Single-cell transcriptome profiling identifies the activation of type I interferon signaling in ossified posterior longitudinal ligament

Abstract

Ossification of the posterior longitudinal ligament (OPLL) is a condition comprising ectopic bone formation from spinal ligaments. This disease is a leading cause of myelopathy in the Asian population. However, the molecular mechanism underlying OPLL and efficient preventive interventions remain unclear. Here, we performed single-cell RNA sequencing and revealed that type I interferon (IFN) signaling was activated in the ossified ligament of patients with OPLL. We also observed that IFN-β stimulation promoted the osteogenic differentiation of preosteoblasts in vitro and activated the ossification-related gene SPP1, thereby confirming the single-cell RNA sequencing findings. Further, blocking the IFN-α/β subunit 1 receptor (IFNAR1) using an anti-IFNAR1 neutralizing antibody markedly suppressed osteogenic differentiation. Together, these results demonstrated that the type I IFN signaling pathway facilitated ligament ossification, and the blockade of this signaling might provide a foundation for the prevention of OPLL.

Keywords

ossification / posterior longitudinal ligament / osteogenic differentiation / IFN signaling

1 Introduction

Ossification of the posterior longitudinal ligament (OPLL) is a degenerative spinal disease characterized by progressive ectopic bone formation in the spine [1,2]. This disease usually develops to produce severe neurological symptoms, such as myelopathy and radiculopathy, due to spinal cord compression. The incidence of OPLL is significantly higher in adults over the age of 50 in Asian countries, especially Japan, China, and the Republic of Korea, compared with that in European and American countries [3,4]. OPLL susceptibility genes have been identified, including those encoding Runx2, collagen 6A1, and interleukin-1β [5–7]. In addition, some systemic hormones and growth factors are thought to be involved in the initiation and development of OPLL, and they include calcium regulating hormones, transforming growth factor-β (TGF-β), and BMP2/4/7 [8]. Moreover, parathyroid hormone is reported to induce osteogenesis in a mouse model of spinal degeneration [9]. Feng et al. found that deletion of suppressor of fused (Sufu) results in spontaneous and progressive ossification [10]. Some non-coding RNAs, such as microRNA-181 and miR-320e, also play important roles in the development of OPLL [11,12]. In addition, an IFN-γ gene polymorphism was found to be associated with OPLL susceptibility in population of the Republic of Korea [13]. However, whether IFN signaling controls ligament ossification remains unknown.

Interferons (IFNs) comprise a heterogeneous family of cytokines originally known as mediators of cellular responses to viral infections. Type I IFNs are mainly involved in the innate immune response, whereas type II IFNs regulate the adaptive immune response [14]. Type I (IFN-α/β) and type II (IFN-γ) IFNs are produced in virus-infected cells or toll-like receptor 9-activated plasmacytoid dendritic cells [15,16]. IFN-α/β binds a heterodimer transmembrane receptor, the IFN-α/β subunit receptor (IFNAR), which comprises IFNAR1 and IFNAR2 subunits. In the canonical type I IFN signaling pathway, this binding activates the receptor-associated protein tyrosine kinases Janus kinase 1 (JAK1) and tyrosine kinase 2 (TYK2), which phosphorylate signal transducer and activator of transcription (STAT) proteins [17]. Phosphorylated STATs translocate into the nucleus, activating IFN-stimulated response genes (ISGs), such as IRF7, OAS, MX, and IFMT, which ultimately mediate an important host defense response against viral infection [18].

Uncontrolled IFN signaling is associated with various diseases such as autoimmune diseases, chronic infections, and cancer [19–21]. Moreover, the type I and II IFN pathways play important roles in bone formation by suppressing osteoclastogenesis. Seeliger et al. found that IFN-β inhibits osteoclastogenesis during the pathogenesis of osteoporosis [22]. In addition, IFNAR1- or IFN-β-deficient mice exhibit a significant increase in osteoclast number and a reduction in trabecular bone mass, indicating a role for type I IFN signaling in regulating osteoclastogenic bone resorption [23]. Furthermore, IFN-γ plays a role in the osteogenic differentiation and mineralization of mesenchymal stem cells (MSCs) by inducing BMP-2 [24,25]. As such, IFN-γ receptor knockout mice exhibit an osteoporotic phenotype, which is associated with a decrease in bone formation [26]. Although IFNs are functionally important for osteoclastogenesis, very little is known about their roles in osteogenic differentiation.

Here, we showed that the type I IFN pathway promotes osteogenic differentiation. Moreover, blockade of the type I IFN pathway using an anti-IFNAR1 neutralizing antibody was found to be effective in suppressing ossification. Our findings provide insight into the potential of the type I IFN pathway to serve as a therapeutic target for OPLL.

2 Materials and methods

2.1 Tissue acquisition and cell preparation

Six unrelated Chinese patients with T-OPLL and four control patients were recruited from the Department of Orthopedics at Peking University Third Hospital (PUTH). The diagnosis of T-OPLL or control patients was confirmed using X-rays, computerized tomography, and magnetic resonance imaging preoperatively in the institution. Ossified ligament tissues were from circumferential decompression (CD) surgery of patients with T-OPLL. The control group comprised patients of lumbar disc herniation who underwent discectomy. The ligament tissues from T-OPLL or control patients were immediately dissected under a microscope after surgery and subjected to primary cell culture in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. The adherent cells that migrated from the tissues were primary ligament cells for further analysis.

