The GSE68303 datasets was downloaded from the Gene Expression Omnibus (GEO) database. GSE68303 datasets which contain 12 ovariectomize (OVX) experimental groups and 11 sham groups were downloaded. Then R software (R-4.3.1) was used to analyze the differentially expressed genes (DEGs) through the “Limma” package with $|$logFC$|$ $\ge$0.5 and p 0.05 as criteria.

Considering the retrieved DEGs, gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed with the Enrichr database [19]. GO enrichment analysis encompassed terms related to biological process (BP), molecular function (MF), and cellular component (CC). KEGG pathway analysis is broadly employed for the identification of functional and metabolic pathways; here the threshold p value for statistical significance was 0.05.

Search tool for the retrieval of interacting genes/proteins (STRING) was utilized to screen genes for protein-protein interaction (PPI) network generation, with an interaction of 0.4 selected. Visualization was carried out with Cytoscape 3.9.1 (http://www.cytoscape.org/). Hub genes were screened in CytoHubba by calculating the density of maximum neighborhood component (DMNC), Clustering Coefficient and edge percolated component (EPC). p 0.05 indicated statistical significance.

Female 8-week-old C57/BL6 mice order from Nanjing Medical University (Nanjing, China) were sham-operated or bilaterally ovariectomized (OVX) as described in a previous report [20]. After 8 weeks, successful ovariectomy was achieved as demonstrated by X-ray analysis of coccygeal vertebrae that showed significantly reduced bone mineral density (data not shown). Next, the OVX group was randomly subdivided into 2 groups of 6 each, administered water and simvastatin (Sigma, St. Louis, MO, USA) at 5 mg/kg/day, respectively, for 4 weeks. The simvastatin dose referred to a previous study [21]. After treatment, the vertebra were removed muscle and soft tissusses ,then collected and used to futhur analysised.

Housing was performed with 5 animals/cage, and a normal diet containing 1.0% calcium, 0.67% phosphorus, and 2.2 IU vitamin D/g was provided (#1010013; Jiangsu Province Collaborative Medicine Bioengineering, China). Assays involving animals were followed the ARRIVE guidelines (Animal Research: Reporting of In Vivo Experiments) [22] and approved by the Committee on the Ethics of Animal Experiments of Nanjing Medical University (Number: IACUC-1802007).

Tibiae and femurs from wild-type animals were obtained aseptically, and DMEM supplemented with 10% FBS, 50 µg/mL vitamin C (Sigma, USA), 10 mM $\beta$-glycerophosphate (Sigma, USA), and 10${}^{-8}$ M dexamethasone (Sigma, USA) was utilized to flush out bone marrow cells. After dispersion by repeated pipetting, single-cell suspensions were obtained by passing the cells through a 22G needle. This was followed by a 10–14 day culture in 6-well plates at 1 million cells per well in proliferation medium (2 mL), which was refreshed at 3-day intervals. Then, cells underwent a PBS wash, fixation with periodate-lysine-paraformaldehyde (PLP), and cytochemical staining for Alkaline phosphatase (ALP) [23] to detect ALP-positive CFU-f (CFU-fap). Briefly, following pre-incubation overnight with 1% magnesium chloride in 100 mM-Tris-maleate buffer (pH 9.2), dewaxed sections were incubated for 2 h at room temperature in 100 mM-Tris-maleate buffer containing naphthol AS-MX phosphate (0.2 mg/mL; Sigma, USA) dissolved in ethylene glycol monomethyl diethyl ether (Sigma, USA) as a substrate and fast red TR (0.4 mg/mL; Sigma, USA) as a stain for the reaction product.

