Osteolineage cells ablation promotes skin aging phenotypes via dysregulated bone marrow macrophages

Summary

Clinical evidence links bone loss to skin thinning during aging, yet the causal role of bone in skin regulation remains unclear. Here, we show that the partial ablation of osteolineage cells induces skin aging, manifested by dermal thinning, reduced epidermal proliferation, delayed wound healing, and impaired hair regeneration. Single-cell RNA sequencing revealed that osteolineage cells ablation profoundly alters bone marrow macrophages, characterized by pro-inflammatory activation, metabolic dysregulation, and enhanced SASP signaling, accompanied by myeloid skewing. These dysfunctional macrophages infiltrated the skin and were associated with aging-like changes in skin parenchymal cells, including impaired epithelial differentiation, hair follicle stem cell dysfunction, and increased inflammation. Integration with published aging skin datasets confirmed that osteolineage cells ablation recapitulates key molecular features of natural skin aging. Together, our findings identify a skeletal-immune-skin axis linking bone marrow homeostasis to peripheral tissue aging.

Graphical abstract

Highlights

• •Osteolineage cells ablation leads to a skin aging phenotype
• •Skin parenchymal cells exhibit aging-related changes after osteolineage cell ablation
• •Osteolineage cell ablation drives myeloid bias and inflammatory macrophage polarization

Immunology

Introduction

Aging is a systemic process that leads to progressive alterations in the morphology and function of multiple tissues and organs. Skin aging, in particular, is characterized by a range of phenotypic changes, including wrinkling, thinning, reduced hair follicle density, delayed wound healing, and increased susceptibility to cancer.^1,2 As the body’s largest organ, the skin plays a critical role in maintaining physiological homeostasis, not only through its protective barrier function but also via its endocrine activities that interact with various other tissues.^3,4,5 For example, in the context of aging, decreased cystatin A expression in the epidermis has been implicated in age-related bone loss, suggesting a potential link between skin and bone in the aging process.⁶

The skeleton itself undergoes substantial changes during aging, marked by a systemic reduction in bone mass and alterations in microarchitecture.⁷ As the most abundant cell population in bone, osteolineage cells are assumed to be involved in bone aging. During natural aging, osteocyte number decreases, bone cavity density reduces, osteocyte dendrites become fewer, and osteocyte morphology is significantly altered.⁸ In addition to constituting the basic structure of bone, osteocytes contribute to the maintenance of overall homeostasis by sensing the local and systemic environment and then engaging in endocrine and paracrine interactions with bone and multiple other systems, including bone marrow, muscle, brain, and vascular system.^9,10 Recent studies suggest that the aging of skin and bone are closely linked, with both tissues experiencing volume loss over time.^11,12,13 Significant premature skin aging has also been observed in some pathological conditions with bone loss, such as fibrodysplasia ossificans progressiva (FOP) and osteogenesis imperfecta.^14,15 However, whether bone has a regulatory role in skin during aging has not been elucidated.

Bone marrow is a vital organ responsible for hematopoiesis—the production of blood and immune cells—and influences the physiology of peripheral tissues and organs through the release of immune mediators, progenitor cells, and metabolic factors. Aging bone marrow exhibits reduced regenerative capacity of hematopoietic stem cells, a myeloid-biased differentiation pattern, and increased progenitor maturation, collectively contributing to systemic homeostatic imbalance.^16,17

In this study, we investigate the role of bone in regulating skin aging using the Dmp1^cre-DTA^ki/wt mouse model, which induces partial osteolineage cells ablation and bone loss. These mice exhibited phenotypes of premature skin aging, including delayed hair regrowth, impaired wound healing, reduced skin thickness, and diminished keratinocyte proliferative capacity. Single-cell RNA sequencing (scRNA-seq) of skin tissues revealed transcriptional profiles characteristic of aging, such as the downregulation of epithelial and hair follicle stem cell markers, impaired keratinization, attenuated Wnt signaling, and metabolic reprogramming. Additionally, elevated inflammatory signaling and increased senescence-associated secretory phenotype (SASP) scores were observed across multiple skin parenchymal cell types. Parallel single-cell analysis of bone marrow demonstrated a shift toward myeloid lineage bias and pro-inflammatory activation of macrophages following osteolineage cells ablation. Together with bone marrow transplantation experiments, these findings reveal a regulatory role of osteolineage cells in skin aging, mediated by bone marrow macrophages.

Results

Osteolineage cells ablation impaired hair regeneration

To investigate the impact of osteolineage cells ablation on skin aging, we generated Dmp1^cre-DTA^ki/wt mice, in which osteolineage cells were selectively ablated by inducing the expression of diphtheria toxin subunit A (DTA) in dentin matrix protein 1 (Dmp1)-positive osteolineage cells. These mice exhibited a markedly higher proportion of empty osteocyte lacunae (approximately 60%), far exceeding that of the WT group (approximately 10%) (Figures S1A and S1B). This finding was consistent with our previous study, which also reported that this mouse line displays reduced body weight, decreased bone mass, and a shorter lifespan compared with WT controls.¹⁸ To exclude the possibility that Dmp1 deletion directly affects the skin, we performed immunofluorescence-based quantification of Dmp1 expression in the skin and bone tissues of Dmp1^CreAi9^fl/fl mice. The results showed that Dmp1 expression in the skin was minimal and significantly lower than in bone tissue (Figure 1A).

Figure 1.

Osteolineage cells ablation in Dmp1^cre-DTA^ki/wt mice induces a skin aging phenotype

(A) Immunofluorescence images of Dmp1 expression in the bones and skin of Dmp1^creAi9^fl/fl mice. Scale bars, 20 μm.

(B and C) Representative images of hair regrowth on the back skin of WT and Dmp1^cre-DTA^ki/wt mice at different time points post epilation and kinetics of hair regrowth (n = 3).

(D and E) Representative H&E staining of skin and quantification of anagen stages in mouse back skin at 8 days post epilation (n = 3). Scale bars, 100 μm.

(F and G) Immunofluorescence staining of EdU in HFs from WT and Dmp1^cre-DTA^ki/wt mice 4 days post-depilation and quantification of EdU⁺ cells per hair follicle at 4 days post epilation (n = 9 HFs from 3 mice per group). Scale bars, 10 μm.

(H and I) Representative macroscopic digital images of the wounds from WT and Dmp1^cre-DTA^ki/wt mice and the quantitative ratios of the open wound size relative to its original size on the day of wounding at different time points during the wound closure process (n = 3). Scale bars, 2 mm.

