HIF-1α attenuates ferroptosis-associated dermal fibroblast senescence via modulation of NF-κB–DPP4 signaling Authors: Shang-Chuan Ng, Shih-Wen Kao, Chih‑Jung Chen, Shinn-Zong Lin, Dennis Jine-Yuan Hsieh, Chia-Hua Kuo, Kuan-Ho Lin, Chih-Yang Huang, Wei-Wen Kuo Published: 2026 ABSTRACT AbstractBackgroundsSkin aging is characterized by progressive structural and functional decline driven in part by chronic oxidative stress. Ferroptosis-associated processes have been increasingly implicated in cellular senescence and are linked to dipeptidyl peptidase-4 (DPP4). Although hypoxia-inducible factor-1α (HIF-1α) is essential for skin homeostasis, its role in dermal fibroblast aging remains incompletely understood. This study aimed to determine whether HIF-1α modulates ferroptosis-associated senescence in human dermal fibroblasts (HDFs) and to explore the involvement of NF-κB–DPP4 signaling.MethodsHIF-1α expression was examined in aged HDFs and sun-exposed human skin. Transcriptomic profiling was performed to assess pathways associated with hypoxia in aging models. Pharmacological modulators of ferroptosis, DPP4, and HIF-1α were applied in vitro to evaluate effects on senescence and ferroptotic markers. NF-κB activity and DPP4 transcription were analyzed using molecular assays. Potential in vivo relevance was evaluated in aged mice treated with a HIF-1α activator or a ferroptosis inhibitor.ResultsHIF-1α expression was reduced in aged HDFs and photoaged skin, and its depletion accelerated cellular senescence. Transcriptomic analysis suggested downregulation of ferroptosis-related pathways under hypoxic conditions. Induction of ferroptosis-associated stress promoted senescence phenotypes in HDFs, whereas hypoxia-induced HIF-1α activation attenuated these effects. Mechanistically, NF-κB functioned as a transcriptional regulator of DPP4, and hypoxia was associated with reduced NF-κB activity, decreased DPP4 expression, and lower secretion of soluble DPP4 (sDPP4). Inhibition of DPP4 mitigated ferroptosis-associated senescence and restrained sDPP4-mediated paracrine propagation to neighboring young fibroblasts. In aged mice, topical treatment with a HIF-1α activator or a ferroptosis inhibitor was associated with improved dermal structure and reduced senescence markers.ConclusionThese findings support a model in which HIF-1α is associated with suppression of NF-κB–dependent DPP4 expression and attenuation of ferroptosis-associated fibroblast senescence. Targeting the HIF-1α/NF-κB/DPP4 axis may represent a potential strategy for mitigating skin aging.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12967-026-08445-y.Keywords: HIF-1α, DPP4, Ferroptosis, HDFs, Dermal aging Backgrounds Skin dermis, composed primarily of fibroblasts, provides structural integrity through the secretion of extracellular matrix (ECM) components, counteracting age-related degeneration [1]. Compared with proliferating cells, senescent cells display increased senescence-associated β-galactosidase (SA-β-gal) activity and secrete a complex mixture of transcriptionally regulated factors collectively termed the senescence-associated secretory phenotype (SASP). SASP components include pro-inflammatory cytokines, chemokines, growth modulators, and matrix metalloproteinases (MMPs), which contribute to a chronic inflammatory milieu, leading to fibroblast senescence [2]. In human skin, senescent fibroblasts accumulate with age, resulting to morphological and functional alterations that underlie skin aging. Atmospheric oxygen levels are around 20%, but its availability differs across various organs. Under physiological conditions, different organs are exposed to varying concentrations of oxygen, depending on their specific metabolic demands and vascular supply. Skin inherently exists in a mildly hypoxic state, and Hypoxia-Inducible Factor alpha (HIF-1α) is consistently expressed in skin cells [3]. In epidermal keratinocytes, HIF-1α regulates antimicrobial peptide production, re-epithelialization during wound healing, and cytoskeletal remodeling to maintain the skin barrier [4, 5]. In dermal fibroblasts, however, the role of HIF-1α is less well characterized. Emerging evidence suggests that hypoxia attenuates fibroblast aging by promoting collagen synthesis [6] and limiting SASPs secretions [7]. Although mammalian cells are generally capable of adapting to fluctuating oxygen concentrations [8], advanced age is associated with impaired stress responses and diminished HIF-1α activity [9]. HIF-1α has been implicated in extending cellular lifespan, maintaining proliferative capacity, and reducing mitochondrial oxidative stress in mammalian cells [10–13], highlighting its potential importance in regulating dermal fibroblast senescence. Excess intracellular ferrous iron (Fe²⁺) can drive ferroptosis, an iron-dependent form of regulated cell death marked by lipid peroxidation, depletion of the antioxidant enzyme glutathione peroxidase 4 (GPX4), and accumulation of toxic aldehydes such as malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE) [14–16]. Ferroptosis promotes chronic, low-grade inflammation (“inflammaging”) and is implicated in age-related disorders including neurodegenerative, cardiovascular, and metabolic diseases [17]. In dermatology, ferroptosis has been studied in melanoma [18], epidermal homeostasis [19], psoriasis [20], vitiligo [21] and UVB-induced keratinocyte death [22]. However, its role in dermal fibroblast aging, particularly in the context of non-lethal ferroptosis-associated stress, remains incompletely understood. Dipeptidyl peptidase 4 (DPP4), a membrane-bound serine protease, can be cleaved by MMPs to generate a soluble enzymatically active form (sDPP4). sDPP4 inhibition is widely used in type 2 diabetes management due to its role in enhancing the degradation of incretin hormones, such as glucagon-like peptide 1 (GLP-1) which can promote insulin secretion [23]. Recent studies have demonstrated that DPP4 is highly expressed on senescent cells [24]. In skin, aberrant DPP4 activity contributes to increased pro-inflammatory cytokine production, disruption of T cell homeostasis, and impaired wound healing [25]. However, whether DPP4 actively contributes to dermal fibroblast senescence or is primarily a marker of this process remains unclear. In this study, using both premature and chronological human dermal fibroblast (HDF) aging models, we investigated the role of HIF-1α in modulating senescence under hypoxic conditions. We found that loss of HIF-1α was associated with increased SA-β-gal–positive cells, elevated SASP factor expression, and reduced proliferative capacity. Transcriptomic and functional analyses suggested an association between hypoxia, ferroptosis-related pathways, and fibroblast senescence. Mechanistically, NF-κB was identified as a transcriptional regulator of DPP4, and modulation of this axis was associated with changes in ferroptosis- and senescence-related phenotypes. In addition, soluble DPP4 contributed to paracrine propagation of senescence-associated signals. Notably, these effects were attenuated under hypoxic conditions in a HIF-1α–dependent manner. Furthermore, pharmacological activation of HIF-1α using dimethyloxalylglycine (DMOG) or inhibition of ferroptosis using ferrostatin-1 (Fer-1) was associated with reduced senescence-related phenotypes in vitro and improved skin characteristics in aged mice. Together, these findings suggest a potential role for the HIF-1α/NF-κB/DPP4 axis in modulating ferroptosis-associated dermal fibroblast senescence. Methods Cell culture and induction of senescence Human dermal fibroblasts (HDFs; #SC-2310) were obtained from ScienCell Research Laboratories (USA). Primary adult HDFs derived from donors aged 18 years (CG1629), 40 years (CG1630), and 65 years (CG1639) were obtained from the Bioresource Collection and Research Center (BCRC, Taiwan). Cells were maintained in Dulbecco’s Modified Eagle’s Medium (Gibco, #31600034) supplemented with 10% fetal bovine serum, 5 µg/mL recombinant human insulin, 5 ng/mL recombinant human fibroblast growth factor, 2 g/L sodium bicarbonate, and 1% penicillin–streptomycin, according to the manufacturer’s protocols. Stress-induced premature senescence was established by treating young HDFs with 250 nM doxorubicin for 24 h or 200 µM H₂O₂ for 2 h, followed by washout and 10 days of culture. For UVB-induced senescence, cells were exposed to 40 mJ/cm² UVB and cultured for 10 days post-exposure. These procedures were performed as described in previous studies [10, 31]. For ferroptosis-associated stress, HDFs were treated with 200 nM RSL3 or 10 µM erastin for 4 h, followed by washout and incubation for 48 h. RSL3, erastin, and doxorubicin were dissolved in dimethyl sulfoxide (DMSO), and corresponding vehicle controls were treated with equivalent concentrations of DMSO. H₂O₂ treatments were performed using sterile water as the vehicle control. All treatments were performed under