiation. Mice lacking eNOS have shown marked bone abnormalities due to impaired osteoblast differentiation resulting in poor maintenance of bone mass [35,36]. Gene expression data from neonatal calvarial osteoblasts from eNOS -/mice have shown downregulation of Runx2, Cbfa-1, and osteocalcin [37]. On the other hand, high concentrations of NO released due to the pathological iNOS expression promote bone resorption through induced osteoclastogenesis [38]. Therefore, an optimum level of NO is important to drive osteogenic differentiation of the MSCs.
In contrast with other NOS family members, eNOS is localized mainly in specific intracellular membrane domains, including the Golgi apparatus [39] and plasma membrane caveolae [40,41]. A previously demonstrated direct interaction of eNOS with wild-type caveolin-1 (CAV-1 WT ) [42] has proposed that CAV-1 WT functions as an endogenous negative regulator of eNOS [43]. In this context, eNOS binds to the caveolin-1 scaffolding domain (CSD; amino acids 82-101) [44] and, furthermore, Thr-90 and Thr-91 (T90 and T91), and Phe-92 (F92) were identified as critical residues for eNOS binding and inhibition [41]. Genetic modification of endothelial cells through overexpression of a mutated version of CAV-1 with a phenylalanine to alanine substitution at the amino acid position 92 (CAV-1 F92A ) resulted in increased NO production, overcoming the inhibitory effect of CAV-1 WT [41].
In the present study, we tested the hypothesis that molecular control of NO synthesis in equine adiposederived stem cells (eASCs), can promote osteogenic differentiation where endogenous eNOS is not available, by recreating the interaction between eNOS and CAV-1 (CAV-1 WT and CAV-1 F92A ) regulates the osteogenic differentiation of eASCs. Our results indicate that the optimum level of NO induces osteogenic differentiation through activation of the downstream canonical Wnt/βcatenin signaling pathway.

Plasmid constructs
The list of cDNA for the genes of interest used in this study is listed in Table 1, and was used for construction of lentiviral vectors. HF Phusion (New England Biolabs; NEB) DNA polymerase was used for all the polymerase chain reactions (PCRs) and all the restriction endonucleases were purchased from NEB unless indicated otherwise.
To construct the CMV promoter-driven eNOS expressing lentiviral vector, a codon optimized eNOS gene was synthesized [4] and subcloned into the pWPT-GFP lentiviral plasmid (Addgene, MA, USA) using BamH1 and Sal1 restriction endonucleases. Doxycycline (DOX) inducible eNOS construct was prepared by amplifying the eNOS gene using the forward (ATCAGAATTCATGGG-CAACCTGAA) and reverse primer (ATCAGAATTCT-CATCAGGGGCTGT) by introducing EcoR1 restriction sites (underlined sequences in both forward and reverse primers) at both the 5' and 3' ends of the final PCR product, followed by subcloning the PCR product into FUW-TetO vector (Addgene). Human wild-type caveolin-1 (CAV-1 WT ) expressing lentiviral vector was constructed by inserting the full length CAV-1 WT (Addgene) into pLVX-AcGFP1-C1 and pLVX-DsRed-C1 (Clontech, CA, USA) using EcoR1 and BamH1 restriction endonucleases. A mutated caveolin-1 (CAV-1 F92A ) in which phenylalanine (F) at the amino acid position 92 was replaced with alanine (A) [41] was synthesized (Geneart), amplified by PCR by introducing BamH1 and Sal1 sites at the 5' and 3' ends of the PCR product, respectively, via forward primer (ATCAGGATCCATGTCTGGGGGCA) and reverse primer (ATCAGTCGACTTATATTTCTTTCTG) (restriction sites are underlined). The PCR product was then ligated into the pWPT-GFP lentiviral vector (Addgene) at the BamH1 and Sal1 restriction sites replacing GFP. Wnt3a and Wnt5a expressing lentiviral plasmids were purchased from DNAsu plasmid repository.

