Skip to main content
NCBI home page
Journal of Cellular and Molecular Medicine logoLink to Journal of Cellular and Molecular Medicine
. 2026 Jan 22;30(2):e71026. doi: 10.1111/jcmm.71026

Transcriptomic and Functional Comparison of Cells Isolated From Healthy and Degenerated Ovine Intervertebral Discs

Paul Humbert 1, Lucie Danet 1, Emmaëlle Carrot 1, Floriane Etienne 1,2, Boris Halgand 1, Frédéric Blanchard 1, Claire Vinatier 1, Jérôme Guicheux 1, Marion Fusellier 1,2, Catherine Le Visage 1, Romain Guiho 1,✉
PMCID: PMC12825213  PMID: 41568907

ABSTRACT

Intervertebral disc degeneration (IVDD) is a leading cause of chronic low back pain, yet its cellular and molecular mechanisms remain incompletely understood. Sheep represent a valuable in vivo and ex vivo model for IVDD due to their anatomical and biomechanical similarities with humans and the possibility to access disc samples at early stages of degeneration. In vitro, isolated annulus fibrosus (AF) and nucleus pulposus (NP) cells may provide insights into age‐associated degenerative processes; this work investigates how well they capture senescence and metabolic alterations observed in vivo. Transcriptomic profiling of AF and NP cells from healthy young lambs and mildly degenerated aged sheep revealed distinct age‐ and tissue‐specific signatures, with upregulation of inflammatory mediators, ECM‐remodelling enzymes, and senescence‐associated genes in aged cells. Cross‐species deconvolution using a human single‐cell RNA‐sequencing reference confirmed conserved transcriptional modules between aged sheep and human degenerated discs, underscoring the model's translational relevance. However, functional assays demonstrated comparable responses of young and aged cells under basal conditions and after exposure to pro‐degenerative stressors (IL‐1β, senescence induction). Altogether, these findings validate sheep cells as a suitable in vitro model for studying disc degeneration mechanisms and for preclinical testing, although aged donors offer no clear additional functional benefits.

Keywords: aging, energy metabolism, in vitro model, intervertebral disc degeneration, ovine spontaneous IVDD, sheep model, spine, transcriptomic analysis

Abbreviations

AF

annulus fibrosus

AFC

annulus fibrosus cells

CEP

cartilaginous endplate

GSEA

geneset enrichment analysis

IVD

intervertebral disc

IVDD

intervertebral disc degeneration

LBP

low back pain

NES

normalised enrichment scores

NP

nucleus pulposus

NPC

nucleus pulposus cells

OCR

oxygen consumption rate

OXPHOS

oxidative phosphorylation

1. Introduction

Chronic low back pain (LBP) is the leading cause of years lived with disability and affects over 600 million patients worldwide [1]. This debilitating condition is particularly prevalent in the aging population and among individuals with decreased levels of physical activity. Therefore, it is expected to affect an increasing number of people in the near future, with a considerable burden, both from a societal and individual patient perspective [2]. The urgent need for better management of LBP is one of the World Health Organization (WHO) priorities, which is reflected by the implementation of the Decade of Healthy Aging for 2021–2030. Chronic LBP has multiple causes, but among them, intervertebral disc degeneration (IVDD) is known to be the major contributor [3].

The intervertebral disc (IVD) is composed of two main regions—the annulus fibrosus (AF), a fibrous outer ring, and the nucleus pulposus (NP), a gel‐like core—connected to adjacent vertebrae by cartilaginous endplates (CEP). Each of those areas contains specialised cells responsible for matrix secretion, but the NP also has a contingent of cells from notochordal origin. IVDD is characterised by a global loss of cellularity, cellular senescence, extracellular matrix (ECM) alterations and dehydration, increased expression of MMPs and inflammatory factors, leading to a loss of IVD biomechanical properties, ectopic ingrowth of sensory nerves and blood vessels, pain and disability. Clinical management, including anti‐inflammatory and analgesic drugs for early degenerative stages, and surgery for advanced ones, only aims at controlling pain but does not address the underlying pathological processes [4]. Given the lack of curative options and the urgent need to identify and test new therapeutic opportunities, the development of relevant preclinical models that faithfully recapitulate human IVDD is essential.

Among laboratory animals for IVDD research, the sheep has emerged as a promising model organism [5, 6, 7, 8, 9] due to anatomical similarities with the human spine [10, 11] (e.g., comparable disc size and biomechanical constraints) and spontaneous radiological changes observed in aging animals resembling those found in human patients [12, 13]. This natural degeneration of sheep IVD may be caused by a decline in the notochordal cell population in adult animals [14], as reported in human physiology [15]. The ovine model of IVDD has the benefits of allowing clinically relevant surgical procedures [16], the generation of whole disc ex vivo explants [17] and primary cultures of isolated cells [18]. This wide range of applications is particularly interesting for translational research. However, despite numerous preclinical and in vitro studies using this animal model, it should be stressed that ovine disc cells and their behaviour in culture remain poorly characterised.

This study explores the relevance of cellular models derived from AF and NP tissues of both lambs and aged sheep, aiming to characterise their behaviour in vitro and gain insights into the age‐related disc degeneration processes. To this end, we performed an in‐depth transcriptomic analysis of expanded cells isolated from AF and NP in lambs and sheep to identify key differences related to the age of the donor animal or the tissue of origin, with a focus on cellular senescence and energy metabolism. First, cellular senescence has emerged as a hallmark of IVDD, contributing to matrix breakdown and the secretion of pro‐inflammatory mediators [19]. Second, an imbalance in energy metabolism between oxidative phosphorylation (OXPHOS) and glycolysis has also been recently identified as a potential driver of disc cell dysfunction, and its alteration could be critical to understand the pathophysiology of IVDD in the years to come [20]. In recent years, the advent of transcriptional studies on single cells has enabled the scientific community to compile important datasets to evaluate the heterogeneity of cellular states in healthy and diseased human discs. To compare the transcriptomes of our ovine cells with those of human discs, we cross‐examined the dataset generated by Cherif et al. [21], a single‐cell RNAseq study of NP and AF from a healthy and a diseased human disc. Ultimately, using relevant functional assays, we compared NP cells derived from aged sheep with those from young lambs, under basal conditions and after pro‐degenerative challenges such as IL‐1β exposure or senescence induction.

2. Materials and Methods

2.1. Sample Collection

As previously described [17], lumbar IVDs were collected from four 6‐month‐old female lambs and four 7–8‐year‐old female sheep (Vendée breed, GAEC HEAS, France) at the accredited Centre of Research and Pre‐Clinical Investigations (ONIRIS, National Veterinary School of Nantes) in accordance with European Directive 2010/63/EU. Post‐euthanasia, lumbar spines underwent sagittal T2‐weighted MRI, and IVDs were graded using the Pfirrmann score [22]. IVDs were then isolated with adjacent vertebral bone using an oscillating saw and transported in sterile medium for subsequent processing (extended protocol available in Supporting Information).

2.2. Histological Analysis

One representative lumbar IVD per animal was processed for histological analysis, after fixation with Paraformaldehyde (PFA, Sigma‐Aldrich) 4% for 5 days, followed by decalcification (Shandon TBD‐2 Decalcifier, Thermo Fisher Scientific) for 3 weeks and a post‐fixation step, after abundant rinsing, in 4% PFA for 24 h. IVDs were then frozen in isopentane cooled with dry ice and embedded in SCEM medium (Section Lab). Coronal cryosections of 10 μm thickness were made with a cryostat (CryoStar NX70, Thermo Fischer Scientific) and stained with either haematoxylin–eosin‐saffron (HES) or Alcian blue, according to standard procedures. Images of the stained sections were acquired with a slide scanner (NanoZoomer, Hamamatsu Photonics) and visualised with NDP.view2 software (Hamamatsu Photonics).

