Abstract Background Erectile dysfunction (ED) is a multifactorial disorder, with mitochondrial dysfunction increasingly recognized as an important contributor to its pathogenesis. Aim This study aimed to characterize the single-cell landscape of ED and investigate the impact of mitochondrial function on cellular heterogeneity. Methods We performed single-cell RNA sequencing analysis on ED samples ([30]GSE206528), screened for ED-related mitochondrial genes, evaluated mitochondrial activity using area under the curve cell scoring at the single-cell level, and conducted subclustering, cell–cell communication, pseudotime trajectory, and pathway enrichment analyses to systematically characterize key cell populations. Outcomes The principal finding is that fibroblasts (FB) and endothelial cells (EC) display significant mitochondrial heterogeneity associated with ED. Results A total of 64 993 high-quality cells were classified into seven major cell types. Among these, FB and EC exhibited significant mitochondrial heterogeneity. Seventy-three ED-related mitochondrial genes were identified, with 11 and six mitochondrial activity-associated genes in FB and EC, respectively. Subclustering analysis revealed six FB and four EC subpopulations, with distinct functional pathways. Cell–cell communication analysis indicated increased tumor necrosis factor, TNF-related apoptosis-inducing ligand, and wingless/integrated signaling in high-mitochondrial-activity groups. Pseudotime analysis suggested FB0 and EC1 as progenitor states, progressing toward FB4 and EC0, respectively. Pathway enrichment highlighted shared metabolic and stress-response pathways in FB and EC. Clinical Implications These results suggest that targeting mitochondrial dysfunction in FB and EC may offer novel therapeutic approaches for ED. Strengths & Limitations The study's strengths lie in its comprehensive single-cell characterization and functional annotation, while limitations include sample representativeness and the lack of direct experimental validation. Conclusion This study provides a comprehensive single-cell landscape of ED, identifying mitochondrial dysfunction as a key contributor to cellular heterogeneity. FB and EC emerged as critical regulators, with potential implications for targeted therapeutic strategies. Keywords: erectile dysfunction, single-cell RNA sequencing, mitochondrial dysfunction, fibroblasts, endothelial cells, pseudotime analysis Introduction Erectile dysfunction (ED) is a common male sexual dysfunction, defined as the persistent or recurrent inability to achieve and/or maintain an erection sufficient for satisfactory sexual performance.[31]^1 ED affects 1%–10% of men under 40 years and 30%–50% of men aged 40–70.[32]^2 By 2025, the global number of ED patients is estimated to reach 322 million.[33]^3 Current treatments for ED focus on symptom relief rather than a permanent cure.[34]^4 Although ED is not life-threatening, it significantly impacts an individual’s quality of life. Additionally, ED is an early symptom and risk factor for vascular and peripheral vascular diseases.[35]^5 Therefore, early diagnosis and active treatment are particularly important. Mechanistic studies suggest that ED results from a multifactorial interplay involving aging, vascular, neurological, hormonal, and psychological factors.[36]^6 Increasing evidence indicates that mitochondria play a crucial role in the onset and progression of ED. Mitochondria are essential organelles responsible for cellular energy metabolism, reactive oxygen species (ROS) generation, and apoptosis regulation. Mitochondrial function is particularly critical for high-energy-demanding tissues such as corpus cavernosum smooth muscle cells (SMC) and endothelial cells (EC).[37]^7 Previous studies have shown mitochondrial damage in smooth muscle and EC in ED rat models[38]^8^,[39]^9; mitochondrial apoptotic pathway activation has also been observed in circulating angiogenic cells from ED patients.[40]^10 Mechanistically, mitochondrial dysfunction impairs the bioavailability of nitric oxide (NO), a key vasodilatory molecule necessary for corpus cavernosum smooth muscle relaxation, by disrupting endothelial nitric oxide synthase activity and increasing ROS production.[41]^11–13 Elevated mitochondrial-derived ROS can inactivate NO via the formation of peroxynitrite, leading to endothelial dysfunction and impaired vasodilation.