BMPER Enhances Bone Formation by Promoting the Osteogenesis-Angiogenesis Coupling Process in Mesenchymal Stem Cells Open Access

Bone repair following fracture is usually a rapid and precise process. After a fracture, bone mesenchymal stem cells (BMSCs) migrate to the injured site and gradually differentiate into osteoblasts, which produce bone matrix and repair the fractured bone [1]. While multiple signaling pathways are involved in regulating BMSC differentiation, bone morphogenetic proteins (BMPs) play a critical role in inducing the differentiation of BMSCs of the osteoblastic lineage and enhancing the differentiated function of osteoblasts [2, 3]. BMP-2, BMP-4, BMP-6, BMP-7 and BMP-9 induce osteogenic differentiation of MSCs in vitro and in vivo [4, 5]. By binding type I and II receptors to form complexes on the cell surface, BMPs exert their biological function through the activation of Smad signaling pathways, which regulate osteogenic-related gene expression [6, 7]. Meanwhile, BMPs are functionally modulated by a class of intracellular and extracellular BMP-binding proteins, termed BMP antagonists [2, 8]. To date, noggin, gremlin, chordin, follistatin, and sclerostin have been identified as osteoblast products that bind BMPs [8, 9]. Some evidence exists that BMP antagonists, such as gremlin-1 and gremlin-2, limit BMP activity during osteogenesis, and the BMP-2-induced osteogenesis of BMSCs can be markedly enhanced by suppression of gremlin-1 and gremlin-2 expression [4, 5]. However, some ‘BMP antagonists’, such as noggin, play varying roles during osteogenesis in different species. Noggin inhibits osteogenesis by preventing BMPs from binding their receptors on the cell surface in some animal models, such as mice [10, 11], and enhances osteogenesis by inducing BMP-2 and osteocalcin in human BMSCs (hBMSCs) [1, 12].

Another ‘BMP antagonist’, BMP-binding endothelial cell precursor-derived regulator (BMPER;also known as Crossveinless 2 [CV2]) exerts both stimulatory and antagonistic effects on BMP actions in different contexts [13, 14]. BMPER binds and modulates at least three BMPs (BMP-2, BMP-4 and BMP-6) and was originally discovered in a screen for differentially expressed proteins in Drosophila embryonic endothelial precursor cells [15]. Loss-of-function models demonstrated that BMPER acts as a pro-BMP modulator most likely by facilitating the binding of BMPs to their respective receptors [13, 16]. A number of studies have clarified the function of BMPER in the vasculature, and research focused on the skeletal system is increasing. Bmper knockout mice show defects in vertebral and cartilage development and renal hypoplasia [17]. Furthermore, mutations in the BMPER gene have been reported in diaphanospondylodysostosis and ischiospinal dysostyosis, both of which result in the skeletal dysostosis phenotype [18, 19]. As a regulator of the BMP signaling pathway, BMPER plays a critical role in vertebral and cartilage development. However, the role of BMPER in hBMSC differentiation is still poorly understood.

Angiogenesis and osteogenesis are two intimately, well-coordinated coupling processes during bone healing [20]. Interactions between MSCs and endothelial cells (ECs) are mediated by numerous angiogenic factors, growth factors and cytokines and have been discovered in the MSC secretome [21]. Conversely, blood vessels contribute to the process of osteogenesis during bone repair [22, 23]. In addition, these processes are also very important for the regulation of endochondral bone growth. Vascular invasion of the primarily avascular hypertrophic zone of the growth plate brings osteoblasts and endothelial precursor cells into future centers of ossification [24]. Patients with skeletal dysplasia ischiospinal dysostosis caused by inactivated BMPER mutations suffer not only from vertebral and rib malformations but also from significantly short statures [18]. Furthermore, the lack of Bmper translates into apparent endothelial thickening, and an increasing number of immature ECs have been observed in Bmper-deficient mice [16, 25]. Thus, we hypothesize that BMPER may be involved in the angiogenesis and osteogenesis processes and even in the coupling process.

