Long-term expandable auricular chondrocytes reveal a conserved nine-gene decline module underlying senescence-independent dedifferentiation

Abstract

Introduction

Primary chondrocytes often lose matrix-forming capacity during in vitro expansion, limiting scalable cartilage engineering. Here, we examined aging-independent regulation of chondrogenic function during long-term expansion of primary auricular chondrocytes and sought molecular features associated with late-passage functional decline.

Methods

Mouse auricular chondrocytes (mACs) were serially expanded in 2D culture up to passage 40 or 50 (P50) under high- or low-seeding conditions. Proliferation, morphology, and senescence status were assessed by growth curves, phase-contrast imaging, and SA-β-gal staining. Chondrogenic capacity was evaluated using 3D pellet culture and toluidine blue staining. Bulk RNA sequencing of matched 2D and 3D samples was followed by PCA, differential expression, and pathway enrichment analyses. Key transcriptional signatures were validated by RT–qPCR in mouse auricular chondrocytes.

Results

Both seeding conditions supported sustained proliferation across extended passaging. Although not immortalized, mACs maintained stable proliferation for more than P 50 with minimal senescence-associated features, exhibiting cell line–like proliferative behavior. Despite preserved growth, proteoglycan-rich matrix production declined at P50. Transcriptomic analyses identified culture dimensionality as the dominant source of variance, with 2D cultures showing passage-dependent drift and 3D pellets retaining a more stable chondrogenic profile. Cross-comparisons identified a conserved nine-gene decline module (Gpx7, Krba1, Plxdc2, Ptgr2, Rarres1, Pxdc1, Ctsh, Pigh, and Diras2), which was validated by RT–qPCR. Enrichment analyses indicated coordinated attenuation of adhesion-, extracellular-, and differentiation-related programs at late passages.

Conclusion

Auricular chondrocytes exhibit cell line–like proliferative capacity with limited senescence activation while progressively losing matrix-forming competence at ultra-late passages. A conserved nine-gene decline module provides molecular benchmarks for passage-resolved assessment and may inform optimization of scalable auricular cartilage manufacturing.

Highlights

• •Primary mouse auricular chondrocytes sustain robust long-term expansion up to passage 50 without classical senescence.
• •Late-passage chondrogenic decline is uncoupled from proliferation, indicating a senescence-independent functional deterioration.
• •Culture dimensionality is the dominant driver of transcriptional variation, with 3D pellets showing comparatively stabilized chondrogenic programs against passage-dependent drift.
• •A conserved nine-gene decline module serves as a robust transcriptomic signature for the loss of chondrogenic identity.

Abbreviation

mACs

Mouse auricular chondrocytes

1. Introduction

A central barrier to cartilage regeneration is that primary chondrocytes lose their stable matrix-forming phenotype during monolayer expansion, accompanied by cell-shape–linked collagen switching, reduced proteoglycan-associated programs, and broad transcriptional remodeling [1,2]. Similar dedifferentiation-associated loss of matrix synthetic capacity has been studied in articular cartilage, particularly in the context of osteoarthritis [[3], [4], [5]].

Multiple redifferentiation strategies, including high-density culture, pellets formation, and hydrogels-based systems, can partially restore chondrogenic programs. However, the efficiency varies widely and deterioration at late-passage remains a major translational bottleneck for scalable cartilage engineering [[6], [7], [8]].

In parallel, auricular (elastic) cartilage has emerged as a clinically relevant cell source for craniofacial reconstruction, including microtia, as patient-derived auricular chondrocytes can be expanded and used to engineer anatomically faithful ear-shaped constructs in preclinical models and early clinical studies [[9], [10], [11], [12]]. Auricular cartilage is non–weight-bearing and surgically accessible, allowing harvest of relatively large cell numbers with lower donor-site morbidity than articular cartilage. Despite these advantages, long-term phenotypic stability, passage limits and scalable manufacturing parameters for auricular chondrocytes remain poorly defined [9,10,13]. Recent studies indicate that auricular cartilage comprises heterogeneous cell populations with variable regenerative potential [[14], [15], [16]]. Moreover, clinical-scale manufacturing can require extremely large chondrocyte yields; for example, auricular cartilage graft constructs have been reported to use on the order of ∼2.4 × 10⁸ expanded chondrocytes from a single patient, with cell–hydrogel formulations loaded at ∼1 × 10⁸ cells/mL, highlighting the practical necessity of robust expansion while preserving chondrogenic potency [17].

Importantly, it remains unclear whether functional decline observed during long-term in vitro expansion reflects classical cellular senescence or instead represents a senescence-independent impairment of chondrogenic capacity, particularly in auricular chondrocytes [[9], [10], [11],16]. Across multiple cartilage systems, serial passaging in 2D monolayer culture is generally associated with progressive loss of cartilage-forming capacity, including reduced 3D matrix production and downregulation of canonical chondrogenic markers such as COL2A1, ACAN, and SOX9, despite continued proliferation. However, the timing and extent of this decline may vary among cartilage cell sources, and the mechanistic basis of this loss remains incompletely resolved. Many studies operationally define “dedifferentiation” primarily by changes in morphology, extracellular matrix output, and type I/II collagen switching, whereas systematic profiling of senescence hallmarks (e.g., p16^INK4a, p21^CIP1, SA-β-gal, γH2AX, and SASP factors such as IL-6 and MMPs) is often limited. In addition, donor age and osteoarthritic status can confound interpretation by conflating organismal aging, replicative aging, and dedifferentiation into a single decline phenotype [3,5,[18], [19], [20]].

