Abstract
Aims:
Ferroptosis has been implicated in the pathogenesis of lupus nephritis (LN), yet its precise role and mechanisms remain unclear. This study aimed to clarify the role of ferroptosis in LN progression and its underlying mechanisms.
Methods:
Transmission electron microscopy (TEM) was used to assess mitochondrial morphology in renal tissues from LN patients and MRL/lpr mice. Multidimensional mass spectrometry-based shotgun lipidomics was applied to analyze lipid alterations in renal cortex, medulla, and isolated renal tubules. Immunoblotting and reverse transcription quantitative PCR were performed to evaluate ferroptosis-related proteins and their messenger RNAs (mRNAs). Primary renal tubular epithelial cells (RTECs) from the distinct renal regions (cortex/medulla) were isolated and exposed to oxidative stress in vitro. Ferroptosis inducer erastin and inhibitor ferrostatin-1 (Fer-1) were used in vivo to determine causal effects.
Results:
TEM revealed typical ferroptotic mitochondrial changes in renal tissues from both LN patients and lupus-prone mice. In MRL/lpr mice, ferroptosis occurred as early as the pre-LN stage (8 weeks) and worsened by 14 weeks, with cortical tubules showing more severe damage than medullary tubules. Lipidomics demonstrated significant increases in lysophospholipids (e.g., 22:4 lysophosphatidylethanolamine, p < 0.001; 20:4 lysophosphatidylcholine, p < 0.01) and HNE species (p < 0.05), along with reductions in plasmalogens (e.g., 18:1–20:4 plasmenylcholine, p < 0.001). Mechanistically, ferroptosis was driven by downregulation of glutathione peroxidase 4 (p < 0.001) and solute carrier family 7 member 11 (p < 0.01) and upregulation of Acyl-CoA synthetase long chain family member 4 (p < 0.05), consistent with mRNA changes. Functionally, cortical RTECs cultured in vitro exhibited higher lipid reactive oxygen species (p < 0.001) and ferrous ion (Fe2+) accumulation (p < 0.01). In vivo, erastin accelerated LN progression, whereas Fer-1 significantly reduced proteinuria, renal pathology, and inflammatory cytokines.
Innovation and Conclusion:
The study provided direct evidence of ferroptosis markers in renal tissues of LN patients. RTECs exhibited the intrinsic abnormalities that trigger ferroptosis, greatly contributing to the progression of LN. Our findings highlighted the critical role of region-specific tubular ferroptosis in driving renal pathology. Early intervention targeting ferroptosis of RTECs in the renal cortex might be an effective strategy for treating LN. Antioxid. Redox Signal. 44, 197–212.
Introduction
As one of the most severe and prevalent complications of systemic lupus erythematosus (SLE), lupus nephritis (LN) is a major risk factor in overall morbidity and mortality, potentially progressing to end-stage renal disease (Hoi et al., 2024; Schwartz et al., 2014; Siegel and Sammaritano, 2024). Clinically, LN manifestations include persistent proteinuria, progressive renal dysfunction, and heightened cardiovascular risks, such as hypertension and accelerated atherosclerosis. These features greatly complicate disease management (Durcan et al., 2019). The pathological hallmarks of LN include glomerular hypercellularity, immune complex deposits, interstitial inflammation, and tubular damage (Giannakakis and Faraggiana, 2011). While glomerulonephritis is recognized as an initial hallmark of LN, increasing evidence highlights the critical role of tubulointerstitial nephritis in disease progression (Dhingra et al., 2014). It is widely accepted that the underlying pathogenesis of LN involves coordinated innate and adaptive immune dysregulation, immune-mediated inflammation, metabolic disturbances, hypoxia, ischemia, and oxidative stress (Bao et al., 2024; Hong et al., 2020). Furthermore, renal tubular epithelial cells (RTECs), essential for maintaining tubular structure and renal homeostasis, are particularly vulnerable to damage during the disease course, thereby promoting renal fibrosis (Mohan et al., 2023). Such damage can produce a vicious cycle of renal function impairment and serves as a key determinant of long-term outcomes in LN patients. However, the exact pathogenic mechanism(s) remains unclear. Thus, identification of the mechanism(s) underlying tubular epithelial injury for LN would contribute to improving clinical outcomes.
Accumulated evidence has demonstrated that ferroptosis is closely associated with the development of LN (Li et al., 2021; Zhao et al., 2024). Ferroptosis has been identified as a distinct iron-dependent, regulated form of cell death driven by excessive lipid peroxidation, membrane damage, and mitochondrial dysfunction (Dixon et al., 2012). It has been demonstrated that ferroptosis participates in regulating the pathological mechanism in the progression of acute kidney injury to chronic kidney disease (Guo et al., 2023; Zhang et al., 2023). In clinical settings, it has also been commonly observed that iron deposits accumulate in the RTECs of patients with primary glomerulonephritis (van Raaij et al., 2018; Vaziri et al., 2015). Iron carrier proteins, including ferritin, have also been investigated as potential urinary biomarkers of LN (Vanarsa et al., 2012). Animal model experiments further demonstrated that restoring iron homeostasis in the RTECs of lupus-prone mice could significantly ameliorate LN (Scindia et al., 2020). Crucially, no prior study has provided direct molecular evidence of ferroptosis in the renal tubular region of LN patients. Our previous studies revealed that lipid peroxidation, the representative characteristics of ferroptosis, was present in the serum of patients with SLE (Hu et al., 2016; Wang et al., 2023). This is closely associated with dysfunction of immunocytes (Hu et al., 2021b). In addition, lipid peroxidation already existed in the renal tissues of lupus-prone mice at pre-LN state (Hu et al., 2021a). Recently, it has been reported that iron accumulates in the renal tubular segments and ferroptosis signature in whole-kidney specimens of aged MRL/lpr mice (Alli et al., 2023). However, this previous study did not provide direct evidence on ferroptosis occurring in RTECs and did not reveal the role of ferroptosis in the pre-LN state. Existing studies based on transcriptomic analyses and bioinformatics have predicted ferroptosis-related gene signatures in glomeruli or tubulointerstitial compartments. However, these predictions remain correlative and lack confirmation at the protein, lipid, or morphological levels (Wang et al., 2022; Xiao et al., 2025). Therefore, depending on the above findings, we hypothesized that ferroptosis in RTECs plays a crucial role in the pathogenesis of LN and is a key initiating factor in the disease development.
