Inhibition of miR-92a Enhances Fracture Healing via Promoting Angiogenesis in a Model of Stabilized Fracture in Young Mice

Authors

  • Koichi Murata,

    1. Department of Orthopaedic Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan
    Search for more papers by this author
  • Hiromu Ito,

    Corresponding author
    1. Department of Orthopaedic Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan
    2. Department of the Control for Rheumatic Diseases, Kyoto University Graduate School of Medicine, Kyoto, Japan
    • Address correspondence to: Hiromu Ito, MD, PhD, Department of the Control for Rheumatic Diseases, Kyoto University Graduate School of Medicine, 54 Kawahara-cho, Shogoin, Sakyo, Kyoto 606-8507, Japan. E-mail: hiromu@kuhp.kyoto-u.ac.jp

    Search for more papers by this author
  • Hiroyuki Yoshitomi,

    1. Department of Orthopaedic Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan
    2. Center for Innovation in Immunoregulative Technology and Therapeutics, Kyoto University Graduate School of Medicine, Kyoto, Japan
    Search for more papers by this author
  • Koji Yamamoto,

    1. Department of Orthopaedic Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan
    Search for more papers by this author
  • Akinobu Fukuda,

    1. Department of Orthopaedic Surgery, Yoshikawa Hospital, Kyoto, Japan
    Search for more papers by this author
  • Junsuke Yoshikawa,

    1. Department of Orthopaedic Surgery, Yoshikawa Hospital, Kyoto, Japan
    Search for more papers by this author
  • Moritoshi Furu,

    1. Department of Orthopaedic Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan
    2. Department of the Control for Rheumatic Diseases, Kyoto University Graduate School of Medicine, Kyoto, Japan
    Search for more papers by this author
  • Masahiro Ishikawa,

    1. Department of Orthopaedic Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan
    Search for more papers by this author
  • Hideyuki Shibuya,

    1. Department of Orthopaedic Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan
    Search for more papers by this author
  • Shuichi Matsuda

    1. Department of Orthopaedic Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan
    Search for more papers by this author

ABSTRACT

MicroRNAs (miRNAs) are endogenous small noncoding RNAs regulating the activities of target mRNAs and cellular processes. Although no miRNA has been reported to play an important role in the regulation of fracture healing, several miRNAs control key elements in tissue repair processes such as inflammation, hypoxia response, angiogenesis, stem cell differentiation, osteogenesis, and chondrogenesis. We compared the plasma concentrations of 134 miRNAs in 4 patients with trochanteric fractures and 4 healthy controls (HCs), and the levels of six miRNAs were dysregulated. Among these miRNAs, miR-92a levels were significantly decreased 24 hours after fracture, compared to HCs. In patients with a trochanteric fracture or a lumbar compression fracture, the plasma concentrations of miR-92a were lower on days 7 and 14, but had recovered on day 21 after the surgery or injury. To determine whether systemic downregulation of miR-92a can modulate fracture healing, we administered antimir-92a, designed using locked nucleic acid technology to inhibit miR-92a, to mice with a femoral fracture. Systemic administration of antimir-92a twice a week increased the callus volume and enhanced fracture healing. Enhancement of fracture healing was also observed after local administration of antimir-92a. Neovascularization was increased in mice treated with antimir-92a. These results suggest that plasma miR-92a plays a crucial role in bone fracture healing in human and that inhibition of miR-92a enhances fracture healing through angiogenesis and has therapeutic potential for bone repair. © 2014 American Society for Bone and Mineral Research.

Introduction

Fracture healing is a complex process orchestrated by a precise sequence of growth factors and cytokines that control activation, proliferation, and differentiation of local mesenchymal stem or progenitor cells. Under optimal conditions, a fractured bone heals without any scar formation and fully recovers its morphological and biomechanical properties. However, accelerated bone regeneration is required in demanding situations such as skeletal reconstruction of large bone defects created by trauma, infection, tumor resection, or skeletal abnormalities. The regenerative processes can be compromised in suboptimal situations such as avascular necrosis, atrophic nonunion, and catastrophic injury with impaired vasculature.[1]

Currently, various strategies are used to augment an impaired or insufficient bone-regeneration process, including distraction osteogenesis, bone transport, administration of growth factors, osteoconductive scaffolds, osteoprogenitor cells, and a number of bone-grafting methods.[1] Most of the current strategies for bone regeneration exhibit relatively satisfactory results, but there are some notable limitations to their effectiveness and availability. Overcoming the limitations of the current methods requires elucidation of bone healing mechanisms to produce more effective and accessible treatment modalities. These reparative processes comprise a variety of molecular and cellular events that recapitulate several aspects of skeletal development including vascularization and recruitment of mesenchymal stem cells,[2-4] but the precise mechanisms responsible for these processes have not been elucidated fully.