In this study, we established an aging-resistant, dedifferentiation-like chondrocyte model by longitudinally profiling primary auricular chondrocytes expanded under defined seeding-density and dimensionality conditions, thereby excluding organismal aging as a confounding factor. Using this controlled cellular system, we sought to delineate late-passage–associated functional decline independently of classical senescence and to identify conserved molecular features that mark the loss of chondrogenic identity.

2. Methods

2.1. Mouse primary auricular chondrocytes

All animal experiments were approved by the Animal Experiment Committee of the Graduate School of Medicine, The University of Tokyo (Approval No. P19-114 and Med-P19-090).

Primary auricular chondrocytes were isolated from the auricle and external auditory canal cartilage of 6-week-old male C57BL/6-Tg (CAG-EGFP) mice (Japan SLC, Shizuoka, Japan), as previously described [21,22]. Mice were euthanized by cervical dislocation, and cartilage tissues were carefully dissected with the perichondrium removed. The tissues were minced and digested in 0.15% collagenase (FUJIFILM Wako, Osaka, Japan) prepared in Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F-12; FUJIFILM Wako, Osaka, Japan) at 37 °C with constant agitation for 8 h. The resulting cell suspension was filtered through a 100-μm cell strainer to remove undigested tissue fragments and centrifuged at 430 × g for 5 min. The isolated chondrocytes were seeded onto 100-mm tissue culture dishes and cultured in chondrocyte growth medium at 37 °C in a humidified atmosphere of 5% CO₂. Upon reaching confluence, the cells were passaged and cryopreserved at passage 1 (P1) using CELLBANKER 1 (Takara Bio Inc., Shiga, Japan).

2.2. Long-term 2D passaging under distinct seeding densities

Cryopreserved chondrocytes were thawed and seeded in monolayer (2D) culture and serially expanded to passage 50 (P50). Cells were plated on 100 mm type I collagen–coated dishes and maintained under standard conditions (37 °C, 5% CO₂, humidified incubator) in complete growth medium consisting of DMEM/F-12 supplemented with 5% fetal bovine serum (FBS) (Gibco, Thermo Fisher Scientific), 0.142 U/mL insulin (Novo Nordisk, Tokyo, Japan), 100 ng/mL FGF-2 (KAKEN PHARMACEUTICAL, Tokyo, Japan), and 1 x penicillin–streptomycin (FUJIFILM Wako, Osaka, Japan). Routine passaging was performed at approximately 80–90% confluence using TrypLE Express (Thermo Scientific, Waltham, MA, USA). Population doubling numbers were recorded at each passage, and cells were collected at the indicated passages (P3, P23, P30, P40, and P50) for downstream analyses. All cultures were maintained in parallel using matched culture vessels and media volumes.

Two seeding-density conditions were used to assess long-term proliferative behavior (Fig. 1a).

• •High seeding density: 6 × 10⁵ cells per 100 mm dish; applied to samples collected at P3, P30, and P40.
• •Low seeding density: 5 × 10⁴ cells per 100 mm dish; the low-density expansion regimen was initiated at P8 to increase replicative pressure; applied to samples collected at P23 and P50.

Fig. 1.

Experimental design and long-term proliferative trajectories of primary mouse auricular chondrocytes.

(a) Schematic of the long-term 2D serial passaging workflow. Primary auricular chondrocytes were isolated from 6-week-old male C57BL/6-Tg (CAG-EGFP) mice (auricle and external auditory canal cartilage), then expanded in monolayer under two seeding-density regimens: high (6 × 10^5 cells/100 mm dish) with sampling at P3, P30, and P40, and low (5 × 10^4 cells/100 mm dish) with sampling at P23 and P50.

(b) Long-term cumulative growth curves of mouse auricular chondrocytes (mACs) under low-versus high-seeding conditions. At each passage, viable cells were counted by trypan blue exclusion, and cumulative cell yields were plotted against passage number after log transformation to accommodate multi-order-of-magnitude expansion.

2.3. 3D chondrogenic culture

To assess chondrogenic potential, cells at passages P3, P23, P30, P40, and P50 (under both seeding-density conditions) were transferred to a three-dimensional (3D) pellet culture system. Chondrogenic differentiation was induced using standard chondrogenic induction medium consisting of high-glucose DMEM supplemented with 1% fetal bovine serum (FBS), 1% ITS-X (Gibco, Thermo Fisher Scientific), 1 × 10⁻⁴ M non-essential amino acids (Gibco, Thermo Fisher Scientific), 1 mM sodium pyruvate (Gibco, Thermo Fisher Scientific), 50 μg/mL ascorbic acid (FUJIFILM Wako, Osaka, Japan), 10 ng/mL BMP-2(Oriental Kobo, Nagoya, Japan), 10 ng/mL GDF5 (PROSPEC, Ness-Ziona, Isreal), 10 ng/mL TGF-β1 (PeproTech, Thermo Fisher Scientific), 0.5 x penicillin–streptomycin [23]. Pellets were harvested after differentiation for RT-qPCR, histological analysis, and RNA sequencing.

2.4. Construction of the long-term growth curve

At each passage, the total number of viable cells was quantified immediately prior to reseeding. Cells were harvested using TrypLE Express (Gibco, Thermo Fisher Scientific), resuspended in growth medium, and counted using a hemocytometer with trypan blue exclusion. The total cell yield at each passage was recorded as the number of cells harvested from the entire culture vessel.