[42]^12 Furthermore, mitochondrial metabolic dysregulation affects ATP production, which is essential for smooth muscle contraction-relaxation cycling, and may promote a pro-apoptotic environment that reduces smooth muscle cell density and increases fibrosis in the corpus cavernosum.[43]^14 Although the involvement of mitochondria in ED pathogenesis is well recognized, few studies have explored the role of mitochondria in ED at the cellular level. In recent years, advancements in single-cell transcriptomics have provided unprecedented opportunities to analyze the cellular heterogeneity of penile tissues and explore the molecular mechanisms underlying ED.[44]^15 Unlike conventional bulk tissue analyses, which obscure cell-specific signals by averaging gene expression across heterogeneous populations, single-cell RNA sequencing (scRNA-seq) enables the identification of distinct cellular subpopulations and their unique gene expression profiles within penile tissues.[45]^16 This approach facilitates the discovery of previously unrecognized cell types, delineates functional states of known cells, and reveals molecular pathways specifically associated with ED pathogenesis. Building on these technological advances, this study aimed to investigate the role of mitochondrial activity in ED through single-cell transcriptomic analysis, elucidating how mitochondrial dysfunction affects cellular states in penile tissues, ultimately contributing to ED. Traditional bulk transcriptomic analyses are limited in their ability to resolve mitochondrial alterations specific to individual cell types. In contrast, our application of scRNA-seq allows for the precise identification of cellular subpopulations exhibiting mitochondrial dysfunction, thereby providing a comprehensive cell-type-specific characterization of mitochondrial activity within the context of ED. A deeper understanding of these molecular mechanisms may provide novel mitochondrial-targeted therapeutic strategies for ED and offer new insights into the pathogenesis of ED. Methods Data acquisition The single-cell dataset [46]GSE206528 was downloaded from the GEO database ([47]http://www.ncbi.nlm.nih.gov/geo/), which includes corpus cavernosum tissue samples from five ED patients and three control individuals. Additionally, 1136 mitochondrial functional genes were collected from the MitoCarta3.0 database ([48]http://www.broadinstitute.org/mitocarta). scRNA-seq analysis Single-cell data analysis was performed using the R package Seurat. Quality control was conducted by setting the filtering criteria as 200 < nFeature_RNA < 10 000, percent.mt < 10%, and min. cell = 3 to remove low-quality single cells.[49]^17 Dimensionality reduction and clustering were performed based on the top 30 principal components in PCA, with resolution set to 0.4 for clustering.[50]^18 The filtering thresholds and the number of principal components were based on quality metrics typically used in single-cell transcriptomic analyses to balance data retention and noise reduction. The clustering resolution was chosen empirically to yield a meaningful number of distinguishable cell populations, validated by marker gene expression and Uniform Manifold Approximation and Projection (UMAP) visualization. Cell type annotation was conducted using marker genes identified from previous literature,[51]^19 which defined different cell populations in the human corpus cavernosum. These marker genes include GZMA, CD8A, CD3D, CD68, CD163, FCER1G, NRXN1, SCN7A, S100B, DES, MYH11, ACTA2, NOTCH3, SSTR2, KCNJ8, LUM, DCN, PDGFRA, CD34, PECAM1, and VWF.[52]^19 These markers were used to assign specific cell types to the identified clusters. Identification of differentially expressed genes between ED and control groups Differential expression analysis was performed for each cell subpopulation using the FindMarkers function, comparing ED vs. Control groups. To control for multiple comparisons, the Benjamini-Hochberg false discovery rate correction was applied, and genes with an adjusted P < .05 and |log₂ fold change| > 1 were considered differentially expressed. The intersection of differentially expressed genes (DEGs) from different cell subpopulations and mitochondrial functional genes was taken to identify ED-related mitochondrial functional genes. Relationship between mitochondria