In the present study, BMP-2 upregulates BMPER expression during hBMSC osteogenic differentiation. Overexpression of BMPER by lentiviruses enhances the osteogenic capability of hBMSCs, while suppression of BMPER exerts the opposite effect. Furthermore, BMPER accelerates bone formation by promoting the osteogenesis-angiogenesis coupling process in ectopic osteogenesis and monocortical defect models. Thus, we provide novel mechanistic insights into hBMSC osteogenesis and show that BMPER is a potential therapeutic target for bone repair in clinical practice.

This experiment was approved by the Institutional Ethics Committee of Xin Hua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. hBMSCs were donated by 3 male and 3 female patients aged 40.7 years (range, 35-47 years) who accepted traumatic femoral shaft fracture treatment by intramedullary nailing. Informed consent was obtained from all donors. None of the patients presented with osteoporosis, other orthopedic diseases or systemic diseases. Bone marrow blood aspirated during reaming was filtered through a 100-µm nylon mesh cell strainer to remove trabecular bone debris. The filtrate (2 mL) from the femurs of each donor was inoculated in a 10-cm dish containing basal medium (BM;α-MEM [HyClone, USA] supplemented with 10% fetal bovine serum [FBS;Gibco, USA], 100 U/mL penicillin G, and 100 mg/L streptomycin [HyClone]) at 37 °C in a humidified atmosphere containing 5% CO₂. After 3 days in culture, non-adherent cells were removed from the culture supernatant. hBMSCs that adhered to the dish were cultured in complete medium that was replaced every 2 days. When the cells reached confluence, they were detached using 0.25% trypsin (Gibco) and then stored or reseeded for culture. Cells were used for subsequent experiments between passages 2 and 5.

hBMSCs (3×10⁵ cells) at passage 3 were incubated with fluorescein-conjugated mouse anti-human monoclonal antibodies for 45 min at room temperature. After the cells were washed with fluorescence-activated cell sorting (FACS) buffer (phosphate buffered saline [PBS] with 10% bovine serum albumin [BSA] and 1% sodium azide) at 300 g for 5 min, the stained cells were resuspended in 250 µL of ice-cold FACS buffer and then subjected to FACS analysis (BD Biosciences, USA). A total of 1×10⁴ events were counted for each sample. The percentage of cells showing a positive signal was analyzed using FlowJo software (Tree Star, USA). The antibodies, including anti-CD34, anti-CD45, anti-CD90, anti-CD29, anti-CD105 and anti-CD44, were purchased from BD Biosciences.

For osteogenic differentiation assays, hBMSCs at passage 3 were seeded at 1×10⁵ cells/well in a six-well culture plate and cultured in BM until confluent. Then, the cells were cultured in BM or osteogenic medium (OM;BM supplemented with 100 ng/mL BMP-2 [PeproTech, USA]) for 21 days for the assessment of alizarin red staining. For chondrogenic differentiation assays, hBMSCs at passage 3 were resuspended in BM at a density of 1×10⁷ cells/mL. Three drops containing a 10-µL cell suspension were carefully added to a 12-well plate. The cells were allowed to adhere at 37 °C in 5% CO₂ for 2 h, followed by the addition of 1 ml of BM or chondrogenic medium (CM;BM supplemented with 10 ng/mL recombination human transforming growth factor-beta 1 [rhTGF-β1;PeproTech], 1× ITS [Invitrogen, USA], 0.5 mg/mL BSA, 4.7 µg/mL linoleic acid, 50 µg/ml L-proline, 100 nM dexamethasone, and 37.5 µg/mL ascorbic acid). The culture medium was changed every 3 days. Micromasses were fixed for paraffin sectioning and alcian blue staining after 21 days. For adipogenic differentiation assays, hBMSCs at passage 3 were seeded at 1×10⁵ cells/well in a six-well plate and then cultured with BM. When confluence was reached, the cells were cultured in BM or adipogenic medium (AM;BM supplemented with 500 nM dexamethasone, 0.5 mM isobutylmethylxanthine, 50 mM indomethacin, and 10 mg/mL insulin). The medium was changed every 3 days. After 14 days in adipogenic culture, the cells were fixed for oil red O staining. Unless otherwise indicated, all other reagents were purchased from Sigma-Aldrich (St. Louis, USA).

