Introduction

The functional spinal units (FSUs) are the building blocks of the spine, representing segmental structures with similar properties of the entire spine1,2. Each FSU consists of two adjacent vertebrae and the connected articulation3. The intervertebral disc, which is positioned between two vertebral bodies, is an avascular cartilaginous structure composed of three components—endplates, annulus fibrosus (AF) and nucleus pulposus (NP)4. Endplates are concave structures that are 1 mm thick in humans and have two distinct layers: the cartilage endplate (CEP) made of hyaline cartilage and the bony endplate (BEP) consisting of semi-porous cancellous bone5,6. During spinal growth until skeletal maturity, the endplates have microscopic blood vessels penetrating from the vertebrae adjacent to the BEP, providing a vascular supply to the intervertebral disc5. Cartilage endplates are vital in the maintenance of disc homeostasis. Endplates are critical regulatory organs for the flow of nutrients into and out of the intervertebral disc and possess a range of physiological properties that are crucial to spinal function5,7,8.

Low back pain (LBP) is one of the most prevalent musculoskeletal system illnesses. It continues to be one of the leading causes of medical expenses and financial strain despite advances in surgery and medication. Intervertebral disc degeneration (IVDD) is thought to be the main pathogenic cause of LBP. The endplate degeneration is a significant factor leading to IVDD as an early morphological change9. Modic changes (MC) refer to endplate changes that can be identified using magnetic resonance imaging (MRI)10. Studies have shown increased outgrowth of vascular vessels and nerve fibers to endplates in degenerative spine, which is considered a possible cause of LBP5,8,11,12. Additionally, recent research has highlighted the link between chondrocytes and immunity, suggesting that immune responses may adversely affect chondrocytes either directly or indirectly13. Further understanding of the cellular composition of the endplate and its connection with cellular immunity during degeneration is necessary.

Single-cell sequencing technology has revolutionized the multidimensional analysis of various tissue cells and diseases. In particular, scRNA-seq is becoming increasingly popular in the study of tissue and cellular heterogeneity14. With the continuous improvement and innovation of high-throughput sequencing technology, scRNA-seq is being applied to investigate the molecular regulatory mechanisms involved in physiological development, pathological processes, inflammation, and immunity across a range of tissues and diseases. ScRNA-seq has been used to detect and analyze the intervertebral disc NP tissue and meniscus tissue, which has expanded our understanding of osteoporosis, IVDD and osteoarthritis (OA)15,16,17,18,19,20. For nucleus pulposus tissue, three empirically defined populations were generated: cartilage progenitor cells (CPCs), fibrochondrocyte progenitors (FCPs), and homeostatic chondrocytes (HomCs)17. In addition, unbiased chondrocyte atlases of articular cartilage have been proposed, including proliferative chondrocytes, prehypertrophic chondrocytes, hypertrophic chondrocytes, fibrocartilage chondrocytes, effector chondrocytes, regulatory chondrocytes and homeostatic chondrocytes20. These findings have significant implications for developing novel therapeutic strategies to target specific cell types or signaling pathways involved in these conditions. However, the scRNA-seq study specifically focused on endplate degeneration is lack.

In this study, we performed a cluster analysis of endplate cells during IVDD and identified nine different cell types including chondroblasts (CB), regulatory chondrocytes (RegC), homeostatic chondrocytes (HomC), prehypertrophic chondrocytes (pre-HTC), fibrocartilage chondrocytes (FC), proliferative chondrocytes (ProC), hypertrophic chondrocytes (HTC), T cells/natural killer cells (T/NK) and macrophages (MA). Macrophages and T/NK cells are both derived from hematopoietic stem cells. They collaborate in immunity, with macrophages focusing on early defense and antigen presentation, and T/NK cells managing targeted responses and cell-mediated cytotoxicity. Our findings provide a comprehensive collection of cell clusters from degenerated endplates and demonstrate the involvement of immune cells in the process of endplate degeneration. This work sheds light on the cellular composition, biological process, transcriptional activity and cell-to-cell communication of degenerative endplates.

