LOX-induced tubulointerstitial fibrosis via the TGF-β/LOX/Snail axis in diabetic mice

Abstract

Background

The partial epithelial-mesenchymal transition (EMT) is emerging as a significant mechanism in diabetic nephropathy (DN). LOX is a copper amine oxidase conventionally thought to act by crosslinking collagen. However, the role of LOX in partial EMT and fibrotic progression in diabetic nephropathy has not been investigated experimentally.

Methods

The bulk RNA sequencing and single-nuclei RNA sequencing (snRNA-seq) analysis were explored to find the role of LOX in diabetic nephropathy. We then investigated the partial EMT and the possible signaling pathway of LOX, both in vivo and in vitro by LOX inhibition experiments in diabetic mice and HK-2 cells. Besides, we further assessed kidney fibrosis and renal function.

Results

LOX expression was elevated in kidneys of diabetic mice. Additionally, snRNA-seq results indicated that LOX expression was higher in partial epithelial-mesenchymal transition proximal tubular (PemtPT) epithelial cells. Moreover, we found that increased LOX prompted partial EMT of renal tubular epithelial cells (RTECs) by modulating the transcription factor Snail both in vivo and in vitro. Remarkably, inhibition of LOX effectively mitigated the partial EMT of RTECs in diabetic mice, thereby attenuating kidney fibrosis and enhancing renal function. Additionally, we identified the TGF-β signaling pathway as an upstream regulator of LOX, and inhibiting LOX partially reversed the partial EMT program in HK-2 cells induced by the TGF-β signaling pathway.

Conclusions

Hyperglycemia induces partial EMT of RTECs via the TGF-β/LOX/Snail axis, thereby contributing to diabetic kidney fibrosis. Inhibiting LOX can effectively reverse the partial EMT of RTECs, diminish diabetic kidney fibrosis, and improve renal function.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-024-06056-z.

Introduction

Diabetic nephropathy (DN), a complication of chronic hyperglycemia, affects around 30% of patients with type 1 diabetes (T1DM) and 40% of those with type 2 diabetes (T2DM). It become the predominant cause of end-stage renal disease (ESRD) worldwide [1–4]. The study found that fibrosis is the main pathological feature of DN associated with high morbidity and mortality rates. The progression of fibrosis can persist due to the continual release of profibrotic mediators, alongside epigenetic alterations, even in cases with glycemia under control [5, 6]. Diabetic kidney fibrosis includes glomerulosclerosis and tubulointerstitial fibrosis (TIF) [7, 8]. Recent studies highlighted the role of tubulointerstitium in the pathogenesis of diabetes nephropathy, as the degree of decline in renal function is more strongly correlated with the severity of tubulointerstitial injury than with glomerular lesions [9, 10]. Furthermore, TIF can heighten the kidney’s susceptibility [11]. However, there are currently no definitive therapies for TIF [12].

TIF is characterized by the accumulation of extracellular matrix, inflammation, fibroblast activation, microvascular rarefaction, and tubular cell loss. Among the constituents of the renal parenchyma, renal tubular epithelial cells (RTECs) play a pivotal role [13]. Recently a new mechanism highlights the relationship between RTECs and TIF, and is identified as partial epithelial-mesenchymal transition (EMT). Unlike complete EMT, where RTECs transform fully into interstitial myofibroblasts, partial EMT describes a scenario where RTECs maintain their tubular localization but exhibit diminished expression of epithelial markers alongside heightened expression of mesenchymal markers [14, 15]. Snail reactivation is a necessary and sufficient condition for partial EMT [14]. This program prompts RTECs to transmit signals to the interstitium, resulting in myofibroblast differentiation and fibrogenesis, as well as inflammation [16, 17]. Moreover, studies reported the presence of partial EMT of RTECs in diabetic mouse models, which leads to several profibrotic signaling pathways that culminate in TIF [13, 18, 19]. Thus, targeting partial EMT presents potential avenues for the development of innovative anti-fibrotic therapies for diabetic nephropathy.

