Long non-coding RNA Homeobox D gene cluster antisense growth-associated long noncoding RNA/microRNA-182-5p/Homeobox protein A10 alleviates postmenopausal osteoporosis via accelerating osteoblast differentiation of bone marrow mesenchymal stem cells

PMOP patients (n = 55) and healthy PM women (n = 55) in The Affiliated Lianyungang Oriental Hospital of Xuzhou Medical University were enrolled, and peripheral blood samples were collected. Lumbar spine bone mineral density (L1-L4) was assessed according to standard operating instructions using a dual-energy X-ray bone densimeter (DXA) (Hologic Discovery Wi, Hologic, USA). T-score is the standard deviation of bone mineral density relative to the mean. Osteoporosis is defined as bone mineral density of the lumbar spine (L1-L4) with a T-score ≤ -2.5. The inclusion criteria are as follows: (1) age ≥ 50 years; (2) Menopause ≥ 1 year; (3) Sign informed consent before participating in the study. Exclusion criteria are as follows: (1) any comorbidities that may significantly affect bone metabolisms, such as thyroid disease, diabetes, cancer, kidney disease, or ankylosing spondylitis; (2) previous anti-osteoporosis medication or hormone therapy (vitamin D and/or calcium supplements are allowed), such as estrogen or glucocorticoids; (3) A history of smoking or alcohol dependence within the past year. The research was approved by the Ethics Committee of The Affiliated Lianyungang Oriental Hospital of Xuzhou Medical University.

Bone marrow was obtained from mice (Guangdong Medical Laboratory Animal Center, Guangdong, China). BMSCs were isolated by the whole bone marrow adhesion method and cultured in a-MEM medium (HyClone) containing 10% fetal bovine serum (FBS) (Gibco) and 1% penicillin (HyClone). After passages, BMSCs were identified by flow cytometry to detect surface markers CD34, CD45, CD73, and CD90.

BMSCs were cultured in 6-well plates and added an osteogenic induction medium containing Dulbecco’s Modified Eagle Medium, 10% FBS, 0.1 mg/mL dexamethasone, 50 mg/mL ascorbic acid, and 10 mmol/L glycerophosphate. The medium was replaced every 3 days.

MiR-182-5p, anti-miR-182-5p, sh-HAGLR/Hoxa10, and their respective negative controls (NCs) were provided by GenePharma (Shanghai, China). Transfection of BMSCs was done with Lipofectamine 3000 reagent (L3000015, Invitrogen) in the light of the manufacturer’s instructions. AD-HAGLR-EGFP and NC adenovirus (AD-EGFP) were produced by Han Biotechnology (Shanghai, China).

When the confluence of BMSCs reached about 100%, osteogenic differentiation was induced by adding osteogenic induction medium. After 14 days,cells were fixed with 95% cold ethanol for 25 min and air-dried. Alizarin Red S (40 mmol/L; A5533, Sigma Aldrich, USA) was dissolved in dH₂O, and cells were stained with the prepared solution at 25℃ for 30 min. Then, 10% (w/v) cetylpyridine chloride (HC0502, HEROCHEM, Shanghai, China) was prepared and used for decolorization. The absorbance at 560 nm was read [29].

Osteoblasts were detached with trypsin and seeded in a 24-well plate. Then, cells were treated with propanol (15 min), incubation solution (6 h), cobalt nitrate (15 min), and ammonium sulfide (5 min) (all 200 μL). Optical density at 490 nm was read on a microplate reader (Varioskan LUX; Thermo Fisher Scientific) [30].

To assess cell apoptosis, 2 mL cell suspension at 1 × 10⁵ cells/mL was seeded into a 6-well plate. After 72 h, cells were centrifuged at 1000 rpm for 3 min and tested by an Apoptosis Detection Kit (559763, BD Biosciences, USA). Cell staining was done using Fluorescein isothiocyanate-Annexin V and propidium iodide, and analysis of cell apoptosis was performed using flow cytometry (BD Biosciences). Cell QuestPro software (BD Biosciences) was used for apoptosis analysis [31].

