Experimental and Molecular Pathology 98 (2015) 158–163
Contents lists available at ScienceDirect
Experimental and Molecular Pathology journal homepage: www.elsevier.com/locate/yexmp
Analysis of altered microRNA expression profile in the reparative interface of the femoral head with osteonecrosis Heng-feng Yuan a,b, Christina Von Roemeling b, Hui-di Gao c, Jing Zhang a, Chang-an Guo a, Zuo-qin Yan a,⁎ a b c
Department of Orthopedics, Zhongshan Hospital, Fudan University, Shanghai, China Department of Cancer Biology, Mayo Clinic, FL, USA Department of Biochemistry and Molecular Biology, Shanghai Medical College of Fudan University, Shanghai, China
a r t i c l e
i n f o
Article history: Received 29 December 2014 Accepted 5 January 2015 Available online 19 January 2015 Keywords: MicroRNA Reparative interface Femoral head Osteonecrosis Genes
a b s t r a c t The reparative reaction is considered to be important during the occurrence of collapse in the femoral head with osteonecrosis (ONFH), but little is known about the long-term reparative process. The aim of this study was to determine and analyze the altered microRNA expression profile in the reparative interface of ONFH, and further validate the expression of the involved genes in the predicted pathways. Microarray analysis was performed comparing the reparative interface of patients with ONFH and normal tissue of patients with fresh femoral neck fracture (FNF) and partly validated by real-time PCR. Potential target genes of differentially expressed miRNAs were predicted by TargetScan and miRanda, and the target genes were used for further bioinformatics analysis such as Gene Ontology and Pathway assay. The filtered miRNAs and genes in the predict pathways were further examined by real-time PCR in another 6 independent ONFH patients. Among the 2578 miRNAs identified, 17 were consistently differentially expressed, 12 of which are up-regulated and 5 down-regulated. GO classification showed that the predicted target genes of these miRNAs are involved in signal transduction, cell differentiation, methylation, cell growth and apoptosis. The Kyoto Encyclopedia of Genes and Genomes (KEGG) classification indicated that these genes play a role in angiogenesis and Wnt signaling pathways. The expression of miR-34a and miR-146a and genes in the predict pathways were significantly up-regulated. This study presented a global view of miRNA expression in the reparative interface of osteonecrosis. In addition, our data provided novel and robust information for further researches in the pathogenesis and molecular events of ONFH. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Osteonecrosis of the femoral head (ONFH) is the pathological process of ischemic changes in cellular constituents of the femoral head including bone, endothelial, adipose and hematopoietic cells under the action of one or more factors that cause cell necrosis and apoptosis (Zalavras and Lieberman, 2014; Fukushima et al., 2010; Assouline-Dayan, 2002). It is a progressive and devastating disease that if left untreated results in collapse of the femoral head, necessitating hip replacement in approximately 70% of patients (Hernigou et al., 2004; Johnson et al., 2014). While there are various theories on the pathogenesis of ONFH, such as extravascular accumulation of fat mediated vascular constriction, the intravascular blood coagulation theory, and the intravascular fat embolism concept (Schroer, 1994), the exact pathogenesis is still unclear. The reparative reaction where osteoclasts mediate reabsorption of necrotic bone is believed to be initiated in early stages of ONFH. During the repair process, imbalanced osteoclast activity over osteoblast mediated bone reformation results in structural weakening, causing collapse of the femoral head in many patients (Hernigou et al., 2004; Nishii, ⁎ Corresponding author. E-mail address: [email protected] (Z. Yan).
http://dx.doi.org/10.1016/j.yexmp.2015.01.002 0014-4800/© 2015 Elsevier Inc. All rights reserved.
2002). The mechanism of collapse is still controversial, however it is generally considered that the reparative reaction rather than the osteonecrosis itself is the primary cause (Li et al., 2009; Kim et al., 2006). To our knowledge, previous studies have mainly focused on parameters of the necrotic zone, including the location, the size, and the pathological changes of this region (Ha et al., 2006; Steinberg et al., 2006; Mont et al., 2010). Currently, the molecular mechanisms contributing to the pathogenesis of ONFH at the reparative interface remain poorly understood. MicroRNAs (miRNAs) belong to a family of non-protein-coding small RNAs that are involved in various physiological and pathological processes (Ambros, 2001; Bartel, 2004). They are approximately ~ 22 nucleotides in length and are expressed in a tissue- or cell-specific manner (Lagos-Quintana et al., 2002). Among them, a set of miRNAs have been confirmed to play fundamental roles in gene regulation of various orthopedic diseases, such as bone tumors (Kafchinski and Jones, 2014), osteoarthritis (Miyaki and Asahara, 2012), and rheumatoid arthritis (Furer et al., 2010). Although several miRNAs have been independently reported (Yamasaki et al., 2012; Sun et al., 2014; Jia et al., 2014), and one preliminary review discussing circulating microRNAs in ONFH has been written (Wang, X. et al., 2014; Wang, Y. et al., 2014), the miRNA expression profile of ONFH remains unclear. Therefore an in depth investigation
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
159
on the role of miRNA in ONFH will help us better understand the molecular mechanisms of this debilitating disease. The objective of our study is to provide a global view of miRNA expression in the reparative interface, and additionally present a novel and comprehensive molecular signature that contributes to the pathogenesis of ONFH that can be used to support future research efforts. We achieved this by (1) to determining the altered microRNA expression profile in the reparative interface of ONFH, (2) analyzing the results of our array using the Gene Ontology and Pathway assays, (3) and using the filtered miRNAs to predict pathways critical in ONFH. 2. Materials and methods 2.1. Ethical statement The study was reviewed and approved by the Ethical Committee of Zhongshan Hospital, Fudan University and patients gave informed consent. 2.2. Patients We selected 9 patients with ONFH who underwent total hip arthroplasty (THA) in our department. ONFH was diagnosed based on imaging examinations and the diagnosis was made according to the guidelines of the Chinese Medical Association for ONFH (Orthopedic Panel, 2012). We evaluated another 6 patients with fresh femoral neck fracture (FNF) who underwent THA in our department (excluding other types of bone and joint diseases) as the normal control group. All study characters were of Chinese Han nationality without any blood relationship to one another. The clinical characteristics of all patients are shown in Table 1. 2.3. Sample preparation When THA was performed, the isolated femoral head was obtained and cut along the coronal plane to check the appearance of the reparative interface zone. The reparative interface is the tissue on the edge of the necrotic zone (Fig. 1). The reparative interface was then cut into smalls pieces of approximately 5 × 5 × 5 mm3, and rapidly immersed in a portable liquid nitrogen tank. The samples from the tank were stored in a −80 °C freezer after leaving the operating room. 2.4. RNA isolation and microRNA microarray The samples were rapidly immersed the liquid nitrogen for 1 h and crushed by a pincer. Total RNA in the samples was isolated using Trizol Reagent (Invitrogen) according to the manufacture's protocol. The
Table 1 Clinical characteristics of the ONFH patients and controls. Patient
Age
Gender
Diagnosis
ONFH classification
Other illness
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
22 31 33 61 46 43 65 63 56 78 84 76 69 88 82
Female Female Female Female Female Female Male Male Male Female Female Female Female Male Female
Glucocorticoid-related ONFH Glucocorticoid-related ONFH Glucocorticoid-related ONFH Idiopathic ONFH Glucocorticoid-related ONFH Glucocorticoid-related ONFH Alcohol-related ONFH Alcohol-related ONFH Idiopathic ONFH FNF FNF FNF FNF FNF FNF
IIIB IIIC IIIB IIIC IIIC IIIC IIIC IV IIIC / / / / / /
SLE SLE SLE / RA SLE / / / / / / / /
Fig. 1. Samples acquired in the reparative interface of the femoral head in ONFH patients (△).
