A total of eight wild-type C57BL/6J mice, obtained from Beijing Beiyou Biotechnology Co., Ltd. (batch no. XH202404160001), included five 2-month-old (young; weight: 22±1.2 g) and three 22-month-old (aged; weight: 35±2.3 g) male mice for the present study. Young male mice were randomly housed in groups on more than three per cage, while aged mice were individually housed to prevent potential conflicts. All the experimental animals were raised in the Specific Pathogen Free (SPF) facility, where the breeding environment was maintained at a temperature 22–25°C, with a humidity range of 50–70%. All mice had free access to food and water. Additionally, the environment was regulated by a 12-h light/dark cycle. The full-thickness skin wound with a diameter of 4 mm was generated on the dorsal skin of mice by biopsy punch, after anesthetized by inhalation of isoflurane (3%). The wound healing progress was measured through digital photography at days 0, 2, 4 and 7, respectively. At the predetermined end of the experiment on day 7, under the same anesthetic conditions, the back skin samples were collected from the wound sites of all mice. Following sample collection, the wounds were dressed and sutured. After incising the back skin and dissecting the subcutaneous tissue, the skin flap was everted to reveal the details of the underlying subcutaneous vascular network, which were then captured using digital photography. All animal experimental procedures were approved by Institutional Animal Care and USE Committee of Chinese PLA General Hospital (Beijing, China; approval no. 2023-407-01) and followed the relevant ethical regulations.

A total of two wild-type neonatal C57BL/6J male mice (weight: 1.4±0.1 g) were obtained from Beijing Beiyou Biotechnology Co., Ltd. (batch no: XH202504300004). Neonatal mice were euthanized by exposure to carbon dioxide for 1 h and immersed in 75% ethanol for 15 min. Following dorsal skin removal, dermal-epidermal separation was performed. The dermal tissue was minced and digested with 0.25% Type I collagenase at 37°C for 30 min. The resultant cell suspension was sequentially filtered through a 40-µm nylon mesh, centrifuged at 200 × g for 10 min at room temperature and resuspended in DMEM medium (Gibco; Thermo Fisher Scientific, Inc.). Fibroblasts were maintained at 37°C/5% CO₂ and passaged at 85% confluence.

For endothelial cell isolation, the initial dermal digestate dermal cells were resuspended in EBM-2 medium supplemented with 1 µl/ml puromycin (MilliporeSigma) and maintained for 3 days under 5% CO₂ at 37°C. After 72-h selection culture under standard conditions (37°C/5% CO₂), cells were transitioned to puromycin-free EBM-2 complete medium with medium renewal every 48 h. The cells were passaged when the confluence rate reached 85%.

Second-passage mouse primary dermal fibroblasts were subjected to senescence induction through chronic hyperglycemic stimulation. Cells were maintained under standard conditions (37°C/5% CO₂) in high-glucose DMEM (Gibco; Thermo Fisher Scientific, Inc.) at final concentrations of 25 (control), 50 and 75 mM for 7 days with medium renewed every 48 h.

A Transwell chamber system (cat. no. 3413; Corning, Inc.) was employed to investigate ECs-Fibs interactions. Second-passage ECs were seeded in the upper chambers (0.4 µm pore size) at a density of 2×10⁴ cells/well. In the lower chambers, two fibroblast populations were respectively cultured: High glucose-induced senescent fibroblasts and non-treated second-passage fibroblasts. The co-culture system was maintained in a 1:1 mixture of DMEM (Gibco; Thermo Fisher Scientific, Inc.) and EBM-2 (Lonza Group, Ltd.) under standard conditions (37°C; 5% CO₂) for 72 h. Fibroblasts from the lower chambers were subsequently harvested for phenotypic characterization.

