Identification of Bone Morphometric Protein-Related Hub Genes and Construction of a Transcriptional Regulatory Network in Patients With Ossification of the Ligamentum Flavum

Abstract

Study Design:

Basic science laboratory study.

Objective:

To identify hub genes related to bone morphogenetic proteins (BMPs) in the ossification of the ligamentum flavum (OLF) and analyze their functional characteristics.

Summary of Background Data:

The exact etiology and pathologic mechanism of OLF remain unclear. BMPs are pleiotropic osteoinductive proteins that may play a critical role in this condition.

Materials and Methods:

The GSE106253 and GSE106256 data sets were downloaded from the Gene Expression Omnibus database. The messenger RNA (mRNA) and long noncoding RNA expression profiles were obtained from GSE106253. The microRNA expression profiles were obtained from GSE106256. Differentially expressed genes were identified between OLF and non-OLF groups and then intersected with BMP-related genes to obtain differentially expressed BMP-related genes. The least absolute shrinkage selection operator and support vector machine recursive feature elimination were used to screen hub genes. Furthermore, a competing endogenous RNA network was constructed to explain the expression regulation of the hub genes in OLF. Finally, the protein and mRNA expression levels of the hub genes were verified using Western blot and real-time polymerase chain reaction, respectively.

Results:

We identified 671 Differentially expressed genes and 32 differentially expressed BMP-related genes. Hub genes ADIPOQ, SCD, SCX, RPS18, WDR82, and SPON1, identified through the least absolute shrinkage selection operator and support vector machine recursive feature elimination analyses, showed high diagnostic values for OLF. Furthermore, the competing endogenous RNA network revealed the regulatory mechanisms of the hub genes. Real-time polymerase chain reaction showed that the mRNA expression of the hub genes was significantly downregulated in the OLF group compared with the non-OLF group. Western blot showed that the protein levels of ADIPOQ, SCD, WDR82, and SPON1 were significantly downregulated, whereas those of SCX and RPS18 were significantly upregulated in the OLF group compared with the non-OLF group.

Conclusion:

This study is the first to identify BMP-related genes in OLF pathogenesis through bioinformatics analysis. ADIPOQ, SCD, SCX, RPS18, WDR82, and SPON1 were identified as hub genes for OLF. The identified genes may serve as potential therapeutic targets for treating patients with OLF.

Ossification of the ligamentum flavum (OLF) is a common cause of thoracic spinal stenosis.¹ The thickened ligamentum flavum (LF), in severe cases, can lead to nerve root lesions, spinal cord lesions, and cauda equina syndrome.² OLF is more common in Asians than in other groups and has the highest prevalence in the lower thoracic segment of T10 to T12.^3,4

The pathologic process of OLF involves endochondral ossification.⁵ The exact etiology and mechanism of progression from normal to ossified ligaments remain unclear. However, OLF pathogenesis is influenced by systemic and local factors, including heredity, abnormal carbohydrate metabolism, abnormal calcium metabolism, abnormal sex hormone secretion, and ligament degeneration. The latter is triggered by mechanical stresses at the enthesis, where the ligament attaches to the bone.⁶ In addition, growth factors are associated with OLF.⁷

Bone morphogenetic proteins (BMPs) are multifunctional growth factors that positively regulate the growth and differentiation of osteoblasts.⁸ The expression profiles of BMPs and BMP receptors in patients with OLF are different from those in patients with nonossified LF.^9–11 In human LF cells, BMP-2 can significantly upregulate the expression of osteogenic phenotype and induce the formation of bone nodules.¹² BMP may be involved in promoting endochondral ossification at heterotopic ossification sites in OLF.^9,13 Miyamoto et al ¹⁴ successfully induced OLF and secondary spinal cord compression symptoms in mice by implanting BMP in the epidural space. In addition, 2 novel variants of BMP-2 induce thoracic OLF in the Han Chinese population.¹³ These findings indicate that BMPs play important roles in OLF pathogenesis.