MicroRNAs (miRNAs) are endogenous, small, noncoding RNAs that mediate mRNA cleavage, translational repression, and mRNA destabilization. Currently more than 2200 human miRNAs are registered (miRBase Release 19).[5] miRNAs play diverse roles in fundamental biological processes such as cell proliferation, differentiation, growth, apoptosis, stress response, and tumorigenesis. miRNAs have been implicated in various aspects of embryonic development, such as neuronal, muscle, and cardiovascular organogenesis.[6, 7] Increasing evidence suggests that miRNAs regulate chondrocyte, osteoblast, and osteoclast differentiation,[8-14] implying important roles in fracture repair. However, to date, no reports have been available regarding miRNAs that regulate fracture healing or treatment strategies using miRNAs to enhance fracture healing.

miRNAs are present in human plasma called as circulating miRNAs in a remarkably stable form, protected from endogenous RNase activity.[15] Altered expression of circulating miRNAs have been reported in patients with rheumatoid arthritis[16] and patients with cancer,[15, 17, 18] suggesting their usefulness as biomarkers and tools for analyzing the pathogenesis of disease or injury. Although circulating miRNAs are expected to be useful biomarkers of fractures or postoperative conditions, no biomarker has been evaluated as an indicator of a patient's condition after fracture or surgery.

In this study, we showed that plasma miR-92a concentrations in human significantly decreased after 24 hours following fracture, compared to healthy controls (HCs). Neovascularization and the bone healing processes were promoted by systemic and local inhibition of miR-92a using a locked nucleic acid (LNA)-stabilized oligonucleotide in a mouse femoral fracture model. These results suggest that plasma miR-92a plays a crucial role in bone fracture healing and that the inhibition of miR-92a has a therapeutic potential in enhancing bone repair.

Subjects and Methods

Preparation of blood samples

Ethical approval for this study was granted by the ethics committee of Kyoto University Graduate School and Faculty of Medicine. Informed consent was obtained from all participants. Plasma samples were obtained from 26 patients with a bone fracture when they first visited the hospital. Three patients with a trochanteric fracture and patients with a lumbar compression fracture provided plasma within 24 hours of injury and days 7, 14, and 21 after the surgery or injury. Blood was also collected from volunteer HCs who were not being treated for arthralgia, heart failure, renal failure, or an autoimmune disease. The backgrounds of the patients with a fracture and HCs are shown in Supplementary Table 1. Blood samples were collected with EDTA-2K–containing tubes to separate the plasma. Samples were centrifuged at 1400g for 7 minutes, and the plasma samples were stored at –80°C until analyzed.

Mouse rib and femoral fracture model

All animal studies were conducted in accordance with principles and procedures approved by Kyoto University Committee of Animal Resources, based on International Guiding Principles for Biomedical Research Involving Animals. A mouse rib or femoral fracture model was created using 6-week-old wild-type (C57BL/6NCrSlc) mice. Details for creating rib fracture models have been described.[19] The surgical protocol for producing a femoral fracture included a longitudinal incision and blunt separation of the underlying muscles with care not to remove the periosteum as described.[20] A transverse osteotomy in the mid-diaphysis of the femur was created with a low-speed rotary diamond disk with saline irrigation. The fractured bones were then repositioned and stabilized by inserting a 23-gauge intramedullary needle.

Total RNA isolation from cellular, tissue samples, and conditioned medium

Total RNA was extracted from cell samples using a High Pure miRNA Isolation Kit (Roche Applied Science, Mannheim, Germany) according to the manufacturer's protocol. For tissue samples, the samples were snap-frozen in liquid nitrogen, homogenized with TriPure Isolation Reagent (Roche Applied Science), incubated for 5 minutes at room temperature, mixed with 0.2 volume of chloroform, shaken vigorously for 15 seconds, incubated for 3 minutes, and centrifuged at 12,000g for 15 minutes at 4°C. Then 300 µL of the aqueous phase was applied to the High Pure miRNA Isolation Kit according to the manufacturer's protocol. Total RNA included in the conditioned medium or plasma was isolated as described.[16]

Quantitative real-time polymerase chain reaction of mature miRNAs

Reverse transcription was performed using NCode VILO miRNA cDNA Synthesis Kit (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's protocol. Using EXPRESS SYBR GreenER qPCR SuperMix (Life Technologies), quantitative real-time polymerase chain reaction (qRT-PCR) was performed on an Applied Biosystems 7500 Thermocycler (Life Technologies) according to the manufacturer's protocol with standard plasmids generated as described[16] or with synthetic first-strand cDNAs with anticipated sequences. A total of 134 miRNAs were measured to screen for plasma biomarkers for bone fractures; the primer sequences for these miRNAs are listed in Supplementary Table 2. The data were analyzed using SDS Relative Quantification Software version 2.06 (Life Technologies). The miRNA expression data for the cellular and tissue samples from mice were normalized against small nuclear RNA MBII-202 (sno202). The absolute concentration of miRNAs in each liquid sample was calculated as described.[16]

Locked nucleic acid

Endotoxin-free antisense oligonucleotides were synthesized as hybrid molecules between deoxyribonucleotides and 2′-O-4′-O-methylene bridge (locked nucleic acid [LNA]) modification of all G and C residues with a complete phosphorothioate backbone by GeneDesign (Ibaraki, Japan). The sequences of the miR-92a inhibitor (antimir-92a) and its scrambled control are 5′-CAGGCCGGGACAAGTGCAATA-3′ and 5′-GCATCAAGACGGTCAGAGCGA-3′, as described.[21] LNA was administered intravenously into the tail vein (75 μg/75 μL) on days 0, 4, 7, 11, and 14 after the surgery, or locally around the fracture site (25 μg/100 μL) on days 4, 7, 11, and 14 after the fracture.