integrity and concentration of all RNA samples were quantified using the NanoDrop 1000 spectrophotometer (Thermo Scientific). All samples met the quality control standards. The total RNA of 3 patients diagnosed with ONFH (Patients 1–3) and 3 patients diagnosed with FNF (Patients 10–12) were hybridized to an Affymetrix GeneChip miRNA 4.0 Array containing 2578 human miRNA sequences. RVM t-test was applied to filter the differentially expressed miRNAs for the ONFH and normal control group because the RVM t-test can raise degrees of freedom effectively in the cases of small samples. After the significant analysis and FDR analysis, we selected the differentially expressed miRNAs according to the p value threshold. P value of b0.05 was considered as significant difference. 2.5. Bioinformatics analysis Bioinformatics analysis (Genminix Informatics Ltd., Shanghai, China) was performed for miRNAs expressed in significant amounts. We used the TargetScan (http://www.targetscan.org/) and miRanda (http://www.microrna.org/microrna/home.do) to predict the intersection of target genes of miRNAs. GO analysis was applied to analyze the main function of the differential expression target genes according to the Gene Ontology (GO, http://www.geneontology.org/) which is the key functional classification of NCBI. Pathway analysis was used to find out the significant pathway of the differential genes according to Kyoto Encyclopedia of Genes and Genomes (KEGG, http://www.genome.jp/ kegg/). GO terms and KEGG Pathway annotation of the miRNA target genes were found using the web-based tool, Database for Annotation, Visualization, and Integrated Discovery (DAVID, http://david.abcc.ncifcrf. gov/). Fisher's exact test and χ2 test were used to classify the GO category and select the significant pathway, and the false discovery rate (FDR) was used to correct the p value. We chose only GO terms with p value of b 0.01 and FDR value of b0.01, and Pathways with p value of b 0.05 and FDR value of b0.05. 2.6. Quantitative real-time PCR (RT-PCR) To validate the microarray data, we used the RNA-tailing and primer-extension RT-PCR method to detect the expression of the miRNA. RNA samples used in RT-PCR validation experiments were the
160
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
Table 2 Primers used in RT-PCR of the four genes. Gene name
Primer sequence
HIF-1α
Sense: 5′-TGATTGCATCTCCATCTCCTACC-3′ Antisense: 5′-GACTCAAAGCGACAGATAACACG-3′ Sense: 5′-CCCACTGAGGAGTCCAACAT-3′ Antisense: 5′-TCCAGGGCATTAGACAGCA-3′ Sense: 5′-CGAAGAGCAAGAATAAAT-3′ Antisense: 5′-GAATGAGGAAAGCAAACT-3′ Sense: 5′-TGGTGGGCTGCAGAAAATGGTT-3′ Antisense: 5′-ACGATGGCCGGCTTGTTGC-3′
VEGF-A Hes1 β-catenin
ones used in the array assay. RT-PCR was performed with a miRNA assay kit (RiboBio Co., Ltd., Guangzhou, China) according to the manufacture's protocol. U6 small nuclear RNA was used as an endogenous control. In addition, we tested two miRNAs (miR-34a and miR-146a) and four genes involved in the predicted pathways in another 6 patients (Patients 4–9) and 6 controls (Patients 10–15). β-actin was used as an endogenous control for the four genes. The primers of the four genes are described in Table 2. The 7500 Real-Time PCR System (ABI) was used for amplification and detection. The CT values were obtained from the amplification plot using SDS software. Finally, the relative miRNA level was normalized to U6 expression, and the gene level to β-actin expression for each sample in triplicate and was expressed as 2−△△CT. The data are presented as the means ± the standard deviations (S.D.s) and were analyzed with one-way ANOVA using SPSS 16.0 (SPSS Inc., Chicago, IL, USA). A p value of b0.05 was considered statistically significant. 3. Results 3.1. MiRNA expression patterns in the reparative interface of ONFH We used Affymetrix GeneChip miRNA 4.0 Array to analyze miRNAs expression in the reparative region, and applied the RVM t-test to filter
Fig. 3. Validation of miRNAs by RT-PCR. The expression levels of the six miRNAs were in concordance with the normalized microarray data.
the differentially expressed miRNAs. Twelve miRNAs (miR-181c-3p, miR-34a-3p, miR-146a-5p, miR-187-3p, miR-181a-3p, miR-30c-1-3p, miR-650, miR-3652, miR-4444, miR-1273e, miR-99a-3p, miR-30645p) were found to be significantly up-regulated (p b 0.05) while five miRNAs (miR-212-3p, miR-212-5p, miR-132-3p, miR-629-3p, miR6836-5p) were markedly down-regulated (p b 0.05) (Fig. 2). In order to validate the microarray results, six representative miRNAs were chosen for RT-PCR using the samples that were differentially expressed in the array assay. The expression levels of all the selected miRNAs were in conformity with the normalized microarray data (Fig. 3).