Lysyl oxidase family (LOX, LOXL1-4) are copper amine oxidases that are secreted to the extracellular space to catalyze the covalent crosslinking of structural ECM components including collagens and elastin, thereby regulating the tensile strength of tissues [20, 21]. Interestingly, research suggests that LOX is distributed in the intracellular space, especially in the nucleus and perinuclear region [22–25]. Recent studies, particularly in oncology, have shown that LOX and LOXLs also play a role in the intracellular space to regulate cell adhesion and migration [22, 26–28]. In diabetic rat nephrons, there is an observed upregulation of LOX and LOXL2, rather than LOXL1 or LOXL3 [29]. Since LOXL2 mainly acts by cross-linking type IV collagen in the kidney, which is an essential component of the glomerular basement membrane, specific inhibition of LOXL2 effectively prevented diabetic glomerulosclerosis [30, 31]. However, in the realm of fibrosis-related investigations, the potential intracellular fibrogenic mechanisms of LOX and the potential efficacy of LOX inhibition in ameliorating diabetic nephropathy have not received adequate consideration.

In the current study, we employed bulk RNA sequencing, single-nuclei RNA sequencing, and both in vitro and in vivo experiments to investigate the role of LOX in diabetic nephropathy. We identified LOX as a significant driver of partial EMT of RTECs in diabetic kidneys and established LOX as a promising therapeutic target for diabetic nephropathy.

Materials and methods

Animals

The animal studies were reviewed and approved in accordance with the policies instituted by the Committee of Animal Experiment Center of Zhejiang University (Approval Number 20210192). Male C57BL/6 mice, 8 weeks old (body weight: 22–25 g), were obtained from the Experimental Animal Center of Zhejiang University. The mice were randomly divided into four groups: control group (Ctrl), diabetic group (DM), control + Beta-aminopropionitrile (BAPN) group (Ctrl + BAPN), and diabetic + Beta-aminopropionitrile group (DM + BAPN). The DM and DM + BAPN groups were administrated 150 mg/kg streptozotocin (STZ) intraperitoneally in a single dose, as previously described [32]. The Ctrl and Ctrl + BAPN groups were received with an equal volume of sodium citrate. BAPN is a specific and irreversible LOX inhibitor [33, 34]. One week after STZ injection, mice in the Ctrl + BAPN and DM + BAPN groups were given BAPN (#A3134, Sigma-Aldrich, USA) at 100 mg/kg every two days intraperitoneally for the next 23 weeks according to previous studies [35–37]. At the same time, The Ctrl and DM groups were treated with an equal volume of sodium citrate. To confirm the reliability of the diabetic model, fasting blood glucose levels of all mice were measured weekly. The mice were deemed to have diabetes mellitus when their fasting blood glucose levels were higher than 16.7 mM. Additionally, weekly measurements of body weight were performed. At the end of 24 weeks after STZ injection, all mice were sacrificed.

Cell culture

Human proximal tubular epithelial cells (HK-2) were a kind gift from professor En-Yin Lai of Zhejiang University. The cells were cultured in DMEM/F-12 (Dulbecco’s modified Eagle’s medium: F-12 medium) with 10% fetal bovine serum and 1% penicillin-streptomycin at 37 °C with 5% CO₂ in a humidified incubator. To induce a partial EMT program, cells were treated with DMEM containing 60 mM D-glucose (high glucose, HG) for 72 h in accordance with previous studies [38–40]. Cells were treated with DMEM containing 5.56 mM D-glucose (normal glucose, NG) as control. Mannitol medium containing 5.56 mM D-glucose and 54.44 mM D-mannitol was used as the osmotic control. BAPN (500µmol/L) was used as an inhibitor of LOX. For TGF-β (#100–21, Peprotech, USA) stimulation, cells were co-cultured with 5 ng/mL TGF-β1 for 48 h.

Data acquisition and analysis

For bulk RNA sequencing analysis, we utilized the datasets GSE197699 and GSE29660, selected from the public Gene Expression Omnibus (GEO) database. GSE197699 was used to investigate the difference in gene expression in the kidneys of wild-type control mice and wild-type diabetic mice, with three biological replicates for each of the groups [41]. GSE29660 was used to investigate the effects of TGF-β1 on human proximal tubular epithelial cells (HK-2) [42]. The dataset comprised three untreated control samples and two experimental group samples treated with TGF-β1. For GSE29660, normalization and log2 conversion were used to filter out the differential expression gene (DEGs) by GEO2R. For all the bulk RNA sequencing datasets, adjusted P value < 0.05 and |log2-fold change (FC)| > 0.5 were considered significant DEGs. All DEGs were displayed as volcano plots using the ggplot2 package in R.