Radiographic analysis of fracture callus formation

Bone radiographs were taken with a soft X-ray instrument (CMB-2; SOFTEX, Ebina, Japan). Radiographic images were scored as follows: 1, no apparent hard callus; 2, slight intramembranous ossification; 3, hard callus without bridging of the fracture gap, fracture line is apparent; 4, hard callus with bridging of the fracture gap, fracture gap is noticeable; 5, unclear boundary between the newly formed hard callus and existing cortical bone; and 6, remodeling. Radiographic images were evaluated independently by two experienced orthopedic surgeons (KM, MI), and when the interpretations differed, the score was decided after discussion with KM, MI, and HI. Fractured femurs were scanned using an TDM1000 micro–computed tomography (µCT) system (Yamato Scientific, Tokyo, Japan) at 1024 views, 16 frames per view, 60 kV, and 60 μA. The callus volume of interest for a fractured bone was determined with TRI/3D-BON64 software (Ratoc System Engineering, Tokyo, Japan). Callus volume was calculated using TRI/3D-BON64 software. We scanned a set of hydroxyapatite (HA) phantoms and defined our mineralization threshold as 300 mg HA/cm3.

μCT analysis of neovascularization of the fracture sites

Vascularity of the fracture callus was evaluated using a μCT-based method.[22-24] Briefly, the entire vascular system was flushed by injecting heparinized (100 units/mL) normal saline, and then a radioopaque, lead chromate-based contrast agent (Flow Tech, Carver, MA, USA) was perfused by intracardiac injection. After perfusion, animals were kept at 4°C for 24 hours to allow the compound to polymerize. The femora were stored at 4°C for 48 hours in 4% paraformaldehyde, soaked for 21 days in 10% ethylenediaminetetraacetic acid (EDTA) solution for decalcification, washed thoroughly using water, and placed in 4% paraformaldehyde until imaging. Specimens were scanned using SMX-100CT-SV-3 (Shimadzu, Kyoto, Japan) at 2400 views, 5 frames per view, 37 kV, and 121 μA. Three-dimensional images of the radioopaque contrast-filled vascular network were rendered and evaluated using VGStudio MAX (Nihon Visual Science Software, Tokyo, Japan).

Histological analysis

Decalcified specimens were processed to obtain paraffin-embedded sections with a thickness of 5 to 7 μm, and the sections were stained with hematoxylin and eosin (HE) and HE/Alcian blue. Immunohistochemical staining was performed as described.[25] Specimens were incubated with rabbit anti-CD31 antibodies (Abcam, Cambridge, MA, USA; 1:50 dilution) overnight at 4°C. The number of CD31-positive vessels was counted in randomly selected areas around the fracture callus. The number of blood vessels was counted in the HE-stained sections. Cross-sectional vessel area was measured with ImageJ software (National Institutes of Health, Bethesda, MD, USA) as described.[26]

Luciferase assay

We inserted a 3′-untranslated region (3′UTR) fragment containing putative binding sites for miR-92a into the NheI-SalI fragment of the pmirGLO vector (Promega, Fitchburg, WI, USA). The sequences of inserted oligonucleotides are shown in Supplementary Table 3.

We transfected C3H10T1/2 cells with the luciferase reporter vector and double-stranded miR-92a (miCENTURY OX miNatural; Cosmo Bio, Tokyo, Japan) or the nonspecific negative control siRNA with using X-tremeGENE siRNA Transfection Reagent (Roche Applied Science). Luciferase activity was measured 24 hours after transfection using the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer's instructions.

Primary osteoblast isolation, cell line, and transfection

Primary osteoblasts were harvested from calvaria of newborn mice by sequential collagenase digestion (Roche Applied Science) and were seeded at a density of 2.5 × 104 cells/cm2 maintained in α modified essential medium (α-MEM) containing 10% fetal bovine serum (FBS) as described.[27]

Antimir-92a or control LNA (final concentration of 150 nM) was transfected using Lipofectamine 2000 (Life Technologies) when the primary osteoblasts reached 70% confluency, and cells were maintained for 14 days with the differentiation medium supplemented with 10 mM β-glycerophosphate (Wako, Osaka, Japan), 80 mg/mL ascorbic acid phosphate (Wako), and with 10−8 M dexamethasone (Nacalai Tesque, Kyoto, Japan) as described.[28] The culture medium was replaced every second day. Alkaline phosphatase (ALP) staining of primary osteoblasts was performed as described,[28] and ALP activity was also evaluated with the tartrate-resistant acid phosphatase (TRACP) and ALP Assay Kit (Takara Bio, Otsu, Japan), according to the manufacturer's instructions.

