Analysis of circulating microRNAs aberrantly expressed in alcohol-induced osteonecrosis of femoral head

Serum miRNAs are potential biomarkers for predicting the progress of bone diseases, but little is known about miRNAs in alcohol-induced osteonecrosis of femoral head (AIONFH). This study evaluated disease-prevention value of specific serum miRNA expression profiles in AIONFH. MiRNA PCR Panel was taken to explore specific miRNAs in serum of AIONFH cases. The top differentially miRNAs were further validated by RT-qPCR assay in serum and bone tissues of two independent cohorts. Their biofunction and target genes were predicted by bioinformatics databases. Target genes related with angiogenesis and osteogenesis were quantified by RT-qPCR in necrotic bone tissue. Our findings demonstrated that multiple miRNAs were evaluated to be differentially expressed with high dignostic values. MiR-127-3p, miR-628-3p, and miR-1 were downregulated, whereas miR-885-5p, miR-483-3p, and miR-483-5p were upregulated in serum and bone samples from the AIONFH patients compared to those from the normal control individuals (p < 0.01). The predicted target genes of the indicated miRNAs quantified by qRT-PCR, including IGF2, PDGFA, RUNX2, PTEN, and VEGF, were presumed to be altered in necrotic bone tissue of AIONFH patients. The presence of five altered miRNAs in AIONFH patients may serve as non-invasive biomarkers and potential therapeutic targets for the early diagnosis of AIONFH.

Alcohol-induced osteonecrosis of the femoral head (AIONFH), as defined by World Health Organization (WHO), is characterized by ischemia to the bone and bone marrow tissues within the constituents of the femoral head due to alcohol consumption 1 . AIONFH is an unpreventable and devastating disease. If left untreated, AIONFH can result in collapse of the femoral head that necessitates hip replacement in approximately 80% of patients 2,3 . While there are various theories regarding the causes of AIONFH, such as the extravascular accumulation of thrombosis mediating vascular constriction, the intravascular blood coagulation theory, the pathogenesis of AIONFH is still unclear 4,5 .
There is currently a lack of accurate and early diagnostic approach for ONFH, and only histological examination and magnetic resonance imaging (MRI) have been employed to date 6 . However, only if patients undergo hip arthroplasty in late stages of the disease (Association Research Circulation Osseous (ARCO) III and IV phase), histological examination can be utilized in the evaluation of the isolated femoral head. MRI enables to detect the AIONFH early to ARCO I phase, but the images captured in this phase are usually at low resolution and easy to be misdiagnosed by physicians. Investigation of microRNAs (miRNAs) in osteonecrosis of femoral head (ONFH) is beneficial to characterization of cellular cross-talk between endothelial cells and osteocytes as well as identification of detailed molecular mechanisms and biomarker regulation patterns. A handful of possible biomarkers for steroid-induced osteonecrosis of femoral head (SIONFH) have been reported [7][8][9] , suggesting that serum biomarkers may be also used to feasibly detect AIONFH in an earlier time point. Furthermore, the screening of biomarkers in AIONFH patients may contribute to the exploration of the microfluidics-based PCR used in AIONFH. However, specific biomarkers in AIONFH patients have not been well established. In this study, we focused on the investigation of specific miRNAs and aimed to identify potentially novel diagnostic biomarkers of AIONFH.
The interest in identifying miRNAs as diagnostic biomarkers in recent years has been focused on serum miRNAs 10,11 . An miRNA is a noncoding, short RNA segment approximately 22 nucleotides in length 12 . In bone microenvironment, miRNAs are considered to be involved in the proliferation, differentiation, and apoptosis of osteoclasts, osteoblasts and osteocytes 13 . They regulate the expression of mRNA in the posttranscription process by inducing the RNA silencing complex 14 . There are two main functions for the silencing mechanism: one results in mRNA target cleavage, and the other leads to mRNA degradation or repression of protein translation. In the differentially expressed miRNAs in SIONFH patients with systemic lupus erythematosus (SLE), 15 miRNAs have been found to be overexpressed while 12 to be downregulated 9 . Other experiments with various gene chips and microarrays have been performed in cells extracted from human tissues or animal model receiving steroid interventions and showed that miRNAs may also play a significant role in regulating bone cells and bone homeostasis [15][16][17][18][19] . However, the role of serum miRNAs is still unclear in human AIONFH patients.