GFP reporter constructs
A 343-bp fragment of the Runx2 enhancer region (sequence information was kindly provided by Toshihisa Komori at the Department of Cell Biology, Nagasaki University) [46] was synthesized together with the sequence of the Hsp68 minimal promoter (GenScript, NJ, USA). The entire fragment was then subcloned into pTRIP-eGFP lentiviral vector at the Mlu1 and BamH1 restriction sites replacing an insulinspecific promoter, upstream of the enhanced GFP (eGFP) coding sequence. The Wnt responsive lentiviral TCF/LEF-dGFP reporter system was purchased from Addgene.

Lentiviral vector production and transduction of equine adipose stem cells
Lentiviral vectors used in this study (Table 1) were generated by four plasmid transfection of HEK293T cells. Briefly, each well of a six-well tissue culture plate was coated with 50 μg/mL of DL-lysine (Sigma-Aldrich) in phosphatebuffered saline (PBS) and incubated for 2 h at 37°C. HEK293T cells were then seeded at a density of 1 × 10 6 cells per well, 24 h prior to transfection of 6.3 μg of packaging plasmid psPAX2 (Addgene), 3.1 μg of Rev expression plasmid pRSV Rev (Addgene), 3.5 μg of VSV-G envelop pMD2.G (Addgene), and 10 μg of the gene of interest expression transfer vector using a standard calcium phosphate transfection method [47]. Seventeen hours posttransfection, the media was changed and supernatant containing lentiviral vectors were collected at 48 h and 72 h post-transfection, combined, and filtered through a 0.45-μM PVDF filter, and used for eASC transduction in the presence of 4 μg/mL Polybrene (Sigma-Aldrich).
After 11 days incubation in OM or control growth medium, cells were washed with PBS and fixed with 4% (w/v) paraformaldehyde (Sigma-Aldrich) for 20 min, washed with distilled water, and then stained with 2% (w/v) Alizarin Red S (pH 4.2) for 20 min. Stained cells were washed with distilled water prior to assessment by light microscopy using a Nikon Eclipse Ti-S inverted microscope (Nikon, Japan).

Alizarin Red S quantification
Quantification of Alizarin Red S staining was performed as previously described [48]. Briefly, after staining the cells with Alizarin Red S for 20 min, 10% acetic acid was added to the 12-well cell culture plate and incubated for 30 min with shaking. The Alizarin Red S stain was extracted and the absorbance was measured at 405 nm in parallel with Alizarin Red S standards comprising of serial 1:2 dilutions of 50 mM Alizarin Red S (pH 4.2).

Quantitative real-time PCR
Total RNA from transduced and control cells after 11 days of incubation in OM or growth medium was isolated using the PureZol reagent (Bio-Rad, CA, USA) according to the manufacturer's instructions, and the concentration of isolated RNA was determined using a Nanodrop spectrophotometer (Thermo Fisher Scientific), treated with RQ1 RNase free DNase (1 U/1 μg RNA; Promega, WI, USA). cDNA was synthesized with 1 μg RNA from all samples using a High Capacity Reverse Transcription Kit (Thermo Fisher Scientific). Quantitative real-time PCR assays were performed on a BioRad CFX96 Real-Time system (Bio-Rad) using the SsoFast EvaGreen Supermix (Bio-Rad). Primer sequences used for target gene amplification are described in Table 2. Assays were performed in triplicate and target gene expression was normalized to equine β-actin mRNA levels using the ΔΔC t method.

