Development of Intervertebral Disc Organoids through Directed Differentiation of Mesenchymal Stem Cells and Hierarchical 3D Printing

Abstract

Organoid models of early tissue development have been generated for organs such as the brain, kidney, and intestine. However, the development of intervertebral disc (IVD) organoids has rarely been reported. Here, we have developed a protocol to directly differentiate nucleus pulposus (NP) organoids and annulus fibrosus (AF) organoids from human mesenchymal stem cells (hMSCs) using differentiation media supplemented with folic acid (FA) and connective tissue growth factor in a 3D environment. Our findings suggested that FA could potentially promote NP cell generation through modulation of the PI3K-AKT and TGF-β pathways. Utilizing digital light processing 3D printing techniques, we hierarchically constructed biomimetic IVD scaffolds consisting of customized host–guest and silk fibroin hydrogels for NP and AF organoids. Subsequently, we manually assembled IVD scaffolds, NP organoids and AF organoids into complex IVD organoids (IVDOs) exhibiting a central NP-like region surrounded by concentric AF-like structures. Furthermore, the implantation of IVDOs into the goat lumbar spine after discectomy results in anisotropic reconstruction of the IVD. These findings highlight the successful establishment of an in vitro IVD organoid model for future research of disc degenerative diseases and demonstrate a translational therapeutic approach for IVD repair.

1. Introduction

Organoids are three-dimensional structures composed of multiple cell types, generated through self-organization by stem cells, and capable of mimicking the structure and function of natural organs. Organoids can be used for both in vivo and in vitro studies and represent one of the latest innovations in the search for models that replicate the physiological processes of an entire organism. Recent advancements in organoid technology and three-dimensional (3D) printing have led to the emergence of pathways for the fabrication of autologous human tissues, which hold significant promise for applications in drug screening and personalized therapy. The ability to engineer organoids via self-assembly utilizing aggregates of stem cells represents a bottom-up strategy that enables the recapitulation of complex organ-specific microarchitectures. However, current protocols for organoid generation, encompassing cerebral, renal, and retinal tissue, typically require extensive periods of time to achieve cellular diversity and size akin to that of native organs. , In contrast, 3D printing can address these issues. Digital light processing (DLP) 3D printing, for instance, utilizes light polymerization to create model layer by layer. This method offers the advantages of high printing speed and high resolution of printed structures (∼40 μm), regardless of the complexity of the layers. ,

In DLP 3D printing, inks need to satisfy several essential criteria, such as biocompatibility, printability, and biomimicry to ensure structural stability. Traditional hydrogel inks, such as hyaluronic acid methacryloyl (HAMA) and gelatin methacryloyl (GelMA), are typically composed of single covalent networks. To align with the mechanical properties of the intervertebral disc (IVD) tissue, these hydrogels typically rely on increasing the density of covalent cross-linking networks to enhance their mechanical performance, which limits cell-matrix and cell–cell interactions, thereby impeding related cellular and tissue functions. , In contrast to GelMA, the reversible cross-links serve as sacrificial bonds to absorb energy when the material undergoes stress, thereby enhancing the material’s toughness without increasing the density of covalent cross-links. Host–guest (HG) cross-linking has been extensively studied among various dynamic cross-linking systems due to its self-healing capabilities, cellular adaptability, and sustained drug release effects.

The IVD is a complex structure comprising the nucleus pulposus (NP), a central hydrated gelatinous core; the annulus fibrosus (AF), a ring-like fibrous laminar structure; and cartilaginous end plates adjacent to vertebral bodies, which together synthesize extracellular matrix (ECM) proteins critical for disc integrity. In terms of the microarchitecture of IVD, the native AF is reinforced by collagen nanofibers, which are arranged at an angle of approximately 30°-45° to the horizontal layer. This type of organizational structure forms a lamellar architecture, which is crucial for the biochemical and biomechanical functions of the AF. The gelatinous NP is located within the collagen fibers that form the AF and is constrained by them. Although organoid models for various tissues have advanced, progress in reconstruction of the IVD organoid has been limited. Research has primarily focused on using differentiated or primary IVD cells to simulate adult IVD tissue. Chen et al. constructed GelMA microspheres with independently adjustable elastic modulus and ligand density, which support the NP-like differentiation of IVD progenitor cells by modulating Yes-associated protein transfer. In terms of AF differentiation, a study employed directional freezing technology to develop a silk protein-based, multilayered IVD, consisting of concentric lamellar sheets. The artificial IVD supported the proliferation and differentiation of native AF cells or hMSCs, and deposited a sufficient amount of ECM. However, these studies are able to reproduce tissues with some aspects of IVD tissue but do not reproduce the integral IVD using stem cells.

IVD degeneration is a primary cause of low back pain, affecting 619 million (95% CI: 554–694 million) individuals globally in 2020, with an age-standardized years lived with disability rate of 832 per 100,000 population (95% CI: 578–1070). Because of the avascular and nonregenerative nature of IVD, current surgical interventions with synthetic polymers or metallic implants fail to achieve structural and biomechanical reconstruction of IVD and often lead to long-term problems of wear and failure. This therapeutic gap underscores the critical need for biological reconstruction strategies. Organoids are promising therapeutic agents for achieving anisotropic reconstruction of the IVD and stimulating regeneration of degenerated IVDs.

Here, we report a method for directing the differentiation of hMSCs into IVD organoids (IVDOs) based on DLP 3D printing ( Figure ). hMSC spheroids were embedded in hydrogels and directed using differentiation medium containing the NP induction factor folic acid (FA) and the AF induction factor connective tissue growth factor (CTGF). Hydrogel with NP/AF like stiffness and niche factors drive the deposition of compartmentalized ECM and arrangement of collagen fibers, and ultimately leading to the construction of NP organoids (NPO) and AF organoids (AFO). Whole-transcriptome sequencing revealed the role of FA in NP organogenesis and led to the identification of the characteristic molecular signatures of NPOs derived from hMSCs. We designed a double-network hydrogel for NP organoid generation, which is composed of β-cyclodextrin (β-CD)-modified HAMA (HAMA-β-CD) and FA-modified GelMA (GelMA-FA). We chose silk fibroin methacryloyl (SilMA) hydrogel for AF organoid generation. Using DLP 3D printing technology, we simulated the structure of the IVD and facilitated the secretion of different collagens from embedded stem cells regulated by FA and CTGF in the localized microenvironment, resulting in the successful formation of IVDOs. Implanting these IVDOs in a goat model of IVD degeneration demonstrated their primary therapeutic effects and potential for clinical translation.

1.

Establishment of NP and AF organoids and the mechanism of FA-induced NP differentiation. (a) Schematic protocol for differentiating NPOs and AFOs from hMSCs. (b) qPCR showing the mRNA levels (fold changes relative to those in 3D hMSC; same in (i), (l) of COL2A1, ACAN, CD24, NCAM1, and PAX1. (c) Volcano plot of 3D hMSC+FA vs 3D hMSC. GSEA analysis of 3D hMSC+FA vs 3D hMSC. Plots are related to (d) DNA replication and (e) the cell cycle. (f) KEGG enrichment analysis of 3D hMSC+FA vs 3D hMSC. (g) Differentially expressed genes (DEGs) associated with NP differentiation. (h) GO biological process enrichment analysis of 3D hMSC+FA vs 3D hMSC. (i) qPCR showing the mRNA levels of TIE2 and B4GALNT1 in 3D hMSC+FA and 3D hMSC groups. (j, k) Western blot analysis of SMAD3, p-SMAD3, COL2, KRT19, CD24, COL1, TNMD, COL5, IBSP, and FBLN1 in 3D hMSC and 3D hMSC+FA groups. (l) qPCR showing the mRNA levels of COL1A1, COL5A1, IBSP, BGLAP, and FBLN1 in 3D hMSC+FA and 3D hMSC groups. The data are expressed as the mean ± SD *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, by Student’s t test (i, j, k) and one-way ANOVA (b, l). n = 3 independent experiments per group.

2. Results

2.1. Generation of an IVD Organoid Model from hMSCs

Figure a shows an optimized compound-based protocol for directing hMSCs toward NP organoids (NPOs) and AF organoids (AFOs) by culturing hMSC spheroids (Figure S1a, b) in NP differentiation medium and AF differentiation medium respectively for 21 days. , The NPO that play a central role in IVD organoid (IVDO) generation, was first analyzed. During the differentiation process of NPO, the regulation of TGF-β pathway is crucial, which is supported by FA-induced shifts toward mitochondrial metabolism, enhancing TGF-β signaling. We included FA in the NP differentiation medium to induce the differentiation of hMSCs into NP for a period of 21 days. And established four groups and conducted qRT-PCR testing after 21 days of culturing. The 3D hMSC group represents hMSC cell spheroids cultured in 3D environments using DMEM/F12 medium. The 3D hMSC+FA group includes the addition of 30 μM concentration of folic acid based on the previous group. The NPO group represents NP organoids generated through the differentiation method shown in Figure a. The NPC group represents nucleus pulposus cells (NPC) derived from human nucleus pulposus tissue. The expression of NP-specific genes, , including KRT19, CD24, ACAN, NCAM1, and COL2A1, was upregulated in NPO group and 3D hMSC+FA group compared to 3D hMSC (Figure b). The expression of NP-specific genes in the NPO group is close to that of NPCs, demonstrating the similarity between NPOs and NPCs (Figure b). In young adults, NP tissue has an aggrecan/collagen II ratio of 27:1, serving as a potential biomarker for discrimination between NP cells and chondrocytes. Accordingly, we quantified the aggrecan/collagen II ratio in NPOs, chondrocytes, and NPCs. The results showed that the aggrecan/collagen II ratio of NPOs was not significantly different from that of natural NPCs (Figure S1c). Parallel qPCR analysis was performed on cell spheroids at different time points (day 0 [hMSCs], day 5 [notochordal stage], and day 14 [NP organoid stage]). The results demonstrated peak expression of the notochordal lineage marker TBXT at day 5. Upregulation of NP progenitor markers (B4GALNT1 and TIE2) was also observed by day 5 of differentiation. Significant elevation of mature NP markers (COL2A1 and ACAN) at day 14 confirmed the successful stepwise differentiation from hMSCs through notochordal-like to NP-like phenotypes (Figure S1d).