Growth curves were generated by plotting cumulative cell numbers against passage numbers for each seeding-density condition. Cell counts were log-transformed prior to plotting to accommodate values spanning multiple orders of magnitude. Low- and high-seeding conditions were plotted independently.

2.5. Phenotypic and histological analyses

Cells at the indicated passages (P3, P23, P30, P40, and P50) were imaged under live culture conditions using a phase-contrast microscope (DMi8; Leica Microsystems, Tokyo, Japan).

Senescence-associated β-galactosidase (SA-β-gal) staining was performed using a commercial staining kit (Cell Signaling Technology, Danvers, MA, USA) according to the manufacturer's instructions, with minor modifications (n = 3). Briefly, cells were rinsed twice with PBS and fixed with the kit-provided fixative for 10 min at room temperature. After fixation, cells were incubated with X-gal staining solution (pH 6.0) at 37 °C in a dry incubator without CO₂ for 12–16 h. Stained cultures were washed with PBS and imaged by All-in-one fluorescence microscope (BZ-X1000, KEYENCE, Osaka, Japan). A549 lung carcinoma cells treated with 7 mM bleomycin, following standard SA-β-gal induction protocols, were used as a positive control to validate assay performance.

For quantitative analysis, five random, non-overlapping fields of view were captured per well at 20x magnification using a bright-field microscope imaging system. In each field, cells-stained blue was identified as β-Gal-positive cells, Senescent cells were identified by the development of a perinuclear blue precipitate of the X-gal chromogen. The percentage of β-Gal-positive cells relative to the total number of cells was calculated as follows:

$$\mathrm{\beta }‐\text{Gal}‐\text{positive}\text{cell}\text{rate}(\%)=\frac{\text{Number}\text{of}\mathrm{\beta }‐\text{Gal}‐\text{positive}\text{cells}}{\text{Total}\text{number}\text{of}\text{cells}}\times 100$$

Pellets after 3 weeks of differentiation. From different passages under high- and low-seeding density conditions were fixed in 4% paraformaldehyde (PFA; FUJIFILM Wako, Osaka, Japan) for 24 h at 4 °C. Following fixation, samples were cryoprotected separately in 10%, 20% and 30% sucrose infiltrated, embedded in OCT compound, and sectioned at 8 μm thickness using a cryostat. Sulfated glycosaminoglycan deposition was evaluated by Toluidine blue (TB) staining. Sections were examined by bright-field microscopy (BZ-X1000, KEYENCE).

2.6. RNA extraction and bulk RNA sequencing

Total RNA was extracted from 2D-expanded auricular chondrocytes at each passage and from 1-week 3D pellet differentiation cultures using ISOGEN reagent (Nippon Gene Co., Ltd., Tokyo, Japan) following the manufacturer's instructions. RNA samples with RNA integrity numbers (RIN) > 8.0 were subjected to bulk RNA sequencing, performed by BGI Japan Co., Ltd. (Tokyo, Japan).

2.7. RNA extraction and real-time quantitative polymerase chain reaction (RT-qPCR)

Complementary DNA (cDNA) was synthesized from total RNA via reverse transcription using PrimeScript™ RT Master Mix (Takara Bio Inc., Shiga, Japan), following the manufacturer's instructions. Gene expression was quantified using THUNDERBIRD® Next SYBR® qPCR Mix (TOYOBO CO., LTD., Tokyo, Japan). Real-time qPCR was performed using a 7500 Fast Real-Time PCR System (Applied BioSystems, CA, USA). Relative expression levels of each gene were analyzed using the ΔΔCt method. Forward and reverse primers utilized for this analysis are detailed in Table 1.

Table 1.

Mouse primers used for qRT-PCR.

Gene	Forward (5′–3′)	Reverse (5′–3′)
Ctsh	ACCGTGAACGCCATAGAAAAG	TGAGCAATTCTGAGGCTCTGA
Diras2	GGGAGAGCTACATCCCGACT	GCTCATAGATGGGCTTGAGTTC
Gpx7	TCCGAGCAGGACTTCTACGAC	TCTCCCTGTTGGTGTCTGGTT
Krba1	CAACTCTGCGGTTCTCGGT	CATGGCGGAACTCAGCTTC
Pigh	GCTACTACTCACCGTACTGCC	GGACCATGCTATTCTCGCACA
Plxdc2	GCCGCAGCAGGAGTTATGTTA	TTCATTCCAAGGAAAAGCGTTTG
Ptgr2	GAGGTGACATCAGCAACGC	GCAACTCCCATGTTTTCCAAGC
Pxdc1	CAACCTCGACTGCATGTCCT	AATACCCAGGCCTGCTGAAC
Rarres1	CTGCGCTGCACTTCTTCAAC	TGCTAAATACCAAGTCCACTTCG
Gapdh	ATGTGTCCGTCGTGGATCTGA	TTGAAGTCGCAGGAGACAACCT

2.8. Bioinformatic analysis

Principal component analysis (PCA) was performed on the normalized gene expression matrix to summarize major axes of transcriptional variation across samples. Scores for the first two principal components (PC1 and PC2), accounting for 39.93% and 24.57% of the total variance, respectively, were plotted to visualize sample relationships. Points were annotated by passage number and culture condition (2D or 3D).

