HUVECs were purchased from Cell Bank at the Chinese Academy of Sciences. Cells were routinely cultured in DMEM medium (Gibco, REF: C11995500BT) supplemented with 10% fetal bovine serum (ExCell Bio, Cat: FSP500) at 37 °C in a humidified incubator with 5% CO₂.

CoQ10 was purchased from MedChemExpress (Cat: HY-N0111). To prepare the CoQ10 stock solution, 10 mg of CoQ10 powder was precisely weighed and dissolved in 1.1583 mL of N, N-dimethylformamide (DMF) (MCE, Cat: HY-Y0345) using a 37 °C water bath and 40 kHz ultrasonication for 10 min to prepare a 20 mM stock solution. When the growth density of the cells to be treated reached 70%–80%, CoQ10-containing culture medium was added, and cells were maintained for 48 h for subsequent experiments.

The RIPA lysis buffer (Epizyme, Cat: PC101) was mixed with protease inhibitor (Abiowell, Cat: AWB0159a) at a ratio of 100:1 and added to cells rinsed with PBS. After thoroughly lysis on ice for 30 min, the lysate was centrifugated at 4 °C, 12,000 rpm for 10 min and the supernatant was collected as the total protein sample. Protein samples were mixed with 5× loading buffer (Cwbio, Cat: CW0027S) at a 4:1 ratio and heated in a 95 °C metal bath for 6–8 min to denature proteins. Equal amounts of protein were separated by SDS-PAGE and electrotransferred to a PVDF membrane (Merck Millipore, Cat: IPVH00010). After transferring, the membrane was blocked at room temperature for 1–2 h with TBST containing 5% skim milk powder, then incubated overnight at 4 °C with the corresponding primary antibody. Primary antibody details are as follows: Coq4 antibody (0.04–0.4 μg/mL, Abcam, Cat: ab126295), 4-Hydroxynonenal antibody (1:2000, Bioss, Cat: BS-6313R), ACSL4 antibody (1:5000, Zenbio, Cat: R24265), SLC7A11 antibody (1:1000, CST, Cat: 98051S), FTH1 antibody (1:1000, Zenbio, Cat: R23306), FSP1 antibody (1:8000, Proteintech, Cat: 20886-1-AP), GPX4 antibody (1:2000, Zenbio, Cat: R381958), Beta Actin antibody (1:5000, Servicebio, Cat: GB15003-100), and GAPDH antibody (1:5000, Servicebio, Cat: GB15004-100). The membrane was then incubated with the secondary antibody (1:10,000, Abcam, Cat: ab6721) at room temperature for 1–2 h followed by the addition of enhanced chemiluminescent reagent (Biosharp, Cat: BL520B) and detection on a chemiluminescence imaging system. ImageJ software was used to perform quantitative analysis of the intensity values of the target bands.

Tissue slides were fixed, paraffin-removed, and subjected to heat-induced antigen retrieval. Afterward, the Triple-Label Four-Color Multiplex Fluorescent Staining Kit (Enhanced) (Aifang Bio, Cat: AFIHC034) was employed for subsequent procedures. Specifically, tissue slices were first incubated overnight with the corresponding primary antibody, then incubated at room temperature away from light for 30 min with the kit-provided fluorescently labeled secondary antibody. This step was repeated using another primary antibody from the same species, followed by incubation with the kit-provided secondary antibody labeled with a different wavelength. Finally, sealing the slides with DAPI-containing anti-fluorescence quenching sealing medium, then observing and capturing images under a fluorescence microscope. Primary antibody including Coq4 antibody (1:200, Proteintech, Cat: 16654-1-AP), ACSL4 antibody (1:200, Zenbio, Cat: R24265) and CD34 antibody (1:200, Proteintech, Cat: 14486-1-AP).

Intracellular Fe²⁺ levels were measured using the FerroOrange fluorescent probe (Dojindo, Cat: F374). Intracellular glutathione (GSH) content was assessed with the GSH Detection Kit (Nanjing Jiancheng Bioengineering Institute, Cat: A006-2-1). Intracellular malondialdehyde (MDA) levels were measured using the MDA Detection Kit (Beyotime, Cat: S0131S), and intracellular coenzyme Q10 (CoQ10) was detected using the CoQ10 Detection Kit (LEAGENE, Cat: 1227A24).

Short hairpin RNA (shRNA) targeting the Coq4 gene sequence (shRNA) (GGT​CGC​GAG​TAT​CTC​CGT​TTC) and a negative control (TTC​TCC​GAA​CGT​GTC​ACG​T) were generated and cloned into the GV493 (pFU-GW-016) vector (purchased from Shanghai Genechem) containing a BsmBI site. Recombinant vectors were verified by DNA sequencing. An appropriate volume of viral solution was added to HUVECs (at approximately 20%–30% confluence) using HitransG P infection buffer. After 48–72 h of infection, puromycin was added for 72 h to conduct continuous selection. Subsequently, cells were dissociated into a single-cell suspension and amplified to sufficient numbers, ultimately yielding a stably transfected cell line for subsequent experiments.