Fig. 2. The miRNA profile differentiates ONFH patients from normal individuals. (A) Hierarchical clustering for the differentially expressed miRNAs (p b 0.05); (B) Scatter plot for all expressed miRNAs, green spots represent down-regulated miRNAs, red spots represent up-regulated miRNAs (p b 0.05); (C) Volcano plot of the microarray data, red spots represent down-regulated miRNAs, blue spots represent up-regulated miRNAs (p b 0.05). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
161
Fig. 4. GO category analysis based on miRNA-targeted genes (p value b 0.01, FDR b 0.05). The vertical axis represents the GO category, and the horizontal axis represents the enrichment of GO.
3.2. Microarray-based Gene Ontology and Pathway analysis According to the GO and KEGG database, we gained functions and pathways of all the predicted target genes of the differentially expressed miRNAs. Fisher's exact test and χ2 test were used to calculate the p value and FDR. According to the standard of p b 0.01 (GO) and p b 0.05 (KEGG), significant functions and pathways were filtered. As shown in Fig. 4, GO results demonstrate upregulation of pathways related to signal transduction, cell differentiation, cell methylation, cell growth and apoptosic processes. In addition, the target genes were also enriched into biological
pathways, such as angiogenesis related pathways, Wnt signaling pathway and PI3K–Akt signaling pathway (Fig. 5). 3.3. Expression of the screened miRNAs and related pathways We tested for expression of miR-34a and miR-146a using the RNAtailing and primer-extension RT-PCR method in patient samples that were independent from the array assay. The results confirmed that expressions of miR-34a and miR-146a were up-regulated by 4 and 5 folds respectively in the patients of ONFH when compared with the
Fig. 5. Pathway analysis based on miRNA-targeted genes (p value b 0.05). The vertical axis represents the pathway category, and the horizontal axis represents the value of −LgP.
162
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
Fig. 6. Expression of miR-34a and miR-146a in the six independent patients of ONFH compared with the controls (***p b 0.001, *p b 0.05).
control samples (Fig. 6). Expression of the predicted target genes for each miR-34a and miR-146a, including HIF-1α in HIF-1 signaling pathway, VEGF-A in VEGF signaling pathway, Hes1 in Notch signaling pathway and β-catenin in Wnt signaling pathway, were all significantly upregulated (p b 0.05) (Fig. 7). 4. Discussion The reparative process plays a critical role in the clinical course and symptoms of ONFH, and is generally considered to be initiated after the occurrence of death of bone or marrow elements (Mont et al., 1998). However, when the lesion is relatively small (15–20 mm) or medially located, the reparative reaction typically has no specific repair function (Lieberman et al., 2012; Glimcher and Kenzora, 1979). While the mechanism is not entirely elucidated, it has been described that in the reparative margin of the infarction, dead trabeculae are resorbed by osteoclasts and are replaced with appositional new bone, resulting in thickened reinforced trabeculae. The thickened interface could prevent penetration of revascularization into the necrotic zone and cause
irreversible damage, possibly leading to eventual collapse (Mont et al., 1998). Further investigations are needed to better understand the reparative process of osteonecrosis so that appropriate preventative or remedial treatment can be applied. In this study we have demonstrated that a number of miRNAs are significantly differentially expressed in the reparative interface. Some of these miRNAs have been characterized for their important roles in bone and joint diseases. MiR-146a has been reported to inhibit osteoclastogenesis and prevent joint destruction in arthritic mice, presenting as a potential novel therapeutic target for bone destruction in rheumatoid arthritis (Nakasa et al., 2011; Ammari et al., 2013). Another specific miRNA, miR-34a, is linked to inhibition of angiogenesis by blocking VEGF production, regulation of cell apoptosis, and additionally has been shown to inhibit the proliferation and metastasis of osteosarcoma cells both in vitro and in vivo (Kumar et al., 2012; Tian et al., 2014; Yan et al., 2012). In order to obtain insights into the classification of miRNAs, we performed GO analysis in order to predict target genes. The result revealed several GOs related to signal transduction including adenosine receptor
Fig. 7. Expression of four genes involved in the predict pathways. A. Expression of HIF-1α in HIF-1 signaling pathway (*p b 0.05); B. Expression of VEGF-A in VEGF signaling pathway (**p b 0.01); C. Expression of Hes1 in Notch signaling pathway (**p b 0.01); D. Expression of β-catenin in Wnt signaling pathway (***p b 0.001).
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
signaling pathway, regulation of dephosphorylation, cGMP catabolic process, transcription initiation from RNA polymerase II promoter, and signal transduction. In addition we found correlations with cell differentiation (osteoclast differentiation, fat cell differentiation, regulation of cell proliferation, chondrocyte differentiation), cell methylation (histone H4-K20 methylation, positive regulation of histone H3-K9 methylation), cell growth (cell cycle, Notch signaling pathway, regulation of cell proliferation) and apoptosis (apoptotic signaling pathway, negative regulation of apoptotic process). Together these groups represent the most of the significantly enriched GO terms. Using KEGG Pathway analysis we were able to functionally categorize predicted target genes. We found strong correlations to angiogenesis related pathways such as HIF-1 signaling pathway, VEGF signaling pathway and Notch signaling pathway. This suggests that angiogenesis plays a crucial role in the reparative process. In addition, the Wnt signaling pathway, which has been reported to play a pivotal role in regulating the balance between bone formation and bone resorption (Zhong et al., 2014; Wang, X. et al., 2014; Wang, Y. et al., 2014), was enriched in our results. We believe this pathway could be responsible for some of the pathological changes observed in the reparative interface, and may provide an avenue for therapeutic intervention, warranting further investigation. PI3K–Akt signaling pathway, which has been reported to be closely relevant with SLE (Beşliu, 2009), was also found in our study. We believe this is reasonable since the ONFH samples we used in the array analysis are all complicated with SLE. In conclusion, our study unveiled the dysregulation of miRNA in the reparative interface of ONFH. Bioinformatics-based analysis of these changes in expression may provide a useful tool for understanding the molecular mechanisms responsible for collapse of the femoral head. Further investigations are needed to clarify the exact roles of identified miRNA in the pathogenesis of reparative process. Disclosure/conflict of interest The authors have no conflict of interest to declare. Acknowledgments This research was supported by the National Natural Science Foundation of China (No. 81371966) and International Cooperation Project of Shanghai Science and Technology Committee (No. 12410710200). References Ambros, V., 2001. microRNAs: tiny regulators with great potential. Cell 107, 823–826. Ammari, M., et al., 2013. Impact of microRNAs on the understanding and treatment of rheumatoid arthritis. Curr. Opin. Rheumatol. 25, 225–233. Assouline-Dayan, Y., 2002. Pathogenesis and natural history of osteonecrosis. Semin. Arthritis Rheum. 32, 94–124. Bartel, D.P., 2004. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116, 281–297. Beşliu, A.N., 2009. PI3K/Akt signaling in peripheral T lymphocytes from systemic lupus erythematosus patients. Roum. Arch. Microbiol. Immunol. 68, 69–79. Fukushima, W., et al., 2010. Nationwide epidemiologic survey of idiopathic osteonecrosis of the femoral head. Clin. Orthop. Relat. Res. 468, 2715–2724.