and ED-associated cellular heterogeneity To further investigate mitochondrial functional heterogeneity in corpus cavernosum cells of ED patients, area under the curve cell (AUCell) in R was used to calculate single-cell mitochondrial functional scores based on the ED-related mitochondrial functional genes identified in the previous step. Single-cell data were divided into high-score and low-score groups according to the median score. A rank-sum test was conducted to compare mitochondrial functional scores between the ED and Control groups. Identification and analysis of key cell subpopulations The proportion of cells in the high-score and low-score groups was compared, and cell types with significant proportion changes were selected as key cell subpopulations. A secondary single-cell clustering and dimensionality reduction analysis was performed on these key cell subpopulations. Differential expression analysis was conducted with the criteria |log2FC| > 2 and P < .05, followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of the differentially expressed marker genes to identify the primary functions of different subpopulations. Cell–cell communication and pseudotime analysis CellChat analysis was conducted with default settings to ensure robust detection of ligand-receptor interactions. It was performed separately for the high-score and low-score groups using the CellChat R package, comparing differences in communication quantity, communication strength, and signaling pathways between the two groups. For pseudotime trajectory analysis, Monocle was employed, allowing for the reconstruction of dynamic differentiation trajectories among relevant cell populations. Molecular pathway enrichment analysis To examine enrichment differences across different cell types, the ReactomeGSA package was used to perform pathway enrichment analysis between key cell subpopulations and other cell types. The top 20 enriched pathways with the largest differences were selected for visualization, with the difference calculated as (key cell enrichment score)—(mean enrichment score of other cell types). Statistical analysis All statistical analyses were performed using R version 4.3.1. The confidence interval was 95% and P < .05 was considered statistically significant. Results Single-cell landscape of ED First, the cellular landscape of ED was explored through a comprehensive single-cell analysis. After stringent quality control, a total of 64 993 high-quality cells were retained for further investigation. PCA was performed using the top 3000 highly variable genes. As shown in [53]Figure 1A, the explained variance of principal components plateaued when 30 components were reached. Subsequently, dimensionality reduction and clustering were conducted, identifying 17 distinct cell clusters, which were visualized using UMAP ([54]Figure 1B). These clusters were annotated based on marker genes curated from the literature ([55]Figure 1C), with detailed information on cell types and their corresponding marker genes provided in [56]Table S1. Ultimately, seven major cell types were identified: EC, fibroblasts (FB), T cells (T), SMC, Schwann cells (SWC), macrophages (MAC), and pericytes (PC) ([57]Figure 1D). Figure 1. [58]Figure 1 [59]Open in a new tab Single-cell atlas of ED. (A) Elbow plot of PCA variance explanation. (B) UMAP plot of cell clusters after dimensionality reduction. (C) UMAP plot of annotated cell subpopulations after dimensionality reduction. (D) Heatmap of marker gene expression across cell subpopulations. Identification of mitochondria-related genes in ED DEGs were identified in the seven annotated cell types. A total of 676, 647, 462, 590, 40, 756, and 531 highly expressed DEGs were detected in EC, FB, T cells, SMC, SWC, MAC, and PC, respectively ([60]Figure 2A). After merging and removing duplicates, 2252 upregulated DEGs in ED were obtained. These DEGs were then intersected with 1136 mitochondria-related genes, resulting in the identification of 73 ED-associated mitochondrial genes ([61]Figure 2B). The full list of these 73 genes is provided in [62]Table S2. Figure 2. [63]Figure 2 [64]Open in a new tab Mitochondrial function-related genes in ED. (A) Volcano plot of differentially expressed genes between ED and control in single-cell subpopulations. (B) Venn diagram of overlapping marker genes and mitochondrial genes