either normoxic (20% O₂) or hypoxic (1% O₂) conditions, depending on the experimental design. Antibodies and reagents The following primary antibodies were used: HIF-1α (GTX127309, GeneTex), ACSL4 (sc-365230, Santa Cruz), DPP4 (#67138, Cell Signaling Technology), 15-LOX2 (sc-365230, Santa Cruz), GPX4 (ab125066, Abcam), GAPDH (sc-32233, Santa Cruz), 4-HNE (ab46545, Abcam), p21 (10355-1-AP, Proteintech), MMP-1 (10371-2-AP, Proteintech), MMP-3 (GB11131, ServiceBio), NCOA4 (ab86707, Abcam), γ-H2A.X (Ser139; #2577, Cell Signaling Technology), COL1A1 (sc-293182, Santa Cruz), COL3A1 (sc-271249, Santa Cruz), Ki67 (sc-23900, Santa Cruz), β-actin (sc-47778, Santa Cruz), NFκB (ab32536, Abcam), and phospho-NFκB (#3033, Cell Signaling Technology) PAR2 (#6976, Cell Signaling Technology). HRP-conjugated secondary antibodies against rabbit (A0545), mouse (A9044), and goat (A5420), as well as 30% H₂O₂ (H1009), were obtained from Sigma-Aldrich (St. Louis, MO, USA). Chemical reagents included doxorubicin (HY-15142), DMOG (HY-15893), Erastin (HY-15763), RSL3 (HY-100218 A), Saxagliptin (HY-10285), Sitagliptin (HY-13749), and recombinant human DPP4 (rhDPP4; HY-N0135), all purchased from MedChemExpress (MCE). Cumulative population doubling levels (CPDLs) CPDL were calculated using the formula “χ = {log10(NH) − log10(N1)}/log10(2)” in which N1 is the inoculum cell number and NH is the cell harvest number. To determine the cumulated doubling level, the population doubling level for each passage will be calculated and then added to the levels of the previous passages. Since the primary cells which we brought were determined for the first time at passage 1, the cumulative doubling number were first calculate for passage 1. The CPDL of HDFs below 10 signifies young cells, whereas a CPDL greater than 40 is indicative of aged cells. Trypan Blue exclusion assay Cells were replated in 24-well culture dishes and subjected to various treatments. Following treatment, cells were detached from the plates by trypsinization and centrifuged at 1500 g for 3 min. The supernatant was discarded, and the cell pellet were resuspended in fresh cell culture medium. A 1:1 ratio of 0.4% trypan blue to the cell suspension will be prepared and incubated at RT for 3 min. A 10 µL aliquot of the trypan blue-treated cell mixture will be loaded into a hemocytometer and counted using the countess 3 automated cell counters (Invitrogen). Expression plasmids, gene construction, siRNA and in vitro gene transfection The pCMV3-DPP4-Flag plasmid (HG10688-CF) and pCMV3-NFκB-Flag plasmid (HG12054-CF) were purchased from Sino Biological (Beijing, China). The luciferase reporter vector pGL4.1[luc2] (E6651) was obtained from Promega Corporation (Madison, WI, USA). DPP4 promoter luciferase constructs, including pGL4.1-DPP4-2 kb (NFκB-BS-WT), pGL4.1-DPP4-2 kb (NFκB-BS-MT), pGL4.1-DPP4-1.3 kb, pGL4.1-DPP4-0.7 kb, and pGL4.1-DPP4-0.2 kb, were synthesized by Protech Technology Enterprise (Taipei, Taiwan). Human siRNAs targeting HIF-1α (IDSASI_Hs01_00332063), c-Fos (IDSASI_Hs01_00184572), SP1 (IDSASI_Hs01_00363664), NFκB (IDSASI_Hs01_00171090), and DPP4 (IDSASI_Hs01_00098446), as well as the negative control siRNA (SIC001), were obtained from Sigma-Aldrich. Plasmid and siRNA transfections were performed using the jetPRIME transfection reagent (Polyplus-transfection, Illkirch, France) according to the manufacturer’s instructions. Luciferase reporter assay Cells were co-transfected with DPP4 reporter constructs and the pCMV3-NFκB-Flag plasmid. Following transfection and treatment, luciferase activity was measured using the luciferase assay system (E2920; Promega, Madison, WI, USA). Briefly, 100 µL of luciferase substrate was added to 20 µL of cell lysate, and luminescence was quantified with a Reporter Microplate Luminometer (Turner Biosystems, Sunnyvale, CA, USA). All experiments were performed in triplicate and repeated independently at least three times. SASP and GSH/GSSG determination Conditioned medium (CM) was collected following the indicated treatments and centrifuged at 300 × g for 5 min to remove cells and debris. For total protein extraction, mouse skin tissues were minced, snap-frozen in liquid nitrogen, and homogenized in cold RIPA buffer supplemented with protease and phosphatase inhibitors using a refrigerated homogenizer (Prep-CB24R; CLUBIO). For HDFs, cells were trypsinized, resuspended in ice-cold PBS, and lysed by sonication until complete disruption. Both tissue and cell lysates were centrifuged at 12,000 × g for 40 min at 4 °C to remove insoluble material. Protein concentrations were determined using the BCA Protein Assay Kit (Cat. #23225; Thermo). ELISA assays were performed on CM according to the manufacturers’ instructions using the following kits: human sDPP4 (Cat. #RE3059H; R&D Systems), human IL-1α (Cat. #RE1047H; Reed Biotech), human IL-1β (Cat. #RE1074H; Reed Biotech), MDA (Cat. #E-EL-0060; Elabscience), human MMP-1 (Cat. #ab215083; Abcam), and human MMP-3 (Cat. #ab269371; Abcam). Cellular GSH and GSSG levels were determined using a commercial enzymatic recycling assay kit based on glutathione reductase (Cat. #703002; Cayman), following the manufacturer’s protocol. The GSH/GSSG ratio was calculated using a standard curve. Flow cytometry analysis DPP4⁺/SA-β-Gal⁺ HDFs were detected using a PE-conjugated mouse anti-human CD26 antibody (555437; BD Biosciences) in combination with the SPiDER-βGal Staining Kit (35302; Cell Signaling). Following the indicated treatments, cells were incubated with Bafilomycin A1 (100 nM) at 37 °C for 1 h to inhibit endogenous β-galactosidase activity. SA-β-Gal substrate solution was then added, and cells were incubated at 37 °C for 2 h. After incubation, cells were washed with stain buffer (554656; BD Biosciences), trypsinized, and collected by centrifugation at 300 × g. The cell pellet was resuspended in cold stain buffer and stained with PE-conjugated anti-CD26 for 40 min at 4 °C in the dark. Finally, cells were washed twice, passed through a 70-µm cell strainer, and resuspended in cold stain buffer for flow cytometry analysis, as previously described [26, 27]. Immunofluorescence staining Ferroptosis and senescence in HDFs were assessed using a combination of 4-HNE immunostaining and the SPiDER-βGal Detection Kit (SG05; Dojindo). Briefly, cells were fixed with 4% paraformaldehyde for 3 min at RT and incubated with the SPiDER-βGal working solution for 30 min at 37 °C in a non-CO₂ incubator. Following staining, cells were permeabilized with 0.1% Triton X-100 for 30 min at RT and blocked with 5% horse serum for 1 h at RT. Cells were then incubated overnight at 4 °C with anti-4-HNE antibody (1:100; bs-6313R, Bioss). Alexa Fluor 594-conjugated goat anti-rabbit IgG (1:200; A11012, Invitrogen) was applied for 2 h at RT. Nuclei were counterstained with DAPI, and images were acquired using a fluorescence microscope. EdU assay After treatment, cells were replated on 8-well slides (1 × 10⁴/well) and cultured for 24 h. Proliferation was evaluated using the Fluor 594-EdU Cell Proliferation Imaging Kit (E-CK-A377; Elabscience). Cells were incubated with 10 µM EdU for 2 h, fixed with 4% paraformaldehyde, permeabilized with 0.3% Triton X-100, and subjected to the Click-iT reaction cocktail for 30 min at RT in the dark. Nuclei were counterstained with 1 µg/mL DAPI, and images were acquired by fluorescence microscopy. Proliferation was expressed as the ratio of EdU⁺ to total DAPI⁺ cells. SA-β-Galactosidase assay SA-β-Gal activity was measured using the cellular senescence assay kit (KAA002, Sigma-Aldrich). Briefly, cells were fixed with 4% paraformaldehyde at RT for 5 min and subsequently stained with X-Gal staining solution (pH 6.0) overnight at 37 °C.-Stained cells were visualized under a bright-field microscope. For quantitative analysis, SA-β-Gal activity was measured using the Cellular Senescence Plate Assay Kit (SG05, Dojindo). Cells were lysed with lysis buffer for 10 min at RT and incubated with SPiDER-βGal working solution at 37 °C for 30 min. After adding the Stop Solution, fluorescence was measured at excitation 535 nm and emission 580 nm using an ELISA plate reader. Cell numbers were normalized using the cell count normalization kit (C544, Dojindo). Intracellular Fe2+ and oxidized lipid determination Cells were incubated with 1 µM FerroOrange (F374; Dojindo, Japan) or 5 µM C11-BODIPY (27086; Cayman) for 30 min at 37 °C in a CO₂ incubator. After FerroOrange staining, Fe²⁺ signals were observed directly under a confocal fluorescence microscope without washing. For C11-BODIPY, cells were washed, counterstained with DAPI, and imaged by confocal microscopy. Fluorescence intensity was quantified using ImageJ from multiple randomly selected fields and averaged across independent experiments. 