After being fixed in a 4% paraformaldehyde solution for a 24-h period at room temperature, the wound tissues were subjected to gradient dehydration and paraffin embedding, following standard protocols. Subsequently, the tissues were sectioned into 5-µm-thick slices. Hematoxylin and eosin (H&E) staining was performed in accordance with standard procedures. First, the slides were baked at 60°C for 2 h. Then, dewaxing was performed in xylene at room temperature twice for 10 min each. This was followed by rehydration through a graded ethanol series (100, 95 and 70%) at room temperature for 3 min each. Next, staining was performed in 10% hematoxylin at room temperature for 10 min. Differentiation was achieved in 1% acid alcohol for 10 sec, followed by bluing in 0.2% ammonia water for 1 min. Counterstaining was done in eosin Y at room temperature for 1 min. Dehydration was performed through a graded ethanol series (70, 95, and 100%) for 30 sec each, and then clearing was done in xylene twice for 5 min each. Finally, the slides were mounted with resin.

For immunohistochemistry: After antigen retrieval, sections were permeabilized with 0.3% Triton X-100 for 15 min at room temperature, followed by blocking with 5% goat serum (cat. no. SL038; Beijing Solarbio) for 1 h at RT. Sections were incubated with primary antibodies against α-smooth muscle actin (α-SMA; 1:500; cat. no. ab7817; mouse; Abcam) and cluster of differentiation (CD)31 (1:4,000; cat. no. ab281583, rabbit; Abcam) at 4°C for 18 h, followed by Coralite594-conjugated anti-mouse IgG (1:300; cat. no. SA00013-3; Proteintech Group, Inc.) and Coralite488-conjugated anti-rabbit IgG (1:300; cat. no. SA00013-2; Proteintech Group, Inc.) at RT for 2 h. Nuclei were counterstained with DAPI (cat. no. 0100-20; SouthernBiotech) for 10 min at room temperature and visualized using laser scanning confocal microscopy (SP8 Falcon; Leica Microsystems, Inc.) at 40× magnification.

For immunofluorescence: Post antigen retrieval and blocking, sections were incubated with TGF-β1 (1:500; cat. no. 21898-1-AP, rabbit; Proteintech Group, Inc.) and Anti-Smad2 + Smad3 (1:200; cat. no. ab202445, rabbit; Abcam) antibodies at 4°C for 18 h, Sections were washed 3×5 min with PBS before incubation with 647-conjugated goat anti-rabbit IgG (1:1,000; cat. no. ab150083; Abcam) for 2 h at RT protected from light. After secondary incubation, sections were washed 3×5 min with PBS. Nuclei were counterstained with DAPI (cat. no. 0100-20; SouthernBiotech) for 10 min at RT. Fluorescent images were captured post-DAPI staining at 40× magnification using confocal microscopy (SP8 Falcon; Leica Microsystems, Inc.).

Cells were harvested for RNA extraction (1×10⁶ cells/ml). Total RNA was extracted from cells using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. Reverse transcription was performed with the PrimeScript RT reagent kit (Takara Biotechnology Co., Ltd.) following the manufacturer's instructions. The synthesized cDNA was amplified using TB Green Premix Ex Taq II (Takara Biotechnology Co., Ltd.) on a QuantStudio 5 system (Thermo Fisher Scientific, Inc.) with the following cycling conditions: 95°C for 30 s (initial denaturation); 40 cycles of 95°C for 5 s (denaturation) and 60°C for 30 s (annealing/extension) according to the manufacturer's protocol. Gene expression were normalized to GADPH and quantified using the 2-ΔΔCq method (1). These experiments were replicated three times. The primer sequences were: GAPDH: F 5′-AGGTCGGTGTGAACGGATTTG-3′; R 5′-TGTAGACCATGTAGTTGAGGTCA-3′; P16: F5′-CGCAGGTTCTTGTCACTGT-3′; R 5′-TGTTCACGAAAGCCAGAGCG-3′; P21: F5′-CCTGGTGATGTCCGACCTG-3′; R 5′-CCATGAGCGCATCGCAATC-3′; Ki67: F 5′-ATCATTGACCGCTCCTTTAGGT-3′; R 5′-GCTCGCCTTGATGGTTCCT-3′; TGFβ1: F 5′-CTCCCGTGGCTTCTAGTGC-3′; R 5′-GCCTTAGTTTGGACAGGATCTG-3′; Smad2: F 5′-TCCGTACCACTACCAGAGAGT-3′; R 5′-GGCGGCAGTTCTGTTAGAATC-3′; Smad3: F 5′-CACGCAGAACGTGAACACC-3′; R 5′-GGCAGTAGATAACGTGAGGGA-3′ and were synthesized by Beijing Tsingke Biotech Co., Ltd.

