ACOD1 deficiency promotes DDX1 methylation–mediated mitochondrial dysfunction and dermal papilla cell senescence in androgenetic alopecia

Abstract

Background

Androgenetic alopecia (AGA), the most prevalent form of hair loss, is driven by the dysfunction of dermal papilla cells (DPCs). Emerging evidence implicates DPC senescence in the pathogenesis of AGA; however, the underlying molecular mechanisms remain incompletely elucidated.

Methods

We employed a multi-faceted experimental approach, including analyses of human scalp tissues, primary DPCs, a dihydrotestosterone (DHT)-induced AGA mouse model, and immortalized DPCs stimulated with DHT. Cellular senescence was evaluated via expression of senescence markers (p16^INK4a, p21, p53) and senescence-associated β-galactosidase staining. Cell proliferation, migration, and apoptosis were also assessed. Mitochondrial function was evaluated using transmission electron microscopy, MitoTracker, MitoSOX, JC-1, and Seahorse assays. RNA sequencing and bioinformatics analyses were performed to identify differentially expressed genes. The interaction between Aconitate Decarboxylase 1 (ACOD1) and DDX1 was verified via co-immunoprecipitation, mass spectrometry, and molecular docking. Metabolomic analysis was performed to profile intracellular metabolic alterations. Functional experiments included ACOD1 knockdown/overexpression and exogenous supplementation with 4-octyl itaconate (4-OI). Hair follicle morphology and hair loss in AGA mice were evaluated following 4-OI treatment.

Results

Senescence markers were significantly elevated in AGA DPCs, accompanied by increased senescence-associated β-galactosidase staining, reduced cell proliferation and migration, and enhanced apoptosis. DHT-induced mitochondrial dysfunction in DPCs was characterized by increased mitochondrial fragmentation, decreased mitochondrial cristae, superoxide accumulation, reduced membrane potential, and impaired oxidative phosphorylation. RNA sequencing identified ACOD1 as significantly downregulated in DHT-treated DPCs. ACOD1 knockdown induced mitochondrial dysfunction, cellular senescence, and functional impairment in DPCs, while ACOD1 overexpression ameliorated these DHT-induced phenotypes. Mechanistically, ACOD1 interacted with DDX1 to inhibit its methylation; ACOD1 knockdown enhanced DDX1 methylation, correlating with mitochondrial dysfunction and DPC senescence. Metabolomic analysis demonstrated that ACOD1 knockdown significantly reduced itaconate levels. Exogenous 4-OI supplementation reduced DDX1 methylation, ameliorated DHT-induced mitochondrial dysfunction and cellular senescence, promoted cell proliferation and migration, suppressed apoptosis, mitigated hair follicle miniaturization, and alleviated hair loss in AGA mice.

Conclusions

Our findings reveal a novel mechanism underlying DPC senescence in AGA, wherein ACOD1 deficiency promotes DDX1 methylation, mitochondrial dysfunction, and subsequent DPC senescence. ACOD1 represents a promising therapeutic target for AGA, and 4-OI may have translational potential for the treatment of AGA.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12916-026-04967-w.

Background

Androgenetic alopecia (AGA) is the most common form of hair loss in clinical practice [1], characterized by progressive miniaturization of hair follicles [1–3] and clinical manifestations such as progressively reduced hair shaft diameter, decreased hair density, and varying degrees of baldness [1, 4, 5]. In China, the prevalence of AGA is approximately 21.3% in men and 6.0% in women [6]. This translates to an estimated 170 million patients in China alone, with hundreds of millions more affected worldwide. Its incidence increases with age [1, 7–9], and by age 70, at least 80% of men and 50% of women are affected [1]. AGA is not merely a cosmetic concern but also a condition that significantly impacts patients’ mental health, social confidence, and quality of life, thereby imposing a significant disease burden [5, 10]. It is now generally accepted that AGA development is driven by androgens in the context of genetic susceptibility [9], although the specific mechanisms are complex and incompletely elucidated [9, 11]. While adjunctive approaches such as low-level laser therapy and platelet-rich plasma are increasingly available, the cornerstone of pharmacological treatment remains topical minoxidil and oral finasteride [12, 13]. Long-term management of AGA may be challenging due to its chronic progression and the interplay of genetic and environmental factors [1, 11]. Therefore, further research into the pathogenesis of AGA is necessary to develop more effective and targeted therapies.

As the “command center” of hair follicles, dermal papilla cells (DPCs) play an irreplaceable role in hair follicle cycle regulation, morphology maintenance, and hair induction through the precise regulation of epithelial-mesenchymal interactions and key signaling pathways such as Wnt/β-catenin and BMP/TGF-β [14]. Hair follicle miniaturization is therefore thought to be due to abnormal dermal papilla (DP) function [2, 15], given that DP volume directly determines the size of the hair bulb and the diameter of the hair shaft, and influences the duration of the anagen phase. Single-cell analysis implicates DPCs as key drivers of AGA [16]. The current consensus is that AGA is androgen-driven in the context of genetic susceptibility, specifically dihydrotestosterone (DHT) acting on androgen receptors (AR) in DPCs at androgen-sensitive sites [17], triggering downstream signaling cascades that lead to shortened anagen and prolonged telogen, culminating in progressive hair follicle miniaturization [9, 11]. However, the mechanism by which DHT causes the progressive functional decline of DPCs remains a key unresolved issue in the field.

Cellular senescence is characterized by the loss of cellular function, cell cycle arrest, alterations in intercellular communication, disruption of protein homeostasis, and massive secretion of pro-inflammatory factors in response to injury and stress [18, 19]. In recent years, cellular senescence—a state of stable cell cycle arrest—has been shown to be closely associated with various age-related diseases [20, 21]. The incidence of AGA also increases with age [4, 7, 8], and consequently, its relationship with DPC senescence has garnered increasing attention. DPCs derived from the balding scalp of AGA patients rapidly exhibit a senescent phenotype in vitro [22], whereas those from the non-balding scalp gradually become senescent after several passages and subsequently lose their ability to induce hair follicle formation [22, 23]. Senescent DPCs secrete elevated levels of IL-6, which inhibits the proliferation of hair follicle keratinocytes and the colony-forming capacity of hair follicle stem cells, and prevents the transition of hair follicles from the telogen to the anagen phase [24]. This effect is consistent with the pathological process of AGA. DHT-induced DPC senescence has been shown to some extent [25–27]. Therefore, DPC senescence is likely a key link between DHT stimulation and hair follicle dysfunction and may directly contribute to the pathological progression of AGA.

There are multiple causes for cellular senescence. In addition to age-related replicative senescence, it is now generally accepted that alterations in cellular physiology induced by a variety of exogenous factors can also promote this process. These changes mainly include mitochondrial dysfunction, DNA damage, telomere attrition, uncontrolled oxidative stress, impaired autophagy, and other cellular perturbations [18]. Mitochondria are the cellular hubs for energy production and metabolism, and their dysfunction leads to redox imbalance (for example, excessive reactive oxygen species production). This imbalance is a key driver of cellular senescence. However, the potential role of mitochondrial dysfunction in DPC senescence and AGA progression, along with its underlying mechanisms, remains poorly understood; thus, further investigation is warranted.

Aconitate decarboxylase 1 (ACOD1), also known as immune-responsive gene 1 (IRG1), is a key enzyme in immunometabolism. It decarboxylates the mitochondrial metabolite cis-aconitate to produce itaconate, an immunomodulatory intermediate in the tricarboxylic acid (TCA) cycle of immune cells [28, 29]. Recently, itaconate has garnered considerable attention due to its critical regulatory roles in immunometabolism and redox balance. In macrophages under inflammatory or oxidative stress, oxidative phosphorylation (OXPHOS) is downregulated, while glycolysis is enhanced. Itaconate exerts anti-inflammatory effects by inhibiting glycolysis [30]. Additionally, itaconate inhibits succinate dehydrogenase activity, thereby reducing the production of mitochondrial ROS (mtROS) driven by mitochondrial complex I, thus suppressing the inflammatory response and improving mitochondrial function [30]. Recent studies have demonstrated that itaconate-derived prodrugs exhibit significant efficacy in the treatment of alopecia areata, effectively alleviating inflammation, protecting hair follicles, and promoting hair regrowth [31, 32]. It remains unclear whether the deficiency of ACOD1 contributes to DHT-induced DPC senescence by disrupting mitochondrial function and TCA cycle metabolism.

In this study, we found that ACOD1 binds to DDX1 and reduces its methylation, thereby maintaining the stability of the TCA cycle. Deficiency of ACOD1 disrupts the TCA cycle, leading to mitochondrial dysfunction and metabolic imbalance, which ultimately promotes DPC senescence and AGA progression. Exogenous supplementation of itaconate partially alleviates DHT-induced DPC senescence and AGA progression, suggesting that the ACOD1 represents a promising therapeutic target.

Methods

Human participants

This study was approved by the Ethics Committee of the First Affiliated Hospital with Nanjing Medical University (Approval number: 2024-SR-638). The study was conducted in accordance with the Declaration of Helsinki, and all participants provided full written informed consent prior to sample and data collection. Human AGA scalp tissues were obtained from 10 AGA patients (age: 34.80 ± 2.82 years) who underwent scalp surgery in the balding area. For healthy controls, scalp tissues were collected from 10 age-matched individuals (age: 40.4 ± 2.35 years) who underwent scalp surgery for conditions unrelated to alopecia or cancer, and were confirmed to be free of AGA prior to surgery. Portions of the scalp tissues were used for subsequent isolation and culture of primary human dermal papilla cells (detailed methods are provided below). For histological sectioning and staining, tissue samples with intact hair follicle structures and well-preserved longitudinal orientation were selected. A total of six AGA patients (age: 32.00 ± 2.94 years) and six healthy controls (age: 36.33 ± 2.62 years) were ultimately included for subsequent histological analyses.

Isolation and culture of primary human DPCs

Primary human DPCs were isolated and cultured according to previously established methods [33, 34]. Briefly, hair bulbs were excised from scalp tissues under a stereomicroscope and digested with 0.1% type I collagenase in Dulbecco’s modified Eagle medium (DMEM; Gibco, USA) at 37 °C for 2–3 h. The digestion process was monitored, and when the dermal papilla tissues appeared partially or fully detached, the reaction was terminated by adding DMEM supplemented with 20% fetal bovine serum (FBS; Gibco, USA). After two rounds of centrifugation, the pelleted dermal papilla tissues were resuspended in culture medium consisting of DMEM, 20% FBS, and 1% penicillin/streptomycin. The tissues were then transferred to a culture flask and incubated to allow for DPC migration. Primary DPCs were passaged at a 1:3 ratio using 0.25% trypsin (Beyotime, China) upon reaching 80% confluence or forming multilayered aggregates. Subcultured DPCs were maintained in growth medium (DMEM supplemented with 10% FBS and 1% penicillin/streptomycin), which was passaged every 3–4 days. Cells from passages 3–5 were used for subsequent experiments.

Immortalization and treatment of DPCs

When primary DPCs were cultured to the third passage and reached 50% confluence, they were infected overnight in a 37 °C, 5% CO₂ incubator with lentivirus expressing the SV40 large T antigen (multiplicity of infection = 80, containing 5 µg/mL puromycin). On the second day after infection, the medium was replaced, and the cells were cultured further, followed by puromycin selection for 4 days. The cells were then digested with 0.25% trypsin and plated into 60-mm dishes. When the cells reached full confluence, they were then routinely subcultured and maintained.