3D cell culture of dermal fibroblast HDFs were pretreated with 200 µM H₂O₂ for 2 h, washed, and cultured under normoxic or hypoxic conditions for 10 days. On days 1, 4, and 7, cells were passaged at 90% confluency, followed by siRNA-mediated knockdown of HIF-1α or negative control. At day 10, cells (4 × 106 cells/mL) were trypsinized, mixed with rat tail collagen type I (08-115; Sigma), and seeded onto 12-well Transwell inserts (0.4 μm PET membrane; Falcon) to generate 3D dermal models. Cultures were maintained under normoxic or hypoxic conditions, with media changes every 3 days. At day 15, tissues were fixed in 10% formalin overnight, embedded, sectioned (5 μm), and stained with H&E for dermal thickness analysis. Western Blotting analysis Cells were cultured in 10-cm dishes, treated, and lysed in buffer on ice for 1 h. Lysates were centrifuged at 12,000 × g for 15 min at 4 °C, and protein concentrations were measured by Bradford assay. Equal protein (20 µg) was resolved on 8–12% SDS-PAGE gels, transferred to PVDF membranes, blocked with 5% milk, and incubated overnight at 4 °C with primary antibodies. After washing, membranes were probed with HRP-conjugated secondary antibodies, and signals were detected using ImageQuant LAS 4000 (GE Healthcare). Protein expression levels were normalized to GAPDH or β-actin, and densitometric analysis was performed using ImageJ. Quantitative RT-PCR Total RNA was extracted using the Direct-zol™ RNA MiniPrep kit (Zymo Research, CA, USA) according to the manufacturer’s instructions. cDNA was synthesized from 1 µg of total RNA using the iScript cDNA Synthesis Kit (Bio-Rad, Cat. #1708891). Quantitative real-time PCR (qRT-PCR) was performed using SYBR Green qPCR Supermix (Bio-Rad, Cat. #1725124) on a real-time PCR system. GAPDH was used as an endogenous control to normalize gene expression levels. Relative gene expression was calculated using the 2⁻ΔΔCt method. Primer sequences are listed in Fig. S7. Chromatin immunoprecipitation (ChIP) assay The binding of NF-κB to the DPP4 promoter was assessed using a chromatin immunoprecipitation (ChIP) assay. ChIP was performed using a commercial kit (Sigma-Aldrich, Cat# CHP1) according to the manufacturer’s instructions. Briefly, HDFs were washed with PBS and crosslinked with 1% formaldehyde for 10 min at room temperature. Crosslinking was quenched with glycine, and cells were harvested for chromatin preparation. The crosslinked chromatin was sheared by sonication to generate DNA fragments of approximately 300–800 bp. Chromatin was then immunoprecipitated overnight at 4 °C with an anti-NF-κB p65 antibody (Cell Signaling Technology, #8242) or normal rabbit IgG as a negative control. Immune complexes were captured, washed, and eluted, followed by reverse crosslinking. DNA was purified and analyzed by quantitative PCR (qPCR) using primers spanning regions of the DPP4 promoter. Enrichment was calculated relative to input DNA and expressed as % input. RNA sequencing and bioinformatic analysis Total RNA was extracted and quality assessed using a Bioanalyzer 2100 with the Agilent RNA 6000 Nano Assay. Libraries were prepared, diluted to 10 nM, pooled equimolarly, and subjected to paired-end sequencing. Data quality, including gene coverage and thermal profiles, was verified. Differential gene expression analysis was performed to identify differentially expressed genes (DEGs), defined by an adjusted P-value < 0.05 and |log₂ fold change| ≥ 1. Functional enrichment analyses were conducted to identify significantly associated biological pathways. GO enrichment analysis was performed using the PANTHER overrepresentation test (version 20240807) via the Gene Ontology online platform (geneontology.org). UniProt IDs were used as input, and enrichment was evaluated within the GO Biological Process Complete dataset (version 19.0; release date: June 17, 2024) [28]. Statistical significance for enrichment was determined using Fisher’s exact test with FDR correction. GSEA was additionally performed to assess pathway-level changes across gene expression profiles. Animal studies All mouse experiments were conducted in accordance with protocols from the National Laboratory Animal Center (NLAC) and were carried out at the China Medical University Animal Center. All procedures were approved by the Institutional Animal Care and Use Committee of China Medical University and conformed to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (CMUIACUC-2024-070). For chronological aging studies, 2-month-old C57BL/6 male mice were obtained and maintained until 24 months of age. These animals were followed as a longitudinal cohort, and mice were sacrificed at 3, 6, 12, 15, 18, 21, and 24 months (n = 3 per time point) for tissue collection. For topical treatment studies, 4-month-old young and 24-month-old aged male C57BL/6 mice (n = 6) were housed under controlled conditions (22 °C, 12-hour light/dark cycle) with ad libitum access to food and water. Dorsal hair was removed under isoflurane anesthesia, and topical treatments were administered daily for 10 consecutive days to a 3 × 2 cm area of dorsal skin. Treatments included Fer-1 (1 mg/kg/day) or DMOG (1 mg/kg/day). Mice were sacrificed on day 11, and dorsal skin tissues were collected, either snap-frozen at − 80 °C for molecular analyses or fixed in formaldehyde for histological studies. Immunohistochemistry (IHC) Formalin-fixed, paraffin-embedded dorsal skin tissues were sectioned at 5 μm, dewaxed in xylene, and rehydrated through graded alcohols. Antigen retrieval was performed in 0.1% trypsin at 37 °C for 20 min, followed by blocking of endogenous peroxidase (SP-6000; Vector Laboratories) and nonspecific binding with 2.5% horse serum (PK-7800; Vector Laboratories). Sections were incubated overnight at 4 °C with primary antibodies against GPX4, 4-HNE, HIF-1α, DPP4, MMP-1, MMP-3, IL-1α, and IL-1β. After washing, sections were incubated with a biotinylated universal secondary antibody and streptavidin–peroxidase complex (PK-7800; Vector Laboratories), and visualized with DAB (SK-4100; Vector Laboratories). Slides were counterstained with hematoxylin (Vector Laboratories) and examined by light microscopy (Olympus). Quantification of staining intensity was performed using ImageJ in a blinded manner. Statistical analysis Statistical analyses were performed using GraphPad Prism 5 (San Diego, CA, USA). Data are presented as mean ± standard deviation (SD) unless otherwise indicated. Statistical significance was defined as P < 0.05. For comparisons between two groups, unpaired two-tailed Student’s t-tests were applied. For multiple group comparisons, one-way analysis of variance (ANOVA) was performed, followed by Tukey’s post hoc test for multiple comparisons. Adjusted P-values < 0.05 were considered statistically significant. Results HIF-1α is reduced with less responsive and more cell viability enhancement following hypoxia exposure in aged skin cells Sun exposure and ultraviolet (UV)-induced damage are well-recognized contributors to skin aging. We utilized a tissue array (SKN1001) collected from various anatomical sites and performed IHC staining (Fig. 1A). Notably, HIF-1α protein levels were lower in sun-exposed regions (face, neck) compared with sun-protected regions (chest, back, groin, anus, thigh) (Fig. 1B, C; Fig. S1). In parallel, we analyzed dorsal skin tissues from mice at different ages (3, 6, 12, 15, 18, 21, and 24 months). Histological and IHC analyses data revealed progressive reductions in epidermal thickness, collagen content, and HIF-1α levels with advancing age (Fig. 1D). Consistently, western blot analysis revealed reduced HIF-1α and collagen proteins (COL1A1, COL3A1), accompanied by increased expression of aging markers (p21, γ-H2AX) and matrix-degrading enzymes (MMP-1, MMP-3) in aged mice skin (Fig. 1E; Fig. S2A). HIF-1α is reduced with less responsive and more cell viability enhancement following hypoxia exposure in aged skin cells. (A) Schematic showing collection sites of normal human skin samples represented in the tissue array (SKN1001). (B, C) IHC staining of HIF-1α in sun-exposed and non–sun-exposed skin. Staining intensity was quantified across different anatomical sites using ImageJ. (D, E) Dorsal skin from mice (3–24 months) was analyzed by H&E, Masson’s trichrome, and HIF-1α IHC staining; skin lysates were assessed by immunoblotting. (n = 3 each). (F) HDFs from donors aged 18, 40, and 65 years were exposed to normoxia (20% O₂) or hypoxia (1% O₂); HIF-1α expression was analyzed by immunoblotting. (G) Young (PD < 10) and aged (PD > 40) HDFs were cultured under normoxia or hypoxia for 24 h; HIF-1α expression was analyzed by immunoblotting. (H, I) Young HDFs were pretreated with or without doxorubicin (250 nM, 24 h) or H₂O₂ (200 µM, 4 h), washed, and cultured under normoxia or hypoxia for 10 days; HIF-1α and p21 expression was analyzed by immunoblotting. (n = 3, 4). (J–L) Gene expression was quantified by qRT-PCR relative to GAPDH. (n = 3 each). (M–O) Protein levels in conditioned media were measured by multiplex ELISA. (n = 6 each). (P) Growth curve showing the cumulative population