2.2 scRNA-seq library construction

Sequencing libraries were obtained using a modified single-cell tagged reverse transcription method, as previously reported [27]. In brief, after isolation, a single cell was suspended in a lysis buffer by using a mouth pipette. Reverse transcription was performed using a 25-nt oligo (dT) primer anchored with an 8-nt cell-specific barcode and 8-nt unique molecular identifiers (UMIs). First- and second-strand cDNAs were then synthesized, followed by 16 cycles of amplification. The amplified cDNAs were pooled, and biotinylated pre-indexed primers were used for further amplification of the amplicons by 4-cycle PCR to introduce biotin tags at the 3ʹ-ends of the amplified cDNAs. Approximately 300 ng of cDNA was then sheared to 300 bp using Covaris S2. The 3ʹ-terminals of the amplified cDNAs were purified using Dynabeads MyOne Streptavidin C1 beads (Thermo Fisher Scientific). Libraries were constructed using the KAPA Hyper Prep Kit (KAPA Biosystems) and then subjected to 150 bp paired-end sequencing on an Illumina HiSeq 4000 platform (Novogene). The raw data have been deposited in GEO under the accession code GSE241505.

2.3 Sequencing data processing

We used the extract tool in UMI-tools v0.5.4 to extract and add the 8-bp barcode sequence and UMIs to the read name of sequencing read pairs. Subsequently, we trimmed the template switch oligo sequence and poly-A tail sequence in read and removed adapter contamination and low-quality bases using costumed script and seqtk tool v1. We then aligned the clean reads to the GRCh38 human reference genome using STAR v2.6.0a, followed by the addition of genomic features to the reads using FeatureCounts from Subread v1.6.2 to count the number of reads uniquely aligned with each gene. We then used the count tool in UMI-tools to remove PCR duplicates based on the barcode and UMI information. Cells were filtered if the number of detected genes (log₁₀ scale) was below the median of all cells minus 3 × the median absolute deviation or during the G2M cell cycling phase. Overall, 1569 single cells were retained for subsequent analysis (900 cells from the patient group and 669 cells from the control group).

2.4 Dimension reduction and cell clustering

The Seurat program (R package, v.4.1.0) and the Scanpy program were applied for analysis of RNA-sequencing data. After normalization and scaling the expression matrix, we selected 4128 high-variable genes for PCA analysis. We integrated the expression matrix using BBKNN to neutralize batch effects. Cells were separated into six clusters by using the top 40 principal components. We calculated the differentially expressed genes (DEGs) using the “FindAllMarkers” function in Seurat package, and genes meeting the criteria of log₂FC threshold > 0.25 and min.pct > 0.25 were regarded as DEGs.

2.5 Monocle-2-based pseudotime trajectory inference

We constructed the single-cell trajectory of cells by using a reversed graph embedding method implemented in the R Monocle 2 package (v. 2.6.3) [28]. DEGs between patients and healthy controls were used to order cells in Monocle to construct the cell DDRtree trajectory plot. The signature of each state was calculated based on DEGs over the other states.

2.6 Transcription factor regulatory analysis

The SCENIC package (version 1.1.2) with default settings was used to infer active TFs and their target genes in all cells. In brief, we constructed co-expression modules with GRNBoost2 and then inferred the regulons by cisTarget. The activities of selected regulons were calculated by AUCell. We also applied DoRotheEA (v1.8.0) to estimate the activity of the most variable transcription factors in the cells. Human regulons (grades A–C) were obtained from the DoRothEA R package. We extracted the most variable transcription factors over clusters. The predicted transcription factors were compared with their corresponding gene expression, and those with low real expression (normalized expression < 0.5) were removed from analysis.

2.7 Transcriptome sequencing (RNA-seq)

A minimum of 3 μg of total RNA was oligo (dT) selected using the Dynabeads mRNA purification kit (Invitrogen). The mRNAs isolated from total RNA were fragmented into short fragments with a fragmentation buffer (Ambion). Double-stranded cDNA was synthesized with these short fragments as templates. The cDNA was end-repaired, ligated to Illumina adapters, size selected on agarose gel (approximately 250 bp), and PCR amplified. The cDNA library was sequenced on a NovaSeq 6000-PE150 sequencing platform (Illumina). Reads were aligned to the mm10 reference genome and counted using HISAT2, Samtools, and FeatureCounts. mFUZZ package (v2.62) was used for time-course gene clustering [29]. The raw data have been deposited in GEO under the accession code GSE241506.

2.8 Alizarin red staining

Murine MC3T3-E1 preosteoblasts cells were induced with OriCell® Mouse MC3T3-E1 Osteogenic Differentiation Basal Medium (MUXMT-90021, Cyagen Biosciences Inc.). After osteogenic induction, cells were fixed with 4% paraformaldehyde, washed with PBS, and stained according to the manufacturer’s instructions.