MC3T3-E1 or ATDC5 cells were seeded at a density of 1 × 104 cells/cm2 or 2 × 104 cells/cm2 18 hours before transfection and transfected using Lipofectamine 2000 (Life Technologies), according to the manufacturer's protocol. The MC3T3-E1 cells were maintained in α-MEM containing osteogenic supplements including, 10% FBS, 10 mM β-glycerophosphate, and 50 µg/mL of ascorbic acid,[29] and ADTC5 cells were grown and maintained in DMEM and F12 at a 1:1 ratio with 5% FBS supplemented with insulin (10 μg/mL) (Sigma, St. Louis, MO, USA), transferrin (5.5 μg/mL) (Sigma), and sodium selenite (5 ng/mL) (Sigma) to induce chondrocyte differentiation, as described.[30]

qRT-PCR of mRNA

Reverse transcription was performed using a Transcriptor High Fidelity cDNA Synthesis Kit (Roche Applied Science) according to the manufacturer's protocol. Using FastStart Universal SYBR Green Master (Roche Applied Science), qRT-PCR was carried out on an Applied Biosystems 7500 thermocycler according to the manufacturer's protocol. The primer sequences are listed in Supplementary Table 4. All gene expression data were normalized against glyceraldehyde-3-phosphate dehydrogenase.

Statistical analysis

The data are presented as the mean ± SEM, unless otherwise noted. The radiographic scores were analyzed using the Mann-Whitney U test. Differences between three groups were analyzed with the Bonferroni method. Unless otherwise noted, Student's t test was used for statistical analysis; p < 0.05 was considered significant.

Results

Plasma miR-92a levels were decreased 24 hours after fracture

miRNAs are present in human plasma in a stable form despite the endogenous RNase activity, called circulating miRNA and are known as noninvasive biomarkers for detection of cancer[15, 17, 18, 31] and of rheumatoid arthritis.[16] To determine whether plasma miRNA concentrations are changed in fracture-healing processes, we quantified the plasma levels of 134 randomly selected miRNAs in 4 patients with a trochanteric fracture and in 4 HCs and compared the average levels of each miRNA between the fracture and the HC group (Fig. 1A). Six miRNAs were identified with a more than a twofold differential expression (p < 0.05) between patients with a trochanteric fracture and HCs (Fig. 1B).

Figure 1.

Plasma concentrations of miRNAs in patients with a bone fracture. (A) Plasma concentrations of 134 miRNAs were compared between patients with a trochanteric fracture and HCs (n = 4 for each group). The averages of the differences in the Ct values between each miRNA and spike-in cel-miR-39 (–ΔCt) were calculated in each group, and the differences of the –ΔCt averages of each miRNA (–ΔΔCt) are demonstrated in the z axis. (B) Plasma miRNAs with more than a twofold differential expression with statistically significant difference (p < 0.05) between patients with a trochanteric fracture and HCs. (C) Plasma concentration of miR-92a in HCs (n = 26) and fracture patients within 24 hours of injury (n = 14) and 24 hours after injury (n = 12). The data are shown as mean ± SD. *p < 0.05. (D) Plasma concentrations of miR-92a in 3 patients with a trochanteric fracture within 24 hours of injury and on postoperative days 7, 14, and 21. The data were normalized against the concentration at the time of the injury. (E) Plasma concentrations of miR-92a in 3 patients with a lumbar compression fracture within 24 hours of injury, and 7, 14, and 21 days after injury. The data were normalized against the concentration at the time of the injury. HC = healthy control; Ct = cycle threshold.

We next measured the concentration of these six miRNAs in 26 patients with a bone fracture and 26 HCs (data not shown). Among these miRNAs, only plasma miR-92a showed changes in concentration in association with fracture healing. Plasma miR-92a levels remained normal within the first 24 hours following fracture, and then significantly decreased compared to HCs (Fig. 1C), whereas there was no significant difference in the concentration when the fractured patients were analyzed altogether (p = 0.09). We also isolated blood from 3 patients with a trochanteric fracture and patients with a lumbar compression fracture within 24 hours of injury and 7, 14, and 21 days after the surgery or injury. Plasma miR-92a concentration was lower on days 7 and 14, and appeared to recover to the normal level by 21 days after the surgery or injury (Fig. 1D, E). These results suggest that plasma miR-92a has the potential to be a biomarker of, and may play an important role in, fracture healing.

Antimir knockdown of miR-92a enhanced endochondral bone formation in fracture healing of the femur

To identify the role of miR-92a in fracture healing, we intravenously introduced either an LNA-stabilized oligonucleotide that is antisense to miR-92a (antimir-92a) or a scrambled control into mice with a femoral fracture. We first analyzed the expression of miR-92a by qRT-PCR in fractured callus of the rib and the expression of miR-92a was not changed on postoperative days 2, 7, and 14 (Supplementary Fig. 1A). Second, qRT-PCR analysis showed successful knockdown of miR-92a in the femurs 24 and 48 hours after the intravenous injection of antimir-92a in mice without a fracture (Supplementary Fig. 1B).