doubling of young HDFs cultured under normoxic and hypoxic conditions for 100 days. (Q, R) Trypan Blue exclusion assays were performed at different time points with or without H₂O₂ or doxorubicin treatment, and cultured under normoxic or hypoxic conditions. Growth curves were generated from day 1 onward in young HDFs. (n = 3 each). Data are presented as mean ± SD. Quantification of the results is shown *P < 0.05, **P < 0.01 and ***P < 0.001 Using models: chronological aging to induce replicative senescence (Fig. 1G) and human skin samples with advancing age (Fig. 1F), we found aged HDFs exhibited reduced activation of HIF-1α protein level and less response to hypoxia. Both the Dox- and H₂O₂-induced premature aging models showed the similar results. However, upregulated P21 induced by Dox and H2O2 was attenuated by hypoxia exposed in aged HDFs (Fig. 1H-I), further indicating the anti-aging effect of HIF-1α. For detailed identification of these effects, it was observed that aged HDFs exposed to hypoxia exhibited significantly lower mRNA expression (Fig. 1J–L) and reduced secretion (Fig. 1M–O) of SASP factors, including MMP-1, MMP-3, IL-1α, and IL-1β, compared with normoxic conditions, confirming the anti-aging effects of hypoxia exposure. To evaluate the long-term effects of hypoxia, cumulative population doubling levels (PDLs) of chronologically aged HDFs and trypan blue exclusion assays of prematurely aged HDFs were assessed. Compared with normoxic controls (PDL, 34.9), hypoxia-exposed HDFs exhibited a higher cumulative PDL (44.4) at day 100, indicating improved growth capacity and extended lifespan in culture (Fig. 1P). Growth kinetics analysis revealed that hypoxia-exposed Dox- and H₂O₂-treated HDFs had significantly higher cell numbers than their normoxic conditions at day 10 (Fig. 1Q, R). Interestingly, more increase of cell growth in response to hypoxia is showed in aged, but not young HDFs (Fig. Q, R). At day 100 of chronological aging, hypoxia-exposed HDFs (PDL, 44.4) exhibited reduced senescence-associated markers, including decreased SA-β-gal positivity and p21 expression, along with increased TERT levels. These cells also showed enhanced proliferative capacity, as evidenced by increased EdU-positive cells, indicating attenuated senescence and preserved proliferative potential (Fig. S2B–H, K). In time-course analyses of premature senescence, hypoxia significantly reduced SA-β-gal–positive cells in doxorubicin-exposed HDFs at day 10 (Fig. S2C, F, I) and in H₂O₂-exposed HDFs at days 8 and 10 (Fig. S2D, G, J). Consistently, hypoxia increased the proportion of EdU-positive cells at day 10 in both doxorubicin- and H₂O₂-exposed HDFs (Fig. S2L, M). Collectively, these findings indicate that aged HDFs exhibit impaired HIF-1α responsiveness, and that hypoxia is associated with reduced SASP expression and attenuated senescence phenotypes. Loss of HIF-1α accelerates senescence and impairs proliferative capacity in HDFs HIF-1α is a key transcription factor activated under hypoxia [4]. To assess its role in dermal fibroblast aging, we employed both chronological aging and H₂O₂-induced premature aging models. SiRNA-mediated knockdown of HIF-1α in aged HDFs led to a marked reduction in HIF-1α protein levels, confirming efficient silencing (Fig. 2A–B). Loss of HIF-1α accelerated cellular senescence, as evidenced by increased SA-β-Gal–positive cells (Fig. 2C, D, F) and elevated mRNA expression and secretion of pro-inflammatory SASP factors, including MMP-1, MMP-3, IL-1α, and IL-1β (Fig. 2J–K). Proliferative capacity was evaluated using EdU incorporation and a 3D collagen matrix dermal fibroblast culture system. HIF-1α knockdown significantly reduced EdU-positive cells (Fig. 2E, G) and decreased dermal thickness (Fig. 2H–I), indicating impaired fibroblast proliferation. Notably, siRNA-mediated knockdown of HIF-1α in young-HDFs did not significantly increased SA-β-Gal activity (Fig. S4A–C). These findings suggest that the effects of HIF-1α depletion on senescence are context-dependent and more pronounced in aged HDFs. These results suggest that HIF-1α contributes to maintaining proliferative capacity and limiting SASP-associated changes in HDFs. Loss of HIF-1α accelerates senescence and impairs proliferative capacity in HDFs. (A, B) Young HDFs were pretreated with H₂O₂ (200 µM, 4 h), washed, and cultured under normoxia or hypoxia for 10 days; cells were passaged at 90% confluency (day 4, 7) and transfected with HIF-1α or NC siRNA (40 nM). Immunoblotting of HIF-1α in H₂O₂-induced senescent and chronologically aged HDFs. (n = 3 each). (C) SA-β-gal staining (green) and EdU staining (red) used to assess senescence and proliferation, respectively; representative images shown. (D, F) Quantification of SA-β-gal–positive cells using an ELISA-based senescence assay, normalized to total cell number. (n = 6 each). (E, G) Quantification of EdU-positive cells normalized to total cell number (corresponding to panel C). (n = 6 each). (H, I) 3D dermal fibroblast models generated in collagen; representative H&E images and dermal thickness quantification. (n = 3 each). (J) Total RNA was extracted and the mRNA of the indicated genes quantified using qRT-PCR relative to housekeeping mRNA levels. (n = 3 each). (K) Protein levels in conditioned media were measured by multiplex ELISA. (n = 6 each). Data are presented as mean ± SD. Quantification of the results is shown *P < 0.05, **P < 0.01, and ***P < 0.001 Transcriptomic profiling of distinct HDF aging models under hypoxic conditions To assess how hypoxia modulates dermal fibroblast aging, RNA-seq was performed on aged HDFs under hypoxic and normoxic conditions in replicative senescence and premature aging models (Dox, H₂O₂). Principal component analysis (PCA) revealed clear separation between hypoxia- and normoxia-exposed cells (Fig. 3A), indicating distinct transcriptional profiles. Differential expression analysis identified 737 DEGs in chronologically aged HDFs, 373 in H₂O₂-treated cells, and 346 in Dox-treated cells (Fig. 3B), with 159 genes co-upregulated and 38 co-downregulated across models (Fig. 3C–D). Gene ontology enrichment analysis revealed that the differentially co-downregulated genes were enriched in ferroptosis-related processes. (Fig. 3E). Notably, several downregulated genes (FTH1, TFRC, EGR1, DPP4, AFF3, PGD, LINC00472) have been reported in the previous literature to be associated with iron metabolism, redox regulation, or stress responses relevant to ferroptosis, although they are not all directly annotated as canonical regulators of ferroptosis (Fig. 3F). In contrast, hypoxia-induced genes were enriched in HIF-1α–related transcriptional programs and extracellular matrix regulation (e.g., TIMP3, TGFB1, COL7A1, COL5A1, DCN, HAS2, P4HA1) (Fig. 3G). Gene set enrichment analysis (GSEA) further suggested suppression of fatty acid metabolism and iron-related pathways, alongside enrichment of skin development and cell–matrix adhesion processes (Fig. 3H). Collectively, these findings suggest that hypoxia is associated with transcriptional changes linked to ferroptosis-related processes and extracellular matrix regulation in dermal fibroblast aging. Transcriptomic profiling of distinct HDFs aging models under hypoxic conditions. (A) PCA showing distinct clustering of aged HDFs exposed to normoxia or hypoxia. (B) Volcano plots of RNA-seq data displaying significantly up- and downregulated genes in three aging models: hypoxia–chronologically aged vs. normoxia–chronologically aged; hypoxia–Doxorubicin vs. normoxia–Doxorubicin; and hypoxia–H₂O₂ vs. normoxia–H₂O₂. (C, D) Overlap of co-upregulated and co-downregulated genes across the three models. (E) GO enrichment analysis of co-downregulated DEGs, performed using Fisher’s exact test with Benjamini–Hochberg correction (FDR < 0.05). (F, G) Heatmaps of co-upregulated and co-downregulated genes. (H) Gene set enrichment analysis (GSEA) comparing hypoxia- and normoxia-treated chronologically aged HDFs Ferroptosis-associated stress promotes senescence in HDFs We next examined whether ferroptosis-associated signaling is activated in aging models. Senescent HDFs exhibited increased expression of ferroptosis-related proteins, including ACSL4 and 15-LOX2, along with accumulation of lipid peroxidation markers (4-HNE) and increased expression of MMP-1 and MMP-3. In contrast, the anti-ferroptotic enzyme GPX4 was reduced (Fig. 4A). The GSH/GSSG ratio was decreased, indicating impaired antioxidant capacity, and SA-β-gal activity was increased (Fig. 4B–C). Ferroptosis-associated oxidative stress was further confirmed by increased lipid peroxidation (C11-BODIPY staining) and elevated intracellular Fe²⁺ levels (FerroOrange staining) (Fig. 4D). Ferroptosis-associated stress promotes senescence in HDFs. (A) Immunoblotting of pro-ferroptosis factors, metalloproteinases, and the anti-ferroptosis factor in aged models (H₂O₂, UVB, Doxorubicin). (n = 3 each). (B, C) Quantification