To investigate the mechanism underlying FA-induced NP differentiation, we conducted mRNA sequencing on 3D hMSC group (cultured using DMEM/F12 basal medium) and 3D hMSC+FA group (cultured using DMEM/F12 basal medium with 30 μM FA). The principal component analysis confirmed intragroup consistency and highlighted significant differences between the FA and control groups (Figure S2a). Figure S2b shows the differentially expressed genes (DEGs) between the two groups. The volcano plot displays DEGs that exhibited statistically significant differences (Figure c). Gene set enrichment analysis (GSEA) indicates that the genes upregulated in the FA group were associated mainly with DNA replication (Figure d) and the cell cycle (Figure e).

In the FA group, characteristic genes linked to NP development were notably upregulated. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis revealed signaling pathways associated with NP cell differentiation, including the ECM-receptor interaction, focal adhesion, PI3K-Akt, and TGF-β signaling pathways (Figure f). In the FA group, the downstream targets of the TGF-β pathway, such as BMP6, SMAD3, FBN1 and RHOA, was markedly upregulated in FA-treated cells, confirming the efficacy of our protocol (Figure g). Gene Ontology analysis under FA conditions highlighted genes related to cell differentiation (Figure h), with primary NP differentiation genes, including COL2A1, SOX9, VIM, CD24 and KRT19 (Figure i), and secondary related genes (Figure S2c), being significantly upregulated. AF-specific genes and inflammatory markers (Figure S2d) were downregulated in the FA group.

The secretion of ECM proteins is one of the core functions of NPOs. Overall, compared to the control group, 71 out of 174 core NP ECM genes were upregulated in the FA group, and the most significant DEGs are shown in Figure S2e. Among these were 9 collagen genes, namely, COL2A1, COL4A1/2/4/6, COL5A2, COL9A3, COL17A1, and COL28A1. Some of these genes, such as SMAD3 and LOXL2, are also characteristic of chondrocytes. In addition, the genes encoding glycoproteins, SPARC, RELN, EMILIN2, THBS3, and laminin (LAMA4/LAMB4), were upregulated. The upregulated genes included integrin genes such as ITGA5/7/8, which interact with the ECM, as well as ECM remodeling factors such as MMP3/8/10/20.

We subsequently examined several key genes that mark the early differentiation of the NP by qPCR and found that cells treated with FA expressed markedly more of the NP progenitor surface genes GD2 and B4GALNT1 than did the control cells (Figure i). These differentiated hMSCs also exhibited upregulation of NP-specific proteins, including KRT19, COL2A1, and CD24 (Figure j). The phosphorylation of SMAD3 is crucial for downstream of TGF signaling pathway. As shown in Figure j and Figure S2f, after exposure of hMSCs to folic acid, the phosphorylation level of Smad3 was significantly increased. Importantly, pharmacological inhibition of the TGF-β receptor with SB431542 abolished the FA-induced enhancement of Smad3 phosphorylation (Figure S2g), demonstrating that folic acid drives nucleus pulposus-like differentiation through activation of the TGF-β signaling pathway. AF-specific proteins, including COL1A1, TNMD, COL5A1, IBSP, and FBLN1, were downregulated (Figure k). We established four groups and conducted qRT-PCR testing after 21 days of AF differentiation. Connective tissue growth factor (CTGF) directs fibroblast differentiation from hMSCs and defines connective tissue healing. The 3D hMSC+CTGF group includes the addition of 100 ng/mL CTGF based on the 3D hMSC group. The AFO group represents AF organoids generated through the differentiation method shown in Figure a. The AFC group represents annulus fibrosus cells (AFC) derived from human annulus fibrosus tissue. Notably, hMSCs treated with AF differentiation medium containing CTGF presented upregulated expression of AF-specific genes, , including TNMD, FBLN1, IBSP, COL1A1, and COL5A1, indicating the formation of AFOs (Figure l). We also included AF cells derived from native AF tissues in qRT-PCR experiments to further demonstrate the similarity between AF organoids and native AF tissues (Figure l).

2.2. Characterization of HG and SilMA Hydrogels and Their Support for 3D Differentiation of NP and AF Organoids

The design of hydrogels for IVDOs is illustrated in Figure a, which shows the formation of hydrogels. We used host–guest (HG) hydrogel to support NP organoids and SilMA hydrogel to support AF organoids. The HG hydrogel is composed of folic acid-conjugated GelMA (GelMA-FA) and hyaluronic acid methacrylate conjugated with β-cyclodextrin (HAMA-β-CD). Folic acid (FA) guides hMSC differentiation toward NP cells. The FA motif peak in the ¹H NMR spectrum of GelMA-FA indicates the conjugation of the FA group to GelMA (Figure S3a). And the degree of substitution (DS) of GelMA is 21.5% according to the eq . The β-CD motif peak in the ¹H NMR spectrum of HAMA-β-CD indicates the conjugation of the β-CD group to HAMA (Figure S3b). The carboxyl group substitution rate of HAMA is 15.2% according to the eq . The HG hydrogel is composed of GelMA-FA and HAMA-β-CD, β-CD and FA groups serve as sacrificial linkages to form host–guest complexes, , providing a polymer network that enables cell migration (Figure b, c). β-CD forms a host–guest complex with FA, enabling the long-term sustained release of FA (Figure d). As shown in Figure d, while GelMA degraded completely within 1 week, the HG hydrogel exhibited a prolonged degradation process lasting up to 40 days. The acute injury response (up to 2 to 4-weeks post injury) is a critical, tissue-dependent time window where essential growth factors and cells coordinate to drive tissue repair. And disc height loss in response to acute injury is typically observed 4 weeks after injury and stabilized after 6 weeks in preclinical models. , The HG hydrogel demonstrated a degradation rate compatible with IVD repair (Figure e) and exhibited excellent antiswelling properties compared with GelMA (Figure f).

2.

The characterization of hydrogels and the differentiation of NP organoids and AF organoids in HG and SilMA hydrogels. (a) A schematic illustration showing the components of HG hydrogels for NP organoids and SilMA hydrogels for AF organoids. (b) Microscopic images of scratch lines at 0, 6, 12, and 24 h. Scale bars: 50 μm. (c) Quantitative assessment of wound healing percentage to evaluate cell migration. (d) Cumulative release of folic acid in vitro from GelMA and the HG hydrogel. (e) Degradation characteristic curves of the HG and GelMA hydrogels in vitro. (f) Swelling ratio of the HG and GelMA hydrogels in vitro. (g) compressive stress–strain relationships and (h) compression modulus of the GelMA, SilMA, SilMA with 4% PEGDA, HG hydrogel and HG hydrogel with 4% PEGDA. (i) Representative fluorescent images of NPO and AFO differentiated in HG and SilMA hydrogel showing markers for NP (COL2, ACAN, CD24, and NCAM1), AF (COL1 and IBSP). Scale bars: 50 μm. (j) UMAP visualization showing the single-nucleus transcriptomes from hMSCs (n = 3951) and IVD organoid (n = 2709). Cells are colored by sample identity (k) UMAP visualization of the four IVD subclusters defined during human early IVD organoid formation. (l) Proportion of cell clusters in the IVD organoid. (m) Pseudotime trajectory axis revealing the progression of IVD organopid. (n) Heatmap revealing the scaled expression of differentially expressed genes for each IVD d subcluster. (o) Representation analysis of GO categories showing different functions for the NP/NC progenitor and iAF clusters. The data are expressed as the mean ± SD *p < 0.05, **p < 0.01, ***P < 0.001, ****P < 0.0001; by one-way ANOVA (c, h) and two-tailed unpaired t test (f). n = 3 independent experiments per group.

SilMA has the effect of promoting hMSCs differentiation into AF, therefore we used 8% w/t SilMA to create AFOs. Regarding the mechanical testing of hydrogels, we conducted compression tests on GelMA, SilMA, HG, SilMA+PEGDA, and HG+PEGDA hydrogels. The results indicated that GelMA, SilMA, and HG hydrogels did not fracture under 60% strain, demonstrating good deformability (Figure g). Among them, GelMA and SilMA hydrogels exhibited characteristics of low modulus. The enhanced mechanical properties of the HG hydrogel can be attributed to the formation of interpenetrating HG networks and covalent networks. This synergistically enhanced its modulus and strength, significantly surpassing the performance of the GelMA and SilMA hydrogel networks (compression modulus of HG hydrogel: 20.8 kPa, which is 3.4 times that of GelMA and 4.0 times that of SilMA, Figure h). The incorporation of PEGDA not only enhanced the 3D printability of the hydrogels but also significantly increased the compressive modulus of both HG hydrogel and SilMA. Alternatively, we conducted cyclic compression tests on HG hydrogel with a maximum compressive strain of 20% to evaluate its mechanical properties under cyclic loads. As shown in Figure S4, after the first cycle, the subsequent nine stress–strain curves almost remain overlapped, indicating minimal energy dissipation (Figure S4a). Significant energy dissipation of the HG hydrogel was observed during continuous 100 cycles of compression, demonstrating its energy absorption capability (Figure S4b).