2.3. Cell Isolation and Culture

AF and NP cells were isolated as previously described [23]. Briefly, the 5 remaining lumbar IVDs were dissected from the vertebral bodies, AF and NP were separated and diced into pieces. Tissue fragments were digested at 37°C successively with 0.05% hyaluronidase (H4272, Sigma‐Aldrich) for 15 min, 0.2% trypsin (T9935, Sigma‐Aldrich) for 30 min, and 0.25% collagenase (C5138, Sigma Aldrich) overnight. The cells were filtered through a 70‐μm‐pore filter and seeded at 10,000 cells/cm2. AF and NP cells were grown at 37°C and 5% CO2 (extended protocol available in Supporting Information).

2.4. RNA Sequencing and Transcriptomic Analysis

Total RNA was extracted from lamb and sheep AF and NP cells at passage 1 (P1) using NucleoSpin RNA XS kit (Macherey‐Nagel, 740902). Library preparation was performed with Illumina Stranded mRNA Prep kit. Sequencing was then performed on an Illumina HiSeq with NovaSeq 6000 SP Reagent Kit (200 cycles) v1.5. Sequencing data was then aligned with STAR using Ovis_aries_rambouillet.ARS‐UI_Ramb_v2.0.111 genome. Secondary analysis was performed using {R 4.3.3} on RStudio (2023.06.0 + 421). Differential expression was performed using {DESeq2 1.38.1}. Geneset Enrichment Analysis (GSEA) was performed using the GSEA software (v4.3.2—UC San Diego and Broad Institute). Gene sets used in this study were either obtained from public databases (gene ontology (GO), KEGG, REACTOME, molecular signatures database—Broad Institute) or published literature: senescence profiling [24, 25] and Senescence‐Associated Secretory Phenotype (SASP) [26]. Weighted Gene Co‐Expression Network Analysis (WGCNA) was performed using {WGCNA 1.72} R package to identify pathways and networks associated with aging. A public dataset of single‐cell RNAseq on human disc cells (GSE199866) [21] has been cross‐analysed using a {Seurat 5.3.0} R pipeline. Briefly, after quality control (filtering on high mitochondrial genes percentage and low number of features, doublet removing with {DoubletFinder 2.0.4}), integration was performed with {rliger 2.2.0} using k = 20 and lambda = 40 settings. Clusters were annotated based on the literature consensus [27, 28, 29, 30]. Bulk deconvolution based on scRNA‐seq was performed with {MuSiC 1.0.0} [31]. scRNA‐seq single sample GSEA (ssGSEA) based on WGCNA module score was performed using {escape 2.4.0}.

2.5. Cellular Senescence Characterisation

NP cells at P3 were seeded in 96‐well plate (Sarstedt) at 5000 cells per well and either: treated with recombinant ovine IL‐1β (5 ng/mL; QP6215‐YE, enQuireBio) for one week with treatment renewal every 3 days; treated with etoposide (10 μM; E1383, Sigma‐Aldrich) for 24 h then maintained in culture for one week; or serum‐starved for one week, as previously described [32]. Senescence‐associated (SA)‐β‐galactosidase and 5‐ethynyl‐2′‐deoxyuridine (EdU) incorporation staining were performed on the same cells. Briefly, at day 7, EdU (10 μM) was added to the culture medium overnight. SA‐β‐galactosidase staining was performed at day 8 using the Senescence Cells Histochemical Staining Kit (CS0030, Sigma‐Aldrich) following the manufacturer's instructions. EdU incorporation staining was performed afterward using Click‐iT EdU Cell Proliferation Kit (Invitrogen) following the manufacturer's instructions. DAPI was used as a nuclear counterstain. Imaging was performed on a Zeiss Macroscope colour fluorescence, using Zen Blue software v3.7. Analysis was performed with CellProfiler software v4.2.5.

2.6. Measurement of Mitochondrial Respiration

NP cells at P3 were seeded in duplicate in XF96 plates at 60,000 cells per well and treated with recombinant ovine IL‐1β (1 ng/mL) or serum‐starved. After 24 h, the culture medium was replaced with DMEM phenol red‐free medium pH 7.4 (5030, Sigma‐Aldrich), supplemented with glutamine (4 mM), glucose (25 mM), and pyruvate (1 mM). The plate was incubated for 45 min at 37°C without CO2 to degas and allow accurate measurements. For the mitostress assay, oxygen consumption rate (OCR) measurements were performed every 5 min before and after the successive injections of oligomycin (2 μM), carbonylcyanide‐3‐chlorophenylhydrazon (CCCP, 4 μM), and a 1:1 mixture of rotenone/antimycine A (1 μM each) using the Seahorse XF Pro analyser (Agilent). Data analysis was performed using Wave software (Agilent) as previously described [33].

2.7. Statistical Analysis

All the results are presented as means ± SEM. When applicable, points on graphs represent the values for the different biological replicates. The statistical analyses were performed using GraphPad Prism software v10.0. Statistical significance was determined using the appropriate test, detailed in each figure legend.

3. Results

3.1. Sheep Develop Spontaneous Mild Intervertebral Disc Degeneration With Age

As the first step of our experimental process (Figure 1A), eight animals were obtained from the same breeder: four female sheep between 7 and 8 years old and 4 sex‐matched female lambs. Six‐month‐old lambs exhibited healthy IVDs based on MRI scoring, with a homogeneous Pfirrmann score of 1 for all discs scored. In contrast, aged sheep showed an increased score around 2, suggesting a homogeneous degenerative profile at the lumbar level (Figure 1B,C). All animals had six lumbar IVDs. Of note, one lamb presented an L5 vertebral deformity, which did not affect the Pfirrmann score or tissue processing. MRI‐based observations were confirmed with histological sections and staining for 1 disc per animal, selected as representative of the mean degeneration score for the animal (Figure 1D). Consistent with our previous report [12], healthy discs showed higher cellularity, while ‘ghost cells’ (i.e., residual space in the matrix after putative cell death) were seen mostly in mildly degenerated IVDs (Figure 1E). The presence of vertebral growth plates and the overall larger size of young discs illustrate that the lambs were not skeletally mature (Figure S1A).

FIGURE 1.

FIGURE 1

Open in a new tab

Sheep develop spontaneous mild intervertebral disc degeneration during aging. (A) Study design. Lumbar IVDs were collected from 4 female lambs (6‐month‐old) and 4 female sheep (7–8 years old). Medullar MRI of the thoracolumbar sheep spines was graded using the Pfirrmann system. One representative IVD per animal was processed for histology; the 5 others were dissected for AF and NP cells, separate isolation. Bulk RNA sequencing was performed for transcriptomic analysis. Functional characterisation was then carried out. (B) Representative MRI images and (C) mean lumbar IVDs Pfirrmann score per animal. Scale bar, 10 cm. p‐values were calculated using the Mann–Whitney test (N = 4 animals per group, n = 6 IVDs scored per animal). (D) Representative haematoxylin–eosin‐saffron (HES) staining images of all discs and (E) magnifications of areas of interest. Scale bars, (D) 10 mm, (E) 200 μm. AF, annulus fibrosus; CEP, cartilaginous endplate; NP, nucleus pulposus.

Cells could be efficiently isolated from both AF and NP tissues for all animals (Figure S1B–D). More NP cells (NPC) tended to be retrieved (Figure S1C), NPC from old animals exhibited a slightly longer doubling time compared to age‐matched AF cells (AFC) (Figure S1D). Overall, primary cultures from both tissues and age groups exhibited comparable growth behaviour under standard culture conditions and a similar morphology (Figure S1B).