Stable AR-overexpressing DPC cell lines were established by infecting immortalized DPCs with pLVX-Neo-AR lentivirus. Referring to concentrations reported in previous studies [25, 35, 36], a DHT treatment concentration of 100 nM was selected. When immortalized DPCs reached 60%–70% confluence, they were washed twice with PBS, and the medium was replaced with fresh medium containing 100 nM DHT (Sigma-Aldrich, USA) for 72 h.

Establishment of the AGA animal model

For animal welfare considerations, all procedures were approved by the Institutional Animal Care and Use Committee of Nanjing Medical University (Approval No.: IACUC-2407031). All animal histological analyses and data quantification were performed by researchers blinded to the group assignments.

According to previous literature [37], 6-week-old male C57BL/6 mice (obtained from Nanjing Medical University) were acclimatized for one week under specific pathogen-free (SPF) conditions with constant temperature and humidity, a 12-hour light/dark cycle, and ad libitum access to standard laboratory chow and water. After the acclimatization period, the mice were randomly divided into a control group and a DHT-treated group (n = 5 per group). Subsequently, the mice were intraperitoneally injected with either corn oil (vehicle control) or DHT (1 mg/day, dissolved in corn oil; KKL MED, USA) for four consecutive days. Following this initial administration period, the mice were anesthetized with sodium pentobarbital, and their dorsal hair was removed using a rosin-wax mixture to synchronize the hair cycle and then photographed. This day was designated as Day 0 (D0). After depilation, administration of DHT or corn oil was continued (5 days/week, once daily) until tissue collection on Day 21. On Days 5, 9, 14, and 21 (before and after depilation), the dorsal skin and hair growth status of the mice were photographed and documented using a digital camera from a fixed distance. On Day 21, all mice were euthanized, and skin tissue samples were collected.

Hematoxylin–Eosin (H&E) staining

Scalp skin samples and mouse skin tissues were collected, dehydrated, embedded in paraffin, and sectioned into 4 μm-thick sections. The paraffin sections were deparaffinized using xylene and rehydrated through a graded ethanol series. After complete deparaffinization, the sections were stained with hematoxylin and eosin. Following staining, the sections were dehydrated through a graded ethanol series, cleared with xylene, and mounted with neutral resin. Images were observed and captured using an optical microscope (Olympus, Japan).

Immunohistochemical (IHC) staining

For immunohistochemical staining, paraffin sections were first thoroughly deparaffinized. Subsequently, antigen retrieval was performed on the deparaffinized sections using a sodium citrate antigen retrieval solution (Beyotime, China). Next, the sections were treated with a 3% hydrogen peroxide solution for 15 minutes at room temperature to inactivate endogenous enzymes, and then washed three times with PBS. Following this, the sections were permeabilized with 0.3% Triton X-100 for 15 minutes at room temperature, and then blocked with 5% BSA (Biosharp, China) for one hour. After diluting the primary antibody according to the manufacturer’s instructions, the diluted primary antibody was incubated with the sections overnight at 4°C. The next day, the primary antibody was discarded, and the sections were washed three times. The HRP-labeled secondary antibody (ZSGB-BIO, China) was diluted according to the manufacturer’s instructions and incubated with the sections for 1 hour at room temperature. The sections were washed three times with PBS to remove any unbound secondary antibody, and then the sections were developed using a 3,3’-diaminobenzidine (DAB) kit. Finally, the nuclei were counterstained with hematoxylin, and the sections were mounted with neutral resin. All images were observed and captured using an optical microscope (Olympus, Japan).

Immunofluorescence (IF) staining

Immunofluorescence staining was performed on thoroughly deparaffinized sections. Antigen retrieval was conducted on the deparaffinized sections using a sodium citrate antigen retrieval solution (Beyotime, China). Subsequently, similar to the IHC staining procedure described above, the sections were permeabilized with 0.3% Triton X-100 for 15 min and then blocked with 5% BSA (Biosharp, China) for 1 h at room temperature. The sections were then incubated with the diluted fluorescent primary antibody overnight at 4 °C. The next day, the primary antibody was discarded, and the sections were washed three times. Following this, the sections were incubated for 1 h at room temperature with either an Alexa Fluor 488- or Alexa Fluor 594-labeled secondary antibody (Beyotime, China) corresponding to the primary antibody species. The sections were washed another three times. Nuclei were stained using a DAPI solution for 10 min. Finally, the sections were mounted using an anti-fade mounting medium. Images of the sections were observed and captured using either a fluorescence microscope or a confocal microscope (Leica, Germany). Fluorescence intensity and the ratio of positive cells were analyzed using ImageJ software.

RNA extraction and quantitative RT-PCR

According to the manufacturer’s instructions, total RNA was extracted from scalp tissues and DPCs using the RNA-easy isolation reagent (Vazyme, China). Subsequently, 1000 ng of total RNA was reverse-transcribed into complementary DNA (cDNA) using HiScript II Q RT SuperMix kit (Vazyme, China). Real-time quantitative PCR (qPCR) was performed on the StepOne Plus PCR System (Applied Biosystems, USA) with SYBR Green PCR Master Mix (Applied Biosystems, USA). Transcript levels were normalized to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and relative gene expression was calculated using the 2^−ΔΔCt method. All experiments were performed with three biological replicates per group. The primer sequences used for qRT-PCR are listed in the table below.

Gene	Forward (5’-3’)	Reverse (5’-3’)
CDKN2A	AGGGCTTCCTGGACACGCTGGTGGT	CGGCATCTATGCGGGCATGGTTA
CDKN1A	TGATTAGCAGCGGAACAAGGAGT	TGGAGAAACGGGAACCAGGACAC
ACOD1	ACAACGAAATGATGCTCAAGTCTA	CGATCTGTCAGGTGTCCCAC
GAPDH	GATTCCACCCATGGCAAATTC	CTGGAAGATGGTGATGGGATT

Western Blot (WB)

Western blot was used to detect the levels of target proteins in tissues and cells. Tissue and cell protein samples were collected using RIPA lysis buffer (Beyotime, China), and protein concentrations were determined with a BCA protein quantification kit (Beyotime, China). After centrifugation to remove cell debris, loading buffer was added to the protein supernatant, followed by heating at 95 °C for 10 min to denature the proteins. The protein samples were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto a polyvinylidene fluoride (PVDF) membrane (Millipore, USA) using transfer buffer (NCM Biotech, China). After blocking with 5% bovine serum albumin (BSA) at room temperature for 1 h, the PVDF membrane was incubated with the corresponding primary antibodies at 4 °C overnight. The next day, the membrane was incubated with the corresponding HRP-labeled secondary antibodies at room temperature for 1 h. After washing with TBST buffer, the protein bands were visualized using an ECL detection reagent (Thermo Fisher Scientific, USA), and the results were recorded using a Tanon 5200 system. Detailed information on the antibodies used in this study is provided in the table below.

Antibody	Source	Catalog No.
p16INK4a(For WB and IHC)	Abcam	ab189034
p16INK4a(For IF)	Cell Signaling Technology	18,769
p21	Abcam	ab109520
p53	Proteintech	10442-1-AP
β-Actin	Proteintech	66009-1-Ig
ACOD1(For WB and IP)	Abcam	ab222411
ACOD1(For IF and IHC)	Invitrogen	PA5-102893
VDAC	Proteintech	10866-1-AP
Histone H3	Proteintech	17168-1-AP
GAPDH	Proteintech	60004-1-Ig
DDX1	Proteintech	11357-1-AP
Flag	Proteintech	20543-1-AP
His	Proteintech	66005-1-Ig
Methylation	Cell Signaling Technology	14,679
Vimentin	Proteintech	60330-1-Ig
GRP75	Santa Cruz	sc-133,137

Senescence-associated β-galactosidase (SA-β-gal) staining

SA-β-gal staining of DPCs was performed using a β-galactosidase staining kit (Beyotime, China). DPCs from different treatment groups were seeded in 24-well plates at a density of 2 × 10⁴ cells per well. According to the manufacturer’s instructions, cell fixation was performed and the staining working solution was prepared separately. The staining working solution was added to the wells, and the plates were incubated overnight in a 37 °C CO₂-free incubator. The next day, the staining solution was discarded, and the cells were washed once with phosphate-buffered saline (PBS). Subsequently, images were observed and captured using an optical microscope (Olympus, Japan), and the positive staining rate was calculated using ImageJ software.

DPC proliferation assay

The proliferative capacity of DPCs from different treatment groups was detected using the EdU assay kit (Ribobio, China). DPCs were seeded in 24-well plates at a density of 2 × 10⁴ cells per well and subjected to different treatments. According to the manufacturer’s instructions, a 50 µM EdU-containing culture medium was prepared, added to the wells, and incubated with the cells for 24 h. After incubation, the procedures for cell fixation, Apollo staining, and nuclear staining were continued as per the kit manual. After staining was completed, the proliferation rates of DPCs in the different treatment groups were observed and recorded using a flow cytometer and a fluorescence microscope, respectively.

DPC migration assay

The migratory ability of DPCs was detected using the Transwell assay and the scratch assay, respectively. In the Transwell assay, DPCs were seeded into the upper chamber of a Transwell insert (0.4 μm pore size, Corning) at a density of 1 × 10⁴ cells per well, and different treatment-conditioned media were added to the lower chamber, followed by incubation for 24 h. After incubation, the upper chamber of the Transwell was fixed with 4% paraformaldehyde (PFA) for 20 min, and the upper surface of the membrane was wiped to remove non-migrated cells. After staining with 0.5% crystal violet for 30 min, the upper chamber was washed three times with PBS. Then, the number of migrated DPCs in each group was observed and counted using an optical microscope (Olympus, Japan). In the scratch assay, DPCs were seeded into 6-well plates until the cell density reached 100% confluence. A sterile 200 µL pipette tip was used to create a scratch on the cell monolayer. Prior to scratch creation, cells were treated with mitomycin C (1 µg/mL) for 1 h to exclude the confounding effect of cell proliferation. After scratching, the cells were gently washed three times with PBS and cultured with complete treatment-conditioned media. To assess the effect of different treatments on DPC migration, cell migration was observed and photographed at 0 h and 24 h.

DPC apoptosis assay

The apoptosis rate of DPCs was detected using the BeyoClick™ EdUTP-594 TUNEL apoptosis detection kit (Beyotime, China). DPCs were seeded in 24-well plates at a density of 2 × 10⁴ cells per well and subjected to group-specific treatments. Following the manufacturer’s instructions, after sequential washing, fixation, and permeabilization, the EdUTP labeling solution was added to label fragmented DNA of apoptotic cells. Subsequently, Hoechst staining was performed for nuclear counterstaining. After staining, the apoptosis of DPCs in different treatment groups was observed and captured using a flow cytometer a fluorescence microscope.

RNA sequencing

Total RNA was extracted from DPCs and DHT-stimulated DPCs using TRIzol reagent (Invitrogen, USA), and treated with Rnase-free Dnase I to eliminate genomic DNA contamination. RNA integrity was assessed by 1.5% agarose gel electrophoresis or a fragment analyzer, while RNA concentration and purity were measured using a NanoDrop spectrophotometer. Sequencing libraries were constructed with the VAHTS Small RNA Library Prep Kit for Illumina (Vazyme, China) following the manufacturer’s protocol. Briefly, 3’ and 5’ adapters were sequentially ligated to the RNA molecules; the ligated products were reverse-transcribed into cDNA; library templates were enriched by PCR amplification; and the target bands were excised and purified from 6% Novex TBE PAGE gel. The constructed libraries were quantified with a Qubit 4.0 Fluorometer (Thermo Fisher Scientific, USA). Sequencing was performed on an Illumina NovaSeq 6000 platform using the paired-end 150 bp (PE150) strategy and NovaSeq 6000 Reagent Kits (Illumina, USA). The sequencing workflow consists of: [1] bridge PCR-mediated amplification of DNA fragments on the flow cell to form clusters; and [2] cyclic sequencing-by-synthesis using fluorescently labeled deoxyribonucleoside triphosphates (dNTPs). For data analysis, gene expression levels were quantified, and differentially expressed genes (DEGs) were identified using the DEGseq tool with thresholds of |fold change| > 2 and p value < 0.05. After standard quality control and alignment of the RNA-Seq data, DESeq2 was used to perform differential expression analysis. The criteria for defining differentially expressed mRNAs were set as |log2 fold change| ≥ 1 and adjusted p value < 0.05. Multiple testing correction was performed using the Benjamini–Hochberg method to control the false discovery rate (FDR).