Extracted femurs and tibias underwent micro–computed tomography [24], and histological, histochemical and immunohistochemical analyses [23]. Briefly, dewaxed and rehydrated paraffin-embedded sections were incubated with methanol: hydrogen peroxide (1:10) to block endogenous peroxidase activity and then washed in Tris-buffered saline (pH 7.6). The slides were then incubated with the primary antibodies overnight at room temperature. After rinsing with Tris-buffered saline for 15 minutes, tissues were incubated with secondary antibody (biotinylated goat anti-rabbit IgG diluted 1:1000, Sigma-Aldrich, St. Louis, MO, USA). Sections were then washed and incubated with the Vectastain Elite ABC reagent (Vector Laboratories, Inc. Ontario, Canada) for 45 minutes. Staining was developed using 3,3-diaminobenzidine (2.5 mg/mL) followed by counterstaining with Mayer’s Hematoxylin. Immunohistochemical staining employed anti-IRF4 (1:800) and anti-Runx2 (1:1000) primary antibodies (Abcam, Cambridge, UK). The staining pictures were taken by microscope (Olympus, Tokyo, Japan) equipted with image acquisition system (Nikon, Tokyo, Japan).

Protein extraction from bone tissue samples followed a previous report [25]. Briefly, bone tissue samples were lysed in lysis buffer containing phenylmethylsulfonyl fluoride (PMSF) (Sigma, USA) supplemented with a protease inhibitor (Sigma, USA) or protease and phosphatase inhibitor cocktail (Roche, Indianapolis, IN, USA). The proteins were separated using gel electrophoresis and electrotransferred to polyvinylidene fluoride membranes. After blocking with 5% bovine serum albumin for 2 h and washing with Tris-buffered saline/Tween-20 (TBS-T) for 30 min, the membranes were incubated overnight at 4 °C with primary antibodies against anti-IRF4 (1:500, Abcam, USA) and anti-$\beta$-actin (1:1000, Bio world Technology, St Louis Park, MN, USA). Subsequently, the membranes were washed with TBS-T for 30 minutes at room temperature and incubated with horseradish peroxidase-conjugated secondary antibodies for 2 h. The membranes were then washed three times with TBS-T. Images of protein bands were obtained using the Chemi Doc XRS+ Imaging System (Bio-Rad, Hercules, CA, USA), and the densitometry values were analyzed by ImageJ (v1.53c, LOCI, University of Wisconsin, Madison, WI, USA). $\beta$-actin signals were employed for normalization.

Total RNA extraction from femur or BM-MSC samples used TRIzol reagent (Invitrogen, Waltham, MA, USA) as directed by the manufacturer. cDNA synthesis was carried out with the Prime Script RT Master Mix (Perfect Real Time; Takara, Kusatsu, Shiga, Japan). Agilent (Santa Clara, CA, USA) Touch Real-Time Detection System were used for RT-qPCR. The 2${}^{-\mathrm{\Delta }\mathrm{\Delta }\text{CT}}$ method was used as a relative quantification strategy for quantitative real-time polymerase chain reaction. Quantitative real-time RT-PCR was performed with the following primers:

IRF4: 5${}^{\prime }$-TCCGACAGTGGTTGATCGAC-3${}^{\prime }$, 5${}^{\prime }$-CTCACGATTGTAGTCCTGCTT-3${}^{\prime }$;

Runx2: 5${}^{\prime }$-ATGCTTCATTCGCCTCACAAA-3${}^{\prime }$, 5${}^{\prime }$-GCACTCACTGACTCGGTTGG-3${}^{\prime }$;

Alp: 5${}^{\prime }$-CTTGCTGGTGGAAGGAGGCAGG-3${}^{\prime }$, 5${}^{\prime }$-GGAGCACAGGAAGTTGGGAC-3${}^{\prime }$;

Ocn: 5${}^{\prime }$-GCTGCCCTAAAGCCAAACTCT-3${}^{\prime }$, 5${}^{\prime }$-GAGGACAGGGAGGATCAAGTTC-3${}^{\prime }$;

Col1a1: 5${}^{\prime }$-GCTCCTCTTAGGGGCCACT-3${}^{\prime }$, 5${}^{\prime }$-CCACGTCTCACCATTGGGG-3${}^{\prime }$.