2.9 RNA isolation and quantitative real-time PCR (qRT-PCR)

Total RNA was extracted using TRIzol reagent (Invitrogen). RNA nuclear or cytoplasmic fraction was performed using the Cytoplasmic & Nuclear RNA Purification Kit (#21000, Norgen, BiotekCorporation, CA) according to the manufacturer’s instructions. A HiScript II Q RT SuperMix Kit (Vazyme, China) was used to synthesize the cDNA. A ChamQ SYBR qPCR Master Mix (Vazyme, China) was used for real-time PCR with a Light Cycler 96 detection system (Roche). Primer sequences were as follows: hSPP1-F-TGAAACGAGTCAGCTGGATG, R-TGAAATTCATGG CTGTGGAA; hIRF7-F-GAGCCCTTACCTCCCCTGTTAT, R-CCACTGCAGCCCCTCATAG; hOAS2-F-ACGTGACATCCTCGATAAAACTG, R-GAACCCATCAAGGGACTTCTG; mSPP1-F-AGACCATGCAGAGAGCGAG, R-GCCCTTTCCGTTGTTGTCCT; mIRF7-F-GAGACTGG CTATTGGGGGAG, R-GACCGAAATGCTTCCAGGG; mOAS2-F-TTGAAGAGGAATACATG CGGAAG, R-GGGTCTGCATTACTGGCACTT.

2.10 Statistical analyses

All statistical analyses were performed using GraphPad Prism software. Unpaired t-test with Welch’s correction or non-parametric test, Mann–Whitney test, was used to analyze the two groups. One-way ANONA with a non-parametric test, Kruskal–Wallis test, was used to analyze multiple groups. All results are presented as mean ± SD. Statistical significance was set at P < 0.05.

3 Results

3.1 Single-cell RNA sequencing (scRNA-seq) reveals disease patterns in patients with OPLL

To clarify the changes in cell population characteristics and reveal pathological gene expression patterns in patients with T-OPLL, we performed scRNA-seq analyses based on six patients with T-OPLL undergoing circumferential decompression-based surgery and four control patients with lumbar disc herniation (Fig.1). We isolated 1920 individual cells from the marginal zone of ossified tissue of patients with T-OPLL and the normal posterior longitudinal ligament of control patients. After low-quality cells were filtered out, the transcriptomes of 1702 high-quality single-cell samples were captured and profiled, and 8537 average genes per sample were detected. We then reduced dimensionality through principal component analysis and visualized the data via Uniform Manifold Approximation and Projection (UMAP) analysis (Fig.1 and 1B). Most single cells were distributed in the G1 phase, conforming to the salient features of ligament cells. To precisely depict the process of ligament cell differentiation, we excluded actively proliferating cells in the G2/M phase from subsequent analyses. The MSC/OPLL expression signatures were evaluated via gene set variation analysis (GSVA) based on the analysis of previously reported DEGs between patients with T-OPLL and control individuals [30]. Notably, single cells from the normal ligaments displayed more MSC-like patterns, whereas marginal cells associated with ossification from patients with OPLL had more ossified features at the transcriptional level (Fig.1). Moreover, DEGs between controls and patients with OPLL (Fig.1) included those involved in canonical ossification (Fig.1).

Fig.1 Distinct subpopulations with transcriptional signatures determined via single-cell RNA-seq analysis of patients with ossification of the posterior longitudinal ligament (OPLL). (A) Diagram of the single-cell sequencing strategy for patients with OPLL (P) and normal donors (N). The image of hematoxylin–eosin staining showing the ossified ligaments from patients with OPLL. The dashed box represents the marginal zone of tissue ossification from which single cells were isolated. The Uniform Manifold Approximation and Projection (UMAP) embedding plot showing cell distribution in the cell cycle. (B) The UMAP embedding plot showing cell distribution from the P group (patients with T-OPLL) and N group (normal donors). Cells in the G2/M phase were excluded. (C) Histogram showing the gene set variation analysis (GSVA) scores for DEG (log2FC > 1.5, P < 0.05) signatures of mesenchymal stem cells (MSCs) from normal donors and patients with OPLL (GSE153829). Statistical testing was performed via a Wilcoxon test. (D) Volcano plot showing the differentially expressed genes between controls and patients with OPLL; P < 0.05, log2(fold-change) ≥ 0.5. Statistical testing was performed via a two-sided Wilcoxon test. (E) Histogram showing the expression levels of canonical ossification marker genes. Statistical testing was performed via a two-sided Wilcoxon test.

Full size|PPT slide

Cells from the control or OPLL groups with similar expression patterns were clustered with a graph-based clustering method. All cells were grouped into six unsupervised clusters, and marker genes of each cluster were selected according to the fold change in differential expression and differences in the proportion of cells expressing them (Fig.2 and 2B). Clusters 0, 3, and 4 were mainly present in the OPLL group, and they were involved in responses to type I IFNs, DNA replication, and other biological processes. Cluster 1 was mostly present in the control group and found to participate in extracellular matrix organization. Cells from Clusters 2 and 5 were distributed in both groups, suggesting that these cells represent a transitional stage from normal ligament cells to marginal cells of ossification (Fig.2). GSEA GO enrichment analysis showed that type I IFN-related responses were significantly enriched in cells from patients with OPLL (Fig.2). We then analyzed the dataset of OPLL RNA-seq from GSE5464 (Fig.2) and intersected the genes whose expression was upregulated in patients with OPLL (GSE5464) and those identified with elevated expression in Cluster 3. The results showed that six genes were related to the type I IFN pathway based on GO enrichment analysis, including OAS2 and IRF7 (Fig.2). Thus, type I IFN pathway-related genes might be activated in cells from the ossified ligament.