For single-nuclei RNA sequencing (snRNA-seq) analysis, we utilized the dataset GSE209821, selected from the public Gene Expression Omnibus (GEO) database, with male ZSF1 lean rats as the control group and male ZSF1 obese rats as the diabetic group, with three biological replicates for each of the groups [43]. The data used had already undergone quality control. Cells of poor quality, defined as having fewer than 200 or more than 3,000 expressed genes or mitochondrial gene percentage exceeding 15%, were excluded. Doublet-like cells were identified using DoubletFinder. Data for control group and diabetic group were obtained. Based on Seurat package (v4.4.0) and harmony package (v1.2.0), normalization, identification of highly variable genes, scaling, linear dimension reduction, correction for potential batch effects, nearest neighbor network creation, clustering, dimension reduction, identification of marker genes and manual annotation were done.

Gene ontology (GO) analysis

Subsequently, Gene Ontology (GO) analysis was performed on DEGs to reveal their enrichment in biological processes (BP). The GO analysis was implemented using the clusterProfiler package in conjunction with either org.Mm.eg.db or org.Hs.eg.db package accordingly in R [44]. The Benjamini-Hochberg (BH) method was used for P value correction to control the false discovery rate (FDR), with the significance threshold set at adjusted P value < 0.05. To visualize the analysis results, a bubble chart was used to display the top 10 GO terms with the smallest P values in the BP category by the ggplot2 package. Additionally, we presented the individual GO terms and corresponding genes in a network diagram, to highlight common genes across multiple GO terms, underscoring their significance.

Gene set enrichment analysis (GSEA)

Gene Set Enrichment Analysis was employed to assess whether DEGs were enriched in gene sets associated with collagen biosynthetic process, collagen fibril organization, fibroblast proliferation, and so on. These gene sets were downloaded from the Molecular Signatures Database (MSigDB) for GSE197699 and GSE29660. For GSE209821, gene sets were downloaded from Human MSigDB Collections and converted to rat orthologues using the R package biomaRt.

Pseudotime analysis

To explore the changes in gene expression patterns during partial EMT program, the PemtPT cluster along with adjacent cells on the UMAP plot were chosen for pseudotime analysis. In total, 8,443 cells were analyzed using Monocle2 [45].

Real time quantitative PCR (qRT-PCR) analysis

Total RNA was extracted from mouse kidney tissue samples and cultured cells using RNAiso Plus (#9019, Takara). RNA was concentrated in chloroform. After centrifugation at 12,000 g for 15 min, supernatants were extracted with isopropanol and then cleaned with ethanol. cDNA was obtained in accordance with the protocol (#RR047A, Takara). Messenger RNA (mRNA) levels for target genes were analyzed by qRT-PCR using the TB Green Premix Ex Taq TM II kit (#RR820, Takara) on a LightCycler480 (#480II, Roche). The expression of the control gene (β-actin) was used to normalize the gene expression, and the 2^{− ΔΔCT} method was applied to calculate the fold changes. The primers used are listed in Supplementary Material 2.

Western blotting analysis

Samples were lysed in RIPA Lysis Buffer (#R0010, Solarbio) containing phosphatase and protease inhibitor (#P1261, Solarbio) and 1 mM PMSF for 30 min on ice followed by ultrasound. After centrifugation at 12,000 g at 4 °C for 15 min, the protein concentration of supernatants was measured using the BCA protein assay kit (P0012, Beyotime). Then the supernatants were boiled with a loading buffer. Protein samples were loaded on gels, and electroblotted to nitrocellulose membranes. After blocking, membranes were incubated overnight at 4 °C with primary antibodies against β-actin (#A5316, Sigma), LOX (#ab174316, Abcam), mature-LOX (#ab31238, Abcam), Snail (#3879, Cell Signaling Technology), GAPDH (#60004, Proteintech), Col I (#ab260043, Abcam), TGF-β1 (#ab215715, Abcam), E-cadherin (#3195, Cell Signaling Technology), Vimentin (#5741, Cell Signaling Technology), α-SMA (#ab124964, Abcam). After washed thrice with 1% Tween 20-TBS (TBS-T), membranes were incubated with goat anti-rabbit IRDye 800CW (LI-COR Biosciences) and goat anti-mouse IRDye 680RD (LI-COR Biosciences) as secondary antibodies for an hour and then washed again (three times for 10 min each). The Odyssey Clx (Li-COR Biosciences, USA) was used to measure the relative expression of proteins.