To investigate the development of organoids in HG and SilMA hydrogels, we performed immunofluorescence staining at day 21 (Figure i, Figure S4c). The results demonstrated that the NPO core exhibited high expression of NP markers (COL2, ACAN, CD24, NCAM1 and KRT19), while the AFO displayed a ring structure with elevated expression of COL1 and IBSP. (Figure i). They exhibit characteristics of organoids, such as possessing cellular composition mimicking native NP and AF tissues and self-organization capacity to form microstructures recapitulating IVD architecture collectively indicating successful organoid maturation. ,

To further validate the functional mimicry and maturity, we subjected the NPOs to physiologically relevant cyclic mechanical compression for 7 days. This biomechanical stimulation significantly upregulated the expression of key NP-specific ECM and marker genes (COL2A1, ACAN, CD24, NCAM1, and PAX1) compared to static controls, confirming their active and mechano-responsive matrix remodeling capacity (Figure S4d). Furthermore, the metabolic profile of the NPOs exhibited characteristics of native NP tissue, evidenced by a lactate production-to-glucose consumption ratio comparable to that of primary human NP cells (Figure S4e). Time-course immunofluorescence analysis from day 7 to day 21 revealed the progressive, zone-specific deposition and organization of key ECM proteins (COL2 and ACAN in the NP region and COL1 in the AF region) confirming dynamic and structured matrix synthesis over time (Figure S 4f). Extended culture to day 56 confirmed the long-term stability of the IVDO model, with maintained structural integrity in both NP and AF regions. Immunofluorescence analysis at this end point demonstrated sustained ECM marker expression and further ECM maturation (Figure S 4f), establishing its phenotypic stability and suitability for prolonged in vitro study.

To systematically evaluate the cellular composition and differentiation efficiency of IVD organoids, we performed single-nucleus RNA sequencing (snRNA-seq) at day 14 of the induction protocol using the 10x Genomics Chromium system. For benchmarking, we included snRNA-seq data from IVD organoid sample (2709 cells) and primary hMSC sample (3951 cells) (Figure j, Figure S5a). Unsupervised clustering and Uniform Manifold Approximation and Projection (UMAP) analysis of the IVD organoid data identified four transcriptionally distinct cell populations, which closely recapitulate the native IVD: NP/Notochordal Progenitors (NP/NC Progenitors, 227 cells, 8.48%), NP Notochordal cells (NP/NC, 1801 cells, 67.28%), Inner AF (iAF, 157 cells, 5.86%), and Outer AF (oAF, 492 cells, 18.38%) (Figure k,l).

The NP/NC Progenitor cluster was characterized by the expression of early discogenic markers (COL2A1, VCAN, PIEZO2) (Figure m, Figure S5b). The NP/NC cluster exhibited high expression of classic notochordal markers, including GDF5, and APOE, confirming the successful commitment of hMSCs toward a specialized notochordal lineage (Figure m, Figure S5c). The fibrocartilaginous components of the organoid were clearly bifurcated into Inner AF and Outer AF populations. The Inner AF cluster showed robust enrichment of chondrogenic signatures (COL5A2, COL6A3, and CDH11), whereas the Outer AF cluster was distinguished by a predominant fibroblastic profile, characterized by high levels of COL1A1, ACTA2, and TAGLN (Figure m, Figure S5d). , The coexistence of these four populations within a single organoid system demonstrates that our 3D culture environment provides the necessary biochemical and mechanical cues to support the spatial heterogeneity and multilineage specification required for functional IVD morphogenesis.

Pseudotime analysis revealed a convergent differentiation trajectory. The NP/NC progenitor cells were predominantly located at the starting point, which subsequently bifurcated into three distinct branches: the NP/NC branch, the iAF branch, and the oAF branch (Figure n). This recapitulates developmental patterning and indicates intrinsic self-organization beyond passive encapsulation. Functional enrichment analysis of marker genes revealed distinct biological programs among the major cell populations within the IVD organoids (Figure o, Figure S5e–j). GO enrichment analysis revealed that the NP/NC progenitor and iAF subpopulations are functionally involved in regulating specialized extracellular matrix, collagen fibril organization and connective tissue development (Figure o). These functional annotations confirm that the organoids are executing disc-specific transcriptional programs indicative of active tissue maturation, beyond mere cell viability within the hydrogel.

2.3. Fabrication and Characterization of Bioink and Hierarchical 3D-Printed Scaffolds for IVDOs

In DLP 3D printing, printability significantly affects the mechanical and biological properties of the printed structure. Rheology and photopolymerization are crucial parameters for assessing printability. For the ink used in 3D printing, we added 4% w/t PEGDA as a plasticizer and 1% w/t carboxymethyl cellulose sodium (CMCNa) as a water retention agent, based on the HG and SilMA hydrogels, to ensure the subsequent 3D printing (Figure a). PEGDA has been widely used for IVD repair, and demonstrated excellent cell-compatibility The introduction of PEGDA influences hydrogen bond formation, resulting in a decrease of 3 °C in the gel–sol transition temperature of the HG hydrogel, which is more conducive to 3D printing (Figure b). Next, we quantified the gelation kinetics by observing the storage modulus under 405 nm light, based on time-sweep rheological measurements (Figure c). Upon exposure to 405 nm light, all inks immediately solidified, with G′ exceeding G′′ and sharply increasing before stabilizing. The HG hydrogels exhibited a shorter gelation time and a higher final G′ value after gelation, indicating stronger mechanical stability. The incorporation of PEGDA significantly reduced the photocuring time, thereby facilitating the 3D printing process. This confirms that SilMA and HG hydrogels demonstrate excellent photopolymerization performance, which is crucial for printing. To assess the compatibility after printing, live/dead cell staining and cell proliferation tests were conducted on hMSCs embedded in HG or SilMA hydrogels. The hydrogels exhibited high cell viability at 1, 3, 7, and 14 days after photopolymerization (Figure S6a–d). The HG hydrogel displayed better cell proliferation characteristics than the GelMA hydrogel (Figure d). Coculture of HG and SilMA hydrogels with macrophages was then performed. Immunofluorescence analysis (Figure S6e) revealed that the HG hydrogel effectively promoted macrophage polarization toward the anti-inflammatory M2 phenotype, as indicated by the upregulation of the M2 marker CD206. This result suggests that the hydrogel possesses intrinsic immunomodulatory properties, which may contribute to establishing a pro-regenerative rather than pro-inflammatory microenvironment. The underlying mechanism may involve the sustained release of folic acid from the HG hydrogel, which exerts anti-inflammatory effects.

3.

Inks and hierarchical 3D-printed scaffolds for IVDOs. (a) Ink NP and AF formed by HG and SilMA hydrogels supplemented with PEGDA, CMCNa, and LAP. HG and SilMA hydrogel (with or without 4% PEGDA) kinetics with (b) increasing temperature and (c) exposure time to 405 nm light. (d) The cell proliferation activity in the hydrogel evaluated by CCK8 assay. (e) Schematic overview of the design and manufacturing process for the IVD scaffold. The shape parameters and external morphology are obtained from MR images. Biomimetic design is conducted based on the internal features of the natural IVD. (f) Engineering models and (g) photographs of 3D-printed IVD scaffolds from different views. Scale bars: 5 mm. The data are expressed as the mean ± SD **p < 0.01, by one-way ANOVA. n = 3 independent experiments per group.

The design strategy and manufacturing processes for IVDO scaffolds depend on the detailed imaging and structural analysis. Figure e illustrates the hierarchical 3D printing process. Magnetic resonance (MR) imaging has been utilized in the design of IVD scaffolds with native dimensions (Figure S7a). The outer boundaries of the IVD and the NP dimensions were determined via MR images. We designed the AF scaffold based on the natural structure of the IVD in which the fibers cross each other approximately 30°-45° to the horizontal layer (Figure f). MSLattices have been proposed for engineering applications, such as tissue engineering scaffolds and in vivo implants. Based on the MSlattices, we designed the NP scaffold and combined it with the AF scaffold to form a complete IVD scaffold (Figure f, g, Figure S7b). Scanning electron microscopy (SEM) images provided a microscale view of the IVD scaffolds (Figure S7c). We determined through SEM that the thickness of each layer of the external AF scaffold is approximately 150 μm.

2.4. 3D-Printed IVD Scaffolds and Cell-Embedded Hydrogels Allow for IVD Organoid Maturation

Regarding the manufacturing process of 3D-printed IVDOs. First, NP differentiation factors (N2, B27, ITS, NEAA, 1% penicillin/streptomycin, 50 μg/mL ascorbic acid-2-phosphate, 40 μg/mL l-proline, 10 nM dexamethasone, 10 ng/mL TGF-β3, 30 μM FA) and AF differentiation factors (N2, B27, ITS, NEAA, 2 mM l-glutamine, 0.1 mM 2-mercaptoethanol, 50 μg/mL ascorbic acid-2-phosphate, 10 ng/mL TGFβ1, 100 nM SAG, 100 ng/mL CTGF, and 100 nM AGN 193109) are added to HG and SilMA hydrogels. Subsequently, on day 6 preliminarily differentiated NP-like and AF-like spheroids are mixed with HG or SilMA hydrogels and then added to the NP and AF regions of the IVD scaffold using a 1 mL Luer-Lok syringe, followed by UV (20s, 405 nm, 20 mW cm-2) irradiation to provide biochemical and biomechanical stimulation (Figure S8). Following 7 days of further differentiation and integration, a complete IVDO is formed. At 28 days (Figure a, Figure S9), hematoxylin-eosin (H&E) staining exhibited a central NP-like region surrounded by concentric AF-like structures. Safranin O & Fast Green images demonstrated that the IVDO group exhibited significantly stronger Safranin O staining (indicating higher GAG content) in NP regions and greater Fast Green affinity (suggesting collagen-rich composition) in AF areas compared to other groups. And the expression of collagen was the highest in the organoid group, as reflected by the Masson staining images. The organoid group also exhibited more COL2- and ACAN-positive staining in the NP region and more COL1-positive staining in the AF region (Figure c, d). These results confirm the successful generation of IVDOs.

4.

Hierarchical 3D-printed IVDOs recapitulate the NP and AF structure. (a) Histological images of H&E, Masson, and safranin O/fast green staining. (b) Immunofluorescence staining results for COL II, COL I, and ACAN demonstrate that the HG and SilMA hydrogels and the NP and AF differentiation media had synergistic effects on the zone-specific matrix secretion of IVD organoids at 28 days. (c) Semiquantitative fluorescence intensity of COL2, COL1 and ACAN. *p < 0.05, **p < 0.01, two-tailed unpaired t test. n = 3 independent experiments per group.

