Bioinformatics analysis of potential key ferroptosis-related genes involved in tubulointerstitial injury in patients with diabetic nephropathy

Abstract

Diabetic nephropathy (DN) is the primary complication of diabetes mellitus. Ferroptosis is a form of cell death that plays an important role in DN tubulointerstitial injury, but the specific molecular mechanism remains unclear. Here, we downloaded the DN tubulointerstitial datasets GSE104954 and GSE30529 from the Gene Expression Omnibus database. We examined the differentially expressed genes (DEGs) between DN patients and healthy controls, and 36 ferroptosis-related DEGs were selected. Pathway-enrichment analyses showed that many of these genes are involved in metabolic pathways, phosphoinositide 3-kinase/Akt signaling, and hypoxia-inducible factor-1 signaling. Ten of the 36 ferroptosis-related DEGs (CD44, PTEN, CDKN1A, DPP4, DUSP1, CYBB, DDIT3, ALOX5, VEGFA, and NCF2) were identified as key genes. Expression patterns for six of these (CD44, PTEN, DDIT3, ALOX5, VEGFA, and NCF2) were validated in the GSE30529 dataset. Nephroseq data indicated that the mRNA expression levels of CD44, PTEN, ALOX5, and NCF2 were negatively correlated with the glomerular filtration rate (GFR), while VEGFA and DDIT3 mRNA expression levels were positively correlated with GFR. Immune infiltration analysis demonstrated altered immunity in DN patients. Real-time quantitative PCR (qPCR) analysis showed that ALOX5, PTEN, and NCF2 mRNA levels were significantly upregulated in high-glucose-treated human proximal tubular (HK-2) cells, while DDIT3 and VEGFA mRNA levels were significantly downregulated. Immunohistochemistry analysis of human renal biopsies showed positive staining for ALOX5 and NCF2 protein in DN samples but not the controls. These key genes may be involved in the molecular mechanisms underlying ferroptosis in patients with DN, potentially through specific metabolic pathways and immune/inflammatory mechanisms.

Keywords:

• Diabetic nephropathy
• tubulointerstitial injury
• ferroptosis
• key gene
• key pathways

Introduction

Diabetic nephropathy (DN) is the main microvascular complication of diabetes mellitus and one of the major causes of mortality in individuals with chronic kidney disease [1]. However, the precise mechanism of DN is complicated and remains to be elucidated. Renal tubulointerstitial lesions are crucial in the development of DN, and proximal tubulopathy has been recognized as an important driver of DN [2].

Ferroptosis is a form of controlled cell death that is induced™ by iron-dependent lipid peroxide accumulation and redox imbalance [3]. Ferroptosis plays an important role in various neoplastic diseases, such as renal cell carcinoma and pancreatic adenocarcinoma [4,5]. The mRNA expression levels of system SLC7A11(Solute carrier family 7 member) and GPX4(Glutathione peroxidase 4) were decreased, while reactive oxygen species (ROS) production and lipid oxidation were enhanced in renal tubular cells in DN [3], indicating that renal tubular cell death in DN may be attributable to ferroptosis [6,7]. Heme oxygenase-1-induced ferroptosis can cause renal tubular injury in db/db mice by degrading heme to accumulate iron [8]. Knockdown of salusin-β inhibited high glucose (HG)-triggered ferroptosis in HK‑2 cells, while overexpressing salusin-β had the opposite effect [9]. Upregulation of Nrf2 by fenofibrate could inhibit diabetes-related ferroptosis, further delaying DN progression [10].

Despite these findings, the role of ferroptosis in DN pathogenesis remains unclear, and detailed information on the relevant molecular mechanisms is still lacking. Exploration of ferroptosis-related variations at the genomic level may help elucidate the pathogenesis and genetic etiology of DN-related ferroptosis, and thus provide new accurate and efficacious treatment directions.