3.2. Disc Cells From Animals of Different Ages and From Different Tissues Maintain a Specific Transcriptomic Profile in Culture

We next performed a bulk RNA sequencing on 16 cell samples at P1 (8 biological replicates isolated from either NP or AF tissues, including 4 young and 4 old animals). As the breed of sheep used in this study was the Vendée breed, the sequencing results were aligned with the genome of a neighbouring breed available on Ensembl: the Rambouillet sheep ( Ovis aries rambouillet). 30,722 automatically identified genes and pseudo‐genes were used for alignment; 12,159 were kept in this study after low count genes pre‐filtering. 91.2% were annotated using NCBI Wikigene. Forty‐eight (0.4%) were manually curated using human orthologous sequences (Figure S2). The first observation following this whole transcriptome analysis was the significant heterogeneity of our different biological samples, a heterogeneity that was particularly pronounced in older animals and in AF tissues, as shown by the dispersion and the absence of clear clustering on the PCA plot (Figure 2A). Although highly heterogeneous, disc cells from animals of different ages nevertheless retained an age‐specific profile in culture (Figure 2B). Differential Gene Expression between cells isolated from young and old animals identified 248 differentially expressed genes (DEGs) for NPC (Table S1) and 455 DEGs for AFC (Table S2). Some of these DEGs are commonly upregulated in AFC and NPC isolated from old animals, including GREB1, ESR1 linked to oestrogen‐response or EEPD1, TDP2, and AKR7A2 potentially involved in DNA damage‐response and stress‐response. On the other side of the volcano plots, developmental factors SOX6 and NCAM1 are overexpressed in both NPC and AFC isolated from young lambs, alongside BASP1, a known marker of juvenile human discs [34]. The whole transcriptome analysis also provides an overall view of the genes encoding collagens, revealing that their transcriptomic signature can distinguish not only cells isolated from AF or NP, but also AFC originating from young or old animals (Figure 2C). Notably, for NPC, the age‐separation based solely on collagens was not as clear‐cut. Interestingly, the genes encoding matrix proteins, previously highlighted in a proteomic study of aging human discs [35], helped distinguish NPC from AFC at any age, although with significant inter‐sample heterogeneity (Figure S3A).

FIGURE 2.

FIGURE 2

Open in a new tab

Transcriptomic profiling and differential gene expression in ovine disc cells from young and old animals. (A) Principal component analysis (PCA) of cells transcriptomes obtained from old (empty dot) and young (plain dot) animals, and from AFC (orange) and NPC (turquoise). (B) Volcano plot of differential gene expression in NPC (left) or AFC (right) from young versus old animals using the Wald test. (C) Clustering on collagens transcripts expression. (D) Geneset Enrichment Analysis (GSEA) young versus old samples using the permutation test. (E) GSEA enrichment plots for 3 genesets relative to cellular senescence (SenMayo) and energy metabolism (glycolysis and OXPHOS) for young NPC versus old NPC, and (F) for young AFC versus old AFC. NES, normalised enrichment score.

To complete the unbiased transcriptomic analysis of our cultured ovine disc cells, we performed a pathway enrichment analysis between cells isolated from young animals vs. old animals (Figure 2D). Notable differences were observed in pathways related to development (e.g., cartilage development, notochord and central nervous system development), which were significantly enriched in young samples. In contrast, cellular stress and inflammation pathways (e.g., DNA repair, NF‐κB activation, innate immune response activation) were enriched in old samples. Moreover, a switch in metabolism‐related processes (e.g., glycophagy and NADP metabolic process in old samples, regulation of intracellular pH in young samples) was observed (Figure 2D).

In order to verify whether predicted cellular senescence and energy metabolism imbalance described in IVDD could be confirmed in cells from aged animals compared to young ones, GSEA was performed separately on NPC and AFC using 3 relevant genesets: the SenMayo geneset [25] for senescence and both hallmark: glycolysis and hallmark: OXPHOS for energy metabolism imbalance. If Normalised Enrichment Scores (NES) seemed to indicate indeed an enrichment for cellular senescence processes in NPC and AFC from old animals (negative scores in favour of the ‘old’ condition, respectively NES = −0.772 and NES = −1.072—Figure 2E,F) and a metabolic switch toward glycolysis at the expense of OXPHOS in NPC from old animals (negative glycolysis NES and positive OXPHOS NES—Figure 2E), p‐values did not reach significance.

On the tissue of origin aspect, disc cells from animals of both ages retained also a tissue‐specific core profile (Figure S3B,C). 253 DEGs have been identified between AFC and NPC from young animals (Table S3). Among these DEGs, the genes encoding for the transcription factors FOXP2 were overexpressed in AFC (Foldchange > 2) and RUNX2 was overexpressed in NPC (Foldchange > 2) (Figure S3B). Surprisingly, the genes recognised to be markers differentiating NP and AF tissues, namely ACAN, COL2A1, and PAX1 expressed by NPC, and COL1A1 and HTRA1 expressed by AFC, failed to reach the significance threshold. CA12, a proposed marker of NPC, was a notable exception with a significant overexpression in NPC from young animals. This contrasts with samples from older animals, where these transcriptomic markers are clearly discriminating AFC from NPC (Foldchange > 1.3, 1.5, 1.4, 1.6 for ACAN, COL2A1, PAX1, CA12 in NPC and > 1.6, 1.4 for COL1A1, HTRA1 in AFC) (Figure S3C). Overall, transcriptomic differences were more importantly marked in cell populations derived from aged animals, with 1117 DEGs between AFC and NPC (Table S4). We then performed pathway enrichment analysis between ovine NPC and AFC (Figure S3D), which highlighted differences in the cellular processes involved (e.g., glucose response and acyl‐CoA biosynthesis for AFC; glycosylation and tryptophan metabolism for NPC). These differences may reflect NPC and AFC different adaptation strategies to the in vitro environment.

To further explain the relatively modest number of DEGs between NPC and AFC transcriptomes, particularly in young animals, we took advantage of the human scRNA‐seq dataset GSE199866 [21]. In line with our observations, the UMAP representation revealed a strong similarity between the transcriptomes of human cells freshly isolated from AF and NP. These cell types clustered together according to their cellular states rather than being separated by tissue of origin (Figure S3E).

Altogether, our data highlighted that ovine cells isolated and cultured from animals of different ages and from distinct tissues maintained an age‐ and tissue‐specific transcriptomic signature.

3.3. Cells Derived From Aged Ovine Discs Exhibit Human Disc Degeneration Signatures

To further dissect the intra‐samples heterogeneity observed in our transcriptomic dataset, we combined two complementary approaches: bulk RNA‐seq deconvolution using a single‐cell reference, and co‐expression network analysis via Weighted Gene Co‐Expression Network Analysis (WGCNA).

In order to identify the sub‐populations of NPC and AFC present in our disc ovine culture, we first use bulk deconvolution technique based on known reference cell annotation (Figure 3A). For this purpose, the scRNA‐seq GSE199866 dataset was used as previously. Cell clusters were annotated (Figures 3B and S4A) and their transcriptomic profile was used as a reference to infer cellular states present in each ovine sample (Figure 3C). Interestingly, we highlighted that in vitro culture of disc cells led to a marked enrichment in fibrotic (65%–75% in AF, ~90% in NP) and progenitor‐like cell states, regardless of donor age (Figure 3C). Notably, these two cell states represented minor populations in the native human disc (respectively 2% and 5% of the disc cells) and were not typically associated with IVDD (Figure S4B).

FIGURE 3.