Immunocytochemistry and confocal microscopy
Immunocytochemical detection of eNOS and caveolin-1 (CAV-1 WT and CAV-1 F92A ) expression in eASCs was performed as follows. Briefly, cells were fixed in 4% paraformaldehyde for 20 min at 37°C, treated with 0.1% Triton-X100 in PBS for 10 min, and blocked in a 10% FBS in PBS solution for 30 min at room temperature. This was followed by a 2-h incubation with a primary mouse monoclonal anti-eNOS antibody (BD Biosciences, CA, USA) or rabbit polyclonal anti-CAV-1 antibody (Cell Signaling Technology, MA, USA), and subsequently with an anti-mouse IgG secondary antibody conjugated with Alexa 488 (Cell Signaling Technology) or anti-rabbit IgG secondary antibody conjugated with Alexa 555 (Cell Signaling Technology) for 1 h and counterstained with DAPI for nuclear staining (Sigma-Aldrich). eNOS and CAV-1 co-localization was observed by confocal microscopy (Nikon).
To detect β-catenin expression, eNOS transduced cells (with or without DOX treatment) and un-transduced cells were fixed in 4% paraformaldehyde for 20 min at 37°C, treated with 0.1% Triton-X100 in PBS for 10 min, and blocked in a 10% FBS in PBS solution for 30 min at room temperature. This was followed by an overnight incubation with a primary rabbit monoclonal anti-βcatenin antibody (Cell Signaling Technology) and subsequently with an anti-rabbit IgG secondary antibody conjugated with Alexa 488 (Cell Signaling Technology) for 1 h and counterstained with DAPI.

Nitric oxide detection
Extracellular NO production was measured using the Griess reagent (Promega) according to the manufacturer's instructions and measurement of absorbance at 540 nm. Triplicates of each sample were measured at each time-point during osteogenic differentiation from day 0 to day 11.

Statistical analysis
All experiments were performed in triplicate and at least three times. Data are presented as mean ± SEM. The statistical significances were determined by one-way analysis of variance (ANOVA) followed by Tukey's test. All tests were performed using the statistical software GraphPad Prism 6 (GraphPad, CA, USA). p < 0.05 was considered statistically significant.

eASC characterization
eASCs were spindle-shaped and adherent to plastic tissue culture dishes (data not shown). We have previously reported their tri-linage differentiation potentials [49].

NO enhances osteogenic differentiation
To examine the role of the NO signaling in eASCs osteogenesis, we first compared the osteogenic differentiation between eNOS transduced (eASC eNOS ) and untransduced eASCs (eASC WT ). A greater number of Alizarin Red S-positive nodules were induced in the eASC e-NOS cultures compared to eASC WT culture after 11 days (Fig. 2b). NO synthesis was also significantly increased in eASC eNOS compared to eASC WT (Fig. 2c). Quantification of calcium deposition showed increased levels of calcium deposition in eASC eNOS compared to eASC WT (Fig. 2d). Quantitative analysis of Runx2 (Fig. 2e) and ALP (Fig. 2f) gene expression were also significantly upregulated in the eASC eNOS cultures compared to the eASC WT cultures. NO-mediated osteogenic differentiation was further highlighted by inhibition of eNOS activity. eASC eNOS were treated with 2 mM of the nitric oxide synthase inhibitor, L-N G -nitroarginine methyl ester (L-NAME) for 11 days. L-NAME treatment resulted in a significant downregulation of osteoblast-specific marker expressions, ALP (Fig. 3a) and Runx2 (Fig. 3b), compared to untreated eASC eNOS .

Exogenous NO donor enhances osteogenesis of eASC in a dose-dependent manner
To confirm the direct role of NO levels on eASC osteogenesis, we treated eASCs with a concentration range of exogenous NO donor (NONOate; Sigma-Aldrich). Treatment with exogenous NO donor promoted osteogenesis from 5 μM to 15 μM but this was reduced with high concentrations of NO donor (20 μM) as evident by Alizarin Red staining (Fig. 4a). NONOate treatment also resulted in a dose-dependent increase in Runx2 (Fig. 4b) and ALP (Fig. 4c) gene expression, in which maximum levels of both were achieved with 15 μM NONOate; notably, significantly lower levels of Runx2 and ALP expression were observed with 20 μM NO donor treatment. When e eNOS, f mutated caveolin-1 (CAV-1 F92A ), and g CAV-1 WT were transduced to eASCs using lentiviral vectors, eNOS showed cytoplasm localization (red) whereas both the CAV-1 F92A and CAV-1 WT showed plasma membrane localization (green). h Confocal microscopy analysis of co-transduction of eASCs with eNOS and CAV-1 F92A resulted in cytoplasmic eNOS expression (red) and membrane localization of CAV-1 F92A (green)