The scRNA-seq data were downloaded from the Genome Sequence Archive (GSA; accession no. CRA010641) of the China National Center for Bioinformation. The subset of dorsal wound healing samples were specifically analyzed for aged (O) vs. young (Y) mice, designated as skin_OD0, skin_OD2, skin_OD4, skin_OD7, skin_YD0, skin_YD2, skin_YD4 and skin_YD7, corresponding to post-wounding days 0, 2, 4 and 7.

In order to align reads of raw data and generate feature-barcode matrices, Cell Ranger software (version 4.0.0; 10×genomics.com/support/software/cell-ranger/downloads) was used to perform preliminary processing of each subset. CellBender software (version 0.2.0; http://github.com/broadinstitute/CellBender) was used to remove possible background RNA bias contamination resulting from technical errors. Then, Seurat R package (5.0.3; http://cran.r-project.org/web/packages/Seurat/) (12) was then used to perform subsequent quality control based on the expression matrix obtained. Cells with a number of detected genes <200 or >7,000, or a proportion of reads mapping to the mitochondrial genome exceeding 20, were excluded from further analysis during the quality control step. Finally, DoubletFinder R package (2.0.4; http://github.com/chris-mcginnis-ucsf/DoubleFinder) was used to identify and filter out potential doublets from the sequencing data in each subset.

To effectively merge the eight datasets of interest aforementioned, the Seurat integration algorithm was used. ‘NormalizeData’, ‘FindVariableFeatures’ and ‘ScaleData’ functions from the Seurat R package (5.0.3) were applied to normalize and scale the integrated datasets, following the standard workflow. Principal component analysis (PCA) dimensions were then calculated using the ‘RunPCA’ function from the Seurat R package (5.0.3). To address batch effects among the eight datasets, the Harmony R package (1.2.0; portals.broadinstitute.org/harmony/) was employed. Cluster identification was performed using the ‘FindNeighbors’, followed by the ‘FindClusters’ function with a resolution parameter set to 0.5. For cell type annotation, the ‘FindALLMarkers’ function was leveraged to identify marker genes and their expression patterns in each cluster. Finally, by mapping the results to the CellMaker 2.0 database (117.50.127.228/CellMarker), 14 distinct cell types were successfully identified.

The ‘FindMakers’ function was used to identify DEGs between skin_OD0 and skin_YD0, skin_OD2 and skin_YD2, skin_OD4 and skin_YD4, as well as skin_OD7 and skin_YD7 in Fibroblast, which were based on the Wilcoxon test. The screening criteria for DEGs were selected by |log2FC| >0.25 and adjust P-value <0.05.

To gain insight into the biological functions associated with the previously identified DEGs, the DAVID database (http://david.nicifcrf.gov/summary.jsp) was used to obtain enriched biological functional annotations of these DEGs. The ‘ggplot2’ R package (version 3.5.0; ggplot2.tidyverse.org) was used to visualize the results.

The CellChat R package (version 1.6.1; http://github.com/sqjin/CellChat) was used to calculate and exhibit the cell-cell communication pattern between cell types of interest. In brief, CellChat compares sequencing datasets with the CellChatDB dataset to compute the expression levels of ligand-receptor pair across various cell types. Following this, using the ‘computeCommunProbPathway’ function, it derives the probabilities of intercellular communication based on this expression. Ultimately, the consolidation of these probabilities across distinct communication pathways yields the comprehensive network of intercellular communication.

To comprehensively elucidate signaling changes between young and aged wounds, a comparative cell-cell communication pattern was conducted. Initially, CellChat was independently applied to both young and aged datasets to generate distinct CellChat objects and quantify the cell-cell communication pattern between Fibroblast and Endothelia cell in each condition. Subsequently, those two objects were merged and a comparative analysis was employed to detect significant alteration between young and age wounds.