MitoTracker and MitoSOX staining

For MitoTracker staining, DPCs from different treatment groups were stained with MitoTracker Deep Red FM (Beyotime, China) to visualize mitochondrial morphology, and with Hoechst 33,342 to label cell nuclei. Images were observed and acquired using a confocal fluorescence microscope (Leica, Germany). For MitoSOX staining, DPCs were incubated with MitoSOX Red (Invitrogen, USA) to detect mitochondrial superoxide accumulation, followed by nuclear labeling with Hoechst 33,342. Images were captured using a fluorescence microscope (Leica, Germany). Quantitative analysis of mitochondrial structural parameters and MitoSOX fluorescence intensity was performed using ImageJ software.

Transmission electron microscopy

To investigate the effects of different treatments on mitochondrial morphology and structure in DPCs, transmission electron microscopy (TEM; Hitachi, Japan) was used for observation and image acquisition. Briefly, DPCs from different treatment groups were collected and fixed with electron microscopy fixative (Servicebio, China) at 4 °C for 2 h, followed by post-fixation with 1% osmium tetroxide for 2 h. After fixation, the samples were dehydrated through a graded series of ethanol concentrations (30%, 50%, 70%, 80%, 90%, 95%, and 100%), with each step lasting 20 min, and subsequently washed twice with acetone (15 min each). Subsequently, the cells were embedded in 812 epoxy resin, polymerized at 37 °C for 6 h, and further cured at 60 °C for 48 h. After polymerization, ultrathin Sects.  (60–80 nm) were cut using an ultramicrotome (Leica, Germany). The sections were double-stained with 2% uranyl acetate and 2.6% lead citrate, and then observed under the TEM to evaluate mitochondrial morphological characteristics, including size, shape, and cristae integrity. Images were acquired for subsequent analysis.

Mitochondrial membrane potential measurement

The mitochondrial membrane potential (ΔΨm) of DPCs from different treatment groups was measured using the JC-1 mitochondrial membrane potential assay kit (Beyotime, China). Briefly, DPCs were collected, washed three times with PBS, and then incubated with 1 mL of JC-1 staining working solution at 37 °C in a 5% CO₂ humidified incubator for 20 min in the dark. After incubation, the staining working solution was aspirated, and the cells were washed twice with JC-1 staining buffer. Subsequently, changes in mitochondrial membrane potential were detected using a flow cytometer (CytoFLEX, Beckman Coulter, USA), and the ratio of JC-1 aggregates to monomers was analyzed using FlowJo software (Version 10.8.1, BD Biosciences, USA).

Seahorse assay

Mitochondrial respiration and glycolytic function of DPCs were measured using the Seahorse XF96 mitochondrial stress test (to assess oxygen consumption rate, OCR) and Proton Efflux Rate (PER) assay (Agilent Technologies, USA), respectively. Briefly, DPCs from different treatment groups were seeded into a Seahorse XF96 cell culture plate at a density of 1 × 10⁴ cells per well and cultured overnight in a 37 °C, 5% CO₂ humidified incubator. One day prior to the experiment, the Seahorse XF96 sensor cartridge was hydrated by adding 180 µL of pre-warmed calibration solution to each assay well, followed by equilibration overnight in a 37 °C CO₂-free incubator. On the day of the experiment, the complete medium was aspirated from the cell culture plate, and the cells were washed three times with pre-warmed assay buffer (as specified in the kit protocol) before being incubated in a 37 °C CO₂-free incubator for 1 h. The cell culture plate and sensor cartridge were then loaded into the Seahorse XF96 Analyzer. According to the manufacturer’s instructions, oligomycin (2 µM), carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (FCCP, 3 µM), and a rotenone/antimycin A mixture (1 µM) were sequentially injected to monitor real-time changes in OCR. For the assessment of glycolytic function, DPCs were seeded and cultured as described above, followed by sequential injection of rotenone/antimycin A (0.5 µM) and 2-deoxy-D-glucose (50 mM) to record dynamic changes in PER. All experimental data were acquired and preliminarily analyzed using Seahorse XF96 Wave software, with OCR and PER values normalized to the corresponding cell count per well.

ATP content assay

ATP content was measured using an ATP assay kit (Beyotime, China). Briefly, DPCs from different treatment groups were collected and washed three times with PBS. An appropriate volume of ATP assay lysis buffer was then added to the cells, followed by thorough mixing. The cells were lysed on ice for 10 min. After complete lysis, the mixture was centrifuged at 4 °C and 12,000 rpm for 5 min, and the supernatant was collected for ATP content determination. According to the kit instructions, standard solutions and samples were added to a 96-well plate, and 100 µL of ATP detection working solution was added to each well immediately afterward. After gentle mixing (avoiding vigorous shaking), the luminescence of each well was measured immediately using a fluorescence microplate reader (Biotek, USA), with the plate being protected from light during the process. A standard curve was generated based on the standard concentrations and their corresponding fluorescence intensities. The ATP content in the samples was calculated using the standard curve and normalized to the protein concentration of the cells in each group. Finally, the relative differences in ATP content among the DPC groups were compared.

Chromatin immunoprecipitation (ChIP)-qPCR

Chromatin immunoprecipitation (ChIP) assays were performed to evaluate AR binding to the ACOD1 promoter in DPCs. Briefly, DPCs were treated with DHT for the indicated time period. Cells were cross-linked with 1% formaldehyde at room temperature for 10 min and subsequently quenched with 125 mM glycine. After washing with cold PBS, DPCs were lysed, and chromatin was fragmented by sonication to an average size of 200–500 bp.

The sheared chromatin was incubated overnight at 4 °C with either an anti-AR antibody or normal rabbit IgG (as a negative control), followed by immunoprecipitation using Protein A/G magnetic beads. After sequential washing, immune complexes were eluted, and cross-links were reversed at 65 °C overnight. DNA was then purified using a commercial DNA purification kit according to the manufacturer’s instructions.

Purified DNA was analyzed by qPCR using primers targeting the predicted AR-binding region within the ACOD1 promoter. qPCR was performed using SYBR Green Master Mix on the StepOne Plus PCR System (Applied Biosystems, USA). ChIP-qPCR enrichment was calculated as the percentage of input chromatin (% input). IgG immunoprecipitation served as the negative control. Relative enrichment levels were compared between the control and DHT-treated groups.

Nuclear, cytoplasmic, and mitochondrial protein extraction

Nuclear and cytoplasmic proteins were extracted using the nuclear and cytoplasmic protein extraction kit (Beyotime, China). Briefly, DPCs were collected and washed twice with PBS. An appropriate volume of cytoplasmic protein extraction reagent A (containing PMSF) was then added. After thorough mixing, the mixture was incubated on ice for 10 min. Subsequently, cytoplasmic protein extraction reagent B was added, and the mixture was gently inverted to mix. Following a 5-minute ice bath, the sample was centrifuged at 4 °C and 12,000 × g for 5 min. The supernatant was transferred to a new centrifuge tube, which constituted the cytoplasmic protein extract. For nuclear protein extraction, the supernatant was discarded after centrifugation, and the pellet was collected. An appropriate volume of nuclear protein extraction reagent (containing PMSF) was added, and the mixture was vortexed until the pellet was completely dispersed. The sample was then incubated on ice for 30 min, with vortexing every 5 min during this period. Afterwards, it was centrifuged at 4 °C and 12,000 × g for 10 min, and the resulting supernatant was the nuclear protein extract.

Mitochondrial proteins were extracted using the mitochondria isolation kit (MedChemExpress, USA). According to the kit instructions, cells were first digested with trypsin and washed three times with PBS. Subsequently, mitochondria isolation reagent was added to the cells, followed by a 15-minute ice bath. The cell suspension was centrifuged at 600 × g for 10 min at 4 °C to remove intact cells and cell debris, and the supernatant (containing mitochondria) was transferred to a new centrifuge tube. Mitochondria were further enriched by centrifugation at 11,000 × g for 10 min at 4 °C, and the resulting pellet was the isolated mitochondria. The mitochondrial pellet was then resuspended in mitochondrial lysis buffer (containing PMSF) and incubated on ice for 20 min to ensure complete lysis. After lysis, the sample was centrifuged at 4 °C and 12,000 × g for 15 min, and the supernatant was collected as the mitochondrial protein extract.

The protein concentration of all samples was determined using the BCA protein quantification kit (Beyotime, China), and the protein concentrations were standardized. Loading buffer was added to each protein sample, which was then denatured by heating at 95 °C for 10 min. The denatured proteins were stored at -20 °C until used for WB detection.

Lentiviral infection of DPCs

ACOD1 knockdown lentiviruses (shACOD1-1: 5’-TGCTTCCAACTGACTACATTA-3’, shACOD1-2: 5’-GGCTATTCACTCCATGGATTT-3’, shACOD1-3: 5’-ACACAGTGGAAAGCCTTATAA-3’), ACOD1 overexpression lentivirus, and the negative control lentivirus were constructed by GenePharma (Shanghai, China). DPCs were seeded in 24-well plates at a density of 2 × 10⁴ cells per well. When the cell confluence reached 70%-80%, the lentiviral suspension was added at a multiplicity of infection (MOI) of 100, along with Polybrene at a final concentration of 5 µg/mL to enhance infection efficiency. After incubation in a humidified 37 °C, 5% CO₂ incubator for 12 h, the medium was replaced with fresh complete medium. At 48 h post-infection, the expression of green fluorescent protein (GFP) was observed under a fluorescence microscope to evaluate infection efficiency. When the infection efficiency exceeded 80%, puromycin selection was performed. For selection, puromycin was added to the culture medium at a final concentration of 2 µg/mL (consistent with the pre-determined MLC). The selection was maintained for 7 days, with the puromycin-containing medium replaced every 2–3 days, to obtain stably infected DPC cell lines for subsequent experiments.

Fluorescence colocalization analysis

DPCs were seeded in confocal dishes (NEST, China) for immunofluorescence colocalization analysis. When the cell density reached 40%, the cells were washed twice with PBS and fixed with 4% paraformaldehyde at room temperature for 20 min. After removing the fixative solution, the cells were washed three times with PBS (5 min per wash). Subsequently, 0.5% Triton X-100 was added for permeabilization at room temperature for 20 min. Following three additional washes with PBS, 5% bovine serum albumin (BSA) blocking solution was added dropwise, and the cells were blocked at room temperature for 1 h. The blocking solution was discarded, and primary antibodies (diluted 1:200 in 5% BSA) were added, followed by incubation at 4 °C overnight. The next day, the cells were washed three times with PBS (5 min per wash), and corresponding fluorescent secondary antibodies were added. The cells were then incubated at room temperature in the dark for 1 h. After incubation, the cells were washed three times with PBS. Finally, DAPI staining solution was added to stain the cell nuclei at room temperature in the dark for 10 min. After a final wash with PBS, the dishes were mounted with anti-fade mounting medium. Images were observed and captured using a Stellaris STED confocal laser microscope (Leica, Germany). The signal intensity distribution along specific lines were measured using Leica LAS X software to assess colocalization.