Next, we radiographically evaluated the time course of endochondral bone formation in femoral fracture healing (Fig. 2A). Hard callus with bridging of the fracture gap was observed, and the fracture gap was noticeable on postfracture day 14 in both the control and antimir-92a groups; however, the callus volume was moderately larger in the antimir-92a group than in the control group. On postfracture day 21 in the control group, the callus volume became larger than on postfracture day 14, and the boundary between the newly formed hard callus and the existing cortical bone was about to diminish. By contrast, in the antimir-92a group, the boundary had disappeared and remodeling processes were observed, indicating accelerated bone repair. Bone formation in X-ray images was assessed using a radiographic score as described in Subjects and Methods, and a significant difference between the control group and antimir-92a group was observed on postfracture day 21 (Fig. 2B).

Figure 2.

Inhibition of miR-92a enhanced endochondral bone formation in mice with a femoral fracture. An LNA-stabilized oligonucleotide that is antisense to miR-92a (antimir-92a) or a scrambled control into mice with a femoral fracture intravenously on days 0, 4, 7, 11, and 14 after the surgery. (A) Time-dependent radiological changes in the femur, 14 and 21 days after the fracture. Representative images are shown. (B) Bone formation in X-ray images was assessed using a radiographic score as described in Subjects and Methods. (C) Representative 3D μCT image of a fractured femur on postoperative day 14. Scale bar = 1 mm. (D) TV and BV of the callus, BV/TV, and BMD on postoperative day 14 and 21 was quantified using μCT. n = 6 mice per group. (E) Histology of the fracture callus stained by HE or HE/Alcian-blue staining 14 days after the femoral fracture. Arrows indicate Alcian-blue–positive cartilage. (F) The expression levels of Col1a1, Col II, and Col X of callus from mice on postoperative day 14 and 21 were quantified by qRT-PCR (n = 5, respectively). Scale bar = 500 μm. The data are shown as mean ± SEM. *p < 0.05. LNA = locked nucleic acid; 3D = three-dimensional; µCT = micro–computed tomography; TV = total volume; BV = bone volume; BMD = bone mineral density; HE = hematoxylin and eosin; qRT-PCR = quantitative real-time polymerase chain reaction.

The callus volume was quantified by μCT. Total volume and bone volume of the callus was 84% and 45% larger in the antimir-92a group than in the control group on postfracture day 14 (Fig. 2C, D). Because the remodeling processes had already occurred in the antimir-92a group, total volume or bone volume of the callus volume did not differ significantly between groups on postfracture day 21. Bone mineral density (BMD) was lower in the antimir-92a group on postfracture day 14. However, BMD significantly increased in the antimir-92a group compared with that in the control group.

Histological analyses on postfracture day 14 showed there was a larger area of cartilage at the fracture junction in the control group, indicating accelerated remodeling in the antimir-92a group (Fig. 2E).

Total RNA was extracted from callus on postfracture days 14 and 21. The expression of Col1a was significantly increased in the antimir-92a group compared to the control group on postfracture day 14, and the difference in Col1a expression became smaller on postfracture day 21 (Fig. 2F). The expression of Col II and Col X was lower in the antimir-92a group than in the control group on postfracture day 14 and the downregulated expressions of Col II and Col X by antimir-92a were diminished on postfracture day 21. These data also support the accelerated bone healing in the antimir-92a group.

Local administration of antimir-92a enhanced endochondral bone formation in fracture healing of the femur

In clinical situations, the dosage of LNA and the possible side effects of an anti-miRNA (antimir) are of concern. Intraperitoneal injection of LNA is reportedly effective for the knockdown of miRNA.[32] Thus, we administered antimir-92a locally to the fracture site to examine whether local administration can induce an equivalent effect of antimir to that of systemic induction.

X-ray showed hard callus with a fracture gap in both the control and antimir-92a groups on postfracture day 14 (Fig. 3A). The callus volume was larger and the fracture gap was diminishing in the antimir-92a group compared with the control group. On postfracture day 21, the boundary between the newly formed hard callus and the existing cortical bone had disappeared and remodeling processes were observed in the antimir-92a group but not in the control group; these results were similar to those found in the samples after systemic administration. A higher score on bone formation in X-ray was observed on postfracture days 14 and 21 between the control group and the antimir-92a group, similar to that observed for systemic administration (Fig. 3B).

Figure 3.

Local administration of antimir-92a enhanced endochondral bone formation in fractured femora in mice. antimir-92a or a scrambled control was administered locally around the fracture site on days 4, 7, 11, and 14 after the fracture. (A) Time-dependent radiological changes in femurs 14 and 21 days after fracture. (B) Bone formation in the X-ray images was assessed by the radiographic score. (C) Representative 3D μCT images of a fractured femur on postoperative day 14. Scale bar = 1 mm. (D) TV and BV of the callus, BV/TV, and BMD on postoperative day 14 and 21 were quantified using μCT. n = 5–10 mice per group. (E) Histology of the fracture callus stained with HE or HE/Alcian blue staining 14 days after the femur fracture. Scale bar = 500 μm. The data are shown as mean ± SEM. *p < 0.05; **p < 0.01. 3D = three-dimensional; µCT = micro–computed tomography; TV = total volume; BV = bone volume; BMD = bone mineral density; HE = hematoxylin and eosin.