Thus, the present study aimed to explore the potential relationship between BMP expression and OLF pathogenesis. Bioinformatics was used to analyze the gene expression profiles of normal LF and OLF to identify BMP-related hub genes in OLF and then analyze their functional characteristics. The messenger RNA (mRNA) and protein expression levels of the hub genes were verified through real-time polymerase chain reaction (RT-PCR) and Western blot, respectively. This study provides a theoretical basis for elucidating the molecular mechanisms underlying BMP-related OLF and developing targeted treatments for OLF.

MATERIALS AND METHODS

Data Source

The GSE106253 and GSE106256 data sets were downloaded from the Gene Expression Omnibus database. The mRNA and long noncoding RNA (lncRNA) expression profiles were obtained from GSE106253. The microRNA (miRNA) expression profiles were obtained from GSE106256. In addition, by searching the term “Bone Morphogenetic Protein” in GeneCards, BMP-related genes were generated and then filtered for genes with scores above 0.5 in the category “Protein Coding.”

Identification of Differentially Expressed Genes

Differentially expressed genes (DEGs) were identified between the OLF and non-OLF groups of GSE106253 with the R package “limma.” Statistically significant DEGs were defined with the cutoff criteria set to log₂ fold-change >1 and P value <0.05.

Enrichment Analysis of Differentially Expressed Bone Morphogenetic Protein-Related Genes

Differentially expressed BMP-related genes (DEBRGs) were obtained by intersecting the DEGs and BMP-related genes. The obtained DEBRGs were subjected to gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses using the R package “clusterProfiler.” GO analysis included 3 categories: biological process, cellular component, and molecular function. The DEBRGs were also subjected to disease ontology enrichment analysis using the R package “DOSE.” Results were considered statistically significant if the adjusted P value <0.05.

Protein-Protein Interaction Network Analysis of Differentially Expressed Bone Morphogenetic Protein-Related Genes

Protein-protein interaction analysis was performed using the STRING website. All DEBRGs were uploaded to the STRING database, and the lowest interaction score was >0.4.

Candidate Hub Gene Screening

Genes significantly associated with the discrimination of OLF and non-OLF samples were identified through least absolute shrinkage selection operator (LASSO) regression using the R package “glmnet.” Furthermore, genes with high discriminative power were identified through support vector machine recursive feature elimination (SVM-RFE) using the R package “e1071” to select the appropriate features. The overlapping genes between the two algorithms were hub genes and evaluated by plotting receiver operating characteristic curves using the “pROC” package. Correlation coefficients between genes were calculated through Spearman correlation coefficient analysis using the R package “corrplot.”

Gene Set Enrichment Analysis

Hub genes were detected and subjected to gene set enrichment analysis to determine further their significant biological functions in OLF. The KEGG gene sets selected in the present study were downloaded from the Molecular Signatures Database.

Construction of the Competing Endogenous RNA Network

Differentially expressed miRNAs (DEmiRNAs) and differentially expressed lncRNAs (DElncRNAs) in OLF were identified from GSE106256 and GSE106253 using the R package “limma” with the cutoff criteria set to log₂ fold-change >1 and P value <0.05. mRNA-miRNA and miRNA-lncRNA pairs were matched using the StarBase database. A competing endogenous RNA (ceRNA) network based on the matched mRNA-miRNA and miRNA-lncRNA pairs was constructed and visualized using Cytoscape software.

Patients and Ligament Samples

Thoracic OLF was diagnosed by the same team of physicians based on clinical symptoms, physical examination, and anteroposterior and lateral plain films of the spine. From November 2020 to June 2022, 7 patients with thoracic OLF were enrolled in the study (4 males and 3 females; mean age, 58.7 yr; age range, 40–68 yr). Nonossified LF samples were obtained from 7 male patients who underwent posterior open decompression laminectomy for thoracic fracture during the same period (mean age, 50.1 yr; age range, 24–66 yr). Patients with cancer, posterior longitudinal ligament ossification, diffuse idiopathic skeletal hyperostosis, ankylosing spondylitis, rheumatoid arthritis, or other systemic autoimmune diseases were excluded from this study.