2.5. Implantation of hMSC-Derived IVDOs Attenuates Disc Degeneration in a Goat Model

We performed lumbar discectomy on goats, followed by implantation of IVDOs and followed them up for 6 weeks, to evaluate the therapeutic efficacy of IVDO therapies, ( Figure a). With the end goal of applying this technique to humans, we selected the goats as the animal model due to their similar intradiscal pressure, range of motion and dimensions to compared to humans. A total of 12 goats were randomly divided into four groups: an intact control group (three goats) that received no discectomy or treatment and goats that underwent discectomy and were subsequently left untreated, treated with our 3D-printed IVD scaffold without cells, or treated with IVDOs. The organoids transplanted into the goat IVD were clearly visible following discectomy via a posterolateral surgical approach (Figure b). During the six-week experimental period, veterinary staff confirmed that no pain, neurological deficits, or mortality were observed in the experimental animals. All goats maintained full lumbar spine function (Movie S1). The stride length was evaluated through the open field walking test, and the results indicated that there were no significant differences in average stride length among the groups (Figure S10a). The organoid group exhibited compressive mechanical properties that were closer to the values of natural motion segments, with a significant increase in modulus relative to the discectomy group (P < 0.01) and scaffold group (P < 0.05) (Figure S10b, c).

5.

Transplantation of IVD organoids improved regenerative capacity in an in vivo goat lumbar spine model. (a) Schematic diagram illustrating the implantation process of IVD organoids. (b) Representative sagittal gross dissection images showing the location of discectomy and organoid transplantation in vivo. Scale bars: 1 cm. (c) Representative axial T2MR images and (d) CT image showing differences in the IVD across treatment groups. High-intensity regions in T2MR images correlate with areas of greater water content. Scale bars: 1 cm. (e) Pfirrmann grading of IVD degeneration and (f) disc height index from the T2MR images compared between groups; the bars denoted statistical significance. (g) H&E, Masson, and safranin O/fast green staining and (h) immunohistochemistry staining of IVDs from the goat lumbar spine following the indicated treatment. Higher-magnification images showing the direction of collagen alignment in the AF and chondrocyte-like cells in the NP region. qPCR showing the mRNA levels associated with (i) IVD regeneration and (j) inflammation. The data are expressed as the mean ± SD **p < 0.01, ***p < 0.001, by one-way ANOVA. n = 3 independent experiments per group.

The T2-weighted MR and CT images depict the IVDs, revealing the therapeutic effects of the IVDOs (Figure c,d; Figure S10d). T2MR revealed severe morphological damage in the IVDs after discectomy, characterized by increased heterogeneity and decreased hydration in the NP, with unclear boundaries between the AF and NP. Treatment with the IVD scaffolds appeared to partially alleviate the loss of T2MR signals. Significantly, the organoid group exhibited marked restoration of hydration and maintenance of the geometric shape of the IVDs, thereby resulting in a morphology most similar to the intact group. The quantitative analysis of MR and CT images further supported the above findings (Figure e), revealing no degeneration in intact IVDs with a Pfirrmann grade of 1. However, the discectomy group exhibited significant degeneration, with a grade of 4.7 ± 0.6. The scaffold group improved to grade 3.3 ± 0.6, indicating that the pure scaffold partially mitigated degeneration. The organoid group had a significantly lower Pfirrmann grade than the discectomy group did (grade 2.3 ± 0.6, P < 0.01). As observed through CT imaging, the Disc Height Index (DHI) also exhibited a similar trend (Figure f). DHI after discectomy was significantly lower than that in the intact group (P = 0.0005), while organoid therapy had DHI values similar to those of intact IVDs (P = 0.13 and P = 0.11). The 3D CT reconstruction revealed the formation of vertebral osteophytes adjacent to the IVD postdiscectomy, with associated end plate disorders. The scaffold treatment alleviated the formation of osteophytes, while the organoid treatment did not result in the formation of osteophytes adjacent to the IVD (Figure S10e).

Histological analysis was used to visualize the morphology and tissue status of the IVD (Figure g). The intact group exhibited a layered structure of AF and homogeneous NPs rich in proteoglycans. In contrast, the discectomy group exhibited distinct and disrupted AF and NP morphology, which was improved in the IVD scaffold group. The organoid group maintained healthy AF lamellae and NP morphology without displaying the loss of DHI. Immunohistochemistry analysis confirmed the morphological differences observed through histology (Figure h, Figure S10f). In the intact group, the anabolism markers in the NP (COL2 and ACAN) region exhibited positive staining. The discectomy group exhibited decreased expression of COL2 and ACAN, and increased expression of COL1, which was improved by the use of IVD scaffolds and further improved by the use of organoids. RT–PCR analysis demonstrated that organoid implantation promoted IVD repair and alleviated the inflammatory response (Figure i,j). Immunofluorescence analysis showed no significant increase in CD68⁺ or CD3⁺ inflammatory cells or in IL-6 expression at the organoid implantation site and adjacent cartilage end plate/subchondral bone regions compared to native tissue (Figure S11). Furthermore, with the exception of the discectomy group, which exhibited increased IL-6 secretion from the cartilage end plate and a reduction in cartilage height, the cartilage end plate and subchondral bone adjacent to the implant in all other groups maintained structural integrity without signs of adverse immune reactions.

These clinically relevant outcomes demonstrated that IVDO implantation improved postinjury disc degeneration within 6 weeks. The untreated discectomy group was treated with the current standard interventions for disc herniation, which contributed to further disc degeneration. Signs of degeneration were less pronounced in all treatment groups than in the discectomy group, with the organoid group showing the best preservation of native IVD morphology.

3. Discussion

We developed a biomaterial platform based on DLP 3D printing to generate high-throughput IVDOs from hMSCs. The organoids feature a central NP-like core encased by a peripheral AF-like region, effectively mimicking the IVD structure. We developed a protocol based on biochemical stimulation and defined the optimal conditions for in vitro induction of NP and AF organoids. Transcriptome analysis of hMSCs under the influence of FA provided a molecular basis for the FA-mediated regulation of NP organoid generation. These molecular features can be used to further guide the differentiation of NP organoids. We engineered a hydrogel system with enhanced cell adaptability by combining covalent bonds and supramolecular interactions. By combining DLP 3D printing technology, we hierarchically constructed a biomimetic heterogeneous structure that provides a tissue-specific microenvironment for IVDO generation and significantly shortens the culture period. Finally, the favorable treatment effect observed in the goat discectomy model suggests the potential for clinical translation of organoid implantation treatment strategies. In summary, the first IVDO we generated is currently the most complete human in vitro IVD model system, providing a new potential clinical treatment strategy for the regeneration and personalized repair of IVDs.

GelMA has become a representative hydrogel formulation and is used in various biomedical applications. Relative to GelMA hydrogel used previously for IVD repair, the HG hydrogel-based ink in this paper possesses reinforced mechanical strength, enables better maintenance of the DHI. The host–guest interaction of FA with β-CD provides a sustainable drug release for stem cell differentiation. And HG hydrogel is more conducive to cell adhesion and cell migration, and exhibit better viscosity at low temperatures, making it more suitable as an ink for printing IVD scaffolds. SilMA is a modified hydrogel based on natural fibrous protein produced by silkworms, which has been used in various biotechnological and biomedical applications, including wound dressings, vascular prostheses, and structural implants. , Numerous studies have demonstrated that SilMA can stimulate hMSCs to differentiate into fibroblasts or osteoblasts, which are commonly used for the repair of fibrous tissues such as tendons and bones. , Therefore, we have chosen SilMA to construct AF organoids.

Balancing biomechanical and biological performance in IVD repair is important. Our design of the HG hydrogel with a compressive modulus of approximately 50 kPa was predicated on the principle of biomechanical mimicry of the native NP. This target value is situated within the reported spectrum of human NP compressive modulus, which ranges from approximately 2 to 60 kPa. Furthermore, broader analyses suggest that for NP replacement constructs, a considerable tolerance exists in the Young’s modulus, spanning from near-native tissue levels (∼10 kPa) to values that mitigate stress shielding and end plate subsidence (∼5 MPa), thereby effectively preventing the abnormal inward bulging of the annulus fibrosus following NP removal. The in vivo efficacy of this mechanically matched construct was substantiated by our 6-week goat model. The organoid-implanted motion segments exhibited a compressive modulus significantly restored toward native levels, confirming successful mechanical integration and load-sharing. Concomitantly, MRI/CT imaging confirmed the stable retention of the implant without dislocation or extrusion, while radiographic and histological analyses demonstrated mitigated disc height loss and a reduction in associated degenerative changes, such as osteophyte formation. Collectively, these findings validate that the selected modulus of ∼ 50 kPa for our HG hydrogel not only replicates the native mechanical niche but also provides a conducive and stable environment for IVD regeneration, effectively supporting functional restoration without inducing adverse biomechanical sequelae.

3D printing techniques can be categorized into inkjet printing, extrusion printing, and light-assisted 3D printing, including DLP and laser-based printing. Inkjet 3D printing and extrusion 3D printing utilize nozzles to dispense ink, typically offering limited resolution that ranges from a few hundred micrometers to several millimeters. In contrast, DLP printers create models layer by layer through photopolymerization, allowing them to disregard layer complexity. Moreover, DLP 3D printing technology has advanced rapidly and is now capable of achieving high resolutions (∼2 μm) and fast printing speeds. , Research has already employed PEG-based DLP inks to fabricate porous 3D hydrogel scaffolds with internal structures resembling trabeculae. These scaffolds support cell adhesion and the formation of a monolayer structure that mimics bone-lining cells, demonstrating the broad application prospects of DLP 3D printing. The diameter of the human IVD can reach up to 50 mm, making it difficult to construct using self-assembled organoids. Additionally, the microstructure of the IVD is highly complex, making it unsuitable for inkjet 3D printing. Therefore, building IVD organoids through DLP 3D printing can achieve layer-by-layer construction of complex structures and solve the problem of too small size in self-assembled organoids.