Furthermore, some aspects of the specific renal region in which ferroptosis of RTECs first appears require clarification. Given the distinct spatial characteristics of renal tubules within the kidney, it is crucial to clarify the differential distribution and microenvironmental exposure of RTECs across regions. Importantly, the heterogeneity of RTECs across different renal regions suggests that these cells exhibit varied responses to metabolic and oxidative stress depending on their locations (Doke and Susztak, 2022). RTECs from cortical and medullary tissues experience different microenvironmental conditions and support specific kidney processes. In the renal cortex, proximal and distal tubular cells rely heavily on mitochondrial oxidative phosphorylation, making them more susceptible to reactive oxygen species (ROS)-induced damage under pathological conditions (Hoogstraten et al., 2024). In contrast, medullary tubular cells, including those in the loops of Henle and collecting ducts, are exposed to low oxygen levels and higher metabolic stress, increasing their risk of mitochondrial dysfunction and lipid peroxidation (Gao et al., 2022). Dysfunctional mitochondria in these regions could release ROS, triggering lipid peroxidation and ferroptosis processes. Identification and recognition of ferroptosis in different renal regions could facilitate understanding of disease pathogenesis and developing more effective treatments for LN.
In this study, mitochondrial morphology in renal tissues from both LN patients and the lupus-prone mice was evaluated via transmission electron microscopy (TEM). A multidimensional mass spectrometry-based shotgun lipidomics (MDMS-SL) approach was used to comprehensively analyze the cellular lipid species associated with lipid peroxidation across different renal regions of lupus-prone mice at 8 and 14 weeks of age, respectively. Immunoblotting analysis and reverse transcription quantitative PCR (RT-qPCR) were performed to determine alterations of ferroptosis-related proteins and their messenger RNAs (mRNAs). Primary RTECs isolated from distinct renal tissues were cultured in vitro to further validate these results. Therefore, the results provide spatial and temporal insights into the role of ferroptosis in the development of LN.
Results
Tubular ferroptosis is involved in the progression of LN
To identify the predominant form of tubular injury in LN, tubular regions from cortical tissue from both healthy controls (HCs) and LN patients were subjected to TEM analysis. The clinical features of active LN patients were characterized by positive anti-double-stranded DNA [dsDNA] antibodies, reduced C3 and C4 levels, proteinuria, urinary transferrin, and severe disease flare (Systemic Lupus Erythematosus Disease Activity Index [SLEDAI] >12) (Supplementary Table S1). Notably, compared with HCs, many typical morphological features of ferroptosis, including mitochondrial vacuolization, increased membrane density, and loss of cristae integrity, were significantly present in the tubular regions of renal samples from patients with severe LN (Fig. 1A). Quantitative morphometry showed reduced mitochondrial area and numerical density, with a higher proportion of abnormal mitochondria in LN, compared with the control group (Fig. 1B–D). To clarify associations between clinical characteristics and ferroptosis features, Spearman’s correlation analysis was performed based on these parameters (Fig. 1E). Our data indicated that mitochondrial area and density were strongly negatively correlated with SLEDAI. In contrast, the proportion of abnormal mitochondria correlated positively with urinary transferrin and SLEDAI. These morphological abnormalities are consistent with ferroptotic features and suggest a potential link between ferroptosis and the progression of LN.
The MRL/lpr mouse line, a widely used lupus-prone model, could spontaneously display clinical manifestations of LN around 12 weeks of age. These features include immune complex-mediated glomerulonephritis, proteinuria, increased serum creatinine, and elevated blood urea nitrogen (Toller-Kawahisa et al., 2015). To further reveal the role of ferroptosis in LN progression, MRL/lpr mice treated with/without either ferroptosis inducer erastin or ferroptosis inhibitor Fer-1 were euthanized at 14 weeks of age (Fig. 2A). Administration of ferroptosis inducer erastin could significantly accelerate the progression of LN, characterized by worsened splenomegaly and lymphadenopathy (Fig. 2B–D). The model mice treated with erastin displayed obvious renal damage, characterized by glomerular hypercellularity, mesangial expansion, and interstitial inflammation, IgG and C3 immune complex deposition, and significantly elevated urinary protein levels (Fig. 2E–I). The model group also exhibited higher levels of serum autoantibodies (e.g., antinuclear antibody [ANA] and dsDNA antibody) and pro-inflammatory cytokines (e.g., IL-6 and TNF-α), which were further enhanced by erastin treatment (p < 0.001) (Fig. 2J–M). In contrast, treatment with ferroptosis inhibitor Fer-1 could significantly alleviate the disease severity of MRL/lpr mice, exhibiting less renal damage (e.g., relieved renal pathological changes, IgG antibody deposition, C3 deposition, and urinary protein) and reduced inflammatory cytokines in serum (Fig. 2).
Collectively, these results strongly demonstrated that ferroptosis served a vital role in the development of LN.
Ferroptosis in renal tissue is present as early as the pre-LN state
As shown in Supplementary Figure S1, the MRL/lpr mice at 8 weeks of age did not exhibit higher levels of autoantibodies and pro-inflammatory cytokines in the serum and significant manifestations of LN. These manifestations include obvious renal pathological changes, IgG antibody deposition, C3 deposition, and urinary protein, indicating that the model mice were at the pre-LN state. However, TEM analysis revealed early signs of mitochondrial cristae loss and fragmentation in the tubular regions of the renal cortex of the MRL/lpr mice at 8 weeks of age (Fig. 3A and Supplementary Fig. S2). Quantitative analysis revealed a mild decline in mitochondrial area and density at 8 weeks in the renal cortex, but a significant increase in abnormal mitochondria (Fig. 3B–D).