BM-MSCs were cultured for 4 days, then the Lipofectamine 3000 kits (Invitrogen, USA) were used to transfect cells according to the manufacturer’s instructions. Sh-IRF4, OV-IRF4, and negative control (NC) plasmids were purchased from Tran Sheep Bio Company (Shanghai, China). Tareget siIRF4-2 sequence: 5${}^{\prime }$-3${}^{\prime }$ CGGCACGCGGGGCATGAACCTGG, scrambled shRNA was used as a negative control. IRF4 cDNA primer Forward (5${}^{\prime }$-3${}^{\prime }$) GCTGATCGACCAGATCGACA Reverse (5${}^{\prime }$-3${}^{\prime }$) CGGTTGTAGTCCTGCTTGC. 24 h later, the culture medium was changed with 2 mL $\alpha$MEM containing 10% FBS. Transfection efficiency could reach 63%–72% in cells detected by qRT-PCR after transfection of 25 nM Sh-IRF4, OV-IRF4 and NC plasmids using Lipofectamine 3000 according to the manufacturer’s protocol.

GraphPad Prism 6.07 (GraphPad Software, Inc., San Diego, CA, USA) was employed for data analysis. One-way ANOVA followed by Dunnett’s multiple comparisons test was performed using GraphPad Prism Software, (Boston, MA USA, www.graphpad.com). All data were analyzed for normal distribution and presented as mean $\pm$ standard deviation (SD) from assays with n = 6/group. The unpaired t-test was performed for comparisons, and two-tailed p 0.05 indicated statistical significance.

To explore the most valuable differential genes in postmenopausal osteoporosis compared to control, GSE68303 datasets which contain 12 OVX experimental groups and 11 sham groups were downloaded. Then R software was used to analyze the differentially expressed genes (DEGs) through the “Limma” package and presented in the form of a Volcano Plot (Fig. 1A). The DEGs were screened with $|$logFC$|$ $\ge$0.5 and p 0.05. Finally, 3 DEGs that were down-regulated and 18 DEGs that were up-regulated were found and presented in the heatmap (Fig. 1B).

Fig. 1.

The difference expressed genes (DEGs) in postmenopausal osteoporosis and functional enrichment analysis of the DEGs. (A) Volcano plot of DEGs between OVX and sham group. (B) The expression heatmap of DEGs. Red or green dots represent upregulated or downregulated genes, respectively. Genes without any significant difference are in black. The differences are set as $|$logFC$|$ $\ge$0.5 and p 0.05; GO and KEGG enrichment of the DEGs. (C) biological processes (D) molecular function (E) cellular component (F) KEGG pathway. OVX, ovariectomize; GO, gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

To better understand the function of the DEGs, the Enricher database [19] were used to analyze the GO enrichment and Kyoto gene and genome pathway encyclopedia (KEGG) enrichment. The main biological processes in which the DEGs were enriched are presented, including B cell receptor signaling pathway, B cell proliferation, and antigen receptor-mediated signaling pathway (Fig. 1C). Based on molecular function, DEGs were significantly enriched in chondroitin sulfotransferase activity,1-alkyl-2-acetylglycerophosphocholine ester activity, and myosin heavy chain binding (Fig. 1D). For cellular components, DEGs are specifically enriched in multivesicular bodies, focal adhesions, and cellular substrate junctions (Fig. 1E). The KEGG pathways program was used to reveal the critical pathway, in which a total of 10 pathways were identified, such as B cell receptor signaling pathway, and pentose and glucuronate interconversion (Fig. 1F).

To determine potential associations among the retrieved DEGs, a PPI network was established in the STRING database based on protein interactions. By calculating the DMNC, Clustering Coefficient and EPC with the Cystoscope plugin cytoHubba, IRF4 and CD79B were selected as the first two hub genes (Fig. 2A–C). In addition, RT-qPCR was used to quantitate their messenger RNA (mRNA) expression levels in lumbar vertebra samples from the OVX and Sham groups. The results revealed that IRF4 expression was significantly elevated in OVX animals in comparison with the Sham group (Fig. 2D). These results suggested that IRF4 might be a key regulator in the development of postmenopausal osteoporosis. Therefore, it was selected for further analysis.

Fig. 2.