of GSH/GSSG ratio and SA-β-gal–positive cells in aged models using an ELISA. (n = 3, 6). (D) Lipid peroxidation and intracellular Fe²⁺ measured with C11-BODIPY and FerroOrange probes; representative images shown. (n = 5, 5). (E) Schematic of RSL3- or erastin-induced ferroptosis-driven aging. HDFs were pretreated with RSL3 (200 nM) or erastin (10 µM) for 4 h, followed by 48 h incubation. (F, G) Immunoblotting of ACSL4, 15-LO2, 4-HNE, GPX4, COL1A1, COL3A1, MMP-1, MMP-3, PARP, c-cas3, and p21 in RSL3/erastin-treated HDFs. (H, I) Quantification of SA-β-gal–positive cells and GSH/GSSG ratio in RSL3- and erastin-treated HDFs using an ELISA. (n = 3, 6). (J) Lipid peroxidation and intracellular Fe²⁺ in RSL3/erastin-treated HDFs detected by C11-BODIPY and FerroOrange. (n = 5, 5). (K) Immunoblotting of dorsal skin lysates from mice aged 3–24 months. Data are presented as mean ± SD. Quantification of the results is shown *P < 0.05, **P < 0.01, and ***P < 0.001 To determine whether ferroptosis contributes to premature senescence, young HDFs were treated with erastin or RSL3 at sub-lethal concentrations, followed by washout (Fig. 4E). These conditions induced a premature senescence phenotype, including increased intracellular Fe²⁺, enhanced lipid peroxidation, elevated MMPs expression, and higher SA-β-Gal activity, together with reduced collagen synthesis, downregulated GPX4, and a decreased GSH/GSSG ratio (Fig. 4F–J). Importantly, doses below 10 µM for Erastin or 200 nM for RSL3 did not induce PARP cleavage or caspase-3 activation, confirming that senescence occurred without apoptosis (Fig. 4F, G). These doses were therefore selected as the ferroptosis-associated senescence model for subsequent experiments. In chronologically aged mouse skin, ferroptosis-related markers (DPP4, ACSL4, 15-LOX-2, and 4-HNE) were similarly upregulated, along with reduced GPX4 expression (Fig. 4K). Together, these findings demonstrate the activation of pro-ferroptotic signaling in both senescent dermal fibroblasts and aged skin. Hypoxia-mediated elimination of DPP4 protects HDFs from ferroptosis-associated senescence DPP4, a surface marker of senescent cells [24], was found among the co-downregulated genes in hypoxia-exposed aged HDFs in our RNA-seq dataset (Fig. 3C, F). However, the role of DPP4 in ferroptosis-associated aging, particularly in dermal fibroblasts, remains poorly defined. Analysis of the Human Protein Atlas database revealed that fibroblasts (clusters 0 and 3) exhibit the highest DPP4 expression among 18 skin cell clusters, indicating a high basal or constitutive level in dermal fibroblasts (Fig. 5A). DPP4 expression was significantly higher in lower-leg skin, which is more exposed to sunlight, compared with suprapubic skin, a sun-protected site (Fig. 5B). A human skin tissue array (SKN1001) further demonstrated that elevated DPP4 protein levels in sun-exposed regions (face, neck) compared with sun-protected regions (chest, back, groin, perineum, hip, thigh) (Fig. 5C, Fig. S3). In order to analyzed the co-occurrence of DPP4 in senescence cells, the double positive (DPP4⁺/SA-β-Gal⁺ HDFs) were analyzed through flow cytometry. To examine the association between DPP4 and senescent cells, the proportion of DPP4⁺/SA-β-Gal⁺ HDFs was quantified by flow cytometry (Fig. S5A). Both SA-β-Gal⁺ and DPP4⁺/SA-β-Gal⁺ populations were increased in Erastin- or RSL3-treated cells compared with controls, and these increases were attenuated under hypoxic conditions (Fig. 5D, E; Fig. S5B). Notably, the proportion of DPP4⁺ cells alone was reduced in both normoxic and hypoxic ferroptotic conditions, suggesting a shift toward DPP4⁺ senescent cells (Fig. S5C). Consistently, DPP4 mRNA expression was decreased in hypoxia-exposed Erastin- or RSL3-treated HDFs (Fig. 5F). We next evaluated whether DPP4 modulation influences ferroptosis-associated senescence in young HDFs under normoxic conditions. MTT assays confirmed that treatment with DPP4 inhibitors (saxagliptin or sitagliptin) at concentrations up to 50 µM did not significantly affect cell viability (Fig. S5D, E). Under these conditions, pharmacological inhibition of DPP4 attenuated Erastin- or RSL3-induced senescence-associated changes, including reduced MMP-1 and MMP-3 expression, preservation of collagen (COL1A1, COL3A1), and decreased p21 levels (Fig. 5G, H). Consistently, siRNA-mediated knockdown of DPP4 recapitulated these effects (Fig. S5F). In contrast, DPP4 overexpression exacerbated Erastin- or RSL3-induced senescence-associated phenotypes, as indicated by increased expression of MMP-1, MMP-3, and p21, along with reduced collagen levels under normoxic conditions (Fig. 5I). Furthermore, enforced DPP4 expression under hypoxic conditions partially attenuated the protective effects of hypoxia (Fig. S4G). Collectively, these findings suggest that DPP4 is associated with ferroptosis- and senescence-related phenotypes in dermal fibroblasts, and that hypoxia-mediated downregulation of DPP4 may contribute to reduced accumulation of senescent HDFs. Hypoxia-mediated elimination of DPP4 protects HDFs from ferroptosis-associated senescence. (A) Human Protein Atlas database analysis of DPP4 expression across 18 skin cell clusters. (B) DPP4 mRNA expression in human skin from suprapubic (non–sun-exposed) and lower leg (sun-exposed) regions (Human Protein Atlas, https://www.proteinatlas.org). (C) Quantification of DPP4 IHC staining in sun-exposed and non–sun-exposed skin using a normal human skin tissue array (SKN1001). (n = 8, 76). (D, E) Flow cytometry of young HDFs treated with erastin (10 µM) or RSL3 (200 nM) for 4 h, washed, and cultured under normoxia or hypoxia for 48 h; quantification of DPP4⁺SPiDER-βGal⁺ cells (Q2 region). (n = 3 each). (F) qRT-PCR quantification of DPP4 mRNA relative to housekeeping controls. (n = 3 each). (G, H) Young HDFs treated with erastin or RSL3 for 4 h, washed, and subsequently treated with saxagliptin (50 µM) or sitagliptin (50 µM) under normoxia for 48 h; immunoblotting of COL1A1, COL3A1, MMP-1, MMP-3 and p21. (n = 3 each). (I) Overexpression of HA-DPP4 or empty vector (3 µg/mL) in young and aged HDFs after ferroptosis induction; immunoblotting of DPP4, COL1A1, COL3A1, MMP-1, MMP-3, and p21. (n = 3 each). Data are presented as mean ± SD. Quantification of the results is shown *P < 0.05, **P < 0.01, and ***P < 0.001 Hypoxia restricts NF-κB–dependent transcriptional activation of DPP4 in aged HDFs To investigate the mechanism underlying DPP4 upregulation in Erastin- or RSL3-induced senescent HDFs, we examined its transcriptional regulation. Previous studies have indicated that several transcription factors (TFs), including SP1, FOS, and NFκB, may bind to the DPP4 promoter [29, 30]. To determine their roles, siRNA-mediated silencing of SP1, FOS, and NFκB was performed under ferroptotic stress. Silencing of SP1, FOS and NFκB markedly reduced DPP4 mRNA levels in aged HDFs (Fig. 6A). Consistently, Western blot analysis confirmed effective protein silencing and demonstrated that NF-κB inhibition resulted in the greatest reduction in DPP4 protein levels (Fig. 6B), suggesting that NF-κB is a major regulator of DPP4 expression in this context. Hypoxia restricts NF-κB–dependent transcriptional activation of DPP4 in aged HDFs. (A) HDFs were pretreated with erastin or RSL3 for 4 h, washed, and transfected with siRNAs (40 nM; NC, SP1, FOS, NFκB) for 48 h under normoxia; DPP4 mRNA expression was analyzed by qRT-PCR. (n = 3 each). (B) Protein expression of DPP4, c-Fos, SP1, and NFκB were analyzed by immunoblotting. (n = 3 each). (C) Prediction of putative NFκB response elements within the DPP4 promoter using the Animal TFDB database. (D, E) Schematic diagram of DPP4 promoter deletion constructs. HDFs were treated with erastin or RSL3 for 4 h, washed, and transfected with DPP4 promoter constructs for 48 h under normoxia or hypoxia; promoter activity was assessed by luciferase reporter assay. (n = 3 each). (F) Sequence analysis identified a putative NFκB binding element in the DPP4 promoter (nt − 1563 to − 1549). (n = 3 each). (G) Chromatin was immunoprecipitated using an anti–NF-κB p65 antibody, and enrichment of the DPP4 promoter region was quantified by qPCR. Results are presented as percent input (% input). (n = 3 each). (H, I) Schematic of wild-type (NFKB-BS-WT) and mutant (NFKB-BS-MT) DPP4 promoter constructs. Promoter activity was measured by luciferase assay in aged HDFs under normoxia or hypoxia. (n = 3 each). (J) HDFs were treated with erastin (10 µM) or RSL3 (200 nM) for 4 h, washed, and cotransfected with DPP4 reporter constructs (2 µg) and pCMV3–NFκB–Flag plasmid (2 µg) under hypoxia; promoter activity assessed by luciferase assay. (n = 3 each). (K) Protein expression of NFκB, DPP4, COL1A1, COL3A1, MMP-1 and p21 were analyzed by immunoblotting in aged HDFs transfected with pCMV3–NFκB–Flag under hypoxia. (n = 3 each). Data are presented as mean ± SD. Quantification of the results is shown *P < 