Bioinformatic methods have been extensively adopted for analyzing microarray data to identify biological variations at the genomic level. However, to the best of our knowledge, there have been almost no bioinformatics-based studies on the role of ferroptosis-associated genes in DN-related tubulointerstitial injury. In the present study, we identified differentially expressed genes (DEGs) in tubulointerstitial tissues of DN and screened the key ferroptosis-related genes. We further investigated the correlations between the expression levels of key genes and the glomerular filtration rate (GFR). High glucose was previously shown to induce ferroptosis in HK-2 cells [9]. Therefore, gene expression was verified using real-time quantitative PCR (qPCR) analysis of HK-2 cells and immunohistochemistry (IHC) analysis of human renal biopsies. Because excessive or deficient ferroptosis may be caused by a dysregulated immune response [11], we also explored the association between ferroptosis and immunity in DN. The primary aims of this study were to broaden our understanding of ferroptosis-related mechanisms in DN tubulointerstitial injury and to identify specific and novel candidate biomarkers with potential diagnostic or therapeutic value.

Materials and methods

Data source and identification of ferroptosis-related DEGs

The overall study design is presented in Figure 1. The microarray datasets GSE30529 [12] and GSE104954 [13] (Table 1) from DN tubulointerstitial areas were downloaded from the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/), which is an international, public, functional genomics data repository of next-generation sequencing information. All data were normalized via log2-scale transformation to ensure standardization. The downloaded annotation information from GPL (GEO platforms) was used to convert the probe ID to the gene symbol in the gene expression data. For genes corresponding to several probes, the average expression value of the probes was calculated as the expression level of the gene. All analyses were performed using R software (version 4.1.2). A total of 388 ferroptosis-related genes, including drivers, suppressors, and markers, were obtained from the FerrDb database [14].

Figure 1. Workflow of the overall study design and bioinformatics analysis.

Table 1. Datasets of transcriptional profiles of human interstitial tubules included in the analysis.

Reference	GEO Dataset	DN, n	Control, n	Platform
Woroniecka et al. [12]	GSE30529	10	12	GPL571 Affymetrix Human Genome U133A 2.0 Array
Grayson et al. [13]	GSE104954	7	18	GPL22945 Affymetrix Human Genome U133 Plus 2.0 Array

The Limma package (version 3.50.0) was applied for differential expression analyses in R software (version 4.1.2). To screen DEGs between kidney tissues from DN patients and controls, the cutoff conditions were set to an adjusted p-value < 0.05 and absolute log-fold change |log2FC| ≥ 0.5. DEGs were visualized using the pheatmap package (Version 1.0.12), and ferroptosis-related DEGs were obtained by examining the intersection of the ferroptosis-related genes and DEGs.

Functional and pathway enrichment analyses

Gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Reactome pathway analyses were performed using DAVID (2021 update) (https://david.ncifcrf.gov/) [15], which offers functional annotation of a large number of genes from a variety of genomic resources. p < 0.05 was considered as the cutoff value. GO analysis [16], which includes biological process (BP), cellular component (CC), and molecular function (MF), is a bioinformatics method for annotating genes and proteins to identify characteristic biological attributes. Both the KEGG and Reactome pathway databases [17,18] include a variety of biochemical pathways and provide resources for understanding the advanced functions and utilities of a biological system.

Construction of a protein-protein interaction (PPI) network and identification of key genes

The FerrDb database includes both human and mouse genes. After excluding the genes from mice, the STRING database (version 11.5; https://cn.string-db.org/, combined score > 0.4 considered significant) [19] was utilized to predict potential interactions between gene candidates using laboratory data, other databases, text mining, and predictive bioinformatics data. Cytoscape software [20] was utilized to perform network analysis using the interaction information downloaded from the STRING database. The Cytoscape plug-in Cytohubba [21] was used to identify key genes by intersecting the results from the five CytoHubba algorithms, MCC, MNC, DMNC, EPC, and Clustering Coefficient [22].

Key gene cross-validation and efficacy evaluation

We cross-verified the expression profiles and levels of the identified key genes in normal and DN tissues using another array dataset, GSE30529, downloaded from GEO. This dataset contains the expression data of human tubulointerstitial samples from 10 DN patients and 12 controls.

Clinical data validation and diagnostic efficacy of key genes in DN evaluation

Information on key genes and clinical features was downloaded from the Nephroseq online open-access platform (http://v5.nephroseq.org) [23] to verify the potential functions of key genes in DN. The mRNA expression levels of key genes were compared between DN samples and controls, and relationships between mRNA expression levels of key genes and GFR were determined by Pearson’s correlation analysis. Receiver operating characteristic (ROC) curves were used to evaluate the diagnostic efficacy of key genes in DN.