Differentially expressed genes (DGEs) among various types of cells were identified. Then filtered DEGs were used to compute communication probabilities and enrichment analysis, such as GO. To evaluate the statistical power of RNA-seq experiments across different cell types, power was calculated using the ‘RNASeqPower’ package in R (bioconductor.org/packages/release/bioc/html/RNASeqPower). The results showed the RNA-seq power for EC and Fib were 0.982 and 0.926. The data are presented as means ± standard deviations (SD) and the statistical analysis was performed using GraphPad Prism Software (version 9.5; Dotmatics). Significant differences between groups were determined using two-way analysis of variance (ANOVA) followed by Bonferroni post hoc test. P<0.05 was considered to indicate a statistically significant difference.

Proper vascularization is essential for effective wound healing, as excessive vascularization may lead to pathological scars, whereas insufficient vascularization can delay healing (13). To investigate the disparities in vascular regeneration between aged and young subjects, we established a murine dorsal wound model. The healing rate was markedly higher in young mice than in aged mice (Fig. 1A and B). This disparity was accompanied by distinct subcutaneous neovascular network patterns: aged wounds exhibited reduced capillary length around wound margins (Fig. 1A and D), whereas young wounds developed robust tubular vascular structures as indicated by H&E staining and CD31/SMA co-staining (Fig. 1C and E). Collectively, these findings demonstrated that vascular regenerative capacity is impaired in aged wounds.

To gain a comprehensive understanding of the dynamic cellular and molecular differences between young and elderly mice during wound healing, single-cell RNA-sequencing datasets from the Genome Sequence Archive (accession no. CRA010641) were analyzed (14). Specifically, a subset of full-thickness excisional skin wounds was selected, as shown in Fig. 2A. Following dataset integration and rigorous quality control, high-quality sequences were obtained from 73,357 cells across various time points: skin-YD0 (8,618 cells), skin-YD2 (7,253 cells), skin-YD4 (7,690 cells), skin-YD7 (5,830 cells), skin-OD0 (9,333 cells), skin-OD2 (18,820 cells), skin-OD4 (7,624 cells) and skin-OD7 (OD7 9,189 cells). Within this comprehensive cellular atlas (Fig. 2B), 15 distinct cell types were identified, including Fibs (Fib, Dpt^HiPdgfra⁺Col4a⁺) (15), ECs (Ec, Ptprb⁺Pecam1⁺Sox17⁺) (16), papillary Fibs (Fp, Lef1⁺), smooth muscle cells (SMC, Lmod1⁺Pln⁺Acta2^Hi) (17), fascia cells (Fasc, Pax7⁺Erfe⁺Gpx3^Hi), type 1 macrophages (Mac1, Stat1^Hi Arg1^Lo), type 2 macrophages (Mac2, Arg1^Hi), neutrophils (Neu, S100a8^HiS100a9^Hi), lymphocytes (Lym, Cd3d^HiCd3g^Hi), T cells (T-cell, Trat1⁺Cd3g^HiCd3e^HiCd3d^HiNkg7^Hi) (18), suprabasal cells (Supra Krt6a⁺Krt6b⁺) (19), spinous and granular keratinocytes (Spin, Krt1^Hi), basal keratinocytes (Basal, Krt5^Hi) (20), epithelial cells (Epi, Krt79^Hi) (21) and germinative layer cells (Germ, Ube2c^Hi), based on top marker expression (Fig. 2C-F).

Next, these cell types and their alterations during wound healing were comprehensively analyzed. As shown in Fig. 3A, inflammatory cells were the predominant cell type across the two groups, with Mac1 at 31.17%, Mac2 at 7.62%, Neu at 16.51%, Lym at 2.22% and T-cell at 0.79%. Consistent with this, a classical inflammatory infiltration-to-resolution process was observed in both groups (Fig. 3B-D). The distribution of ECs clearly differed between the two groups, with young wounds exhibiting markedly higher cell numbers across all time points. Young wounds exhibited a faster recovery rate in terms of Fib count (Fig. 3B-F). Fp, known to promote hair follicle formation (22,23), were predominantly observed in OD0 (Fig. 3B-D). These findings of impaired EC restoration and delayed Fib recovery underscore the diminished regenerative capacity and compromised healing potential of aged tissues.