Human BMPER genes were ligated into the pLVX-IRES-Puro vector using PCR primers to amplify the coding region. The pLVX-IRES-Puro and pLVX-IRES-Puro-BMPER constructs were transfected into the HEK293T viral packaging cell line together with pSPAX2 and pMD2.G. Viral supernatant was used for the infection of hBMSCs.

siRNA transfection was carried out using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. The sequences of siRNA targeting BMPER (GenePharma Co., Ltd., China) were as follows:

siRNA1: sense 5’-GCGCUGUGUUGUUCAUUGUTT dTdT-3’ and antisense 5’-ACAAUGAACAACACAGCGCTT dTdT-3’;

siRNA2: sense 5’-GACGGUCGGACAUUUAACUTT dTdT-3’ and antisense 5’-AGUUAAAUGUCCGACCGUCTT dTdT-3’;and

siRNA3: sense 5’-GCAACUACAAUGGACAUAATT dTdT-3’ and antisense 5’-UUAUGUCCAUUGUAGUUGCTT dTdT-3’.

The sequences of BMPER negative control siRNA (BMPER NC siRNA) were as follows: sense 5’-UUCUCCGAACGUGUCACGUTT dTdT-3’ and antisense 5’-ACGUGACACGUUCGGAGAATT dTdT-3’.

Total RNA was extracted using TRIzol reagent and reverse transcribed using the PrimeScript RT Master Mix cDNA Synthesis Kit (Takara, Japan) to obtain first strand cDNA. Real-time PCR reactions were performed on the VIIA7 instrument (Applied Biosystems, USA) using the SYBR Green PCR Kit (TaKaRa). The real-time PCR conditions were set as follows: denaturation at 95 °C for 30 s, followed by 50 cycles of 95 °C for 10 s and 60 °C for 30 s. All quantitation was normalized to an endogenous control, GAPDH, and data were analyzed using the 2^-ΔΔCT method. The sequences of the gene primers used are listed in Table 1. All experiments were repeated independently in triplicate.

Total proteins were extracted from hBMSCs using RIPA buffer (Beyotime), and the protein concentrations were determined using a BCA protein assay (Thermo Fisher). Lysate proteins (30 µg) were separated by 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes. The membranes were then blocked with 5% skim milk in TBS-Tween (TBS: 0.05 M Tris and 0.15 M NaCl, pH 7.4; 0.2% Tween-20) for 1 h and incubated with primary antibodies diluted in 1% (w/v) skim milk powder in TBS-Tween overnight at 4 °C. A horseradish peroxidase-labeled secondary antibody was added and visualized using an enhanced chemiluminescence detection system (Millipore, USA) in accordance with the manufacturer’s instructions. All experiments were repeated independently in triplicate.

The cells were washed twice with PBS, fixed in 4% polyoxymethylene for 15 min and stained with alkaline phosphatase reagent (ALP;Shanghai Hongqiao Le Xiang Institute of Biomedical) or alizarin red (Sigma) staining solution at 37 °C for 30 min. After three washes with distilled water, the stained cells were photographed. For quantitative ALP measurements, the cells were rinsed with PBS in triplicate and lysed with RIPA lysis buffer. Next, the cell supernatants were collected in 96-well plates. Substrates and p-nitrophenol from the Alkaline Phosphatase Assay Kit (Beyotime) were added to the 96-well plates, and ALP activity was determined at a wavelength of 405 nm after a 30-min incubation at 37 °C. Finally, the ALP levels were normalized to the total protein content determined by the Bicinchoninic Acid Protein Assay Kit (Beyotime). For the quantification of calcium deposits, 0.1N NaOH was added to dissolve extracellular calcium deposits, and 100 µL of the dissolved solution was detected using a microplate reader (TECAN, Switzerland) at 548 nm. All experiments were repeated independently in triplicate.

hBMSCs were seeded on a six-well plate at a density of 1×10⁵ cells/well and induced by OM after lentiviral infection or siRNA transfection. The medium was changed twice a week, and the conditional medium was collected at the second medium change for the following experiments. The concentrations of VEGF and endostatin in the conditional medium were measured using an ELISA kit (Thermo Fisher, USA) according to the manufacturer’s instructions. All experiments were repeated independently in triplicate.