between ED and control groups. Relationship between mitochondria and cellular heterogeneity in ED To investigate the significance of the 73 mitochondrial genes in ED, AUCell was used to calculate the mitochondrial gene set scores at the single-cell level. The distribution of mitochondrial gene set scores and the classification of high- and low-scoring single-cell subpopulations are shown in [65]Figure 3A-B. A higher proportion of high mitochondrial scores was observed in EC and SMC, whereas a greater proportion of low mitochondrial scores was found in FB. To validate the reliability of the AUCell scores, correlations were assessed between AUCell scores and those derived from the AddModuleScore and PercentageFeatureSet algorithms ([66]Figure 3C-D). The AUCell score was significantly correlated with the AddModuleScore (r = 0.642, P < .0001) and the PercentageFeatureSet score (r = 0.442, P < .0001). Figure 3. [67]Figure 3 [68]Open in a new tab Relationship between single-cell heterogeneity and mitochondrial function in ED. (A) UMAP plot of AUCell scores for mitochondrial gene sets. (B) UMAP plot showing the proportion of cells with high and low mitochondrial function gene set scores across subpopulations. (C, D) Scatter plots showing the correlation between AUCell scores and AddModuleScore or PercentageFeatureSet scores. (E) Violin plot comparing AUCell scores between ED and control groups. (F) Bar plot showing the proportion of cells in different scoring groups. Additionally, a comparison between the normal and ED groups revealed a significant increase in mitochondrial scores in ED (P < .01, [69]Figure 3E). Then, proportions of cell populations in the high- and low-mitochondrial-score groups were analyzed. The most notable change was observed in FB, where the proportion decreased by 23.4% in the high-score group compared to the low-score group ([70]Figure 3F). In contrast, EC showed a 15.0% increase in the high-score group compared to the low-score group ([71]Figure 3F). These findings suggest that FB and EC may be key cellular subpopulations associated with mitochondrial function in ED. Identification of DEGs associated with mitochondrial activity in FB and EC To explore the relationship between mitochondrial activity and the key cell populations identified (FB and EC), DEGs between high/low mitochondrial activity groups were analyzed within these two cell types. As shown in [72]Figure 4A, the most significantly upregulated genes in the high-activity FB group included TFPI2, SELE, MFSD2A, PMAIP1, and ADAMTS4. Similarly, in the high-activity EC group, the most significantly upregulated genes were MT1A, MT1G, SLC25A25, SOD2, and CIRBP-AS1 ([73]Figure 4B). Subsequently, these upregulated DEGs in the high-activity groups were intersected with the 73 previously identified ED-associated mitochondrial genes. As a result, 11 and six mitochondrial activity-related DEGs were identified in FB and EC, respectively ([74]Figure 4C). The detailed gene list is provided in [75]Table 1. Three common genes existed in FB and EC, including SLC25A25, PDK4, and SOD2. Figure 4. [76]Figure 4 [77]Open in a new tab DEGs associated with mitochondrial activity in key cell subpopulations. (A) Volcano plot of DEGs between high and low mitochondrial activity groups in FB cells. (B) Volcano plot of DEGs between high and low mitochondrial activity groups in EC cells. (C) Bar plot showing the overlap between upregulated DEGs and mitochondrial function genes in key cell subpopulations. Table 1. Mitochondrial activity-related DEGs in FB and EC. FB EC SLC25A25 SLC25A25 ARG2 SLC25A44 SLC25A33 PMAIP1 MCL1 SOD2 NOCT TMEM70 PMAIP1 PDK4 IFI27 SOD2 PDK4 CYCS TSTD1 [78]Open in a new tab Subclustering of FB and EC To further investigate the role of FB and EC in ED, these two cell types were subjected to subclustering analysis. First, FBs from the ED group were extracted, and PCA was performed using the top 3000 highly variable genes. The variance in principal component information stabilized when the number of principal components reached 20 ([79]Figure 5A). UMAP coordinates were then computed using all Harmony-aligned coordinates, followed by clustering, which resulted in the identification of six distinct FB subclusters FB0-FB5 ([80]Figure 5B). GO and KEGG enrichment analyses were performed for marker genes of each cluster, and [81]Figure 5C presents