(J and K) H&E staining of back skin from WT and Dmp1^cre-DTA^ki/wt mice and quantification of skin thickness, skin epidermal thickness, and cell numbers (n = 6). Scale bars, 100 μm.

(L and M) Co-immunostaining of Ki67 and Krt14 in the skin sections and quantification of the proportion of Ki67-positive cells among Krt14-positive cells in the skin (n = 6). Scale bars, 20 μm.

Normal mouse hair follicles, driven by hair follicle stem cells (HFSCs), undergo cyclical phases of growth (anagen), regression (catagen), and rest (telogen).¹⁹ In aged mice, HFSCs exhibit prolonged resting phases and shorter growth phases, with a delayed response to tissue regeneration signals, resulting in a prolonged transition from telogen to anagen.^20,21 To investigate whether the activation potential of HFSCs is compromised in Dmp1^cre-DTA^ki/wt mice under depilation-induced conditions, we applied depilatory cream to the dorsal skin of these mice, thereby synchronizing the entry of hair follicles into the anagen phase. The onset of anagen was monitored by assessing hyperpigmentation in the depilated area.¹⁹ Our results showed that hair regeneration was significantly delayed in Dmp1^cre-DTA^ki/wt mice compared to WT controls 15 days post-depilation (Figures 1B and 1C). Histological analysis on day 8 revealed that in WT mice, hair follicles were elongated, with hair bulbs extending deep into the dermal fat layer, and most follicles had progressed to anagen III. In contrast, hair follicles in Dmp1^cre-DTA^ki/wt mice were shorter, with most hair bulbs confined to the dermis, and the majority of follicles were in phases anagen I-II (Figures 1D and 1E). To further explore HFSC activation dynamics following depilation, we performed a 5-ethynyl-2′-deoxyuridine (EdU) labeling assay to assess cell proliferation in the hair follicles. We found a significant reduction in EdU-labeled cells in Dmp1^cre-DTA^ki/wt hair follicles compared to WT controls, indicating a marked impairment in HFSC activation (Figures 1F and 1G). These findings suggest that, similarly to naturally aged mice, HFSCs in Dmp1^cre-DTA^ki/wt mice exhibit substantially impaired activation in response to depilation, leading to a significant delay in the transition of hair follicles into the anagen phase.

Osteolineage cells ablation resulted in delayed skin wound healing

Wound healing is a complex biological process that involves the coordinated interaction of multiple cell types and tissues, including cell migration, proliferation, extracellular matrix deposition and remodeling, and angiogenesis.^22,23 Impaired wound healing is a hallmark of skin aging, often resulting in delayed tissue repair and increased susceptibility to chronic wounds.²⁴ To investigate whether osteolineage cells ablation affects skin wound healing, we established a full-thickness wound model on the dorsal skin of Dmp1^cre-DTA^ki/wt mice, following previously described protocols. An 8 mm skin biopsy punch was used to create standardized wounds, and wound closure was monitored every two days post-surgery.²⁵ Our findings revealed that Dmp1^cre-DTA^ki/wt mice exhibited significantly delayed wound healing compared to WT controls (Figures 1H and 1I).

Osteolineage cells ablation induces epidermal atrophy and impaired keratinocyte proliferation

Given that skin aging is often characterized by a progressive thinning of both the epidermis and dermis, we performed histomorphometric analysis using H&E staining. Our findings revealed significantly reduced skin and epidermal thickness, along with a decreased number of epidermal cells, in Dmp1^cre-DTA^ki/wt mice compared to controls (Figures 1J and 1K). Keratinocytes, the predominant cells in the epidermis, are essential for epidermal regeneration and stratification, ultimately forming the stratum corneum and maintaining skin barrier function.²⁶ During aging, keratinocyte proliferation declines, leading to reduced epidermal turnover and progressive thinning.²⁷ To assess keratinocyte proliferative capacity in Dmp1^cre-DTA^ki/wt mice, we performed co-localization immunofluorescence staining for the proliferation marker Ki67 and the keratinocyte marker Krt14. We observed a significant reduction in proliferating keratinocytes in Dmp1^cre-DTA^ki/wt mice compared to WT controls, indicating a marked impairment in keratinocyte proliferation (Figures 1L and 1M). In conclusion, we found that osteolineage cells ablation resulted in skin thinning and impaired keratinocyte proliferation.

ScRNA-seq analysis of dentin matrix protein 1^cre-diphtheria toxin subunit A^ki/wt mouse skin

To further investigate the changes in skin aging phenotype caused by osteolineage cells ablation, we conducted scRNA-seq analysis on the skin tissues of Dmp1^cre-DTA^ki/wt and WT mice to identify the genetic alterations in each component cell of the skin. After quality control, we retained 39,274 single cells. Clustering analysis and annotation identified 13 major cell types (Figure 2A). The distribution of cell proportions was relatively balanced between the two groups (Figure 2B). Within the immune cells, we identified Cd86⁺ M1 macrophages and Cd163⁺ M2 macrophages. Among parenchymal cells, we annotated epithelial cells with basal characteristics (Krt5 and Krt14) and epidermal spinous cells (Krt1 and Krt10) (Figure 2C). Furthermore, three distinct types of hair follicle (HF) cells were characterized: HF_1 (Sostdc1⁺/Fst⁺), located in the bulge region of the hair follicle, bears the closest resemblance to stem cells; HF_2 (Krt17⁺/Krt79⁺), enriched in the isthmus, represents terminally differentiated hair follicle keratinocytes; and HF_3 (Scd1⁺/Fabp5⁺/Mgst1⁺), associated with regions adjacent to sebaceous glands. HFSCs exhibited high expression of markers such as Postn, Angptl7, Cd34, and Grem1.²⁸ The overall gene expression profile of the skin revealed a higher SASP score in the Dmp1^cre-DTA^ki/wt group (Figure 2D). Differential gene expression analysis revealed a general downregulation of spinous epithelial markers (Krt1 and Krt10) in the Dmp1^cre-DTA^ki/wt mice (Figure 2E). Immunofluorescence staining for Krt10 was performed on dorsal skin sections to further validate these transcriptomic findings. Consistent with the single-cell RNA sequencing results, Krt10 expression in the epidermis of Dmp1^cre-DTA^ki/wt mice was markedly reduced compared with WT controls (Figures 2F and 1G). These markers were involved in desmosome formation and cell adhesion, contributing to the mechanical strength and stability of the epidermis, and their downregulation may have an impact on skin thickness and hair regeneration.