FIGURE 3

Open in a new tab

Cells from ovine aged disc express human disc degeneration signatures. (A) Ovine bulk RNA‐seq deconvolution approach based on a human scRNA‐seq reference. The expression profiles of annotated cellular states served as a basis for computational deconvolution. (B) UMAP of scRNA‐seq human dataset (GSE199866) combining AFC and NPC transcriptome from one healthy and one IVDD intervertebral disc. Cellular states were annotated using literature consensus. (C) Estimated proportion of cellular states in ovine samples by MuSiC deconvolution strategy. (D) Bioinformatic approach to the weighted correlation network analysis (WGCNA) enabling the identification of gene modules associated with the age of ovine IVD. The scores associated with these modules were calculated for healthy and degenerated human IVD cells. (E) Ranked GSEA of the genes associated with Cyan and Green (up in young animals) and Pink (up in old animals) modules. (F) Top‐gene modules used as ssGSEA score on human progenitor and fibrotic cells from healthy (clear) and IVDD (grey) disc. p‐values were calculated with Wilcoxon signed‐rank test, ns, not significant, *p < 0.05; **p < 0.01; ***p < 0.001.

To explore underlying gene expression programs, we applied WGCNA across all 16 bulk RNA‐seq samples (Figure 3D) and identified six co‐expression modules (Figure S4C). Besides gene modules associated with AF tissue (Yellow and Magenta modules) or NP tissue (Tan and Brown modules), WGCNA identified 3 modules associated with young (Cyan and Green modules) or old age (Pink module) (Figure S4D). For subsequent analysis, we focused on these 3 modules of interest. Functional annotation of these modules highlighted key biological processes differentially active with aging (Figure 3E). The Green module was positively enriched for pathways related to cartilage development and negatively enriched for DNA repair and cellular senescence. The Cyan module showed enrichment in chondrocyte differentiation and NADH metabolic processes, while being negatively associated with microautophagy, mitochondrial fission, and innate immune signaling. In contrast, the Pink module, upregulated in aged samples, was enriched for myeloid cell activation, oxidative stress response, and unsaturated fatty acid metabolism, while being negatively associated with DNA replication initiation and mitochondrial genome maintenance. These distinct co‐expression signatures suggest that aging is accompanied by a coordinated shift away from anabolic and developmental programs toward inflammatory, metabolic, and stress‐related responses, consistent with hallmarks of disc degeneration.

To assess the relevance of these ovine transcriptional programs to human IVDD, we projected the top 30 hub genes from each module onto the human scRNA‐seq dataset (Figure 3F and Table S5). Using single‐sample GSEA (ssGSEA), we computed enrichment scores for each module across individual fibrotic and progenitor cells, two states commonly observed in vitro according to the bulk deconvolution. This analysis revealed a significantly higher proportion of cells with high Green and Cyan scores in healthy human discs compared to IVDD samples. Conversely, cells with high Pink module scores were significantly more frequent in fibrotic cells from degenerated discs (Figure 3F). These results indicate that the transcriptional modules enriched in young ovine cells reflect a healthy disc state, while those upregulated with age align with molecular signatures of human disc degeneration.

3.4. NPC From Young and Aged Ovine Disc Share Similar Senescence and Energy Metabolic Responses to Stress‐Related Stimuli

Based on the observed discrepancy in enriched pathways related to inflammation, senescence, and energy metabolism, we sought to better characterise cellular functions associated with these processes. Given their central position in the pathophysiology of IVDD, the role of NPC in disc degeneration has been the subject of numerous studies in the literature. In this section, we chose to functionally explore the stress‐response of these cells at baseline and after stimuli. Cell proliferation, SA‐β‐Gal expression, and mitochondrial respiratory profiles of NPC from young and old animals in response to various stimuli were evaluated (Figures 4 and S5). IL‐1β treatment mimicked inflammatory processes, etoposide treatment served as an efficient senescence inducer, and serum starvation was used as a basic culture condition known for influencing cell proliferation and metabolism. To validate the responsiveness of NPC to IL‐1β, quantification of MMP3 and IL6 transcript expression was performed, showing significant overexpression of these markers in cells from both young and aged animals (Figure S5A).

FIGURE 4.

FIGURE 4

Open in a new tab

NPC from young and aged ovine disc share common senescence and mitochondrial energetic responses to stress‐related stimuli. (A) Co‐immunofluorescence of EdU incorporation, an indicator of cell proliferation, and SA‐β‐Gal staining, indicating high lysosomal content and potential marker of cellular senescence. DAPI is used as a nuclear marker. Arrows highlight SA‐β‐Gal positive blue staining. (B) Quantification of % of NPC from young and old animals positive for EdU incorporation and SA‐β‐Gal staining. (C) Functional investigation of mitochondrial energetic profiles changes by Seahorse evaluating Oxygen Consumption Rate (OCR) measurements using the mitostress test assay in NPC from young and old animals. (D) Quantification of maximal respiration and ATP production. ns, not significant, p‐values were calculated using the Kruskal–Wallis test, and Dunn's post hoc test; N = 4 animals per group.

In control conditions, NPC from both young and old animals were proliferating similarly, with around 80% of them incorporating EdU overnight, and only marginally expressing SA‐β‐Gal (Figures 4A,B and S5B). All treatments greatly restricted cell proliferation, down below 40% EdU incorporation with serum starvation and even under 10% with IL‐1β or etoposide treatment. Etoposide was the only treatment increasing SA‐β‐Gal expression up to 10% of stained cells. However, no difference in cell state was observed between cells from young and old animals, at baseline and after treatments.

Accordingly, NPC from young and aged animals exhibited similar mitochondrial energetic profiles in basal conditions and in responses to either IL‐1β or serum starvation (Figures 4C,D and S5C). This included comparable maximal respiration and ATP production for NPC from young and old animals at baseline, and the same increase of maximal respiration and decrease of ATP production in the IL‐1β‐treated condition. Lastly, serum starvation reduced maximal respiration and ATP production similarly in young and old NPC.

Therefore, these functional experiments highlighted that NPC derived from mildly degenerated discs remain responsive to inflammatory, senescence‐inducing, and metabolic stress stimuli, similarly to NPC derived from healthy young IVD.

4. Discussion

The sheep model is already widely used and recognised for ex vivo and in vivo studies of IVDD [11], positioning ovine‐derived cell culture models as particularly valuable tools to bridge basic and translational research. Considering that its naturally occurring degenerative process resembles the human one, cellular models from sheep are more likely to mimic what could be observed in human cells. Moreover, the availability of animals across a wide age range offers the opportunity to investigate developmental, maturational, and degenerative processes. However, this advantage comes with significant limitations. A key limitation of our study is the exclusive use of female animals. Livestock management practices primarily dictated this choice: in farming, male lambs are generally slaughtered for meat, whereas females are retained for breeding and milk production and can reach advanced ages (6–7 years). As a result, aged males are rarely available, and using females across all age groups avoids introducing a sex bias between cohorts. While sex‐specific differences in IVDD have not been clearly established in sheep, they are well documented in humans. Biological factors such as hormones, immunity, behaviour, and gene expression contribute to sex‐related differences in chronic diseases. Importantly, low back pain, a clinical manifestation of IVDD, shows consistently higher prevalence and incidence in women than in men across all age groups, with even more pronounced differences in older populations [36]. Therefore, although driven by practical considerations, the exclusive use of female sheep in this study may also increase the translational relevance of the model. Finally, this approach is consistent with the 3R's principle of reduction, as we routinely obtain female animals from local farms, avoiding the need to breed males for research purposes specifically.

A clear strength of the ovine model lies in the anatomical size of the IVD close to the human ones, allowing for macroscopically distinguishing and dissecting the AF from the NP. This facilitates the independent isolation and culture of the two main disc cell types, a significant technical advantage over smaller animal models where such dissection is challenging and imprecise. In this study, we efficiently isolated and expanded AFC and NPC from all individuals across both age groups. AFC and NPC retained somewhat distinct transcriptomic signatures in vitro that were more marked in cells isolated from older animals, likely reflecting full maturation of the IVD tissue. Notably, AFC, often understudied in the context of IVDD, were readily available and displayed age‐related transcriptomic features, supporting their inclusion in studies investigating disc degeneration and in the preclinical evaluation of biomaterials, such as those designed for herniated disc repair [37, 38].