Repairing IVD degeneration is challenging due to its lack of blood supply and limited nutrient exchange. Most cell-based therapies induce IVD regeneration through nutritional mechanisms, where implanted cells can only survive temporarily and secrete growth factors that stimulate resident progenitor cells. Due to the limited regenerative potential of resident cells, the repaired tissue exhibits inferior functionality compared to the undamaged IVD. The tissue engineering method based on the in vitro expansion of NP cells and AF cells can address these issues. However, the subsequent translational application is limited due to their restricted cell availability. On the other hand, small animal models have limitations due to differences in the cellular composition of IVDs and significant biomechanical disparities. Similar to observations in humans, ovine and canines rapidly lose notochordal cells early in life, making these models more suitable for preclinical validation. A study has confirmed the effectiveness of biomaterial repair strategy in sheep lumbar IVD degeneration model. However, in this study, cells were not administered in conjunction with the biomaterials. Delivering cells through biomaterial scaffolds for IVD regeneration offers the advantages of enhanced cell retention, structural support, and tissue integration. Since the IVDO is composed of IVD cells and IVD ECM, it provides an appropriate environment for these cells, which facilitates cell survival and tissue healing. Therefore, we designed a hydrogel system for IVDO, and induced the differentiation of MSCs to produce NPO and AFO in a 3D in vitro environment. By combining these with 3D-printed scaffolds, we created a transplantable IVDO, ultimately achieving the restoration of IVD structure and function.

The limitations of this study include an observation period of up to 6 weeks in animal experiments. A longer observation period is needed to confirm whether IVDO can replace IVD. The RNA-seq data presented in this article will provide new insights into the signals inducing NP cells, which will aid in the development of differentiation protocols for hMSC-derived NP cells. However, this article lacks information on the signals inducing AF cells, and further research in this area is needed. The differentiation method for generating IVDO from hMSCs in vitro can be further refined to obtain well-developed IVDOs in the future. Beyond biological refinement, the clinical translation of this technology demands thorough investigation of long-term safety, including biomechanical performance (e.g., dynamic loading, cyclic compression, and shear stress) to assess their fatigue resistance and long-term mechanical stability as implants.

It is also necessary to analyze the tumorigenicity and other safety issues of the IVDO implanted in the spine. Overall, further research is needed before clinical translation, but our results indicate that we have successfully generated the first IVDO, and that IVDO is a potential candidate construct for treating IVD degeneration.

4. Conclusion

In summary, this study successfully established the first comprehensive human intervertebral disc organoid (IVDO) model derived from mesenchymal stem cells (hMSCs). We developed a protocol utilizing folic acid (FA) and connective tissue growth factor (CTGF) to direct the differentiation of hMSCs into nucleus pulposus organoids (NPOs) and annulus fibrosus organoids (AFOs), respectively, within a 3D environment. Mechanistic studies revealed that FA promotes NP differentiation by modulating the PI3K-AKT and TGF-β signaling pathways. Furthermore, we engineered a biomimetic, hierarchically structured IVD scaffold using innovative host–guest hydrogels for NP and SilMA hydrogels for AF, fabricated via high-resolution DLP 3D printing. The synergistic integration of these biochemically and biomechanically tailored hydrogels with predifferentiated organoids enabled the generation of functional IVDOs exhibiting a central NP-like core surrounded by concentric AF-like lamellae.

Crucially, the implantation of these bioengineered IVDOs into a goat lumbar spine discectomy model demonstrated significant therapeutic efficacy. The IVDOs promoted anisotropic tissue reconstruction, restored disc height, reduced degeneration (as evidenced by improved Pfirrmann grades), enhanced extracellular matrix deposition (COL2, ACAN in NP; COL1 in AF), and mitigated inflammation compared to discectomy-only or acellular scaffold controls. This highlights the potential of IVDOs for structural and functional IVD regeneration.

While further optimization of differentiation protocols, long-term safety assessments, and investigation into AF-specific inductive signals are warranted, this study provides a robust platform for fundamental research on disc degeneration and a promising translational strategy for biological IVD repair. The IVDO model represents a significant advancement toward personalized therapies for disc degenerative disease.

5. Experimental Section

Ethics Approval

Bone marrow, NP, AF and cartilage end plate (CEP) samples from healthy volunteers complied with the protocols by The Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology (No. S347, No. S341). All participants gave informed consent. Protocols of animal experiments were approved by The Institutional Animal Care and Use Committee at Tongji Medical College, Huazhong University of Science and Technology (No. S3553). The study was performed in accordance with the Declaration of Helsinki.

Chemicals and Materials

Gelatin methacryloyl (EFL-GM-30, Engineering For Life, China), folic acid (Aladdin, China), carboxymethyl cellulose sodium (CMCNa, Aladdin, China), poly­(ethylene glycol) diacrylate (PEGDA, Aladdin, China; 400Mw), N-hydroxysuccinimide (NHS, Macklin, China), dimethyl sulfoxide (DMSO, Aladdin, China), Tartrazine (Aladdin, China), methacrylic anhydride (MA, Macklin, China), N-(3Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC, Aladdin, China), ethylenediamine-β-cyclodextrin (EDA-β-CD, Zhiyuan Biotechnology, China), hyaluronic acid methacryloyl (HAMA, 150 kDa, Engineering For Life, China), China), recombinant human connective tissue growth factor (CTGF, PeproTech, USA), recombinant human transforming growth factor-b3 (TGF-β3, PeproTech, USA), recombinant human transforming growth factor-b1 (TGF-β1, PeproTech, USA), lithium phenyl (2,4,6-trimethylbenzoyl) phosphinate (LAP, Engineering For Life, China), fetal bovine serum (Gibco, USA), 1% penicillin–streptomycin (15140122, Thermo Fisher), DMEM/F12 (Gibco, USA), 1 × PBS pH 7.4 (10010023, Thermo Fisher), Trypsin-EDTA (25200056, Thermo Fisher), SAG (566661, Sigma).

Cell Isolation and Culture

Bone marrow samples were obtained from the iliac crest of volunteers. hMSCs were isolated from human bone marrow using density gradient centrifugation and cultured in expansion medium (DMEM/F12 containing 1% penicillin-streptomycin and 15% fetal bovine serum). The multilineage differentiation potential of MSCs was evaluated by staining with Oil Red O and Alizarin Red after induction in the corresponding differentiation media (Cyagen).

Human NP, AF and CEP samples classified as grade I were obtained from volunteers who were diagnosed with idiopathic scoliosis, as previous described (Table S1). , NP, AF and CEP samples obtained via were immersed in phosphate-buffered saline (PBS; Biosharp, China) and minced into 1–3 mm³ fragments. The fragments were washed three times with PBS to remove residual debris, followed by digestion with 0.4% Type II collagenase (Invitrogen, USA) or Type I collagenase (Invitrogen, USA) at 37 °C for 4–6 h. The resulting cell suspension was centrifuged at 800 rpm for 5 min, washed three times with PBS, and resuspended in growth medium (DMEM/F12, 1:1; Gibco, USA) supplemented with 15% fetal bovine serum (FBS; Gibco, USA) and 1% penicillin-streptomycin (Gibco, USA). Cells were maintained at 37 °C in a 5% CO₂ atmosphere. The medium was first refreshed after 7 days and subsequently replaced every 3 days until cells reached >95% confluence. Cells were passaged at a 1:3–1:4 ratio. NP, AF and CEP in the second passage were regarded as normal state and used to perform experiments.

Synthesis of GelMA-FA

The interaction between GelMA (DS: 30%) and FA occurred in an EDC/NHS/DMSO solution. Briefly, FA (30 mg) and GelMA (2.5 g) was completely dissolved in DMSO (5 mL) on magnetic stirrer (37 °C, 2 h). Then, EDC (30 mg) and NHS (50 mg) were added and thoroughly mixed (37 °C, 12 h) and dialyzed for 5 days. Then freeze-dried for 3 days and collected GelMA-FA. The FA motif peak in the 1H NMR spectrum (δ=1.79 and δ=2.57) indicated the conjugation of the FA group to GelMA. In addition, the relatively weak peak at 7.16 ppm in the complex could be attributed to the resonance of aromatic protons of GelMA and FA. By using the reported FA extinction coefficient of 6230 M^–1 cm^–1 and monitoring the absorbance of GelMA-FA at 350 nm, the DS of FA binding to GelMA was calculated to be 21.5% according to the following equation:

$$\mathrm{DS\; of\; FA}=\frac{\mathrm{Mole\; of\; FA\; conjugated\; to\; GelMA}}{\mathrm{Total\; mole\; of\; lysine\; present\; in\; the\; GelMA}}\times 100\%$$ 1

Synthesis of β-Cyclodextrin (β-CD)-Modified Hyaluronic Acid Methacryloyl (HAMA)

The linkage between HAMA and EDA-β-CD occurred within EDC/NHS/HAMA in phosphate buffer solution (PBS). Briefly, 25 mg of NHS and 50 mg of EDC were fully dissolved in a solution of 25 mg of HAMA in 7.5 mL of 0.1 mol PBS (room temperature, 30 min). Then, 73 mg of β-CD-EDA in 10 mL of PBS was added with sufficient mixing. After the mixture was stirred and dialyzed for 5 days at room temperature. Then freeze-dried for 3 days and collected HAMA-β-CD. The β-CD motif peak in the ¹H NMR spectrum (H1 at δ= 5.0 ppm) indicated the conjugation of the β-CD group to HAMA. DS was calculated to be 15.2% according to eq . In this equation, A2 ppm was the area of the characteristic peak at 2 ppm chemical shift for HAMA in the ¹H NMR spectrum, and A5 ppm was the area of the characteristic peak at 5 ppm chemical shift for β-CD in the ¹H NMR spectrum.

$$\mathrm{DS}\mathrm{of}\mathit{\beta }\mathrm{‐CD}=\frac{\mathrm{A}5\mathrm{ppm}}{\mathrm{A}2\mathrm{ppm}}\times \frac{3}{7}\times 100\%$$ 2

Preparation of GelMA, HG, and SilMA Hydrogels

For the GelMA hydrogels, 8% w/t GelMA was dissolved in PBS. For the HG hydrogel, 4% w/t GelMA-FA and 4% w/t HAMA-β-CD were dissolved in PBS. For the SilMA hydrogels, 8% w/t SilMA was dissolved in PBS. For all the hydrogels, 0.5% w/t LAP was added and dissolved at 37 °C. Then, the solution was exposed to 20 mW cm^–2 405 nm UV radiation (30 s).