As previously described, enhanced lipid peroxidation could result in elevated levels of lysophospholipids, including lysophosphatidylcholine (lysoPC) and lysophosphatidylethanolamine (lysoPE). In contrast, it could reduce plasmalogen lipids that serve as endogenous antioxidant reagents, including plasmenylcholine (pPC) and plasmenylethanolamine (pPE), along with increased hydroxyalkenal (HNE) species (Hu et al., 2021a). The MDMS-SL analysis showed that, compared with the control, the amount of lysophospholipids species in renal cortex from the model mice at 8 weeks of age was already increased (e.g., LysoPC, p < 0.05). This was accompanied by a decrease in plasmalogen levels (e.g., pPC, p < 0.05) and a trend toward HNE accumulation. Similar alterations also existed in renal medulla (Fig. 3F). These results clearly demonstrated that lipid peroxidation, one of the most important indices of ferroptosis, was already present in renal tissue of the model mice as early as the pre-LN state. However, the expressions of ferroptosis-related proteins and their mRNA were not significantly changed at this point (Supplementary Fig. S3).
Ferroptosis in the renal tissue of LN displayed prominent regional features
As mentioned above, TEM analysis revealed early signs of ferroptosis in the tubular regions of the renal cortex from the model mice at 8 weeks of age, while the medulla only displayed minor pathological changes (Fig. 3A). Similar alterations were also displayed in the tubular regions from the model mice at 14 weeks of age. Specifically, the tubular regions of the renal cortex had pronounced ferroptosis-related morphological changes. These included marked cristae reduction, fragmentation, degeneration, and increased mitochondrial membrane density. Conversely, mitochondrial changes in the medulla were less severe, exhibiting reduced and fragmented cristae (Fig. 3B–E).
Lipidomics analysis further confirmed the regional characteristics of ferroptosis in renal tissue. Briefly, the lipid peroxidation in renal cortex from the model mice at 8 weeks of age was more significant compared with the medulla (Fig. 3F). Moreover, the lysophospholipid levels in renal cortex from the model mice at 14 weeks of age exhibited further elevations. This was especially true in species containing polyunsaturated fatty acids (PUFAs), such as 22:4 LysoPE (p < 0.001), 18:2 LysoPC (p < 0.01), and 20:4 LysoPC (p < 0.01) (Fig. 3F, G). The levels of plasmalogen were significantly reduced, specifically in species containing PUFAs, including 16:1–20:4 pPE (p < 0.001), 18:2–22:6 pPE (p < 0.05), 18:1–20:4 pPC (p < 0.001), and 18:0–22:6 pPC (p < 0.001) (Fig. 3F, H). Consistently, the accumulation of HNE in the renal cortex was increasingly evident at this stage (p < 0.05) (Fig. 3I). Notably, accumulation of lysoPE, lysoPC, and HNE was also observed in the renal medulla during the peak stage of LN at 14 weeks (p < 0.05, p < 0.01, and p < 0.001, respectively) (Fig. 3F, I). However, the level of plasmalogen in the renal medulla showed no significant reduction (Fig. 3F, H).
Mitochondria are essential for ferroptosis, functioning as the main generators of ROS. Cardiolipin (CL), especially T18:2 CL, deficiency or disruption in its remodeling process can directly impair mitochondrial function (Chicco and Sparagna, 2007). Therefore, CL levels in different renal regions were detected to evaluate the mitochondrial function during LN progression. It was found that the total amount of T18:2 CL in renal cortex from the model group was significantly increased at 8 weeks of age compared with that of the control group (p < 0.05). At 14 weeks of age, T18:2 CL levels in the model group were significantly lower than those of age-matched controls (p < 0.001), suggesting that LN-associated mechanisms may interfere with the physiological maturation of CL (Fig. 3J). Furthermore, the increased total amount of CL species in the model group was mainly attributed to the elevated level of T18:2 CL (p < 0.001) (Fig. 3F, J). However, this change was not observed in the renal medulla.
To further confirm our findings, the expression levels of ferroptosis-related regulatory proteins were assessed in renal cortex and medulla across different disease stages in lupus-prone mice (Stockwell et al., 2017). It is well-known that glutathione peroxidase 4 (GPX4) and solute carrier family 7 member 11 (SLC7A11) are two critical ferroptosis suppressors, acting synergistically to attenuate ferroptosis by reducing phospholipid hydroperoxides (PLOOHs) in membranes. In contrast, Acyl-CoA synthetase long chain family member 4 (ACSL4) promotes PUFA activation, increasing the pool of lipid peroxidation substrates. The protein expression of ACSL4 was significantly upregulated in both regions at 14 weeks of age in MRL/lpr mice (p < 0.05) (Supplementary Fig. S3B). Conversely, the protein levels of SLC7A11 were significantly decreased in both the cortex and medulla (p < 0.05 and p < 0.01, respectively) (Supplementary Fig. S3B). In support, the protein level of GPX4 also showed a decreasing trend in both regions, with a significant decline in mRNA expression at this stage (p < 0.001 and p < 0.01, respectively) (Supplementary Fig. S3B, D).
Collectively, these findings strongly demonstrated that the progression of LN was accompanied by exacerbated ferroptosis across the renal tissue, with pronounced regional characteristics. Specifically, the renal cortex displayed earlier and more obvious mitochondrial structural abnormalities, whereas medullary involvement became evident mainly at the later stage.