Immunofluorescence

The HK-2 cells were grown on the coverslips and fixed with 4% paraformaldehyde for 10 min. Renal sections and coverslips with cultured cells were washed, permeabilized, and blocked. Then they were incubated with primary antibodies overnight at 4 °C. Briefly, kidney sections were incubated with rabbit anti-LOX (#ab174316, Abcam); or rabbit anti-E-cadherin (#3195, Cell Signaling Technology) and mouse anti-α-SMA (#ab7817, Abcam); or rabbit anti-Vimentin (#5741, Cell Signaling Technology). Coverslips with cultured cells were incubated with rabbit anti-α-SMA (#ab124964, Abcam); or rabbit anti-Vimentin (#5741, Cell Signaling Technology). After incubating the primary antibodies overnight, kidney sections and coverslips with cultured cells were washed three times and incubated with secondary antibodies. Then they were washed and mounted with an antifade mounting medium with DAPI to visualize the nuclei. Images were pictured using a confocal microscope (FV1000, Olympus).

Lysyl oxidase (LOX) activity assay

LOX activity was measured by Amplite Fluorimetric Lysyl Oxidase Assay Kit (#15255, AAT Bioquest, USA). Supernatants from the kidneys and culture media of cells were prepared in a solid black 96-well plate and incubated with LOX working solution at 37 ℃ for 30 min protected from light. The fluorescence increase was monitored with a fluorescence plate reader (Varioskan Flash, Thermo Scientific, USA) at Ex/Em = 540/590 nm.

Masson, Sirius Red, and periodic acid-Schiff (PAS) staining

The kidney samples were harvested and fixed with 4% paraformaldehyde. After dehydration in 10–20–30% sucrose individually, kidney samples were embedded in paraffin and cut into 3 μm thickness. Masson’s trichrome staining, Sirius Red staining, and Periodic Acid-Schiff staining were performed in accordance with the manufacturer’s protocols. Images were observed using an Olympus microscope.

Measurement of blood urea nitrogen (BUN) and serum creatinine (SCr)

Serum samples were collected and stored at -80℃. Concentrations of BUN and SCr were measured using Cobas c311 (Roche, USA).

Statistical analysis

Data were presented as mean ± SD. Statistical analyses were done using either unpaired t-test or one-way ANOVA followed by Tukey’s multiple comparisons test. Statistical analyses were done using GraphPad Prism Version 9. Statistical significance was considered at P < 0.05.

Results

LOX expression was upregulated in renal tubular epithelial cells of diabetic mice

We first utilized the dataset GSE197699 from Gene Expression Omnibus (GEO) database to explore differential expression genes (DEGs) between the kidneys of wild-type control mice and wild-type diabetic mice. We observed that LOX was highly expressed in the diabetic group (Fig. 1A). We then conducted Gene Ontology (GO) analysis on the differential expression genes (DEGs) and identified the top 10 GO terms with the lowest p values (Fig. 1B). We observed enrichment of immune-related and fibrosis-related Gene Ontology (GO) terms, such as immune response-regulating signaling pathway and extracellular matrix organization. Since fibrosis is a critical pathological mechanism in diabetic nephropathy, we further investigated the enrichment of fibrosis-related gene sets using Gene Set Enrichment Analysis (GSEA). In diabetic groups, there was a marked upregulation of collagen biosynthetic process, collagen fibril organization, fibroblast proliferation, and so on (Fig. 1C). We further explored the expression and location of LOX in vivo. LOX mRNA expression in kidney tissues was enhanced in the diabetic group versus the control group (p < 0.05, Fig. 1D). Additionally, western blotting analysis also confirmed the heightened expression of LOX in diabetic kidneys (p < 0.05, Fig. 1E-F). We confirmed the upregulation of LOX expression by immunofluorescence. We observed that LOX was present in both the cytoplasm and nuclei of tubular epithelial cells, with a higher concentration in the nuclear or perinuclear region, suggesting an intracellular function of LOX (Fig. 1G).