Raw read count matrices were imported into the iDEP web platform for downstream analyses. Differentially expressed genes (DEGs) were identified using DESeq2 with a ≥2-fold change threshold and Benjamini–Hochberg–adjusted p-values <0.05.

Bar graphs were generated to examine passage- and condition-dependent expression patterns for two predefined gene sets: chondrogenesis-associated genes and senescence-associated genes. RNA-seq data were normalized to fragments per kilobase of exon per million mapped reads (FPKM)values. To enable cross-gene and cross-condition comparability, the transformed expression values were standardized into row-wise Z-scores (mean = 0, standard deviation = 1). These Z-score normalized expression matrices were then visualized as bar graphs using the iDEP web platform, stratifying samples by passage (P3, P23, P30, P40, P50) and culture condition (2D monolayer vs. 3D pellet). Differential expression was annotated based on predefined fold-change thresholds (FC > 3, ∗∗; FC > 10, ∗∗∗).

DESeq2 results generated in iDEP were exported to MATLAB (MathWorks) for downstream analysis. For passages P3, P23, P30, and P40, DESeq2 contrasts against P50 were performed separately for 2D and 3D cultures. Genes showing a ≥3-fold decrease at P50 relative to the earlier passage were identified, and for each passage, only genes meeting this criterion in both 2D and 3D were retained as passage-common decline sets. The intersection of the four passage-common sets was visualized using a four-way Venn diagram, and the shared nine genes were defined as the conserved core gene-decline module.

Volcano plots were generated for comparisons between P50 and P3, P23, P30, and P40 to visualize differential-expression patterns and assess the consistency of the nine-gene decline (Gpx7, Krba1, Plxdc2, Ptgr2, Rarres1, Pxdc1, Ctsh, Pigh, and Diras2) across contrasts.

Gene set enrichment analysis (GSEA) was performed using transcriptomic profiles from P3 and P50 samples. Genes were ranked by signed differential-expression statistics and tested against curated Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Gene Ontology (GO) collections. GO enrichment was evaluated for Biological Process, Cellular Component, and Molecular Function categories. Enrichment results were summarized using normalized enrichment scores and FDR-adjusted q-values.

2.9. Statistics

For mouse primary auricular samples, RT–qPCR validation was conducted with three biological replicates (n = 3; Fig. 7a–i), and differences among groups were tested by one-way ANOVA followed by Tukey's multiple-comparisons test.

Fig. 7.

RT–qPCR validation of the conserved nine-gene decline module in mouse auricular chondrocytes (mACs).

(a–i) RT–qPCR quantification of Plxdc2 (a), Gpx7 (b), Ctsh (c), Diras2 (d), Pigh (e), Krba1 (f), Ptgr2 (g), Rarres1 (h), and Pxdc1 (i) in mACs maintained as 2D monolayers and 3D pellets at P3, P40, P50. For each condition, three independent biological replicates were analyzed (n = 3); technical triplicates were performed for each sample and averaged within each biological replicate. Relative expression was calculated using the ΔΔCt method. Statistical significance between passages within each culture format is indicated in the panels.

3. Results

3.1. Proliferative trajectories of long-term passaging

Primary auricular chondrocytes isolated from CAG-EGFP mice were subjected to long-term culture to assess replicative senescence. Under standard culture conditions, the cells retained proliferative capacity through at least passage 8, indicating no overt senescence during early passaging.

To compare proliferative behavior under different initial plating densities, cumulative cell yields were tracked across 50 serial passages under low-density conditions and 40 serial passages under high-density conditions (Fig. 1a). Cells seeded at low density exhibited a consistently higher growth trajectory than those seeded at high density (Fig. 1b). Both low- and high-density cultures showed increasing cumulative cell numbers without an apparent reduction in expansion rate, although cells cultured at high density displayed a lower overall proliferation rate. Together, these results indicate that while low-density seeding supports more robust long-term expansion, both seeding conditions permit sustained proliferative capacity.

3.2. Morphological stability and senescence status during long-term passaging

Phase-contrast imaging was used to examine morphological differences between the two plating conditions over time (Fig. 2a). Cells were examined at P3, P30, and P40 under high-density conditions, and at P23 and P50 under low-density conditions. At P30 and P40 under high-density conditions, as well as at P23 and P50 under low-density conditions, cultures formed confluent monolayers and preserved a dense morphology with predominantly rounded cell shapes, without evidence of cellular enlargement or vacuolization.

Fig. 2.

Morphology and senescence status during long-term passaging of mouse auricular chondrocytes.

(a) Representative phase-contrast images of 2D-expanded mACs across serial passages under high seeding (P3, P30, P40; 6 × 10^5 cells/100 mm dish) and low seeding (P23, P50; 5 × 10^4 cells/100 mm dish) conditions. Scale bar: Left 200 μm, Right 20 μm

(b) Senescence-associated β-galactosidase (SA-β-gal) staining and qualification of mACs at the indicated passages undSAer high- and low-seeding conditions (n = 3). Bleomycin-treated A549 cells were included as a positive control to confirm assay performance. Scale bar: Left 100 μm, Right 50 μm

(c) Quantification of senescence-associated β-galactosidase (SA-β-gal) staining, showing the percentage of SA-β-gal-positive cells in bleomycin-treated A549 cells (A549+Bleo) and in the indicated P groups (P3, P23, P30, P40, P50) (n = 3).

Overall, auricular chondrocytes maintained a relatively stable proliferative morphology across extended serial passaging under both low- and high-density seeding conditions.