0.05, **P < 0.01, and ***P < 0.001 To predict potential regulatory elements, we used the integrated Animal TFDB resource, which incorporates hTFtarget, HOCOMOCO, and CIS-BP databases, and identified a putative NFκB response element (− 1563 to − 1549 nt) within the DPP4 promoter (Fig. 6C). The results of luciferase reporter assays with DPP4 promoter deletion constructs demonstrated that the − 2 kb promoter fragment was strongly activated by Erastin or RSL3 under both normoxic and hypoxic conditions, whereas shorter promoter constructs (− 1.3 kb, − 0.7 kb, − 0.2 kb) failed to respond, indicating that the regulatory region is localized within the (− 2000 to − 1300 nt) fragment (Fig. 6D–E). Hypoxia significantly reduced DPP4 promoter activity in Erastin- and RSL3-treated HDFs compared with normoxia (Fig. 6E), suggesting hypoxia-associated attenuation of transcriptional activation. The conserved NFκB-binding motif (5′-TCGGGAAATTCCCTA-3′) was localized to this region between − 1563/−1549 nt (Fig. 6F). A conserved NF-κB–binding motif (5′-TCGGGAAATTCCCTA-3′) was identified within the DPP4 promoter at positions − 1563 to − 1549 nt (Fig. 6F). To determine whether NF-κB p65 binds to this putative response element, chromatin immunoprecipitation (ChIP) assays were performed. The DPP4 promoter region was enriched using an NF-κB–specific antibody, confirming direct binding. This interaction was enhanced in Erastin- or RSL3-treated cells, whereas hypoxia markedly reduced NF-κB binding to the DPP4 promoter (Fig. 6G), suggesting that hypoxia suppresses NF-κB–mediated transcriptional regulation of DPP4. To further validate the functional relevance of this binding site, a mutant construct (NFκB-BS-MT) was generated. Mutation of the NF-κB–binding site abolished Erastin- and RSL3-induced promoter activation compared with the wild-type construct (NFκB-BS-WT) under both normoxic and hypoxic conditions (Fig. 6H–I). To further validate this interaction, NFκB was overexpressed using a pCMV-NFκB-Flag plasmid. NFκB overexpression restored DPP4 promoter activity under hypoxia in the wild-type construct but not in the NFκB-BS-MT, confirming the essential role of the − 1563/−1549 nt binding site (Fig. 6J). Functionally, NF-κB overexpression increased DPP4 protein expression and was associated with enhanced senescence-related phenotypes in Erastin/RSL3-treated HDFs, including increased expression of MMP-1 and p21 and reduced collagen expression (COL1A1, COL3A1) (Fig. 6K). Taken together, these findings demonstrate that NFκB directly regulates DPP4 transcription via its binding site at − 1563/−1549 nt of the promoter and that hypoxia is associated with reduced NF-κB–dependent transcriptional activation of DPP4. Hypoxia restrains sDPP4-mediated paracrine induction of ferroptosis-associated stress and senescence in dermal fibroblasts Transmembrane DPP4 can be cleaved by MMPs to generate sDPP4 to affect other cells. We next examined the role of extracellular sDPP4 signaling in Erastin- and RSL3-induced ferroptotic aging of HDFs. To assess whether hypoxia restricts sDPP4 secretion from senescent HDFs to neighboring young HDFs, conditioned medium (CM) was collected from drug-treated donor HDFs and transferred to recipient young HDFs for 3 days (Fig. 7A). sDPP4 levels in donor HDF media revealed that hypoxia markedly reduced Erastin/RSL3-induced sDPP4 secretion (Fig. 7B). Recipient HDFs exposed to CM from hypoxia-treated aged cells displayed reduced induction of lipid peroxidation (MDA), MMP-1, MMP-3, and senescence marker p21, together with higher collagen (COL1A1, COL3A1) expression, compared with CM from normoxia-treated aged HDFs (Fig. 7C, E). Fluorescent co-staining results further confirmed these findings. Recipient young HDFs treated with CM from hypoxia-exposed aged HDFs showed lower 4-HNE (red) and SPIDER-β-gal (green) signals, with diminished colocalization (yellow), indicating reduced co-occurrence of ferroptosis-associated stress and senescence compared with CM from normoxia-treated cells (Fig. 7D). To assess the contribution of DPP4 to these paracrine effects, DPP4 was silenced in donor HDFs prior to CM collection. CM from DPP4-silenced cells showed a reduced capacity to induce lipid peroxidation and senescence in recipient HDFs, while preserving collagen expression (Fig. 7F, G). Conversely, treatment of young HDFs with recombinant human DPP4 (rhDPP4) induced lipid peroxidation and increased senescence-associated markers (Fig. 7H, I), supporting a contributory role of sDPP4 in these processes. Given that sDPP4 has been reported to interact with PAR2 and activate pro-inflammatory signaling [31], we further investigated whether this pathway mediates sDPP4-induced effects. Inhibition of PAR2 reduced lipid peroxidation and decreased the expression of aging-related markers (p21 and γ-H2AX) in response to sDPP4 treatment (Fig. 7J, K). Collectively, these findings support that sDPP4 contributes to paracrine propagation of ferroptosis-associated stress and senescence in dermal fibroblasts, and that hypoxia is associated with attenuation of this extracellular signaling, at least in part through reduced sDPP4 availability. Hypoxia restrains sDPP4-mediated paracrine induction of ferroptosis and senescence in dermal fibroblasts (A) Schematic overview of the experimental design to test whether ferroptotic cell–derived medium (CM) can induce senescence via sDPP4 release. (B) Secreted sDPP4 in CM of donor HDFs were analyzed using an ELISA. (n = 4 each). (C) CM from erastin- or RSL3-treated donor HDFs under normoxia or hypoxia was transferred to recipient young HDFs for 48 h; protein expression of COL1A1, COL3A1, MMP-1, MMP-3 and p21 was analyzed by immunoblotting. (n = 3 each). (D) Representative images of recipient HDFs stained for 4-HNE (red) and senescence probe SPiDER-βGal (green). (E) Relative intracellular MDA levels in recipient HDFs cultured with CM. (n = 3 each). (F) Donor HDFs was transfected with si-DPP4 or si-NC (40 nM), and CM was transferred to recipient HDFs for 48 h; protein expression in donor and recipient HDFs was analyzed by immunoblotting. (n = 3 each). (G) Relative intracellular MDA levels in recipient HDFs cultured with CM from si-DPP4 or si-NC donor HDFs. (n = 4 each). (H) HDFs treated with rhDPP4 (20, 50, 100 ng/mL) for 48 h; protein expression was analyzed by immunoblotting. (I) Relative intracellular MDA levels in HDFs treated with rhDPP4 (100 ng/mL, 48 h). (n = 4 each). (J, K) HDFs were transfected with siRNA targeting PAR2 or si-NC, followed by rhDPP4 treatment for 48 h; protein expression and MDA levels were analyzed by immunoblotting and ELISA, respectively. (n = 3, 4). Data are presented as mean ± SD. Quantification of the results is shown *P < 0.05, **P < 0.01, and ***P < 0.001 Hypoxia attenuates Erastin/RSL3-induced sDPP4 secretion and aging in HDFs through HIF-1α–dependent modulation of NF-κB activity To determine whether hypoxia-induced HIF-1α mediates NF-κB inactivation and attenuates erastin/RSL3-induced ferroptosis-associated stress and cellular senescence, aged HDFs were transfected with HIF-1α siRNA and subsequently cultured under hypoxic conditions. Western blot analysis confirmed efficient knockdown of HIF-1α protein expression (Fig. 8A). Silencing of HIF-1α enhanced Erastin/RSL3-induced NF-κB activity under hypoxia, as evidenced by increased DPP4 promoter luciferase activity and enhanced binding of NF-κB to the DPP4 promoter, as determined by ChIP-qPCR. These changes were accompanied by elevated DPP4 protein expression and increased secretion of soluble DPP4 (sDPP4) (Fig. 8A–D). Given that phosphorylation and subsequent degradation of IκBα facilitate NF-κB p65 nuclear translocation, we examined NF-κB subcellular localization. Knockdown of HIF-1α increased nuclear NF-κB levels, accompanied by a reduction in cytosolic IκBα in hypoxia-treated, Erastin/RSL3-exposed HDFs (Fig. 8E, F, H). Moreover, HIF-1α depletion markedly aggravated erastin/RSL3-induced ferroptosis-associated senescence, as indicated by increased expression of pro-inflammatory SASP factors, including MMP-1, and the senescence-associated marker p21, along with a significant increase in the proportion of SA-β-Gal–positive cells (Fig. 8A, E, G). Collectively, these findings support that HIF-1α negatively regulates NF-κB activity under hypoxic conditions and is associated with attenuation of sDPP4 levels and ferroptosis-associated senescence in aged HDFs. Hypoxia restrains sDPP4-mediated paracrine induction of ferroptosis-associated stress and senescence in dermal fibroblasts (A) HDFs transfected with HIF-1α siRNA or NC siRNA (40 nM) were subjected to erastin/RSL3-induced ferroptotic stress and cultured under normoxia or hypoxia for 48 h; proteins expression was analyzed by immunoblotting. (n = 3 each). (B) DPP4 promoter luciferase activity was accessed by cotransfected with wild-type (NFKB-BS-WT) and HIF-1α siRNA or NC siRNA and subjected to erastin/RSL3-induced