Fifty Balb/c mice were randomly divided into the ovariectomized model (OVX) and the sham operation control (sham) groups. The OVX mice were anesthetized with 5% ketamine and sterilized normally. The ovaries on both sides of the mice were removed in a biologically clean environment. After complete hemostasis, the abdominal wound was sutured. Only a small amount of fat was removed in sham surgery. Two groups of mice were fed separately with free food and water. All experiments on mice met the standard guidelines for the use of animals in scientific research.

Four weeks after ovariectomy, PMOP mice were injected with AD-HAGLR-EGFP, NC adenovirus (AD-EGFP), and sh-Hoxa10. On days 1 to 3 of weeks 1 and 4, mice (n = 8) were administered at a dose of 7 mg/kg via the caudal vein. PMOP mice injected with normal saline were regarded as controls (PMOP; n = 8). Six weeks after the first injection, the bilateral femurs were collected from mice.

BMD levels in the left femur of mice were measured using a Lunar DPX-IQ dual-energy X-ray absorptiometry with a PIXImus II absorptiometry (Lunar Corporation, Madison, WI). Elastic modulus, maximum load, and maximum bending stress were tested according to the requirements of the three-point bending test using a computer-controlled mechanical testing machine (SANS-10404043, Shenzhen, China). The sample distance is 23 mm and the plunger speed is 2.0 mm/min [32].

Tibias of mice were fixed with paraformaldehyde (P0099, Beyotime Biotechnology, China) for 1 week and rinsed 3 times to remove excess paraformaldehyde. Tibias were embedded in paraffin and cut into 5 μm sections, followed by treatments with xylene and graded ethanol. HE staining was performed using Hematoxylin and Eosin (G1120, Solarbio, China) according to the manufacturer's instructions. The morphology of the tibias was observed under a microscope. HE staining was performed on the left hind leg, and the mice in each group were examined on the same side [33].

To test cell apoptosis, the tissues were treated with 2-methoxyethyl acetate and then incubated in 10 mM citrate buffer (pH 6.3) at 90℃ for 15 min. Subsequently, the sections were incubated with 0.5% pepsin at 37℃ for 30 min and detected by an in situ cell death detection kit (12,156,792,910, Roche, USA). Sections were stained with diaminobenzidine (Sigma) after incubation with POD at 37℃ for 1 h and then examined under a light microscope. TUNEL-positive cells were counted in 5 random fields [34].

Total RNA was extracted from tissues or cells using TRIzol (15,596,026, Invitrogen) and reverse-transcribed using the PrimeScript RT Master Mix kit (RR036B, Takara, Dalian, China) according to the manufacturer's recommended instructions. Then, RT-qPCR was performed in the ABI Prism 7900HT sequence detection system (Applied Biosystems) using SYBR Green Real Time PCR Master Mix (QPK-201, Toyobo, Osaka, Japan). Gene quantification was performed using the 2^−ΔΔCT method. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and U6 were used as loading controls, respectively. The primer sequences for RT-qPCR were presented in Table 1 [35]. The agarose gel images are presented in the supplementary Fig. 1.

Radioimmunoprecipitation assay lysis buffer (P0013B, Beyotime) was added to extract the total protein. Subsequently, proteins were separated by electrophoresis and electroblotted onto polyvinylidene fluoride membranes (Millipore). After incubation with 5% skim milk for 2 h, the membrane was reacted overnight with primary antibodies Hoxa10 (ab191470), Runt-related transcription factor 2 (RUNX2) (ab76956), osteopontin (OPN) (ab214050) (1: 1000), Osteocalcin (OCN) (ab93876, 1: 500) and GAPDH (ab8245, 1: 2000, Abcam). Membranes were incubated with the corresponding secondary antibody for 2 h and exposed to enhanced electrochemiluminescence (Thermo Fisher Science) to develop protein bands. Data analysis was performed using Image J software (NIH, Bethesda) [36].