Ferroptosis is significantly enhanced in cortical tubular segments of LN
Iron resorption and accumulation primarily occur in the tubular compartment of the kidneys, which are abundant in mitochondria-key organelles involved in ferroptosis (Morel and Scindia, 2024). Combined with the mitochondrial damage observed in RTECs of LN, it was reasonable to suppose that ferroptosis predominantly occurs in this segment of the nephron. To further resolve RTEC-specific lipid alterations that may be masked in whole-tissue analyses, we performed lipidomics analysis on isolated tubular segments from cortex and medulla. Therefore, tubular segments from distinct regions were isolated for further investigation of ferroptosis in this area. Lipidomics analysis confirmed this hypothesis by revealing a distinct lipid profile in the renal tubules. The alterations in relative abundance of lipid classes have been summarized in Figure 4A. Notably, the total amount of T18:2 CL was significantly decreased in both cortical and medullary tubules of the model group at 14 weeks of age (Fig. 4B). Briefly, in the cortical tubules, the model group showed a significant increase in total amount of lysoPE (p < 0.01). Remarkably, the most significantly increased lysoPE species included 16:0 (p < 0.001) and 22:6 (p < 0.01) (Fig. 4C). However, these significant differences in lysoPE species were not observed in the medullary tubules, where only an upward trend was observed (Fig. 4A, C). In addition, the levels of most PC and PE species, including pPC and pPE species, markedly decreased in cortical tubules of the model group (p < 0.05) (Fig. 4A). The most significantly reduced plasmalogen species were those containing PUFAs, such as 20:4 and 22:6 (p < 0.05) (Fig. 4D). In the medullary tubules, there was also a remarkable reduction in the total amount of pPE species (p < 0.05), particularly 16:0–20:4 pPE (p < 0.001) (Fig. 4A, D). However, no significant change was observed in the total amount of pPC, with only 18:1–20:4 and 18:0–22:6 pPC species showing notable differences (p < 0.01) (Fig. 4D).
Furthermore, it was found that the cortical tubular segments in the model group showed significantly elevated protein levels of ACSL4 (p < 0.05), along with decreased protein levels of SLC7A11 (p < 0.05) and GPX4 (p < 0.001) (Fig. 4E). Conversely, in the medullary tubular segments, only the ACSL4 protein expression significantly increased. In contrast, the protein levels of SLC7A11 and GPX4 changed minimally (Fig. 4F). Correspondingly, RT-qPCR analysis revealed an upregulation expression of ACSL4 mRNA (p < 0.01). This was accompanied by a significant downregulation expression of SLC7A11 mRNA (p < 0.05) and GPX4 mRNA (p < 0.001) in cortical tubular segments of the model group (Fig. 4G), further supporting the observed alterations of their proteins. However, only ACSL4 mRNA upregulation was observed in the medullary segments (Fig. 4H). Taken together, these data outlined a structured cascade from lipid alterations to protein dysregulation and transcriptional changes, highlighting the greater susceptibility of cortical tubular segments to ferroptotic injury of LN.
Inherently impaired ROS defense of RTECs initially triggered ferroptosis in LN
To further reveal the microenvironmental or intrinsic factors that initially caused the ferroptosis of RTECs, primary RTECs were isolated from different renal regions of the control and the model mice. These RTECs were examined under both basal and stressed conditions in vitro to determine the regional capacity for eliminating lipid peroxidation. The isolated RTECs were identified using flow cytometry based on the specific expression of cytokeratin-18 in the cytoplasm, resulting in a purity of ∼ 72% (Supplementary Fig. S4). Specifically, the stress condition was induced by lipid ROS generator tert-butyl hydroperoxide (TBHP) (Wenz et al., 2018). Using BODIPY 581/591-C11 as a lipid peroxidation reporter, confocal microscopy revealed that cortex-derived RTECs from the model mice exhibited significantly higher baseline levels of lipid peroxidation (green fluorescence) than the control mice (p < 0.05). Furthermore, the difference became more apparent following TBHP exposure (p < 0.001) (Fig. 5A, C). Although there were no remarkable baseline differences in medulla-derived RTECs from the control and model mice, lipid peroxidation of the model was significantly elevated after TBHP exposure in comparison with that of the control (p < 0.05) (Fig. 5B, D). Flow cytometry analysis also revealed that the total intracellular ROS levels were significantly higher in both cortex- and medulla-derived RTECs from model mice compared with the control mice (p < 0.001) (Fig. 5E, F).
Bioactive iron serves a crucial role in the formation of lipid peroxides and initiation of ferroptosis via the Fenton reaction (Shan et al., 2023). Under basal conditions, baseline levels of labile ferrous ion (Fe2+) were significantly higher in cortex-derived RTECs from the model mice in comparison with that of the control mice (p < 0.01), as detected by FerroOrange staining. Exposure to TBHP further elevated labile Fe2+ in cortex-derived RTECs from MRL/lpr mice (p < 0.05) (Fig. 5G). Meanwhile, no significant differences in labile iron were observed in medulla-derived RTECs between the two strains (p > 0.05) (Fig. 5H).
Immunoblotting analysis revealed that TBHP treatment exacerbated the dysregulation of ferroptosis-related proteins in cortex-derived RTECs from MRL/lpr mice, suppressing the expression of GPX4 and SLC7A11 while enhancing ACSL4 levels (Fig. 5I). Notably, these changes were more pronounced in RTECs from MRL/lpr mice compared with the control, consistent with their inherently impaired antioxidant defense capacity. This enhanced sensitivity underscores the intrinsic vulnerability of lupus-prone RTECs to ferroptotic injury under oxidative stress conditions.
Collectively, these results strongly supported that impaired lipid ROS defense driven by the inherent dysregulation of antioxidant proteins in cortex-derived RTECs led to ferroptosis and contributed to the exacerbation of tubular damage of LN.