In addition to the technical and logistical burden that is represented by IVDs' isolation from these large animals, the main drawback highlighted here is the pronounced inter‐individual variability, illustrated by the absence of clear clustering on RNA‐seq PCA plot. While this mirrors the biological heterogeneity seen in human populations and may thus increase translational relevance compared to studies using inbred rodent models or cell lines, it presents a challenge for statistical analyses and experimental reproducibility. The use of bioinformatics tools built around human gene sets further complicates interpretation, as species‐specific differences in gene annotation and pathway mapping can introduce biases or lead to incomplete pathway inference. Nevertheless, cross‐species comparison of our data with a human dataset indicates that mainly ‘fibrotic’ and, to a lesser extent, ‘progenitor‐like’ IVD cell states are represented after in vitro expansion. Importantly, this enrichment is likely influenced by culture conditions, as monolayer expansion on rigid plastic under high‐nutrient conditions may selectively favour fibroblastic phenotypes and the proliferation of progenitor‐like cells, or drive phenotypic convergence through dedifferentiation, rather than reflecting the native cellular composition of the disc [39]. Future studies using ovine disc cell culture models should aim to disentangle intrinsic disc cell heterogeneity from culture‐induced phenotypic selection or convergence. In particular, a direct comparison with freshly isolated ovine disc cells would be required to determine the extent to which fibrotic and progenitor‐like profiles pre‐exist in vivo versus being induced or amplified during in vitro expansion, which represents an important limitation of the present study. Importantly, linking the aging‐related transcriptional signature of sheep cells with data from human samples nonetheless reinforces the translational relevance and overall similarity between ovine and human IVDD.

In the functional assays, sheep NP cells were compatible with all experimental technical approaches used with minimal adjustment to standard protocols. We used commercially available ovine recombinant IL‐1β, but cells were also responsive to the human recombinant cytokine (data not shown). However, a notable limit of this model may come with the use of antibody‐based approaches for which anti‐sheep protein specificity may often be unproven or unavailable. Cells from young and old animals showed similar mitochondrial energetic profiles and comparable susceptibility to senescence, independently of the treatment tested. However, these observations must be interpreted with caution. The overall low level of senescence detected in our cultures, together with the preferential survival and expansion of proliferative cells during in vitro passaging, may mask more subtle age‐related differences in mitochondrial function. In addition, the use of higher‐passage cells for functional assays, required to obtain sufficient cell numbers, likely contributed to phenotypic convergence between age groups. Consequently, conclusions regarding age‐dependent mitochondrial function should be validated in experimental systems that better preserve native disc cell states. On the bright side, these findings show that NPC derived from mildly degenerated discs remain viable and thus potentially responsive to regenerative stimuli in vivo. However, this functional result contrasts with studies in rabbits, where age‐associated mitochondrial energetic dysfunction and increased senescence were reported in cultured IVD cells [40]. While this discrepancy may reflect species‐specific differences in disc biology (i.e., rabbits develop spontaneous IVDD [41] yet retain notochordal cells throughout adulthood, in contrast to sheep [14]), it may also be influenced by methodological differences between studies. In particular, more robust or prolonged senescence induction protocols used in rabbit models could have amplified age‐related mitochondrial and senescence‐associated phenotypes. The maintenance of a notochordal‐derived progenitor pool in rabbits, together with such experimental differences, may therefore contribute to divergent aging processes, IVDD mechanisms, and regenerative capacities between species.

Altogether, our findings support the use of sheep‐derived NP and AF cells as relevant in vitro models of early‐stage IVDD. While cells from animals of all ages are suitable for experimentation, our findings suggest that donor age alone may not be the primary determinant of functional outcomes in expanded cultures, particularly when aging‐related markers are subtle and only a subset of cells survives and proliferates in vitro. To better preserve native disc cell phenotypes and limit progenitor overexpansion and dedifferentiation, future studies should prioritise the use of freshly isolated cells encapsulated in three‐dimensional systems or biomimetic matrices. Such approaches may constrain cell spreading and proliferation while maintaining key biophysical and biochemical cues of the disc microenvironment, thereby improving physiological relevance. In this context, strategies including culture in three‐dimensional spheroids [42, 43], encapsulation in hydrogels mimicking disc matrix properties [44, 45], or maintenance in hyperosmolar [46] or hypoxic conditions [47, 48, 49] may help preserve native cellular states that are partially lost upon monolayer expansion.

In conclusion, our study demonstrates that ovine NPC and AFC represent a promising model for the study of mechanisms implicated in disc aging and degeneration, and for the early evaluation of potential regenerative therapies, while suffering from the same limitations as any cell culture model. Similarities found with the transcriptomic signature of human cells reinforce the translational relevance of the ovine model from in vitro to in vivo experiments.

Author Contributions

R.G., J.G. and C.L.V. conceived and designed the study. P.H., L.D., E.C. and M.F. performed the formal analyses. J.G., C.L.V., F.B. and R.G. acquired funding. P.H., L.D., E.C., B.H. and F.E. carried out investigations. P.H., L.D., R.G., C.V., M.F. and F.B. developed the methodology. P.H. and R.G. drafted the original manuscript. All authors read, amended, and approved the final version of the manuscript.

Ethics Statement

The animals used in this study were housed in accordance with good practices and animal welfare standards, under the supervision of veterinarians. Euthanasia was performed according to good practices and legislation (European Directive 2010/63/EU) in the accredited Centre of Research and Pre‐clinical Investigations (CRIP) at the ONIRIS‐National Veterinary School of Nantes. All efforts were made to minimise the number of animals used and their suffering.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Relative to Main Figure 1. (A) Representative Alcian Blue staining images. Scale bar: 10 mm. (B) Cell morphology at passage P1. Scale bar: 100 μm. (C) Cell count after cell isolation. (D) Cellular doubling time between passage P0 and P1. n.s., not significant; *p < 0.05. p‐values were calculated using Mann Whitney test; N = 4 young and N = 4 old animals.

Figure S2: Schematic bioinformatic analysis pipeline. (A) From primary to secondary analysis pipeline.

Figure S3: Relative to main Figure 2. (A) Heatmap showing sample clustering based on matrix‐related transcripts expression from DIPPER analysis [35]. (B) Volcano plot of differential gene expression in NPC versus AFC from young and (C) old animals using the Wald test. (D) Geneset Enrichment Analysis (GSEA) AFC versus NPC samples using the permutation test. (E) UMAP of scRNA‐seq human dataset (GSE199866) split between one healthy and one IVDD intervertebral disc, cells from NP tissue are highlighted in turquoise and from AF tissue in orange.

Figure S4: Relative to main Figure 3. (A) UMAP of scRNA‐seq human dataset (GSE199866) used as basis for bulk RNA‐seq deconvolution combining AFC and NPC transcriptome, split between one healthy and one degenerated intervertebral disc. Cellular states were annotated using literature consensus. (B) Proportion of the different cellular states present in healthy and IVDD condition. (C) Dendrogram plot showing hierarchical clustering of the 12159 genes in 14 modules after merging, a different colour represents each module. (D) Dendrogram and heatmap generated with the module eigengenes for the 14 modules identified. Significant association between modules and origin of the cells are highlighted below (Yng: young). p‐values were calculated with the Student asymptotic p‐value for correlation.