RNA Isolation and Real-Time Polymerase Chain Reaction

Hub genes were selected for RT-PCR assays, and the experiment was performed following the steps described in our previous research.¹⁵ The sequences of primers used for RT-PCR are listed in Table 1.

TABLE 1.

Gene and Primer Information

Gene	Login number	Sequence of primers	Length of product (bp)	Temperature of annealing (°C)
ADIPOQ	NM_004797.4	F: GTCCTAAGGGAGACATCGGT R: AACGTAAGTCTCCAATCCCACA	147	59
RPS18	NM_022551.3	F: ACCAAGAGGGCGGGAGAA R: TGGCTAGGACCTGGCTGTATT	145	60
SCD	NM_005063.5	F: TTGCGATATGCTGTGGTGCT R: TCATAGGCCAGACCGAGGG	251	60
SCX	NM_001080514.3	F: GACGGCGAGAACACCCAG R: TGCCCAGCTCAGGTCCAA	198	60
SPON1	NM_006108.4	F: CACCACCAGCGGGAACA R: TCCTTTGGGTGTGTCTTCTCG	223	59
WDR82	NM_025222.4	F: GCGAGGCTCCACAGTCCA R: GAAGCTCCGCAACACGCT	230	59
Actin	NM_001101.3	F: TGGCACCCAGCACAATGAA R: AGGGTGTAACGCAACTAAGTCATAG	200	59

Extraction of Total Tissue Protein and Western Blot Analysis

The OLF and non-OLF tissue blocks were washed three times with prechilled phosphate-buffered saline to remove blood stains and then sectioned into fine pieces. The experiment was performed following the steps described in our previous research.¹⁵ Antibodies used for Western blotting are listed in Table 2.

TABLE 2.

Antibodies Used in Western Blotting

Proteins (Dilution)	Catalog Number	Source
ADIPOQ (1:1000)	ab13347	Abcam
SCD (1:1000)	ab23686	Abcam
SCX (1:1000)	ab58655	Abcam
RPS18 (1:1000)	DF3679	Affinity Biosciences
WDR82 (1:1000)	29317-1	Signalway Antibody
SPON1(1:1000)	40372-1	Signalway Antibody
Actin (1:1000)	GB12001	Servicebio

Statistical Analyses

All statistical analyses were performed using R software (version 3.6.3) and SPSS 25.0 for Windows. Data are presented as mean ± SD. Student t test was used to compare statistical data. Each experiment was repeated at least thrice. Statistical significance was considered at P <0.05.

RESULTS

Identification of Differentially Expressed Genes and Differentially Expressed Bone Morphogenetic Protein-Related Genes in Ossification of the Ligamentum Flavum

The detailed workflow diagram of this study is depicted in Figure 1. In total, 671 DEGs were identified, of which 327 were upregulated and 344 were downregulated (Figure 2A). The expression patterns of the DEGs are shown in Figure 2B. The 32 DEBRGs obtained after intersecting the DEGs and 2555 BMP-related genes were visualized for their expression patterns using a heatmap (Figure 2C, D).

Figure 1.

Flowchart of the analysis process. BMP indicates bone morphogenetic protein; ceRNA, competing endogenous RNA; DEG, differentially expressed gene; GEO, Gene Expression Omnibus; GO, gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; LASSO, least absolute shrinkage selection operator; miRNA, microRNA; PPI, protein-protein interaction; ROC, receiver operating characteristic; RT-PCR, real-time polymerase chain reaction; SVM-RFE, support vector machine recursive feature elimination.

Figure 2.

Identification and expression analysis of DEGs and DEBRGs in patients with OLF. A, Volcano plot showing the number and distribution of differential DEGs; red and blue dots represent upregulated and downregulated genes, respectively. B, Heat maps of gene expression in the OLF and non-OLF groups; red and blue indicate high and low gene expression levels, respectively. C, Venn diagram showing the intersection between DEGs and BMP-related genes to obtain DEBRGs. D, Heat maps of DEBRG expression in the OLF and non-OLF groups; red and blue indicate high and low gene expression levels, respectively. BMP indicates bone morphogenetic protein; DEBRG, differentially expressed BMP-related gene; DEG, differentially expressed gene; OLF, ossification of the ligamentum flavum.