Cell Migration

To evaluate cell migration, hMSCs were seeded into a 24-well plate. When the cells reached 70–80% confluence, use a 200 μL pipet tip to create a straight scratch in the center of the well. The liquid in the wells was aspirated, and 8% GelMA, 8% SilMA, and HG hydrogels were cast into the 24-well plate. To eliminate the influence of cell proliferation, serum-free medium was used to replace complete medium. The wound healing process was observed at 0, 6, 12, and 24 h using a bright field microscope, and the scratch area was measured with ImageJ.

Hydrogel Swelling Tests

GelMA and HG hydrogels (2 mm thick, Φ 8 mm) were immersed in PBS (37 °C). The weight of the hydrogels (W0) was subsequently measured. The weights of the hydrogels (W12, W24, W48, and W72) were measured at different swelling durations (12, 24, 48, and 72 h). The swelling rate was calculated according to the following equation:

$$\mathrm{swelling\; ratio}=\frac{W_n-W_0}{W_0}\times 100\%$$ 3

Degradation of Hydrogels In Vitro

Circular samples (2 mm thick, Φ 8 mm) of GelMA and HG hydrogels were incubated in DMEM/F12 (37 °C) media containing 50 μM TBHP to simulate the inflammatory microenvironment of IVD degeneration. The degradation rate was calculated by thoroughly washing the hydrogel in deionized water, freeze-drying and weighing, and finally applying the following formula:

$$\mathrm{Degradation\; ratio}=\frac{W_{o,dry}-W_{t,dry}}{W_{o,dry}}\times 100\%$$ 4

where W_o,dry represents the initial dry weight of the sample, while W_t,dry denotes the dry weight at each incubation time point.

In Vitro Release of Folic Acid

The in vitro release of FA from GelMA and HG hydrogels was evaluated by immersing HG hydrogel samples (2 mm thick, Φ 8 mm) in PBS (2 mL, 37 °C). extract 100 μL of PBS for absorbance measurement, then add another 100 μL of PBS and return it to the incubator. Use NanoDrop to measure the absorbance of the sample at 288 nm, and calculate the concentration based on the FA standard curve dissolved in the same medium.

Mechanical Properties of Hydrogels and Motion Segments

Mechanical property tests were conducted using the EUT2000 electromechanical testing machine (Shenzhen Sansi Testing Co., Ltd., China) with reference to GB/T 1041–2008 standard at room temperature. The crosshead speed was set at 10 mm min^–1, and the tests involved compression of cylindrical hydrogel samples with a height of 8 mm and a diameter of 10 mm. For the cyclic compression tests, the gel sample was compressed to a strain of 20% and then unloaded back to 0. This cycle was repeated for 100 times. Each of the mechanical tests was repeated for three times with distinct samples. Motion segments were prepared for compression testing by carefully clearing the soft tissue adjacent to the vertebral bodies (with adjacent muscle and tendon left intact).

Differentiation of NPO and AFO through 3D Culture

For NPO differentiation, hMSCs were passaged to P3, and on day 0, wash the cells twice with 1 × PBS. Subsequently, incubate with trypsin-EDTA at 37 °C and 5% CO2 for 1 min. Then, the cells were collected and resuspend at a concentration of 1 × 10⁶ cells ml^–1 in expansion medium. Next, 200 μL of the cell suspension was added to each well of a 96-well ultralow attachment round-bottom microplate (7007, Corning). On day 1, encapsulate cell spheroids in GelMA hydrogel (10% GelMA and 0.25% w/t LAP) HG hydrogel (4% w/t GelMA-FA, 4 wt % HAMA-β-CD and 0.25% w/t LAP) or SilMA hydrogel (8% SilMA and 0.25% w/t LAP) using a 1 mL Luer-Lok syringe (605–002601, Winner, China) and a female Luer-threaded coupler. Dissolve the HG hydrogel in 1 × PBS and mix with the cell spheroids in the syringe, followed by deposition into a 12-well plate and exposed to UV irradiation (20s, 405 nm, 20 mW cm^–2). The expansion medium was replaced with DMEM/F12 medium supplemented with N2 Supplement (Cat. No. 17502001, from Thermo Fisher), B27 Supplement (Cat. No. 17504044, from Thermo Fisher), ITS Supplement (Cat. No. 41400045, from Thermo Fisher), 1% Non-Essential Amino Acids (NEAA, Cat. No. 11140050, from Thermo Fisher), 2 mM l-glutamine (Cat. No. 25030081, from Thermo Fisher), 0.1 mM 2-mercaptoethanol (Cat. No. 21985023, from Thermo Fisher), and 10 ng/mL Activin A (Cat. No. 120–14E, from PeproTech) for 2 days at step 1. Subsequently, additional cell growth factors were introduced, including 10 ng/mL of FGF2 (Catalog No. 100–18B, sourced from PeproTech), 50 ng/mL of Noggin (Catalog No. SRP4675, sourced from Sigma), 3 μM of CHIR99021 (Catalog No. SML1046, sourced from Sigma), and 10 μM of AGN193109 (Catalog No. SML2034, sourced from Sigma) for notochordal-like cell differentiation for 3 days in the second step. In step 3, the medium is changed to DMEM-HG supplemented with the following components for NP-like cell differentiation for 15 days: N2, B27, ITS, NEAA, 1% penicillin/streptomycin, 50 μg/mL ascorbic acid-2-phosphate (A8960, Sigma), 40 μg/mL l-proline (P0380, Sigma), 10 nM dexamethasone (D1756, Sigma), 10 ng/mL TGF-β3, 30 μM FA.

For AF organoid differentiation, on day 1, the expansion medium was replaced with DMEM/F12 medium supplemented with N2, B27, ITS, NEAA, 2 mM l-glutamine, 0.1 mM 2-mercaptoethanol, 0.5 mM ascorbic acid, 10 ng/mL Activin A, 500 nM LDN-193189, 1% penicillin/streptomycin, and ascorbic acid-2-phosphate for 2 days at step 1 to induce embryoid body formation. In step 2, 5 μM CHIR99021, 100 nM AGN 193109 was added to enhance paraxial mesoderm differentiation for additional 3 days. In Step 3, the medium was changed to DMEM-HG supplemented with N2, B27, ITS, NEAA, 2 mM l-glutamine, 0.1 mM 2-mercaptoethanol, 50 μg/mL ascorbic acid-2-phosphate, 10 ng/mL TGFβ1, 100 nM SAG, 100 ng/mL CTGF, and 100 nM AGN 193109, and the cells were further cultured for 15 days. On day 21 NPOs and AFOs were collected for for subsequent experiments.

Single-Nucleus RNA Sequencing (snRNA-seq) Library Preparation and Sequencing

Single-nucleus suspensions were prepared using a modified protocol based on the 10x Genomics standard. Briefly, flash-frozen tissue samples (30–50 mg) were transferred from −80 °C storage to ice-cold homogenization buffer (10 mM Tris-HCl, 10 mM NaCl, 3 mM MgCl2, 0.1% NP-40, 0.2 U/μL RNase inhibitor) and mechanically dissociated on ice using a chilled Dounce homogenizer (10 strokes with the loose pestle, followed by 10–15 strokes with the tight pestle). The homogenate was filtered through a 40-μm cell strainer and centrifuged at 500g for 5 min at 4 °C. The pellet was resuspended in wash/resuspension buffer (PBS containing 1.0% BSA and 0.2 U/μL RNase inhibitor). To remove debris and myelin contamination, the nuclear suspension was further purified by OptiPrep density gradient centrifugation or double filtration. Nuclear concentration and integrity were assessed using a Countess II FL Automated Cell Counter; only suspensions with >90% nuclear integrity were used for downstream library preparation. Libraries were constructed through standard steps of fragmentation, end repair, A-tailing, and adapter ligation, with quality evaluated on an Agilent 2100 Bioanalyzer. Finally, libraries were pooled and sequenced on an Illumina NovaSeq 6000 platform using a paired-end 150-bp (PE150) strategy.

Bioinformatic Analysis of snRNA-seq Data

Bioinformatic processing was performed using R (v4.5.0) primarily with the Seurat (v5.4.0) toolkit. Raw gene expression matrices were normalized using the LogNormalize method in Seurat. Quality control was applied by removing nuclei with a high percentage of mitochondrial reads and low gene complexity. To reduce technical noise, mitochondrial, ribosomal, immunoglobulin, and T-cell receptor genes, along with common technical artifacts defined in the SignatuR blocklists, were excluded prior to downstream analysis. Highly variable genes (HVGs) were then selected for dimensionality reduction. Principal component analysis (PCA) was performed, and the top principal components (PCs) capturing 80% of the cumulative variance were retained for downstream nonlinear dimensionality reduction using both Uniform Manifold Approximation and Projection (UMAP) and t-distributed Stochastic Neighbor Embedding (t-SNE). Unsupervised cell clustering was conducted using a shared nearest neighbor (SNN) graph-based algorithm in Seurat at a resolution of 0.5. Differential expression analysis between clusters was performed using the Wilcoxon rank-sum test via the presto package. Cell clusters were annotated as NP/NC Progenitor, NP/NC, iAF, and oAF based on the expression of established lineage-specific marker genes. For trajectory analysis, pseudotemporal ordering was inferred using Slingshot. Genes dynamically expressed along pseudotime within each lineage were identified by fitting generalized additive models (GAMs) with the tradeSeq package, and their significance was assessed using association and start-vs-end tests. The overall trajectory structure was independently validated using Monocle 3 to ensure robustness.