Figure 2.

ScRNA-seq of Dmp1^cre-DTA^ki/wt mice skin

(A) UMAP plot of major cell types in the skin.

(B) Cell type proportions in WT and Dmp1^cre-DTA^ki/wt mice.

(C) Dot plot shows the canonical markers in skin cell types.

(D) Gene set score and distribution of single cells in the skin of mouse SASP-associated genes between WT and Dmp1^cre-DTA^ki/wt mice. Dotted line indicated mean score.

(E) Differentially expressed genes in selected cell types between WT and Dmp1^cre-DTA^ki/wt mice.

(F and G) Immunostaining of Krt10 in the epidermis and quantification of relative immunofluorescence intensity in the epidermis (n = 6). Scale bars, 20 μm.

Bone marrow showed the potential to regulate skin aging through macrophages

To determine the upstream cause of this phenotype, we next focused on the bone marrow, as it was one of the major compartments affected by osteolineage cells ablation. Aged bone marrow can induce the aging of peripheral tissues and organs, among which bone marrow macrophages are the main factor driving the aging process.^29,30 To further investigate whether bone marrow exerts potential regulatory effects on peripheral cutaneous tissues following osteolineage cells ablation in mice, we initially validated via bone marrow transplantation experiments that bone marrow cells can traffic into cutaneous tissues. Notably, bone marrow-derived macrophages in the skin tissue progressively increased at 1, 2, and 4 weeks post-transplantation (Figure 3A). In bone marrow-transplanted recipient mice, immunofluorescence co-staining of skin sections for F4/80 with CD86 or CD163 identified GFP⁺ bone marrow-derived macrophages (GFP⁺CD86⁺ and GFP⁺CD163⁺; white arrows) as well as GFP⁻ tissue-resident macrophages (GFP⁻CD86⁺ and GFP⁻CD163⁺; red arrows) (Figure S2A). These results indicated that infiltrating bone marrow-derived cells contribute to the cutaneous macrophage pool and can display CD86⁺/CD163⁺ M1/M2-like phenotypes.

Figure 3.

ScRNA-seq of Dmp1^cre-DTA^ki/wt mice bone marrow

(A) Immunofluorescence staining of donor-derived cells (GFP⁺) from CAG^cre; mT/mG mice and the macrophage marker F4/80 in skin tissues after bone marrow transplantation, and quantification of the ratio of GFP⁺F4/80⁺ double-positive cells relative to total F4/80⁺ macrophages. Scale bars, 20 μm.

(B) UMAP plot of major cell types in the bone marrow.

(C) Cell type proportions in WT and Dmp1^cre-DTA^ki/wt mice.

(D) Dot plot shows the canonical markers in bone marrow cell types.

(E) Top 10 ranked GO and KEGG enrichment terms and associated differentially expressed genes in bone marrow macrophages between WT and Dmp1^cre-DTA^ki/wt mice.

(F) Ridge plot of representative differentially expressed genes in bone marrow macrophages.

(G) Gene set score and distribution of single cells in bone marrow of mouse SASP-associated genes between WT and Dmp1^cre-DTA^ki/wt mice. Dotted line indicated mean score.

Osteolineage cells are known to regulate bone marrow niche cells through both endocrine and paracrine mechanisms.³¹ To investigate the changes in bone marrow following osteolineage cells ablation, we performed scRNA-seq analysis on bone marrow cells from Dmp1^cre-DTA^ki/wt and WT mice. After quality control of sequencing data and gene expression matrices, this analysis yielded a total of 21,233 single cells. Downstream analysis resulted in a uniform manifold approximation and projection (UMAP) plot displaying 15 major cell types with distinct expression profiles after annotation (Figure 3B).³² The canonical immune cell markers were selected for cell type annotation (Figure 3C). Single-cell profiling revealed comparable total cell numbers between the two groups (Figure 3D), with notable lineage-specific alterations of B lineages, characterized by an enrichment of early-stage subsets, including naive and immature B cells. This skewed distribution suggests a potential differentiation block in B cell development. Conversely, the proportions of myeloid cells, including neutrophils, macrophages, and CD86⁺ M1 macrophages, were increased. The proportions of other cell types, such as hematopoietic stem and progenitor cells (HSPCs), DCs, and T cells, showed no substantial changes. These findings indicate a selective impact on lymphoid and myeloid lineages, while preserving HSPC and certain immune cell compartments in Dmp1^cre-DTA^ki/wt mice.

To assess whether specific functions were perturbed in Dmp1^cre-DTA^ki/wt mice, we performed variable gene expression analysis of specific cell types and function enrichment. Comprehensive differential gene expression analysis of various immune cell types revealed macrophages as the most affected. Gene ontology (GO) enrichment analysis in the Dmp1^cre-DTA^ki/wt group indicated the upregulation of functions related to enhanced translation, along with leukocyte chemotaxis and migration (Figure 3E). Notably, metabolic alterations included the upregulation of positive regulation of reactive oxygen species metabolic processes, contrasting with downregulated processes such as oxidative stress response and glycosphingolipid catabolic process (regulated by Psap and Prkcd) in macrophages. Furthermore, KEGG pathway enrichment suggested upregulation in osteoclast differentiation (regulated by Mmp8, Sirpb1c, Fos, and Junb) and the IL-17 signaling pathway.³³ There was a decline in antigen presentation functions of macrophages in Dmp1^cre-DTA^ki/wt mice. Representative differentially expressed genes in macrophages, such as Retnlg, Mmp8, and Chil3, were implicated in the regulation of chronic inflammation, osteoclast precursor cell migration and adhesion, and macrophage polarization (Figure 3F).^34,35 Dysregulated glycolipid metabolism contributed to a pro-inflammatory, pro-senescent microenvironment. In Dmp1^cre-DTA^ki/wt mice, stromal alterations drove macrophage dysfunction, facilitating osteoclast differentiation and chronic inflammation, thereby accelerating overall bone marrow aging. This is further reflected in the enhanced SASP observed within Dmp1^cre-DTA^ki/wt group (Figure 3G).³⁶ Taken together, our findings demonstrated that osteolineage cells ablation induced distinct compositional and functional changes in bone marrow macrophages. These altered macrophages not only contributed to the aging of the bone marrow itself but also exerted potential regulatory effects on peripheral skin tissue aging, suggesting a mechanistic link between skeletal health and systemic tissue aging.