Enrichment Analysis and Protein-Protein Interaction Network Construction of Differentially Expressed Bone Morphogenetic Protein-Related Genes

The results obtained from the 3 parts of the GO enrichment analysis are shown in Figure 3A. The DEBRGs were mainly enriched in “transmembrane receptor protein serine/threonine kinase signaling pathway (ontology: biological process),” “focal adhesion (ontology: cellular component),” and “receptor ligand activity (ontology: molecular function).” The top 5 enriched KEGG pathways of the DEBRGs were the “transforming growth factor (TGF)-beta signaling pathway,” “Wnt signaling pathway,” “focal adhesion,” “lipid and atherosclerosis,” and “transcriptional misregulation in cancer” (Figure 3B). The DEBRGs were enriched in different disease ontology terms, such as “rheumatoid arthritis,” “osteoporosis,” “hepatitis B,” “non-small cell lung cancer,” and “breast cancer” (Figure 3C). The protein-protein interaction network of the DEBRGs was analyzed using the STRING website. The interaction network of the DEBRG proteins is shown in Figure 3D.

Figure 3.

Enrichment analysis and PPI network construction of DEBRGs. A, GO functional analysis results illustrating the top 10 significantly enriched terms for DEBRGs. B, KEGG analysis showing the top 10 pathways of DEBRGs. C, DO enrichment results illustrating the top 10 significantly enriched diseases for DEBRGs. D, PPI network of DEBRGs. DEBRG indicates differentially expressed BMP-related gene; DO, disease ontology; GO, gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; PPI, protein-protein interaction;

Identification of Hub Genes and Verification of Their Diagnostic Characteristics

Hub genes were screened using 2 algorithms. The DEBRGs were narrowed down through LASSO regression (Figure 4A, B), which identified 8 variables as hub genes for OLF. A subset of 14 features among the DEBRGs was determined using SVM-RFE (Figure 4C, D). Six overlapping features (ADIPOQ, SCD, SCX, RPS18, WDR82, and SPON1) between these two algorithms were ultimately selected (Figure 4E). As shown in Figure 4F, the 6 hub genes showed favorable abilities to discriminate OLF from the non-OLF samples, with an area under the curve values of 0.90, 1.00, 0.81, 1.00, 1.00, and 0.88 in ADIPOQ, SCD, SCX, RPS18, WDR82, and SPON1, respectively. The expression of all 6 genes decreased in OLF (Figure 4G). Subsequently, correlation analysis of the 6 genes (Figure 4H) revealed that SCD and WDR82 (correlation = 0.89) had the highest positive correlation.

Figure 4.

Identification and validation of diagnostic feature biomarkers. A, LASSO coefficient spectrum of DEBRGs. B, Ten cross-validations for adjusting parameters in LASSO analysis. C, Accuracy of the SVM-RFE algorithm. D, Error of the SVM-RFE algorithm. E, Venn diagram showing the intersection among genes screened by the LASSO and SVM-RFE algorithms as the hub genes. F, ROC curves for assessing the diagnostic accuracy of the hub genes. G, Expression of hub genes in the OLF and non-OLF groups. H, Correlation analysis of hub gene expression. DEBRG indicates differentially expressed BMP-related gene; LASSO, least absolute shrinkage selection operator; OLF, ossification of the ligamentum flavum; ROC, receiver operating characteristic; SVM-RFE, support vector machine recursive feature elimination.

Gene Set Enrichment Analysis of Six Hub Genes

As shown in Figure 5 (sorted by fold change of DEGs), “cytokine-cytokine receptor interaction” and “olfactory transduction” were enriched in the highly expressed genes of RPS18, SCD, SCX, SPON1, and WDR82, and “MAPK signaling pathway” and “pathways to cancer” were enriched in the lowly expressed genes of RPS18, SCD, SCX, SPON1, and WDR82. However, all 4 pathways were enriched in the lowly expressed gene of ADIPOQ.