the top 100 most significant marker genes and the top five most enriched pathways for each subcluster. A similar subclustering approach was applied to ECs, yielding four distinct EC subclusters EC0-EC3 ([82]Figure 5D-E). The top 100 most significant marker genes and top five most enriched pathways for each EC subcluster are shown in [83]Figure 5F. Figure 5. [84]Figure 5 [85]Open in a new tab Secondary clustering of key cell subpopulations. (A) Elbow plot of PCA variance explanation in FB cells. (B) UMAP plot of secondary clustering in FB cells. (C) Heatmap of marker gene expression in FB secondary clusters with the top five enriched pathways. (D) Elbow plot of PCA variance explanation in EC cells. (E) UMAP plot of secondary clustering in EC cells. (F) Heatmap of marker gene expression in EC secondary clusters with the top five enriched pathways. Cell–cell communication for FB and EC To further explore intercellular interactions, we performed cell–cell communication analysis by dividing disease samples into high- and low-activity groups based on mitochondrial scores. The number and intensity of cellular interactions in the low-activity group are shown in [86]Figure 6A, while those in the high-activity group are displayed in [87]Figure 6B. Several communication pathways showed significant differences between the two groups, with tumor necrosis factor (TNF), TNF-related apoptosis-inducing ligand (TRAIL), and wingless/integrated (WNT) being upregulated in the high-activity group, whereas cell adhesion molecule (CADM), reelin (RELN), and selectin p ligand (SELPLG) pathways were downregulated ([88]Figure 6C-D). Further analysis identified the specific cellular interactions associated with these altered pathways, revealing that in the high-activity group, EC, FB0, and FB4 upregulated TNF receptor signaling, EC and FB4 exhibited increased TRAIL ligand expression, EC1, EC2, and EC3 upregulated TRAIL receptor expression, and EC0, EC1, FB4, and FB5 demonstrated increased WNT receptor expression. In contrast, pathways such as lymphocyte-specific protein tyrosine kinase, neuregulin, cadherin, and SELPLG were significantly downregulated and completely absent in the high-activity group ([89]Figure 6E-H). Figure 6. [90]Figure 6 [91]Open in a new tab Cell communication analysis. (A) Network diagram of communication quantity and strength in the low-activity group. (B) Network diagram of communication quantity and strength in the high-activity group. (C) Bar plot of overall cell communication quantity and strength distribution. (D) Stacked bar plot of pathway strength in high- and low-activity groups. (E) Heatmap of upregulated communication pathways (source) in high- and low-activity groups. (F) Heatmap of upregulated communication pathways (target) in high- and low-activity groups. (G) Heatmap of downregulated communication pathways (source) in high- and low-activity groups. (H) Heatmap of downregulated communication pathways (target) in high- and low-activity groups. Pseudotime trajectory analysis To investigate the potential developmental trajectory of different FB and EC subclusters during ED progression, we conducted pseudotime trajectory analysis. The results suggested that within the FB lineage, FB0 served as the developmental starting point, while FB4 represented the terminal state, indicating a progressive cellular transition towards this endpoint ([92]Figure 7A). Similarly, in the EC lineage, EC1 was identified as the potential developmental origin, whereas EC0 was predicted to be the final differentiated state, suggesting a structured and directional progression of EC subclusters in the context of ED ([93]Figure 7B). Figure 7. [94]Figure 7 [95]Open in a new tab Pseudotime analysis. (A) Pseudotime developmental trajectory of FB0-5 subpopulations. (B) Pseudotime developmental trajectory of EC0-3 subpopulations. Pathway enrichment analysis in FB and EC Finally, pathway enrichment analysis was conducted to further explore the functional significance of FB and EC in ED. Results showed the top 20 most significantly upregulated pathways in these cell types compared to others ([96]Figure 8A-B). Among these pathways, pyrophosphate hydrolysis, sterols are 12-hydroxylated by CYP8B1, NEIL3-mediated resolution of ICLs, and threonine catabolism were found to be commonly upregulated in both FB and EC, suggesting their potential