Aging-associated effects on skin

We further evaluated the previously reported epidermal (Figure S3A) and HF (Figure S3B) marker genes³⁷ in the two groups, which were consistent with the expression patterns between young and aged mice. Functional perturbations in skin parenchymal cells indicated a downregulation of epithelial keratinization and oxidative phosphorylation within the HF (Figure S3C). Consistent with the downregulation of Krt1 and Krt10, there was a downregulation of desmosome organization and epithelial development in the epidermis. Fibroblasts and endothelial cells exhibited decreased intercellular junctions, lymphatic vessel development, and ion transport, respectively. These results suggested the presence of an aging phenotype characterized by impaired proliferation, repair, and hair development in the skin. In the HFSCs, Dmp1^cre-DTA^ki/wt mice displayed a downregulation of the Wnt pathway, which is related to stem cell proliferation and differentiation, alongside epithelial development and intercellular adhesion (Figure 4A). To validate these transcriptomic findings, quantitative PCR analysis was performed for key Wnt pathway markers, including Axin2, Sfrp1, and Wnt4a. Consistent with the single-cell sequencing data, Axin2 and Wnt4a were significantly downregulated, whereas the Wnt inhibitor Sfrp1 was upregulated in the dorsal skin of Dmp1^cre-DTA^ki/wt mice (Figure 4B). Gene set variation analysis (GSVA) analysis of hallmark gene sets in HFSCs revealed down-regulation of oxidative phosphorylation and up-regulation of inflammatory pathways (Figure 4C). GSVA analysis of the KEGG pathways suggested a process of metabolic reprogramming within the HFSCs in the Dmp1^cre-DTA^ki/wt group. There was an upregulation in metabolic pathways related to fatty acid metabolism, phenylalanine metabolism, taurine and hypotaurine metabolism, and cysteine and methionine metabolism. However, the TCA cycle energy metabolic pathway was impeded (Figure 4D). In macrophages aligned with these alterations, there was an upregulation of the acute inflammatory response pathway corresponding with HF changes (Figure S3D). Inflammatory response pathways in the epithelium were also elevated in the Dmp1^cre-DTA^ki/wt group (Figure 4E). An increased SASP score was observed in various epithelial cells and fibroblasts, including M1 macrophages (Figure S3E).

Figure 4.

Integration of scRNA-seq results in Dmp1^cre-DTA^ki/wt and aging mice skin

(A) Top 10 ranked GO enrichment terms and associated differentially expressed genes in skin HFSCs between WT and Dmp1^cre-DTA^ki/wt mice.

(B) RT-qPCR of Axin2, Sfrp1, and Wnt4a expression in WT and Dmp1^cre-DTA^ki/wt mice dorsal skin (n = 5). Expression was normalized to β-actin and plotted relative to WT.

(C) Heatmap shows the GSVA results of hallmark genes in HFSCs WT and Dmp1^cre-DTA^ki/wt mice.

(D) Heatmap shows the GSVA results of KEGG pathway genes in HFSCs between WT and Dmp1^cre-DTA^ki/wt mice.

(E) Single-cell GSEA enrichment of inflammatory pathways in skin epithelium between WT and Dmp1^cre-DTA^ki/wt mice.

(F) Dot plots shows the enrichment of regulons in HFSC (up) and HF (later in discussion).

(G) Heatmap shows the expression correlation between Dmp1^cre-DTA^ki/wt and aging mice skin cell types.

(H) Network shows the best hit correlation between Dmp1^cre-DTA^ki/wt and aging mice skin cell types (Ge_2020_Old and Tabula_2020_Old).

(I) Quantitative expression levels of Krt1 and Krt10 in the epidermal tissue of aged mice.

Subsequently, we utilized SCENIC to enrich the analysis of transcription factor (TF) activity within skin parenchymal cells (Figure S3F).³⁸ In HFSCs, the Zfp605 and Lhx2, involved in cell cycle regulation and stem cell differentiation, and Nfatc1, which modulates stem cell quiescence/proliferation and inflammation, were relatively downregulated in the Dmp1^cre-DTA^ki/wt group (Figure 4F). Similarly, in normal HF, a series of transcription factors related to follicle proliferation, immune cell activation, and inflammation regulation also showed decreased activity. In fibroblasts, TFs associated with lineage differentiation, such as Shox2 (skeletal), Vax2 (neural), and Rarb (retinoic acid metabolism related to skin aging), were relatively downregulated in the Dmp1^cre-DTA^ki/wt group (Figure S3G). In endothelial cells (ECs), while the activity of Hoxd1/9, enriched in the Dmp1^cre-DTA^ki/wt group, increased during skin aging, Sox18, which is involved in the endothelial-mesenchymal transition during aging, was also enriched in the Dmp1^cre-DTA^ki/wt mice.^39,40 These findings collectively supported the transcriptional regulatory patterns across different cell types that underpin the phenotypic changes associated with skin aging in the Dmp1^cre-DTA^ki/wt group. Finally, we integrated expression profile data from two previously reported studies on aged mouse skin to explore the correlation of cell types within the Dmp1^cre-DTA^ki/wt mice (Figures 4G and 4H).^37,41 The high AUROC clustering results suggested that HFSCs, along with epithelial epidermal, basal, and proliferative cells, exhibited expression characteristics similar to those of corresponding cell types in aged skin data. The expression of Krt1 and Krt10 was also decreased in aging mouse skin epithelium (Figure 4I). These findings revealed that Dmp1^cre-DTA^ki/wt mice exhibit transcriptional and functional alterations in skin parenchymal cells consistent with aging phenotypes, including impaired regeneration, increased inflammation, and metabolic reprogramming.

Discussion

Our study demonstrates that osteolineage cells ablation plays a regulatory role in skin aging, with bone marrow macrophages potentially acting as intermediaries in this process. Partial ablation of osteolineage cells in Dmp1^cre-DTA^ki/wt mice led to phenotypes associated with skin aging, including dermal thinning, delayed wound healing, and impaired hair regeneration. These phenotypes may be driven by myeloid skewing and a pro-inflammatory shift in bone marrow macrophages following osteolineage cells ablation, suggesting a systemic mechanism by which bone-derived signals influence peripheral tissue aging.