Discussion
This study confirmed that ferroptosis in renal tubules was present in LN and revealed that it occurred as early as the pre-LN stage, greatly contributing to the development of LN. Although many studies have indicated that ferroptosis is involved in the progression of acute kidney injury to chronic kidney disease (Guo et al., 2023; Zhang et al., 2023), this study provided direct evidence that ferroptosis existed in tubular regions of patients with LN as its morphological features were clearly displayed in TEM image (Fig. 1A). Functionally, mitochondrial vacuolization and cristae disruption in RTECs can impair tubular reabsorption and epithelial integrity, leading to proteinuria and progressive renal dysfunction. Moreover, ferroptosis releases lipid peroxidation products that trigger innate immune pathways and autoantigen presentation, thereby linking structural mitochondrial injury to the active disease activity observed in LN patients. The relevance of podocyte ferroptosis in LN has been discussed (Liu et al., 2024, 2025). However, the pathophysiological role of tubular cells has received insufficient attention. The early signs of mitochondrial damage in the cortical tubular regions of the model mice at 8 weeks of age were revealed by TEM analysis. Lipidomics analysis also demonstrated the remarkable lipid peroxidation with marked increases in lysophospholipids and decreases in plasmalogen levels in the renal cortex, collectively suggesting that ferroptosis in renal tissue already existed at the pre-LN stage (Fig. 3). A slight dysregulation of ferroptosis-related proteins (e.g., GPX4, SLC7A11, and ACSL4) was observed at this stage. This may result from the presence of other structural components in the kidney, obscuring the early sensitivity of ferroptosis detection at the protein and gene expression levels (Supplementary Fig. S3).
Furthermore, administration of the ferroptosis inducer exacerbated the severity of LN, while the ferroptosis inhibitor effectively mitigated these alterations. The severity of ferroptosis was associated with worsened renal pathology, elevated antibody levels, and increased pro-inflammatory cytokines, including IL-6 and TNF-α. Therefore, ferroptosis drives tissue damage and amplifies immune dysregulation. Lipid peroxidation products associated with ferroptosis disrupt the integrity of RTECs, leading to cell death and damage-associated molecular patterns release, which trigger innate immune pathways and inflammasome activation (Murao et al., 2021). Furthermore, ferroptotic debris enhances autoantigen presentation, which further exacerbates immune dysfunction. This cascade of events exacerbates immune complex deposition and local inflammation, thereby accelerating the progression of LN. From a broader perspective, ferroptosis involves both metabolic and oxidative stress features in RTECs. Our lipidomic analysis revealed PUFA enrichment and plasmalogen depletion, indicating lipid metabolism remodeling that drove ferroptosis. Meanwhile, mitochondrial damage and high ROS levels reflected oxidative stress that worsened ferroptotic injury. Taken together, these results indicated that ferroptosis integrated immune activation, metabolic remodeling, and oxidative stress as part of its pathological signature in LN. Unlike prior investigations relying on predictive gene expression models (Wang et al., 2022; Xiao et al., 2025) or ferroptosis signatures in whole-kidney tissue of lupus-prone mice (Alli et al., 2023), this study directly visualized ferroptotic mitochondrial features in patient tissues and lupus-prone mice. It systematically profiled lipidomic signatures and functional vulnerability of RTECs across renal regions. These findings establish our study as the first to reveal RTEC ferroptosis as an initiating pathomechanism, rather than a secondary effect of inflammation or glomerular damage.
In addition, ferroptosis in RTECs of LN showed significant regional features. Renal tubules are susceptible to hypoxia, proteinuria, toxins, metabolic disorders, and senescence (Chen et al., 2023; Xu et al., 2019; Yao et al., 2022). Ferroptosis in the renal cortex and medulla was different due to the distinct physiological and metabolic characteristics between these regions. The application of MDMS-SL in this study enables high-sensitivity detection of lipid peroxidation signatures across renal regions. It provides comprehensive coverage of lipid species, including low-abundance molecules that are critical to ferroptosis (Hu et al., 2017). The renal cortex showed more significant abnormalities in lipid peroxidation and dysregulated expression of key ferroptosis-related enzymes (Fig. 3 and Supplementary Fig. S3). Ultrastructural analysis via TEM revealed remarkable mitochondrial morphological changes in the cortical RTECs (Fig. 3A–E). Moreover, renal tubules isolated from the cortex in vitro exhibited a more extensive landscape of ferroptosis at 14 weeks of age. Cortex-derived RTECs displayed higher levels of labile Fe2+. While medulla-derived RTECs did not show such iron overload under basal conditions, they became more vulnerable to oxidative stress when challenged with external ROS inducers. These differences likely reflect the intrinsic biological complexity of cortical versus medullary tubules. Cortical RTECs rely heavily on oxidative phosphorylation and face high metabolic ROS loads, whereas medullary tubules function in relatively hypoxic conditions with distinct transport demands, conferring differential thresholds for ferroptosis initiation. In contrast, the medulla-derived RTECs were initially more resistant to ferroptosis (Figs. 4 and 5). Collectively, the results supported that cortex-derived RTECs were more susceptible to ferroptotic injury due to a combination of impaired lipid ROS defense and iron overload.