ferroptotic stress. (n = 3 each). (C) Levels of secreted DPP4 (sDPP4) in conditioned medium (CM) were measured by ELISA. (n = 3 each). (D) Chromatin was immunoprecipitated using an anti–NF-κB p65 antibody, and enrichment of the DPP4 promoter region was quantified by qPCR. Results are presented as percent input (% input). (n = 3 each). (E) Representative SA-β-gal staining (green) and p-NFĸB (green) images showing senescence and nuclear localization of NFĸB, respectively. (F) Quantification of nuclear NF-κB–positive cells from panel (D) using ImageJ. (n = 5 each). (G) Quantification of SA-β-gal–positive cells using an ELISA-based senescence assay, normalized to total cell number. (n = 6 each). (H) Cytoplasmic and nuclear fractions were isolated using a nuclear–cytosol separation kit and analyzed by immunoblotting. Data are presented as mean ± SD. Quantification of the results is shown *P < 0.05, **P < 0.01, and ***P < 0.001 Inhibition of ferroptosis or stabilization of HIF-1α attenuates Erastin/RSL3-induced aging phenotypes Given that hypoxia reduced NF-κB activity in senescent cells, we next examined whether pharmacological stabilization of HIF-1α using the hypoxia mimetic DMOG, or scavenging of lipid peroxidation with Fer-1, could recapitulate these effects in HDFs. Our results showed that treatment with either DMOG or Fer-1 attenuated Erastin- or RSL3-induced NF-κB activation, as evidenced by reduced nuclear NF-κB levels accompanied by increased cytosolic IκBα (Fig. 9A, B, G-J). Consistently, both treatments reduced NF-κB–dependent transcriptional activation of the DPP4 promoter and decreased secreted sDPP4 levels in Erastin- or RSL3-exposed HDFs (Fig. 9C–F). Pharmacological stabilization of HIF-1α produced effects comparable to hypoxic exposure, supporting an inverse association between HIF-1α signaling and NF-κB–mediated DPP4 transcription (Fig. 9E, F). Given that lipid peroxidation–derived ROS can act as upstream activators of NF-κB signaling, scavenging lipid ROS by Fer-1 may attenuate this redox-sensitive pathway, thereby reducing NF-κB transcriptional activity and downstream DPP4 expression. In addition, both compounds mitigated Erastin- or RSL3-induced extracellular matrix remodeling, as indicated by increased COL1A1 and COL3A1 levels and reduced expression of sDPP4, MMP-1, MMP-3, and the senescence-associated marker p21 (Fig. 9A, B). Collectively, these findings support that stabilization of HIF-1α or inhibition of ferroptosis is associated with attenuation of NF-κB activity and reduction of ferroptosis-associated and senescence-related phenotypes in HDFs. Inhibition of ferroptosis or stabilization of HIF-1α attenuates Erastin/RSL3-induced aging phenotypes. (A, B) HDFs were treated with erastin or RSL3 for 4 h, washed, and incubated with DMOG, 100 µM or Fer-1, 10 µM for 48 h under normoxia; protein expression was analyzed by immunoblotting. (n=3 each). (C, D) Secreted sDPP4 in CM of donor HDFs were analyzed using an ELISA. (n=6 each). (E, F) DPP4 promoter luciferase activity was accessed by transfected with wild-type (NFKB-BS-WT). (n=3 each). (G) Nuclear translocation of NF-κB was assessed by immunofluorescence, with nuclei counterstained using DAPI. (H) Quantification of nuclear NF-κB–positive cells from panel (G) using ImageJ. (n=5 each). (I, J) Cytoplasmic and nuclear fractions were isolated using a nuclear–cytosol separation kit and analyzed by immunoblotting. Data are presented as mean ± SD. Quantification of the results is shown *P<0.05, **P<0.01, and ***P<0.001 Topical application of a HIF-1α activator or a ferroptosis inhibitor attenuates the hallmarks of dermal aging in naturally aged mice To evaluate whether stabilization of HIF-1α or inhibition of ferroptosis could mitigate skin aging in vivo, DMOG or Fer-1 was topically applied to a 3 × 2 cm area of dorsal skin in 24-month-old C57BL/6 mouse for 10 consecutive days (Fig. 10A). Histological analyses using H&E and Masson’s trichrome staining revealed that aged skin exhibited characteristic atrophic features, including reduced epidermal thickness and diminished dermal collagen content. Treatment with DMOG or Fer-1 was associated with partial restoration of these structural features, including increased epidermal thickness and improved collagen density and organization (Fig. 10B-D). Immunohistochemical analyses further showed that pro-inflammatory markers (MMP-1, MMP-3, IL-1α, and IL-1β) were elevated in aged dermal tissue and were reduced following DMOG or Fer-1 treatment (Fig. 10F; Fig. S6B, C). In parallel, expression of HIF-1α and the anti-ferroptotic enzyme GPX4, which were decreased in aged skin, were increased following treatment (Fig. 10F). Conversely, ferroptosis-associated markers, including DPP4 and lipid peroxidation adducts (4-HNE and MDA), were elevated in aged skin and were attenuated upon DMOG or Ferrostatin-1 application (Fig. 10E, F). Collectively, these findings suggest that pharmacological stabilization of HIF-1α or inhibition of ferroptosis is associated with attenuation of aging-related structural and molecular changes in murine skin. Topical application of a HIF-1α activator or a ferroptosis inhibitor attenuates the hallmarks of dermal aging in naturally aged mice. (A) Chronologically aged mice (24 months) were topically treated with Fer-1 (1 mg/kg) or DMOG (1 mg/kg) on a 3 × 2 cm dorsal skin area for 10 consecutive days. (B) H&E and Masson’s trichrome staining were performed to assess dermal morphology and collagen content. (n=6 each). (C) Quantification of epidermal thickness. (D) Quantification of dermal collagen levels. (E) MDA levels in dorsal skin were measured. (n=6 each). (F) IHC staining of HIF-1α, DPP4, GPX4 and 4-HNE in skin tissues. (n=3 each). Data are presented as mean ± SD. Quantification of the results is shown *P<0.05, **P<0.01, and ***P<0.001. (G) Proposed model of HIF-1α regulation in ferroptosis-induced senescence of HDFs Discussion Our findings support a model in which activation of HIF-1α is associated with suppression of NF-κB–dependent DPP4 transcription, accompanied by attenuation of ferroptosis-associated stress and senescence in dermal fibroblasts. Pharmacological stabilization of HIF-1α or inhibition of ferroptosis was associated with reduced aging-related alterations in vitro and in vivo, including dermal atrophy and collagen loss. A schematic summary of this working model is presented in Fig. 10G. Cells respond to low oxygen availability through the oxygen-sensitive protein HIF-1α. The activation of HIF-1α represents a double-edged sword in skin biology: on one hand, it serves as a critical regulator of skin barrier function, preventing external invasion and promoting tissue regeneration and remodeling during wound healing; on the other hand, excessive or dysregulated HIF- 1α activation can exert pro-tumorigenic effects, facilitating the transformation of benign cells into highly aggressive cancers [4]. Recent studies have drawn attention to the role of hypoxia-mediated HIF-1α in the regulation of cellular aging. For instance, fibroblasts under hypoxia show reduced SASP secretion, and in vivo, hypoxia mimetic treatment in aged mice alleviates muscle weakness and systemic SASPs expression across multiple organs [7]. Similarly, another study reported that HDFs cultured under 5% O₂ showed enhanced collagen synthesis and reduced aging markers compared with cells under 21% O₂ [12]. Consistent with these observations, our data demonstrate that hypoxia (1% O₂) attenuates senescence in HDFs, as evidenced by reduced SASP production, decreased SA-β-gal activity, and enhanced proliferative capacity (Figs. 1 and 2). Notably, HIF-1α levels declined with aging in both cultured fibroblasts and skin tissues, suggesting impaired hypoxia responsiveness in aged cells. These findings prompted further investigation into the mechanisms by which HIF-1α modulates dermal fibroblast aging. Disruption of iron homeostasis is a hallmark of aging and contributes to intracellular Fe²⁺ accumulation and redox imbalance [32]. Ferroptosis, an iron-dependent process characterized by lipid peroxidation, has been increasingly linked to aging and age-related diseases [33–36]. In the skin, ferroptosis-associated features have been implicated in UVB-induced damage and inflammatory skin conditions [22, 37–40]. However, its role in dermal fibroblast aging remains incompletely understood. Our transcriptomic analysis revealed that ferroptosis-related pathways were downregulated under hypoxic conditions in aged HDFs (Fig. 3), suggesting a potential link between hypoxia signaling and ferroptosis-associated processes. Importantly, our study models sub-lethal ferroptosis-associated stress rather than canonical ferroptotic cell death. Erastin and RSL3 were applied at concentrations that did not induce apoptosis or overt cell death (Fig. 4), but instead triggered lipid peroxidation and redox imbalance, leading to senescence phenotypes. This distinction may help