Bone and skin share similar developmental origins, anatomical structures, and immune environments. As the largest organ of the human body, the skin is subjected to both extrinsic (e.g., UV radiation, smoking) and intrinsic factors that drive aging. Skin aging, as part of systemic aging, is modulated by multiple organs and tissues.⁴² For example, insulin-like growth factor 1 (IGF-1), which is synthesized by the liver, skin, and resident fibroblasts in other connective tissues, declines with natural aging.⁴³ IGF-1 deficiency inhibits cellular growth in connective tissue-rich organs such as bone and skin by reducing ribosome biogenesis and protein translation, ultimately leading to tissue atrophy and aging.⁴⁴ In postmenopausal women, the loss of 17β-estradiol is a major intrinsic factor contributing to skin aging.⁴⁵ This hormonal decline results in reduced collagen content and increased skin dryness. Approximately 30% of skin collagen is lost in early menopause, followed by an annual decline of about 2%.⁴⁶ Notably, the rate of skin thinning mirrors the loss of bone mineral density in postmenopausal women, further supporting the hypothesis of a reciprocal regulatory relationship between bone and skin. Our findings support this notion, showing that osteolineage cells can modulate skin aging via bone marrow macrophages. However, this regulatory axis likely represents only part of the complex systemic network influencing skin aging. Other systemic alterations induced by osteolineage cells ablation—such as in the brain or gonads—may also contribute to skin aging and warrant further investigation. Furthermore, this mechanism requires further validation in additional models of osteolineage cells reduction, such as ovariectomized mice and naturally aged mice. In addition to targeting osteocytes, osteoblasts located on the bone surface, and preosteoblasts at the cartilage-bone junction of the metaphysis, Dmp1 is also expressed in several non-skeletal tissues. Previous studies have reported that Dmp1^cre can label cells in extra-skeletal tissues such as skeletal muscle fibers, specific cell populations in the cerebellum and hindbrain, as well as Pdgfra⁺ mesenchymal cells in the gastrointestinal tract.⁴⁷ Therefore, we cannot fully exclude the possibility that the ablation of Dmp1-positive cells may lead to severe dysfunction in other organs, which could secondarily impact skin aging.

Mechanistically, osteolineage cells ablation in Dmp1^cre-DTA^ki/wt mice induces systemic aging features, including severe sarcopenia, osteoporosis, and kyphosis, ultimately leading to shortened lifespan.¹⁸ These systemic manifestations suggest that osteolineage cell loss exerts widespread aging effects, rather than being confined to the skeletal system. Macrophages play a critical role in aging regulation.^48,49 They are major contributors to age-associated inflammation, and aging alters macrophage phenotypes and functions, affecting processes such as phagocytosis, wound healing, and polarization. Senescent macrophages secrete a spectrum of pro-inflammatory factors collectively known as the SASP, which exacerbates chronic inflammation in aged tissues.^50,51 We propose that osteolineage cells ablation-induced alterations in bone marrow macrophages—specifically their inflammatory shift—mediate the effects of osteolineage cells on skin aging. Indeed, our findings indicate a clear pro-inflammatory transition in bone marrow macrophages and a concomitant myeloid skewing in Dmp1^cre-DTA^ki/wt mice. Nevertheless, our study does not fully elucidate the exact mechanisms by which bone marrow macrophages regulate skin aging, highlighting the need for future research in this area. Moreover, osteolineage cells secrete bone-derived factors, often referred to as “osteokines,” which regulate both bone metabolism and systemic homeostasis. Sclerostin (Sost) is a secreted glycoprotein mainly expressed by mature osteocytes and functions as a negative regulator of bone formation. Studies have shown that Sost(−/−) mice exhibit impaired B cell development, characterized by increased apoptosis during early maturation, developmental arrest at the precursor stage, and reduced numbers of mature B cells in the bone marrow. These alterations are directly associated with elevated apoptosis in pre-B, immature, and recirculating B cell populations.⁵² Consistent with these findings, our single-cell sequencing results revealed similar changes in B cell populations within the bone marrow of Dmp1^cre-DTA^ki/wt mice. Moreover, fatty acid-binding protein 3 (FABP3), which is abundantly secreted by senescent osteocytes, can be internalized by dermal fibroblasts and activate the p53/p21 signaling pathway, leading to fibroblast senescence.⁵³ By comparing the expression profiles of osteokines in cortical bone between Dmp1^cre-DTA^ki/wt and WT mice, we found that multiple osteokines were generally downregulated in Dmp1^cre-DTA^ki/wt mice. These results suggest that osteokines may play an important role in mediating both bone marrow alterations and skin aging following osteolineage cells ablation, an aspect that warrants further investigation in future studies.

Skin wound healing is a dynamic process that occurs in three sequential stages: inflammation, re-epithelialization, and tissue remodeling. It involves various cell types, including keratinocytes, fibroblasts, adipocytes, endothelial cells, and macrophages. Aging impairs cellular regeneration capacity, weakens immune function, and alters inflammatory responses. In contrast, hair regeneration follows a cyclical process consisting of three stages, with HFSC function and regenerative capacity declining during aging. This process involves the participation of HFSCs, dermal papilla cells, dermal fibroblasts, endothelial cells, and macrophages. In both models of aging, macrophages play crucial roles. During the early stages of wound healing, monocytes migrate to the injury site in response to chemotactic factors, differentiate into macrophages, and polarize into M1 pro-inflammatory macrophages to clear apoptotic cells and neutrophils. Later, they switch to M2 macrophages, which secrete cytokines and growth factors to mediate tissue repair. In aging, macrophage function is compromised, with a shift toward M1 polarization, leading to increased pro-inflammatory cytokine production and reduced M2 polarization, ultimately impairing tissue repair. In hair follicle cycles, macrophages also exhibit key regulatory roles. Studies have shown that macrophages aggregate around hair follicles and correlate with changes in the hair cycle. They may influence the transition to the catagen phase by secreting proteins such as FGF-5.⁵⁴ Additionally, research has demonstrated that a subset of TREM2+ macrophages in the mouse skin can secrete Oncostatin M to maintain HFSC quiescence and inhibit hair growth.⁵⁵ Moreover, M1 macrophage polarization through IL-18 and IL-1β has been shown to induce HFSC apoptosis, leading to hair loss.⁵⁶ As we have highlighted in our study, although the cell types, processes, and pathways involved in the two different aging phenotypes may differ, the overall decline in multi-cellular function due to aging leads to a generalized aging phenotype. Macrophages, in particular, play an important role in regulating these changes. Our research further supports this by demonstrating that in Dmp1^cre-DTA^ki/wt mice, various skin cell types, including epithelial cells, HFSCs, fibroblasts, and macrophages, undergo functional shifts consistent with the aging phenotype.