Total volume and bone volume of the callus analyzed by μCT was 122% and 107%, respectively, larger in the antimir-92a group than in the control group on postfracture day 14 (Fig. 3C, D). No significant differences in total volume and bone volume of the callus were observed between the groups on postfracture day 21. There was no significant difference in BMD on postoperative day 14. However, BMD on postoperative day 21 in the antimir-92a group was significantly higher than in the control group.

Histological analysis on postfracture day 14 showed a greater proportion of mineralized tissues and a smaller proportion of the cartilage area in the antimir-92a group (Fig. 3E) compared with the control group. These findings indicate an accelerated healing in the antimir-92a group.

Inhibition of miR-92a did not affect osteoblast or chondrocyte differentiation

To examine the roles of miR-92a in osteoblast and chondrocyte differentiation, we tried to find new target genes of miR-92a, which should be highly expressed in osteoblasts or chondrocytes, by a computational screening for genes with complementary sites for miR-92a in their 3′UTR using open access software including TargetScan and miRanda. More than 4000 putative target genes were estimated, and miRanda identified alpha-2 type I collagen (Col1a2) and angiopoietin1 (ANGPT1) as potential targets of miR-92a. We conducted a luciferase reporter assay by cloning the 3′UTR of a putative target gene downstream of the firefly luciferase reporter gene in the pmirGLO vector and cotransfected these vectors with double-stranded miR-92a or mimics into C3H10T1/2. The luciferase activities were not downregulated (Supplementary Fig. 2), indicating that these genes were not direct targets of miR-92a.

Next, to analyze the effects of antimir-92a on osteoblast and chondrocyte differentiation, we isolated primary osteoblasts from neonatal calvariae. We transfected primary osteoblasts with antimir-92a and incubated the cells in differentiation medium for 14 days. The expression of Runx2, Osterix, Osteocalcin, Col1a1, and ALP did not differ between the control and the antimir-92a groups (Fig. 4A). ALP staining for the culture did not show a difference and there was no significant difference in ALP activity of the culture between the control and the antimir-92a groups (Fig. 4B), indicating no acceleration or deceleration of osteoblast differentiation by antimir-92a.

Figure 4.

Effects of miR-92a inhibition on osteoblast and chondrocyte differentiation. Primary osteoblasts were isolated from neonatal calvariae and transfected with antimir-92a or control LNA and incubated in differentiation medium for 14 days. (A) Total RNA was extracted, and the expression of each gene was analyzed by qRT-PCR (n = 5, respectively). (B) ALP activity was measured by ALP staining (representatives are shown) and ALP assay kit (n = 8, respectively). (C) ATDC5 cells were transfected with antimir-92a or control LNA and incubated in differentiation medium for 48 hours. Total RNA was extracted, and the expression of each gene was analyzed by qRT-PCR. (n = 6, respectively). The data are shown as mean ± SEM. *p < 0.05. LNA = locked nucleic acid; qRT-PCR = quantitative real-time polymerase chain reaction; ALP = alkaline phosphatase; Ct = cycle threshold; OD = optical density.

We also transfected MC3T3-E1 with antimir-92a and incubated the cells in differentiation medium for 2 days. The expression of Osteocalcin did not differ between the control and the antimir-92a groups (data not shown).

Similarly, no significant differences in the expression of type II collagen (Col II), type X collagen (Col X), and Sox9 were observed in ATDC5 cells transfected with antimir-92a (Fig. 4C), indicating again no effect of miR-92a inhibition on chondrocyte differentiation.

These results suggest that antimir-92a did not affect either of osteoblast and chondrocyte differentiation.

Suppression of miR-92a enhanced angiogenesis during fracture healing

Because miR-92a has been shown to inhibit vascular development by targeting ITGA5,[33] we used μCT imaging of contrast-perfused, decalcified specimens to evaluate vascular growth within the fracture callus of mice (Fig. 5A). The vessel volume was 90% larger 14 days after surgery in the fracture callus of mice intravenously administered antimir-92a compared with those treated with scrambled control (Fig. 5B).

Figure 5.

Suppression of miR-92a enhanced angiogenesis during fracture healing. (A) Vascularity in the fractured femora of mice was visualized by μCT using a radioopaque silicon polymer medium on postfracture day 14. LNA was administered intravenously on days 0, 4, 7, 11, and 14 after the fracture. Representative images are shown. (B) On day 14 after surgery, vessel volume was quantified on 3D μCT images. n = 5–6 mice per group. The data are shown as mean ± SEM. *p < 0.05. (C) Tissues surrounding the fracture callus on day 14 after surgery were stained with HE (upper panels) or stained immunohistochemically with anti-CD31 (lower panels). Scale bar = 200 μm. (D) The number of vessels in HE-stained samples (left panel) and CD31-positive vessels (middle panel) were counted, and the ratio of vessel area was measured in HE-stained samples (right panel) (n = 5–7 per group). (E) The fractured callus on postoperative day 14 were stained with anti-CD31 (lower panels). Representatives are shown and boxed areas are enlarged on right bottom. Scale bar for original images = 200 μm. (F) The expression levels of VEGF-A, ANGPT1, and ITGA5 of callus from mice on postoperative day 14 were quantified by qRT-PCR (n = 5, respectively). The data are shown as mean ± SEM. *p < 0.05; **p < 0.01. LNA = locked nucleic acid; 3D = three-dimensional; µCT = micro–computed tomography; HE = hematoxylin and eosin; Ct = cycle threshold.