To determine whether long-term passaging induced senescence-associated changes, SA-β-galactosidase staining was performed at P3, P30, and P40 under high-density seeding conditions, and at P23 and P50 under low-density seeding conditions (Fig. 2b), and the results were quantified (Fig. 2c). Across all examined passages, nearly no SA-β-gal–positive cells was observed, including at P40 and P50, with staining levels comparable to those of early-passage cultures and substantially lower than those of the positive control (bleomycin-treated A549 cells). The β-Gal-positive cell rate was 52.98 ± 11.90% in bleomycin-treated A549 cells (positive control), whereas it was 0% in all passage groups (P3, P23, P30, P40, and P50). The Kruskal-Wallis test revealed a significant difference among the groups (H = 16.83, p = 0.0048).

These results indicate that prolonged serial passaging under both seeding conditions was not associated with a marked increase in senescence-associated β-galactosidase activity in mACs.

3.3. Alteration of chondrogenic matrix production capacity

To evaluate the effects of prolonged passaging on the chondrogenic capacity of auricular chondrocytes, pellet cultures were generated under high- and low-seeding conditions and analyzed by toluidine blue (TB) staining (Fig. 3a).

Fig. 3.

Chondrogenic matrix production during long-term passaging of mouse auricular chondrocytes.

(a) Toluidine blue (TB) staining of cryosections from 3D chondrogenic pellets generated from the indicated passages under high- and low-seeding conditions after chondrogenic induction. Scale bar: 500 μm, Right 50 μm.

Under high-seeding conditions, cells at P3 formed cartilaginous pellets with intense and relatively homogeneous metachromatic staining, indicative of abundant extracellular matrix deposition. At P30, pellets remained intact but showed less organized matrix architecture, although metachromasia was still evident. At P40, pellets displayed cohesive morphology with continuous extracellular matrix deposition.

In contrast, under low-seeding conditions, pellets at P23 exhibited strong metachromatic staining with clear proteoglycan deposition, whereas pellets at P50 showed reduced TB staining intensity with limited metachromasia, particularly in peripheral regions.

Overall, TB staining revealed passage-dependent changes in matrix deposition under both seeding conditions, with reduced staining intensity observed at P50.

3.4. Emergence of divergent molecular trajectories by transcriptomic profiling

To characterize transcriptional changes associated with long-term expansion of auricular chondrocytes, a structured bulk transcriptomic workflow was established (Fig. 4a). Cells were collected at P3, P30, and P40 under high-seeding conditions, and at P23 and P50 under low-seeding conditions and subsequently maintained as either 2D monolayers or 3D pellets prior to RNA sequencing. This experimental design enabled systematic comparison of passage-dependent transcriptional changes across culture conditions.

Fig. 4.

Bulk transcriptomic profiling reveals a culture-dimensionality–driven transcriptional divergence and preserved chondrogenic identity in 3D pellets.

(a) Schematic of the bulk RNA-seq experimental design. Auricular chondrocytes were sampled at P3, P30, and P40 under high seeding conditions and at P23 and P50 under low seeding conditions and processed as either 2D monolayers or 3D pellets prior to RNA sequencing.

(b) Principal component analysis (PCA) of normalized gene-expression profiles. Samples were separated primarily by culture dimensionality (2D vs 3D) along PC1 (39.93% variance) with additional variation captured by PC2 (24.57% variance).

(c) RNA-seq–derived Bar graph of chondrogenesis-related genes (Mmp13, Comp, Col2a1, Acan, Sox6, Prg4, Sox9, Sox5) and representative fibroblastic/mesenchymal markers (Acta2, Thy1) across serial passages, P3, P23, P30, P40 and P50, in 2D and 3D conditions.

(d) RNA-seq–derived Bar graph of senescence-related genes (Mmp3, Cdkn1a, Foxo1, Foxo3, Il6, Ccl2, Cdkn2a, Prdx3) displayed across 2D and 3D samples and passages.

The y-axis is shown on a logarithmic scale. Each condition's own P50 served as the reference, and FC_signed = log₂(FPKM(t)/FPKM(P50)) was computed. ∗∗|FC| > 3, ∗∗∗|FC| > 10 or a zero-versus-non-zero transition. NA, not available (filtered out at the mfuzz preprocessing step).

Principal component analysis (PCA) revealed clear separation of samples according to culture dimensionality (Fig. 4b). In 2D culture, samples exhibited a passage-associated shift along PC1 and PC2, with later passages positioned progressively farther from early-passage samples. High- and low-seeding 2D samples largely overlapped along this trajectory, indicating that passage number, rather than initial seeding density, accounted for the dominant source of transcriptional variation in 2D culture.

In contrast, 3D-pellet samples clustered within a distinct region of the PCA space and showed limited dispersion across passages. Pellets from different passages occupied overlapping positions along both principal components, indicating reduced passage-dependent transcriptional divergence relative to 2D cultures. Notably, 3D pellets at P40 were positioned closer to the regions occupied by P23, P30, and P40 2D samples than earlier-passage 3D pellets. In addition, 3D pellets at P50 approached the P50 2D region, indicating partial transcriptomic similarity at the latest passage.

3.5. Maintenance of chondrogenesis and limited senescence activation

To further characterize transcriptomic differences between 2D and 3D cultures, expression patterns of genes associated with chondrogenesis and cellular senescence markers were examined (Fig. 4c and d).