Fig. 1.

LOX expression was upregulated in renal tubular epithelial cells of diabetic mice. (A) Volcano plot displaying up-regulated (red) and down-regulated (blue) differential expression genes (DEGs) in diabetic kidneys compared with control kidneys. (B) Bubble chart of the top 10 GO terms with the smallest P values in the Biological Process (BP). (C) Gene Set Enrichment Analysis of differential expression genes. (D) Validation of LOX mRNA expression in kidney tissues by qRT-PCR analysis (n = 4). (E) Protein levels of LOX in kidney tissues by western blotting. (F) Quantification of LOX expression (n = 3). (G) Representative immunofluorescence staining images displaying LOX expression. *p < 0.05, **p < 0.01

Single-nuclei RNA sequencing (snRNA-seq) revealed that LOX was expressed in partial epithelial-mesenchymal transition proximal tubular (PemtPT) epithelial cells

To further investigate the function of LOX in diabetic nephropathy, GSE209821 from Gene Expression Omnibus (GEO) database was used to elucidate the mechanism of LOX at a single-cell resolution. We analyzed the data of 115,904 cell nuclei from both control group and diabetic group. The cells were firstly annotated as proximal tubule epithelial cells (Prox tub, PT), loop of Henle (LOH), endothelial cells (Endo), distal convoluted tubule (DCT), connecting tubule (CNT), collecting duct principal cells (PC), collecting duct intercalated cells (IC), stroma cells (Stroma), podocytes (Podo), and immune cells (Immune) (Fig. 2A-B, S1A). To further annotate the kidney structures, we refined the proximal tubule epithelial cells (Prox tub, PT) into high osteopontin PT (PT Spp1⁺), proximal straight tubule (PST), proximal convoluted tubule (PCT), high mitochondrial gene content PT (mitoPT), proximal straight tubule S2 segment (PST S2), injured PT (PTinj) and partial epithelial-mesenchymal transition PT (PemtPT). The loop of Henle (LOH) was further detailed into thick ascending limb (TAL), descending thin limbs (DTL) and ascending thin limb (ATL) of Henle’s loop. Additionally, the stroma was further subdivided into fibroblasts (Fib), smooth muscle cells (SMC) and pericytes (Peri) (Fig. 2A, C). We investigated the expression of LOX family members across different cell types. LOX was primarily expressed in PemtPT and Fib/Peri clusters, with the highest level observed in the PemtPT cluster. However, LOXL1-3 were primarily expressed in the Fib/Peri cluster (Fig. 2D). Additionally, more than 80% of the cells within the PemtPT cluster were derived from the diabetic group (Fig. 2E). These results indicated that the upregulation of LOX in diabetic kidney was potentially associated with the occurrence of partial EMT. To elucidate the changes in gene expression patterns during the partial EMT program, we selected the PemtPT cluster along with adjacent cells on the UMAP plot for pseudotime analysis (Fig. S1B). We observed that the PT (Spp1⁺) cluster differentiated towards the PemtPT cluster, with their gene expression patterns transitioning towards those of the Fib/Peri cluster (Fig. 2F-H). We presented the expression of Prox tub and Stroma markers shown in Fig. 2B along with the pseudotime. We observed a decline in the expression of Prox tub markers, including Slc34a1 and Cubn, while an increase in Stroma markers, including Col12a1 and Pdgfrb (Fig. 2I). Besides, Gene Set Enrichment Analysis (GSEA) demonstrated that the PemtPT cluster exhibited a significant upregulation in EMT and response to TGF-β compared to other Prox tub (Fig. 2J). We further detected the expression of LOX mRNA in HK-2 cells by qRT-PCR, As seen in Fig. 2K, LOX mRNA expression was higher in high glucose treated HK-2 cells compared to those treated with normal glucose (p < 0.01) or D-mannitol (p < 0.01). We also found that the LOX activity of high glucose treated HK-2 cells was significantly higher compared to HK-2 cells with normal glucose (p < 0.01) or D-mannitol (p < 0.01) treatment (Fig. 2L).