Mechanistically, the abnormalities of RTECs, especially in the renal cortex, initially caused ferroptosis in LN. The intrinsic vulnerability of RTECs to ferroptosis in MRL/lpr mice was evident in vitro, where they exhibited a markedly higher ferroptotic tendency than the control mice (Fig. 5). This heightened susceptibility persisted despite the removal of systemic inflammatory and microenvironmental factors, suggesting that ferroptosis in LN was not merely a secondary consequence of immune dysregulation but rather an inherent defect within RTECs independently. As key regulators of tubular integrity and function, RTECs are fundamental initiators of oxidative stress-induced renal damage in LN. This finding shifted the perspective from viewing ferroptosis as an outcome of inflammatory injury to recognizing RTECs as an intrinsic ferroptotic trigger and a critical therapeutic target in lupus-associated renal pathology. Further investigation revealed that the susceptibility of RTECs to ferroptosis in LN was attributed to the disruption of antioxidant defenses. This study of RTECs in vitro was conducted to demonstrate that the dysregulation of ferroptosis-related proteins significantly impaired detoxification mechanisms. Key regulators of ferroptosis, such as SLC7A11, GPX4, and ACSL4, were disrupted in MRL/lpr mice. In addition, it should be emphasized that GPX4 represents a central component of the cellular antioxidant system in the context of ferroptosis-specific defenses. The cystine-glutamate antiporter subunit SLC7A11 provides cystine for glutathione synthesis, which in turn supports GPX4 activity to mitigate lipid peroxidation by reducing PLOOHs and inhibiting membrane damage. The conversion of PLOOHs to nontoxic phospholipid alcohols by GPX4 limits membrane damage and blocks the progression of lipid peroxidation (Yang et al., 2014). However, reduced GPX4 activity in RTECs impaired the critical defense, exacerbating oxidative damage. Furthermore, ACSL4 activates PUFAs to acyl-CoA for insertion into phospholipids by LysoPC acyltransferase 3 (Doll et al., 2017). This lipid remodeling altered membrane composition, increasing susceptibility to ferroptosis-related damage in RTECs. Upon exposure to the external inducer, RTECs from MRL/lpr mice had a compromised ability to eliminate lipid peroxidation, with increased levels of lipid ROS, intracellular ROS, and labile Fe2+ levels. Two main mechanisms have been shown to induce cellular ferroptosis: excessive oxidative modification of PUFAs and the inhibition of GPX4 (Stockwell et al., 2017). Thus, strategies that restore GPX4 activity or inhibit ACSL4 of RTECs in the renal cortex might provide therapeutic benefits to alleviate oxidative stress and renal damage in LN.
One of the limitations of this study is that lipidomic analysis was performed in renal cortical and medullary tissues as well as in isolated renal tubules enriched in RTECs, rather than in purified cultured RTECs. Although tubule-enriched samples provide a reasonable approximation of the in vivo lipid metabolic state of RTECs, they may not fully capture the lipidomic profile of individual tubular epithelial cells. Future studies integrating lipidomic profiling of purified RTECs will help clarify cell-specific metabolic alterations. Another limitation is that different techniques were applied in patient samples, animal models, and cultured cells due to differences in tissue availability and experimental feasibility. While this heterogeneity may limit direct comparability across models, the integrated approaches provided complementary evidence consistently supporting the role of ferroptosis in LN. Future studies will aim to integrate multi-omics analyses in larger cohorts of patient samples to provide more comprehensive validation. In addition, regarding the pharmacological modulation experiments, it should be noted that although erastin and Fer-1 are widely used as canonical ferroptosis modulator compounds, potential off-target effects cannot be completely excluded. Nonetheless, our complementary in vitro experiments using isolated RTECs exposed to oxidative stress confirmed the ferroptosis-specific alterations, thereby strengthening the causal link between ferroptosis and tubular injury in LN.
In conclusion, this study provides direct evidence of ferroptosis in RTECs from LN patients and lupus-prone mice, revealing it as an early and region-specific pathological event rather than a secondary consequence of immune injury. By integrating ultrastructural, molecular, and lipidomic analyses, we establish RTEC ferroptosis, particularly in the renal cortex, as a key driver of LN progression. These findings shift the current paradigm from glomerulus-centered mechanisms to tubular pathogenesis and highlight ferroptosis as a promising therapeutic target. Early intervention in RTEC ferroptosis may offer novel and more precise strategies for LN treatment beyond traditional immunosuppression.
Materials and Methods
Patients and healthy controls
SLE patients enrolled in this study met the 2019 European League Against Rheumatism/American College of Rheumatology classification criteria (Aringer et al., 2019). Disease activity was evaluated using the modified SLEDAI-2000 index (Uribe et al., 2004). Renal tissue samples were obtained from newly diagnosed and untreated female patients with active LN, defined by a SLEDAI score >12 with typical serological and urinary abnormalities. HCs were age- and gender-matched individuals without autoimmune, inflammatory, or infectious diseases. The study was conducted in accordance with the principles of the Declaration of Helsinki and approved by the Medical Ethics Committee of Zhejiang Chinese Medical University (No. 2024-KL-1136–1). Written informed consent was obtained from all of the participants.
Animal experiments
Female MRL/MpJ and MRL/lpr mice (6 weeks old) were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. and housed under specific pathogen-free conditions at the Zhejiang Chinese Medical University laboratory animal research center.
A total of 20 female MRL/lpr mice were randomly assigned to experimental groups. Five were euthanized at 8 weeks of age as the untreated 8-week model group. The remaining 15 mice were euthanized at 14 weeks of age and divided into three groups: the untreated 14-week model group, the ferroptosis inducer group, and the ferroptosis inhibitor group. Mice in the ferroptosis inducer group received erastin by intraperitoneal injection (2 mg/kg every other day from 6 to 14 weeks of age), which inhibits system Xc− and impairs GPX4 activity to trigger ferroptosis. Mice in the ferroptosis inhibitor group received Ferrostatin-1 (Fer-1) by oral gavage (2 mg/kg every other day for the same duration), a radical-trapping antioxidant that blocks lipid peroxidation.
In parallel, 10 female MRL/MpJ mice served as the untreated control group, with 5 euthanized at each time point. All animal experiments were approved by the Ethics Committee of Zhejiang Chinese Medical University (No. IACUC-20211101–04). Mice were housed in cages under controlled conditions (22°C–25°C and relative humidity of 60%–70%) with ad libitum access to food and water.
Histopathological evaluation
Renal tissues from the right kidneys were fixed in 10% formaldehyde for 48 h, paraffin-embedded, and sectioned. A subset of sections underwent hematoxylin and eosin. Pathological scoring utilized previous methods (Ou et al., 2024), ranging from 0 to 4, with scores per section calculated as the total score divided by the number of glomeruli observed. For immunofluorescence analysis, sections were incubated overnight at 4°C with Alexa Fluor 488 goat anti-mouse IgG (1:1000 dilution, #ab150113, Abcam), mounted with 4′,6-diamidino-2-phenylindole (DAPI), and imaged using a virtual slide microscope (OLYMPUS, Japan). Mean fluorescence intensity (MFI) was analyzed using ImageJ 2.1.0 software. For immunohistochemical analysis, sections were incubated overnight at 4°C with anti-C3 antibody (1:2000 dilution, #ab200999, Abcam), followed by incubation with goat anti-rabbit horseradish peroxidase-conjugated immunoglobulin G (IgG-HRP) for 30 min at 37°C. Visualization utilized DAB substrate solution, and sections were digitally scanned with a NanoZoomer digital pathology system (Hamamatsu, Japan).