NO promotes endogenous Runx2 expression in differentiating eASCs
To monitor the effect of NO on endogenous Runx2 expression in differentiating eASCs, we generated a GFP lentiviral reporter system under the control of a Runx2 enhancer fused to the Hsp68 promoter (Runx2.Hsp68-eGFP; Fig. 5a) based on a novel Runx2 enhancer. eASCs which were stably transduced with the Runx2 reporter showed low levels of GFP expression in mostly undifferentiated cells (Fig. 5b), whereas GFP expression as a result of Runx2 promoter activity in differentiating eASCs was increased (Fig. 5b). When eASCs were transduced with eNOS (eASC eNOS ), the GFP signals were increased compared to eASCs co-transduced with eNOS and CAV-1 WT (eASC eNOS+CAV-1WT ) and un-transduced control (eASC WT ) (Fig. 5b). Interestingly, we observed that endogenous Runx2 activity was remarkably increased when the eASCs were co-transduced with eNOS and CAV-1 F92A (eASC eNOS+CAV-1F92A ) (Fig. 5b) compared to eASC eNOS , eASC eNOS+CAV-1WT , and eASC WT . Osteogenic nodule Fig. 3 Inhibition of endothelial nitric oxide synthase (eNOS) through L-N G -nitroarginine methyl ester (L-NAME) treatment downregulates eASC osteogenic differentiation. Relative mRNA transcripts analysis by qPCR showing that 2 mM L-NAME treatment decreased the expression of osteoblast markers a alkaline phosphatase (ALP) and b Runx2 in eNOS transduced eASCs compared to untreated eNOS transduced cells. *p < 0.05, versus eNOS (L-NAME), eASC (osteogenic induction), and eASC (growth medium) Fig. 4 The exogenous nitric oxide donor NONOate promotes osteogenic differentiation in a dose-dependent manner. a Alizarin Red S staining of eASCs after 11 days in osteogenic induction medium or growth medium. Relative mRNA transcript analysis shows a significant increase of osteoblast markers b Runx2 and c alkaline phosphatase (ALP) when the eASCs were treated with 15 μM NONOate. *p < 0.05, versus 0 μM, 5 μM, 10 μM, and 20 μM NONOate, and control (growth medium) formation was significantly increased in the eASC eNOS +CAV-1F92A (Fig. 5c), and these result suggests that endogenous Runx2 expression is less active in undifferentiated eASCs and its expression is significantly increased through NO signaling during osteogenic differentiation.

NO modulates Wnt signaling to promote osteogenic differentiation
To examine the role of canonical and non-canonical Wnt signaling during NO-mediated osteogenic differentiation, expression of Wnt3a, Wnt8a, and Wnt5a was assessed by quantitative real-time PCR. Non-canonical Wnt5a expression was reduced in eASC eNOS (Fig. 6c), and was significantly further decreased in eASC eNOS+CAVF92A (Fig. 6c). However, expression of canonical Wnt ligands Wnt3a (Fig. 6a) and Wnt8a (Fig. 6b) was upregulated in eASC eNOS and significantly further increased in eASC eNOS+CAVF92A ( Fig. 6a and b, respectively). Treatment with 2 mM L-NAME showed downregulation of Wnt3a expression (Fig. 6d) and upregulation of Wnt5a (Fig. 6e) in eASC eNOS , indicating that NO modulates Wnt signaling pathway in eASCs.
Furthermore, treatment with NO donor (NONOate) also resulted in increased expression of canonical Wnt ligands Wnt3a (Fig. 7a) and Wnt8a (Fig. 7b), and downregulation of non-canonical Wnt5a expression (Fig 7c) in a dose-dependent manner from 5 μM to 15 μM of NON-Oate. Interestingly, when the NO donor concentration was increased up to 20 μM, the effect was completely reversed by downregulating Wnt3a (Fig. 7a) and Wnt8a (Fig. 7b), and upregulating Wnt5a expression (Fig. 7c). Control cells (eASCs in normal growth medium) also showed increased expression of Wnt5a (Fig. 7c), suggesting that induction of osteogenic differentiation of eASCs requires activation of canonical Wnt signaling and suppression of non-canonical Wnt5a expression.
To further analyze the relationship between NO-induced osteogenic differentiation and Wnt signaling, eNOStransduced eASCs were treated with 20 ng/mL of the Wnt signaling inhibitor, Dickkopf-related protein 1 (Dkk-1). Dkk-1 treatment resulted in a significant downregulation of the osteoblast specific markers ALP (Fig. 8a) and Runx2 (Fig. 8b) compared to untreated eNOS transduced cells.