Fig. 2.

Single-nuclei RNA sequencing (snRNA-seq) revealed that LOX was expressed in partial epithelial-mesenchymal transition proximal tubular (PemtPT) epithelial cells. (A) Integrated UMAP of 115,904 high-quality nuclei from six rat kidney samples. High-level clustering: Prox tub, proximal tubule (PT) epithelial cells; LOH, loop of Henle; Endo, endothelial cells; DCT, distal convoluted tubule; CNT, connecting tubule; PC, collecting duct principal cells; IC, collecting duct intercalated cells; Stroma, stroma cells; Podo, podocytes; Immune, immune cells. Low-level clustering for Prox tub: PT Spp1⁺, Spp1⁺ PT; PST, proximal straight tubule; PCT, proximal convoluted tubule; mitoPT, high mitochondrial gene content PT; PST (S2), proximal straight tubule S2 segment; PTinj, injured PT; PemtPT, partial epithelial-mesenchymal transition PT. Low-level clustering for LOH: TAL, thick ascending limb of Henle’s loop; DTL, descending thin limbs of Henle’s loop; ATL, ascending thin limb of Henle’s loop. Low-level clustering for Stroma: Fib, fibroblasts; SMC, smooth muscle cells; Peri, pericytes. (B) Dot plot showing cell-specific markers for high-level clustering. (C) Heatmap of marker genes for low-level clustering. (D) Heatmap of the expression of LOX family members across different cell types. (E) Pie chart representing the proportion of cells from different groups in the PemtPT cluster. Ctrl, control group; DM, diabetic group. (F) Cellular trajectory showing partial EMT program with color-coded by cell type. (G) Cellular trajectory showing partial EMT program with color-coded by pseudotime. (H) Heatmap illustrating the dynamics of DEGs during partial EMT program along pseudotime. (I)The expression of Prox tub or Stroma markers along with the pseudotime. (J) Gene Set Enrichment Analysis of differential expression genes between the PemtPT cluster and other Prox tub. (K) Validation of LOX mRNA expression in HK-2 cells by qRT-PCR analysis (n = 4). NG, normal glucose group; D-mannitol, D-mannitol group; HG, high glucose group. (L) Measurement of LOX activity in culture media of HK-2 cells (n = 4). *p < 0.05, **p < 0.01

Inhibition of LOX ameliorated high glucose induced partial EMT program in HK-2 cells

To further investigate the function of LOX in partial EMT program, HK-2 cells were exposed to high glucose and treated with BAPN, a LOX inhibitor. We extracted the total protein of the cells and detected the expression of mature LOX by western blotting. We found that mature LOX was significantly upregulated in the high glucose group (p < 0.01 vs. NG group), while BAPN treatment significantly inhibited the increase in mature LOX expression caused by high glucose (p < 0.05, Fig. 3A, B). It was reported that LOX could stabilize Snail by interacting with it intracellularly [23, 28]. Additionally, Snail reactivation was a necessary and sufficient condition for partial EMT, which promoted kidney fibrosis [14]. Therefore, we further investigated whether inhibiting LOX could suppress the partial EMT program induced by high glucose in HK-2 cells by affecting Snail stability. Western blotting analysis showed that Snail protein levels in HK-2 cells rose significantly under high glucose stimulation (p < 0.05 vs. NG group). Moreover, LOX inhibition reduced the high glucose mediated increase of Snail protein (p < 0.05 Fig. 3A, C). As partial EMT was characterized by a decrease of epithelial cell biomarkers and an increase of mesenchymal cell biomarkers, E-cadherin and Vimentin were used as markers of epithelial and mesenchymal cells respectively, to detect the occurrence of the partial EMT program. Western blotting analysis demonstrated that HK-2 cells with high glucose treatment exhibited a significant decrease in E-cadherin expression and a slight increase in Vimentin expression. Furthermore, LOX inhibition restored the expression of E-cadherin and downregulated Vimentin (Fig. 3A, D, E). In addition, we used immunofluorescence to verify the partial EMT process. Mesenchymal cell biomarkers, such as α-SMA and Vimentin, were upregulated in HK-2 cells exposed to high glucose. BAPN treatment reduced the upregulation of α-SMA and Vimentin. Meanwhile, High glucose induced cytoskeletal rearrangement in HK-2 cells, leading to morphological changes from polygonal to spindle-shaped, which was prevented by BAPN treatment (Fig. 3F).