Assessment of disease severity
Fresh urine samples from various experimental groups were collected using metabolic cages and centrifuged at 2000 rpm for 5 min at 4°C. The supernatant was then analyzed for protein concentration using an automatic biochemical analyzer (TOSHIBA TBA-120FR). Upon euthanasia, the spleen, bilateral axillary lymph nodes, bilateral inguinal lymph nodes, and body weights of mice were recorded. The spleen or lymph node index was calculated by dividing the fresh weight of the organ (combined axillary and inguinal lymph nodes for the lymph node index) by the body weight of the mouse.
Enzyme-linked immunosorbent assay
Serum samples from different experimental groups of mice were analyzed for autoantibodies (i.e., ANA and dsDNA) and cytokines (i.e., IL-6 and TNF-α) using detection kits (Jingmei Biotechnology Co., Jiangsu, China). The quantification of cytokines was conducted according to the manufacturer’s instructions, and the concentrations were determined based on standard curves.
Transmission electron microscopy
Fresh renal samples from the left kidneys of mice and renal cortical tissue samples from patients were collected. All specimens, each no larger than 1 mm3, were promptly fixed in 2.5% glutaraldehyde at 4°C overnight. The remaining portion of the left kidney of mice was snap-frozen in liquid nitrogen and stored at −80°C for subsequent molecular and biochemical analyses. After thorough rinsing, the samples underwent fixation for 1 h in a solution containing 1% osmium tetroxide and 1.5% potassium ferricyanide at 4°C. They were then treated with tannic acid chromate at room temperature for 30 min and 1% osmium tetroxide at 4°C for 1 h. After a final rinse, the samples were immersed overnight in 2% uranyl acetate at 4°C. Subsequently, the fixed samples were dehydrated using an ethanol series, embedded, sectioned via ultramicrotomy, and mounted onto copper grids. TEM was at magnifications ranging from 20,000× to 40,000× using equipment from Hitachi, Japan. The percentage of mitochondrial area is relative to the total image area, the mitochondrial density (number of mitochondria per μm2), and the proportion of abnormal mitochondria, defined as mitochondria exhibiting swelling, disrupted cristae, or loss of membrane integrity (Gao et al., 2025).
Isolation of tubules from renal tissue
Renal tubules were isolated according to the previous report (Yang et al., 2022). The renal cortex and medulla were dissected, sliced into 1 mm pieces, and digested in Dulbecco’s Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F-12) containing 1 mg/mL Collagenase Type I (#17018029, Gibco) at 37°C for 30 min. The enzyme reaction was terminated by adding fetal bovine serum (FBS) to a final concentration of 5% (v/v), and then the tubule suspension was sequentially filtered through 40- and 70-μm cell meshes (#352340/352350, Falcon). The retained tubules on the 40-μm cell meshes were resuspended in DMEM/F-12 and centrifuged at 1500 rpm for 5 min. The supernatant was discarded, and 1 mL erythrocyte lysate (#NH4CL2009, TBD) was added. Next, the solution was centrifuged at 1500 rpm for 5 min. Eventually, protein, RNA, and lipidomes were extracted from the sediment.
Lipidomics analysis
For lipidomic analysis, renal tissue samples from distinct regions (cortex and medulla) and renal tubules isolated as described above were used. Renal tissue samples were snap-frozen in liquid nitrogen immediately after dissection and stored at −80°C until use. As previously described (Bligh and Dyer, 1959), lipids were extracted from renal samples using a modified Bligh and Dyer method, incorporating an internal standard. The lipid extracts were resuspended in 2 mL of chloroform/methanol (1:1, v/v)/mg protein and stored at −20°C for subsequent analysis. Derivatization of the primary amine in phosphoethanolamine-containing species with fluorenylmethoxycarbonyl chloride and of HNE species with carnosine was performed using established methods (Wang et al., 2012).
A triple-quadrupole mass spectrometer (Thermo TSQ Quantiva) equipped with an automated nanospray ion source (TriVersa NanoMate, Advion Bioscience Ltd.) was utilized in this study. The instrument was operated under Xcalibur system software, following previously reported protocols (Han et al., 2008). Lipid identification and quantification across various classes and species were performed using MDMS-SL as described previously. Mass spectrometry data were analyzed using a custom macro program within Microsoft Excel (Yang et al., 2009).
Immunoblotting and antibodies
Renal tissues were homogenized in RIPA lysis buffer (#P0013B, Beyotime) supplemented with protease inhibitors (#P1005, Beyotime) and phosphatase inhibitors (#CW2383S, CWBIO), then lysed for 30 min on ice. The lysate was isolated through centrifugation at 14,000 g at 4°C for 10 min, and the protein concentration was determined using a BCA Protein Assay Kit (#23225, Thermo Fisher Scientific). Subsequently, the lysate was added with a proper volume of 5× loading buffer and then boiled at 100°C for 10 min. Samples were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on the 12.5% gel and transferred to polyvinylidene fluoride (PVDF) membranes (#IPVH00010, Millipore). The membranes were blocked for 1.5 h at room temperature in TBST (Tris-buffered saline with 0.1% Tween-20) containing 5% skim milk to minimize nonspecific binding. They were then incubated overnight with primary antibodies at 4°C, followed by incubation with the secondary antibodies. Western blot results were scanned and analyzed using Odyssey CLx. Gel quantification was conducted with ImageJ 2.1.0 software.