To gain a deeper insight into the transcriptomic changes in ECs and Fibs throughout the healing process in both young and aged mice, ECs and Fibs were extracted and differentially expressed genes (DEGs) identified at four time points, as shown in Figs. 3G-J and 4A-D.

On day 0, regulator of G protein signaling 5 (Rgs5), a recognized marker of endothelial function and vascular remodeling (24) and platelet-derived growth factor A (Pdgfr), which is critical for vascular development (25), were downregulated specifically in aged wounds ECs. Concurrently, chronic inflammation-related genes, including cytochrome P450 1A1 (Cyp1a1) and angiopoietin-like protein 4 (Angptl 4) (26,27), were upregulated in aged wounds ECs (Fig. 3G). On days 2 and 4, significant DEGs (highlighted in red in the figure) were relatively sparse, probably reflecting reduced EC populations. However, S100a8, IL1b and Cdkn2b, factors related to the regulation of inflammation and aging (28), remained upregulated in aged wounds ECs (Fig. 3H and I). Notably, mitochondrial genes critical for metabolism, including mt-Atp8, mt-Nd3 and mt-Nd4, were downregulated. Notably, platelet-derived growth factor receptor-β (Pdgfrb), essential for endothelial differentiation (29), was markedly downregulated in aged wounds ECs on day 7 (Fig. 3J). Most of the upregulated DEGs on day 7, such as H2-Ab1, H2-Eb1, H2-Aa and Cd74, were related to immune activation and antigen presentation (30).

Dynamic transcriptomic changes in Fibs are critical for wound healing progression (31). As shown in Fig. 4A, on day 0, uninjured aged skin exhibited significant downregulation of key Fib markers, including adenylate cyclase 1 (Adcy1; myoFib activation), collagen type I α 1 (Col1α1; fibrosis driver) and Fib growth factor receptor 4 (Fgfr4; proliferation/migration) (32). On days 2–4, genes associated with tissue aging and chronic inflammation, including S100a8, cd74 cd52, Ccl6 and S100a4, were upregulated in aged wound Fibs (Fig. 4B and C). By day 7, delta-like homolog-1 (Dlk-1; adipogenesis inhibitor), elastin (Eln; extracellular matrix components), renin-angiotensin-aldosterone system (RAAS; fibrosis-related genes) and angiotensin II receptor 1A (Agtr1a; transcription growth factor (TGF)-β signaling pathway) were markedly downregulated (33–36).

The present study identified 31 genes that were consistently and markedly downregulated in aged vs. young wounds across all time points (Fig. 4 E), including fibrosis-associated genes (Meg3, Sparc, Nrp1 and Peg3) (37–40). To elucidate the biological significance of these alterations, Gene Ontology (GO) term enrichment analysis was performed. Downregulated DEGs in both uninjured skin and healing aged skin were consistently markedly associated with cell proliferation, cell division and angiogenesis (Fig. 4G-J).

Using CellChat, distinct intercellular communication patterns were identified in the two age groups. Given the similarity in principal cell types and the more representative wound differences on day 7, the OD7 and YD7 datasets were selected for analysis. First, the two datasets were integrated to construct a unified cell communication network, categorized into four functional clusters (Fig. 5A and B). Pathways involving IL2, IL4, WNT and BMP clustered together, suggesting their universal roles in wound healing. By contrast, the CCL, GDF, PERIOSTIN, ACTIVIN, TGF-β and NRG pathways formed distinct clusters, highlighting their differential contributions to age-related healing outcomes (Fig. 5B).

These pathways demonstrated substantial divergence in Euclidean distances within the shared two-dimensional space (Fig. 5C). Of 52 pathways, 16 showed significant activity in information flow analysis, albeit with varying intensities (Fig. 5D). A total of three pathways were markedly upregulated in OD7: midkine (inflammation/tissue repair regulator) (41,42), CCL (inflammatory pathway;) (43) and TGF-β (wound healing/fibrosis mediator) (44–46).