Co-immunoprecipitation (Co-IP)

The Co-IP experiment was performed according to the instructions of the Co-immunoprecipitation kit (Beyotime, China). Briefly, an appropriate amount of Protein A + G magnetic beads was mixed with antibodies and incubated on a rotating mixer at room temperature for 1 h to allow the antibodies to bind to the Protein A + G magnetic beads. The beads were then washed three times with Tris-buffered saline (TBS). The washed antibody-bound beads were incubated with the protein samples on a rotator at 4 °C overnight. After incubation, the beads were separated and the supernatant was removed. The beads were washed three times with a mild wash buffer. After washing, SDS-PAGE Sample Loading Buffer (1X) was added, and the mixture was heated at 95 °C for 5 min. Subsequently, the beads were separated, and the supernatant was collected for SDS-PAGE electrophoresis or WB detection of the target protein and its interacting proteins.

Coomassie brilliant blue staining

Coomassie brilliant blue staining of the gel strips was performed using BeyoBlue™ plus coomassie blue superfast staining solution (Beyotime, China). Following electrophoretic separation, the protein gel strips were immersed in the staining solution and incubated with gentle shaking at room temperature for 50 min. The strips were then rinsed repeatedly with distilled water or destaining solution until the protein bands were clearly visible and the background was substantially cleared. Subsequently, the stained gel strips were scanned and imaged using the Tanon 5200 gel imaging system, and the images were stored for subsequent protein band analysis.

Immunoprecipitation and mass spectrometry (IP/MS)

To identify proteins interacting with ACOD1, Co-IP combined with mass spectrometry analysis was performed. Briefly, Protein A + G magnetic beads were incubated with ACOD1 antibody to prepare antibody-conjugated beads. The antibody-bound beads were then incubated with protein samples to form antigen-antibody complexes. Normal IgG (incubated with Protein A + G beads under the same conditions) was set as a negative control to exclude non-specific binding. After elution of the immune complexes, the captured proteins were separated by SDS-PAGE. Subsequently, the gel regions corresponding to the target protein bands were excised and subjected to standard in-gel tryptic digestion. The digested peptides were extracted with 50% acetonitrile containing 0.1% formic acid and vacuum-dried prior to MS analysis. MS analysis was performed using an EASY-nLC 1200 nanoflow liquid chromatography system coupled with a Q Exactive HF-X mass spectrometer (Thermo Scientific™, USA). Peptides were separated on a reversed-phase C18 column with a linear gradient of 2% to 100% mobile phase B (80% acetonitrile, 0.1% formic acid) over 80 min. The mass spectrometer was operated in data-dependent acquisition (DDA) mode: a full scan was acquired at a resolution of 60,000, followed by higher-energy collisional dissociation (HCD) of the top 20 most intense precursor ions (normalized collision energy: 27%), with a MS/MS scan resolution of 15,000. Raw MS data were searched against the Swiss-Prot Human database using Proteome Discoverer 2.4 software. Finally, only high-confidence peptide and protein identifications (FDR < 1% for both peptides and proteins) were retained for subsequent analysis.

Plasmid construction and transfection

Human Flag-tagged ACOD1, Flag-tagged ACOD1-F1 (1–240 aa), Flag-tagged ACOD1-F2 (241–481 aa), His-tagged DDX1, His-tagged DDX1-H1 (1–370 aa), and His-tagged DDX1-H2 (371–741 aa) plasmids were constructed by SYNBIO Biotech (Suzhou, China). These recombinant plasmids were transfected into DPCs and HEK 293T cells using Lipofectamine 3000 reagent (Thermo Fisher Scientific, USA) in accordance with the manufacturer’s instructions.

Molecular docking

Molecular docking simulations between human DDX1 and human ACOD1 were performed using the HDOCK server (http://hdock.phys.hust.edu.cn/). The full-length three-dimensional (3D) structure of human DDX1 was constructed by integrating two crystal structures retrieved from the RCSB Protein Data Bank (RCSB PDB; https://www.rcsb.org/): the S86-Q279 residue segment was derived from the crystal structure with PDB ID: 4XW3, and the M1-Q71 and A286-V673 residue segments were obtained from the crystal structure with PDB ID: 8TBX. Missing fragments (Q72-A85, T280-N285, and E674-F740) were modeled using the AlphaFold2-predicted structure, yielding the complete 3D structure of human DDX1. The dimeric crystal structure of human ACOD1 was retrieved from the RCSB PDB (PDB ID: 6R6U). For the docking simulation, human DDX1 was designated as the receptor and human ACOD1 as the ligand. The HDOCK server automatically predicted their binding interactions using a hybrid algorithm that combines template-based modeling and template-free docking. The top-ranked binding conformation from the HDOCK output was selected as the most plausible DDX1-ACOD1 complex structure. The interaction interfaces of the complex were analyzed using the interface analysis module of WeMol3, and the results were visualized with PyMOL (Schrödinger, LLC; Version 2.5.0).

Q300 fully quantitative metabolomics assay

Fully quantitative metabolomics analysis was performed using the Q300 kit (Metabo-Profile, Shanghai, China). Briefly, collected cell samples were homogenized with pre-chilled zirconium beads and deionized water. Subsequently, a methanol solution containing internal standards was added to extract metabolites. After centrifugation, the supernatant was transferred to a 96-well plate. Freshly prepared derivatization reagent was added to the samples using a Biomek 4000 automated workstation (Beckman Coulter, Inc., Brea, CA, USA), followed by derivatization at 30 °C for 60 min. Post-derivatization, the samples were dried and reconstituted with pre-chilled 50% methanol (v/v) solution. The reconstituted mixture was incubated at -20 °C, then centrifuged, and the final supernatant was collected for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. Analyses were conducted on an ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) system (ACQUITY UPLC-Xevo TQ-S, Waters). Chromatographic separation was achieved on an ACQUITY BEH C18 column at 40 °C. The mobile phase consisted of 0.1% (v/v) formic acid in water (phase A) and acetonitrile/isopropanol (70:30, v/v) (phase B), with a gradient elution program at a flow rate of 0.40 mL/min. Mass spectrometric detection was performed using an electrospray ionization (ESI) source, operating in both positive and negative ion modes with capillary voltages set to 1.5 kV and 2.0 kV, respectively. The source temperature was maintained at 150 °C, and the desolvation gas was set to 550 °C with a flow rate of 1000 L/h. Raw data were processed using the iMAP platform (v1.0), and differential metabolites were screened by integrating multivariate statistical analysis methods.

4-octyl itaconate (4-OI) treatment in vitro and in vivo

4-OI (MedChemExpress, USA) was dissolved in PBS containing 10% dimethyl sulfoxide (DMSO) to prepare a 250 mM stock solution. The stock solution was aliquoted into single-use portions and stored at -20 °C to avoid repeated freeze-thaw cycles. Based on previous studies [38, 39] and our preliminary dose-escalation experiments, the working concentration of 4-OI for in vitro cell experiments was set at 100 µM, and cells were treated for 12 h. For in vivo studies, 4-OI was administered to androgenetic alopecia (AGA) model mice via intraperitoneal injection at a dose of 5 mg/kg/d, once daily (n = 5 per group).

Statistical analysis

Experimental data are presented as the mean ± standard deviation (SD). Graphs were generated using GraphPad Prism software (Version 10.1.2). Differences between two independent groups were analyzed by Student’s t-test, while comparisons among three or more groups were performed using one-way analysis of variance (ANOVA) followed by a post hoc test (e.g., Tukey’s honestly significant difference test or Dunnett’s test) for pairwise comparisons. A p value < 0.05 was considered statistically significant in all analyses.

Results

Cellular senescence occurs in DPCs in AGA

To assess cellular senescence in the hair follicles of AGA patients, scalp tissues were collected from both AGA patients and healthy controls. Hair follicles from AGA patients exhibited miniaturization compared with those from healthy controls. Furthermore, IHC revealed a significant increase in p16^INK4a expression in the hair follicles of AGA patients (Fig. 1A). To further evaluate senescence, IF co-staining for p16^INK4a and vimentin showed a higher percentage of p16^INK4a-positive cells in AGA hair follicles than in those from healthy controls (Fig. 1B). Additionally, mRNA expression levels of CDKN1A (encoding p21) and CDKN2A (encoding p16^INK4a) were upregulated in hair follicles of AGA patients compared with those from healthy controls (Additional file 1: Fig. S1A). We next isolated DPCs from the hair follicles. Compared with DPCs from healthy controls, both CDKN2A and CDKN1A mRNA levels were significantly increased in AGA patient-derived DPCs (Fig. 1C). Similarly, the protein levels of senescence-related markers (p16^INK4a, p21, and p53) were also elevated in AGA patient-derived DPCs (Fig. 1D).

Fig. 1.

Cellular senescence occurs in DPCs in AGA. A) Representative H&E and p16^INK4a immunohistochemical staining of scalp tissues from healthy controls and AGA patients. Scale bar = 200 μm. n = 6. B) Representative images of p16^INK4a immunofluorescence staining and quantification of p16^INK4a-positive cells in scalp tissues. Scale bar = 200 μm. n = 6. C) qPCR analysis of CDKN2A and CDKN1A expression in primary DPCs. n = 6. D) Western blot analysis of senescence markers in primary DPCs and corresponding quantification. n = 3. E) Representative images of dorsal skin of mice after depilation. n = 5. F) Quantification of skin color score after depilation. G) Representative H&E staining of mouse dorsal skin tissues. Scale bar = 500 μm. n = 5. H) Quantification of hair follicle numbers per field, hair bulb diameter, skin thickness, and dermal thickness relative to total skin thickness. n = 5. I) Representative p16^INK4a immunohistochemical staining in mouse dorsal skin tissues. Scale bar = 100 μm. n = 5. J) Representative p16^INK4a immunofluorescence staining in mouse dorsal skin tissues. Scale bar = 100 μm. n = 5. p values were determined by two-tailed unpaired Student’s t-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Following the method described in reference [37], we induced an AGA mouse model by intraperitoneal injection of DHT; the specific administration regimen is provided in Additional file 1 Fig. S1B. Photographic comparisons on days 5, 9, and 14 post-depilation revealed a slower transition from pink to black skin in the DHT group, indicating a delayed entry into anagen. By day 21, depilated skin in the DHT group appeared pinker than that of controls, suggesting a premature entry into telogen (Fig. 1E and F). This recapitulates the pathological hallmark of human AGA—shortened anagen and prolonged telogen [11, 37]. H&E staining confirmed miniaturization of hair follicles in the DHT group (Fig. 1G). Quantitative analysis showed reduced hair bulb diameter and overall skin thickness in the DHT group compared with the control group. Although absolute dermal thickness was reduced, its relative proportion within the entire skin layer was increased. No significant difference was found in the number of hair follicles per field between the two groups (Fig. 1H). Collectively, these changes validate the successful establishment of a DHT-induced AGA mouse model in this hair cycle. We next assessed cellular senescence in this model. Both IHC and IF co-staining for p16^INK4a and vimentin revealed elevated p16^INK4a expression in DPCs of the DHT group(Fig. 1I and J).

In summary, these results demonstrate that cellular senescence occurs in DPCs in AGA.