Commercially available antibodies were used as follows: ACSL4 (1:1000 dilution; #ab15528, Abcam), SLC7A11 (1:1000 dilution; #ab175186, Abcam), GPX4 (1:2000 dilution; #ab125066, Abcam), GAPDH (1:1000 dilution; #5174S, Cell Signaling Technology), β-Actin (1:1000 dilution; #9145S, Cell Signaling Technology), and IRDye 800CW Goat anti-Rabbit IgG (1:5000 dilution; #926-32211, LI-COR Biosciences).
RNA extraction and RT-qPCR analysis
Total mRNA was extracted using Tissue/Cell Fast RNA Extraction Kit (#RK30120, ABclonal) and then transcribed into cDNA using Eastep® RT Master Mix (#LS2050, Promega) according to the manufacturer’s instructions. Quantitative PCR was conducted in a 20 μL reaction mixture using Eastep qPCR Master Mix (#LS2062, Promega). Subsequently, the PCR program was run as follows: 95°C, 2 min; 40 cycles (95°C, 15 s; 60°C, 50 s for each cycle). All data were normalized to the housekeeping gene β-Actin. The primers used in this study were listed as follows: ACSL4-F: CTCACCATTATATTGCTGCCTGT; ACSL4-R: TCTCTTTGCCATAGCGTTTTTCT; SLC7A11-F: GATGCTGTGCTTGGTCTTGA; SLC7A11-R: GCCTACCATGAGCAGCTTTC; GPX4-F: CAGGAGCCAGGAAGTAAT; GPX4-R: CAGCCGTTCTTATCAATGAG; Actin-F: GGCTGTATTCCCCTCCATCG; Actin-R: CCAGTTGGTAACAATGCCATGT.
Culture of primary RTECs
Cultures of primary RTECs from renal tissues of the mice at 8 weeks of age were established following the reported method (Yang et al., 2022). Briefly, tubules were resuspended in DMEM/F-12 culture media (#C11330500BT, Gibco) supplemented with 10% FBS, 1% penicillin–streptomycin solution, 1% insulin-transferrin-selenium-A (#51300044, Invitrogen), 20 ng/mL recombinant human epidermal growth factor (#AF-100-15-100, PeproTech), and 50 nmol/L hydrocortisone (#S31360, Shanghai Yuanye Bio-Technology Co.). Cells were cultured for 4 days in a humidified atmosphere with 5% carbon dioxide (CO2) at 37°C. The quality of isolated RTECs was confirmed by flow cytometry using FITC-anti-Cytokeratin 18 antibody (#MA110326, Thermo Fisher Scientific) (Mattila et al., 1989).
Lipid peroxidation measurement
For confocal microscopy imaging of lipid peroxides, cells were seeded in confocal dishes with a glass bottom. Following treatment with either vehicle or 1 mM tert-butyl hydroperoxide (TBHP) (#458139, Sigma) for 1 h, cells were washed with PBS for three times and then incubated in culture media containing 2 μM BODIPY 581/591-C11 dye (#D3861, Thermo Fisher Scientific) in the dark at 37°C for 20 min. After another PBS wash, images were acquired using a confocal laser scanning microscope (#LSM880, Zeiss). Excitation at 488 nm detected oxidized C11 BODIPY, whereas excitation at 543 nm detected the nonoxidized.
ROS measurement
Cells were seeded in 12-well plates and subjected to respective treatments. Following treatment, cells were detached using trypsin digestion (#25200056, Gibco) and suspended in culture media containing 10 μM 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) (#S0033S-1, Beyotime) in the dark at 37°C for 20 min. Subsequently, cells were washed twice with staining buffer (#420201, Biolegend), resuspended in buffer, and analyzed by flow cytometry (#CytoFlex S, Beckman). A minimum of 20,000 events was recorded per sample, with results shown as the MFI using FlowJo 10.8.1 software.
Intracellular Fe2+ determination
Following the above description in the section on ROS measurement, cells were harvested and stained with 1 μM FerroOrange probe (#F374, Dojindo) in the dark at 37°C for 20 min after digestion. Subsequently, cells were washed with staining buffer and subjected to analysis using a flow cytometer with a 561 nm laser for excitation. A minimum of 20,000 events was recorded per sample, and the relative MFI was calculated using FlowJo 10.8.1 and presented in the results.
Statistical analysis
All data were expressed as mean ± standard error of the mean (SEM). Statistical analysis was performed with IBM SPSS V.26.0 software. For comparison of two groups significance was determined using a two-tailed, unpaired Student’s t-test. For comparison of more than two groups, a one- or two-way analysis of variance (ANOVA) was performed as appropriate with multiple comparisons performed using Fisher’s least squares difference test. Two-sided p-values < 0.05 were considered to be statistically significant. p-Values and sample size can be found in the main and supplementary figure legends. An electronic laboratory notebook was not used.
Authors’ Contributions
W.J.: Writing, methodology, investigation, data curation, and conceptualization. H.C.: Investigation, data curation, and validation. S.Z.: Visualization, investigation, and formal analysis. X.X.: Visualization and investigation. P.Z.: Visualization and software. X.H.: Resources. G.X.: Resources. Y.D.: Funding acquisition. C.W.: Resources. L.L.: Editing, supervision, funding acquisition, and conceptualization. C.H.: Writing, project administration, methodology, funding acquisition, and conceptualization.
Footnotes
Acknowledgments
The authors thank all the study participants for their support.
Author Disclosure Statement
The authors declare that they have no conflicts that could have appeared to influence the work reported in this article.
Funding Information
This work was partially supported by National Natural Science Foundation of China (No. 82274230 and 82104849), Natural Science Foundation of Zhejiang Province of China (No. LQ23H290002), and Research Project of Zhejiang Chinese Medical University (No. 2024JKZKTS02).
Data Availability Statement
The data used to support the findings of this study are available from the corresponding author upon request. All the lipids identified by the lipidomics analysis are provided in Supplementary Data S2. Source data are included with this article, and uncropped Western blot images with size marker indication are presented in 
.
Supplemental Material
Supplemental Material
Supplemental Material
Abbreviations
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