AIONFH is an unpreventable musculoskeletal disease occurring in the femoral head that frequently requires total hip arthroplasty. Hence, it is imperative to identify noninvasive diagnostic and predictive biomarkers for AIONFH. In this study, we investigated whether specific miRNAs are differentially expressed in patients with AIONFH. Our results revealed miRNAs as novel potential biomarkers for the early diagnosis of AIONFH and might promote development of innovative treatment for AIONFH.

Materials and Methods
Subjects and samples. A total of forty AIONFH subjects were recruited from Department of Orthopedic, the First Affliated Hospital of Guangzhou University of Chinese Medicine in Guangzhou, China between June 2016 to Feburary 2017. AIONFH was diagnosed based on MRI evaluations and then classified according to the Association Research Circulation Osseous (ARCO) Stage system. Patients were invovled with the following criteria to reduce the unexpected impacts of biases: (a) with no interventions contraindicated, (b) age younger than 50 years; (c) presumptive long-term alcohol consumption; and (d) MRI and X-ray examinations revealing ARCO stage II to III disease duration. The exclusion criteria were as follows: (a) personal history of taking steroidal hormones; (b) ARCO stage IV. We also recruited the same number of age−/sex-matched volunteers as the control group. The inclusion criteria for the control group were as follows: (1) healthy individuals with no history of serious diseases or hormone use; (2) without ONFH; (3) suffered from traumatic femoral neck fracture or underwent total hip arthroplasty. All the study participants were of Chinese Han nationality with no blood relationship to one another. The detailed personal characteristics of the patient and control groups were profiled in Sup. Table 1. Participants involved in the study have been informed by details of study process and written informed consents were obtained from all participants. RNA extraction and miRNA expression profiles. Total miRNAs were measured from serum consisting of five AIONFH serum samples and five healthy serum samples to identify differentially expressed miRNAs. miRNAs were isolated and extracted from serum in a standard method according to the manufacturers' recommendations (Invitrogen, Carlsbad, CA, USA). Briefly, the serum samples from both groups were centrifuged at 13000 × g for 5 min at 4 °C to eliminate pellet and cell debris. The supernatant serum was collected and added to TRIzol-LS (Invitrogen, Carlsbad, CA, USA), and then high-speed centrifugation was performed to obtain the aqueous phase. To precipitate the total RNA, serum was mixed with isopropyl alcohol and subjected to one further high-speed centrifugation at 10000 × g for 5 min at 4 °C. The total RNA was eluted in RNase-free water and stored at −80 °C until analysis. The amount and purity of RNA was estimated by ultraviolet spectrophotometer.
MiRNAs were profiled with miRCURY LNA TM Universal RT microRNA PCR Panels (Exiqon, Vedbaek, Denmark) as described 20 . Briefly, RNA samples were diluted to 1.5-1.8 ng/μl using nuclease-free water. The specific miRNA primers for reverse transcription (RT) were used to amplify miRNAs in an RT reaction mix (Exiqon, Denmark) and subjected to amplification using SYBR ™ Green Master Mix (Exiqon, Denmark) with an ABI PRISM 7900 Real-time PCR System (Applied Biosystems, USA). Quantitative miRNA expression levels were analyzed by 2 −ΔCt/ΔCt calculation with GenEx qPCR analysis software (Exiqon, Vedbaek, Denmark). Fold changes >2-fold or p < 0.05 were used to identify miRNAs that had significantly altered expression between the patient group and the healthy control group.

miRNA validation by Real-time PCR.
To further validate the miRNA array data, the differentially expressed miRNAs were analyzed with quantitative RT-PCR (qRT-PCR) in duplicate using an miRNA assay kit (GenePharma, Shanghai, China) in 30 serum samples (15 AIONFH patients and 15 healthy controls with femoral neck fracture) and 30 bone samples (15 AIONFH patients and 15 healthy controls with femoral neck fracture) as described 21,22 . Serum and femoral head isolated or detached from patients and volunteers were stored in −80 °C. Necrotic bone tissue was partially evaluated by histological examination and tissue with empty lacuna rate more than 50% was identified as necrotic area (Sup. Fig. 1). To isolate miRNA from serum and bone tissue, extraction with TriReagent (Invitrogen life technologies) was performed after milling of the tissue using liquid nitrogen.