reconcile conflicting reports in the literature. While robust ferroptosis induction has been shown to eliminate senescent cells (senolytic effect) [41–44], sub-lethal ferroptosis-associated stress may instead promote senescence and tissue dysfunction [22, 37, 39, 40, 45]. Consistent with this concept, our data demonstrate that ferroptosis-associated signaling contributes to HDF senescence. Notably, HIF-1α depletion further exacerbated ferroptosis- and senescence-associated phenotypes (Fig. 8), suggesting that HIF-1α acts as a negative modulator of ferroptosis-associated aging. DPP4 has been widely recognized as a surface marker of senescent cells [24], yet its functional role in aging remains incompletely defined. Emerging evidence indicates that DPP4 may play context-dependent roles in aging and disease. For example, targeted elimination of DPP4⁺ senescent cells via ferroptosis induction have been proposed as a senolytic strategy [41], whereas pharmacological inhibition of DPP4 has been reported to ameliorate aging phenotypes and improve tissue function [46, 47], highlighting its context-dependent role. In our study, DPP4 expression was increased in senescent HDFs and was negatively regulated by HIF-1α under hypoxic conditions (Fig. 8). Both genetic and pharmacological inhibition of DPP4 attenuated ferroptosis-associated senescence, supporting a contributory role for DPP4 in this process (Fig. 5). These findings suggest that, in dermal fibroblasts, DPP4 is not only a marker but also participates in senescence-associated signaling under ferroptosis-related stress conditions. At the transcriptional level, our data identify NF-κB as a major regulator of DPP4 expression in this system. Consistent with previous reports implicating NF-κB in skin inflammaging [48], inhibition of NF-κB reduced DPP4 expression, whereas NF-κB activation enhanced DPP4 transcription. We further demonstrated that NF-κB directly binds to a defined response element within the DPP4 promoter (− 1563/−1549 nt), and that hypoxia is associated with reduced NF-κB binding and transcriptional activity at this site (Fig. 6). Mechanistically, our data from subcellular fractionation and signaling analyses (Figs. 8 and 9) indicate that HIF-1α may modulate canonical NF-κB activation at the level of the IκBα. Specifically, HIF-1α depletion was associated with enhanced IκBα degradation and increased nuclear accumulation of NF-κB, whereas pharmacological stabilization of HIF-1α or inhibition of ferroptosis attenuated NF-κB activation, accompanied by preservation of cytosolic IκBα (Fig. 9). These findings suggest that HIF-1α may restrain NF-κB signaling by limiting upstream IKK activation and subsequent IκBα turnover. Although hypoxia has been reported to exert context-dependent effects on NF-κB signaling, our findings support that HIF-1α is associated with suppression of NF-κB–dependent DPP4 transcription in dermal fibroblasts. However, the precise molecular interface between HIF-1α and NF-κB—such as competition for shared coactivators (e.g., p300/CBP) remains to be elucidated. In addition to intracellular effects, our study highlights a paracrine role for DPP4. Senescent cells can propagate aging phenotypes through secretion of soluble factors, including SASPs [49, 50]. We demonstrate that conditioned medium from aged HDFs induces ferroptosis-associated stress and senescence in neighboring young fibroblasts, and that this effect is attenuated under hypoxic conditions (Fig. 7). Notably, sDPP4 was identified as a contributing mediator of this paracrine signaling. Reduction of DPP4 in donor cells diminished the pro-senescent effects of conditioned medium, whereas recombinant sDPP4 induced lipid peroxidation and senescence markers in recipient cells. These findings suggest that hypoxia limits paracrine propagation of senescence, at least in part, by reducing sDPP4 levels. Fer-1 is a selective ferroptosis inhibitor that scavenges lipid-based reactive oxygen species (ROS) and prevents lipid peroxidation of cellular membranes [51]. Previous studies have shown that Fer-1 attenuates ferroptosis and has been applied in models of inflammatory skin conditions, including psoriasis [20], UVB-induced inflammation [22, 39], and dermatitis [52]. DMOG, a competitive inhibitor of prolyl hydroxylase (PHD), blocks the hydroxylation and subsequent proteasomal degradation of HIF-1α, thereby promoting its stabilization [53]. Topical application of HIF-1α activators in mouse skin has been reported to enhance hair follicle growth [54–56]. Moreover, HIF-1α activation plays an important role in regulating epidermal proliferation, differentiation, and barrier function [57–59], as well as extracellular matrix (ECM) homeostasis, including collagen synthesis in dermal fibroblasts [6, 12]. While these studies highlight roles of ferroptosis inhibition and HIF-1α activation in skin biology, our findings extend this framework by showing that treatment with Fer-1 or DMOG is associated with attenuation of ferroptosis-associated senescence in vitro and reduction of aging-related features in chronologically aged mouse skin (Figs. 9 and 10). Together, these results support a model in which hypoxia-driven HIF-1α activation contributes to the maintenance of dermal homeostasis, at least in part by limiting oxidative stress–associated damage and preserving ECM integrity. However, several limitations should be considered. DMOG broadly inhibits PHD enzymes, and its effects may not be exclusively mediated through HIF-1α, but may also involve additional PHD-dependent pathways. This highlights the need for more selective approaches to delineate HIF-1α–specific mechanisms. In addition, the translational relevance of these findings should be interpreted with caution, as murine skin differs substantially from human skin in structure, thickness, and physiology. Future studies using human-relevant systems, such as ex vivo human skin, organotypic skin equivalents, or clinical samples, will be important to validate the efficacy and safety of topical Fer-1 or DMOG in the context of human skin aging. Importantly, the mechanistic model proposed in this study should be considered a working framework rather than a strictly linear pathway. While our data support a functional link between HIF-1α, NF-κB, and DPP4 in regulating ferroptosis-associated senescence, both NF-κB signaling and ferroptosis are integrated within broader stress-response networks. Thus, additional regulatory mechanisms may also contribute to the observed phenotypes and warrant further investigation. Conclusion In conclusion, our findings support a model in which Erastin- or RSL3 induced ferroptosis-associated stress is linked to activation of NF-κB and subsequent upregulation of DPP4, accompanied by increased expression of p21 and SASP factors and impaired collagen synthesis, thereby contributing to dermal aging. In parallel, cleavage of membrane-bound DPP4 generates sDPP4, which acts on neighboring fibroblasts to propagate a pro-aging microenvironment and reinforce ferroptosis- and senescence-associated changes. Our data further indicate that stabilization of HIF-1α is associated with suppression of NF-κB–dependent DPP4 transcription and attenuation of these pathological features. Collectively, these findings suggest that modulation of the HIF-1α–NF-κB–DPP4 axis may represent a potential therapeutic strategy for senescence-associated skin disorders. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Abbreviations dimethyloxalylglycine Dipeptidyl peptidase-4 extracellular matrix ferrostatin-1 glucagon-like peptide 1 glutathione peroxidase 4 human dermal fibroblasts Hypoxia-inducible factor-1α 4-hydroxynonenal senescence-associated β-galactosidase senescence-associated secretory phenotype malondialdehyde matrix metalloproteinases Author contributions Shang-Chuan Ng: Investigation, Project administration, Conceptualization, Writing – original draft. Shih-Wen Kao: Conceptualization, Investigation. Chih‑Jung Chen: Investigation. Shinn-Zong Lin: Conceptualization, Investigation Dennis Jine-Yuan Hsieh: Investigation. Chia-Hua Kuo: Investigation. Kuan-Ho Lin: Investigation. Chih-Yang Huang: Conceptualization, Supervision, Writing – review & editing. Wei-Wen Kuo: Conceptualization, Supervision, Writing – review & editing. Funding This study was supported by grants from the National Science and Technology Council, Taiwan (NSTC 114-2320-B-039-034 -MY3) and China Medical University (CMU114-MF-20). Data availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval All procedures were approved by the Institutional Animal Care and Use Committee of China Medical University and conformed to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (CMUIACUC-2024-070). Consent for publication Not applicable. Competing interests The authors declared no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Chih-Yang Huang and Wei-Wen Kuo contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.