FSP1 overexpression and control vector (purchased from RIBOBIO, Guangzhou, China) were transfected into Coq4 stably knocked-down endothelial cells at density of approximately 70%–80% using Lipo3000 transfection reagent (Thermo Fisher, REF: L3000015). Subsequent experiments were conducted 48 h post-transfection.

Coq4 heterozygote mice (Coq4^+/−) on a C57BL/6 background were purchased from Jiangsu Jicui Yao Kang Biological Co., Ltd. (B6/JGpt⁃ Coq4em2Cd3877/Gpt), which were crossed to generate Coq4 homozygous (Coq4^−/−) embryos and wild-type (WT) embryos.

All animals used in this study were housed individually in accordance with the Mouse Welfare and Ethics Regulations in University of South China. Housing conditions included a 24-h light-dark cycle (light: 8:00–20:00; dark: 20:00–8:00), ambient temperature of (23 ± 2) °C, relative humidity of 55%–60%, and ad libitum access to food and water. Bedding was changed regularly, and feed and water were replenished as needed. All animal procedures in this study complied with the standards of the Ethics Committee for experimental animals in University of South China.

Statistical analysis was performed using Prism 10.0 (GraphPad). Experimental data are presented as mean ± standard deviation. Comparisons between two groups were conducted using t-tests, while comparisons among multiple groups were performed using one-way analysis of variance (ANOVA), Differences were considered statistically significant at P < 0.05. Two-way ANOVA was used to assess interactions between two factors, with P < 0.05 indicating statistical significance, the Bliss independence model was employed to evaluate synergistic effects between factors, calculated using the formula: Synergy Score = E_observed - E_expected, A Synergy Score >0 was defined as synergistic interaction.

To investigate the physiological function of the Coq4 gene in vivo, we successfully generated a Coq4 knockout (Coq4^−/−) mouse model. Histological analysis revealed that at embryonic day 9.5 (E9.5), angiogenesis in the placenta of Coq4^−/− mice was significantly reduced compared to wild-type (WT) mice (Figure 1A). To further evaluate vascular development, we employed immunofluorescence staining to detect expression of the neonatal endothelial cell marker CD34 in placental tissue. Results showed a significant reduction in CD34-positive staining in the placenta of Coq4^−/− mice compared to WT mice, indicating markedly decreased neovascularization (Figure 1B).

Based on the above findings, to further investigate the specific mechanisms by which Coq4 functions in vascular development and homeostasis maintenance, we selected human umbilical vein endothelial cells (HUVECs) as an in vitro model for subsequent studies. First, we established stable Coq4 knockdown (shCoq4) and negative control (shNC) HUVEC cell lines using shRNA technology. Western blotting revealed significantly reduced Coq4 protein expression in the shCoq4 group compared to the shNC group (P < 0.01, Figure 2A), confirming successful establishment of the stable cell lines. Next, cell scratch assays evaluated migration capacity, revealing that Coq4 deficiency significantly inhibited cell migration (P < 0.01 at 24h; P < 0.05 at 48 h, Figure 2B). Simultaneously, the matrix gel formation assay evaluated angiogenesis capacity, revealing that Coq4 deletion significantly inhibited endothelial cell tube formation ability (P < 0.05, Figure 2C). These results collectively demonstrate that Coq4 deletion significantly suppresses both endothelial cell migration and tube formation capacity.

Subsequently, we performed RNA-seq analysis on the established cell models. By calculating the FPKM values (fragments per kilobase per million mapped reads) for all genes in each sample, we obtained intra- and inter-group correlation coefficients and generated a correlation heatmap. Results showed that the intra-group correlation coefficient R² exceeded 0.98 for biological replicates, while the inter-group correlation coefficient R² between shCoq4 and shNc groups exceeded 0.90 (Figure 3A). To further validate intergroup differences and intragroup sample replication, principal component analysis (PCA) was performed. The PCA plot revealed significant dispersion between the shCoq4 and shNc groups while showing tight clustering within each group (Figure 3B). These results indicate significant intergroup differences, good intragroup sample reproducibility, and high data quality.

Differential gene expression analysis was performed using DESeq2 (threshold for differential gene screening: |log2FoldChange| > 1.0 and Padj < 0.05), and a volcano plot was generated. The results revealed that 1,923 genes were upregulated and 1,344 genes were downregulated in this cell model (Figure 3C). Subsequently, we categorized the top 20 significantly enriched pathways and identified ferroptosis pathways as significantly enriched, clearly classified as an upregulated pathway (Figures 3D,E). Subsequently, we analyzed significantly differentially expressed genes within the ferroptosis-enriched pathways. Compared to the shNc group, four genes were downregulated in the shCoq4 group, including key components of the anti-ferroptosis system Xc⁻: SLC7A11 and SLC3A2. Concurrently, ten genes were upregulated, including ferroptosis-promoting protein ACSL4 (Figure 3F).

These findings suggest at the transcriptomic level that Coq4 deficiency may promote ferroptosis in endothelial cells by modulating the expression of ferroptosis-related genes.