In conclusion, we identify a previously unrecognized regulatory axis wherein osteolineage cells ablation influences skin aging via bone marrow macrophages. These findings provide new insights into the systemic regulation of aging and underscore the importance of understanding inter-organ communication in the aging process. They also suggest potential therapeutic targets for the prevention of skin aging and osteoporosis.

Limitations of the study

Although our study identifies a regulatory axis linking osteolineage cells, bone marrow macrophages, and skin aging, this pathway likely represents only one component of a broader systemic aging network. Osteolineage cells ablation induces widespread aging-related phenotypes, and therefore, we cannot fully exclude the possibility that alterations in other organs may indirectly contribute to the observed skin changes. In addition, the use of the Dmp1^cre-DTA^ki/wt model has inherent limitations, as Dmp1 is expressed not only in osteolineage cells but also in several extra-skeletal tissues, raising the possibility of secondary effects resulting from the ablation of Dmp1-positive cells outside the skeleton. Finally, although we observed clear myeloid skewing and a pro-inflammatory shift in bone marrow macrophages, the precise molecular mechanisms by which these altered macrophages mediate skin aging remain to be fully elucidated. Moreover, although both male and female mice were included, the study was not designed or powered to systematically assess sex-specific effects, and potential sex-dependent differences were therefore not examined.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Yi-Gang Huang (yiganghuang@sjtu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The scRNA-seq datasets generated in this study have been deposited in the Gene Expression Omnibus (GEO) and are publicly available upon publication. The RNA-seq data have been deposited in the Gene Expression Omnibus under accession number GSE305068 and are publicly available. ALL accession numbers are provided in the STAR Methods section.
• •This study did not generate new code.
• •Additional information required to reanalyze the data reported in this article is available from the lead contact upon reasonable request.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (No. 82072525 to Y.G.H).

Author contributions

S.Z. performed the experiments and drafted the article; Z.Z., F.Y., B.W., and S.F. analyzed the data; J.J., Y.P., and Y.H. provided feedback on the report. All authors read and approved the final article.

Declaration of interests

No potential conflict of interest was reported by the authors.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit polyclonal anti Ki-67	Proteintech	Cat#28074-1-AP; RRID: AB_2918145
Rabbit polyclonal anti Krt10	Proteintech	Cat#18343-1-AP; RRID: AB_1086365
Mouse polyclonal anti Krt14	Santa Cruz Biotechnology	Cat#sc-53253; RRID: AB_2134820
Rat monoclonal anti F4/80	Abcam	Cat#ab6640; RRID: AB_1140040
Mouse monoclonal anti CD86	Proteintech	Cat#83213-6-RR; RRID: AB_3670898
Rabbit monoclonal anti CD163	Abcam	Cat#ab316218
Donkey anti-rabbit IgG (H+L), Alexa Fluor 488	Invitrogen	Cat#A-21206; RRID: AB_2535792
Donkey anti-mouse IgG (H+L), Alexa Fluor 488	Invitrogen	Cat#A-32766; RRID: AB_2762823
Donkey anti-rabbit IgG (H+L), Alexa Fluor 647	Invitrogen	Cat#A-31573; RRID: AB_2536183
Donkey anti-mouse IgG (H+L), Alexa Fluor 647	Invitrogen	Cat#A-32787; RRID: AB_2762830
Goat anti-rat IgG (H+L), Alexa Fluor 555	Invitrogen	Cat#A-48263; RRID: AB_2896332
Chemicals, peptides, and recombinant proteins
EdU(5-ethynyl-2' -deoxyuridine)	Beyotime	Cat#ST067
Hematoxylin	Servicebio	Cat#G1004
Eosin	Servicebio	Cat#G1001
Dapi	FUSHENBIO	Cat#FS1206
ProLong™ Diamond Anti-Quenching Mounting Medium	Invitrogen	Cat#P36970
BSA	Sigma	Cat#SRE0096
Triton X-100	Solarbio	Cat#T8200
Avertin	Sigma	Cat#T48402
2 × Color SYBR Green qPCR Mix	EZBioscience	Cat#A0001-R1
Critical commercial assays
Click-iT™ EdU Imaging Kit	Beyotime	Cat#C0071S
Tissue RNA Purification Kit PLUS	EZBioscience	Cat#EZB-RN001-plus
4× Reverse Transcription Mix	EZBioscience	Cat#EZB-RT2GQ
Deposited data
Single-cell RNA-seq of mouse skin	This paper	GSE305068
Single-cell RNA-seq of bone marrow macrophages	Previously published	GSE202516
Single-cell RNA-seq of aging mouse skin	Previously published	GSE124901; GSE132042
Experimental models: Organisms/strains
C57BL/6J	Charles River	Strain #219
Dmp1cre mice	Dr. J. Q. (Jerry) Feng	Strain #023047
Ai9 mice	The Jackson Laboratory	Strain #007909
MT/mG mice	The Jackson Laboratory	Strain #007676
Rosa26em1Cin (SA-IRES-Loxp-ZsGreen-stop-Loxp-DTA) heterozygous mice	The Jackson Laboratory	Strain #T009408
CAG cre mice	GemPharmatech	Strain #T004055
Oligonucleotides
β-actin forward: 5′-CATTGCTGACAGGATGCAGAAGG-3′	This paper	–
β-actin reverse: 5′-GTTCCACAGGCGTCATCTCCTT-3′	This paper	–
Wnt4a forward: 5′-GAGAACTGGAGAAGTGTGGCTG-3′	This paper	–
Wnt4a reverse: 5′-CTGTGAGAAGGCTACGCCATAG-3′	This paper	–
Sfrp1 forward: 5′-CAATACCACGGAAGCCTCTAAGC-3′	This paper	–
Sfrp1 reverse: 5′-GCAAACTCGCTTGCACAGAGATG-3′	This paper	–
Axin2 forward: 5′-ATGGAGTCCCTCCTTACCGCAT-3′	This paper	–
Axin2 reverse: 5′-GTTCCACAGGCGTCATCTCCTT-3′	This paper	–
Software and algorithms
ImageJ	Schneider et al.	https://imagej.net/ij/
GraphPad Prism 10	GraphPad Software	https://www.graphpad-prism.cn/