Immune cell infiltration in DN interstitial tubules

Immuno-Oncology Biological Research (IOBR) is a computational tool for immunobiology research [24]. The immune infiltrating cell score was calculated for each sample from the GSE104954 and GSE30529 matrix using the R package IOBR, then selecting the CIBERSORT method to distinguish 22 human hematopoietic cell phenotypes [25]. In addition, we evaluated the relationships of immune cells with the expression levels of key genes, as well as the relationships of immune cells with GFR.

Cell culture and treatment

HK-2 cells were obtained from the American Type Culture Collection (Cat. No. CRL-2190) and were cultured in DMEM/F-12 medium containing 10% fetal bovine serum (FBS), streptomycin (100 mg/mL), and penicillin (100 U/mL) under a humidified atmosphere of 5% CO₂ at 37 °C. The HK-2 cells were divided into control and HG-treated groups. Cells in the HG group were cultured with 30 mM D-glucose in the medium for 48 h.

qPCR

Total RNA was extracted from samples using TRIzol reagent (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA). The primers used for qPCR amplification are shown in Supplementary Data 1. qPCR was performed in duplicate using Power SYBR Green PCR Master Mix (Applied Biosystems, Thermo Fisher Scientific) on an ABI 7500 sequencing detection system according to the manufacturer’s protocol. Target gene mRNA levels were calculated using the 2^-ΔΔCt method after normalization to GAPDH mRNA levels.

IHC

To verify the protein levels of hub genes in human renal biopsy samples, we selected five DN cases and five minimal change disease (MCD) cases as controls. The current study was approved by the ethics committee of Beijing Friendship Hospital affiliated with Capital Medical University. All patients provided written informed consent. The formalin-fixed, paraffin-embedded renal sections were incubated with primary antibodies against ALOX5 (#3289, Cell Signal Technology), PTEN (ab267787, Abcam, Cambridge, UK), and NCF2 (ab109366, Abcam), then analyzed using the streptavidin peroxidase detection system (Abcam) according to the manufacturer’s protocol. We used DAB (Sinopharm, China) as the horseradish peroxidase (HRP)-specific substrate (chromogen).

Statistical analysis

Statistical analyses were performed using R and GraphPad Prism 9.0.0 software. The differences between DN and control samples were statistically evaluated using unpaired t-tests or Mann-Whitney U tests. The efficiency of the hub genes for diagnosing DN was analyzed by calculating the area under the ROC curve (AUC) values. Relationships between mRNA levels, immune cell scores, and GFR values were determined by Pearson’s correlation analysis. A two-tailed value of p < 0.05 was considered statistically significant.

Results

Ferroptosis-related DEGs

GSE104954 included seven DN and 18 control tubulointerstitial samples that were collected from living donors and diagnostic kidney biopsies. A total of 879 DEGs involved in DN tubulointerstitial lesions were screened. The heat map shows the significantly different distribution of the 879 DEGs, including 455 upregulated and 424 downregulated genes (Figure 2(a)).

Figure 2. Heatmap of the differentially expressed genes (DEGs) and Venn diagram of the ferroptosis-related genes. (a) Heatmap of the 879 DEGs screened using the Limma package. Gene expression levels are color coded: red, high expression; blue, low expression. (b) Venn diagram of the DEGs identified from the GSE104954 dataset and ferroptosis-related genes to identify candidate ferroptosis-related genes in diabetic neuropathy (DN).

A total of 388 ferroptosis-related genes were obtained from the FerrDb database, with 36 ferroptosis-related DEGs obtained after intersecting them with the GSE104954 DEGs (Figure 2(b)). These 36 ferroptosis-related DEGs are shown in a Venn diagram [26] (Table 2).

Table 2. Ferroptosis-related differentially expressed genes (DEGs), including ferroptosis drivers, suppressors, and markers.

Driver	Suppressor	Marker
ACSF2^a, CYBB, TFRC^a, GOT1 ALOX5^a, ACO1, GABARAPL1 WIPI1, DPP4, POR, PTEN MMD, AGPS, ACADSB, SLC39A14	AKR1C2, GCLC, CDKN1A CD44, ZFP36, GCH1, RRM2^a, NR4A1, AHCY	ACSF2^a, TFRC^a, ALOX5^a RRM2^a, DUSP1, NCF2, ALB, DDIT4, TSC22D3, DDIT3, PCK2, VEGFA, IL33, NNMT, CAPG, YWHAE

^aGenes characterized as both driver and marker or as both suppressor and marker.