Rheological Testing of Inks

The rheological properties of HG hydrogels, with and without 4% PEGDA, were studied using a rheometer (HAAKE MARS, Thermo Scientific) before UV irradiation. The temperature sweep test was performed by placing an appropriate amount of ink onto a rheometer, cooling it to 4 °C to form a gel state, and recording the viscosity within the temperature range of 4–50 °C at a heating rate of 2 °C/min (σ = 20 Pa, f = 1 Hz). For the sweep test under 405 nm 20 mW cm^–2 light irradiation. The solution of HG and SilMA hydrogel was placed on the rheometer at 30 °C, which operated at 10 Hz and 1% strain. The light source was turned on at the 60 s and was maintained it until the storage modulus reached a stable state.

Fabrication of IVD Scaffolds

The ink solution was based on HG and SilMA hydrogels. We added 4% w/t PEGDA and 1% w/t CMCNa to ensure subsequent 3D printing, after which the solution was dissolved in PBS. Subsequently, 0.25% w/t LAP and 0.05% w/t tartrazine were added to the ink as photoinitiator and photoabsorber. A 3D printer (AUTOCERA-R, TenDimensions Technology, China) was used to print IVD scaffolds. First, we prepared the NP and AF inks and placed them into two separate ink cartridges, respectively. Materialise Magics 22 was used to design the 3D model, which was output as a StereoLithography file. Subsequently, the model was programmatically sliced in the Z direction. The predesigned printing parameters (layer thickness and curing time) and the 3D model were input into the 3D printer. Printing was carried out by controlling the transmission of images to the projector and the movement of the build platform. During this process, the ink was exposed to 405 nm 20 mW cm^–2 light. The printing conditions included a curing time of 20 s and a layer thickness of 25 μm. After the printing of the NP section on a layer was finished, the ink cartridge was replaced with one containing AF ink to proceed with the printing of the AF section on the same layer. The obtained scaffold was named the IVD scaffold.

Fabrication of 3D-Printed IVDO

The ″preliminary differentiation″ phase encompasses the first 5 days of differentiation. For NPO differentiation, hMSCs were passaged to P3, and on day 0, wash the cells twice with 1 × PBS. Subsequently, incubate with trypsin-EDTA at 37 °C and 5% CO2 for 1 min. Then, the cells were collected and resuspendh at a concentration of 1 × 10⁶ cells ml^–1 in expansion medium. Next, 200 μL of the cell suspension was added to each well of a 96-well ultralow attachment round-bottom microplate (7007, Corning). On day 1, The expansion medium was replaced with DMEM/F12 medium supplemented with N2, B27, ITS, NEAA, 2 mM l-glutamine, 0.1 mM 2-mercaptoethanol, and 10 ng/mL Activin A for 2 days at step 1. Subsequently, additional cell growth factors were introduced, including 10 ng/mL of FGF2, 50 ng/mL of Noggin, 3 μM of CHIR99021, and 10 μM of AGN193109 for notochordal-like cell differentiation for 3 days in the second step. In step 3, the medium is changed to DMEM-HG supplemented with the following components for NP-like cell differentiation for 15 days: N2, B27, ITS, NEAA, 1% penicillin/streptomycin, 50 μg/mL ascorbic acid-2-phosphate, 40 μg/mL l-proline, 10 nM dexamethasone, 10 ng/mL TGF-β3, 30 μM FA.

For AF organoid differentiation, on day 1, the expansion medium was replaced with DMEM/F12 medium supplemented with N2, B27, ITS, NEAA, 2 mM l-glutamine, 0.1 mM 2-mercaptoethanol, 0.5 mM ascorbic acid, 10 ng/mL Activin A, 500 nM LDN-193189, 1% penicillin/streptomycin, and ascorbic acid-2-phosphate for 2 days at step 1. In step 2, 5 μM CHIR99021, 100 nM AGN 193109 was added to enhance paraxial mesoderm differentiation for additional 3 days. In Step 3, the medium was changed to DMEM-HG supplemented with N2, B27, ITS, NEAA, 2 mM l-glutamine, 0.1 mM 2-mercaptoethanol, 50 μg/mL ascorbic acid-2-phosphate, 10 ng/mL TGFβ1, 100 nM SAG, 100 ng/mL CTGF, and 100 nM AGN 193109, and the cells were further cultured for 15 days. On day 21 the NP and AF cell spheroids were collected and suspended in HG hydrogel (containing N2, B27, ITS, NEAA, 1% penicillin/streptomycin, 50 μg/mL ascorbic acid-2-phosphate, 40 μg/mL l-proline, 10 nM dexamethasone, 10 ng/mL TGF-β3, 30 μM FA.) and SilMA hydrogel (containing N2, B27, ITS, NEAA, 2 mM l-glutamine, 0.1 mM 2-mercaptoethanol, 1% penicillin/streptomycin, 50 μg/mL ascorbic acid-2-phosphate,10 ng/mL TGFβ1, 100 nM SAG, 100 ng/mL CTGF, and 100 nM AGN 193109) were respectively deposited into NP and AF section of IVD scffolds using a 1 mL Luer-Lok syringe, followed by UV (20s, 405 nm, 20 mW cm^–2) irradiation. Subsequently, the cells were cultured in DMEM/HG medium for 7 days. On day 28, an integrated IVDO was formed and collected for subsequent experiments.

Scanning Electron Microscopy (SEM)

Assembled NP and AF scaffolds were prepared and allowed to reach swelling equilibrium in PBS at 37 °C. Subsequently, the samples were freeze-dried under vacuum conditions at −80 °C. The dried hydrogels were observed using SEM (TESCAN MIRA LMS, TESCAN, Czech Republic). Using ImageJ, a line was drawn along the scale bar with the Straight-Line Tool. The scale was set under Analyze > Set Scale. The thickness of each hydrogel scaffold layer was measured by tracing with the Straight-Line Tool, the thickness data were recorded by pressing Ctrl+M. and the entire process was repeated three times to obtain an average value.

Cell Proliferation Assay

The CCK-8 assay kit was used to conduct a cell proliferation experiment to verify the in vitro cell compatibility of the hydrogel. hMSCs resuspended in hydrogels were seeded into a 24-well plate (15,000 cells/well), with three replicates per group. The CCK-8 assay was performed after 1, 3, 7, and 12 days of 3D culture. The specific experimental steps were as follows: A cell culture medium containing 10% CCK-8 solution (HYCEZMBIO, HYCCK8) was prepared, the culture medium from the wells was removed and washed with PBS, 400 μL of 10% CCK-8 culture medium was added to each well, and the plate was incubated at 37 °C for 2.5 h. The liquid from the 24-well plate was transferred to a 96-well plate, and the absorbance at 450 nm for each well was measured using the VICTOR Nivo multimode plate reader (PerkinElmer, Waltham, USA).

Cell Viability Assay

Cells were embedded in hydrogels and photopolymerized. Cell culture medium was added immediately and incubated in a CO2 incubator. Cell viability was assessed using the Calcein-AM/PI Cell Viability/Cytotoxicity Assay Kit (Beyotime, China). The cells were stained on the first, third, seventh, and 14th days after 3D printing. The stained cells were observed after they had incubated for 30 min. ImageJ software was used for image analysis, and cell viability was calculated based on the percentage of live cells.

Coculture of Hydrogels with Macrophages

To evaluate the immune response elicited by the hydrogels, direct coculture experiments with macrophages were performed. Sterile HG and SilMA hydrogel discs (8 mm in diameter, 2 mm in height) were prepared by UV photopolymerization. Prior to coculture, the hydrogel discs were equilibrated in complete DMEM for 24 h. Macrophages were then seeded directly onto the surface of the hydrogels placed in 24-well plates at a density of 1 × 10⁴ cells per well. Cells seeded on tissue culture plastic (TCP) served as a control.

The coculture was maintained for 48 h. Subsequently, the macrophages on the hydrogels were processed for downstream immunofluorescence to assess polarization markers, including the M2 marker CD206 and the M1 marker CD86.

RNA Sequencing

The experiment included a control group and a folate group. Total RNA was extracted from the aforementioned hMSCs using TRIzol reagent, and the quality and integrity of the samples were assessed. After quantifying the RNA samples, an RNA sequencing library was prepared. BGI Genomics (Shenzhen, China) collaborated in processing the raw data, and the human reference genome was mapped using STAR software. The cutoff values for the fold change and p-value were set at 2 and 0.05, respectively.

qRT–PCR

Cellular RNA was lysed and purified using TRIzol (Invitrogen, 15596026, USA). First, freeze the hydrogel containing organoids in liquid nitrogen. For lysis, 1 mL of TRIzol reagent was added per 50–100 mg tissue equivalent, followed by thorough mechanical homogenization. The resulting lysate was incubated at room temperature for 5 min to facilitate complete dissociation of nucleoprotein complexes. Subsequent RNA purification involved phase separation, isopropanol precipitation, and ethanol washes according to standard protocols. RNA quality control metrics included spectrophotometric analysis (NanoDrop 2000, Thermo Fisher Scientific; acceptable A260/A280 ratios of 1.8–2.2) and evaluation of RNA integrity (Agilent 2100 Bioanalyzer; RNA Integrity Number [RIN] ≥ 7 required for downstream applications). Reverse transcription was performed using a cDNA synthesis kit (Vazyme, Nanjing, R312–01). The Bio-Rad fluorescence quantitative PCR system was used for amplification with the following parameters: initial denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s and 60 °C for 30 s. All reactions contained 0.3 μM of each primer (18S primer sequences are provided in Table S1). Target genes were quantified by normalizing their expression to that of 18sRNA. The gene expression values were log2-transformed to normalize the data scale, enabling simultaneous visualization of both high- and low-abundance genes in the heatmap. Table S2 displays the primer sequences used.