Materials and methods

Cartilage endplate sample

Cartilage endplate tissue was obtained from two patients diagnosed with modic changes at the levels of L3-L4 or L4-L5, who were undergoing surgery for disc herniation. The patients were a 58-year-old female and a 66-year-old male, respectively. All participants included were confirmed with MRI images before surgery. This study was approved by the Life Science Ethics Review Committee of Zhengzhou University. All experiments were performed in accordance with relevant guidelines and regulations. And all the patients signed informed consent forms before participation. The cartilage endplate tissue of the intervertebral discs resected during the surgery was transported immediately to the laboratory under 4 °C cold storage for single-cell suspension preparation.

Preparation of single cell suspensions of vertebral endplate samples

The vertebral endplate tissues were washed twice with sterile PBS buffer to remove any blood and prevent contamination, left to stand, and the supernatants were aspirated. The tissues were cut quickly into small 1–2 mm pieces with pre-cooled ophthalmic scissors and placed on ice. And added pre-warmed digestive enzymes, including 1 mg/ml collagenase type I (Gibco; #17100-017), 2 mg/ml dispase II (Sigma-Aldrich; #D4693-1G) and 1 unit/ml DNase I (NEB; #M0303S), mixed gently with a pasteurizing pipette and digested in a 37 °C water bath for 20–25 min, mixed gently with a pasteurizing pipette every 5 min. At the end of digestion, added an equal volume of complete medium containing FBS to terminate the digestion. The isolated cells were filtered using 70-μm disposable filter and washed twice with PBS (Gibco; # 13151014), centrifuged for 5 min at 500×g at 4 °C to collect the cell precipitate. To remove red blood cells, incubated with 1X RBC lysis buffer (Biogems; 64010-00-100) at room temperature for 5-10 min, centrifuged for 5 min at 500×g, collected cell precipitate. Added PBS (0.04% BSA, Sigma; 9048–46-8) and washed 1–2 times, centrifuged for 5 min, discarded the supernatant and resuspend in PBS. For scRNA-seq, the cell density was subsequently changed to 106/ml.

10 × Genomics vertebral endplate cell single cell marker and library construction and sequencing

Single cell suspensions of vertebral endplate tissue were prepared and tested (cell viability > 80%). They were subjected to single cell library construction according to the Single Cell 3' Reagent Kits manual (10 × Genomics, USA). Using a microfluidic chip, cells and gel beads with cell labeling sequences were wrapped in a droplet, and the cell-wrapped droplet was collected. In the droplet, the cells were lysed so that the mRNA in the cells was connected to the cell Barcode of the bead, forming Single Cell GEMs. These were subjected to RT reactions in the droplet, followed by the disruption of emulsions and cDNA library construction. Cells and reaction reagents were placed in one lane on the microfluidic chip and gel beads in the other lane to form a GEM. Reverse transcription was performed independently in each GEM. Afterward, the labeled cDNA was mixed, then amplified, and the library was constructed.

Single-cell high-throughput sequencing

Library size was measured by a bioanalyzer (Agilent Technologies, United States) and the Qubit™ dsDNA HS Assay Kit (Thermo Fisher) was used to quantify the library products. The NovaSeq 6000 system (Illumina) was used for sequencing with the PE150 strategy.

Processing of single-cell RNA-seq raw data

Sequencing reads were aligned to the human genome reference sequences (GRCh38), and gene-level unique molecular identifier (UMI) count matrices were obtained by using Cell Ranger 3.1.0 (10X Genomics).

Clustering analysis of scRNA-seq data

Generated read counts matrices were analyzed in R (v3.5.2, https://www.R-project.org) via Seurat (v3.1.1)21 following its filtering, variable gene selection, normalization, scaling, dimensionality reduction, clustering and visualization. Cells that with gene counts greater than 200 per cell, UMI counts within outlier thresholds (Q1-1.5IQR < UMI < Q3 + 1.5IQR), and a percentage of mitochondrial genes < 20% were retained for downstream analysis. Multiple samples were integrated by CCA-based anchors method21,22. Following principal component analysis (PCA), the first 30 principal components were selected for clustering the cells. A resolution of 0.5 was used with t-SNE visualization. Differentially expressed genes (DEGs) of each cluster were identified by Seurat’s implementation function FindAllMarkers with Wilcoxon rank-sum test. According the expression of the known markers and specifically expressed genes we assigned 9 known cell types.