DHT induces cellular senescence and functional decline in DPCs

Given that primary DPCs lose their characteristic properties with serial passaging [22, 23, 40], we immortalized DPCs and treated them with DHT to mimic the in vivo conditions of AGA. After DHT treatment, we observed a significant increase in the percentage of senescent cells, as shown by senescence-associated β-galactosidase staining (Fig. 2A). This finding was corroborated by an increase in p16^INK4a-positive cells detected through IF co-staining for p16^INK4a and vimentin (Fig. 2B). At the molecular level, DHT stimulation significantly upregulated mRNA expression of CDKN2A and CDKN1A in DPCs compared with untreated controls (Fig. 2C). A corresponding increase was also observed at the protein level for senescence markers p16^INK4a, p21, and p53 (Fig. 2D).

Fig. 2.

DHT induces cellular senescence and functional decline in DPCs. A) Representative SA-β-gal staining and quantification of β-gal-positive cells after DHT stimulation. Scale bar = 100 μm. n = 6. B) Representative p16^INK4a immunofluorescence staining and quantification after DHT stimulation. Scale bar = 100 μm. n = 6. C) qPCR analysis of CDKN2A and CDKN1A expression after DHT stimulation. n = 6. D) Western blot analysis of senescence markers and quantification after DHT stimulation. n = 3. E) Flow-cytometric analysis of proliferation and quantification after DHT treatment. n = 6. F) Representative EdU staining and quantification after DHT stimulation. Scale bar = 200 μm. n = 6. G) Representative Transwell images showing DPC migration after DHT stimulation. Scale bar = 100 μm. H) Quantification of Transwell migration. n = 6. I) Representative wound-healing assay images after DHT stimulation. Scale bar = 100 μm. J) Quantification of migration area at 24 h. n = 6. K) Flow-cytometric analysis of apoptosis after DHT stimulation and quantification. n = 6. L) Representative TUNEL staining and quantification after DHT stimulation. Scale bar = 200 μm. n = 6. p values were determined by two-tailed unpaired Student’s t-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

We next investigated the functional impairments resulting from DHT-induced senescence in DPCs. An EdU incorporation assay revealed that DHT treatment significantly impaired DPC proliferation. This impairment was confirmed by both flow cytometry (Fig. 2E) and IF, which showed a significantly lower proportion of EdU-positive cells in DHT-treated DPCs compared with the control group (Fig. 2F). Similarly, DHT treatment markedly inhibited DPC migration. Transwell assays revealed a significant reduction in the number of migrated cells (Fig. 2G and H), and wound-healing assays confirmed delayed wound closure in the DHT group compared with the control group (Fig. 2I and J). Finally, we evaluated apoptosis using a TUNEL assay. DHT stimulation significantly increased the apoptotic rate in DPCs, as quantified by flow cytometry (Fig. 2K) and visualized as an increase in TUNEL-positive cells via IF (Fig. 2L).

In summary, these results demonstrate that DHT stimulation induces cellular senescence in DPCs, which contributes to the observed increase in apoptosis and impairment of proliferation and migration.

DHT induces mitochondrial dysfunction in DPCs

To investigate the mechanism underlying DPC senescence in AGA, we performed RNA sequencing (RNA-Seq) analysis on DHT-stimulated and control DPCs. Compared with the control group, we identified differentially expressed mRNAs in the DHT-treated group (Fig. 3A). Gene Ontology (GO) enrichment analysis revealed that these changes were enriched in terms related to senescence and mitochondria (Fig. 3B). Gene Set Enrichment Analysis (GSEA) indicated impaired OXPHOS in DHT-stimulated DPCs, suggesting mitochondrial dysfunction (Fig. 3C). Since mitochondrial dysfunction is a major contributor to cellular senescence, we further assessed mitochondrial function in DHT-stimulated DPCs. MitoTracker staining and TEM showed that, compared with the control group, DHT-treated cells exhibited morphological alterations, including increased mitochondrial fragmentation and reduced mitochondrial cristae (Fig. 3D and E). Additionally, we observed increased mitochondrial superoxide levels and a decrease in mitochondrial membrane potential in DHT-stimulated DPCs (Fig. 3F and G).

Fig. 3.

DHT induces mitochondrial dysfunction in DPCs. A) Heatmap of RNA-sequencing analysis of DPCs with or without DHT stimulation. B) GO enrichment analysis of differentially expressed mRNAs. C) GSEA analysis of the “OXPHOS” gene set in control versus DHT-treated DPCs. D) Representative Mitotracker staining showing mitochondrial morphology and quantification of mean mitochondrial branch length. Scale bar = 10 μm. n = 6. E) Representative TEM images (mitochondria indicated by red arrows) and quantification of mitochondrial cristae number. Scale bar = 500 nm. n = 6. F) Representative MitoSOX staining of mitochondrial superoxide and quantification of fluorescence intensity. Scale bar = 50 μm. n = 6. G) JC-1 assay of mitochondrial membrane potential and JC-1 aggregate/monomer ratios. n = 6. H–I) Oxygen consumption rate (OCR) and quantification of mitochondrial respiration. n = 5. J–K) Proton efflux rate (PER) and quantification of basal and compensatory glycolysis. n = 5. L) Quantification of intracellular ATP levels. n = 6. p values were determined by two-tailed unpaired Student’s t-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Mitochondria produce ATP primarily through OXPHOS to supply energy for cellular functions. In contrast, glycolysis is a less efficient ATP-producing pathway that occurs in the cytoplasm. The balance between these two processes is crucial for maintaining cellular energy homeostasis. We used Seahorse assays to assess mitochondrial function in DHT-stimulated DPCs. Measurements of the oxygen consumption rate (OCR) and proton efflux rate (PER) indicated impaired mitochondrial respiration and a compensatory increase in glycolysis upon DHT stimulation (Fig. 3H–K). Consistent with this metabolic shift, ATP assays confirmed an overall reduction in total cellular ATP production (Fig. 3L).

Collectively, these results demonstrate that DHT induces significant mitochondrial dysfunction in DPCs, suggesting it as a potential mechanism driving cellular senescence.

ACOD1 deficiency contributes to mitochondrial dysfunction and promotes cellular senescence in DPCs

We further analyzed the RNA-Seq data to identify potential mechanisms underlying mitochondrial dysfunction and cellular senescence in DPCs after DHT stimulation. The analysis revealed a significant decrease in ACOD1 expression in DHT-stimulated DPCs (Fig. 4A). ACOD1 catalyzes the production of itaconate, an important metabolite in the TCA cycle [29], and plays a critical role in mitochondrial energy metabolism, reactive oxygen species (ROS) production, and overall cellular function [28, 30]. To validate the sequencing results, we performed PCR and WB analysis to verify ACOD1 expression levels. The results showed that both ACOD1 mRNA and protein levels were significantly reduced in DHT-stimulated DPCs compared with the control group (Fig. 4B and D). IF staining of scalp tissues from patients with AGA also confirmed decreased ACOD1 expression compared with healthy controls (Fig. 4C). ChIP-qPCR results demonstrated a significant increase in AR binding to the ACOD1 promoter region following DHT stimulation (Fig. 4E), providing a preliminary insight into the mechanisms underlying downregulation of ACOD1 expression in DPCs after DHT treatment.

Fig. 4.

ACOD1 deficiency contributes to mitochondrial dysfunction and promotes cellular senescence in DPCs. A) Volcano plot of RNA-sequencing analysis in DPCs. B) qPCR analysis of ACOD1 expression in DPCs. n = 6. C) Representative ACOD1 immunofluorescence staining in scalp tissues from healthy controls and AGA patients. Scale bar = 100 μm. n = 6. D) Western blot analysis of ACOD1 protein levels in DPCs. n = 3. E) ChIP-qPCR analysis of AR recruitment to the ACOD1 promoter in DPCs. n = 3. F) Western blot analysis of intracellular ACOD1 subcellular localization. n = 3. G) Representative images of ACOD1 colocalization with the mitochondrial marker GRP75 and fluorescence intensity profiles along the indicated line. Scale bar = 5 μm. n = 3. H) Western blot analysis of ACOD1 knockdown efficiency using lentiviral shACOD1. n = 3. I) Representative MitoSOX staining of control and shACOD1 DPCs and quantification of mean fluorescence intensity. Scale bar = 50 μm. n = 6. J) Representative TEM images (mitochondria indicated by red arrows) and quantification of average mitochondrial cristae number in DPCs. Scale bar = 500 nm. n = 6. K–L. OCR and quantification of mitochondrial respiration in DPCs. n = 4. M–N. PER and quantification of basal and compensatory glycolysis in DPCs. n = 5. O) Quantification of intracellular ATP levels in DPCs. n = 6. P) Western blot analysis of senescence markers in DPCs. n = 3. Q) Representative SA-β-gal staining and quantification of β-gal-positive cells in DPCs. Scale bar = 100 μm. n = 6. R) Representative p16^INK4a immunofluorescence staining in DPCs. Scale bar = 100 μm. S) Quantification of p16^INK4a-positive cells. n = 6. p values were determined by two-tailed unpaired Student’s t-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Subcellular fractionation followed by WB confirmed the primary mitochondrial localization of ACOD1 (Fig. 4F). This finding was supported by IF co-localization analysis, which showed that ACOD1 co-localized with the mitochondrial marker GRP75 (Fig. 4G). These results indicate that ACOD1 is a mitochondrial protein, suggesting that its downregulation may be linked to mitochondrial dysfunction and senescence in DPCs.

To investigate the functional role of ACOD1 in DPCs, we knocked down its expression using lentiviral infection (Fig. 4H). IF analysis revealed a significant increase in mitochondrial superoxide in ACOD1-knockdown DPCs (Fig. 4I). TEM further demonstrated aberrant mitochondrial morphology, including increased fragmentation and reduced cristae (Fig. 4J). Seahorse assays indicated that ACOD1 knockdown impaired mitochondrial function, as evidenced by decreased OXPHOS, reduced ATP production, and enhanced glycolysis (Fig. 4K–O). In addition to mitochondrial dysfunction, ACOD1 knockdown induced a senescent phenotype, characterized by elevated protein levels of p16^INK4a, p21, and p53 (Fig. 4P), an increased percentage of SA-β-galactosidase-positive cells (Fig. 4Q), and a higher proportion of p16^INK4a-positive cells in co-staining with vimentin (Fig. 4R, S). Subsequently, we assessed the functional consequences of ACOD1 knockdown. The proliferation capacity of shACOD1 DPCs was significantly impaired (Additional file 1: Fig. S2A and B), alongside a marked reduction in migration capacity (Additional file 1: Fig. S2C–F). Conversely, the apoptosis rate was significantly increased (Additional file 1: Fig. S2G and H). Together, these results demonstrate that ACOD1 deficiency induces mitochondrial dysfunction, promotes cellular senescence, and consequently leads to diminished proliferation and migration capabilities, as well as increased apoptosis in DPCs.

To further validate the protective role of ACOD1 in DPCs, we overexpressed ACOD1 using a lentiviral vector (Fig. 5A). Using MitoSOX Red staining, we found that ACOD1 overexpression significantly attenuated the DHT-induced increase in mitochondrial superoxide (Fig. 5B and C). TEM demonstrated that ACOD1 overexpression mitigated the DHT-induced mitochondrial morphological damage (Fig. 5D and E). Seahorse assays indicated that ACOD1 overexpression ameliorated the DHT-induced mitochondrial dysfunction, restoring OXPHOS and ATP production (Fig. 5F–J). Furthermore, ACOD1 overexpression markedly suppressed the DHT-induced senescent phenotype. This was evidenced by reduced protein levels of p16^INK4a, p21, and p53 (Fig. 5K), a lower percentage of SA-β-galactosidase-positive cells (Fig. 5L), and a decreased proportion of p16^INK4a-positive cells in IF staining (Fig. 5M). Additionally, ACOD1 overexpression ameliorated DHT-induced functional impairments, leading to improved proliferation (Additional file 1: Fig. S3A and B), enhanced migration capacity (Additional file 1: Fig. S3C–F), and reduced apoptosis (Additional file 1: Fig. S3G and H).