Western Blot Analysis

Cells were washed twice with precooled PBS, and then were lysed using RIPA lysis buffer containing 1% PMSF (Solarbio). The lysates were then centrifuged (12,000g, 15 min). The lysed protein concentration was measured using the BCA method. SDS–PAGE 5× loading buffer (Beyotime) was added and boiled (95–100 °C, 5 min) to denature the protein. The proteins were separated via SDS–PAGE (120 V, 90 min) based on their molecular weight. The loading amount of protein antigen was 30 μg, and the thickness of the glass plate gap was selected to be 1.5 mm. After electrophoresis, the PVDF membrane (Millipore) were used to transfer the proteins, with a current of 300 mA for 60 min. Skim milk powder (5% w/t) was used to block the membrane (1 h) with low-speed shaking and was then washed with TBST. After the membrane washing was completed the primary antibody, which had been diluted in advance, was immediately added and incubated (4 °C, 8 h). The membrane was incubated with HRP-conjugated secondary antibody at room temperature for 1.5 h. The antibodies used can be found in Table S3. Protein bands were visualized using ECL reagent (Affinity) and analyzed with ImageJ (NIH, USA). The experiment was repeated three times.

Dynamic Mechanical Stimulation of the NPOs In Vitro

To simulate the physiological mechanical environment of the intervertebral disc, nucleus pulposus cells were subjected to dynamic compressive stimulation using a cell-based dynamic compression system (Compressor, Cell&Force, China). The NPOs were stimulated with the following parameters: a stress amplitude of 0.1–0.4 MPa, a frequency of 1.0 Hz, and a duration of 2 h per day. After 7 days of culture under this regimen, the cells were harvested for subsequent analysis.

Animal Preparation and In Vivo Test

IVD organoids for in vivo test were prepared referring to the protocol of ‘Differentiation of NPO and AFO through 3D culture’. Under anesthesia, a lateral approach between the lumbar muscle and peritoneum was used to simulate clinical intervertebral disc (IVD) excision surgery, during which a circular defect approximately 17 mm in diameter was created to ensure the implantation of IVDOs. IVDOs were implanted into the excised discs, along with 3D-printed IVD scaffold without cells or no implantation as a control. An untreated disc was used as the intact control. Euthanasia was performed on all animals 6 weeks later.

Open Field Gait Test

During the adaptation period, goats were allowed to acclimate to the testing arena. The goats were placed in a clean open field-testing arena, and gait testing was conducted using the footprint trail method (with the hind hooves marked with different colored inks). The goats were trained to walk along a 9-m-long and 2-m-wide pathway, referring to the previously described method. During the gait testing, the average stride length (measured and averaged from all right and left hind hoof steps) were quantified via ink footprint patterns.

CT and MRI Examination

Prior to euthanasia, lateral CT images and T1 and T2 magnetic resonance images, including axial, sagittal, and coronal planes, were acquired. Three blinded evaluators quantitatively analyzed the IDD using the Pfirrmann grading system based on the signal intensity and uniformity of the nucleus pulposus (NP) in sagittal and axial T2MR images, as well as the boundary distinction between the NP and AF, grading the lesions from 1 to 5. The Pfirrmann grade for each IVD was based on the median. The disc height index (DHI) was calculated by dividing the disc height by the height of adjacent vertebral bodies to measure the intervertebral disc height. The posttreatment intervertebral disc DHI was normalized against the unaffected portion within the same spine to account for interanimal differences.

Histological Evaluation

Organoid and goat intervertebral disc samples were collected, fixed, decalcified and dehydrated. The dehydrated tissue was then embedded in paraffin, sectioned into 4 μm slices, and subjected to safranin O/fast green staining, Masson staining, and H&E staining.

Immunofluorescence Assessment

The cells or tissues were fixed with 4% paraformaldehyde (Biosharp, China), wash three times with Tris-buffered saline for 10 min each. Then incubated in Tris-buffered saline containing 0.3% Triton X-100 (w/v) and 5% normal donkey serum (w/v) for 1 h to block nonspecific reactions. Subsequently, the sections were incubated overnight at 4 °C or for 3 h at room temperature with the antibodies listed in Table S2. After washing three times with Tris-buffered saline for 10 min each, the cells were incubated at 22–28 °C with FITC, CY3, and CY5-conjugated secondary antibodies (Thermo Fisher) for 1 h. Prior to observation under a fluorescence microscope (Olympus), staining was performed using DAPI (Thermo Fisher).

Immunohistochemistry Assessment

For IHC, tissue sections were fixed in 4% paraformaldehyde (Biosharp, China) for 24 h at 4 °C, followed by deparaffinization and rehydration. Sections were incubated for 1 h at room temperature using COL-1 (Proteintech, 14695-1-AP), COL-2 (Proteintech, 28459–1-AP), and ACAN (Proteintech, 13880-1-AP) as primary antibodies. And were then incubated with biotin-labeled goat antirabbit IgG secondary antibody (1:200 dilution; Abcam, ab6720) for 1 h at room temperature. Finally, DAB peroxidase substrate kit (AR1000, Boster) was used for staining.

Antibody Validation for Immunoassays in Caprine Tissues

Primary antibodies used for immunohistochemistry and immunofluorescence in goat tissues were selected and validated based on a multitiered strategy, as explicit cross-reactivity with goat (Capra hircus) was not listed for all reagents. First, antibodies with manufacturer-confirmed reactivity for goat were prioritized (e.g., IL-6). For other antibodies, protein sequence homology between the human immunogen and the goat ortholog was analyzed in silico using BLAST. Antibodies targeting proteins with high predicted sequence homology (>85%; e.g., COL2A1, ACAN, COL1A1) were considered highly likely to exhibit cross-reactivity. For targets with moderate predicted homology (CD68, CD3), their selection was supported by peer-reviewed literature documenting their successful application in species with comparable or lower sequence homology. Ultimately, the specific, robust, and biologically consistent staining patterns achieved for all antibodies (e.g., zonal deposition of COL2 in NP and COL1 in AF, specific immune cell labeling) served as definitive empirical validation of their specificity and effectiveness in our experimental system.

Quantification and Statistical Analysis

Data were expressed as mean ± SD of three independent experiments. Comparisons between multiple groups were performed using One-way analysis of variance (ANOVA). The significance of differences between two groups was assessed using the Student’s t test. P < 0.05 was considered statistically significant, P > 0.05 is considered not significant (ns) (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001). n = 3 independent experiments per group.

The RNA-seq data generated in this study are available on the National Center for Biotechnology Information (NCBI) database under PRJNA1253563 by visiting https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1253563/ and corresponding primer sequences used for qRT-PCR are listed in Table S2. Information on antibodies is provided in Table S3. All data supporting the findings of this study are available within the article and its Supporting Information. Source data are available for Figure 1c–i, l, Figure 2c–h, j–o, Figure 3b–d, Figure 4c, Figugre 5e, f, i, j and Figure S1c, d, Figure S2, Figure S4, Figure S5, and Figure S10a–c, f in the associated source data files. Source data are provided with this paper. The RPKM data of RNA-sequencing supporting the results are available in the Supporting Information. The data supporting the findings of this study are available from the corresponding author upon reasonable request. The information on primer sequences used in qRT–PCR and antibodies used in this study are provided in the Supporting Information. All data supporting the findings of this study are available within the article and its Supporting Information. Any additional requests for information can be directed to, and will be fulfilled by, the corresponding authors. Source data are provided with this paper.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.5c14391.

• Figures S1–S11 and Tables S1–S3: qPCR primers, antibody lists, patient demographics, NMR spectra, RNA sequencing, histological staining, mechanical testing, imaging, and schematics related to the characterization, and in vivo evaluation of intervertebral disc organoids (PDF)

• Movie S1: All goats maintained full lumbar spine function (MP4)

• Movie S2: The goats in the discectomy group maintained full lumbar spine function (MP4)

#.

D. Wu, D. Zhu, X. Zhou, and G. Li contributed equally to this work. D. Wu, D. Zhu, X. Zhou, and G. Li conceived and designed the study. X. Zhang, L. Ma, H. Liang, B. Tong, J. Lei, H. Wang, W. Ke, X. Feng, and K. Wang performed the experiments and analyzed the data. L. Tan, Y. Shi, C. Yang, and B. Wang supervised the project. D. Wu and D. Zhu drafted the manuscript. All authors participated in the interpretation of the data and production of the final manuscript.

The authors acknowledge the support from the National Natural Science Foundation of China (Nos. 82130072, 82072505, 82202725, 22175058, 82372380, 82472511 and 82402871), the Natural Science Foundation of Hubei Province (Nos. 2023AFB805, 2023AFB770), the Natural Science Foundation of Hubei Province Innovative Group Project (No. 2021CFA007), Shenzhen Science and Technology Program (SGDX20230116093544006, JCYJ20240813153421028, JCYJ20250604190927038), and the Fundamental Research Funds for the Central Universities (HUST. YCJJ20242118). We are thankful for the technical support by the Huazhong University of Science and Technology Laboratory Animal Center, the Huazhong University of Science and Technology Analytical & Testing Center, Medical Subcenter, Wuhan Center for Magnetic Resonance, Innovation Academy for Precision Measurement Science and Technology, and Chinese Academy of Sciences. We thank Springer Nature for providing language editing services for this manuscript.

A preprint of this study was deposited on Research Square and was withdrawn prior to this submission: Bingjin Wang; Di Wu; Dingchao Zhu; et al. Development of Intervertebral Disc Organoids through Directed Differentiation of Mesenchymal Stem Cells and Hierarchical Bioprinting. 2024, rs.3.rs-4945920/v1. 10.21203/rs.3.rs-4945920/v1 (accessed February 28, 2026).

The authors declare no competing financial interest.