Sub-clustering of chondrocyte cell types

The cells of several chondrocyte cell types were performed an unsupervised sub-clustering by Seurat with the first 30 principal components. Using recently reported marker genes, we identified 9 sub cell types, including CB, RegC, HomC, pre-HTC, FC, ProC, HTC, T/NK, and MA.

Pseudotime analysis

Monocle 2 (version 2.10.1)23 was used to estimate a pseudotemporal path of chondrocyte cell differentiation. Monocle is an unsupervised algorithm that increases the temporal resolution of transcriptome dynamics using single-cell RNA-Seq data collected at multiple time points. Monocle can in principle be used to recover single-cell gene expression kinetics from a wide array of cellular processes, including differentiation, proliferation and oncogenic transformation. Cells were ordered in pseudotime along a trajectory using reduceDimension with the DDRTree method and orderCells functions.

Ligand–receptor interactions

To assess cellular crosstalk between different cell types, we used UMI count data and the Python package CellPhoneDB v2.1.224 with the human database v2.0.0 to infer cell–cell communication networks from single cell transcriptome data. Only interactions pass the p values threshold (p < 0.05) were considered.

Motif enrichment and regulatory network

The gene regulatory network of stem cells was built by SCENIC (version 1.1.2)25. With the default parameters, SCENIC generated the enrichment scores of 160 motifs. We used Wilcoxon rank-sum test to calculate the differential enriched motifs when compared between case patients and controls.

Gene functional analysis

Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) terms were applied by clusterProfiler (v3.10.1)26 with the database org.Hs.eg.db (v3.7.0).

Results

Identification of different cell types in endplate during degeneration

The endplate, indicated as Modic Changes in the IVDD by MRI, was isolated and digested from two patients with lumbar spine surgery. After single-cell transcriptome sequencing and quality control, a total of 8534 cells were obtained (Fig. 1A).

Fig. 1
figure 1

Identification of unique clusters of endplate cells during degeneration. (A) Schematic diagram of sample source and experimental workflow. (B) UMAP of the 8,534 cells profiled nine main cell types. (C) Percentage of each cell type in the two samples. (D) t-SNE plots showing the expression of marker genes for nine cell subclusters. (E) Heatmap of top 10 DEGs for nine clusters; each column represents a cell cluster, and each row represents a DEG for a cluster. CB, chondroblasts; RegC, regulatory chondrocytes; HomC, homeostatic chondrocytes; pre-HTC, prehypertrophic chondrocytes; FC, fibrocartilage chondrocytes; ProC, proliferative chondrocytes; HTC, hypertrophic chondrocytes; MA, macrophages; T/NK, T cells/natural killer cells.

Full size image

Based on the single-cell transcriptomics data, previous literature and databases, nine distinct cell clusters were identified after unbiased analysis (Fig. 1B–D and Supplementary Fig. S1): 1) Cluster 1 (average 36.69% of all cells): CB is enriched with marker gene chondromodulin (CNMD). At the same time, the genes lumican (LUM) and decorin (DCN) representing chondrogenicity are also expressed in this cluster27; 2) Cluster 2 (average 23.34% of all cells): RegC is characteristically highly expressed marker genes complement factor H (CFH) and frizzled related protein (FRZB)20; 3) Cluster 3 (average 14.89% of all cells): HomC is specifically highly expressed marker genes ras related dexamethasone induced 1 (RASD1)20; 4) Cluster 4 (average 14.89% of all cells): pre-HTC is enriched with marker gene TNF alpha induced protein 6 (TNFAIP6)20; 5) Cluster 5 (average 7.34% of all cells): FC is particularly intensely expressed marker gene collagen type I alpha 1 chain (COL1A1)20; 6) Cluster 6 (average 1.34% of all cells): ProC is powerful expressed marker gene topoisomerase II alpha (TOP2A)28; 7) Cluster 7 (average 0.45% of all cells): HTC is enriched with marker gene collagen type X alpha 1 chain (COL10A1)20; 8) Cluster 8 (average 0.84% of all cells): T/NK) is specifically highly expressed marker gene granulysin (GNLY)29; and 9) Cluster 9 (average 0.22% of all cells): MA is most expressed marker gene lysozyme (LYZ)30. Further, the differential expression analysis showed the top 10 differentially expressed genes for each cluster (Fig. 1E). Specially, CB was newly defined in the present study, suggesting further expansion of the impression of chondrocytes.