Across serial passaging, 3D culture more effectively preserved the cartilage-associated transcriptional program than 2D expansion, as evidenced by consistently higher expression of key matrix-related chondrogenic markers, including Acan, Col2a1, Comp, and Prg4, at most passages. In contrast, these matrix genes were markedly attenuated in 2D cultures and approached minimal levels at later passages, indicating progressive loss of mature chondrogenic features during monolayer expansion. Although transcription factors such as Sox5, Sox6, and Sox9 showed gene- and passage-dependent variation rather than a strictly monotonic pattern, the overall expression landscape still supported superior maintenance of chondrogenic identity under 3D conditions.

Fibroblastic or dedifferentiation-associated markers did not show progressive enrichment in 3D culture. Thy1 remained low across most samples, while Acta2 displayed only transient or discontinuous changes, arguing against sustained fibroblastic conversion in the 3D setting. In addition, Mmp13 was detected mainly in selected 3D samples, suggesting context-dependent matrix remodeling or partial hypertrophic-like activity rather than global erosion of the chondrogenic program.

Senescence-related transcriptional changes were relatively modest overall and did not support a pattern of robust, progressive senescence activation across passaging in either culture system. In 2D cultures, several senescence- or stress-associated genes, including Cdkn1a, Cdkn2a, Mmp3, and Prdx3, were elevated at specific passages, but these changes were intermittent and gene-specific rather than uniformly increasing over time. By contrast, 3D cultures generally maintained lower expression of canonical senescence effectors such as Cdkn2a and Mmp3, consistent with a more restrained senescence-associated transcriptional state.

Notably, genes linked to stress adaptation or senescence modulation, including Foxo1, Foxo3, Il6, and Ccl2, showed variable passage-dependent fluctuations rather than coordinated induction. This pattern suggests that 3D culture does not fully eliminate stress-responsive signaling, but it limits the emergence of a coherent senescence-associated program during long-term expansion. Collectively, these data support the conclusion that 3D culture preserves chondrogenic identity while restricting overt senescence activation during extended passaging.

3.6. Pathway enrichment revelation of functional convergence

Pathway enrichment analyses were performed for genes expressed at higher levels at P3 compared with P50 (Fig. 5). KEGG analysis identified Cell adhesion molecules as the most significantly enriched pathway, followed by Mucin type O-glycan biosynthesis, Arachidonic acid metabolism, and Cytokine–cytokine receptor interaction (Fig. 5a).

Fig. 5.

Pathway enrichment analysis of genes upregulated at P3 relative to P50 (P3 vs P50).

(a) KEGG pathway enrichment analysis for genes expressed at higher levels at P3 relative to P50.

(b) GO Biological Process enrichment analysis for genes expressed at higher levels at P3 relative to P50.

(c) GO Cellular Component enrichment analysis for genes expressed at higher levels at P3 relative to P50, highlighting enrichment for extracellular region and plasma-membrane–associated localization.

(d) GO Molecular Function enrichment analysis for genes expressed at higher levels at P3 relative to P50, including enrichment for signaling receptor–related activities. For all panels, the x-axis indicates enrichment significance as $-{\log }_{10}(\text{FDR})$; dot size denotes the number of genes mapped to each term, and dot color represents fold enrichment.

Gene Ontology analyses showed enrichment of biological processes related to cell adhesion, receptor signaling, BMP-related pathways, and cell migration (Fig. 5b). Enriched cellular components were predominantly associated with the extracellular region and plasma membrane (Fig. 5c). Molecular function analysis highlighted enrichment of calcium ion binding and signaling receptor–related activities (Fig. 5d). Collectively, these convergent KEGG/GO signals point to a passage-sensitive “cell-surface/ECM-interface” program, centered on adhesion and receptor–ligand communication, providing a rationale to prioritize a representative cell-surface candidate when nominating mechanistically interpretable genes for follow-up.

3.7. A conserved core nine-gene decline module

Because many chondrogenic and lineage-associated transcripts showed reduced expression at P50, we examined whether a common set of genes exhibited consistent downregulation across expansion conditions. Genes significantly reduced at P50 in both 2D and 3D cultures were identified for each earlier passage, yielding partially overlapping sets of downregulated genes for P3, P23, P30, and P40 (Fig. 6a). Integration of these passage-specific gene sets by Venn analysis revealed a shared intersection of nine transcripts that were consistently downregulated across all early-passage versus P50 comparisons (Fig. 6b). These nine transcripts (Gpx7, Krba1, Plxdc2, Ptgr2, Rarres1, Pxdc1, Ctsh, Pigh, and Diras2) were defined as a core gene-decline module, based on their consistent downregulation at P50 across passages, seeding densities, and culture conditions.

Fig. 6.

Identification of a conserved nine-gene decline module during long-term expansion

(a) Schematic of the filtering and integration strategy used to define a conserved decline signature. For each early passage (P3, P23, P30, P40), differential expression was computed against P50 separately in 2D monolayers and 3D pellets; genes that were expressed at least threefold lower at P50 in both culture formats were retained as “passage-common decline sets,” and the intersection across early passages was used to nominate a conserved core module.

(b) Four-way Venn diagram showing overlap among the passage-common decline sets derived from P3, P23, P30, and P40 comparisons versus P50. Numbers denote gene counts in each region; the shared intersection identifies the conserved core set.