Fig. 5.

ACOD1 overexpression protects against mitochondrial dysfunction and senescence in DPCs. A) Western blot validation of ACOD1 overexpression. n = 3. B) Representative MitoSOX staining in control and oeACOD1 DPCs. Scale bar = 50 μm. C) Quantification of mean MitoSOX fluorescence intensity. n = 6. D) Representative TEM images of mitochondria (red arrows) in DPCs. Scale bar = 500 nm. E) Quantification of average mitochondrial cristae number. n = 6. F–G) OCR and quantification of mitochondrial respiration in DPCs. n = 3. H–I) PER and quantification of basal and compensatory glycolysis in DPCs. n = 3. J) Quantification of intracellular ATP levels. n = 6. K) Western blot analysis of senescence markers in DPCs. n = 3. L) Representative SA-β-gal staining and quantification of β-gal-positive cells. Scale bar = 100 μm. n = 6. M) Representative p16^INK4a immunofluorescence staining and quantification of p16^INK4a-positive cells. Scale bar = 100 μm. n = 6. p values were determined by two-tailed unpaired Student’s t-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

In summary, ACOD1 deficiency leads to mitochondrial dysfunction, cellular senescence, and functional decline in DPCs, whereas ACOD1 overexpression can counteract the detrimental effects induced by DHT.

ACOD1 interacts with DDX1

To further investigate the specific mechanism by which ACOD1 deficiency contributes to mitochondrial dysfunction and promotes cellular senescence in DPCs, we employed Co-IP combined with proteomic analysis to identify proteins that interact with ACOD1. IP/MS revealed a potential interaction between ACOD1 and DDX1 (Fig. 6A and B). DDX1 is a member of the DEAD-box (DDX) family of RNA helicases and is involved in RNA metabolic processes [41]. In addition to its role in RNA metabolism, previous studies have shown that DDX1 can modulate mitochondrial function by regulating Ca²⁺ distribution [42, 43]. To examine whether the ACOD1-DDX1 interaction influences mitochondrial function, we first sought to confirm the physical association and determine the binding mode. Immunofluorescence confocal microscopy demonstrated co-localization of ACOD1 and DDX1 in DPCs (Fig. 6C and D), indicating that both proteins are present in mitochondria. Endogenous Co-IP experiments confirmed the interaction between ACOD1 and DDX1 (Fig. 6E and F). Furthermore, when exogenous Flag-ACOD1 and His-DDX1 were transfected into HEK 293T cells, Co-IP assays also validated their interaction (Fig. 6G). We then used molecular docking to predict the binding sites between ACOD1 and DDX1 (Fig. 6H and I). To further characterize the specific domains mediating the interaction, we constructed truncated plasmids (Fig. 6J). After transfecting these plasmids into HEK 293T cells and performing Co-IP, we found that the F1 domain (residues 1-240) of ACOD1 binds to the H1 domain (residues 1-370) of DDX1 (Fig. 6K and L). Overall, these results demonstrate that ACOD1 interacts with DDX1, suggesting that ACOD1 deficiency may impair mitochondrial function and promote cellular senescence by affecting DDX1.

Fig. 6.

ACOD1 interacts with DDX1. A) Coomassie brilliant blue staining of ACOD1 immunoprecipitated protein bands. B) IP/MS analysis showing DDX1 as an interacting partner of ACOD1. C) Representative immunofluorescence colocalization images of ACOD1 and DDX1. Scale bar = 5 μm. n = 3. D) Fluorescence intensity profiles of ACOD1 and DDX1 along the indicated line. E–F) Endogenous Co-IP validation of ACOD1–DDX1 interaction in DPCs. n = 3. G) Exogenous Co-IP validation of Flag-ACOD1 and His-DDX1 interaction in HEK 293T cells. n = 3. H–I) Molecular docking analysis showing binding interfaces between ACOD1 and DDX1. J) Construction of ACOD1 and DDX1 truncation plasmids. K–L) Co-IP mapping of interaction regions between Flag-ACOD1 and His-DDX1 in HEK 293T cells.

ACOD1 deficiency may promote mitochondrial dysfunction and induce cellular senescence by enhancing DDX1 methylation

Having confirmed the interaction between ACOD1 and DDX1, we further investigated whether ACOD1 deficiency affects mitochondrial function and cellular senescence by influencing DDX1. It has been reported that the methyltransferase EZH2 catalyzes DDX1 methylation, thereby promoting nucleus pulposus cell senescence and apoptosis and contributing to intervertebral disc degeneration [44]. Therefore, we asked whether ACOD1 deficiency leads to mitochondrial dysfunction and cellular senescence in DPCs by regulating DDX1 methylation. WB analysis showed that protein methylation levels were increased upon ACOD1 knockdown compared with the control group (Fig. 7A). Endogenous and exogenous Co-IP experiments further confirmed elevated DDX1 methylation after ACOD1 knockdown (Fig. 7B and C). Transfection of the ACOD1 F1 domain fragment into cells significantly reduced DDX1 methylation (Fig. 7D). This result not only verified the binding between the ACOD1 F1 domain and DDX1 but also indicated that ACOD1 deficiency enhances DDX1 methylation. In ACOD1-knockdown cells, the methyltransferase inhibitor MS023 attenuated the effects of ACOD1 knockdown and suppressed DDX1 methylation, whereas the methyl donor S-adenosylmethionine (SAM) increased DDX1 methylation (Fig. 7E). Taken together, these findings reveal that ACOD1 deficiency is associated with increased DDX1 methylation, which may represent one of the potential mechanisms underlying the interaction between ACOD1 and DDX1.

Fig. 7.

ACOD1 deficiency promotes mitochondrial dysfunction and induces cellular senescence by enhancing DDX1 methylation. A) Western blot analysis of global protein methylation levels in DPCs. n = 3. B) Co-IP analysis of DDX1 methylation in shNC and shACOD1 DPCs. n = 3. C) Co-IP analysis of His-DDX1 methylation in shNC and shACOD1 HEK 293T cells. n = 3. D) Co-IP analysis of His-DDX1 methylation after transfection of Flag-ACOD1-FL and Flag-ACOD1-F1 into HEK 293T cells. n = 3. E) Co-IP analysis of His-DDX1 methylation in different treatment groups in HEK 293T cells. n = 3. F–G) OCR and quantification of mitochondrial respiration in differently treated DPCs. n = 3. H–I) PER and quantification of basal and compensatory glycolysis. n = 3. J) Quantification of intracellular ATP levels. n = 6. K) Western blot analysis of senescence markers in differently treated DPCs. n = 3. L) Representative SA-β-gal staining in differently treated DPCs. Scale bar = 100 μm. M) Quantification of β-gal-positive cells. n = 6. N) Representative p16^INK4a immunofluorescence staining in differently treated DPCs. Scale bar = 100 μm. O) Quantification of p16^INK4a-positive cells. n = 6. p values were determined by one-way ANOVA with Tukey’s multiple-comparisons. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Seahorse analysis revealed that the methyltransferase inhibitor MS023 significantly attenuated the mitochondrial dysfunction induced by ACOD1 knockdown, which was characterized by impaired OXPHOS, reduced ATP production, and enhanced glycolysis. In contrast, the methyl donor SAM exacerbated these impairments compared with the DMSO control (Fig. 7F–J). Consistently, MS023 treatment also counteracted the upregulation of senescence-associated proteins p16^INK4a, p21, and p53 following ACOD1 knockdown (Fig. 7K). Similarly, MS023 alleviated the cellular senescence phenotype caused by ACOD1 knockdown, as evidenced by a significant reduction in the percentage of SA-β-galactosidase-positive and p16^INK4a-positive cells. Conversely, SAM further increased the proportions of these senescence markers relative to the DMSO control (Fig. 7L–O).

Together, these results suggest that elevated DDX1 methylation is associated with mitochondrial dysfunction and cellular senescence in DPCs, and that ACOD1 deficiency may partially contribute to these processes by affecting DDX1 methylation. Although ACOD1 may provide partial protection by suppressing DDX1 methylation, its functional specificity requires further validation.

Itaconate may reduce DDX1 methylation to preserve mitochondrial function and mitigate cellular senescence and AGA progression

Furthermore, we performed metabolomic analysis on DPCs following ACOD1 knockdown. The results showed that this knockdown induced marked and highly consistent metabolic perturbations in DPCs (Fig. 8A). Analysis of specific metabolites indicated that ACOD1 knockdown significantly disrupted organic acid metabolism (Fig. 8B), suggesting profound changes in energy metabolism or impairment of specific metabolic pathways. Consistent with this, heatmap analysis confirmed that ACOD1 knockdown triggered metabolic reprogramming in DPCs (Fig. 8C). Specifically, ACOD1 knockdown significantly decreased the levels of aspartate, glutamine, and alanine, while increasing those of citrate, malate, fumarate, α-ketoglutarate, lactate, and succinate (Fig. 8D–L). These findings further confirm that the TCA cycle was significantly impaired, whereas glycolysis was upregulated, in ACOD1-knockdown DPCs, which is consistent with our previous observations. As expected, since ACOD1 catalyzes the production of itaconate—a key regulator of the TCA cycle—metabolomic profiling confirmed a significant decrease in itaconate levels in ACOD1-knockdown DPCs (Fig. 8M).

Fig. 8.

ACOD1 deficiency alters DPC metabolism. A) PCA score plot from Q300 metabolomics analysis. B) Bar plot of differential metabolites in shACOD1 DPCs. C) Heatmap of metabolic reprogramming induced by ACOD1 knockdown. D–M) Quantification of intracellular metabolite levels in control and shACOD1 DPCs, including aspartate, glutamine, citrate, malate, fumarate, α-ketoglutarate, lactate, succinate, alanine, and the ACOD1-derived metabolite itaconate. n = 6. p values were determined by two-tailed unpaired Student’s t-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Next, we further examined whether exogenous itaconate supplementation could mitigate the effects induced by DHT. To this end, we treated DPCs with 4-OI, a cell-permeable itaconate derivative [45]. Compared with the vehicle control group, 4-OI attenuated the DHT-induced upregulation of DDX1 methylation in DPCs (Fig. 9A and B). Seahorse analysis also indicated that 4-OI was associated with improved mitochondrial function in DHT-stimulated DPCs, as evidenced by enhanced OXPHOS, increased ATP production, and reduced glycolysis (Fig. 9C–G). Moreover, 4-OI treatment was associated with decreased protein levels of p16^INK4a, p21, and p53 in DHT-stimulated DPCs (Fig. 9H). Consistent with this, it was also associated with reduced percentages of SA-β-galactosidase-positive and p16^INK4a-positive cells in DHT-stimulated DPCs (Fig. 9I–K). In addition, 4-OI significantly ameliorated the DHT-induced suppression of DPC proliferation (Additional file 1: Fig. S4A and B) and migration (Additional file 1: Fig. S4C–F), while reducing apoptosis (Additional file 1: Fig. S4G and H). Collectively, these findings suggest that exogenous itaconate supplementation partially reduces DDX1 methylation, which is associated with improvements in mitochondrial function, indicators of cellular senescence, and proliferation and migratory capacity, as well as a reduction in apoptosis, in DHT-stimulated DPCs.

Fig. 9.