Enriched pathways and analysis of ferroptosis-related DEGs

GO analysis indicated that these genes were significantly enriched in the cytosol, mitochondria, melanosome, Nicotinamide Adenine Dinucleotide Phosphate Oxidase (NADPH) oxidase complex, and apical plasma membrane (Figure 3). KEGG pathway analysis suggested that these ferroptosis-related DEGs were mainly enriched in ferroptosis, phosphatidylinositol 3-kinase (PI3K)/Akt signaling, hypoxia-inducible factor 1 (HIF-1) signaling, Forkhead Box Protein O (FOXO) signaling, cysteine and methionine metabolism, autophagy – animal, and metabolic pathways. Reactome pathway analysis revealed that the ferroptosis-related DEGs were mainly enriched in metabolism, cellular responses to stress, cellular responses to stimuli, cross-presentation of particulate exogenous antigens (phagosomes), and metabolism of ingested selenomethionine. The abovementioned KEGG and Reactome pathways are shown in Figure 4. Notably, many ferroptosis-related DEGs were enriched in interleukin (IL)-4 and IL-13 signaling and cross-presentation of particulate exogenous antigen pathways.

Figure 3. Significantly enriched gene ontology (GO) terms (BP, MF, and CC) of ferroptosis-related differentially expressed genes (DEGs) in diabetic neuropathy (DN). The size of the node indicates the number of gene counts and the color of the node (red to blue) indicates ascending p-value. BP, biological process; MF, molecular function; CC, cellular component.

Figure 4. Significantly enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) and Reactome pathways of ferroptosis-related differentially expressed genes (DEGs) in diabetic neuropathy (DN). The size of the node indicates the number of gene counts and the color of the node (red to blue) indicates ascending p-value.

PPI network analysis of ferroptosis-related DEGs and key ferroptosis-related gene identification

To screen and identify the key genes, we uploaded the ferroptosis-related DEGs to STRING for further analysis, resulting in 31 nodes plus 51 edges. After hiding disconnected nodes in the network, the data file was processed using Cytoscape (Figure 5(a)). A total of 10 key genes were identified by intersecting the results from the five CytoHubba algorithms (MCC, MNC, DMNC, EPC, and Clustering Coefficient) (Figure 5(b)). The 10 identified key genes were CD44, PTEN, CDKN1A, DPP4, DUSP1, CYBB, DDIT3, ALOX5, VEGFA, and NCF2.

Figure 5. Protein-protein interaction (PPI) network and nine key genes. (a) PPI network of ferroptosis-related differentially expressed genes (DEGs) in diabetic neuropathy (DN). The interaction network between ferroptosis-related DEGs contained 36 nodes and 57 edges. Nodes represent genes and edges represent gene-gene associations. Node color represents neighborhood connectivity and node size represents degree of connectivity. (b) Venn diagram showing the screening of nine key genes identified by intersecting the results from five CytoHubba algorithms.

Key gene validation and efficacy evaluation

We downloaded and analyzed the GSE30529 dataset to cross-verify the key genes in another DN-related dataset. Six of the ten identified key genes, CD44, PTEN, DDIT3, ALOX5, VEGFA, and NCF2, showed similar upregulation/downregulation patterns between the datasets (Table 3). Detailed information on these six key genes was extracted from the NCBI database (https://www.ncbi.nlm.nih.gov/gds/) (Table 4).

Table 3. Five cross-validated key ferroptosis-related differentially expressed genes (DEGs) in diabetic neuropathy (DN).

Gene symbol	GSE104954		GSE30529
	Log FC	p Value	Log FC	p Value
CD44	0.667	7.79E-05	0.673	0.001
PTEN	0.520	1.45E-05	0.512	0.003
DDIT3	−0.507	1.59E-05	−0.750	0.002
ALOX5	0.785	9.62E-09	0.997	3.73E-05
NCF2	0.639	0.001	1.223	0.001
VEGFA	−0.840	9.56E-07	−0.561	0.005

Table 4. Significant key ferroptosis-related differentially expressed genes (DEGs) in diabetic neuropathy (DN).