163
Furer, V., et al., 2010. The role of microRNA in rheumatoid arthritis and other autoimmune diseases. Clin. Immunol. 136, 1–15. Glimcher, M.J., Kenzora, J.E., 1979. The biology of osteonecrosis of the human femoral head and its clinical implications: II. The pathological changes in the femoral head as an organ and in the hip joint. Clin. Orthop. Relat. Res. 283–312. Ha, Y.C., et al., 2006. Prediction of collapse in femoral head osteonecrosis: a modified Kerboul method with use of magnetic resonance images. J. Bone Joint Surg. Am. 88 (Suppl. 3), 35–40. Hernigou, P., et al., 2004. Fate of very small asymptomatic stage-I osteonecrotic lesions of the hip. J. Bone Joint Surg. Am. 86-A, 2589–2593. Jia, J., et al., 2014. MiR-17-5p modulates osteoblastic differentiation and cell proliferation by targeting SMAD7 in non-traumatic osteonecrosis. Exp. Mol. Med. 46, e107. Johnson, A.J., et al., 2014. Treatment of femoral head osteonecrosis in the United States: 16-year analysis of the Nationwide Inpatient Sample. Clin. Orthop. Relat. Res. 472, 617–623. Kafchinski, L.A., Jones, K.B., 2014. MicroRNAs in osteosarcomagenesis. Adv. Exp. Med. Biol. 804, 119–127. Kim, H.K., et al., 2006. RANKL inhibition: a novel strategy to decrease femoral head deformity after ischemic osteonecrosis. J. Bone Miner. Res. 21, 1946–1954. Kumar, B., et al., 2012. Dysregulation of microRNA-34a expression in head and neck squamous cell carcinoma promotes tumor growth and tumor angiogenesis. PLoS One 7, e37601. Lagos-Quintana, M., et al., 2002. Identification of tissue-specific microRNAs from mouse. Curr. Biol. 12, 735–739. Li, W., et al., 2009. Distribution of TRAP-positive cells and expression of HIF-1alpha, VEGF, and FGF-2 in the reparative reaction in patients with osteonecrosis of the femoral head. J. Orthop. Res. 27, 694–700. Lieberman, J.R., et al., 2012. Which factors influence preservation of the osteonecrotic femoral head? Clin. Orthop. Relat. Res. 470, 525–534. Miyaki, S., Asahara, H., 2012. Macro view of microRNA function in osteoarthritis. Nat. Rev. Rheumatol. 8, 543–552. Mont, M.A., et al., 1998. Osteonecrosis of the femoral head. Potential treatment with growth and differentiation factors. Clin. Orthop. Relat. Res. (355 Suppl.), S314–S335. Mont, M.A., et al., 2010. The natural history of untreated asymptomatic osteonecrosis of the femoral head: a systematic literature review. J. Bone Joint Surg. Am. 92, 2165–2170. Nakasa, T., et al., 2011. The inhibitory effect of microRNA-146a expression on bone destruction in collagen-induced arthritis. Arthritis Rheum. 63, 1582–1590. Nishii, T., 2002. Progression and cessation of collapse in osteonecrosis of the femoral head. Clin. Orthop. Relat. Res. 149–57. Orthopedic Panel, 2012. Diagnostic criteria for adult osteonecrosis (2012 edition). Chin. J. Orthop. 32, 606–610. Schroer, W.C., 1994. Current concepts on the pathogenesis of osteonecrosis of the femoral head. Orthop. Rev. 23, 487–497. Steinberg, D.R., et al., 2006. Determining lesion size in osteonecrosis of the femoral head. J. Bone Joint Surg. Am. 88 (Suppl. 3), 27–34. Sun, J., et al., 2014. Downregulation of PPARγ by miR-548d-5p suppresses the adipogenic differentiation of human bone marrow mesenchymal stem cells and enhances their osteogenic potential. J. Transl. Med. 12, 168. Tian, Y., et al., 2014. MicroRNA-199a-3p and microRNA-34a regulate apoptosis in human osteosarcoma cells. Biosci. Rep. 34, 479–485. Wang, X., et al., 2014. Preliminary screening of differentially expressed circulating microRNAs in patients with steroid-induced osteonecrosis of the femoral head. Mol. Med. Rep. 10, 3118–3124. Wang, Y., et al., 2014. Wnt and the Wnt signaling pathway in bone development and disease. Front. Biosci. 19, 379–407. Yamasaki, K., et al., 2012. Angiogenic microRNA-210 is present in cells surrounding osteonecrosis. J. Orthop. Res. 30, 1263–1270. Yan, K., et al., 2012. MicroRNA-34a inhibits the proliferation and metastasis of osteosarcoma cells both in vitro and in vivo. PLoS One 7, e33778. Zalavras, C.G., Lieberman, J.R., 2014. Osteonecrosis of the femoral head: evaluation and treatment. J. Am. Acad. Orthop. Surg. 22, 455–464. Zhong, Z., et al., 2014. WNT signaling in bone development and homeostasis. Wiley Interdiscip. Rev. Dev. Biol. 3, 489–500.