To verify whether Coq4 deficiency indeed induces ferroptosis in vascular endothelial cells, we first conducted in vivo validation. Immunofluorescence results revealed that compared to WT mice, Coq4^−/− mice exhibited significantly elevated expression levels of ACSL4, a key regulator of ferroptosis, in placental endothelial cells (Figure 4A). We then further examined whether ferroptosis also occurred in the human umbilical vein endothelial cell model (shCoq4-HUVECs). First, the impact of Coq4 knockdown on cell viability was assessed. CCK-8 assay results revealed significantly reduced viability in the shCoq4 group compared to the shNc (P < 0.01, Figure 4B). Furthermore, intracellular levels of the primary antioxidant glutathione (GSH) were markedly decreased in the shCoq4 group relative to shNc (P < 0.01, Figure 4C), while lipid peroxidation products malondialdehyde (MDA) (P < 0.001, Figure 4D) and 4-hydroxy-2-nonenal (4-HNE) (P < 0.05, Figure 4E) were significantly elevated. FerroOrange probe detection revealed significant accumulation of Fe²⁺ in shCoq4 cells compared to shNc cells (Figure 4F). At the ultrastructural level, transmission electron microscopy (TEM) examination showed classical ferroptosis morphological features in the shCoq4 group, including reduced mitochondrial cristae, decreased mitochondrial volume, and increased membrane density, compared to the shNc group (Figure 4G). Additionally, Western blotting analysis revealed that compared to the shNc group, the shCoq4 group exhibited significantly upregulated expression of (ACSL4) (P < 0.05) and (FTH1) (P < 0.05). Conversely, expression of SLC7A11 was significantly downregulated (P < 0.001) (Figure 4H).

The above findings collectively reflect the hallmark features of ferroptosis, namely impaired antioxidant defense systems, disrupted iron metabolism, and accumulation of lipid peroxides. Therefore, we conclude that Coq4 deficiency induces ferroptosis in human endothelial cells, consistent with the bioinformatics analysis described above.

In the anti-ferroptosis pathway, GPX4 and FSP1 exert parallel and independent roles (Yang et al., 2025). To investigate the specific mechanism by which Coq4 deficiency affects ferroptosis, we first assessed the protein expression levels of GPX4 and FSP1 in shCoq4-HUVECs. Western blotting revealed that although intracellular GSH levels were significantly reduced in shCoq4 cells compared to shNc cells, GPX4 protein expression showed no significant change, whereas FSP1 protein levels were markedly decreased (P < 0.01, Figure 5A). GSH serves as an essential cofactor for GPX4 enzymatic reactions, and its depletion typically inhibits GPX4 activity (Li et al., 2024; Ursini et al., 2020). However, in this study, GPX4 protein expression did not significantly change with GSH reduction, suggesting that ferroptosis induced by Coq4 knockdown may not rely on GPX4 protein degradation. The specific regulatory mechanism requires further elucidation. Subsequently, we focused on the FSP1 pathway. Given that the classical anti-ferroptosis pathway for FSP1 involves FSP1/CoQ10, and Coq4 is a key factor in CoQ10 biosynthesis, we further assessed CoQ10 levels. Results showed that intracellular CoQ10 levels were significantly reduced in the shCoq4 group compared to the shNc group (P < 0.01, Figure 5B). We therefore hypothesized that Coq4 knockdown may impair CoQ10 synthesis, leading to functional loss or reduced stability of its downstream effector FSP1 due to insufficient “substrate,” thereby weakening this anti-ferroptosis defense pathway parallel to GPX4. Therefore, we chose to restore CoQ10 to test this hypothesis. However, after CoQ10 supplementation, intracellular FSP1 protein levels in the shCoq4 group did not show significant recovery compared to the shNc group (Figure 5C). This indicates that FSP1 downregulation is not primarily caused by insufficient CoQ10 synthesis but is more likely a direct or indirect consequence of Coq4 deficiency.

Further experiments were conducted to assess the synergistic effects of FSP1 and CoQ10 in regulating ferroptosis through combined rescue strategies. First, we constructed an FSP1 overexpression (shCoq4+FSP1-OE) model based on shCoq4-HUVECs via plasmid expression. Western blotting revealed significantly elevated FSP1 protein levels in the shCoq4+FSP1-OE group compared to the empty vector control (shCoq4+Vector) (P < 0.01, Figure 6A), confirming successful model establishment. Subsequently, in the context of Coq4 knockdown, we performed a combined rescue experiment with FSP1 overexpression and CoQ10 supplementation. Western blotting analysis revealed that, compared to the control group (shCoq4+DMF+Vector), both ACSL4 and FTH1 expression were significantly downregulated in the treatment group (shCoq4+Coq10+FSP1-OE) (P < 0.05, Synergy Score > 0, Figure 6B), while SLC7A11 protein levels remained unchanged (Figure 6B). Concurrently, the treated group (shCoq4+CoQ10+FSP1-OE) exhibited significantly reduced levels of malondialdehyde (MDA), a lipid peroxidation product, (P < 0.001, Synergy Score > 0, Figure 6C), and showed significantly improved cell viability (P < 0.001, Synergy Score > 0, Figure 6D).