Fig. 3.

Inhibition of LOX ameliorated high glucose induced partial EMT program in HK-2 cells. (A) Protein levels of mature LOX, Snail, E-cadherin and Vimentin in HK-2 cells by western blotting. (B-E) Quantification of mature LOX, Snail, E-cadherin and Vimentin expression (n = 3). (F) Representative immunofluorescence staining images displaying α-SMA and Vimentin expression. *p < 0.05, **p < 0.01

Inhibition of LOX ameliorated partial EMT in diabetic mice

We further investigated whether LOX induced partial EMT in tubular epithelial cells of diabetic mice. Mice were treated with BAPN as a LOX inhibitor. Kidneys from different groups were harvested at 24 weeks after the streptozotocin (STZ) injection. As seen in Fig. 4A, Blood glucose levels in the DM group and DM + BAPN group increased to more than 16.7 mmol/L and were much higher than those in the Ctrl group and Ctrl + BAPN group. In addition, the body weights in the DM group and DM + BAPN group were lower than those in the Ctrl group and Ctrl + BAPN group. (Fig. 4B). We then measured the expression and activity of mature LOX in different groups of mice. Western blotting analysis showed that the level of mature LOX in the kidneys of DM group mice was higher than that of the Ctrl group (p < 0.01), while the level of mature LOX in the kidneys of DM + BAPN group mice was significantly lower than that of the DM group (p < 0.05, Fig. 4C, D). Correspondingly, LOX activity assay also indicated that LOX activity was significantly increased in the kidneys of DM group mice, and BAPN treatment significantly reduced the LOX activity (p < 0.05, Fig. 4E). To investigate whether LOX inhibition could reduce the expression of Snail in vivo, we detected the expression of Snail in kidneys of different groups. Snail was upregulated in the kidneys of DM group mice compared with the Ctrl group, and BAPN treatment decreased the Snail expression of diabetic mice (p < 0.05, Fig. 4C, F). We then examined the biomarkers related to partial EMT. We found that the epithelial cell biomarker, E-cadherin, was significantly reduced in diabetic mice, while BAPN treatment restored the expression of E-cadherin (p < 0.05, Fig. 4C, G). At the same time, the mesenchymal cell biomarkers, such as α-SMA and Vimentin, were significantly upregulated in diabetic mice, while BAPN treatment decreased the upregulation of α-SMA and Vimentin (p < 0.05, Fig. 4C, H, I). We further confirmed the effect of BAPN treatment on partial EMT of tubular epithelial cells in diabetic mice by immunofluorescence. Immunofluorescence results showed that the expression of α-SMA increased, while the expression of E-cadherin decreased in the kidneys of diabetic mice. Moreover, the tubular epithelial cells still resided within the basement membrane, which was consistent with the definition of partial EMT. Meanwhile, BAPN treatment increased the expression of E-cadherin and decreased the expression of α-SMA in the kidneys of diabetic mice, alleviating partial EMT (Fig. 4J). α-SMA was a marker of myofibroblasts [46]. We found that the expression of α-SMA in fibroblasts was significantly higher than that in tubular epithelial cells, suggesting that the upregulation of α-SMA in kidney tissue shown by western blotting (p < 0.05, Fig. 4C, H) was more suitable for reflecting the myofibroblast differentiation in diabetic kidneys, which was also an important consequence of partial EMT. We further stained Vimentin by immunofluorescence, and we found that the expression of Vimentin in tubular epithelial cells of diabetic mice was significantly upregulated, while BAPN decreased the expression of Vimentin, alleviating partial EMT of tubular epithelial cells in diabetic mice (Fig. 4K).