Fig.2 Type I interferon-related pathway is enriched in Cluster 3 from patients with ossification of the posterior longitudinal ligament (OPLL). (A) Uniform Manifold Approximation and Projection (UMAP) embedding plot showing clusters. Cells were colored according to their clusters. (B) A dot plot showing marker genes of each cluster. (C) Heatmap showing distinct expression features of each cluster. The color key from blue to red indicates low to high gene expression, respectively. The left side shows significantly enriched biological process terms based on GO enrichment analysis. (D) Enrichment plot showing that type I interferon-related responses were enriched in cells from patients with OPLL. (E) A volcano plot showing the differentially expressed genes between controls and patients with OPLL from GSE5464; P < 0.05, log2(fold-change) ≥ 2. Statistical testing was performed via a two-sided Wilcoxon test. (F) Venn diagram visualizing the intersection between genes for which expression was upregulated in patients with OPLL from the GSE5854 data set and genes identified with elevated expression in Cluster 3.

Full size|PPT slide

3.2 Type I IFN signaling pathway is activated during the progression of ossification

Monocle 2 was used to establish a pseudo-time map of OPLL and uncover the developmental trajectory from normal ligament cells to marginal cells of ossification. Regarding the pseudo-time trajectory of OPLL, one significant branch point and three distinct states were observed (Fig.3). Cells derived from the normal ligaments were distributed in state 1, whereas cells from patients with OPLL were mainly located at two distinct branches, states 2 and 3. The cells in Clusters 1 and 2 were mainly distributed in state 1. Cells in Clusters 0 and 3 accounted for the largest proportion of states 2 and 3, respectively (Fig.3). Furthermore, dynamically changing genes among trajectories were estimated and divided into two gene sets (Fig.3). Genes enriched in Gene Set 1 were obviously upregulated during transdifferentiation from state 1 to state 3, and these genes were mainly involved in ossification and the type I IFN pathway (Fig.3). The expression levels of some IFN-related genes, such as OAS2, OAS3, IRF7, MX1, MX2, and STAT1, were higher in state 3 than in states 1 and 2 (Fig.3). Consistently, SCENIC analysis identified that the IFN-related transcription factors, IRF7 and STAT1, were activated in the cells of Cluster 3 in the OPLL patient group (Fig.3). The expression levels of IFN-related genes including OAS2, OAS3, IRF7, and the ossification-related gene SPP1 were higher in the OPLL group than in the control group (Fig.4). To confirm the scRNA data, we extracted total RNA from normal or ossified ligaments to perform RT-qPCR. The mRNA levels of these IFN-related genes increased in the OPLL patient group (Fig.4). Hematoxylin and eosin and fluorescence staining showed that OAS2 was highly expressed in the ossified region of the ligaments from patients with OPLL (Fig.4). These data indicated that the type I IFN pathway was significantly activated in ossified tissues.

Fig.3 Reconstruction of the differentiation trajectories of cell subpopulations in a pseudotime manner. (A) The developmental trajectory of cells was constructed by Monocle2. Cells were colored according to the pseudotime, state, and group. (B) Histogram showing percentages of cell subpopulations in different states. (C) Heatmap showing the patterns of gene expression along with the pseudotime. The color scheme represents the z-score distribution from −3 (blue) to 3 (red). The right side shows the enriched biological process terms in Gene Set 1 based on GO enrichment analysis. (D) Violin plots showing the expression levels of representative IFN genes in each state. (E) Heatmap showing the binary regulon activity of transcriptional factors.

Full size|PPT slide

Fig.4 High expression of IFN-related genes in the ossified ligaments from patients with OPLL. (A) Histogram showing the expression levels of IFN-related genes in normal donors and patients with OPLL from scRNA-seq. (B) Primary cells of ligament tissues from patients with OPLL and control donors were cultured. Total RNA was extracted for RT-qPCR analysis. GAPDH was used as reference gene for RT-qPCR analysis. Data represent the means ± SD from three patients, *P < 0.05 (Mann–Whitney test). (C) Hematoxylin–eosin (HE) staining and fluorescence staining were performed in ligament tissues of patients with OPLL. The white dotted lines show the boundary between ossified tissues (right panel) and normal connective tissues (left panel).

Full size|PPT slide

3.3 Type I IFN signaling promotes the osteogenic differentiation of preosteoblasts

To ascertain the role of the type I IFN pathway in osteogenic differentiation in vitro, we profiled the transcriptome in MC3T3-E1 cells at 0, 14, and 21 days after osteogenic induction, with or without IFN-β treatment (Fig.5). Overall, DEGs between non- and IFN stimulation groups were analyzed (Fig.5), and these were considered to be involved in the cellular response to IFN-β, positive activation of innate immunity, defense against virus symbionts, negative viral genome replication, and viral processes (Fig.5). To reveal the gradual trend in DEGs from day 0 to day 21, we performed a time-series clustering pattern analysis of the DEGs, which revealed six distinct gene clusters (Fig.5). The expression levels of the IFN-related genes in Cluster 1 showed a rapid increase from day 0 to day 14 after IFN-β treatment and then remained constant (Fig.5 and 5E). Compared with that in the control group, the expression level of cell cycle-related genes in Cluster 4 displayed an obvious decrease, suggesting that IFN-β inhibited preosteoblast proliferation. The expression levels of genes in Cluster 6 increased steadily with IFN-β stimulation, which was involved in ossification and bone development (Fig.5 and 5E). Compared with that in the control group, the expression level of ossification-related genes of Cluster 6, such as SPP1, was much higher in the IFN-β-treated group (Fig.5). OPN, encoded by SPP1, is a major component of the bone matrix and is involved in bone remodeling. Thus, SPP1 might be involved in the regulation of IFN-β signaling during ossification.