involvement in shared biological processes. Further quantification of pathway activity scores across different cells revealed that prophosphate hydrolysis, serols are 12-hydroxylated by CYP8B1, and treonine catabolism were broadly expressed in most cells, whereas NEIL3-mediated resolution of ICLs exhibited relatively limited expression, suggesting a more restricted functional role ([97]Figure 8C-F). Figure 8. [98]Figure 8 [99]Open in a new tab Molecular pathway enrichment analysis. (A) Heatmap of the top 20 significantly upregulated pathways in EC cells compared to other cells. (B) Heatmap of the top 20 significantly upregulated pathways in FB cells compared to other cells. (C) UMAP plot of pyrophosphate hydrolysis score. (D) UMAP plot of sterols are 12-hydroxylated by CYP8B1 score. (E) UMAP plot of NEIL3-mediated resolution of ICLs score. (F) UMAP plot of threonine catabolism score. Discussion This study conducted a comprehensive single-cell analysis of ED, with a particular focus on mitochondrial dysfunction and its role in cellular heterogeneity. Through rigorous quality control and clustering, we identified seven major cell types within the ED microenvironment, among which FB and EC exhibited significant changes in mitochondrial activity. Mitochondria play a crucial role in cellular energy metabolism, oxidative stress regulation, and apoptosis, and their dysfunction is associated with various pathological conditions.[100]^20 Our findings are consistent with previous studies indicating that mitochondrial dysfunction leads to endothelial dysfunction and fibrosis,[101]^21^,[102]^22 both of which are key features of ED.[103]^23 One of the key findings of our study is that EC and FB exhibit the most substantial changes under different mitochondrial activity states. In ED, mitochondrial activity significantly increased in EC, whereas decreased in FB. These findings suggest that mitochondrial metabolism may have cell-type-specific roles in the pathogenesis of ED. Endothelial injury is one of the most prominent pathological changes in cavernous disorders and may occur at the early stages of ED.[104]^24 Notably, ECs, which are critical for maintaining vascular homeostasis and promoting angiogenesis, exhibited a significant increase in mitochondrial genome scores under ED conditions. This aligns with previous evidence showing that endothelial dysfunction in ED is closely related to increased oxidative stress and mitochondrial dysregulation. In castrated rat cavernous tissues, EC undergo atrophy, mitochondrial numbers decrease, and NO bioavailability is impaired.[105]^9^,[106]^25 Given that NO is essential for vasodilation and erectile function, our findings reinforce the hypothesis that mitochondrial dysfunction in EC may impair vasodilation and cavernous blood perfusion, ultimately contributing to ED development (PMID: 26878803). Previous studies have suggested that perivascular FBs serve as a fundamental component of ED and that their heterogeneity is a hallmark of age-related ED.[107]^26 However, few studies have linked mitochondrial dysfunction in ED to FBs. In pulmonary fibrosis, mitochondrial dysfunction in lung FBs has been shown to enhance sensitivity to fibrotic activation.[108]^27 The observed decline in mitochondrial activity in FBs suggests that fibroblast-mediated fibrosis may be driven by metabolic reprogramming. To further establish the correlation between mitochondrial function and EC and FB in ED, we identified 73 ED-related mitochondrial genes, among which 11 and six mitochondrial activity-associated DEGs were found in FB and EC, respectively. Notably, SLC25A25, PDK4, and SOD2 were common to both FB and EC, highlighting their potential roles as key mitochondrial regulators in the ED microenvironment. SLC25A25 is a mitochondrial carrier protein that has been shown to act downstream of TRPP2 and is involved in Ca^2+ signaling transduction.[109]^28 SOD2 is a crucial mitochondrial antioxidant enzyme that plays a key role in neutralizing superoxide radicals and mitigating oxidative stress. Reduced SOD2 expression has been linked to endothelial dysfunction and impaired vascular relaxation.[110]^29 Similarly, PDK4, a regulator of mitochondrial pyruvate metabolism, is involved in metabolic adaptation during oxidative stress. Previous studies have reported PDK4 upregulation in both ED patients and ED model rats.