We also used immunohistochemistry to analyze the tissues surrounding the femoral fracture callus 14 days after surgery. The number of total and CD31-positive blood vessels increased 174% and 106%, respectively, in the antimir-92a group compared with the control group (Fig. 5C, D). The ratio of the vessel areas was also 300% higher in the antimir-92a group than in the control group.

Immunohistochemical staining analysis showed increased CD31-positive capillary invaded into the cartilage 14 days after surgery by the intravenous administration of antimir-92a (Fig. 5E). In accordance with the results above, the expression levels of VEGF-A and ANGPT1 in fractured callus 14 days after surgery were significantly higher in the antimiR-92a group (Fig. 5F), suggesting that antimir-92a stimulates the angiogenesis in fractured callus and surrounding tissues. There was no significant difference in mRNA expression level of ITGA5, a known target of miR-92a, in fractured callus between the antimir-92a and the control group. Capillary invasion to the fractured callus by antimir-92a coincide with the results from microCT analysis, where no significant increase in BMD on postfracture day 14 was observed (Fig. 2D) and quick invasion might remain the Alcian blue-positive chondrocytes observed in the center of the callus (Fig. 2E).

Taken together, miR-92a plays a crucial role in inhibition of neovascularization in bone repair processes and that inhibition of miR-92a enhances fracture healing. Systemic and local administration of antimir are both effective for the enhancement of fracture repair.

Discussion

miRNAs have been shown to exist in a stable form in plasma despite endogenous RNase activity and are thought to have potential noninvasive biomarkers for detection of cancer[15, 17, 18, 31] and of tissue injury.[34, 35] However, no biomarker is currently available for evaluation of fracture healing in humans. Plasma miR-92a concentration was lower in fracture patients 24 hours after injury than in HCs and remained low 7 and 14 days after injury, suggesting that it can potentially indicate the status of fracture healing. Furthermore, the results suggest that miR-92a plays a crucial role in fracture healing. Although we cannot exclude the possibility that downregulation of miR-92a reflects the response to inflammation or tissue injury, previous studies and the present data collectively indicate that changes of plasma miR-92a concentration show the status of neovascularization after a fracture. Indeed, there are both clinical and preclinical studies that show that markers of angiogenesis define both the progression of fracture healing and may be prognostic of delayed or failed bone healing.[36, 37] Further analyses using a larger number of samples, including age-matched controls with various injuries, are needed.

miR-92a is encoded in the miR-17-92 cluster, which is highly expressed in human endothelial cells. The miR-17-92 cluster is essential for vertebrate development, because universal disruption of Mirn17 in mice results in smaller embryos and immediate postnatal death. This is due to severely hypoplastic lungs and ventricular septal defects in the hearts of mice lacking miR-17-92.[38] These miRNAs are normally highly expressed in embryonic lung and decrease as mice reach maturity. Recently, haploinsufficiency of the miR-17-92 cluster was shown to impair skeletal bone formation as well as osteoblast proliferation and differentiation.[39] However, suppression of miR-92a was not sufficient in their mice and individual functions of miRNAs in the miR-17-92a cluster in osteoblasts have not been fully investigated.

miR-92a targets ITGA5 and MKK4[33, 40] and is considered to inhibit angiogenesis. Angiogenesis is obviously important for bone fracture healing, because new blood vessels bring oxygen and nutrients to the highly metabolically active regenerating callus and serve as a route for inflammatory cells and mesenchymal progenitor cells to reach the injury site.[40] Defective angiogenesis at a fracture site can cause poor outcomes, including nonunion of the fracture, a clinical diagnosis made when a patient has clinical symptoms of pain at the fracture site, a pathological instability, and radiographic findings of unfused bone. We show here for the first time that miR-92a plays an essential role in bone healing in a mouse fracture model. Our results indicate that miR-92a inhibits neovascularization during the fracture healing processes. The possible reasons and mechanisms for the downregulation of plasma miR-92a remain to be investigated, but our study showed clearly that the inhibition of miR-92a can enhance neovascularization in a fracture model.

A possible concern would be that antagonizing a factor expressing in “normal fracture” may inhibit a normal repair process. However, this study demonstrated that an inhibitory factor for angiogenesis, miR-92a, is downregulated in normal fracture healing, and that even more downregulation of miR-92a accelerates fracture healing. Our hypothesis is that antiangiogenic factor(s) suppress unnecessary angiogenesis in normal conditions, but, once a tissue is injured, the factor(s) are suppressed, and neovascularization occurs. If one can inhibit those factors more effectively, the repair process would be accelerated. Side effects of the proangiogenic strategy are of concern, but any side effects were not found in this study. The proangiogenic therapeutic strategy has great possibilities for the future in accelerating fracture healing.