Fig.5 IFN-β promotes the differentiation of murine MC3T3-E1 preosteoblasts. (A) For RNA sequencing, total RNA was extracted from IFN-β (50 mM)-treated MC3T3-E1 cells at 0, 14, and 21 days after osteogenic induction. The heatmap shows Spearman’s correlations of global mRNA expression across samples. Color scale: 0.8 (blue color) to 1 (red color). (B) Volcano plot showing the differentially expressed genes between IFN-stimulated and unstimulated groups; P < 0.05, log2(fold-change) ≥ 2. Statistical testing was performed via a two-sided Wilcoxon test. (C) Heatmap showing the expression of upregulated genes in the IFN-stimulated groups among samples. The color scheme represents the z-score distribution from −2 (blue) to 2 (red). DotPlot showing the enrichment results of upregulated genes. The number of enriched genes is presented based on the size of the dot, and the adjusted P value is presented based on the color. (D) Line graph showing six different gene clusters according to expression changes with osteogenic induction and IFN stimulation. (E) Bar plot showing the GO enrichment results of biological process terms for genes in Clusters 1, 4, and 6. (F) Heatmap showing the expression pattern of genes in Cluster 6. The color scheme represents the z-score distribution from −2 (blue) to 2 (red).

Full size|PPT slide

Subsequently, we treated murine MC3T3-E1 preosteoblasts with IFN-β and found that it promoted calcium deposition at 21 and 28 days after osteogenic induction (Fig.6 and 6B), confirming that the IFN signaling pathway plays a positive role in accelerating osteogenic differentiation. We then reduced the duration of IFN-β treatment to evaluate the effect of IFN-β signaling on the expression of the ossification-related gene SPP1 (which encodes osteopontin, OPN). At 72 h after IFN-β treatment, STAT1 and IRF7, known IFN-stimulated genes, were markedly activated. Concurrently, the protein and mRNA levels of OPN were upregulated (Fig.6 and 6D). To further validate the effect of the type I IFN pathway on SPP1, we used fludarabine, an inhibitor of STAT1, to treat MC3T3-E1 cells, with or without IFN-β treatment. The results showed that STAT1 inhibition blocked the SPP1 activation induced by IFN-β (Fig.6), indicating that the IFN-β signaling pathway activated SPP1 expression via STAT1.

Fig.6 IFN-β accelerates the osteogenic differentiation of murine MC3T3-E1 cells. (A, B) IFN-β (50 ng/mL) was added to the osteogenic differentiation medium to stimulate MC3T3-E1 cells. Alizarin red S staining was performed at different time points of osteogenic induction. The right panel shows the quantification of positive alizarin red-stained cells. Data represent the means ± SDs from three fields, *P < 0.05 (Kruskal–Wallis test). (C) IFN-β (100 ng/mL) was used to treat MC3T3-E1 cells, and total protein was extracted at different time points for Western blot analysis. (D) IFN-β (100 ng/mL) was used to treat MC3T3-E1 cells, and total RNA was extracted for RT-qPCR. Data represent the means ± SDs (Mann–Whitney test). (E) MC3T3-E1 cells were treated with a STAT1 inhibitor (fludarabine, 10 nM) with or without IFN-β (100 ng/mL) stimulation for 72 h, followed by RT-qPCR. Data represent the means ± SDs; *P < 0.05, **P < 0.01, ***P < 0.001 (Kruskal–Wallis test).

Full size|PPT slide

3.4 Anti-IFNAR1 neutralizing antibody inhibits osteogenic differentiation

The type I IFN pathway promotes ligament ossification; therefore, blocking this pathway should prevent ossification. Prior to treating murine MC3T3-E1 preosteoblasts with IFN-β, we used an anti-IFNAR1 neutralizing antibody to block the binding of IFN-β to IFNAR1 and found that this partially inhibited the activation of SPP1 stimulated by IFN-β (Fig.7). Alizarin red S staining was performed to evaluate the effect of the anti-IFNAR1 neutralizing antibody. After 28 days of osteogenic induction, we found that it clearly inhibited calcium deposition compared with that in the control group. Thus, blocking IFNAR1 using an anti-IFNAR1 neutralizing antibody significantly impeded the calcium deposition induced by IFN-β (Fig.7 and 7C).

Fig.7 Anti-IFNAR1 neutralizing antibody inhibits osteogenic differentiation. (A) MC3T3-E1 cells were treated with IFN-β (50 ng/mL) for 48 h, with or without pre-treatment with an anti-IFNAR1 neutralizing antibody (IFNAR1 Ab) (1 μg/mL, 3 h), followed by RT-qPCR. (B) MC3T3-E1 cells were pre-treated with an anti-IFNAR1 neutralizing antibody (1 μg/mL). After 3 h, IFN-β (50 ng/mL) was added to the osteogenic differentiation medium to stimulate MC3T3-E1 cells. At 28 days of osteogenic induction, alizarin red S staining was performed. Representative pictures are shown. (C) Positive alizarin red cell staining was quantified. Data represent the means ± SDs from three fields. (D) At 28 days of osteogenic induction, the media from the four groups in (A) were collected to detect OPN levels using an ELISA kit. Data represent the mean ± SD (A, C, and D); ***P < 0.001, ****P < 0.0001(Kruskal–Wallis test).

Full size|PPT slide

After osteogenic induction, cell culture media from IFN-β and/or IFNAR1 treatment groups were collected to detect the level of the secretory glycoprotein OPN via ELISA. We found that treatment with IFN-β led to a marked increase in its secretion, whereas blocking IFNAR1 decreased OPN protein levels in the medium (Fig.7). These results showed that blockade of the type I IFN pathway mediated by an anti-IFNAR1 neutralizing antibody could be effective against ossification.