Analyzes of biological functions across seven chondrocyte clusters

To further understand the heterogeneity and biological function of seven different chondrocyte clusters, Gene ontology (GO) analysis (Fig. 2A), cell-cycle analysis (Fig. 2B), and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways enrichment analysis (Fig. 3A–F) were performed. (1) For CB, the analysis showed that its function was mainly related to cartilage development, showing a tendency towards chondrocyte proliferation and differentiation. (2) We also found that the function of FC was mainly associated with extracellular matrix (ECM) organization and ECM-receptor interaction via collagen fibers and focal adhesion, indicating the involvement of FC in degeneration. (3) As HomC, GO analysis suggested that it was mainly involved in cellular metabolism and homeostasis. In addition, it was associated with MAPK signaling pathway, by KEGG analysis. (4) It was indicated that HTC was related to the regulation of ECM, such as ECM disassembly and peptidase regulator activity. (5) pre-HTC was shown to involve in immune responses. KEGG pathways enrichment analysis demonstrated that pre-HTC was also enriched in ECM regulation. (6) In regard to ProC, it controlled cell proliferation through functions involved in the cell division process. Meanwhile, most of the cells in this cluster were in the G2/M phase of cell division, suggesting that ProC was in an active state of cell proliferation (Fig. 2B). (7) GO analysis suggested that RegC was related to protein localization and the response to reactive oxygen species and metal ion. Disturbances in transition metal ion homeostasis can lead to increased oxidative stress and reactive oxygen species production.

Fig. 2
figure 2

Gene ontology (GO) analysis of chondrocyte subsets. (A) Bar plots show the upregulated terms of seven chondrocyte clusters respectively by GO analysis. (B) The proportion of cells annotated as the different cell cycle phases in different cell clusters. NC, other chondrocytes; CB, chondroblasts; RegC, regulatory chondrocytes; HomC, homeostatic chondrocytes; pre-HTC, prehypertrophic chondrocytes; FC, fibrocartilage chondrocytes; ProC, proliferative chondrocytes; HTC, hypertrophic chondrocytes.

Full size image
Fig. 3
figure 3

Bar plots show the upregulated pathways of chondrocyte clusters respectively by KEGG analysis. (A) CB, chondroblasts. (B) FC, fibrocartilage chondrocytes. (C) HomC, homeostatic chondrocytes. (D) pre-HTC, prehypertrophic chondrocytes. (E) ProC, proliferative chondrocytes. (F) RegC, regulatory chondrocytes.

Full size image

Transcriptional regulatory activities in different chondrocyte subsets during endplate degeneration

The TFs could control different chondrocyte phenotypes. To further identify the transcriptional networks in different subsets, the single-cell regulatory network inference and clustering (SCENIC) method were conducted. Based on the distribution of activity across cells, the activity of each regulon in each cell was quantified, and it was then binarized as “on” or “off”. The SCENIC results indicated strong enrichment of EGR1 and JUND regulon activity in all chondrocytes; CEBPB activity in RegC; and ATF3 and BHLHE40 activity in HTC (Fig. 4A). Enrichment analysis of chondrocytes based on chondrogenic function-related TFs revealed that the highly different and specific TF networks (Fig. 4B). In detail, HomC was enriched for EGR3, RORA, MAFB and IRF1. These TFs were involved in the regulation of rhythmic genes (ERR3h and RORA)31,32 and immune response (MAFB and IRF1)33. This is consistent with the GO analysis, implying the participation of HomC in the regulation of homeostasis. In addition, the higher regulatory activity of FOXO1 and FOSL1 in HomC and FC suggested that they have the ability to regulate metabolic homeostasis (FOXO1)34,35,36 as well as cell proliferation, differentiation and transformation (FOSL1)31. The REL, one of the NF-kappaB family, exhibited increased activity in HTC, indicating the inflammatory features of HTC37. RAD21 and MYBL1 were ProC-specific regulators that were crucial for cell cycle and DNA damage repair38,39. MYC as well as CEBPG were upregulated in RegC. MYC manages the cell cycle progression, apoptosis and cellular transformation37. Additionally, EGR1 regulon regulating activities in CB were significantly increased, which is required for differentiation and mitosis. Based on the functional correlation of transcription factors and predicted target genes, we selected EGR1, FOXO1 and REL for analysis and found that EGR1-DCN, FOXO1-collagen type III alpha 1 chain (COL3A1) and REL- interferon regulatory factor 1 (IRF1) were highly related transcription factor-target gene pairs. TF and the infer target genes are co-expression show the t-SNE and related with pseudotime (Fig. 4C,D). Further suggesting that they may play important regulatory roles in CB (EGR1), FC (FOXO1) and HTC (REL), respectively. Therefore, the investigation of potential regulatory TFs will help us understand chondrocyte heterogeneity and their possible involvement in physiological processes.