Contents lists available at ScienceDirect
Experimental and Molecular Pathology journal homepage: www.elsevier.com/locate/yexmp
Analysis of altered microRNA expression profile in the reparative interface of the femoral head with osteonecrosis Heng-feng Yuan a,b, Christina Von Roemeling b, Hui-di Gao c, Jing Zhang a, Chang-an Guo a, Zuo-qin Yan a,⁎ a b c
Department of Orthopedics, Zhongshan Hospital, Fudan University, Shanghai, China Department of Cancer Biology, Mayo Clinic, FL, USA Department of Biochemistry and Molecular Biology, Shanghai Medical College of Fudan University, Shanghai, China
a r t i c l e
i n f o
Article history: Received 29 December 2014 Accepted 5 January 2015 Available online 19 January 2015 Keywords: MicroRNA Reparative interface Femoral head Osteonecrosis Genes
a b s t r a c t The reparative reaction is considered to be important during the occurrence of collapse in the femoral head with osteonecrosis (ONFH), but little is known about the long-term reparative process. The aim of this study was to determine and analyze the altered microRNA expression profile in the reparative interface of ONFH, and further validate the expression of the involved genes in the predicted pathways. Microarray analysis was performed comparing the reparative interface of patients with ONFH and normal tissue of patients with fresh femoral neck fracture (FNF) and partly validated by real-time PCR. Potential target genes of differentially expressed miRNAs were predicted by TargetScan and miRanda, and the target genes were used for further bioinformatics analysis such as Gene Ontology and Pathway assay. The filtered miRNAs and genes in the predict pathways were further examined by real-time PCR in another 6 independent ONFH patients. Among the 2578 miRNAs identified, 17 were consistently differentially expressed, 12 of which are up-regulated and 5 down-regulated. GO classification showed that the predicted target genes of these miRNAs are involved in signal transduction, cell differentiation, methylation, cell growth and apoptosis. The Kyoto Encyclopedia of Genes and Genomes (KEGG) classification indicated that these genes play a role in angiogenesis and Wnt signaling pathways. The expression of miR-34a and miR-146a and genes in the predict pathways were significantly up-regulated. This study presented a global view of miRNA expression in the reparative interface of osteonecrosis. In addition, our data provided novel and robust information for further researches in the pathogenesis and molecular events of ONFH. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Osteonecrosis of the femoral head (ONFH) is the pathological process of ischemic changes in cellular constituents of the femoral head including bone, endothelial, adipose and hematopoietic cells under the action of one or more factors that cause cell necrosis and apoptosis (Zalavras and Lieberman, 2014; Fukushima et al., 2010; Assouline-Dayan, 2002). It is a progressive and devastating disease that if left untreated results in collapse of the femoral head, necessitating hip replacement in approximately 70% of patients (Hernigou et al., 2004; Johnson et al., 2014). While there are various theories on the pathogenesis of ONFH, such as extravascular accumulation of fat mediated vascular constriction, the intravascular blood coagulation theory, and the intravascular fat embolism concept (Schroer, 1994), the exact pathogenesis is still unclear. The reparative reaction where osteoclasts mediate reabsorption of necrotic bone is believed to be initiated in early stages of ONFH. During the repair process, imbalanced osteoclast activity over osteoblast mediated bone reformation results in structural weakening, causing collapse of the femoral head in many patients (Hernigou et al., 2004; Nishii, ⁎ Corresponding author. E-mail address: [email protected] (Z. Yan).
http://dx.doi.org/10.1016/j.yexmp.2015.01.002 0014-4800/© 2015 Elsevier Inc. All rights reserved.
2002). The mechanism of collapse is still controversial, however it is generally considered that the reparative reaction rather than the osteonecrosis itself is the primary cause (Li et al., 2009; Kim et al., 2006). To our knowledge, previous studies have mainly focused on parameters of the necrotic zone, including the location, the size, and the pathological changes of this region (Ha et al., 2006; Steinberg et al., 2006; Mont et al., 2010). Currently, the molecular mechanisms contributing to the pathogenesis of ONFH at the reparative interface remain poorly understood. MicroRNAs (miRNAs) belong to a family of non-protein-coding small RNAs that are involved in various physiological and pathological processes (Ambros, 2001; Bartel, 2004). They are approximately ~ 22 nucleotides in length and are expressed in a tissue- or cell-specific manner (Lagos-Quintana et al., 2002). Among them, a set of miRNAs have been confirmed to play fundamental roles in gene regulation of various orthopedic diseases, such as bone tumors (Kafchinski and Jones, 2014), osteoarthritis (Miyaki and Asahara, 2012), and rheumatoid arthritis (Furer et al., 2010). Although several miRNAs have been independently reported (Yamasaki et al., 2012; Sun et al., 2014; Jia et al., 2014), and one preliminary review discussing circulating microRNAs in ONFH has been written (Wang, X. et al., 2014; Wang, Y. et al., 2014), the miRNA expression profile of ONFH remains unclear. Therefore an in depth investigation
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
159
on the role of miRNA in ONFH will help us better understand the molecular mechanisms of this debilitating disease. The objective of our study is to provide a global view of miRNA expression in the reparative interface, and additionally present a novel and comprehensive molecular signature that contributes to the pathogenesis of ONFH that can be used to support future research efforts. We achieved this by (1) to determining the altered microRNA expression profile in the reparative interface of ONFH, (2) analyzing the results of our array using the Gene Ontology and Pathway assays, (3) and using the filtered miRNAs to predict pathways critical in ONFH. 2. Materials and methods 2.1. Ethical statement The study was reviewed and approved by the Ethical Committee of Zhongshan Hospital, Fudan University and patients gave informed consent. 2.2. Patients We selected 9 patients with ONFH who underwent total hip arthroplasty (THA) in our department. ONFH was diagnosed based on imaging examinations and the diagnosis was made according to the guidelines of the Chinese Medical Association for ONFH (Orthopedic Panel, 2012). We evaluated another 6 patients with fresh femoral neck fracture (FNF) who underwent THA in our department (excluding other types of bone and joint diseases) as the normal control group. All study characters were of Chinese Han nationality without any blood relationship to one another. The clinical characteristics of all patients are shown in Table 1. 2.3. Sample preparation When THA was performed, the isolated femoral head was obtained and cut along the coronal plane to check the appearance of the reparative interface zone. The reparative interface is the tissue on the edge of the necrotic zone (Fig. 1). The reparative interface was then cut into smalls pieces of approximately 5 × 5 × 5 mm3, and rapidly immersed in a portable liquid nitrogen tank. The samples from the tank were stored in a −80 °C freezer after leaving the operating room. 2.4. RNA isolation and microRNA microarray The samples were rapidly immersed the liquid nitrogen for 1 h and crushed by a pincer. Total RNA in the samples was isolated using Trizol Reagent (Invitrogen) according to the manufacture's protocol. The
Table 1 Clinical characteristics of the ONFH patients and controls. Patient
Age
Gender
Diagnosis
ONFH classification
Other illness
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
22 31 33 61 46 43 65 63 56 78 84 76 69 88 82
Female Female Female Female Female Female Male Male Male Female Female Female Female Male Female
Glucocorticoid-related ONFH Glucocorticoid-related ONFH Glucocorticoid-related ONFH Idiopathic ONFH Glucocorticoid-related ONFH Glucocorticoid-related ONFH Alcohol-related ONFH Alcohol-related ONFH Idiopathic ONFH FNF FNF FNF FNF FNF FNF
IIIB IIIC IIIB IIIC IIIC IIIC IIIC IV IIIC / / / / / /
SLE SLE SLE / RA SLE / / / / / / / /
Fig. 1. Samples acquired in the reparative interface of the femoral head in ONFH patients (△).