4 Discussion

Although IFN signaling promotes bone formation, most studies focused on its role in suppressing osteoclastogenesis [22]. Only a few studies have clarified the role of IFN signaling in promoting osteoblastogenesis. Ma et al. reported that alendronate, commonly used for the treatment of osteoporosis, promotes osteoblast differentiation in rats with osteoporosis by upregulating the IFN-β/STAT1 signaling pathway [31]. That study provided evidence that alendronate administration can promote the regulatory effect of IFN-β signaling on osteoblastogenesis. However, the direct relationship between IFN-β signaling and osteoblast differentiation remains unclear. Our current study revealed that the IFN-β signaling pathway played a positive role in osteoblast differentiation. The communication between osteoblasts and osteoclasts is critical for bone homeostasis [32], and the disruption of this homeostasis can lead to ectopic ossification. Thus, IFN-β signaling might play a dual role in promoting osteoblastogenesis and suppressing osteoclastogenesis during the initiation and progression of ligament ossification.

Type I IFNs initiate signaling by binding to IFNAR1 and IFNAR2, and they activate JAK1 and TYK2, which subsequently induce the phosphorylation of STATs, leading to the activation of ISGs. The STAT1-STAT2-IRF9 tri-complex (ISGF3) binds to the IFN-stimulated response element (ISRE), which is involved in antiviral responses such as those associated with IRF7. The STAT1 homodimer binds to a distinct GAS sequence and participates in the inflammatory response [18]. In this current study, IRF7 was activated in ossified ligaments, so we speculated that IFNs produced during antiviral responses might promote ligament ossification. In this study, we found that STAT1 and SPP1 were activated in the ossified tissues. Inhibition of STAT1 blocked the activation of SPP1 induced by IFN-β. Thus, we believe that type I IFN signaling pathway promotes ossification by regulating the effect of STAT1 on SPP1. Consistently, the regulation of STAT1 on SPP1 has been reported in the endometrial tissue of patients with PCOS [33]. STAT1 could bind to the promoter of SPP1, and the binding significantly increased after IFN stimulation in the endometrial cells.

In addition to SPP1, IFN-b stimulation activated some ossification-related genes such as Cbfb, Vegfa, Nab1, Map3k7, Foxc1, and Ext1 (Fig.5). Cbfb regulates bone development by stabilizing Runx family proteins [34]. Vegfa and Map3k7 are essential for bone formation and bone homeostasis [35]. Foxc1 is required for normal chondrocyte differentiation and function [36]. EXT1 shows elevated expression levels in the proliferating chondrocytes of endochondral bones [37]. Given that ligament ossification is a multifactorial and multi-stage disease, we cannot rule out that other genes might also be involved in the regulation of the IFN signaling pathway on ossification.

BMP signaling stimulates MSC differentiation and initiates OPLL development. In the current study, we found that BMP signaling was active in Cluster 5 cells, which were distributed in states 1 and 3 near the branch point. Moreover, the IFN signaling pathway was active in Cluster 3 cells, which accounted for approximately 70% of state 3 cells and 20% of state 2 cells. Thus, BMP signaling activation may occur earlier than IFN signaling during ligament ossification.

Given the earlier onset and more diffuse progression of OPLL in the thoracic spine than that in the cervical spine, we chose patients with T-OPLL as the object of this study. Samples from the posterior longitudinal ligament were limited because of the operative procedure. The current analysis of scRNA-seq was based on six patients with T-OPLL and four patients with lumbar disc herniation. Although this is a small sample size, the key data were validated with further in vitro experiments using cell lines. To further strengthen our findings, we used the OPLL mouse model, such as tiptoe-walking mice, to confirm the role of IFN signaling in OPLL. In addition, it is difficult to digest and isolate single cells from the ossified area of the ligaments; therefore, we used cells at the edge of the ossified area to perform scRNA-seq. All cells used for scRNA-seq expressed the MSC markers, CD44 and CD90 (data not shown), indicating that cells from the control and OPLL groups had the characteristics of MSCs. MSCs accumulate in the collagenous matrix and around the blood vessels of the ossified ligament [38]. MSCs from the OPLL group displayed a more ossified signature than those from the non-ossified control group, suggesting that IFN-β signaling might directly promote the osteogenic differentiation of MSCs, which provides an opportunity for future study.

References

[1] Le HV, Wick JB, Van BW, Klineberg EO. Ossification of the posterior longitudinal ligament: pathophysiology, diagnosis, and management. J Am Acad Orthop Surg 2022; 30(17): 820–830

[2] Zhang Q, Zhou D, Wang H, Tan J. Heterotopic ossification of tendon and ligament. J Cell Mol Med 2020; 24(10): 5428–5437

[3] Koike Y, Takahata M, Nakajima M, Otomo N, Suetsugu H, Liu X, Endo T, Imagama S, Kobayashi K, Kaito T, Kato S, Kawaguchi Y, Kanayama M, Sakai H, Tsuji T, Miyamoto T, Inose H, Yoshii T, Kashii M, Nakashima H, Ando K, Taniguchi Y, Takeuchi K, Ito S, Tomizuka K, Hikino K, Iwasaki Y, Kamatani Y, Maeda S, Nakajima H, Mori K, Seichi A, Fujibayashi S, Kanchiku T, Watanabe K, Tanaka T, Kida K, Kobayashi S, Takahashi M, Yamada K, Takuwa H, Lu HF, Niida S, Ozaki K, Momozawa Y; Genetic Study Group of Investigation Committee on Ossification of the Spinal Ligaments; Yamazaki M, Okawa A, Matsumoto M, Iwasaki N, Terao C, Ikegawa S. Genetic insights into ossification of the posterior longitudinal ligament of the spine. eLife 2023; 12: e86514