NO promotes the canonical Wnt signaling pathway by promoting nuclear translocation of β-catenin
To further explore mechanisms by which NO promotes canonical Wnt signaling, we used a lentiviral vector expressing GFP reporter under the control of the TCF/LEF promoter (TOPFLASH; Addgene). In the canonical Wnt signaling pathway, β-catenin translocation to the nucleus is promoted by the activation of canonical Wnt signaling [52,53]. Accordingly, eASCs were introduced with the lentiviral TCF/LEF-dGFP reporter, and those cells were then co-transduced with doxycycline inducible eNOS expressing lentiviral vector (Fig. 10a). As a readout for nuclear translocation of β-catenin, the TCF/LEF-driven GFP mRNA expression levels were measured by quantitative real-time PCR. Under osteogenic induction conditions, increased GFP mRNA expression was demonstrated compared to non-osteogenic induction conditions (cells in growth medium) (Fig. 10b). When doxycycline was added to the OM, GFP mRNA expression was significantly increased in the eNOStransduced cells (eASC eNOS ) (Fig. 10b). Interestingly, when doxycycline was removed from the medium, GFP mRNA expression in eASC eNOS was similar to other osteogenic induction conditions (Fig. 10b).
Using a β-catenin-specific monoclonal antibody, we further investigated the effect of NO on β-catenin nuclear translocation. eASCs were transduced with the doxycycline Fig. 8 Inhibition of Wnt signaling through Dickkopf-related protein 1 (Dkk-1) downregulates eASC osteogenic differentiation. Relative mRNA transcript analysis by qPCR shows that Dkk-1 treatment (20 ng/mL) decreased the expression of osteoblast markers a alkaline phosphatase (ALP) and b Runx2 in endothelial nitric oxide synthase (eNOS)-transduced eASCs compared to untreated eNOS-transduced eASCs. *p < 0.05, versus eNOS (Dkk-1), eASC (osteogenic induction with Dkk-1), eASC (osteogenic induction), and eASC (growth medium) Fig. 9 Lentiviral expression of canonical Wnt3a promotes osteogenesis and non-canonical Wnt5a results in suppressed osteogenesiss. a Alizarin Red S staining after 11 days in osteogenic induction medium or growth medium showing that lentiviral Wnt3a transduction increased the calcium deposition, whereas Wnt5a decreased calcium deposition levels. Relative mRNA transcript analysis by qPCR showing that Wnt3a upregulates the expression of osteoblast markers b alkaline phosphatase (ALP) and c Runx2 expression, whereas Wnt5a resulted in downregulation. *p < 0.05, versus Wnt5a, eASC (osteogenic induction), and eASC (growth medium) inducible eNOS lentiviral vector followed by immunostaining with β-catenin-specific monoclonal antibody (Cell Signaling Technology). When doxycycline was added to the OM, the expression of β-catenin was observed in eNOStransduced cells (eASC eNOS ) (Fig. 11a) in both the nucleus and cytoplasm, and when doxycycline was removed from the medium β-catenin expression in eASC eNOS was reduced (Fig. 11a) to that seen in un-transduced control cells (Fig. 11a). Furthermore, we observed nuclear co-localization of β-catenin and DAPI only in doxycycline-treated eASC eNOS suggesting that NO may promote nuclear localization of βcatenin (Fig. 11b). As a control for the primary antibody, immunostaining was carried out in the absence of the primary antibody specific to β-catenin (Additional file 2: Figure S2).
Together, these findings support the paradigm that cellular environments rich in bioavailable NO through either genetic modification or exogenous sources can modulate Wnt signaling, by upregulating the canonical and downregulating the non-canonical pathways resulting in increased osteogenic differentiation (Fig. 12).