Itaconate reduces DDX1 methylation to preserve mitochondrial function and mitigate cellular senescence and AGA progression. A) Western blot analysis of protein methylation levels in DPCs after itaconate treatment. n = 3. B) Co-IP analysis of DDX1 methylation after itaconate treatment in DPCs. n = 3. C–D) OCR and quantification of mitochondrial respiration after itaconate treatment. n = 3. E–F) PER and quantification of basal and compensatory glycolysis. n = 3. G) Quantification of intracellular ATP levels. n = 6. H) Western blot analysis of senescence markers after itaconate treatment. n = 3. I) Representative SA-β-gal staining and quantification of β-gal-positive cells. Scale bar = 100 μm. n = 6. J) Representative p16^INK4a immunofluorescence staining after itaconate treatment. Scale bar = 100 μm. K) Quantification of p16^INK4a-positive cells. n = 6. L) Representative images of dorsal skin in AGA mice after itaconate administration. n = 5. M) Quantification of skin color score in itaconate-treated AGA mice after depilation. N) Representative H&E staining of mouse skin after itaconate treatment. Scale bar = 500 μm. n = 5. O) Quantification of hair follicle numbers per field, hair bulb diameter, skin thickness, and dermal thickness relative to total skin thickness. n = 5. P) Representative p16^INK4a immunohistochemical staining in mouse dorsal skin tissues. Scale bar = 100 μm. n = 5. Q) Representative p16^INK4a immunofluorescence staining in mouse dorsal skin tissues. Scale bar = 100 μm. n = 5. p values were determined by one-way ANOVA with Tukey’s multiple-comparisons. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Furthermore, we assessed the efficacy of exogenous itaconate in a DHT-induced AGA mouse model via 4-OI supplementation. Photographic analysis on days 5, 9, and 14 post-depilation revealed that, compared with mice treated with DHT alone, those co-treated with DHT and 4-OI exhibited accelerated dorsal skin darkening from pink, indicative of an earlier transition to the anagen phase. By day 21, the dorsal skin of the 4-OI-treated group remained darker, suggesting a prolonged anagen phase and delayed entry into telogen, thus mitigating AGA progression (Fig. 9L and M). H&E staining demonstrated that 4-OI treatment mitigated DHT-induced hair follicle miniaturization (Fig. 9N). Quantitative analysis revealed increased hair bulb diameter and total skin thickness, a reduced relative proportion of dermal thickness within the total skin layer, and no significant difference in the number of hair follicles per field among the groups (Fig. 9O). Collectively, these morphological alterations demonstrate delayed AGA progression in 4-OI-treated mice during the hair cycle. Accordingly, both IHC for p16^INK4a and IF co-staining for p16^INK4a and vimentin demonstrated reduced p16^INK4a expression in DPCs of the DHT + 4-OI group compared with the DHT-alone group (Fig. 9P and Q). In summary, these in vivo findings indicate that itaconate supplementation may alleviate DHT-induced DPC senescence and AGA progression.

Discussion

According to classical theory, the pathogenesis of AGA is primarily attributed to DHT, the active metabolite of testosterone [9]. Upon binding to AR in DPCs of the hair follicle, DHT induces hair follicle miniaturization by regulating downstream target genes, ultimately leading to hair thinning, softening, and loss [9, 11, 17]. However, in recent years, there has been growing recognition that other mechanisms also contribute to AGA pathogenesis, including hair follicle microinflammation [46, 47], perifollicular fibrosis [46, 48], microcirculatory dysfunction [35, 49], and oxidative stress [50, 51]. Simultaneously, as cellular senescence has been increasingly implicated in various age-related diseases, its role in AGA pathogenesis has garnered growing interest, although the underlying mechanisms remain to be fully elucidated. In this study, by integrating evidence from clinical tissues, animal models, and cellular experiments, we demonstrated that DPCs in the context of AGA exhibit a typical senescent phenotype. This phenotype was characterized by elevated protein levels of p16^INK4a, p21, and p53, an increased percentage of SA-β-gal-positive cells, impaired proliferation and migration, and enhanced apoptosis. Our findings identify DPC senescence as a key driver of AGA pathogenesis and further support targeting this process as a promising therapeutic strategy for AGA [52].

Compared with healthy controls, primary DPCs isolated from the balding scalps of AGA patients exhibited increased mRNA levels of senescence markers CDKN2A and CDKN1A, as well as elevated protein levels of p16^INK4a, p21, and p53. This observation is consistent with previous reports documenting differences in senescence marker expression between primary DPCs from balding and non-balding scalps of AGA patients [22]. However, due to ethical constraints and patients’ aesthetic concerns, it is challenging to recruit a large cohort of AGA patients to donate hair follicles from both balding and non-balding regions simultaneously, limiting the acquisition of primary DPCs for extensive experimental studies. Additionally, prior literature has shown that during in vitro passage, primary DPCs lose their intrinsic hair-inductive capacity and concomitantly accumulate senescent cells [22, 23, 40]. This phenomenon may be attributed to the 2D culture conditions for human primary DPCs, which disrupt mechanical signaling and alter cellular phenotype [23, 40]. In contrast, 2D-cultured immortalized DPCs derived from both balding and non-balding regions retain hair follicle-inductive properties, constituting them a valuable model for investigating AGA pathogenesis and hair biology [23]. Therefore, we generated immortalized DPCs for subsequent mechanistic analyses. These immortalized DPCs maintained expression of AR as well as profiles of dermal papilla-specific markers. The use of immortalized DPCs not only circumvents the limitation of sample scarcity but also eliminates experimental variability arising from inter-individual differences associated with repeated primary DPC isolation. Such variability would otherwise lead to inconsistent responses to experimental manipulations. Furthermore, the induction of senescent phenotypes in immortalized DPCs upon DHT stimulation more directly demonstrates that this senescence is not caused by replicative exhaustion but is a DHT-specific effect, thereby providing further evidence for the role of cellular senescence in AGA progression.

Mitochondria, as the “energy factories” of cells, are closely associated with cellular senescence [18, 53]. Mitochondria are key regulators of signaling pathways that drive hair follicle development [54]. However, most current studies on the impact of mitochondria on hair follicles focus on signaling pathways related to hair follicle morphogenesis, with limited research on the specific mechanisms underlying mitochondrial dysfunction in AGA [54]. In this study, RNA-Seq and bioinformatics analysis revealed that in DHT-stimulated DPCs, differentially expressed genes were enriched in pathways related to cellular senescence and mitochondrial function. GSEA further confirmed a significant impairment of OXPHOS, suggesting that mitochondrial dysfunction may be a key mechanism in DHT-induced DPC senescence. Subsequent morphological and functional validations showed that after DHT stimulation, DPC mitochondria exhibited abnormalities such as increased mitochondrial fragmentation, reduced cristae structure, mitochondrial superoxide accumulation, and decreased membrane potential—ultimately leading to reduced ATP production. To sustain energy supply, cells were forced to enhance the glycolytic pathway, resulting in metabolic reprogramming characterized by “impaired OXPHOS and a compensatory enhancement of glycolysis”. This inefficient energy production mode further exacerbates cellular functional exhaustion and accelerates the senescence process. This finding aligns with recent studies indicating that, compared to dermal papilla aggregates from non-balding regions, those from balding regions exhibit altered mitochondrial metabolism—specifically characterized by upregulation of mitochondria-related gene expression, impaired mitochondrial function, and elevated oxidative stress levels [51]. Impaired OXPHOS and ROS accumulation are core features of mitochondrial dysfunction [53]. Furthermore, previous studies have shown that impaired mitochondrial OXPHOS is a common characteristic of senescent cells [53]. Therefore, our study demonstrates that DHT-induced mitochondrial dysfunction in DPCs is a crucial intrinsic mechanism promoting DPC senescence and AGA progression, a finding consistent with recent research [25, 26, 55].

To elucidate the intrinsic mechanism by which DHT leads to mitochondrial dysfunction in DPCs, we utilized RNA-Seq data and identified ACOD1 as a key regulator regulating mitochondrial function and the senescence of DPCs. We further demonstrated that ACOD1 expression is downregulated in AGA patients and DHT-treated DPCs. Subcellular localization analysis confirmed its primary mitochondrial localization, suggesting that ACOD1 may be involved in DPC senescence by regulating mitochondrial function. Functional experiments confirmed that ACOD1 knockdown recapitulated the effects of DHT by inducing mitochondrial dysfunction, upregulating senescence-related markers, and subsequently inhibiting cell proliferation and migration while promoting apoptosis. Conversely, ACOD1 overexpression alleviated the aforementioned phenotypes induced by DHT. ACOD1, also known as IRG1, is a mitochondrial-localized enzyme that primarily regulates immunometabolism during inflammation and infection through its role in itaconate metabolism, oxidative stress, and antigen processing [56]. In activated immune cells (for example, macrophages and monocytes), ACOD1 activation limits pathogen infection [57, 58]. Furthermore, ACOD1 upregulation promotes embryo implantation [59]. However, abnormally high ACOD1 expression can also lead to immune paralysis [58, 60], tumor progression [61], and neurodegeneration [62], indicating that ACOD1 plays a dual role in immunity and disease [56]. In macrophages, ACOD1 promotes fatty acid utilization for OXPHOS and increases mtROS production, thereby enhancing antibacterial activity [57].

ACOD1 serves not only as a key metabolic regulator in immune cells, but also as a critical hub connecting metabolic status with cellular function in various mesenchymal cells through multiple mechanisms, including metabolic reprogramming, redox balance maintenance, inflammatory regulation, cell differentiation control, and tissue homeostasis preservation. ACOD1 and its metabolic product itaconate function as a critical metabolic checkpoint during bone remodeling, exhibiting dual roles characterized by both anti-resorptive and osteoanabolic potential [63–65]. However, in endometriosis, the aberrant overexpression of ACOD1 and excessive secretion of itaconate in ectopic stromal cells promote disease progression through paracrine pathways [66]. Furthermore, the ACOD1/itaconate axis has been identified as an endogenous protective mechanism in pulmonary fibrosis, exerting anti-fibrotic effects by regulating fibroblast-myofibroblast differentiation [67]. This axis also demonstrates direct cytoprotective effects against obesity-induced pulmonary microvascular endotheliopathy, contributing to the maintenance of vascular homeostasis [68]. In our study, ACOD1 overexpression alleviated DHT-induced mitochondrial dysfunction and the upregulation of senescence-related markers in DPCs. Mechanistically, our study provides the first evidence linking ACOD1 to DPC senescence. This finding elucidates a novel function of ACOD1 in non-immune cells and expands the known regulatory mechanisms within hair follicle biology.