Figure S5: Relative to main Figure 4. NPC from young and aged ovine disc share common senescence and mitochondrial energetic responses to stress‐related stimuli. (A) IL6 and MMP3 transcript expression by RT‐qPCR of NPC from young and old animals following 24 h of IL‐1 β treatment or serum starvation. (B) Quantification of % of NPC from young and old animals positive for EdU incorporation and SA‐β‐Gal staining. (C) Quantification of maximal respiration and ATP production. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. p‐values were calculated using the Repeated Measures ANOVA test, and Tukey's post‐hoc test; N = 8 animals.

JCMM-30-e71026-s004.pdf (3.1MB, pdf)

Table S1: Differential Expression Results between NPC from young and old animals. Standard differential expression analysis results using DESeq2 pipeline. BaseMean = mean of normalised read count across all samples. log2 fold change = log2 Fold change between groups, positive indicates upregulated in NPC isolated from old animals, padj = p‐value adjusted for multiple testing using Benjamini–Hojberg correction.

JCMM-30-e71026-s001.xlsx (21.8KB, xlsx)

Table S2: Differential Expression Results between AFC from young and old animals. Standard differential expression analysis results using DESeq2 pipeline. BaseMean= mean of normalised read count across all samples. log2 fold change = log2 Fold change between groups, positive indicates upregulated in AFC isolated from old animals, padj = p‐value adjusted for multiple testing using Benjamini–Hojberg correction.

JCMM-30-e71026-s006.xlsx (34.8KB, xlsx)

Table S3: Differential Expression Results between NPC and AFC isolated from young animals. Standard differential expression analysis results using DESeq2 pipeline. BaseMean = mean of normalised read count across all samples. log2 fold change = log2 Fold change between groups, positive indicates upregulated in NPC, padj = p‐value adjusted for multiple testing using Benjamini–Hojberg correction.

JCMM-30-e71026-s003.xlsx (22.2KB, xlsx)

Table S4: Differential Expression Results between NPC and AFC isolated from old animals. Standard differential expression analysis results using DESeq2 pipeline. BaseMean = mean of normalised read count across all samples. log2 fold change = log2 Fold change between groups, positive indicates upregulated in NPC, padj = p‐value adjusted for multiple testing using Benjamini–Hojberg correction.

JCMM-30-e71026-s002.xlsx (76.9KB, xlsx)

Table S5: List of the top‐30 genes associated with WGCNA modules. List of the 30 genes most associated with the different modules identified by WGCNA, ranked by importance in the module. The modules included are the modules associated with AFC (Yellow and Magenta modules), with NPC (Tan and Brown modules), with young and (Cyan and Green modules) and with old age (Pink module).

JCMM-30-e71026-s005.xlsx (12KB, xlsx)

Acknowledgements

The authors thank O. Gauthier, A. Lafragette, P. Roy, S. Madec, D. Rouleau, P. Monmousseau from the Center for Research and Preclinical Investigation (C.R.I.P.; Oniris Nantes Atlantic College of Veterinary Medicine, Food Science and Engineering, Nantes, France) for their professionalism and precious help in this study. The authors thank C. Pecqueur (CRCI2NA, Nantes Université, INSERM U1307, CNRS 6075, Nantes, France) for her help with Seahorse technology, with financial support from ITMO Cancer of Aviesan within the framework of the 2021‐2030 Cancer Control Strategy. The authors acknowledge the SC3M platform from the Inserm/NU/ONIRIS UMR1229 RMeS Laboratory and SFR Bonamy. They also acknowledge the MicroPICell core facility (SFR Bonamy, BioCore, Inserm UMS 016, CNRS UAR 3556, Nantes, France), member of the Scientific Interest Group (GIS) Biogenouest, IBISA, and the national infrastructure France‐Bioimaging supported by the French national research agency (ANR‐10‐INBS‐04). The authors thank the Genomics Core Facility GenoA, member of Biogenouest and France Genomique, and the Bioinformatics Core Facility BiRD, member of Biogenouest and Institut Français de Bioinformatique (ANR‐11‐INBS‐0013) for the use of their resources and their technical support. This study was supported by Grants from the French Society of Rheumatology (Société Française de Rhumatologie—SFR, to R.G.), the French Agence Nationale de la Recherche (ANR‐19‐CE18‐0020‐01, EXCELLDISC, to C.L.V.), the European Commission's Horizon2020 funding program for the iPSpine project [grant number 825925, to J.G.]and the Fondation pour la Recherche Médicale (DISCAFF project), E.C. received a PhD fellowship from the Pays de la Loire Region, and L.D. was supported by a doctoral fellowship funded by the French Ministry of Higher Education and Research. Schematics were created with Biorender.com.

Humbert P., Danet L., Carrot E., et al., “Transcriptomic and Functional Comparison of Cells Isolated From Healthy and Degenerated Ovine Intervertebral Discs,” Journal of Cellular and Molecular Medicine 30, no. 2 (2026): e71026, 10.1111/jcmm.71026.

Data Availability Statement

The raw and processed transcriptomic data that support the findings of this study have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE302551. The complete bioanalysis pipeline is available at https://gitlab.univ‐nantes.fr/guihomics/ovine_IVD_bulkRNAseq. Other raw data and materials are available from the corresponding author on reasonable request.