Discussion
NO plays an important role in osteogensis, bone remodeling, and metabolism [54][55][56]. It has been reported that both iNOS and eNOS play a role in osteogenesis of embryonic stem cells [57]. We [4] and others [58] have shown that MSCs do not express eNOS. Therefore, in order to investigate the role of eNOS in osteogenic differentiation of eASCs, in this study eASCs were genetically modified by lentiviral vector-based eNOS. ASCs are promising candidates for stem cell-based therapy for bone repair [59], and the role of eNOS-mediated NO synthesis and its downstream effect on osteogenesis of MSCs remains to be explored. We found that eNOS gene transfer by lentiviral vector promoted osteoblast-specific gene expressions ( Fig. 2e and f), contributing to the matrix mineralization as visualized by Alizarin Red S staining ( Fig. 2b and d). Noteworthy, this osteogenic potential of eASCs eNOS was significantly abrogated by L-NAME treatment (Fig. 3), suggesting that NO derived from eNOS plays a major role in enhancing osteogenesis in eASCs.
CAV-1 is a key negative regulator of eNOS activation and thus inhibits the production of NO [41,60] and, importantly, CAV-1 is expressed endogenously in MSCs [61]. The scaffolding domain (82-101 amino acids) of CAV-1 protein interacts with eNOS at the plasma membrane and this interaction inhibits the eNOS activation reducing NO synthesis [41]. An alanine scanning approach revealed that substitution of phenylalanine at the amino acid position 92 with alanine to produce CAV-1 F92A mutant restored the eNOS  Nitric oxide promotes nuclear translocation of β-catenin. a Immunostaining with a β-catenin-specific monoclonal antibody reveals that the expression of β-catenin in endothelial nitric oxide synthase (eNOS) transduced cells when doxycycline (DOX) is available in the medium. b nuclear localization of beta catenin in eNOS transduced cells in DOX containing medium. Arrows indicate nuclear localisation of β-catenin activation and promoted NO synthesis [41]. Thus, in order to understand the contribution of caveolin-1 on the control of NO synthesis in eASC osteogenesis, we modified eASCs by expressing CAV-1 WT (as a negative regulator for eNOS activation) or CAV-1 F92A (as a positive regulator for eNOS activation) together with eNOS. Confirming a previous observation [41], we found that co-expression of eNOS and CAV-1 F92A increased NO production while eNOS and CAV-1 WT co-expressed eASCs showed reduced NO production (Fig. 2c), suggesting that CAV-1 is an important regulator of NO production in eASCs. We further found that these controlled levels of NO synthesis regulate osteogenesis, where eNOS together with CAV-1 F92A resulted in increased osteogenic differentiation of eASCs.
To explore the molecular basis of NO-mediated osteogenesis, we investigated the effect of NO on Wnt signaling. Wnt signaling pathways have been shown to regulate osteoblastogenesis [62], in which canonical Wnt ligands promote osteogenesis [63,64], and non-canonical Wnt5a can inhibit the canonical Wnt signaling [65]. In the canonical Wnt pathway, binding of canonical Wnt ligands such as Wnt3a and Wnt8a to cell surface frizzled receptors results in the nuclear translocation of β-catenin [66], which ultimately binds with the TCF/LEF region to initiate the transcription of osteogenic genes such as Runx2 [62]. On the other hand, binding of non-canonical Wnt5a ligand to the ROR2 member of the Ror-family of RTKs inhibits canonical Wnt signaling by promoting β-catenin degradation, and downregulation of β-catenin reduced osteoblast-specific gene expression [67]. Our results revealed that genetic manipulation of eASCs with eNOS and CAV-1 