BMSCs were transfected with pMS2bp-GFP and MS2, MS2-HAGLR, or MS2-mutant (MUT)-HAGLR and collected after 48 h. Biotin-conjugated RNA complexes were pulled down and checked for RIP according to the manufacturer's instructions using the Magna RIP kit (Millipore, USA). Samples were incubated with anti-GFP and anti-immunoglobulin G, and miR-182-5p expression was detected by RT-qPCR [37].

BMSCs were transfected with biotinylated miR-182-5p and HAGLR. After 72 h, BMSCs were collected, and cell lysates were bound to M-280 streptavidin magnetic beads (Sigma) to pull down the biotin-conjugated RNA complexes. Then, the RNA-bound beads were purified with TRIzol. HAGLR or miR-182-5p expression was tested by RT-qPCR.

Wild-type (WT) or MUT HAGLR binding to miR-182-5p was subcloned into the pGL3 vector. BMSCs were co-transfected with miR-182-5p (RiboBio, Guangzhou, China) and 10 μg of pur-WT-HAGLR or pur-MUT-HAGLR. WT or MUT Hoxa10 and miR-182-5p were subcloned into a pGL3-based vector (Promega). miR-182-5p (RiboBio) was co-transfected with 10 μg pluco-WT-Hoxa10 or pluco-MUT-Hoxa10. After transfection of 48 h, a test of luciferase activity was done via the dual luciferase detection system (Promega Corporation, Fitchburg, WI, USA) [38].

Data analysis was performed using SPSS 21.0 (SPSS, Inc, Chicago, IL, USA) statistical software. After the Kolmogorov–Smirnov test, the data were normally distributed, and the results were expressed as mean ± standard deviation (SD). Two-group comparisons were done with t test and comparisons among multiple groups were done with one-way analysis of variance (ANOVA) and Fisher’s least significant difference t-test. The enumeration data were expressed as rates or percentages, and the chi-square test was used for comparative analysis. P was a two-sided test, and P < 0.05 was accepted as indicative of distinct differences.

To explore the biological functions of HAGLR, miR-182-5p, and Hoxa10 in PMOP, their expression levels in the peripheral blood of PMOP patients and healthy PM women were tested by RT-qPCR. HAGLR and Hoxa10 were down-regulated and miR-182-5p was elevated in PMOP patients compared with healthy controls (Fig. 1A–C). Pearson linear regression analysis noted that HAGLR was negatively linked with miR-182-5p and positively associated with Hoxa10, while miR-182-5p was negatively associated with Hoxa10 (Fig. 1D–F).

BMSCs were isolated from mouse bone marrow. After 1 week of primary culture, BMSCs were polygonal with a limited distribution area. The second-generation cells were long spindle fibroblasts with large nuclei and abundant cytoplasm. The third generation is typical of bipolar spindle cells. When the confluence reached 90%, the cells were observed to be spirally shaped (Fig. 2A). Additionally, flow cytometry identified that BMSCs (the third generation) were positive for CD73 and CD90 and negative for CD34 and CD45 (Fig. 2B). These results confirmed that the cells isolated from mice were BMSCs.

Subsequently, BMSCs were induced, RT-qPCR tested HAGLR, miR-182-5p, and Hoxa10 at 0, 7, and 14 d. HAGLR and Hoxa10 levels were augmented, while miR-182-5p was suppressed time-dependently (Fig. 2C–E). Alizarin red S staining showed manifested intense staining of BMSCs and BMSC mineralization (Fig. 2F). ALP staining showed increased ALP expression (Fig. 2G). Additionally, mRNA and protein expressions of OD-associated genes RUNX2, OPN, and OCN were gradually elevated as well (Fig. 2H–I). Flow cytometry revealed that with the increase of OD degree of BMSCs, apoptosis decreased significantly on days 7 and 14 (Fig. 2J). These experiments suggest that HAGLR/miR-182-5p may be involved in BMSC OD.