Gene symbol	Official full name	Description
ALOX5	Arachidonate 5 Lipoxygenase	A member of the lipoxygenase gene family and plays a dual role in the synthesis of leukotrienes from arachidonic acid.
NCF2	Neutrophil Cytosolic Factor 2	Encodes neutrophil cytosolic factor 2, the 67-kilodalton cytosolic subunit of the multi-protein NADPH oxidase complex found in neutrophils
PTEN	Phosphatase and Tensin Homolog	Encodes a phosphatidylinositol-3,4,5-trisphosphate 3-phosphatase and functions as a tumor suppressor by negatively regulating AKT/PKB signaling pathway
CD44	CD44 Molecule (Indian blood group)	The protein encoded by this gene is a cell-surface glycoprotein involved in cell-cell interactions, cell adhesion and migration. This protein participates in a wide variety of cellular functions including lymphocyte activation, recirculation and homing, hematopoiesis, and tumor metastasis.
DDIT3	DNA Damage Inducible Transcript 3	This gene encodes a member of the CCAAT/enhancer-binding protein (C/EBP) family of transcription factors. The protein is implicated in adipogenesis and erythropoiesis, is activated by endoplasmic reticulum stress, and promotes apoptosis.
VEGFA	Vascular Endothelial Growth Factor A	This gene is a member of the PDGF/VEGF growth factor family. This growth factor induces proliferation and migration of vascular endothelial cells, and is essential for both physiological and pathological angiogenesis. This gene is upregulated in many known tumors and its expression is correlated with tumor stage and progression.

Clinical data validation and diagnostic efficacy of key genes in DN evaluation

Nephroseq v5 showed that the mRNA expression levels of ALOX5, NCF2, PTEN, and CD44 in renal interstitial tubules were upregulated, while DDIT3 and VEGFA mRNA levels were downregulated, in DN patients compared with controls. Furthermore, the mRNA expression levels of ALOX5, NCF2, PTEN, and CD44 in renal interstitial tubules were negatively correlated with GFR in DN patients, indicating that these upregulated ferroptosis-related DEGs may promote the progression of DN. In contrast, the mRNA expression levels of DDIT3 and VEGFA were positively correlated with GFR, indicating that these ferroptosis-related DEGs may prevent DN progression. The AUC values of the six ferroptosis-related DEGs were all > 0.8. The diagnostic values of the key genes were: ALOX5 (AUC, 1.000), NCF2 (AUC, 0.825), PTEN (AUC, 1.000), CD44 (AUC, 1.000), VEGFA (AUC, 0.943), and DDIT3 (AUC, 0.833). The threshold (cutoff) values of the ROC curves of the key genes for distinguishing between DN and controls were: ALOX5, 1.118; NCF2, 0.727; PTEN, 65.300; CD44, 10.740; VEGFA, 1.228; and DDIT3, 4.140 (Figure 6).

Figure 6. Validation and efficacy evaluation of key genes in renal interstitial tubules. (a) ALOX5, (b) NCF2, (c) PTEN, and (d) CD44 mRNA levels were upregulated in diabetic neuropathy (DN) compared with normal samples, while (e) DDIT3 and VEGFA mRNA levels were downregulated in DN compared with normal samples. Expression levels of (f) ALOX5, (g) NCF2, (h) PTEN, and (i) CD44 were negatively correlated with glomerular filtration rate (GFR), while (j) expression levels of DDIT3 and VEGFA were positively correlated with GFR. p < 0.05 was considered statistically significant.

Infiltration of immune cells in DN interstitial tubules

Because excessive or deficient ferroptosis has been linked to a dysregulated immune response [11], we investigated immune cell infiltration in renal interstitial tubules in DN. Analysis (CIBERSORT method) of GSE30529 and GSE104954 showed increased infiltration of M1 macrophages and resting mast cells, and decreased infiltration of activated mast cells in DN tubulointerstitial areas compared with normal interstitial tubules. Furthermore, the immune score of resting mast cells was negatively correlated with GFR in DN patients. The mRNA expression levels of CD44 were positively correlated with M0 macrophages, while the mRNA expression levels of VEGFA were negatively correlated with gamma delta T cells (Figure 7 and Table 5).