HUVECs were seeded (1×10⁵ cells/well) in 1% gelatin-coated 24-well plates (Corning, Netherlands). Confluent cells were serum-deprived for 16 h, and linear wounds were created in the monolayers by scratching with a sterile pipette tip. Next, the monolayers were washed with PBS to remove floating cells, and the conditional medium was added. After an additional 24 h, cell migration into the wound was assessed by microscopy. All experiments were repeated independently in triplicate.

A transwell migration assay was performed using Transwell inserts (BD Biosciences) with an 8-µm pore size filter. A total of 5×10⁴ cells were seeded in serum-free medium into the upper chamber of the insert precoated with Matrigel, and 700 µl of conditional medium was added to the lower chamber. After 24 h of incubation, the cells were fixed with 75% ethanol and stained with crystal violet. Then, the cells on the top surface of the membrane were carefully wiped off, and the cells on the lower surface were examined with a microscope. Five random fields were photographed for counting purposes, and the average number of migrated cells was used as a measure of migration capacity. All experiments were repeated independently in triplicate.

HUVECs were serum starved for 16 h and then seeded at a density of 4×10⁴ per well on growth factor-depleted Matrigel (BD Biosciences) in 48-well plates. Conditional medium was added, and the results were quantified after 4 h. Microscopic fields containing the tube structures formed in the gel were photographed at low magnification (×10). Five fields per test condition were examined. Before being photographed, the cells were fixed with 4% paraformaldehyde. All experiments were repeated independently in triplicate.

Immunostaining was performed using a standard protocol. After stimulation with BMP-2 for 7 days, hBMSCs were incubated with the following primary antibodies: BMPER rabbit pAb (Abcam, USA), VEGF mouse pAb (Abcam) and COL1A1 mouse pAb (Abcam) overnight at 4 °C. Then, the primary antibodies were detected using Cy3- or FITC-conjugated anti-mouse or rabbit IgG secondary antibodies (Life Technologies, USA). After the final wash, nuclei were counterstained with Hoechst 3342 (Life Technologies) in PBS for 5 min before imaging. The stained sections were immediately observed by laser confocal microscopy (ZEISS, Germany). All experiments were repeated independently in triplicate.

This animal experiment was approved by the Institutional Ethics Committee of Xin Hua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. Subcutaneous stem cell implantation was performed as previously described [26, 27]. Briefly, BMSCs were infected with different lentiviruses. The cells were collected 24 h after infection, and approximately 3.0×10⁶ cells were mixed with β-TCP ceramic particles (50 mg, Shanghai Bio-lu Biomaterials Co., Ltd, China). This mixture was subcutaneously implanted into the dorsal surfaces of nude mice. After 6 or 12 weeks, the implants were harvested, fixed in 4% paraformaldehyde, decalcified with an ultrasonic decalcification instrument, and embedded in paraffin. For histological analyses, the sections (7 µm) were stained with hematoxylin and eosin (HE) and Masson’s Trichrome stain.

Sections were deparaffinized and rehydrated. Antigen retrieval was performed by digestion with pepsin (Sigma) at 37 °C for several minutes. The sections were blocked with 5% goat serum and 0.1% Tween-20 in PBS for 30 min at room temperature and then incubated with an anti-CD31 primary antibody (Abcam) overnight at 4 °C. A streptavidin-horse radish peroxidase (HRP) detection system was used to detect the antigens, which were visualized with 2, 2′-diaminobenzidine tetrahydrochloride. The sections were then counterstained with hematoxylin and sealed with mounting medium.