(c–f) Volcano plots visualizing transcriptome-wide differential expression for each early-passage comparison versus P50: P3 vs P50 (c), P23 vs P50 (d), P30 vs P50 (e), and P40 vs P50 (f). The x-axis shows log 2 fold change (early passage relative to P50), and the y-axis shows −log 10 (adjusted P value). The nine conserved genes are annotated, illustrating their consistent, strong downregulation at P50 across contrasts.

Samples from P3 to P40 maintained moderate to high expression of these genes, with 3D cultures generally exhibiting slightly higher expression levels than 2D cultures. In contrast, at P50, both 2D and 3D cultures showed uniformly low expression of the nine genes.

In volcano plot analyses, these nine genes consistently showed large negative log₂ fold changes with high statistical significance in all comparisons against P50, indicating marked downregulation during long-term expansion. Several genes, including Plxdc2, Ptgr2, Ctsh, and Pxdc1, exhibited log₂ fold changes exceeding 10, depending on the comparison. This pattern was observed across all early-passage versus P50 contrasts (Fig. 6c–f).

3.8. RT-qPCR validation of the nine-gene module in mACs

To validate the transcriptomic changes observed in the nine-gene module, RT–qPCR analyses were performed using mouse auricular chondrocytes (mACs) under 2D and 3D culture conditions (Fig. 7a–i). All nine genes (Ctsh, Diras2, Gpx7, Krba1, Pigh, Plxdc2, Ptgr2, Pxdc1, and Rarres1) showed expression levels generally higher in 3D cultures than in 2D monolayers (Fig. 7a–i). Expression of all nine genes decreased at later passages and was markedly reduced at P50 in both culture systems.

3.9. A coordinated stress and pro-inflammatory signature emerges at ultra-late passage

We next investigated genes significantly downregulated at P50 in the volcano plots (Fig. 6c–f) to complement the nine-gene decline module. Analysis revealed a consistent set of genes, prominently featuring alarmins (S100a8, S100a9), matrix proteases (Mmp3, Mmp 10, Mmp13, Adamts4), and chemokines (Ccl7, Cxcl5). This signature points to a non-canonical, senescence-independent inflammatory and catabolic program activated in response to sustained long-term proliferative stress.

4. Discussion

A central challenge in auricular cartilage engineering is not whether primary chondrocytes can be expanded to clinically relevant numbers, but whether cellular identity and functional potency can be preserved throughout the extensive expansion required for manufacturing. In this study, we established a long-term expansion framework for primary auricular chondrocytes and integrated histological assessment with transcriptomic analyses to examine how passage number, seeding density, and culture dimensionality influence long-term cell states.

Across more than 50 serial passages, mouse auricular chondrocytes maintained sustained proliferative capacity with minimal induction of classical senescence markers. Although the auricular chondrocytes analyzed in this study were primary cells, their ability to sustain proliferation over more than 50 passages confers cell line–like properties. In contrast, chondrogenic output, as assessed by proteoglycan-rich matrix formation, declined at P50 under low-seeding conditions.

In many chondrocyte systems, serial monolayer expansion is associated with progressive dedifferentiation, characterized by reduced expression of cartilage-specific matrix genes and, in some contexts, increased expression of fibroblastic or stress-related programs [8,[24], [25], [26], [27]]. In this study, long-term expanded auricular chondrocytes exhibited limited accumulation of senescence-associated features even at late passages, suggesting that early entry into canonical senescence is not the dominant outcome of prolonged culture. Instead, the primary change observed was a gradual reduction in lineage-associated transcriptional programs and extracellular matrix production capacity. Together, this long-term expandable, non-immortalized chondrocyte population provides a practical platform for dissecting passage dependent functional decline and for evaluating manufacturing-relevant quality attributes.

An organizing feature emerging from the transcriptomic analyses is the strong association between culture dimensionality and passage-dependent transcriptional patterns. Cells maintained in 2D monolayer culture exhibited a largely unidirectional shift with increasing passage number, whereas 3D pellet cultures occupied a comparatively constrained transcriptional space across extended expansion. These observations are consistent with previous reports showing that three-dimensional culture conditions through factors such as altered cell shape, enhanced cell–cell interactions, pericellular matrix formation, and mechano-transduction can support maintenance of chondrogenic gene expression programs in expanded chondrocytes [24,25,[28], [29], [30]].

Notably, molecular and functional readouts diverged at P40_3D and P50_3D. Although bulk RNA-seq revealed broadly distinct transcriptomic tendencies at the gene-expression level, only P40 formed a mature, proteoglycan-rich cartilage pellet, whereas P50 failed to do so in the 3-week histological assay. This mismatch likely reflects differences in culture duration. Our results suggest that auricular chondrocytes retain substantial chondrogenic competence through P40, despite the general expectation that serial passaging progressively impairs cartilage formation. The 1-week 3D transcriptome likely captures early chondrogenic engagement, which remains partially detectable even in P50_3D cells. In contrast, the 3-week pellet assay interrogates sustained matrix synthesis, pericellular matrix assembly, and tissue maturation, capacities that collapse specifically at P50. These findings indicate that early transcriptional activation and durable functional cartilage formation are distinct properties, with the critical failure at P50 emerging during the later phase of matrix production and maintenance. Mechanistically, this discrepancy may arise from (i) kinetic insufficiency, whereby late-passage cells initiate but cannot maintain ECM accumulation; (ii) compositional effects, in which a reduced fraction of fully competent chondrogenic cells at P50 biases early bulk RNA-seq toward transient responders; and/or (iii) cumulative 3D culture constraints (e.g., diffusion or stress microenvironments) that disproportionately impair P50 pellets. In this study, P50 was selected as the endpoint to sharpen the distinction between senescence and dedifferentiation despite sustained proliferation and limited canonical senescence induction, P50 exhibits a coordinated erosion of chondrogenic identity and matrix-forming potency, enabling interrogation of senescence-independent functional decline.