On the one hand, to determine the role of HAGLR in OD, recombinant vectors were transfected to increase HAGLR. On the other hand, HAGLR gene-silencing BMSCs were induced. The transfection efficiency was verified by RT-qPCR (Fig. 3A). Silenced HAGLR constrained BMSC mineralization ability, while elevating HAGLR had the opposite effects (Fig. 3B). In the meantime, during BMSC OD, ALP content was decreased after silencing HAGLR, while it was augmented after elevating HAGLR (Fig. 3C). Additionally, silencing HAGLR decreased expressions of osteogenic markers (OPN, OCN, and RUNX2), whereas overexpressing HAGLR elevated expressions of these markers (Fig. 3D, E). Repression of HAGLR strengthened BMSC apoptosis, while elevated HAGLR suppressed cell apoptosis (Fig. 3F). These results suggest that HAGLR silencing inhibits OD of BMSCs.

Possible miRNA binding to HAGLR was predicted by StarBase. As shown in Fig. 4A, HAGLR may target miR-182-5p. Luciferase assay was used to determine the target interaction between HAGLR and miR-182-5p. As shown in Fig. 4B, miR-182-5p decreased the luciferase activity of WT-HAGLR but had no effect on the luciferase activity of MUT-HAGLR. RNA pull-down assay and RIP assay further verified the relationship between miR-182-5p and HAGLR. As manifested in Fig. 4C, miR-182-5p was decreased by biotin-labeled WT-HAGLR but not MUT-HAGLR. MiR-182-5p was abundant in the HAGLR group but not in the mut-HAGLR group (Fig. 4D). Besides, suppression of HAGLR augmented miR-182-5p expression, while elevation of HAGLR reduced miR-182-5p expression (Fig. 4E). In short, HAGLR performed as a sponge for miR-182-5p to suppress miR-182-5p expression.

BMSCs were transfected with anti-NC or anti-miR-182-5p to explore the role of miR-182-5p in BMSC OD. In addition, miR-182-5p mimic was transfected into HAGLR-silenced BMSCs to further clarify the association between miR-182-5p and HAGLR (Fig. 5A). The experiment manifested that suppression of miR-182-5p boosted BMSC mineralization ability, elevated ALP content, and increased mRNA and protein expressions of OPN, OCN, and RUNX2 (Fig. 5B–E). Suppression of miR-182-5p decreased BMSC apoptosis (Fig. 5F). Additionally, repression of miR-182-5p turned around the effects of silenced HAGLR on BMSC OD (Fig. 5B–F). To sum up, suppression of miR-182-5p ameliorated BMSC OD.

Bioinformatics analysis clarified that miR-182-5p and Hoxa10 had complementary binding sites (Fig. 6A). Co-transfection of miR-182-5p and wt-Hoxa10 repressed the luciferase activity (Fig. 6B). Elevated Hoxa10 mRNA (Fig. 6C) and protein (Fig. 6D) expression was detected in BMSCs knocking down miR-182-5p. These results indicate that miR-182-5p negatively regulates Hoxa10 expression.

To further verify that HAGLR alleviates PMOP, OVX mice were constructed and injected with the corresponding lentivirus (Fig. 7A). Compared with Sham mice, OVX mice showed a decrease in BMD, elastic modulus, maximum load, and maximum bending stress, confirming that the model was successfully constructed. These indices increased after overexpression of HAGLR. Furthermore, BMD and three biomechanical parameters decreased after sh-Hoxa10 injection (Fig. 7B–E), These results indicated that Hoxa10 is involved in HAGLR/miR-182-5p axis to participate in OVX mouse development. HE staining showed that the subchondral trabecular bone volume decreased in OVX mice. After overexpressing HAGLR, the bone cortex was thickened, the bone trabecular was increased, and the intertrabecular connectivity was better (Fig. 7F). TUNEL assay manifested that elevated HAGLR reduced TUNEL-positive chondrocytes in OVX mice, while sh-Hoxa10 turned around the effect of elevated HAGLR on chondrocyte apoptosis (Fig. 7G). In summary, overexpression of HAGLR alleviates PMOP, and down-regulation of Hoxa10 reverses the effect of overexpression of HAGLR.