[4] Ikegawa S. Genomic study of ossification of the posterior longitudinal ligament of the spine. Proc Jpn Acad, Ser B, Phys Biol Sci 2014; 90(10): 405–412

[5] Nakajima M, Takahashi A, Tsuji T, Karasugi T, Baba H, Uchida K, Kawabata S, Okawa A, Shindo S, Takeuchi K, Taniguchi Y, Maeda S, Kashii M, Seichi A, Nakajima H, Kawaguchi Y, Fujibayashi S, Takahata M, Tanaka T, Watanabe K, Kida K, Kanchiku T, Ito Z, Mori K, Kaito T, Kobayashi S, Yamada K, Takahashi M, Chiba K, Matsumoto M, Furukawa K, Kubo M, Toyama Y, Ikegawa S; Genetic Study Group of Investigation Committee on Ossification of the Spinal Ligaments; Ikegawa S. A genome-wide association study identifies susceptibility loci for ossification of the posterior longitudinal ligament of the spine. Nat Genet 2014; 46(9): 1012–1016

[6] Kim KH, Kuh SU, Park JY, Lee SJ, Park HS, Chin DK, Kim KS, Cho YE. Association between BMP-2 and COL6A1 gene polymorphisms with susceptibility to ossification of the posterior longitudinal ligament of the cervical spine in Korean patients and family members. Genet Mol Res 2014; 13(1): 2240–2247

[7] Yan L, Gao R, Liu Y, He B, Lv S, Hao D. The pathogenesis of ossification of the posterior longitudinal ligament. Aging Dis 2017; 8(5): 570–582

[8] Li H, Jiang LS, Dai LY. Hormones and growth factors in the pathogenesis of spinal ligament ossification. Eur Spine J 2007; 16(8): 1075–1084

[9] Ling Z, Crane J, Hu H, Chen Y, Wan M, Ni S, Demehri S, Mohajer B, Peng X, Zou X, Cao X. Parathyroid hormone treatment partially reverses endplate remodeling and attenuates low back pain in animal models of spine degeneration. Sci Transl Med 2023; 15(722): eadg8982

[10] Feng H, Xing W, Han Y, Sun J, Kong M, Gao B, Yang Y, Yin Z, Chen X, Zhao Y, Bi Q, Zou W. Tendon-derived cathepsin K-expressing progenitor cells activate Hedgehog signaling to drive heterotopic ossification. J Clin Invest 2020; 130(12): 6354–6365

[11] Liu N, Zhang Z, Li L, Shen X, Sun B, Wang R, Zhong H, Shi Q, Wei L, Zhang Y, Wang Y, Xu C, Liu Y, Yuan W. MicroRNA-181 regulates the development of ossification of posterior longitudinal ligament via epigenetic modulation by targeting PBX1. Theranostics 2020; 10(17): 7492–7509

[12] Xu C, Zhang Z, Liu N, Li L, Zhong H, Wang R, Shi Q, Zhang Z, Wei L, Hu B, Zhang H, Shen X, Wang Y, Liu Y, Yuan W. Small extracellular vesicle-mediated miR-320e transmission promotes osteogenesis in OPLL by targeting TAK1. Nat Commun 2022; 13(1): 2467

[13] Kim KT, Kim DH, Chung JY, Lee S, Joo J, Nah SS, Song HY, Kim HJ. Association of interferon gamma polymorphism with ossification of the posterior longitudinal ligament in the Korean population. Immunol Invest 2012; 41(8): 876–887

[14] Pestka S, Krause CD, Walter MR. Interferons, interferon-like cytokines, and their receptors. Immunol Rev 2004; 202(1): 8–32

[15] Lee AJ, Ashkar AA. The dual nature of type I and type II interferons. Front Immunol 2018; 9: 2061

[16] Wathelet MG, Lin CH, Parekh BS, Ronco LV, Howley PM, Maniatis T. Virus infection induces the assembly of coordinately activated transcription factors on the IFN-β enhancer in vivo. Mol Cell 1998; 1(4): 507–518

[17] Hu X, Li J, Fu M, Zhao X, Wang W. The JAK/STAT signaling pathway: from bench to clinic. Signal Transduct Target Ther 2021; 6(1): 402

[18] Ivashkiv LB, Donlin LT. Regulation of type I interferon responses. Nat Rev Immunol 2014; 14(1): 36–49

[19] Chen K, Liu J, Cao X. Regulation of type I interferon signaling in immunity and inflammation: a comprehensive review. J Autoimmun 2017; 83: 1–11

[20] Crow YJ, Stetson DB. The type I interferonopathies: 10 years on. Nat Rev Immunol 2022; 22(8): 471–483

[21] Snell LM, McGaha TL, Brooks DG. Type I interferon in chronic virus infection and cancer. Trends Immunol 2017; 38(8): 542–557

[22] Seeliger C, Schyschka L, Kronbach Z, Wottge A, van Griensven M, Wildemann B, Vester H. Signaling pathway STAT1 is strongly activated by IFN-β in the pathogenesis of osteoporosis. Eur J Med Res 2015; 20(1): 1