Identification of the PPI networks and the hub genes. (A) The PPI network acquired from the String database. (B) Clustering Coefficient. (C) EPC of Cytoscape plugin cytoHubba. Two hub genes were selected by overlapping the top five genes based on three methods, including DMNC, Clustering Coefficient and EPC. (D) RT-qPCR of the mRNA exacted from OVX and Sham lumbar vertebrae. **, p 0.01 compared with Sham group (n = 6). PPI, protein-protein interaction; EPC, edge percolated component; DMNC, density of maximum neighborhood component; RT-qPCR, reverse transcription quantitative real-time polymerase chain reaction; IRF4, interferon regulatory factor 4; mRNA, messenger RNA.

To further determine whether IRF4 plays a key role in postmenopausal osteoporosis, immunoblot was used to detect IRF4 protein expression in OVX and Sham mice firstly. As depicted in Fig. 3A,B, these results showed that IRF4 protein expression was significantly elevated in OVX animals in comparison with the Sham group. Then immunohistochemical staining was used to detect IRF4 in lumbar vertebra samples from OVX and Sham mice. The results showed that IRF4-positive cells had significantly higher percentage in OVX animals versus the Sham group (Fig. 3C,D). These results suggest that IRF4 plays a crucial role in the pathogenetic mechanisms of postmenopausal osteoporosis.

Fig. 3.

IRF4 is essential in the occurrence of postmenopausal osteoporosis. (A) Western blots of IRF4 detected in thoracic vertebra extracts from OVX and Sham mice. (B) Protein level relative to $\beta$-actin assessed by densitometric analysis. (C) Representative micrographs from paraffin embedded sections of lumbar vertebrae stained immunohistochemically, 400$\times$. (D) The percentage of IRF4 positive cells in the vertebrae tissues relative to total cells. Values are means $\pm$ standard deviation (SD) of six determinations per group. **, p 0.01 compared with Sham group (n = 6).

To further study IRF4’s impact on osteogenic differentiation in postmenopausal BM-MSCs, the latter cells were transfected with IRF4 overexpression (OV-IRF4) and knockdown (Sh-IRF4) plasmids, the efficiency of transfection for overexpression or silencing plasmids were showed in the Supplementary Fig. 1. As depicted in Fig. 4A,B, IRF4 overexpression significantly suppressed ALP activity, while IRF4 knockdown promoted ALP activity in BM-MSCs. These results indicate that IRF4 might regulate osteoporosis via inhibiting osteogenic differentiation in BM-MSCs.

Fig. 4.

IRF4 overexpression inhibits osteogenic differentiation of postmenopausal BM-MSCs. BM-MSCs of WT mice were transfected with OV-IRF4, Sh-IRF4 or NC and then were stained with alkaline-phosphatase (ALP) (A) and (B) CFU-f–positive areas. (C) Representative radiographs and three-dimensional reconstructions of Micro-CT. Quantitative bone formation analysis of the BV/TV (D) and N.Ob/T (E). Six animals were analyzed in each group. (F) Immunohistochemical staining of Runx2. (G) Runx2 positive cells (%). (H) Osteogenesis-related mRNA levels relative to NC group, *, p 0.05, **, p 0.01, compared with shame or NC group; #, p 0.05, ##, p 0.01, compared with sh-IRF4 or OVX group. BM-MSCs, bone marrow mesenchymal stem cells; WT, wild-type; CFU-f, colony-forming unit fibroblast; BV/TV, bone volume/total volume; N.Ob/T, osteoblast number/total area.

To examine IRF4’s role in PMOP-associated osteogenic differentiation, simvastatin (an inhibitor of IRF4) was used in OVX mice. After 4 weeks of treatment, Micro-CT examination was utilized to analyze bone density and bone volume fraction (BV/TV) after IRF4 inhibition. The results showed reduced bone density and BV/TV in the OVX group compared with Sham animals, while both indexes were significantly increased in simvastatin-treated OVX groups compared with the Sham and OVX groups (Fig. 4C–E). Meanwhile, immunohistochemical staining also showed that Runx2-positive cells (Fig. 4F,G) were significantly elevated in simvastatin-treated OVX groups. In OVX animals treated with simvastatin, osteogenic Runx2, ALP, type I collagen (Col-I) and osteocalcin (OCN) mRNA amounts were significantly elevated (Fig. 4H). These results suggest that IRF4 inhibition can prevent postmenopausal osteoporosis by inducing osteogenic differentiation in BM-MSCs.