Importantly, the stabilizing effect of 3D culture appeared to be partial rather than absolute. At the latest passage examined, 3D samples showed transcriptional changes that approached those observed in 2D-expanded cells, particularly for genes within the decline-associated module. This convergence suggests that prolonged passaging is associated with transcriptional changes that are not fully reversible by changes in culture context alone. While the underlying mechanisms were not directly examined in this study, these findings are compatible with the possibility that late-passage alterations involve more persistent regulatory changes, such as durable transcriptional or epigenetic modifications, alterations in receptor or extracellular matrix organization, or cumulative cellular stress responses [27,31].

Seeding density represents an additional process parameter that modulates long-term chondrocyte behavior, primarily through effects on expansion kinetics and late-passage outcomes. High-density culture has been widely used to preserve or partially re-induce chondrocytic features compared with sparse monolayer conditions [28,32,33], potentially through contact-dependent signaling and local accumulation of autocrine or paracrine factors.

In this study, the influence of seeding density on global transcriptomic patterns in 2D culture was relatively modest, with passage number accounting for the dominant source of variation. However, density-associated effects became more apparent at later passages and within gene sets related to stress responses and fate regulation. These observations suggest that while seeding density does not override passage-dependent transcriptional drift, it can modulate the severity or timing of late-stage changes.

An important outcome of this study is the identification of a conserved nine-gene core-decline module (Gpx7, Krba1, Plxdc2, Ptgr2, Rarres1, Pxdc1, Ctsh, Pigh, and Diras2) that shows consistent downregulation at P50 across multiple early-passage comparisons. The reproducibility of this pattern across different starting passages and culture conditions indicates that these transcripts are associated with late-stage expansion rather than reflecting context-specific effects.

Although the causal roles of individual genes were not directly examined, the module includes genes linked to processes relevant to chondrocyte maintenance, such as redox regulation, lysosomal or proteostatic function, membrane or adhesion-related signaling, and lipid mediator pathways. These functional categories have been implicated in maintaining cellular stability under prolonged culture conditions [24,25,29,34]. In the present dataset, auricular chondrocytes maintained sustained proliferative capacity with limited induction of classical senescence markers yet exhibited coordinated attenuation of chondrogenic programs at ultra-late passages. These findings indicate that proliferative capacity and matrix-forming potential can become uncoupled during prolonged expansion. In contrast to the muted canonical senescence program, ultra-late passage cells further gain a decompensated stress and inflammatory phenotype marked by alarmins (S100a8/a9), chemokines, and a suite of ECM-degrading enzymes. This signature represents a crucial mechanistic bridge between the passive loss of the chondrogenic program (the decline module) and the active, functional failure of matrix production at P50. It suggests that long-term expansion does not merely silence anabolism but actively induces a pro-catabolic, ‘osteoarthritic-like’ state independent of classical p16/p21 signaling [35], [36], [37], [38]]. This gain-of-function phenotype serves as a functional readout of eroded cellular quality that is not captured by proliferation or senescence assays alone, reinforcing the need for multiparametric quality benchmarks for scalable auricular cell manufacturing.

However, the mechanisms by which the key regulators identified in this study contribute to passage-associated functional decline remain unclear and were not directly addressed here. Further studies will be required to determine the specific mechanistic roles of these genes in passage-associated functional decline.

Several limitations should guide interpretation. Functional readouts in this study emphasize pellet histology; further studies should incorporate quantitative matrix biochemistry and in vivo implantation to link molecular erosion to clinically meaningful performance.

5. Conclusion

In summary, auricular chondrocytes with cell line–like properties exhibit sustained proliferative capacity with limited activation of classical senescence pathways, while undergoing coordinated transcriptional changes that culminate in reduced chondrogenic matrix production and the emergence of a conserved nine-gene decline module at P50 under low-seeding conditions. Culture dimensionality was strongly associated with transcriptional stability, with three-dimensional microenvironments partially preserving chondrocyte identity during prolonged expansion. Together, these findings provide a framework for understanding passage-associated functional decline and support the development of molecular and functional benchmarks to guide optimization of auricular chondrocyte expansion for cartilage regeneration.

Author contributions

D.A., A.H., and N.T. conceived the study and designed the experimental strategy. D.A. performed the experiments, acquired the data, drafted the figures and manuscript and acquired funding. D.A. and N.T. analyzed and interpreted the data. T.S. provided technical support for data acquisition. Y.-Y.C. contributed to chondrocyte handling. N.T., A.H., and K.H. supervised the study. All authors reviewed and edited the manuscript and approved the final version.

Declaration of competing interest

Atsuhiko Hikita was affiliated with an endowed chair supported by FUJISOFT INCORPORATED (until October 31, 2020) and with an endowed chair supported by CPC corporation, Kyowa Co., Ltd., Kanto Chemical Co. Inc., and Nichirei corporation (July 1, 2021), and has been affiliated with the Social Cooperation Program of Kohjin Bio Co., Ltd. Since July 1 2022. Naoki Tsuji was affiliated with the Social Cooperation Program of Kohjin Bio Co., Ltd. from April 1, 2023,to March 31,2026.