This animal experiment was approved by the Institutional Ethics Committee of Xin Hua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. The femoral monocortical defect model was established as previously described [4, 28]. Briefly, nude mice were placed under general anesthesia, and the lateral aspects of their right femurs were exposed. The overlying soft tissues were pushed aside, and the periosteum was carefully preserved. A monocortical osseous hole (0.8 mm diameter) was created on the femur lateral surface using a round burr attached to a dental drill. Irrigation with saline was used to remove bone dust and fragments. Approximately 5.0×10⁴ hBMSCs were exposed to different treatments for 24 h, resuspended in a mixture of medium and Matrigel, and then transplanted into the osseous hole. After one month, the right femurs were fixed in 4% paraformaldehyde for 24 h at 4 °C. The samples were scanned using high-resolution microcomputed tomography (µCT) (SCANCO, Switzerland) with a spatial resolution of 5 µm. Sagittal image sections of the model femurs were used to perform 3D histomorphometric analysis. We defined the regions of interest as (i) the hole region between the interrupted cortical bone ends, (ii) injured bone marrow, and (iii) periosteal callus outside the hole. Old bone fragments remaining from the drilling were excluded from the regions of interest. A total of 100 consecutive images were used for 3D reconstruction and analysis, covering most of the injured region and periosteal callus. Structural parameters included bone mineral density (BMD) and bone volume percentage (bone volume/total volume, BV/TV). Fixed samples were decalcified and embedded in paraffin for HE staining.

Statistical significance was assessed using two-tailed Student’s t-tests or analysis of variance (ANOVA). All statistical analyses were performed using SPSS software version 19.0. Statistical significance was considered at P<0.05(*) and P< 0.01 (**). Data are presented as the means ± S.Ds.

Isolated hBMSCs expressed the MSC markers CD29, CD44, CD90, and CD105 and did not express the hematopoietic marker CD34 or the leukocyte marker CD45 (Fig. 1A). The adipogenic differentiation capacity of hBMSCs was verified by oil red O staining. Lipid droplets were formed in hBMSCs after 14 days of adipogenic induction but were not observed in hBMSCs cultured in BM (Fig. 1B). The osteogenic differentiation assay indicated that most cells formed mineralized calcium deposits, as confirmed by alizarin red staining after 21 days of induction (Fig. 1C). Compared with chondrogenic induction in BM, chondrogenic induction in CM resulted in higher alcian blue staining in the hBMSC micromass after 21 days (Fig. 1D).

The mRNA expression levels of all the involved BMP inhibitors showed varying degrees of change during osteogenic differentiation, and BMPER expression increased over time except for a decrease on day 21 compared with that on day 14 (Fig. 2A). Western blot analysis showed that BMPER was significantly upregulated over time, with an approximately 5-fold increase in protein levels after 21 days in the osteogenic induction group compared with those in the non-induction group (Fig. 2B and C).

To confirm the effect of BMPER on hBMSC osteogenesis, we used lentiviruses expressing BMPER and siRNA targeting BMPER. Three siRNAs were synthesized, and while they all showed inhibition effects (data not shown), only siRNA 1 was used in the subsequent experiment because it exhibited the highest inhibition efficiency. The levels of BMPER mRNA and protein expression were significantly increased by pLVX-BMPER (Fig. 3A and B) and suppressed by siRNA 1 (Fig. 3C and D). The ALP assay showed that pLVX-BMPER infection increased the ALP activity in the cells after 7 days of osteogenic induction, while transfection of the siRNA exerted the opposite effect (Fig. 3E and G). Qualitative alizarin red staining and the quantitative calcium assay showed that more calcium deposits were produced by the BMPER overexpression group than by the negative control (NC) and pLVX-vector groups after 21 days of osteogenic induction (Fig. 3F and H).

After 7 days of osteogenic induction, the mRNA expression levels of osteogenesis-related genes, including OSX, OCN, and COL1A1, in the pLVX-BMPER group were higher than those in the NC and pLVX-vector groups. The mRNA levels of these genes in the siRNA group were lower than those in the NC and NC-siRNA groups (Fig. 4A-C). Consistent with the mRNA expression levels, the protein expression levels of OSX, OCN, and COL1A1 were increased with BMPER overexpression and decreased with BMPER suppression (Fig. 4D). In addition, BMPER overexpression increased phosphorylated Smad1/5/8, which is important the activation of BMP-2 signaling during osteogenesis (Fig. 4D). Furthermore, the immunofluorescence assay confirmed that COL1A1 expression increased with BMPER overexpression and decreased with BMPER suppression (Fig. 4E).