integrity and concentration of all RNA samples were quantified using the NanoDrop 1000 spectrophotometer (Thermo Scientific). All samples met the quality control standards. The total RNA of 3 patients diagnosed with ONFH (Patients 1–3) and 3 patients diagnosed with FNF (Patients 10–12) were hybridized to an Affymetrix GeneChip miRNA 4.0 Array containing 2578 human miRNA sequences. RVM t-test was applied to filter the differentially expressed miRNAs for the ONFH and normal control group because the RVM t-test can raise degrees of freedom effectively in the cases of small samples. After the significant analysis and FDR analysis, we selected the differentially expressed miRNAs according to the p value threshold. P value of b0.05 was considered as significant difference. 2.5. Bioinformatics analysis Bioinformatics analysis (Genminix Informatics Ltd., Shanghai, China) was performed for miRNAs expressed in significant amounts. We used the TargetScan (http://www.targetscan.org/) and miRanda (http://www.microrna.org/microrna/home.do) to predict the intersection of target genes of miRNAs. GO analysis was applied to analyze the main function of the differential expression target genes according to the Gene Ontology (GO, http://www.geneontology.org/) which is the key functional classification of NCBI. Pathway analysis was used to find out the significant pathway of the differential genes according to Kyoto Encyclopedia of Genes and Genomes (KEGG, http://www.genome.jp/ kegg/). GO terms and KEGG Pathway annotation of the miRNA target genes were found using the web-based tool, Database for Annotation, Visualization, and Integrated Discovery (DAVID, http://david.abcc.ncifcrf. gov/). Fisher's exact test and χ2 test were used to classify the GO category and select the significant pathway, and the false discovery rate (FDR) was used to correct the p value. We chose only GO terms with p value of b 0.01 and FDR value of b0.01, and Pathways with p value of b 0.05 and FDR value of b0.05. 2.6. Quantitative real-time PCR (RT-PCR) To validate the microarray data, we used the RNA-tailing and primer-extension RT-PCR method to detect the expression of the miRNA. RNA samples used in RT-PCR validation experiments were the
160
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
Table 2 Primers used in RT-PCR of the four genes. Gene name
Primer sequence
HIF-1α
Sense: 5′-TGATTGCATCTCCATCTCCTACC-3′ Antisense: 5′-GACTCAAAGCGACAGATAACACG-3′ Sense: 5′-CCCACTGAGGAGTCCAACAT-3′ Antisense: 5′-TCCAGGGCATTAGACAGCA-3′ Sense: 5′-CGAAGAGCAAGAATAAAT-3′ Antisense: 5′-GAATGAGGAAAGCAAACT-3′ Sense: 5′-TGGTGGGCTGCAGAAAATGGTT-3′ Antisense: 5′-ACGATGGCCGGCTTGTTGC-3′
VEGF-A Hes1 β-catenin
ones used in the array assay. RT-PCR was performed with a miRNA assay kit (RiboBio Co., Ltd., Guangzhou, China) according to the manufacture's protocol. U6 small nuclear RNA was used as an endogenous control. In addition, we tested two miRNAs (miR-34a and miR-146a) and four genes involved in the predicted pathways in another 6 patients (Patients 4–9) and 6 controls (Patients 10–15). β-actin was used as an endogenous control for the four genes. The primers of the four genes are described in Table 2. The 7500 Real-Time PCR System (ABI) was used for amplification and detection. The CT values were obtained from the amplification plot using SDS software. Finally, the relative miRNA level was normalized to U6 expression, and the gene level to β-actin expression for each sample in triplicate and was expressed as 2−△△CT. The data are presented as the means ± the standard deviations (S.D.s) and were analyzed with one-way ANOVA using SPSS 16.0 (SPSS Inc., Chicago, IL, USA). A p value of b0.05 was considered statistically significant. 3. Results 3.1. MiRNA expression patterns in the reparative interface of ONFH We used Affymetrix GeneChip miRNA 4.0 Array to analyze miRNAs expression in the reparative region, and applied the RVM t-test to filter
Fig. 3. Validation of miRNAs by RT-PCR. The expression levels of the six miRNAs were in concordance with the normalized microarray data.
the differentially expressed miRNAs. Twelve miRNAs (miR-181c-3p, miR-34a-3p, miR-146a-5p, miR-187-3p, miR-181a-3p, miR-30c-1-3p, miR-650, miR-3652, miR-4444, miR-1273e, miR-99a-3p, miR-30645p) were found to be significantly up-regulated (p b 0.05) while five miRNAs (miR-212-3p, miR-212-5p, miR-132-3p, miR-629-3p, miR6836-5p) were markedly down-regulated (p b 0.05) (Fig. 2). In order to validate the microarray results, six representative miRNAs were chosen for RT-PCR using the samples that were differentially expressed in the array assay. The expression levels of all the selected miRNAs were in conformity with the normalized microarray data (Fig. 3).