References

  • 1. Knezevic N. N., Candido K. D., Vlaeyen J. W. S., Van Zundert J., and Cohen S. P., “Low Back Pain,” Lancet 398, no. 10294 (2021): 78–92. [DOI] [PubMed] [Google Scholar]
  • 2. Ferreira M. L., de Luca K., Haile L. M., et al., “Global, Regional, and National Burden of Low Back Pain, 1990–2020, Its Attributable Risk Factors, and Projections to 2050: A Systematic Analysis of the Global Burden of Disease Study 2021,” Lancet Rheumatology 5, no. 6 (2023): e316–e329. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Lyu F. J., Cui H., Pan H., et al., “Painful Intervertebral Disc Degeneration and Inflammation: From Laboratory Evidence to Clinical Interventions,” Bone Research 9, no. 1 (2021): 1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4. Xin J., Wang Y., Zheng Z., Wang S., Na S., and Zhang S., “Treatment of Intervertebral Disc Degeneration,” Orthopaedic Surgery 14, no. 7 (2022): 1271–1280. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Hong J. Y., Kim H., Jeon W. J., et al., “Animal Models of Intervertebral Disc Diseases: Advantages, Limitations, and Future Directions,” Neurology International 16, no. 6 (2024): 1788–1818. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Tang S. N., Bonilla A. F., Chahine N. O., et al., “Controversies in Spine Research: Organ Culture Versus In Vivo Models for Studies of the Intervertebral Disc,” JOR Spine 5, no. 4 (2022): e1235. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Daly C., Ghosh P., Jenkin G., Oehme D., and Goldschlager T., “A Review of Animal Models of Intervertebral Disc Degeneration: Pathophysiology, Regeneration, and Translation to the Clinic,” BioMed Research International 2016 (2016): 5952165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Malli S. E., Kumbhkarn P., Dewle A., and Srivastava A., “Evaluation of Tissue Engineering Approaches for Intervertebral Disc Regeneration in Relevant Animal Models,” ACS Applied Bio Materials 4, no. 11 (2021): 7721–7737. [DOI] [PubMed] [Google Scholar]
  • 9. Peng Y., Qing X., Shu H., et al., “Proper Animal Experimental Designs for Preclinical Research of Biomaterials for Intervertebral Disc Regeneration,” Biomaterials Translational 2, no. 2 (2021): 91–142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Wilke H. J., Kettler A., Wenger K. H., and Claes L. E., “Anatomy of the Sheep Spine and Its Comparison to the Human Spine,” Anatomical Record 247, no. 4 (1997): 542–555. [DOI] [PubMed] [Google Scholar]
  • 11. Fusellier M., Clouet J., Gauthier O., Tryfonidou M., Le Visage C., and Guicheux J., “Degenerative Lumbar Disc Disease: In Vivo Data Support the Rationale for the Selection of Appropriate Animal Models,” eCM 39 (2020): 17–48. [DOI] [PubMed] [Google Scholar]
  • 12. Bouhsina N., Decante C., Hardel J. B., et al., “Comparison of MRI T1, T2, and T2* Mapping With Histology for Assessment of Intervertebral Disc Degeneration in an Ovine Model,” Scientific Reports 12, no. 1 (2022): 5398. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Nisolle J. F., Bihin B., Kirschvink N., et al., “Prevalence of Age‐Related Changes in Ovine Lumbar Intervertebral Discs During Computed Tomography and Magnetic Resonance Imaging,” Comparative Medicine 66, no. 4 (2016): 300–307. [PMC free article] [PubMed] [Google Scholar]
  • 14. Hunter C. J., Matyas J. R., and Duncan N. A., “Cytomorphology of Notochordal and Chondrocytic Cells From the Nucleus Pulposus: A Species Comparison,” Journal of Anatomy 205, no. 5 (2004): 357–362. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Warin J., Vedrenne N., Tam V., et al., “In Vitro and In Vivo Models Define a Molecular Signature Reference for Human Embryonic Notochordal Cells,” iScience 27, no. 2 (2024): 109018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Decante C., Clouet J., Hamel A., et al., “Collateral Effects of Targeting the Nucleus Pulposus via a Transpedicular or Transannular Surgical Route: A Combined X‐Ray, MRI, and Histological Long‐Term Descriptive Study in Sheep,” European Spine Journal 30, no. 2 (2021): 585–595. [DOI] [PubMed] [Google Scholar]
  • 17. Frapin L., Clouet J., Chédeville C., et al., “Controlled Release of Biological Factors for Endogenous Progenitor Cell Migration and Intervertebral Disc Extracellular Matrix Remodelling,” Biomaterials 253 (2020): 120107. [DOI] [PubMed] [Google Scholar]
  • 18. Carrot E., de Villemagne P., Humbert P., et al., “Replication of an Intervertebral Disc Using an In‐House Bioprinter: A Proof‐of‐Concept Study,” International Journal of Bioprinting 11, no. 1 (2024): 470–484. [Google Scholar]
  • 19. Patil P., Niedernhofer L. J., Robbins P. D., Lee J., Sowa G., and Vo N., “Cellular Senescence in Intervertebral Disc Aging and Degeneration,” Current Molecular Biology Reports 4, no. 4 (2018): 180–190. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Wu J., Zhou X., Xu X., and Xie J., “A Molecular Chemical Perspective: Mitochondrial Dynamics Is Not a Bystander of Cartilage Diseases,” ACS Pharmacology & Translational Science 8, no. 6 (2025): 1473–1497. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Cherif H., Mannarino M., Pacis A. S., et al., “Single‐Cell RNA‐Seq Analysis of Cells From Degenerating and Non‐Degenerating Intervertebral Discs From the Same Individual Reveals New Biomarkers for Intervertebral Disc Degeneration,” International Journal of Molecular Sciences 23, no. 7 (2022): 3993. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Pfirrmann C. W., Metzdorf A., Zanetti M., Hodler J., and Boos N., “Magnetic Resonance Classification of Lumbar Intervertebral Disc Degeneration,” Spine (Philadelphia, PA 1976) 26, no. 17 (2001): 1873–1878. [DOI] [PubMed] [Google Scholar]
  • 23. Le Moal B., Lepeltier É., Rouleau D., et al., “Lipid Nanocapsules for Intracellular Delivery of MicroRNA: A First Step Towards Intervertebral Disc Degeneration Therapy,” International Journal of Pharmaceutics 624 (2022): 121941. [DOI] [PubMed] [Google Scholar]
  • 24. Fridman A. L. and Tainsky M. A., “Critical Pathways in Cellular Senescence and Immortalization Revealed by Gene Expression Profiling,” Oncogene 27, no. 46 (2008): 5975–5987. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Saul D., Kosinsky R. L., Atkinson E. J., et al., “A New Gene Set Identifies Senescent Cells and Predicts Senescence‐Associated Pathways Across Tissues,” Nature Communications 13, no. 1 (2022): 4827. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Acosta J. C., Banito A., Wuestefeld T., et al., “A Complex Secretory Program Orchestrated by the Inflammasome Controls Paracrine Senescence,” Nature Cell Biology 15, no. 8 (2013): 978–990. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Jia S., Liu H., Yang T., et al., “Single‐Cell Sequencing Reveals Cellular Heterogeneity of Nucleus Pulposus in Intervertebral Disc Degeneration,” Scientific Reports 14, no. 1 (2024): 27245. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Han S., Zhang Y., Zhang X., et al., “Single‐Cell RNA Sequencing of the Nucleus Pulposus Reveals Chondrocyte Differentiation and Regulation in Intervertebral Disc Degeneration,” Frontiers in Cell and Developmental Biology 10 (2022): 824771. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Chen F., Lei L., Chen S., et al., “Serglycin Secreted by Late‐Stage Nucleus Pulposus Cells Is a Biomarker of Intervertebral Disc Degeneration,” Nature Communications 15, no. 1 (2024): 47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Tu J., Li W., Yang S., et al., “Single‐Cell Transcriptome Profiling Reveals Multicellular Ecosystem of Nucleus Pulposus During Degeneration Progression,” Advanced Science 9, no. 3 (2022): 2103631. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Wang X., Park J., Susztak K., Zhang N. R., and Li M., “Bulk Tissue Cell Type Deconvolution With Multi‐Subject Single‐Cell Expression Reference,” Nature Communications 10, no. 1 (2019): 380. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Georget M., Defois A., Guiho R., et al., “Development of a DNA Damage‐Induced Senescence Model in Osteoarthritic Chondrocytes,” Aging (Albany NY) 15, no. 17 (2023): 8576–8593. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Defois A., Bon N., Charpentier A., et al., “Osteoarthritic Chondrocytes Undergo a Glycolysis‐Related Metabolic Switch Upon Exposure to IL‐1b or TNF,” Cell Communication and Signaling 21, no. 1 (2023): 137. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Tang X., Jing L., Richardson W. J., et al., “Identifying Molecular Phenotype of Nucleus Pulposus Cells in Human Intervertebral Disc With Aging and Degeneration,” Journal of Orthopaedic Research 34, no. 8 (2016): 1316–1326. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Tam V., Chen P., Yee A., et al., “DIPPER, a Spatiotemporal Proteomics Atlas of Human Intervertebral Discs for Exploring Ageing and Degeneration Dynamics,” eLife 9 (2020): e64940. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36. Saravanan A., Jaber R., Zhao B., et al., “Prevalence and Sociodemographic Correlates of Low Back Pain in a Diverse U.S. Population: Insights From the All of Us Research Program,” Pain Management Nursing 26 (2025): 611–619, S1524‐9042(25)00200‐0. [DOI] [PubMed] [Google Scholar]
  • 37. Gluais M., Clouet J., Fusellier M., et al., “In Vitro and In Vivo Evaluation of an Electrospun‐Aligned Microfibrous Implant for Annulus Fibrosus Repair,” Biomaterials 205 (2019): 81–93. [DOI] [PubMed] [Google Scholar]
  • 38. Falcoz C., Chaaban M., Paniagua C., et al., “Design and Ex Vivo Evaluation of a PCLA Degradable Device to Improve Annulus Fibrosus Repair,” ACS Applied Bio Materials 8, no. 2 (2025): 1097–1107. [DOI] [PubMed] [Google Scholar]
  • 39. Williams R. J., Laagland L. T., Bach F. C., et al., “Recommendations for Intervertebral Disc Notochordal Cell Investigation: From Isolation to Characterization,” JOR Spine 6, no. 3 (2023): e1272. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40. Hartman R., Patil P., Tisherman R., et al., “Age‐Dependent Changes in Intervertebral Disc Cell Mitochondria and Bioenergetics,” European Cells and Materials 36 (2018): 171–183. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Clouet J., Pot‐Vaucel M., Grimandi G., et al., “Characterization of the Age‐Dependent Intervertebral Disc Changes in Rabbit by Correlation Between MRI, Histology and Gene Expression,” BMC Musculoskeletal Disorders 12 (2011): 147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Zhang X., Guerrero J., Croft A. S., Albers C. E., Häckel S., and Gantenbein B., “Spheroid‐Like Cultures for Expanding Angiopoietin Receptor‐1 (Aka. Tie2) Positive Cells From the Human Intervertebral Disc,” International Journal of Molecular Sciences 21, no. 24 (2020): 9423. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Kushioka J., Kaito T., Chijimatsu R., et al., “A Novel and Efficient Method for Culturing Mouse Nucleus Pulposus Cells,” Spine Journal 19, no. 9 (2019): 1573–1583. [DOI] [PubMed] [Google Scholar]
  • 44. Yuan M., Leong K. W., and Chan B. P., “Three‐Dimensional Culture of Rabbit Nucleus Pulposus Cells in Collagen Microspheres,” Spine Journal 11, no. 10 (2011): 947–960. [DOI] [PubMed] [Google Scholar]
  • 45. Tan X., Jain E., Barcellona M. N., et al., “Integrin and Syndecan Binding Peptide‐Conjugated Alginate Hydrogel for Modulation of Nucleus Pulposus Cell Phenotype,” Biomaterials 277 (2021): 121113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Laagland L. T., Bach F. C., Creemers L. B., Le Maitre C. L., Poramba‐Liyanage D. W., and Tryfonidou M. A., “Hyperosmolar Expansion Medium Improves Nucleus Pulposus Cell Phenotype,” JOR Spine 5, no. 3 (2022): e1219. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Jaworski L. M., Kleinhans K. L., and Jackson A. R., “Effects of Oxygen Concentration and Culture Time on Porcine Nucleus Pulposus Cell Metabolism: An In Vitro Study,” Frontiers in Bioengineering and Biotechnology 7 (2019): 00064. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. Novais E. J., Choi H., Madhu V., et al., “Hypoxia and Hypoxia‐Inducible Factor‐1α Regulate Endoplasmic Reticulum Stress in Nucleus Pulposus Cells: Implications of Endoplasmic Reticulum Stress for Extracellular Matrix Secretion,” American Journal of Pathology 191, no. 3 (2021): 487–502. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49. Humphreys M. D., Ward L., Richardson S. M., and Hoyland J. A., “An Optimized Culture System for Notochordal Cell Expansion With Retention of Phenotype,” JOR Spine 1, no. 3 (2018): e1028. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S1: Relative to Main Figure 1. (A) Representative Alcian Blue staining images. Scale bar: 10 mm. (B) Cell morphology at passage P1. Scale bar: 100 μm. (C) Cell count after cell isolation. (D) Cellular doubling time between passage P0 and P1. n.s., not significant; *p < 0.05. p‐values were calculated using Mann Whitney test; N = 4 young and N = 4 old animals.