Figure 5.

Gene set enrichment analysis results of ADIPOQ, SCD, SCX, RPS18, WDR82, and SPON1.

Construction of a Competing Endogenous RNA Network

A total of 70 DEmiRNAs (Figure 6A) and 186 DElncRNAs (Figure 6B) were identified between the OLF and non-OLF groups. Subsequently, 404 miRNAs were predicted to target the 6 diagnostic mRNAs by using the StarBase database. Twenty-eight overlapped miRNAs (mRNA-miRNA pairs) were obtained between the predicted DEmiRNAs and miRNAs (Figure 6C). Similarly, 1116 lncRNAs were predicted to interact with 28 miRNAs by using the StarBase database. Fifteen overlapping lncRNAs (miRNA-lncRNA pairs) were identified between the predicted DElncRNAs and lncRNAs (Figure 6D). Figure 6E shows the constructed mRNA-miRNA-lncRNA ceRNA network, including 6 mRNAs, 15 miRNAs, and 15 lncRNAs.

Figure 6.

Prediction of hub gene regulatory networks. A, Volcanic plot showing the number and distribution of DEmiRNAs in the GSE106256 data set; red and blue dots represent upregulated and downregulated genes, respectively. B, Volcanic plot showing the number and distribution of DElncRNAs in the GSE106253 data set; red and blue dots represent upregulated and downregulated genes, respectively. C, Venn diagram showing the overlapping miRNAs predicted using the StarBase database and DEmiRNA. D, Venn diagram showing the overlapping lncRNAs predicted using the StarBase database and DElncRNA. E, OLF associated mRNA-miRNA-lncRNA competitive endogenous RNA triple regulatory network. DElncRN indicates differentially expressed lncRNAs; DEmiRNA, differentially expressed miRNA; lncRNA, long noncoding RNA; miRNA, microRNA; mRNA, messenger RNA; OLF, ossification of the ligamentum flavum.

Messenger RNA Expression of Hub Genes

The expression levels of the hub genes were paralleled through microarray and RT-PCR analyses (Figure 7).

Figure 7.

Validation of the expression of these genes through RT-PCR. mRNA levels of ADIPOQ, SCD, SCX, RPS18, WDR82, and SPON1 were significantly downregulated in the OLF group compared with the non-OLF group. **P < 0.01. CON indicates Nonossified ligamentum flavum; mRNA, messenger RNA; OLF, ossification of the ligamentum flavum; RT-PCR, real-time polymerase chain reaction.

Protein Expression of Hub Genes

The protein expression levels of ADIPOQ, SCD, WDR82, and SPON1 were significantly lower in the OLF group than in the non-OLF group (P < 0.01), whereas the protein expression levels of SCX and RPS18 were significantly higher in the OLF group than in the non-OLF group (P < 0.05, Figure 8).

Figure 8.

Protein levels of ADIPOQ, SCD, WDR82, and SPON1 were significantly downregulated, whereas those of SCX and RPS18 were significantly upregulated in the OLF group compared with the non-OLF group. *P < 0.05, **P < 0.01. CON indicates Nonossified ligamentum flavum; OLF, ossification of the ligamentum flavum.

DISCUSSION

OLF is a multifactorial disease, and BMPs may be influential in the pathogenesis of this disease process.^9,16 However, the specific biological mechanisms mediated by BMPs remain unclear. Therefore, this study focused on BMP-related genes involved in OLF pathogenesis.