Fig. 2. The miRNA profile differentiates ONFH patients from normal individuals. (A) Hierarchical clustering for the differentially expressed miRNAs (p b 0.05); (B) Scatter plot for all expressed miRNAs, green spots represent down-regulated miRNAs, red spots represent up-regulated miRNAs (p b 0.05); (C) Volcano plot of the microarray data, red spots represent down-regulated miRNAs, blue spots represent up-regulated miRNAs (p b 0.05). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
161
Fig. 4. GO category analysis based on miRNA-targeted genes (p value b 0.01, FDR b 0.05). The vertical axis represents the GO category, and the horizontal axis represents the enrichment of GO.
3.2. Microarray-based Gene Ontology and Pathway analysis According to the GO and KEGG database, we gained functions and pathways of all the predicted target genes of the differentially expressed miRNAs. Fisher's exact test and χ2 test were used to calculate the p value and FDR. According to the standard of p b 0.01 (GO) and p b 0.05 (KEGG), significant functions and pathways were filtered. As shown in Fig. 4, GO results demonstrate upregulation of pathways related to signal transduction, cell differentiation, cell methylation, cell growth and apoptosic processes. In addition, the target genes were also enriched into biological
pathways, such as angiogenesis related pathways, Wnt signaling pathway and PI3K–Akt signaling pathway (Fig. 5). 3.3. Expression of the screened miRNAs and related pathways We tested for expression of miR-34a and miR-146a using the RNAtailing and primer-extension RT-PCR method in patient samples that were independent from the array assay. The results confirmed that expressions of miR-34a and miR-146a were up-regulated by 4 and 5 folds respectively in the patients of ONFH when compared with the
Fig. 5. Pathway analysis based on miRNA-targeted genes (p value b 0.05). The vertical axis represents the pathway category, and the horizontal axis represents the value of −LgP.
162
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
Fig. 6. Expression of miR-34a and miR-146a in the six independent patients of ONFH compared with the controls (***p b 0.001, *p b 0.05).
control samples (Fig. 6). Expression of the predicted target genes for each miR-34a and miR-146a, including HIF-1α in HIF-1 signaling pathway, VEGF-A in VEGF signaling pathway, Hes1 in Notch signaling pathway and β-catenin in Wnt signaling pathway, were all significantly upregulated (p b 0.05) (Fig. 7). 4. Discussion The reparative process plays a critical role in the clinical course and symptoms of ONFH, and is generally considered to be initiated after the occurrence of death of bone or marrow elements (Mont et al., 1998). However, when the lesion is relatively small (15–20 mm) or medially located, the reparative reaction typically has no specific repair function (Lieberman et al., 2012; Glimcher and Kenzora, 1979). While the mechanism is not entirely elucidated, it has been described that in the reparative margin of the infarction, dead trabeculae are resorbed by osteoclasts and are replaced with appositional new bone, resulting in thickened reinforced trabeculae. The thickened interface could prevent penetration of revascularization into the necrotic zone and cause
irreversible damage, possibly leading to eventual collapse (Mont et al., 1998). Further investigations are needed to better understand the reparative process of osteonecrosis so that appropriate preventative or remedial treatment can be applied. In this study we have demonstrated that a number of miRNAs are significantly differentially expressed in the reparative interface. Some of these miRNAs have been characterized for their important roles in bone and joint diseases. MiR-146a has been reported to inhibit osteoclastogenesis and prevent joint destruction in arthritic mice, presenting as a potential novel therapeutic target for bone destruction in rheumatoid arthritis (Nakasa et al., 2011; Ammari et al., 2013). Another specific miRNA, miR-34a, is linked to inhibition of angiogenesis by blocking VEGF production, regulation of cell apoptosis, and additionally has been shown to inhibit the proliferation and metastasis of osteosarcoma cells both in vitro and in vivo (Kumar et al., 2012; Tian et al., 2014; Yan et al., 2012). In order to obtain insights into the classification of miRNAs, we performed GO analysis in order to predict target genes. The result revealed several GOs related to signal transduction including adenosine receptor
Fig. 7. Expression of four genes involved in the predict pathways. A. Expression of HIF-1α in HIF-1 signaling pathway (*p b 0.05); B. Expression of VEGF-A in VEGF signaling pathway (**p b 0.01); C. Expression of Hes1 in Notch signaling pathway (**p b 0.01); D. Expression of β-catenin in Wnt signaling pathway (***p b 0.001).
H. Yuan et al. / Experimental and Molecular Pathology 98 (2015) 158–163
signaling pathway, regulation of dephosphorylation, cGMP catabolic process, transcription initiation from RNA polymerase II promoter, and signal transduction. In addition we found correlations with cell differentiation (osteoclast differentiation, fat cell differentiation, regulation of cell proliferation, chondrocyte differentiation), cell methylation (histone H4-K20 methylation, positive regulation of histone H3-K9 methylation), cell growth (cell cycle, Notch signaling pathway, regulation of cell proliferation) and apoptosis (apoptotic signaling pathway, negative regulation of apoptotic process). Together these groups represent the most of the significantly enriched GO terms. Using KEGG Pathway analysis we were able to functionally categorize predicted target genes. We found strong correlations to angiogenesis related pathways such as HIF-1 signaling pathway, VEGF signaling pathway and Notch signaling pathway. This suggests that angiogenesis plays a crucial role in the reparative process. In addition, the Wnt signaling pathway, which has been reported to play a pivotal role in regulating the balance between bone formation and bone resorption (Zhong et al., 2014; Wang, X. et al., 2014; Wang, Y. et al., 2014), was enriched in our results. We believe this pathway could be responsible for some of the pathological changes observed in the reparative interface, and may provide an avenue for therapeutic intervention, warranting further investigation. PI3K–Akt signaling pathway, which has been reported to be closely relevant with SLE (Beşliu, 2009), was also found in our study. We believe this is reasonable since the ONFH samples we used in the array analysis are all complicated with SLE. In conclusion, our study unveiled the dysregulation of miRNA in the reparative interface of ONFH. Bioinformatics-based analysis of these changes in expression may provide a useful tool for understanding the molecular mechanisms responsible for collapse of the femoral head. Further investigations are needed to clarify the exact roles of identified miRNA in the pathogenesis of reparative process. Disclosure/conflict of interest The authors have no conflict of interest to declare. Acknowledgments This research was supported by the National Natural Science Foundation of China (No. 81371966) and International Cooperation Project of Shanghai Science and Technology Committee (No. 12410710200). References Ambros, V., 2001. microRNAs: tiny regulators with great potential. Cell 107, 823–826. Ammari, M., et al., 2013. Impact of microRNAs on the understanding and treatment of rheumatoid arthritis. Curr. Opin. Rheumatol. 25, 225–233. Assouline-Dayan, Y., 2002. Pathogenesis and natural history of osteonecrosis. Semin. Arthritis Rheum. 32, 94–124. Bartel, D.P., 2004. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116, 281–297. Beşliu, A.N., 2009. PI3K/Akt signaling in peripheral T lymphocytes from systemic lupus erythematosus patients. Roum. Arch. Microbiol. Immunol. 68, 69–79. Fukushima, W., et al., 2010. Nationwide epidemiologic survey of idiopathic osteonecrosis of the femoral head. Clin. Orthop. Relat. Res. 468, 2715–2724.