Fig. 4
figure 4

Transcriptional regulation of seven chondrocyte subclusters. (A) Heatmap of regulons specific to different cell types, as inferred by the SCENIC. Regulon activity was binarized to “on” (black) or “off” (white). (B) Heatmap indicating the activity of chondrogenic function-related TFs analyzed with SCENIC in seven chondrocyte subclusters. (C) tSNE depiction of regulon activity, TF gene expression, and predicted target gene expression for EGR1 in CB, FOXO1 in FC, and REL in HTC. (D) The trend of gene expression changes with pseudotime. CB, chondroblasts; RegC, regulatory chondrocytes; HomC, homeostatic chondrocytes; pre-HTC, prehypertrophic chondrocytes; FC, fibrocartilage chondrocytes; ProC, proliferative chondrocytes; HTC, hypertrophic chondrocytes; MA, macrophages; T/NK, T cells/natural killer cells.

Full size image

The trajectory branch of all chondrocyte subsets

To investigate the conversion of different chondrocyte clusters and the corresponding gene expression in endplate degeneration, we used the pseudotime trajectories to undertake a trajectory analysis of seven chondrocyte subtypes with two endpoints representing two different cell fates (Fig. 5A). By examining the dispersion patterns of several chondrocyte subtypes in the pseudotime trajectory we found: ProC was mainly located close to the root. CB, pre-HTC and FC were dispersed along the route, whereas RegC, HomC and HTC generally occupied the extremities of the two fates (Fig. 5A,B). Interestingly, ProC, pre-HTC and HTC compromised a complete evolutionary route. And ProC and HTC formed independent head ends and tail ends of the trajectory, respectively (Fig. 5C). This result confirms that these three clusters have a closer association with each other than the rest.

Fig. 5
figure 5

Trajectory branch of chondrocyte subsets. (A) Monocle pseudotime trajectory contains three branches showing the dynamic progression of chondrocyte clusters. (B) Projection of individual cell types onto the trajectory of (A). (C) ProC, pre-HTC and HTC compromise a complete evolutionary route without branches. (D) Heatmap showing the standardized kinetic curves of the genes in trajectory from root to fate1 or fate2. (E) Pseudotime kinetics of specific representative genes from the root of the trajectory to fate1 (solid line) and fate2 (dashed line). (F) Bar plots showing the top annotated GO terms in five gene sets, which are hierarchically clustered from (D). CB, chondroblasts; RegC, regulatory chondrocytes; HomC, homeostatic chondrocytes; pre-HTC, prehypertrophic chondrocytes; FC, fibrocartilage chondrocytes; ProC, proliferative chondrocytes; HTC, hypertrophic chondrocytes.

Full size image

In order to determine the effect of the differential genes in cell fates, we then deconstructed the gene transformation patterns based on the findings of the trajectory analysis (Fig. 5D). Five different chondrocyte clusters were obtained based on the different gene expression. In the transition to fate1, Cluster 3-enriched genes, included CFH, chitinase 3 like 1 (CHI3L1), FRZB and ceruloplasmin (CP), that were related to regulation of complement activation, cytokine action, and immune-inflammatory response were significantly increased. However, cluster 5-enriched genes, included COL10A1 and integrin binding sialoprotein (IBSP), did not show significant enhancement during the transition to fate 1 or fate 2. COL1A1, AE binding protein 1 (AEBP1) and cartilage oligomeric matrix protein (COMP) were genes enriched in clusters 1 and 4 that were expressed in the direction of fate 2. And they were related to the ECM structure. Notably, Cluster 2-enriched genes, included Jun proto-oncogene (JUN) and FUS RNA binding protein (FUS), which were associated with cellular homeostasis and cellular metabolism, were significantly elevated during the transition to fate2 (Fig. 5D–F). Taken together, the two separate paths of differentiation provided functional diversity in chondrocytes of IVDD, one relating to cellular immunity and the other affecting the ECM. There were high correlations between the two cell fates, as well as the process of inflammatory reactions and ECM modifications. This demonstrated that the treatment of IVDD may benefit from research on the direction of chondrocytes differentiation, especially the fate 1.