DEBRGs in OLF were identified and subjected to enrichment analyses. KEGG enrichment analysis showed that the DEBRGs were enriched in the “TGF-beta signaling pathway” and “Wnt signaling pathway.” The TGF-β signaling pathway is involved in bone formation during mammalian development and in various types of ectopic ossification.^17,18 TGF-β is expressed in all surgically resected OLF lesions.¹¹ Using whole exome sequencing in patients with thoracic OLF, Qu et al ¹³ found many possible pathogenic variations in genes involved in the TGF-β signaling pathway in patients with long-segment thoracic OLF, suggesting that thoracic OLF is associated with mutations of genes involved in the TGF-β signaling pathway. Wnt signaling plays a key role in ossification maturation. Yayama et al ¹⁹ found that Wnt signaling pathway molecules are highly expressed in OLF, and their upregulation can accelerate and control chondrocyte maturation. Therefore, we speculated that BMPs affect OLF pathogenesis through the abnormal expression of TGF-β and Wnt signaling pathway molecules, but the exact mechanisms require further research.

In this study, we identified the ADIPOQ, SCD, SCX, RPS18, WDR82, and SPON1 hub genes as novel BMP-related diagnostic markers in OLF through LASSO and SVM-RFE and evaluated the diagnostic performance of these genes through receiver operating characteristic curve analysis. All hub genes could be used to distinguish between the OLF and non-OLF groups.

ADIPOQ, an important adipocytokine secreted by adipocytes, plays an important role in regulating inflammation, glucose and lipid metabolism, and oxidative stress.²⁰ As a typical anti-inflammatory agent, ADIPOQ reduces inflammatory responses to multiple stimuli by modulating signaling pathways in various cell types.^21,22 ADIPOQ knockout upregulates the expression of tumor necrosis factor (TNF)-α and interleukin (IL)-6 in mice.²³ In addition, this gene regulates the action and production of TNF-α in various tissues.²² TNF-α and IL-6 play key roles in OLF.²⁴ In the present study, the mRNA and protein expression of ADIPOQ was downregulated in the case group. Therefore, we speculate that ADIPOQ participates in OLF pathogenesis by altering metabolism and reducing the inhibitory effect of inflammatory responses.

SCD belongs to the fatty acid desaturase family, of which SCD1 is the major subtype and is widely expressed in human tissues.²⁵ Inhibition of SCD1 upregulates the expression of proinflammatory genes, which promote cellular inflammation and stress.²⁶ Downregulation of SCD1 indirectly upregulates the gene expression of proinflammatory cytokines, such as IL-6 in adipocytes and TNF-α in macrophages.²⁷ In vivo studies in mice have shown that a lack of SCD1 increases plasma IL-6 levels.²⁸ In addition, SCD regulates ADIPOQ, and inhibition of SCD1 can significantly downregulate ADIPOQ expression.²⁹ In the present study, the mRNA and protein expression levels of SCD in the OLF group decreased. We speculate that SCD participates in OLF pathogenesis by directly regulating the expression of proinflammatory genes or indirectly regulating ADIPOQ to promote inflammatory responses.

As a basic helix-loop-helix transcription factor, SCX is essential for the normal development and maturation of tendons and ligaments, and its late expression is limited to developing tendons and ligaments.³⁰ The expression of SCX is regulated by mechanical force and upregulated with an increase in tension load.^31,32 Mechanical stress is a potential cause of OLF development.⁷ In the present study, SCX protein expression was significantly higher in the OLF group than in the non-OLF group. This finding indirectly confirms that mechanical stress triggers OLF development.

In most cases of OLF, BMP receptors are strongly expressed in chondrocytes surrounding the calcified zone.⁹ Dey et al ³³ and Agarwal et al.³⁴ reported that the hyperactivity of BMP receptors induces the chondrogenic differentiation of SCX lineages to provide a substrate for ectopic endochondral ossification, resulting in heterotopic ligament ossification. Therefore, SCX lineages may be involved in OLF pathogenesis through recruitment and chondrogenic differentiation at LF ossification sites.

RPS18 is a component of the 40S ribosomal subunit that is involved in translation initiation.³⁵ Kusui et al ³⁶ suggested that RPS18 affects the binding of ribosomes and the actin cytoskeleton by interacting with the actin-binding protein cofilin, thus affecting protein synthesis. RPS18 is stably expressed in different types of cells and tissues.³⁷ However, in the present study, the protein expression of RPS18 was significantly higher in the OLF group than in the non-OLF group. This result may be related to the increased demand for protein synthesis caused by metabolic changes. Therefore, we speculated that RPS18 is involved in the abnormal metabolism associated with OLF. Meanwhile, the mRNA level of RPS18 was lower in the OLF group than in the non-OLF group. Thus, RPS18 may have other transcriptional translation regulatory mechanisms.