163
Furer, V., et al., 2010. The role of microRNA in rheumatoid arthritis and other autoimmune diseases. Clin. Immunol. 136, 1–15. Glimcher, M.J., Kenzora, J.E., 1979. The biology of osteonecrosis of the human femoral head and its clinical implications: II. The pathological changes in the femoral head as an organ and in the hip joint. Clin. Orthop. Relat. Res. 283–312. Ha, Y.C., et al., 2006. Prediction of collapse in femoral head osteonecrosis: a modified Kerboul method with use of magnetic resonance images. J. Bone Joint Surg. Am. 88 (Suppl. 3), 35–40. Hernigou, P., et al., 2004. Fate of very small asymptomatic stage-I osteonecrotic lesions of the hip. J. Bone Joint Surg. Am. 86-A, 2589–2593. Jia, J., et al., 2014. MiR-17-5p modulates osteoblastic differentiation and cell proliferation by targeting SMAD7 in non-traumatic osteonecrosis. Exp. Mol. Med. 46, e107. Johnson, A.J., et al., 2014. Treatment of femoral head osteonecrosis in the United States: 16-year analysis of the Nationwide Inpatient Sample. Clin. Orthop. Relat. Res. 472, 617–623. Kafchinski, L.A., Jones, K.B., 2014. MicroRNAs in osteosarcomagenesis. Adv. Exp. Med. Biol. 804, 119–127. Kim, H.K., et al., 2006. RANKL inhibition: a novel strategy to decrease femoral head deformity after ischemic osteonecrosis. J. Bone Miner. Res. 21, 1946–1954. Kumar, B., et al., 2012. Dysregulation of microRNA-34a expression in head and neck squamous cell carcinoma promotes tumor growth and tumor angiogenesis. PLoS One 7, e37601. Lagos-Quintana, M., et al., 2002. Identification of tissue-specific microRNAs from mouse. Curr. Biol. 12, 735–739. Li, W., et al., 2009. Distribution of TRAP-positive cells and expression of HIF-1alpha, VEGF, and FGF-2 in the reparative reaction in patients with osteonecrosis of the femoral head. J. Orthop. Res. 27, 694–700. Lieberman, J.R., et al., 2012. Which factors influence preservation of the osteonecrotic femoral head? Clin. Orthop. Relat. Res. 470, 525–534. Miyaki, S., Asahara, H., 2012. Macro view of microRNA function in osteoarthritis. Nat. Rev. Rheumatol. 8, 543–552. Mont, M.A., et al., 1998. Osteonecrosis of the femoral head. Potential treatment with growth and differentiation factors. Clin. Orthop. Relat. Res. (355 Suppl.), S314–S335. Mont, M.A., et al., 2010. The natural history of untreated asymptomatic osteonecrosis of the femoral head: a systematic literature review. J. Bone Joint Surg. Am. 92, 2165–2170. Nakasa, T., et al., 2011. The inhibitory effect of microRNA-146a expression on bone destruction in collagen-induced arthritis. Arthritis Rheum. 63, 1582–1590. Nishii, T., 2002. Progression and cessation of collapse in osteonecrosis of the femoral head. Clin. Orthop. Relat. Res. 149–57. Orthopedic Panel, 2012. Diagnostic criteria for adult osteonecrosis (2012 edition). Chin. J. Orthop. 32, 606–610. Schroer, W.C., 1994. Current concepts on the pathogenesis of osteonecrosis of the femoral head. Orthop. Rev. 23, 487–497. Steinberg, D.R., et al., 2006. Determining lesion size in osteonecrosis of the femoral head. J. Bone Joint Surg. Am. 88 (Suppl. 3), 27–34. Sun, J., et al., 2014. Downregulation of PPARγ by miR-548d-5p suppresses the adipogenic differentiation of human bone marrow mesenchymal stem cells and enhances their osteogenic potential. J. Transl. Med. 12, 168. Tian, Y., et al., 2014. MicroRNA-199a-3p and microRNA-34a regulate apoptosis in human osteosarcoma cells. Biosci. Rep. 34, 479–485. Wang, X., et al., 2014. Preliminary screening of differentially expressed circulating microRNAs in patients with steroid-induced osteonecrosis of the femoral head. Mol. Med. Rep. 10, 3118–3124. Wang, Y., et al., 2014. Wnt and the Wnt signaling pathway in bone development and disease. Front. Biosci. 19, 379–407. Yamasaki, K., et al., 2012. Angiogenic microRNA-210 is present in cells surrounding osteonecrosis. J. Orthop. Res. 30, 1263–1270. Yan, K., et al., 2012. MicroRNA-34a inhibits the proliferation and metastasis of osteosarcoma cells both in vitro and in vivo. PLoS One 7, e33778. Zalavras, C.G., Lieberman, J.R., 2014. Osteonecrosis of the femoral head: evaluation and treatment. J. Am. Acad. Orthop. Surg. 22, 455–464. Zhong, Z., et al., 2014. WNT signaling in bone development and homeostasis. Wiley Interdiscip. Rev. Dev. Biol. 3, 489–500.