F92A (eASC eNOS+CAV-1F92A ) increased canonical Wnt3a and Wnt8a expression, whereas eASC eNOS+CAV-1WT decreased Wnt3a and Wnt8a expression ( Fig. 6a and  b), suggesting that NO levels may regulate Wnt ligand expression and promote osteogenesis. Confirming the role of Wnt signaling on osteogenesis, inhibition of canonical Wnt signaling through Dkk-1 treatment of eNOSexpressing cells attenuated osteogenesis as evident by downregulation of osteoblast-specific Runx2 and ALP expression (Fig. 8). On the other hand, the effect of noncanonical Wnt5a expression was completely the opposite (Fig. 6c) to the canonical Wnt3a and Wnt8a expression profiles, suggesting that molecular control of NO synthesis through eNOS/CAV-1 interaction or exogenous NO treatment (Fig. 7) results in differential regulation of Wnt ligand expression and their subsequent effect on osteogenic differentiation. Furthermore, we also found that NO modulates Wnt signaling and promotes osteogenesis when a differentiation environment is enhanced with an optimum concentration of exogenous NO (Figs. 4 and 7).
It has been shown that Wnt3a can directly promote osteogenesis [68], whilst Wnt5a plays a role in selfrenewal of stem cells [69]. We further investigated the direct effect of canonical Wnt3a and non-canonical Wnt5a on eASC osteogenesis through lentiviral vector overexpression. Interestingly, corroborating our results on NO-mediated Wnt-regulated osteogenesis, Wnt3a promoted osteogenesis (Fig. 9a-c) whereas Wnt5a inhibited osteogenesis (Fig. 9a-c). It was shown that increased levels of β-catenin can promote bone formation through increasing the expression of osteoblast-specific genes [70,71], whilst abnormal osteoblast differentiation has been observed with β-catenin knockdown [70,72]. Thus, it is possible that Wnt3a promotes osteogenesis by increasing β-catenin stability and Wnt5a may suppress osteogenesis by degrading β-catenin. NO may regulate this mechanism by increasing Wnt3a and suppressing Wnt5a ligand availability to modulate nuclear localization of β-catenin via the canonical Wnt ligand transduction pathway. In support of this, we observed that eNOStransduced cells promoted the expression of β-catenin and its nuclear localization (Fig. 11), and a TCF/LEF-dGFP reporter assay demonstrated responsiveness in a NO-rich cellular environment (Fig. 10b), which could be controlled through the expression of DOX-inducible eNOS.

Conclusions
In summary, our findings provide an insight into the role of NO in promoting eASC osteogenic differentiation in a cellular environment of optimum levels of NO through interaction with Wnt signaling pathways. This may lead to the development of novel cell-based therapeutic approaches for bone repair, in particular in vitro modification of MSCs by NO to optimize the endogenous Wnt signaling pathway to promote osteogenic differentiation upon subsequent transplantation.

Additional files
Additional file 1: Figure S1. Primary antibody controls for Fig. 1. As a control for specific primary antibody binding, immunostaining was performed in the absence of primary antibodies, with only fluorophore tagged secondary antibody present. We confirm no florescence signal was detected these controls for eNOS and CAV-1. Additional file 2: Figure S2. Primary antibody controls for Fig. 11. As a control for specific β-catenin primary antibody binding, immunostaining was carried out in the absence of the primary antibody with no florescence signal detected. (a) Detection of nuclear staining by Dapi and (b) fluorescence detection without rabbit monoclonal anti-β-catenin antibody with an antimouse IgG secondary antibody conjugated with Alexa 488. (PPT 102 kb)