WDR82, a component of mammalian histone H3 at the lysine4 trimethylated (H3K4me3) methyltransferase complex,^38,39 is a key epigenetic factor. Hou et al ⁴⁰ suggested that histone H3K4me3 modification creates a favorable microenvironment for Osterix and Runx2 to access osteogenic genes during OLF, indicating that histone H3K4me3 plays an important role in OLF. Transcription factors Osterix and Runx2 play important roles in OLF development.⁴¹ Therefore, we hypothesized that WDR82 regulates OLF development by creating a favorable microenvironment for transcription factors (Osterix and Runx2) to access osteogenic genes through histone H3K4me3 modification during OLF.

The expression of WDR82 is downregulated during stem cell differentiation.⁴² The pathogenesis of OLF involves the differentiation of mesenchymal stem cells in the LF into osteoblasts, after which the ligament tissue is replaced by the bone matrix secreted by osteoblasts.⁴⁰ In the present study, the mRNA and protein expression levels of WDR82 were downregulated in the case group, providing novel indirect evidence for the involvement of mesenchymal stem cells in OLF pathogenesis.

SPON1 is an important member of the thrombospondin family.⁴³ Decreased SPON1 activity reduces TGF-β levels, thereby activating the Smad family of transcription factors that, together with RUNX2, promote BMP-driven osteoblast differentiation and bone formation.⁴⁴ Palmer et al ⁴⁵ demonstrated that SPON1 loss results in the local and systemic reduction of TGF-β in adult mice, leading to enhanced BMP signaling and increased bone deposition. SPON1 also exerts antiangiogenic activity.⁴⁶ Angiogenesis is tightly controlled by the balance between angiogenic factors and inhibitors. A large number of new blood vessels grow into the cartilage in the early stages of OLF.¹⁶ In the present study, the mRNA and protein expression levels of SPON1 in the OLF group were downregulated. Therefore, SPON1 might be involved in OLF pathogenesis by regulating TGF-β expression and angiogenesis.

Our study has limitations. First, the sample size for screening hub genes in OLF is small and limited in the public database. Second, the effects of the hub genes on osteogenesis and the molecular mechanisms in regulating OLF were not explored through in vitro and in vivo experiments. Third, their diagnostic value in the real clinical world was not verified. In future studies, we plan to collect more examples of OLF and normal LF tissue and evaluate their association with OLF through hematoxylin-eosin staining and immunohistochemical staining. In addition, we will study the effects of the hub genes on osteogenesis in vitro by using alizarin red and alkaline phosphatase staining through overexpression and knockdown. Furthermore, based on our ceRNA network results, we will investigate the mechanisms by which the hub genes regulate OLF.

CONCLUSION

This study identified 6 BMP-related genes in OLF and evaluated their diagnostic performance. Molecular biology experiments were used to verify the results of bioinformatics analysis. This study is the first to identify and functionally characterize BMP-related genes involved in OLF pathogenesis through bioinformatics analysis. The results of this study serve as a theoretical basis for elucidating the molecular mechanisms underlying OLF pathogenesis and provide novel biomarkers for the diagnosis and treatment of patients with OLF.

Key Points

• ADIPOQ and SCD may participate in the occurrence and development of OLF by altering metabolic and inflammatory responses in the body.

• SCX lineages may be involved in OLF pathogenesis through recruitment and chondrogenic differentiation at LF ossification sites.

• RPS18 may be involved in abnormal metabolism associated with OLF.

• WDR82 may regulate OLF development by creating a favorable microenvironment for transcription factors to access osteogenic genes.

• SPON1 may be involved in OLF pathogenesis by regulating TGF-β expression and angiogenesis.