The RT reaction was performed under the following reaction conditions: 30 min at 25 °C, 30 min at 42 °C, and 5 min at 85 °C, followed by maintenance at 4 °C. The selected miRNAs were confirmed with SYBR Green I dye (Takara, Dalian, China) with an ABI PRISM 7300 Real-time PCR System (Applied Biosystems, USA) at 95 °C for 3 min, followed by 40 cycles at 95 °C for 12 s and 62 °C for 40 s. Cel-miRNA-39-3p, a nonhuman miRNA, was spiked into the RNA samples as a control for the extraction and amplification steps. GAPDH was used for normalization of serum samples following the miRNA PCR array analysis.
Furthermore, receiver operating characteristic (ROC) curves were used to determine the diagnostic potential of serum miRNAs, which were generated after logarithmic transformation of all the samples included in the validation.
Bioinformatics analysis of miRNA data. To estimate the potential biofunction and signaling pathways as well as target genes of miRNAs validated by real-time PCR, several bioinformatics analysis technologies were employed. Known putative targets of the selected miRNAs were generated by total three gene databases: TargetScan (http://www.targetscan.org/), miRanda (http://www.microrna.org/microrna/home.do), and MiRDB (http://www.mirbase.org). The predicted target genes from the three gene databases were summarized and ranked in order to obtain objective results with context scores and to minimize the false positive rates. Gene ontology (GO) analysis was utilized to identify the main biofunctions of the specific differentially expressed miRNAs, according to GO (http://www.geneontology.org/). The Kyoto Encyclopedia of Genes and Genomes (KEGG, http://www.genome.jp/kegg/) was used as a tool in our study to manage pathway analysis. All the involved data were considered based on their conserved sequences and the prospect of regulation of genes or pathways via miRNA activity.
Quantitative real-time PCR of angiogenesis-and osteogenesis-related genes. The qRT-PCR system was used to quantified miRNAs related to mechanisms of osteogenesis and angiogenesis. These include insulin-like growth factor 2 (IGF2), semaphorin 3D (SEMA3D), runt-related transcription factor 2 (RUNX2), platelet-derived growth factor subunit A (PDGFA), superoxide dismutase 1 (SOD1), telomerase-associated protein 1 (TEP1), and vascular endothelial growth factor A (VEGFA). cDNA from ten necrotic bone tissue samples were used to compare with ten control bone tissue samples from femoral neck fracture patients. The relative expression levels of each gene were determined by qRT-PCR using the 2−ΔΔ CT method as described previously 23 , with β-actin as a normalized control. The sequences of both the forward and reverse primers of all genes involved are listed in Sup. Table 2.
Statistical analyses. All statistical results are performed using SPSS (SPSS, IBM, USA) and GraphPad Prism (GraphPad Software, San Diego, CA, USA). Data was presented as the mean ± SD. Differences in the relative amount of miRNA between AIONFH patients and those in the control group were compared by using Student's t-test or Welch's t-test for equal or unequal variance. A two-tailed Mann-Whitney U test was utilized to compare different expression levels from samples from AIONFH patients compared with those from non-AIONFH patients. Fisher's exact test and the 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-values < 0.01 and FDRs <0.01, and KEGG with p-values < 0.05 and FDRs <0.05. To identify the diagnostic sensitivity and specificity of serum miRNAs, ROC curves were used. The p-value tested the null hypothesis that the area under the curve was equal to 0.50. The cut-off points with the highest sensitivity and specificity were measured. The minimum level for significant differences was p < 0.05 or p < 0.01.

miRNA PCR array-based GO and pathway analysis.
We explored the functions and potential signaling pathways of all the selected miRNAs with differential expression using the GO database and KEGG pathway analysis. Fisher's exact test and the χ2 test were used to calculate p-values and FDRs. According to the standards of p < 0.05 (GO) and p < 0.05 (KEGG), significant functions and pathways were filtered. As shown in Sup. Fig. 2, the GO results demonstrated enrolled pathways related to signal transduction, cell differentiation, cell methylation, cell growth and apoptotic processes. The KEGG results showed that the Wnt signaling pathway and the PI3K-Akt and cancer pathways had the highest correlation with the selected miRNAs (Sup. Fig. 3).