[111]^30^,[112]^31 In the study by Ock et al., miR-148a-3p was suggested to enhance cavernous neurovascular regeneration in diabetes-induced ED mice by inhibiting PDK4.[113]^32 We hypothesize that the dysregulation of SLC25A25, SOD2, and PDK4 may alter energy homeostasis in ECs and FBs, thereby exacerbating ED progression. Cell–cell communication analysis revealed dynamic intercellular interactions regulated by mitochondrial activity, further elucidating the complex regulatory network of ED. Notably, TNF, TRAIL, and WNT signaling pathways were significantly upregulated in the high mitochondrial activity group, whereas CADM, RELN, and SELPLG pathways were downregulated. The upregulation of TNF and TRAIL signaling suggests that an inflammatory microenvironment may contribute to vascular dysfunction and tissue remodeling. Previous studies have reported a strong association between elevated TNF-α levels and ED.[114]^33 Chronic TNF-α-mediated inflammation can lead to endothelial dysfunction and increased oxidative stress, thereby aggravating ED severity.[115]^34^,[116]^35 Moreover, WNT dysregulation has been implicated in ED pathogenesis, particularly in fibrosis and EC/SMC dysfunction.[117]^36 The observed upregulation of WNT signaling in the high-activity group suggests that mitochondrial function may be closely associated with fibrotic remodeling and angiogenesis in ED. Conversely, the downregulation of the CADM pathway may indicate impaired cell adhesion and extracellular matrix interactions, potentially contributing to tissue deterioration in ED. The identification of distinct FB and EC subpopulations further clarifies the heterogeneity of these key cell types and their potential functional roles in ED. Notably, pseudotime trajectory analysis indicated a progressive transition from FB0 to FB4 and from EC1 to EC0, providing insights into cellular differentiation dynamics. Given that FB4 exhibited increased expression of TNF and TRAIL receptors, it may represent a pro-inflammatory and pro-fibrotic fibroblast subtype, contributing to pathological remodeling in ED. Similarly, EC0 was predicted to be the final differentiated state of the EC lineage, potentially playing a critical role in vascular dysfunction and impaired angiogenesis. Pathway enrichment analysis further highlighted the shared metabolic and regulatory pathways between FBs and ECs, including pyrophosphate hydrolysis, sterols are 12-hydroxylated by CYP8B1, NEIL3-mediated resolution of ICLs, and threonine catabolism, all of which were significantly upregulated. These pathways are closely related to mitochondrial metabolism, oxidative stress regulation, and DNA damage repair,[118]^37^,[119]^38 emphasizing the fundamental role of mitochondria in maintaining cellular function in ED. The upregulation of the NEIL3-mediated resolution of ICLs pathway suggests an adaptive response to oxidative DNA damage,[120]^39 a common consequence of mitochondrial dysfunction.[121]^40 However, the expression of NEIL3 is relatively limited across different cell types, indicating that it may play a cell type-specific role in the pathophysiology of ED. Conclusion In conclusion, our findings suggest that mitochondrial dysfunction is a central feature of ED, influencing cell heterogeneity, inflammatory signaling, and intercellular communication. Genes and pathways associated with mitochondrial activity in FBs and ECs represent potential therapeutic targets for ED intervention. Given the critical roles of oxidative stress and metabolic reprogramming in the pathogenesis of ED, future studies should focus on strategies aimed at restoring mitochondrial homeostasis, such as the use of antioxidants or agents that enhance mitochondrial function. Targeting key mitochondrial regulators such as SOD2 and PDK4 also represents a promising approach; although no clinical therapies directly targeting these proteins have been approved to date, ongoing preclinical studies support their potential as therapeutic targets. Additionally, further investigation into the functional consequences of the identified subpopulations and their contributions to ED pathogenesis is essential for enhancing our understanding of the disease and identifying effective therapeutic strategies. Supplementary Material supplementary_tables_qfaf049 [122]supplementary_tables_qfaf049.docx^ (17.7KB, docx) Acknowledgments