Complex intercellular interaction network in degenerative endplate cells

CellPhoneDB analysis was performed to assess intercellular communication in all subpopulations, showing the higher interactions between pre-HTC and FC or CB or ProC (Fig. 6A). Then, we conducted the analysis of ligand-receptor pairs between different clusters (Fig. 6B and Supplementary Fig. S2). The results demonstrated that pre-HTC was most closely related to ProC and FC by ligand-receptor pairs (Fig. 6B). Furthermore, considering the important regulative features of macrophages and T/NK cells in local micro environment, we investigated the ligand-receptor pairs between them and other chondrocyte subsets (Fig. 6C,D). Among these chondrocyte subpopulations, the strength of MA interactions with them varied according to cell types, being the stronger with CB, pre-HTC, FC, and ProC. Notably, we found that the ligand-receptor pairs of FN1(Fibronectin)-a4b7 complex and FN1-a4b1 complex were specifically involved in the interactions of MA and all chondrocyte subsets, but not in the interactions of T/NK cells and chondrocytes (Fig. 6C). These two ligand-receptor pairs mainly participated in the cell adhesion process, indicating the important regulative function of MA. Whereas, the ligand-receptor pairs of CD74-APP, CD74-COPA, and CD74-MF were engaged in the interactions of T/NK cells and chondrocyte subsets, indicating T/NK cells may be widely involved in the regulation of endplate cell degeneration through CD74 (Fig. 6D).

Fig. 6
figure 6

Complex intercellular interaction network in degenerative endplate cells. (A) Interaction network constructed by CellPhoneDB analysis. Thicker lines indicate more interaction with other types of cells. (B) Heatmap showing the number of potential ligand-receptor pairs among the predicted cell types. (C) The detected ligand-receptor pairs of MAs and all chondrocyte subsets by CellPhoneDB. (D) The detected ligand-receptor pairs of T/NK cells and all chondrocyte subsets by CellPhoneDB. CB, chondroblasts; RegC, regulatory chondrocytes; HomC, homeostatic chondrocytes; pre-HTC, prehypertrophic chondrocytes; FC, fibrocartilage chondrocytes; ProC, proliferative chondrocytes; HTC, hypertrophic chondrocytes; MA, macrophages; T/NK, T cells/natural killer cells.

Full size image

Discussion

Through scRNA-seq, we conducted a detailed analysis of endplate cells during IVDD, providing a comprehensive overview of their cellular composition and gene expression patterns. Our findings confirm previous chondrocyte research, identify key cell subsets and TFs, offer new insights into cell fates and differentiation, and demonstrate chondrocyte-macrophage communication.

We defined CB, linked to chondrogenesis and chondrocyte differentiation., and identified new genes CFH and FRZB, differing from the previous RegC definition by Quanbo Ji et al20.The role of transition metal ions in oxidative stress and inflammation is well-established40. Previously, TOP2A and CENPF were characteristic of CPC17. Based on differential genes, GO analysis, and cell trajectory analysis, we classify ProC and CPC as the same cluster.

Excess iron ions generate oxygen species (ROS) and oxidative stress, damaging cellular components and causing chronic inflammation. We found CP enrichment in RegC, involved in metal ion transport and redox processes, suggesting RegC's role in maintaining metal ion homeostasis and protecting against oxidative stress. Future studies can explore chondrocyte interactions with metal ions and oxidative stress in endplate degeneration.