Figure 7. Immune infiltration analysis and Pearson correlation analysis between immune scores and expression levels of hub genes. (a) Immune infiltration scores of the GSE104954 dataset. (b) Immune infiltration scores of the GSE30529 dataset. Red boxes represent diabetic neuropathy (DN) samples and blue boxes represent control samples. (c) Pearson correlation analysis between immune scores and expression levels of hub genes in GSE104954. (d) Pearson correlation analysis between immune scores and expression levels of hub genes in GSE30529. The color from white to deep green refers to the correlation coefficient from small to large and the color from white to deep red refers to the -log10 p-value.

Table 5. Relationships between immune cell scores and glomerular filtration rate (GFR) in the GSE30529 dataset.

	Immune score			Correction between immune cells and Log2 GFR
Immune cells	Control (Mean ± std)	DN (Mean ± std)	p Value	p Value	r	95% confidence interval
B_cells_naive	7.2e-3 ± 8.4e^-3	0.05 ± 0.06	0.04	0.60	0.20	[−0.53 to 0.76]
T_cells_CD8	0.03 ± 0.07	0.13 ± 0.14	0.04	0.41	0.31	[−0.44 to 0.81]
T_cells_regulatory	1.8e^-3±4.2e^-3	0.07 ± 0.08	0.02	0.69	0.15	[0.57 to 0.74]
T_cells_gamma_delta	0.18 ± 0.14	0.02 ± 0.05	4.50e^-3	0.0021	0.87	[0.50 to 0.97]
NK_cells_resting	0.01 ± 0.03	0.05 ± 0.06	0.04	0.23	−0.44	[−0.86 to 0.31]
Monocytes	0.03 ± 0.03	4.6e^-3±0.02	0.04	0.74	−0.13	[−0.73 to 0.59]
Macrophages_M0	0.01 ± 0.04	0.08 ± 0.09	0.04	0.53	0.24	[−0.50 to 0.78]
Macrophages_M1	0.05 ± 0.04	0.0e^+0±0.0e^+	1.70e^-03	0.06	0.65	[−0.02 to 0.92]
Macrophages_M2	0.07 ± 0.04	8.8e^-3±0.02	1.30e^-03	0.43	0.30	[−0.45 to 0.81]
Mast_cells_resting	0.13 ± 0.08	7.6e-3 ± 0.03	9.50e^-04	0.01	−0.81	[−0.96 to −0.31]
Mast_cells_activated	0.00	0.12 ± 0.12	4.30e^-03	NA	NA	NA

ALOX5, PTEN, and NCF2 were upregulated, while DDIT3 and VEGFA were downregulated in HG-treated HK-2 cells

qPCR assays showed that the mRNA expression levels of ALOX5, PTEN, and NCF2 were upregulated in HK-2 cells following HG treatment for 48 h, while the mRNA levels of DDIT3 and VEGFA were downregulated. There was no statistically significant difference in CD44 mRNA levels between the two groups (Figure 8(a)).

Figure 8. (a) ALOX5, PTEN, DDIT3, and VEGFA mRNA levels in human renal tubular epithelial cells from the control and high-glucose (HG)-treated groups (n = 3 to 5). *p < 0.05, ***p < 0.001. (b) Positive ALOX5 and NCF2 IHC staining was observed in the diabetic neuropathy (DN) group, with negative staining in the minimal change disease (MCD) group. PTEN IHC staining was negative in the DN group, but positive in the MCD group.

ALOX5 and NCF2 were positive in DN, while PTEN was negative

As shown in Figure 8(b), positive IHC staining was observed for both ALOX5 and NCF2 in DN renal tubulointerstitial areas in the DN group, but negative IHC staining was found for these proteins in the MCD samples.

Discussion

In this study, we identified 36 ferroptosis-related DEGs between renal tubulointerstitial tissues in DN patients and normal controls, including 24 downregulated and 12 upregulated genes. These genes were mainly enriched in metabolic pathways and the PI3K/Akt and HIF-1 signaling pathways. In particular, the genes were enriched in metabolic pathways in both the KEGG and Reactome analyses, suggesting their significance in the ferroptosis-related pathological mechanism of DN.