Experimental model and study participant details

All mouse lines were maintained on a C57BL/6J genetic background. Dmp1^cre mice were obtained from Dr. J. Q. (Jerry) Feng at Texas A&M College of Dentistry (Jackson Laboratory, stock no. 023047). Rosa26^{em1Cin (SA-IRES-Loxp-ZsGreen-stop-Loxp-DTA)} heterozygous mice were sourced from GemPharmatech (strain ID: T009408). To generate osteolineage cells ablation models during development, Dmp1^cre mice were crossed with Rosa26^{em1Cin (SA-IRES-Loxp-ZsGreen-stop-Loxp-DTA)} homozygotes, producing Dmp1^cre Rosa26^{em1Cin (SA-IRES-Loxp-ZsGreen-stop-Loxp-DTA)} (Dmp1^cre-DTA^ki/wt). We used B6.Cg-Gt (ROSA)26Sortm9 (CAG-tdTomato) Hze/J mice (commonly referred to as Ai9 or Ai9 (RCL-tdT); Strain #: 007909; RRID: IMSR_JAX:007909) mice and mT/mG reporter mice (B6.129 (Cg)-Gt (ROSA) 26Sor-tm4 (ACTB-tdTomato,-EGFP) Luo/J, JAX #007676) for this study. MT/mG mice were crossed with CAG^cre transgenic mice to generate CAG^cre; mT/mG offspring. Dmp1^cre Ai9^fl/fl mice and CAG^cre; mT/mG mice were raised to eight weeks of age. All animal procedures were approved by the Animal Care and Use Committee of Shanghai Sixth People’s Hospital (permit numbers: 2021-0028). All procedures were performed in accordance with institutional guidelines to minimize animal suffering. For interventions requiring anesthesia, mice were anesthetized with isoflurane or Avertin (1.25% 2,2,2-tribromoethanol and 2.5% 2-methyl-2-butanol; Sigma).

Method details

Epilation and hair regeneration analysis

For depilation-induced hair regrowth assays, dorsal hair was initially trimmed using electric clippers, followed by complete removal with depilatory cream applied to the shaved area for exactly 5 minutes before being wiped clean. Standardized photographs were taken on the day of depilation (day 0) and at subsequent time points. Hair regrowth was quantified either by measuring gray intensity using ImageJ software or by calculating the percentage of pigmented skin relative to the baseline on day 0. Stages of hair follicles were determined based on established morphological criteria from previous studies.¹⁹

Wounding and wound-healing experiments

Before surgery, mice were anesthetized with Avertin. Dorsal hair was removed using depilatory cream, and a full-thickness excisional wound was created on the dorsal skin using an 8-mm biopsy punch. Wound images were captured at the indicated time points using a digital camera, and wound areas were quantified using ImageJ.

Histological sections and staining

Mouse femoral tissue was fixed in 4% paraformaldehyde at 4 °C for 48 hours, followed by decalcification in 10% ethylenediaminetetraacetic acid (EDTA, pH 7.2) at 4 °C for one week, with the decalcification solution refreshed every other day. Skin tissue samples were fixed in 4% paraformaldehyde at 4 °C for 24 hours. Following fixation, all tissues underwent a graded ethanol dehydration process, were embedded in paraffin, and sectioned at a thickness of 5 μm. Hematoxylin and eosin (H&E) staining was performed according to standard protocols. Skin thickness was measured using ImageJ software.

Immunofluorescence

Fixed tissue samples were cryoprotected in 15% and 30% sucrose solutions (w/v in PBS) for 24 h each at 4 °C. Samples were then embedded in optimal cutting temperature (OCT) compound and cryosectioned at a thickness of 12 μm using a cryostat at −20 °C. Frozen sections were washed three times with phosphate-buffered saline (PBS) for 5 min each, permeabilized with 0.1% Triton X-100 for 10 min, and blocked with 3% bovine serum albumin (BSA) in PBS for 2 h at room temperature. Sections were incubated overnight at 4 °C with primary antibodies, followed by fluorescently labeled secondary antibodies for 2 h at room temperature. After final PBS washes, sections were mounted with fluorescence-compatible mounting medium and imaged under a confocal microscope.

EdU labeling assay

To identify proliferating cells in the skin, 5-ethynyl-2'-deoxyuridine (EdU) was administered intraperitoneally on day 3 following depilation. EdU was dissolved in sterile PBS and injected intraperitoneally at a dose of 25 mg/kg body weight, twice within a 24-hour period prior to tissue collection. Skin tissues were harvested 24 hours after the first injection. EdU incorporation was detected using the Click-iT™ EdU Imaging Kit (Biyotime, C0071S) according to the manufacturer’s instructions.

RNA extraction and real-time PCR

Total RNA was extracted from skin tissue using a tissue RNA extraction kit (EZbioscience, Cat. No. EZB-RN001-plus) according to the manufacturer’s instructions. For reverse transcription, 1,000 ng of total RNA was converted into cDNA using a 4× Reverse Transcription Mix (EZbioscience, Cat. No. EZB-RT2GQ). Quantitative real-time PCR (qPCR) was performed with a 2× SYBR Green qPCR Mix (EZbioscience, Cat. No. A0001-R1) following the manufacturer’s recommendations on a QuantStudio™ 7 Flex Real-Time PCR System (Thermo Fisher Scientific). Relative gene expression levels were calculated using the 2^ΔΔCT method and normalized to β-actin expression. All primers were synthesized by Tsingke Biotechnology company. The primer sequences used in this study were as follows:

Axin2 forward: 5′-ATGGAGTCCCTCCTTACCGCAT-3′; reverse: 5′-GTTCCACAGGCGTCATCTCCTT-3′.

Sfrp1 forward: 5′-CAATACCACGGAAGCCTCTAAGC-3′; reverse: 5′-GCAAACTCGCTTGCACAGAGATG-3′.

Wnt4a forward: 5′-GAGAACTGGAGAAGTGTGGCTG-3′; reverse: 5′-CTGTGAGAAGGCTACGCCATAG-3′.

β-actin forward: 5′-CATTGCTGACAGGATGCAGAAGG-3′; reverse: 5′-TGCTGGAAGGTGGACAGTGAGG-3′.