SCENIC results provide insights into transcriptional activities in chondrocyte subtype. There are a few pathways for HomC to involve homeostasis and metabolism. According to previous findings, Fosl1 regulates MMPs expression41, IRF1 upregulates inflammatory factors42, and SOX4 influences cell fate in HomC43. In RegC, MYC promotes growth-associated genes and VEGFA production44,45, while CEBPG binds MYC to activate the PI2K-AKT pathway46, may link RegC to angiogenesis in IVDD. EGR1, a specific modulator of CB, plays a pivotal role in cellular processes such as survival, proliferation, and apoptosis47. DCN, one of the chondrogenic genes, is correlated with EGR1 as its predicted target gene. We speculate that CB may be involved in chondrogenesis by regulating DCN through EGR1.

Pseudo-temporal trajectory analysis shows endplate chondrocytes diverging into two fates in IVDD, one links to inflammatory and the other to homeostasis and ECM regulation. Notably, consistent with previous results, pre-HTC represents an intermediate state between ProC and HTC, indicating the continuous dynamic evolution among these chondrocyte clusters20.

CellPhoneDB analysis reveals MA’s closer relation to chondrocytes than T/NK cells, with chondrocytes-macrophages interaction relying on FN1-a4b7 complex and FN1-a4b1 complex. FN1, a major non-collagenous ECM glycoprotein, regulates cell adhesion, differentiation and growth, and is linked to the pro-fibrotic, anti-inflammatory phenotype of M2-like macrophages48,49. This suggests a unique macrophages-chondrocytes relationship may be related to cellular immunity. Interestingly, macrophages promote inflammation of NP via IL-6 and TNF in IVDD17. Here, our study provides valuable insights into the potential role of FN1 in mediating chondrocytes-macrophages crosstalk in endplate degeneration. Yamawaki et al. showed chondrocytes-macrophages contact during the early phase of cartilage regeneration50. There may be more possibilities between chondrocytes and macrophages that must be further investigated and discussed.

HomC’s lack of close associated with macrophages in cell-to-cell communication analysis is noteworthy. Relevant ligand-receptor pairs include TNFRSF1B-GRN, TNFRSF1A-GRN, HLA-C-FAM3C, TGFB1-TGFBR3 and LGALS9-CD47. TGFB1-TGFBR3 is involved in angiogenesis-related VEGF signaling pathways, with IVDD potentially triggered by reduced nutrient diffusion in vertebral endplate vasculature. HomC may also impact IVDD progression.

Single-cell sequencing of the endplate has revealed various cell types, notably immune cells. Pseudo-temporal trajectory analysis indicates chondrocytes significantly influence IVDD by maintaining homeostasis and regulating ECM, thereby altering the endplate microenvironment. The ECM ensures the intervertebral disc's mechanical integrity51, while the endplate facilitates nutrient exchange52. Changes in chondrocyte types, which modulate ECM and nutrient exchange, may drive disc degeneration. Additionally, chondrocytes are linked to immune function. Recent studies have focused on immune cell-mediated endplate changes in disc research. Our previous research demonstrated that senescent macrophages increase pSTAT3 in endothelial cells via IL10 secretion, leading to angiogenesis and sclerosis in the endplates53. CellPhoneDB analysis shows chondrocytes interact with macrophages through FN1. Further studies are needed to elucidate how chondrocytes affect macrophages. Given the crucial roles of chondrocytes and immune cells in IVDD, future treatments may involve altering chondrocyte types and regulating immune cells.

It should be noted that our study has several limitations. First, the healthy endplate tissues were absent in this study due to the ethical reasons and challenging human sample collection at a clinic. Consequently, at scRNA-seq, the data might not accurately distinguish the cell populations. Moreover, we did not use extensive clinical samples and data to validate the majority of the findings. Further large-scale experiments are needed to overcome the aforementioned difficulties and evaluate the mechanism of IVDD.

Overall, this study provides valuable reference information for future research on endplate degeneration as a potential therapeutic target. The identification of specific cell subsets and TFs may inform the development of novel therapeutic strategies that target these cells and their regulatory pathways. Additionally, the analysis of cell–cell interactions reveals a previously unrecognized link between chondrocytes and macrophages, suggesting the potential for new treatments that modulate this interaction to alleviate inflammation and promote tissue repair. Overall, our study opens up new avenues for investigation into the mechanisms underlying cartilage endplate degeneration and the development of targeted therapies for this debilitating condition.