Several metabolic pathways, including the tricarboxylic acid cycle, lipid metabolism, amino acid metabolism, and urea cycle, have been correlated with DN progression [27,28]. Abnormal levels of metabolites have been associated with oxidative stress, which has in turn been demonstrated to be an important factor in podocyte injury, proteinuria, and renal tubulointerstitial fibrosis [29]. In addition, an increasing number of metabolic pathways have been associated with ferroptosis. Notably, certain metabolic pathways, such as (selenium) thiol metabolism, fatty acid metabolism, iron handling, mevalonate pathway, and mitochondrial respiration, can directly affect the susceptibility of cells to lipid peroxidation and ferroptosis [30–32]. These studies suggest that metabolic pathways may be an essential pathological mechanism of ferroptosis in DN.

Adaptation of cells and tissues to low oxygen tension (hypoxia) induces the transcription of a series of genes that participate in angiogenesis, iron metabolism, glucose metabolism, and cell proliferation/survival. HIF-1 is an oxygen-sensitive transcriptional activator that acts as the primary factor mediating this response [33]. Sorafenib is a drug that can attenuate liver fibrosis by triggering hepatic stellate cell ferroptosis via the HIF-1α/xCT (cystine/glutamate antiporter) pathway [34]. Moreover, miR-34a-5p potentially promotes cell proliferation and fibrosis in DN via the sirtuin 1/HIF-1α signaling pathway [35]. These previous studies suggest that the HIF-1 signaling pathway is related to both ferroptosis and DN, and thus may be a pathological mechanism for ferroptosis in DN.

The PI3K/Akt/mammalian target of rapamycin (mTOR) signaling pathway plays an important role in the regulation of cell survival, growth, and proliferation [36]. Ginsenoside Rh1 has also been shown to alleviate type 2 DN through adenosine monophosphate (AMP)-activated protein kinase/PI3K/Akt-mediated inflammation and apoptosis signaling pathways [37]. In addition, lapatinib can induce doxorubicin-induced mitochondrial dysfunction in cardiomyocytes through inhibition of the PI3K/Akt signaling pathway, thereby enhancing oxidative stress and ferroptosis [38]. These results suggest that the PI3K/Akt signaling pathway may be a pathological mechanism of ferroptosis in DN.

In this study, six ferroptosis-related DEGs, ALOX5, NCF2, PTEN, CD44, DDIT3, and VEGFA, were identified as key genes. Arachidonate 5-lipoxygenase (5-LO), encoded by the ALOX5 gene, has recently been shown to play a central role in ferroptosis [39]. 5-LO is an essential enzyme that mediates lipid peroxidation by generating lipid peroxides [40], which is a critical feature of ferroptosis [41]. Numerous studies have shown that 5-LO is a target for ferroptosis, including in neurological disorders and renal failure [1,42–44]. Pharmacological inhibition of 5-LO by zileuton exhibited a neuroprotective role in glutamate-induced HT22 cells by blunting ferroptosis [42], while inhibiting 5-LO could also protect neurons from ferroptosis-related death in mice with hemorrhagic stroke via neutralizing lipid peroxides [43]. Overexpression of three selected ALOX isoforms (ALOX5, ALOX12, and ALOX15) in human embryonic kidney (HEK293) cells sensitized them to erastin-induced cell death [44]. Interference with 5-LO can alleviate inflammation and fibrosis in HG-induced renal mesangial cells [45]. Numerous studies have demonstrated that inflammation-related processes and immune cells are potentially involved in DN progression [46]. Leukotrienes (LTs) are a family of lipid mediators that function as pro-inflammatory mediators. LTs are synthesized in leukocytes from arachidonic acid, which is metabolized by 5-LO [47]. We therefore hypothesized that 5-LO might be a key molecule in the ferroptosis process in DN.

Phosphatase and tensin homolog (PTEN) is a potent tumor suppressor protein that can antagonize the effects of the proto-oncogenic PI3K/Akt signaling pathway and regulate essential cellular metabolic processes. In addition, PTEN plays an important role in the regulation of tubulointerstitial fibrosis and the epithelial-mesenchymal transition, both of which are significant features of DN [1]. A recent study also demonstrated that activating mutations in the PI3K gene or loss of PTEN function could endow cancer cells with resistance to ferroptosis [48]. IHC assays in this study demonstrated negative PTEN staining in the DN renal tubulointerstitial areas in the DN group, but positive PTEN staining in the MCD samples. Further studies are therefore needed to confirm its connection to ferroptosis in kidney disease. The FerrDb database [49] showed that the expression levels of NCF2, encoding neutrophil cytosolic factor 2, were upregulated during ferroptosis induced by erastin. NCF2 was thus inferred to be a ferroptosis-related molecule. Here, we observed positive NCF2 IHC staining in DN samples, indicating that NCF2 may participate in ferroptosis in cells from DN patients. Further investigation is needed to clarify the role of NCF2 in DN ferroptosis.