To determine whether BMPER is involved in the osteogenesis-angiogenesis coupling process, we cultured HUVECs in conditional medium collected from osteogenic hBMSCs with altered BMPER expression. HUVECs cultured in conditional medium from the pLVX-BMPER group exhibited a higher migration rate than those from the pLVX-vector group, while the siRNA-mediated downregulation of BMPER significantly weakened the migration ability of HUVECs (Fig. 5A, B, D and E). HUVECs seeded on Matrigel in conditional medium from the pLVX-BMPER group formed many branches, termed anastomosing tubes, resulting in a meshwork of capillary-like structures (Fig. 5C). By contrast, many HUVECs remained spherical and isolated when maintained in conditional medium from hBMSCs treated with BMPER siRNA, with small cellular nests and short tubes being observed (Fig. 5C). In addition, secretion of the angiogenic factor VEGF and the angiostatic factor endostatin by hBMSCs was detected. Compared with that in the pLVX-vector group, the pLVX-BMPER group showed significantly enhanced VEGF secretion and suppressed endostatin secretion (Fig. 5F and G), indicating that BMPER stimulated VEGF expression and inhibited endostatin expression. The immunofluorescence assay further confirmed that VEGF expression increased with BMPER overexpression and decreased with BMPER suppression (Fig. 5H). These findings suggested that upregulated BMPER in hBMSCs enhances angiogenesis in vitro.

Two animal models, the femoral monocortical defect model and the ectopic bone formation model, were established to examine the effects of BMPER on osteogenesis and angiogenesis in vivo. In the ectopic bone formation model, hBMSCs treated under the indicated condition were combined with β-TCP and then subcutaneously implanted into mice. The implants were harvested at six weeks for immunohistochemical analysis and at twelve weeks for HE and Masson’s Trichrome staining. The immunohistochemical analysis confirmed that the upregulation of BMPER in hBMSCs promoted angiogenesis during ectopic bone formation, as shown by the marked increases in the numbers of CD31-positive cells and capillaries in the implants (Fig. 6E). HE and Masson’s Trichrome staining confirmed that overexpression of BMPER enhanced ectopic bone formation in hBMSCs, as shown by the more robust bone formation and better mineralization (Fig. 6F). In the femoral monocortical defect model, lateral views of the 3D reconstruction of injured femurs showed more mineralized tissue in the defects of pLVX-BMPER femurs than in those of control femurs and less mineralized tissue in the defects of BMPER siRNA femurs than in those of NC femurs (Fig. 6A-C). HE staining showed that new bone covered the defective region and almost filled the bone marrow cavity in the pLVX-BMPER group, while only a small area of the hole was covered with new bone in the BMPER siRNA group (Fig. 6D). Therefore, these findings indicate that BMPER plays important roles in the osteogenic differentiation of hBMSCs and that overexpression of BMPER in hBMSCs promotes angiogenesis and increases bone formation in vivo.

Enhancing bone formation and angiogenesis is key for the treatment of fractures, bone nonunion, delayed healing of fractures and other related diseases. BMPs play a critical role in the osteogenic differentiation of hBMSCs. However, in-depth research into the relationship between BMP inhibitors and bone formation is lacking. In this study, overexpression of one ‘BMP inhibitor’, BMPER, remarkably increased the BMP-2-induced osteogenic differentiation of hBMSCs both in vitro and in vivo. Furthermore, mixing β-TCP ceramic particles with hBMSCs led to extensive angiogenesis and vascularization and more mature new bone when hBMSCs overexpressed BMPER. This study extends our understanding of the role of BMPER in the osteogenesis-angiogenesis coupling process in hBMSCs, revealing that BMPER might be beneficial to human bone formation.