Bone marrow transplantation

Seven days prior to bone marrow transplantation, ciprofloxacin was added to the drinking water of all male wild-type (WT) recipient mice at a concentration of 1 g/1.4 L of water. Recipient mice were subjected to total body irradiation using an RS2000 irradiator, receiving a total dose of 9 Gy administered in two fractions of 4.5 Gy each, with a 2-hour interval between doses. Donor male CAG^cre; mT/mG mice were euthanized, and their femurs and tibias were harvested under sterile conditions. The proximal and distal epiphyses were removed with sterile scissors, and bone marrow was flushed from the medullary cavities using cold sterile PBS. The isolated bone marrow cells were suspended in sterile PBS at a concentration of 5.0 × 10⁷ cells/mL. A total volume of 200 μL of the cell suspension (equivalent to 1.0 × 10⁷ cells) was administered via tail vein injection into each irradiated recipient mouse, 24 hours post-irradiation. Following transplantation, ciprofloxacin-supplemented water was continued for an additional two weeks.

Single-cell collection, library construction, and sequencing

Bone marrow

Previously published single-cell RNA sequencing (scRNA-seq) data of bone marrow cells from Dmp1^cre-DTA^ki/wt and WT mice were obtained from our earlier study (GEO accession number: GSE202516). Briefly, bone marrow cells were flushed from femurs, filtered through a 70 μm cell strainer, and red blood cells were removed. Single-cell suspensions were stained with AO/PI for viability assessment.¹⁸ Bone marrow was flushed and kept in MACS Tissue Storage Solution (Miltenyi Biotec). Samples were filtered through 40 μm cell strainers and centrifuged at 300×g for 5 min. Pelleted cells were suspended in red blood cell lysis buffer (Miltenyi Biotec) to lyse red blood cells.

Skin

For skin scRNA-seq, dorsal skin tissues were collected from mice after depilation and careful removal of subcutaneous fat. The tissues were immediately transferred to ice-cold, RNase-free 1× PBS without calcium and magnesium and cut into ∼0.5 mm² fragments. Fragments were washed thoroughly to remove surface blood and lipid residues, then incubated in a dissociation solution containing 0.35% collagenase IV, 2 mg/mL papain, and 120 U/mL DNase I at 37°C for 20 minutes with gentle agitation (100 rpm). Digestion was stopped by adding 10% fetal bovine serum (FBS) in PBS, followed by gentle trituration (5–10 times) to promote dissociation while minimizing shear stress. The suspension was filtered sequentially through 70 μm and 30 μm strainers and centrifuged at 300 × g for 5 minutes at 4°C. The cell pellet was resuspended in 1× PBS containing 0.04% BSA. Red blood cells were lysed using 10× RBC lysis buffer (MACS, Cat#130-094-183), followed by two rounds of washing and centrifugation (300 × g, 3 minutes, 4°C). Dead cells were removed using Dead Cell Removal MicroBeads (MACS, Cat#130-090-101) according to the manufacturer’s protocol. The final cell suspension was resuspended in 1× PBS (0.04% BSA), and cell viability (>85%) was confirmed by trypan blue staining. Cell counts were determined using a hemocytometer, and the concentration was adjusted to 700–1200 cells/μL.

ScRNA-seq

For bone marrow and skin samples, single-cell suspensions were loaded onto the 10× Genomics Chromium system (Chromium Single Cell 3′ Reagent Kits (v3)). Gel beads in emulsion (GEM) generation, barcoding, reverse transcription, and library construction were performed following standard protocols. Libraries were sequenced on an Illumina NovaSeq 6000 system.

Processing of scRNA-seq data

Raw sequencing data were processed using Cell Ranger (v3.1.0) with default parameters and aligned to the mouse mm10 reference genome (Ensembl annotation). Cells with fewer than 500 detected genes or with mitochondrial UMI counts >15% were excluded from further analysis. Downstream analyses were performed using Seurat (v4.4.0) in R. Data were normalized and scaled after regressing out total UMI counts and mitochondrial gene percentage. The top 2,000 highly variable genes were identified using the “FindVariableFeatures” function. Batch effects across samples and conditions were corrected using Harmony integration.⁵⁷ Principal component analysis (PCA) was performed, and the top 20 principal components were used for clustering with the “FindClusters” function (resolution = 0.5). Dimensionality reduction and visualization were achieved using Uniform Manifold Approximation and Projection (UMAP).

Cell type identification, differential gene expression, and functional enrichment analysis

For bone marrow and skin samples, cell types were identified using canonical marker genes obtained from publicly available online databases and published literature. The same general strategy was applied to bone marrow samples for annotation of major hematopoietic and stromal lineages. Differentially expressed genes (DEGs) between groups or clusters were identified using the “FindMarkers” function in Seurat with the Wilcoxon rank-sum test. A threshold of absolute log fold change > 0.25 and adjusted p-value < 0.05 was applied, and only genes expressed in at least 10% of cells in either group (min.pct > 0.1) were considered significant. Gene function enrichment analyses were performed using the clusterProfiler R package and Metascape to identify significantly enriched Gene Ontology (GO) terms and KEGG pathways.^58,59 To evaluate pathway activity and SASP–related signatures at the single-cell level, Gene Set Variation Analysis (GSVA) and single-cell gene set enrichment analysis were performed using the GSVA R package in combination with gene sets retrieved from the MSigDB database. The Wilcoxon test was adopted for statistical comparisons of enrichment scores between WT and Dmp1^cre-DTA^ki/wt groups, and p-values < 0.05 were considered statistically significant.

TF analysis

SCENIC (python version) was used to perform transcription factor enrichment analysis.³⁸ We combined the AUCell score matrices with the orthologous TFs and the matrices of identified TFs. We identified regulon specificity scores with default pipeline and visualized them using dot plots.

Correlation analysis

Single-cell gene expression correlation across skin cell types and published datasets mentioned in the results were performed using MetaNeighbour with the top 1,000 variable genes.⁶⁰ The AUROC correlation was visualized in a heatmap. Publicly available single-cell RNA-seq datasets (GEO accession numbers: GSE124901 and GSE132042) were included for comparative correlation analysis.^37,41

Quantification and statistical analysis

For each experiment, the exact value of n and what n represents are indicated in the corresponding figure legends. All data were analyzed using GraphPad Prism (v9.0.2) software for statistical significance. Statistical significance was assessed using unpaired two-tailed Student’s t-tests or one-way ANOVA with appropriate multiple-comparisons tests. Data are summarized as mean ± SEM for statistical analysis. Differences were considered statistically significant at P < 0.05, and exact P values are reported where applicable.

Published: February 2, 2026