Various growth factors in the angiogenic cascade, including vascular endothelial growth factor, fibroblast growth factor, and platelet-derived growth factor, are potential targets for upregulation and direct administration to accelerate bone repair.[41] However, there are no reports describing an miRNA-based approach in treating bone fracture. miRNAs are considered to be an attractive target for therapeutic manipulation, compared with the growth factors, due to the fact that one miRNA can regulate dozens of genes and can thus act as an amplifier. miRNAs are relatively stable small molecules whose expression levels can be manipulated by a growing number of technologies. To inhibit specific miRNAs, as shown in this study, oligonucleotides complementary to either the mature miRNA or its precursors can be designed such that the miRNA will be functionally arrested and will be unable to bind to the target mRNA subset.[42]

Plasma miRNAs are secreted in various form through cell-derived microvesicles including microparticles and exosomes, complexes with proteins including Ago2 protein and high-density lipoproteins, and apoptotic bodies.[43-46] The possible functions and protective mechanisms of circulating miRNAs remain unclear, but these forms of miRNAs can transfer a gene-silencing signal between living cells in vitro and in vivo, and they are considered to be used as cell-to-cell communication. Therefore, we systemically introduced LNA-stabilized antimir in mice with a femoral fracture, and the systemic induction strategy was proved to be effective for bone fracture healing.

The LNA-antimir technology adopted in this study has been used in miRNA inhibition experiments in vivo, and a phase II clinical trial is ongoing for SPC3649, an LNA-based antisense molecule against miR-122 for the treatment of hepatitis C.[32, 47] In our study, successful downregulation was achieved even in the mouse femur (Supplementary Fig. 1B), suggesting that this LNA-based antisense therapy has unique potential in the treatment of bone injuries or disorders. Further modification to provide more effective transfer would enable this technology to be useful in therapeutic strategies in human diseases.

There are concerns that therapeutically induced angiogenesis may increase the risk of developing hemangioma, recurrence of malignant tumors, or deterioration of diabetic retinopathy.[48] Although we could not identify any side effects of antimir treatment given systemically or locally in this study, ensuring tissue-specific delivery and cellular uptake of sufficient amounts of synthetic oligonucleotide is a major challenge in the use of miRNA therapeutics in vivo and has been studied extensively.[49, 50] In our study, local administration of an antimir was associated with bone fracture healing similar to that induced by systemic administration, suggesting that a lower dosage of an antimir can be administered locally yet produce similarly successful bone healing with potentially fewer side effects compared with systemic administration.

The present study has a few limitations. First, the dosage and interval of administration of antimir-92a was not optimized. A previous study showed that silencing of miR-122 and subsequent reduction in plasma cholesterol level could be achieved with 3 mg/kg of LNA,[32] and we adopted this dosage. It remains unclear when inhibition of miR-92a is necessary during the growth of new blood vessels, or when angiogenesis should be induced to enhance fracture healing. Additionally, local administration of an antimir may impair the healing of surrounding tissue unless it is administered at an appropriate time and in an appropriate volume. Optimization of the dosage, volume, timing, and delivery methods to an injured tissue is necessary.

Another limitation of this study is that the number of human samples was not sufficient for judging miR-92a as a useful biomarker. Patients with various bone fractures were included in this study, and HCs were younger than were patients with fractures. Samples from patients with a bone bruise may be appropriate as an optimal control group. Moreover, a longitudinal study with a sufficient sample size is mandatory for assessing the time-dependent changes in various fractures and injuries. Further analyses using a larger number of samples including age-matched controls with various injuries are required. However, this study shows, for the first time, that the plasma miRNA concentrations are downregulated during fracture healing.

In conclusion, inhibition of miR-92a by antimir enhanced bone healing in a mouse fracture model by promoting angiogenesis.

Disclosures

All authors state that they have no conflicts of interest.

Acknowledgments

This work was supported by a grant-in-aid from the Ministry of Education of Japan (No. 21591942, 23659719) and a grant from the Japan Orthopaedics and Traumatology Foundation, Inc. (Grant No. 262). The authors are grateful to all patients and volunteers who kindly provided blood for use in this study. We thank the staff at Yoshikawa Hospital for collecting the blood samples. We also thank Dr. Tomoki Aoyama, Dr. Toshiyuki Kitaori, and Ryoko Nakanishi (Kyoto University Graduate School of Medicine) for their valuable technical assistance.

Authors' roles: Study design. KM, HI, and KY. Study conduct: KM, HI, HY, KY, AF, JY, FM, MI, and HS. Data collection: KM, HI, KY, HS, and MI. Data interpretation: KM, HI, KY, MI, HY, and SM. Drafting the manuscript: KM and HI. Revising the manuscript content: KM, HI, and HY. Approving the final version of the manuscript: KM, HI, HY, MF, AF, JY, MI, HS, and SM. HI takes responsibility for the integrity of the data analysis.

Ancillary