Most BMP inhibitors act as inhibitors by competing with BMP receptors for binding to BMP ligands [8, 9]. Consistent with previous studies, we recently demonstrated that one BMP inhibitor, GREM2, negatively regulates the BMP-2-induced osteogenesis of hBMSCs via suppressing the BMP-2/Smad/Runx2 signaling pathway [4]. To comprehensively understand the role of other BMP inhibitors in the osteogenic differentiation of hBMSCs, we examined the mRNA expression levels of eleven BMP inhibitors during osteogenic differentiation. Although the expression of NOGGIN changed the most (Fig. 2A), we did not study this protein in depth because it plays a positive role during the osteogenic differentiation of hBMSCs [1, 12]. Numerous studies have clarified the function of BMPER in vasculature;however, little is known about the role of BMPER in osteogenesis. Since differential expression of BMPER was detected in the osteogenic differentiation process of hBMSCs (Fig. 2A-C), we propose that BMPER participates in the regulation of hBMSC differentiation. ALP and alizarin red staining confirmed that BMPER plays a positive role in the osteogenic differentiation process (Fig. 3E and F). Activation of the BMP-Smad signaling pathway is a critical process during osteogenic differentiation. When BMPs combine with BMP receptors on the BMSC membrane, Smad1/5/8 is phosphorylated and integrates with Smad4 to form a complex and is then transferred into the nucleus to promote the expression of Runx2 and other related osteogenic target genes, such as COL1A1, OCN and OSX [29]. In this study, overexpression of BMPER significantly elevated the phosphorylation of Smad1/5/8 and other related osteogenic target genes during the BMP-2-induced osteogenic process in hBMSCs (Fig. 4). Since BMP-2 remarkably promoted the expression of BMPER (Fig. 2), and upregulation of BMPER had an enhancing effect on the BMP2 signaling pathway (Fig. 4), a positive feedback mechanism accelerating osteogenic differentiation may exist. By contrast, BMPER suppression by siRNA decreases this positive feedback signal and delays the differentiation process, which may partially explain the vertebral dysplasia resulting from BMPER malfunctional mutation [18, 19]. However, further studies are needed to reveal the specific cellular and molecular mechanisms involved in the BMPER regulation of osteogenesis.

Bone regeneration occurs in close spatial and temporal association with angiogenesis [30]. Thus, promoting vascularization by optimizing effective osteogenesis-angiogenesis coupling is essential for fracture healing [31]. In a previous study, BMPER was ascribed a pro-BMP role and shown to promote sprouting and migration in HUVECs [16]. In this study, conditional medium from hBMSCs infected with pLVX-BMPER enhanced the migration (Fig. 5A, 5B) and network formation (Fig. 5C) abilities of HUVECs in vitro. A possible mechanism underlying this phenomenon is increased secretion of the angiogenic factor VEGF and suppressed secretion of the angiostatic factor endostatin induced by BMPER overexpression in hBMSCs during osteogenic differentiation (Fig. 5F and G). As a suitable anti-BMPER neutralizing antibody is lacking, we could not reveal the precise mechanism underlying the enhancing effect of the conditional medium on angiogenesis, and whether BMPER directly acts on ECs, indirectly affects ECs by promoting VEGF secretion, or exerts both effects remains unclear. CD31 is a cell adhesion molecule that is mainly expressed on the cell membranes of both developing and mature vascular ECs and is considered a marker of vascular ECs [32]. Our data showed that upregulation of BMPER in hBMSCs markedly increased the numbers of CD31-positive cells and capillaries in ectopic osteogenesis implants (Fig. 6E). Combined with our in vitro experiments, there is a possibility that BMPER overexpression stimulates hBMSCs to secrete more VEGF and less endostatin, thus affecting angiogenesis. However, we cannot exclude the possibility that modified hBMSCs communicate with ECs and their progenitors by other means. The mechanism underlying how the osteogenesis-angiogenesis coupling process is promoted by BMPER overexpression merits further investigation.

Our study demonstrates that BMPER functions as a positive regulator of osteogenesis. Importantly, our results show that upregulating BMPER enhances bone formation by promoting the osteogenesis-angiogenesis coupling process in hBMSCs, suggesting a novel therapeutic role of BMPER in the regenerative capacity of bone repair.

This work was supported by grants from the Ministry of Science and Technology of China (No. 2015DFG32200), National Natural Science Foundation of China (No.81772347, 81572123), Science and Technology Commission of Shanghai Municipality (No. 16430723500), Shanghai Municipal Education Commission-Gaofeng Clinical Medicine Grant Support (No.20161314), Shanghai Shen Kang hospital development center (No.16CR2036B) and China Postdoctoral Science Foundation (No.2017M621501).

The authors declare no Disclosure Statement.

F. Xiao, CD. Wang and CL. Wang contributed to the work equally.