Figure S2: Schematic bioinformatic analysis pipeline. (A) From primary to secondary analysis pipeline.

Figure S3: Relative to main Figure 2. (A) Heatmap showing sample clustering based on matrix‐related transcripts expression from DIPPER analysis [35]. (B) Volcano plot of differential gene expression in NPC versus AFC from young and (C) old animals using the Wald test. (D) Geneset Enrichment Analysis (GSEA) AFC versus NPC samples using the permutation test. (E) UMAP of scRNA‐seq human dataset (GSE199866) split between one healthy and one IVDD intervertebral disc, cells from NP tissue are highlighted in turquoise and from AF tissue in orange.

Figure S4: Relative to main Figure 3. (A) UMAP of scRNA‐seq human dataset (GSE199866) used as basis for bulk RNA‐seq deconvolution combining AFC and NPC transcriptome, split between one healthy and one degenerated intervertebral disc. Cellular states were annotated using literature consensus. (B) Proportion of the different cellular states present in healthy and IVDD condition. (C) Dendrogram plot showing hierarchical clustering of the 12159 genes in 14 modules after merging, a different colour represents each module. (D) Dendrogram and heatmap generated with the module eigengenes for the 14 modules identified. Significant association between modules and origin of the cells are highlighted below (Yng: young). p‐values were calculated with the Student asymptotic p‐value for correlation.

Figure S5: Relative to main Figure 4. NPC from young and aged ovine disc share common senescence and mitochondrial energetic responses to stress‐related stimuli. (A) IL6 and MMP3 transcript expression by RT‐qPCR of NPC from young and old animals following 24 h of IL‐1 β treatment or serum starvation. (B) Quantification of % of NPC from young and old animals positive for EdU incorporation and SA‐β‐Gal staining. (C) Quantification of maximal respiration and ATP production. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. p‐values were calculated using the Repeated Measures ANOVA test, and Tukey's post‐hoc test; N = 8 animals.

JCMM-30-e71026-s004.pdf (3.1MB, pdf)

Table S1: Differential Expression Results between NPC from young and old animals. Standard differential expression analysis results using DESeq2 pipeline. BaseMean = mean of normalised read count across all samples. log2 fold change = log2 Fold change between groups, positive indicates upregulated in NPC isolated from old animals, padj = p‐value adjusted for multiple testing using Benjamini–Hojberg correction.

JCMM-30-e71026-s001.xlsx (21.8KB, xlsx)

Table S2: Differential Expression Results between AFC from young and old animals. Standard differential expression analysis results using DESeq2 pipeline. BaseMean= mean of normalised read count across all samples. log2 fold change = log2 Fold change between groups, positive indicates upregulated in AFC isolated from old animals, padj = p‐value adjusted for multiple testing using Benjamini–Hojberg correction.

JCMM-30-e71026-s006.xlsx (34.8KB, xlsx)

Table S3: Differential Expression Results between NPC and AFC isolated from young animals. Standard differential expression analysis results using DESeq2 pipeline. BaseMean = mean of normalised read count across all samples. log2 fold change = log2 Fold change between groups, positive indicates upregulated in NPC, padj = p‐value adjusted for multiple testing using Benjamini–Hojberg correction.

JCMM-30-e71026-s003.xlsx (22.2KB, xlsx)

Table S4: Differential Expression Results between NPC and AFC isolated from old animals. Standard differential expression analysis results using DESeq2 pipeline. BaseMean = mean of normalised read count across all samples. log2 fold change = log2 Fold change between groups, positive indicates upregulated in NPC, padj = p‐value adjusted for multiple testing using Benjamini–Hojberg correction.

JCMM-30-e71026-s002.xlsx (76.9KB, xlsx)

Table S5: List of the top‐30 genes associated with WGCNA modules. List of the 30 genes most associated with the different modules identified by WGCNA, ranked by importance in the module. The modules included are the modules associated with AFC (Yellow and Magenta modules), with NPC (Tan and Brown modules), with young and (Cyan and Green modules) and with old age (Pink module).

JCMM-30-e71026-s005.xlsx (12KB, xlsx)

Data Availability Statement

The raw and processed transcriptomic data that support the findings of this study have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE302551. The complete bioanalysis pipeline is available at https://gitlab.univ‐nantes.fr/guihomics/ovine_IVD_bulkRNAseq. Other raw data and materials are available from the corresponding author on reasonable request.

Articles from Journal of Cellular and Molecular Medicine are provided here courtesy of Blackwell Publishing

RESOURCES