NAs. Sequencing analyses of the TargetScan, miRanda Microcosm and MiRDB databases demonstrated that
several genes were predicted to be induced targets mediated by the potential miRNAs for both osteoblasts and osteoclasts (Sup. Table 3 and Sup. Fig. 4). In addition, several potential genes related to angiogenesis and osteogenesis functions were found to be correlated with the identified miRNAs in the literature research (Pubmed, https://www.ncbi.nlm.nih.gov/pubmed/). To declare the correlation between indicated miRNAs and their predicted target genes, the mRNA transcript in necrotic bone samples from AIONFH patients was tested (Fig. 5). In our report, VEGFA and PDGFA are predicted to be upregulated by miR-1 and VEGFA is modulated by miR-127-3p. By contrast, PTEN and RUNX2 are predicted to be suppressed by miR-628-3p and IGF2 is downregulated by miR-483-5p. However, the expression of SEMA3D, which is regulated by miR-885-5p, and SOD1, which is regulated by miR-1, are not significantly different in samples from AIONFH individuals.

Discussion
In ONFH, interruption of vessels and an imbalance in osteoblast-osteoclast coupling causes alterations in bone mass and structure. miRNAs play a critical role in skeletal diseases such as osteonecrosis 6,19 . The diagnostic potential of measuring miRNAs in the serum of ONFH patients has been reported in patients with SIONFH. In particular, several miRNAs present in ONFH tissues and cells have been correlated with osteogenesis 18,19,25 . In the mesenchymal stem cells (MSCs) of patients with ONFH, SMAD3 and SMAD7 have been identified as the target genes of miR-708 and miR-17-5p, respectively, and their downstream signaling pathways are involved in osteogenic differentiation 7,25 . In hormone-induced rat models of osteonecrosis, miR-672-5p and miR-146a have been found to improve osteoblast formation 26 . Furthermore, some miRNAs involved in angiogenesis were also found to enhance the activity of angiogenic factors including VEGF, basic fibroblast growth factor (bFGF), tumor necrosis factor alpha (TNF-α) and proliferating cell nuclear antigen (PCNA) 8 , while some miRNAs, such as miR-34a, were found to be active in osteoblastic differentiation and endothelial coupling activity 27 .
To better understand the underlying mechanisms of AIONFH and to develop a potential early diagnostic biomarker for AIONFH, we investigated differentially expressed miRNAs in the serum of AIONFH patients with microarray analyses and identified seven specific miRNAs by RT-PCR. Among six of them, miR-127, miR-628-3p Figure 2. miRNAs are differently expressed in serum of AIONFH patients. Scatter plots provide the expression levels of specific miRNA in serum of AIONFH patients (n = 15) compared with health control group patients (n = 15). The expression of miR-127, miR-628-3p, miR-1, miR-432-5p, miR-885-5p, miR-483-5p, miR-483-3p were found to be significantly different (*p < 0.05, **p < 0.01 for the comparison indicated by Mann-Whitney U test).  www.nature.com/scientificreports www.nature.com/scientificreports/ and miR-1 were downregulated, whereas miR-885-5p, miR-483-3p, and miR-483-5p were upregulated both in the serum and necrotic bone tissue samples from AIONFH patients. The expression level of each miRNA in the serum of AIONFH patients and controls was identified by relative concentration analysis with miR-432-5p, which is easy to yield better conclusion by comparing them with each other. In addition, we established the sensitivity and specificity with measurements of miRNAs using ROC curve to identify their diagnostic value in AIONFH.  www.nature.com/scientificreports www.nature.com/scientificreports/ All the above mentioned miRNAs identified in the serum of participants in our experiment, except for miR-885-5p, have effects on osteoblast, osteoclast and endothelial cell development for osteonecrosis healing. Induction of these particular miRNAs was highly effective in posttranscriptional control, which inhibited or enhanced the expression of various cytokines, thus allowing regulation of angiogenesis and osteogenesis in AIONFH. To establish the network of miRNAs posttranscriptional control and figure out their downstream genes, we combined the results of target genes prediction and literature review, and have quantitative measurements of those involved factors by using qRT-PCR.