Beyond this, we have revealed a novel non-canonical pathway through which ACOD1 exerts its molecular function: through its direct interaction with the RNA helicase DDX1. DDX1 is a member of the DEAD-box (DDX) family of RNA helicases, defined by the conserved D-E-A-D (Asp-Glu-Ala-Asp) motif. As an RNA-binding protein, it is involved in key RNA metabolic processes, including pre-mRNA splicing, mRNA transport, RNA editing, RNA decay, and ribosome biogenesis [41]. In recent years, DDX1 has been reported to serve as a substrate for EZH2-mediated methylation. Increased methylation of DDX1 can induce nucleus pulposus cell senescence and apoptosis, ultimately contributing to intervertebral disc degeneration [44]. We demonstrated that ACOD1 binds to the H1 domain of DDX1 through its own F1 domain, thereby inhibiting DDX1 methylation. In the context of ACOD1 deficiency, elevated DDX1 methylation levels are closely associated with mitochondrial dysfunction and the senescent phenotype of DPCs—consistent with previous findings that increased DDX1 methylation induces cellular senescence and apoptosis [44]. Treatment with the global methylation inhibitor MS023 alleviated the adverse effects caused by ACOD1 deficiency, partially suggesting that aberrant DDX1 methylation links ACOD1 deficiency to downstream mitochondrial dysfunction and cellular senescence. This discovery organically connects protein methylation modification, metabolic regulation, and cellular senescence, thus establishing a complete signaling axis. However, a direct experimental validation of the functional consequences of ACOD1–DDX1 interaction (specifically, its direct regulation of mitochondrial dysfunction or cellular senescence) is still lacking in the present study. Interestingly, DDX1 has been reported to form large cytoplasmic aggregates with RNA and Ca²⁺ in early mouse embryos, which regulate Ca²⁺ distribution [42]. Ddx1 gene knockout disrupts Ca²⁺ homeostasis, impairs mitochondrial function, and results in embryonic lethality [42]. Additionally, DDX1 enhances Ca²⁺ transient activity in astrocyte microdomains, promoting brain repair after ischemic stroke [43]. Therefore, we speculate that abnormal DDX1 methylation may contribute to DHT-induced AGA progression by impairing mitochondrial function through the disruption of mitochondrial Ca²⁺ homeostasis in DPCs. Although this hypothesis is supported by studies linking DHT-induced alterations in mitochondrial Ca²⁺ homeostasis to AGA [25], whether DDX1 acts specifically through this mechanism warrants further investigation.

To analyze the impact of ACOD1 on mitochondrial metabolism in DPCs, we performed metabolomic profiling following ACOD1 knockdown. The results revealed that ACOD1 silencing induced significant metabolic alterations in DPCs. These findings not only corroborated the aforementioned DHT-induced metabolic reprogramming—characterized by impaired TCA cycle activity and mitochondrial dysfunction—but also demonstrated that ACOD1 knockdown perturbed organic acid metabolism, directly reducing the levels of its metabolic product, itaconate. To further validate this, we employed exogenous itaconate supplementation to assess the involvement of ACOD1 in DHT-induced mitochondrial dysfunction and cellular senescence in DPCs. We found that exogenous itaconate partially reduced the DHT-induced elevation of DDX1 methylation, which was associated with improvements in mitochondrial function, indicators of cellular senescence, and proliferation and migratory capacity, as well as a reduction in apoptosis in DPCs. Furthermore, our in vivo experiments suggested that exogenous itaconate supplementation may alleviate DPC senescence in AGA model mice and delay hair follicle miniaturization and hair cycle abnormalities—reinforcing the role of mitochondrial metabolic reprogramming in DPC senescence. This result is consistent with findings from studies on itaconate in Caenorhabditis elegans, where exogenous itaconate extended the healthy lifespan of worms by preserving mitochondrial integrity, increasing ATP content, and reducing ROS levels—an effect that was shown to be dependent on mitochondrial functional homeostasis [69]. Overall, our data establish ACOD1 as a crucial regulator of DPC functional homeostasis and identify its deficiency as a key step in DHT-induced DPC senescence and mitochondrial dysfunction—thereby providing a novel perspective on the molecular mechanisms underlying AGA.

Itaconate, catalyzed by ACOD1, is not only an important metabolite in the TCA cycle but, more significantly, a key immunometabolic signal that links metabolic reprogramming to immunoregulation in immune cells [30, 70]. Previously, itaconate was believed to exert anti-inflammatory effects in the body, playing a protective role in conditions such as autoimmune diseases [71, 72], sepsis [73], virus-induced inflammation [74], and ischemia-reperfusion injury [38, 75]. However, recent literature has reported that, contrary to its effects on bone marrow-derived macrophages, itaconate exerts pro-inflammatory effects in alveolar macrophages—aggravating lung injury, promoting the production of pro-inflammatory cytokines, and enhancing NLRP3 inflammasome activation [76]. In contrast, 4-OI can suppress the inflammatory response in alveolar macrophages [76]. This suggests that the biological functions of itaconate and its derivatives exhibit significant context-dependence and vary across different cell types, species, and experimental models [70], highlighting the necessity for additional research before clinical application. Therefore, the exact mechanism of action and therapeutic window of itaconate in DPCs require further detailed evaluation. It is worth noting that, as a highly polar α, β-unsaturated dicarboxylic acid, itaconate may have limited transmembrane transport in cells [30, 45]. We therefore used 4-OI—a cell-permeable derivative of itaconate that can be hydrolyzed into itaconate in vivo—in our subsequent experiments [30, 45, 77].

However, our study has several limitations. Our study delineated how ACOD1 downregulation in DPCs promotes AGA via DDX1 hypermethylation-mediated mitochondrial dysfunction and cellular senescence. However, the upstream mechanisms by which DHT suppresses ACOD1 expression remain elusive. Previous studies have shown that DHT-activated AR undergoes nuclear translocation to modulate target gene transcription [25]. Our preliminary results also indicated that the DHT-AR complex downregulated ACOD1 expression by binding to the ACOD1 gene promoter region; however, the underlying mechanism remains to be further elucidated through approaches such as promoter reporter gene assays and genetic interventions. Although we have established that the ACOD1-DDX1 interaction is linked to mitochondrial dysfunction and cellular senescence in DPCs, the functional role of DDX1 methylation in directly regulating these processes, as well as the specific downstream effector molecules through which it may act (such as whether it affects the processing and stability of mitochondria-related mRNAs or mitochondrial Ca²⁺ homeostasis), remains to be further explored. Given that methyltransferase inhibitors such as MS023 or methyl donors such as SAM broadly alter global methylation status, subsequent studies should adopt more specific strategies, such as blocking DDX1 methylation modifications through site-directed mutagenesis or overexpressing DDX1-specific demethylases, to more directly validate the causal relationship between DDX1 methylation changes and mitochondrial dysfunction and cellular senescence in DPCs. Additionally, as the view that hair follicle microinflammation contributes to AGA development is increasingly accepted [46, 47], it is also worth investigating whether itaconate, beyond its effect on DDX1, acts to delay AGA progression through anti-inflammatory pathways in vivo. Metabolomic analysis following ACOD1 knockdown revealed that, in addition to decreased itaconate levels, other mitochondrial metabolites also underwent significant alterations. Subsequent studies should employ supplementation or depletion of specific metabolites combined with targeted metabolic flux analysis to identify the key metabolic changes driving mitochondrial dysfunction and cellular senescence. Importantly, the primary isolation and functional validation of DPCs derived from AGA model mice is currently lacking, making it impossible to directly demonstrate the effects of 4-OI supplementation on mitochondrial function, DDX1 methylation levels, and cellular senescence in DPCs in vivo. This represents an important direction for future research. Furthermore, the safety and effective dosage range of itaconate both in vitro and in vivo still require systematic evaluation, especially given its potential tissue-specific effects. Future research might integrate single-cell sequencing and spatial metabolomics to analyze the dynamic changes of the ACOD1–itaconate axis within the hair follicle microenvironment, and further develop locally administered itaconate nanoformulations to enhance treatment specificity and safety.

In summary, our research provides new insights into the role of DPC senescence in the pathogenesis of AGA. Our study significantly enhances the theoretical status and clinical value of scalp anti-aging strategies for the comprehensive treatment of AGA. This concept aligns well with current research trends in materials science and regenerative medicine for AGA treatment [52, 78], providing valuable insights for the subsequent development of novel interventions targeting the aging of the hair follicle microenvironment. Mechanistically, we delineate a pathway wherein ACOD1 deficiency leads to increased DDX1 methylation, which subsequently disrupts the TCA cycle, induces mitochondrial dysfunction, promotes DPC senescence, and ultimately contributes to AGA progression. This pathway not only deepens our understanding of AGA pathogenesis by highlighting that maintaining mitochondrial redox homeostasis protects DPC function, but also identifies ACOD1 as a novel therapeutic target, with itaconate supplementation emerging as a promising treatment strategy for AGA. This suggests that abnormalities in the ACOD1/DDX1 pathway may be associated with “treatment resistance” in some AGA patients. In the future, identification of patient subpopulations suitable for targeted therapy could be achieved by detecting the expression or methylation levels of key molecules in this pathway, thereby enabling “personalized treatment”; alternatively, ACOD1 agonists or itaconate could be delivered via nanocarriers and combined with existing drugs (such as minoxidil and finasteride) to achieve synergistic efficacy. This lays a multidisciplinary foundation for the development of novel therapeutic approaches for AGA.

Conclusions

In summary, our research provides new insights into the role of DPC senescence in the pathogenesis of AGA. We delineate a pathway in which ACOD1 deficiency leads to increased DDX1 methylation, which subsequently disrupts the TCA cycle, impairs mitochondrial function, promotes DPC senescence, and ultimately contributes to AGA progression. This pathway not only deepens our understanding of AGA pathogenesis by highlighting that maintaining mitochondrial redox homeostasis protects DPC function, but also identifies ACOD1 as a novel therapeutic target, with itaconate supplementation emerging as a promising treatment strategy. However, our findings are limited by the small sample size and the need to clarify several key issues in future studies, such as the specific downstream effector molecules of DDX1 methylation affecting mitochondrial function, the potential in vivo anti-inflammatory mechanisms of itaconate beyond DDX1, and its safety profile and effective dosage range.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AGA

androgenetic alopecia

DPC

dermal papilla cell

DHT

dihydrotestosterone

ACOD1

aconitate decarboxylase 1

DP

dermal papilla

AR

androgen receptor

TCA

tricarboxylic acid

OXPHOS

oxidative phosphorylation

mtROS

mitochondrial reactive oxygen species

DDX1

DEAD-box 1

H&E

hematoxylin–eosin

IHC

immunohistochemical

IF

immunofluorescence

SA-β-gal

senescence-associated β-galactosidase

TEM

transmission electron microscopy

OCR

oxygen consumption rate

PER

proton efflux rate

Co-IP

co-immunoprecipitation

IP/MS

immunoprecipitation and mass spectrometry

4-OI

4-octyl itaconate

Author contributions

MZ: Contributed to conceptualization, methodology, validation, formal analysis, investigation, data curation, writing – original draft, and visualization. QW: Contributed to validation, formal analysis, investigation, and writing – review & editing. YD1: Contributed to methodology, validation, formal analysis, investigation, and data curation. CL: Contributed to validation, formal analysis, and investigation. YD2: Contributed to methodology, validation, formal analysis, and visualization. XL: Contributed to investigation, data curation, visualization, and writing – review & editing. LB: Contributed to methodology, validation, and formal analysis. CW: Contributed to methodology, validation, formal analysis, and writing – review & editing. JJ: Contributed to validation, investigation, and writing – review & editing. WS: Contributed to resources, data curation, writing – review & editing, supervision, and project administration. WF: Contributed to conceptualization, investigation, resources, data curation, writing – review & editing, supervision, project administration, and funding acquisition. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (NO. 81972954).

Data availability

The RNA-Seq raw data have been deposited in the National Center for Biotechnology Information (NCBI) under the BioProject accession number PRJNA1434740. Source data of the metabolomic analysis have been deposited in the EMBL-EBI MetaboLights database with the identifier MTBLS14024. The authors declare that all the data supporting the findings of this study are available within the article and its supplementary information files or from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of the First Affiliated Hospital with Nanjing Medical University (Approval number: 2024-SR-638). The study was conducted in accordance with the Declaration of Helsinki, and all participants provided full written informed consent prior to sample and data collection. For animal welfare considerations, all procedures were approved by the Institutional Animal Care and Use Committee of Nanjing Medical University (Approval No.: IACUC-2407031).

Consent for publication

We confirm that all the authors have reviewed and approved the final version of the manuscript and agreed to its publication.

Competing interests

The authors declare no competing interests.