Previous work indicated that vascular endothelial growth factor A (VEGFA) is involved in glomerular endothelial cell proliferation in membrane-proliferative glomerulonephritis (MPGN) [50]. In addition, VEGFA is potentially involved in the inflammatory response by triggering expression of intracellular adhesion molecules in endothelial cells [51]. In this study, VEGFA expression levels were negatively correlated with the infiltration levels of immune cells, indicating that VEGFA may induce ferroptosis in DN patient cells through altering inflammatory responses in the immune microenvironment. VEGFA is therefore a possible molecular target for DN therapy.

DDIT3, which encodes DNA damage inducible transcript 3, was upregulated (≥ 2-fold) in erastin-treated samples in a previous study [52]. Our data suggest that DDIT3 mRNA levels were downregulated in both GEO DN datasets and our qPCR experiments. Further studies are therefore needed to confirm its specific connection to ferroptosis in DN.

We also analyzed immune cell infiltration in DN interstitial tubules. Ferroptosis can affect immune cells in two fundamentally different ways. First, ferroptosis can impact the number and function of immune cells. Second, cells undergoing ferroptosis can also be identified by immune cells, leading to a range of inflammatory [11]. Accumulating evidence indicates that macrophage polarization and ferroptosis can interact with each other at the cell-autonomous level. M1 macrophages are resistant to ferroptosis because of the loss of arachidonate 15-lipoxygenase (ALOX15) activity, compared with M2 microglia (brain macrophages) [53]. In addition, both B1 and marginal zone B cells expressed higher levels of the fatty acid transporter protein CD36, contributing to increased fatty acid intake and ferroptosis sensitivity [54]. Several in vitro and in vivo studies demonstrated that the activities and functions of cytotoxic T cells (CD8+) and helper T cells (CD4+) were regulated by lipid peroxidation and ferroptosis [55,56]. We therefore speculated that humoral and cellular immunity may be transformed in DN interstitial tubules, and there also may be crosstalk among ferroptosis, immune cells, and interstitial tubules in DN.

This study did have some limitations. We only considered DN cases and compared them with healthy controls because of a lack of samples from diabetes patients without nephropathy and patients with non-diabetic nephropathy. In addition, a variety of drugs used by patients with DN may have a certain impact on the expression of key DN-related genes. Further validation of our findings using animal model experiments is also necessary.

In conclusion, this study identified several novel key ferroptosis-related genes with potential roles in DN tubulointerstitial injury. Several essential pathways, especially metabolic pathways, were significantly involved in ferroptosis-related DN tubulointerstitial injury. In addition, there may be crosstalk among ferroptosis, immune cells, and interstitial tubules in DN. The results of this study provide valuable reference information regarding the ferroptosis-related pathological mechanism of DN. Further clinical and basic studies are needed to elucidate the specific mechanistic details of the key genes in ferroptosis in DN-related tubulointerstitial injury.

Ethical approval

This study was approved by the Ethics Committee of Beijing Friendship Hospital affiliated with Capital Medical University (2023-P2-017-01). All subjects provided written informed consent in accordance with the World Medical Association Declaration of Helsinki.

Author contributions

WHL and ZLD conceived and designed the study. LLM and YB performed the bioinformatics and correlation analyses. LLM wrote the first draft. ZLD and YB revised the manuscript.

Acknowledgments

The authors thank Dr. Wei Chen from the experimental and translational center of Beijing Friendship Hospital for their kind suggestions. We also thank Susan Furness, PhD, and J. Iacona, PhD, from Liwen Bianji (Edanz) (www.liwenbianji.cn) for editing the English text of a draft of this manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets used or analyzed during the current study are available from the corresponding author upon reasonable request. These data were derived in the public domain GEO.

Additional information

Funding

This work was supported by the Construction of Wu Jieping Medical Foundation under Grant code [32067502021-11-26]; the Beijing Municipal Administration of Hospitals Incubating Program under Grant code [PX2022003]; the Beijing Natural Science Foundation under Grant [7232036]; and the Beijing Natural Science Foundation under Grant [7232030].