Age-related differences in cell type-specific communication were pronounced. Aged wounds exhibited reduced incoming interaction strength for ECs and Fibs when compared with young wounds (Fig. 5E and F). After consolidation and standardization, Fibs displayed the most pronounced signal strength alterations in both incoming and outgoing communications, while ECs also showed substantial variations (Fig. 5G). These findings indicate that intercellular communication, particularly between ECs and Fibs, is markedly decreased in elderly wounds.

Pathway-specific visualization revealed the key contributors to communication differences (Fig. 5H-J). In Fibs, the diminished incoming signal intensity in aged wounds primarily involved the Wnt, EGF, TGF-β, GAS, IL-6, PROS and Fas ligand (FASLG) pathways. Similarly, ECs exhibited reduced incoming signals primarily via the Wnt, TGF-β and FASLG pathways, which were all upregulated in YD7.

To investigate age-related differences in EC-Fib communication during wound healing, CellChat was employed to quantify the differences in communication probabilities between OD7 and YD7. As shown in Fig. 6A, young wounds exhibited markedly increased numbers and strength of EC-Fib interactions compared with aged wounds, regardless of signal origin. Network analysis corroborated the distinct communication patterns: Young wounds showed 16 novel Fib-EC interactions and three additional EC-Fib connections compared with aged wounds, accompanied by enhanced interaction strength (Fig. 6C).

Pathway analysis revealed significant differences in Fib-related signaling pathways between the age groups. Notably, the TGF-β pathway demonstrated divergent regulation in young wounds, with increased incoming but decreased outgoing signaling from Fibs (Fig. 6B). Further examination of the entire TGF-β pathway network revealed enhanced EC-EC communication, a feature absent in aged tissue (Fig. 6D). Moreover, young ECs exhibited broader signaling versatility, functioning as both signal senders and receivers, whereas aged ECs primarily acted as signal sources (Fig. 6E). Notably, EC-Fib communication via TGF-β was substantially augmented in young wounds (Fig. 6F), mediated via enhanced signaling through Acvr1-TGF-βR1and TGF-βR1-TGF-βR2 ligand-receptor pairs (Fig. 6G). Mechanistically, the age-related downregulation of TGF-β1 on ECs and TGF-βR2on Fibs probably contribute to impaired TGF-β signaling in aged wounds (Fig. 6H).

Together, these findings demonstrate the age-dependent dysregulation of EC-Fib communication, particularly via TGF-β signaling. This impaired crosstalk may underline the delayed healing kinetics in aged wounds.

To verify the impaired interaction between ECs and Fibs through the TGF-β pathway in aged wounds, we induced Fib senescence via high-glucose treatment in vitro and evaluated the effects of ECs on young Fibs vs. aged Fibs through Transwell-based indirect co-culture system. First, a senescent Fib model was established by culturing primary mouse dermal fibroblasts in gradient concentrations of high-glucose medium (50 and 75 mmol/l) to determine the optimal senescence-inducing dose. qPCR analysis revealed that the senescence marker genes P16 and P21 were markedly upregulated, while the proliferation marker gene Ki67 was markedly downregulated in high-glucose groups (Fig. 7A-C). Protein-level validation via immunofluorescence demonstrated that the 75 mmol/l high-glucose group exhibited markedly higher senescence protein markers and reduced proliferative capacity compared with normal glucose controls (Fig. 7D-G). These results confirmed the successful establishment of the senescent Fib (old) model in vitro. Based on these findings, Fib (old) showing the most pronounced senescence phenotype from the 75 mmol/l high-glucose group was selected for subsequent experiments.

Following 3 days of co-culture with ECs, senescent Fibs (Fib-old) displayed significant downregulation of TGFβ1, Smad2 and Smad3 genes, critical fibrotic markers in the TGFβ signaling pathway, which was further confirmed by immunofluorescence staining (Fig. 7H-R).