[23] Takaoka A, Taniguchi T. New aspects of IFN-α/β signalling in immunity, oncogenesis and bone metabolism. Cancer Sci 2003; 94(5): 405–411

[24] Duque G, Huang DC, Macoritto M, Rivas D, Yang XF, Ste-Marie LG, Kremer R. Autocrine regulation of interferon gamma in mesenchymal stem cells plays a role in early osteoblastogenesis. Stem Cells 2009; 27(3): 550–558

[25] Rifas L. T-cell cytokine induction of BMP-2 regulates human mesenchymal stromal cell differentiation and mineralization. J Cell Biochem 2006; 98(4): 706–714

[26] Duque G, Huang DC, Dion N, Macoritto M, Rivas D, Li W, Yang XF, Li J, Lian J, Marino FT, Barralet J, Lascau V, Deschenes C, Ste-Marie LG, Kremer R. Interferon-gamma plays a role in bone formation in vivo and rescues osteoporosis in ovariectomized mice. J Bone Miner Res 2011; 26(7): 1472–1483

[27] Ji Q, Zheng Y, Zhang G, Hu Y, Fan X, Hou Y, Wen L, Li L, Xu Y, Wang Y, Tang F. Single-cell RNA-seq analysis reveals the progression of human osteoarthritis. Ann Rheum Dis 2019; 78(1): 100–110

[28] Butler A, Hoffman P, Smibert P, Papalexi E, Satija R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat Biotechnol 2018; 36(5): 411–420

[29] Kumar L, Futschik ME. Mfuzz: a software package for soft clustering of microarray data. Bioinformation 2007; 2(1): 5–7

[30] Cai Z, Liu W, Chen K, Wang P, Xie Z, Li J, Li M, Cen S, Ye G, Li Z, Su Z, Ma M, Wu Y, Shen H. Aberrantly expressed lncRNAs and mRNAs of osteogenically differentiated mesenchymal stem cells in ossification of the posterior longitudinal ligament. Front Genet 2020; 11: 896

[31] Ma X, Xu Z, Ding S, Yi G, Wang Q. Alendronate promotes osteoblast differentiation and bone formation in ovariectomy-induced osteoporosis through interferon-β/signal transducer and activator of transcription 1 pathway. Exp Ther Med 2018; 15(1): 182–190

[32] Kim JM, Lin C, Stavre Z, Greenblatt MB, Shim JH. Osteoblast-osteoclast communication and bone homeostasis. Cells 2020; 9(9): 2073

[33] Paravati R, De Mello N, Onyido EK, Francis LW, Brusehafer K, Younas K, Spencer-Harty S, Conlan RS, Gonzalez D, Margarit L. Differential regulation of osteopontin and CD44 correlates with infertility status in PCOS patients. J Mol Med (Berl) 2020; 98(12): 1713–1725

[34] Qin X, Jiang Q, Matsuo Y, Kawane T, Komori H, Moriishi T, Taniuchi I, Ito K, Kawai Y, Rokutanda S, Izumi S, Komori T. Cbfb regulates bone development by stabilizing Runx family proteins. J Bone Miner Res 2015; 30(4): 706–714

[35] Hwang C, Marini S, Huber AK, Stepien DM, Sorkin M, Loder S, Pagani CA, Li J, Visser ND, Vasquez K, Garada MA, Li S, Xu J, Hsu CY, Yu PB, James AW, Mishina Y, Agarwal S, Li J, Levi B. Mesenchymal VEGFA induces aberrant differentiation in heterotopic ossification. Bone Res 2019; 7(1): 36

[36] Almubarak A, Lavy R, Srnic N, Hu Y, Maripuri DP, Kume T, Berry FB. Loss of Foxc1 and Foxc2 function in chondroprogenitor cells disrupts endochondral ossification. J Biol Chem 2021; 297(3): 101020

[37] Kitsoulis P, Galani V, Stefanaki K, Paraskevas G, Karatzias G, Agnantis NJ, Bai M. Osteochondromas: review of the clinical, radiological and pathological features. In Vivo 2008; 22(5): 633–646

[38] Chin S, Furukawa K, Ono A, Asari T, Harada Y, Wada K, Tanaka T, Inaba W, Mizukami H, Motomura S, Yagihashi S, Ishibashi Y. Immunohistochemical localization of mesenchymal stem cells in ossified human spinal ligaments. Biochem Biophys Res Commun 2013; 436(4): 698–704

Acknowledgements

This work was supported by grants from the Ministry of Science and Technology of the People’s Republic of China (No. 2022YFA1104003 to Hongquan Zhang and No. 2021YFC2501003 to Hongquan Zhang); and the National Natural Science Foundation of China (No. 82073076 to Yu Yu, No. 82372992 to Yu Yu, No. 82230094 to Hongquan Zhang and No. 81972616 to Hongquan Zhang, and No. 82372451 to Xiaoguang Liu).

Compliance with ethics guidelines

Conflicts of interest Xiao Liu, Lei Zhang, Ge Wang, Wei Zhao, Chen Liang, Youzhi Tang, Yenan Fu, Bo Liu, Jing Zhang, Xiaoguang Liu, Hongquan Zhang, and Yu Yu declare that they have no conflict of interest.

The study was approved by the PUTH Institutional Review Board (IRB00006761-M2019410), and the study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Informed consent was obtained from all patients for being included in the study.

RIGHTS & PERMISSIONS

2024 Higher Education Press