mRNA ID Forward and reverse primers bp
As previously reported, overexpression miR-628-3p has a suppressive effect on osteogenesis and osteoblast differentiation via the downregulation of RUNX2 mRNA or protein levels 28 . We found 0.4-fold significant downregulation of miR-628-3p in the serum and bone of AIONFH patients. The downregulation may lead to the suppression of RUNX2 gene expression, resulting in the absence of more mature osteoblasts in AIONFH patients than in control individuals, which may explain the bone loss and microstructural disorders.
MiR-483-5p, which are both encoded within the intron of host IGF2 gene, acts as suppressive factors 29,30 . In osteoporosis, only upregulation of miR-483-5p has a negative effect on IGF2 and inhibits osteoblast differentiation by combining with the 5′-UTR of the fetal IGF2 promoter transcript 31,32 . Our investigation in serum showed a strong upregulation of miR-483-5p, and the PCR analysis showed that IGF2 expression was suppressed in bone samples from AIONFH patients, which suggested that miR-483 was sufficient to predict the disease precisely and inhibit osteoblast in AIONFH through IGF2. It is also mentioned that upregulation of miR-483-5p and miR-483-3p have an negative effect on the migration endothelial progenitor cells and tube formation by targeting in serum response factors 29,30 . However, it is not consistent with the upregulated trend of VEGF. Further studies are required to clarified the competing results. No potential correlation with target genes was found in miR-483-3p.  Table 3. Sensitivity and Specificity of the Regulated miRNAs in the Serum of AIONFH and Non-AIONFH Patients. www.nature.com/scientificreports www.nature.com/scientificreports/ MiR-127 contributes greatly to osteogenesis and angiogenesis in microenvironment. Specifically, the downregulated expression of miR-127 precursors has a potential effect on improving osteoblast mineralization and suppressing osteoclast differentiation in ovariectomized mice 33 . It is indicated that miR-127 is greatly downregulated in our study. Hence, we hypothesized that miR-127 may act as a reactive and positive role in AIONFH even osteoblast differentiation was suppressed in the femoral head. One analysis of ionizing radiation incidentally revealed a correlation between downregulated miR-127 and increased sensitivity of endothelial cells 34 . Due to a positive effect on angiogenic factors miR-127 hold, we considered the upregulated expression of VEGF is correlated to the identified miR-127 in AIONFH patients. This may lead to compensatory increases in vascularization in necrotic bone tissues.
The function of miR-1 has been comprehensively investigated in multiple previous reports. The critical role of miR-1 in angiogenesis was first identified in cardiovascular repair 35 . As mentioned, miR-1 is highly expressed in cardiomyocytes and cardiomyogenesis 36,37 . Lu also confirmed the specific function of miR-1 in cardiomyocytes and identified human frizzled-7 (FZD7) and fibroblast growth factor receptor substrate 2 (FRS2) as direct targets of miR-1 38 . In addition, miR-1 increased the expression levels of VEGFA in muscle in a zebrafish model and induced angiogenic signaling in the endothelium 39 , while the opposite results were found in osteosarcoma cells and gastric cells 40,41 . Furthermore, enhancer of zeste homolog 2 (EZH2) can suppress miR-1 transcription and promote angiogenesis 42 . In contrast, other reports have found that miR-1 increases the proliferation and migration or invasion of tumor endothelial cells 43,44 . During osteogenesis, miR-1 has been found to be related to VEGFA, FGF2, etc 45 . Interestingly, we detected lower expression levels of miR-1 and higher expression of VEGF in bone samples from AIONFH patients than in those from the control participants. However, regarding to bone metabolism there was still no direct evidence proving the correlation between PDGFA and miR-1 in literature study, even it was predicted in GO and KEGG analysis.
Taken together, our research evaluated serum miRNAs in AIONFH patients and investigated seven miRNAs that were differentially expressed in serum samples from these patients. In necrotic bone tissue, six differentially expressed miRNAs were identified. Three of these were concurrently downregulated in serum while the others were upregulated. These several significantly different circulating miRNAs might serve as novel biomarkers for the early diagnosis of AIONFH, as serum collection is convenient and noninvasive. Furthermore, our findings might provide potential novel targets for the pharmacological treatment of AIONFH in the future.