Fig. 4.

Inhibition of LOX ameliorated partial EMT in diabetic mice. (A) Effect of BAPN on fasting blood glucose level. (B) Effect of BAPN on body weight. (C) Protein levels of mature LOX, Snail, E-cadherin, α-SMA and Vimentin in kidney tissues by western blotting. (D) Quantification of mature LOX expression (n = 3). (E) Measurement of LOX activity in kidney homogenates (n = 4). (F-I) Quantification of Snail, E-cadherin, α-SMA and Vimentin expression (n = 3). (J) Representative immunofluorescence staining images displaying expression of α-SMA (green) and E-cadherin (red). (K) Representative immunofluorescence staining images displaying Vimentin expression. *p < 0.05, **p < 0.01

Inhibition of LOX ameliorated tubulointerstitial fibrosis and restored normal renal function in diabetic mice

As partial EMT played an essential role in diabetic kidney fibrosis [18], we further explored whether LOX inhibition could reduce kidney fibrosis and improve renal function in diabetic mice. Masson and Sirius red staining were performed to detect and evaluate the severity of kidney fibrosis. Compared with the Ctrl group, an increase in extracellular matrix (ECM) accumulation was observed in the renal tubulointerstitium of the DM group (p < 0.01). Meanwhile, kidneys in the DM + BAPN group displayed a lower degree of extracellular matrix (ECM) accumulation in the renal tubulointerstitium compared with the DM group (p < 0.01, Fig. 5A-D). Periodic Acid-Schiff (PAS) staining revealed thickening of the tubular basement membrane (TBM) in diabetic mice, which was alleviated by BAPN treatment (Fig. 5E). Additionally, western blotting analysis showed increased expression of Col I in the kidneys of diabetic mice (p < 0.05 vs. Ctrl group), which was attenuated by BAPN treatment (Fig. 5F-G). Furthermore, we measured the renal function indicators like blood urea nitrogen (BUN) and serum creatinine (SCr) and found that the diabetic mice had elevated blood urea nitrogen and serum creatinine levels (p < 0.01 vs. Ctrl group), which were ameliorated by treatment with BAPN (p < 0.05, Fig. 5H-I).

Fig. 5.

Inhibition of LOX ameliorated tubulointerstitial fibrosis and restored normal renal function in diabetic mice. (A) Representative Masson staining images and the collagen area is blue. (B) Quantification of results from Masson staining. (C) Representative Sirius red staining images and the collagen area is red. (D) Quantification of results from Sirius red staining. (E) Representative Periodic Acid-Schiff staining images. (F) Protein levels of Col I in kidney tissues by western blotting. (G) Quantification of Col I expression (n = 3). (H) Effect of BAPN on blood urea nitrogen (BUN) (n = 4). (I) Effect of BAPN on serum creatinine (SCr) (n = 4). *p < 0.05, **p < 0.01

TGF-β1 upregulated LOX expression and triggered the partial EMT program in HK-2 cells

TGF-β1 was reported to be significantly upregulated in diabetic mice and was a key modulator of kidney fibrosis [47]. Additionally, the diabetic group exhibited an increase in response to TGF-β (Fig. 1C). So, we used qRT-PCR and western blotting to detect the expression of TGF-β1 in the kidneys of diabetic mice, confirming the increased expression of TGF-β1 (Fig. 6A-C). Moreover, the TGF-β signaling pathway was reported to be upstream of the LOX family in many diseases [35, 48]. Therefore, we further studied the effect of TGF-β1 on the expression of LOX family members in HK-2 cells. So, we utilized the dataset GSE29660 from Gene Expression Omnibus (GEO) database to investigate the effects of TGF-β1 on HK-2 cells. TGF-β1 stimulation only upregulated LOX, but not other family members including LOXL1-4 in HK-2 cells (Fig. 6D). We then performed Gene Ontology (GO) analysis on the differential expression genes (DEGs) and listed the top 10 GO terms with the smallest p values (Fig. 6E). We found that fibrosis-related GO terms like extracellular matrix organization and extracellular structure organization were enriched. The network diagram showed the GO terms related to fibrosis and TGF-β signaling pathway and the core DEGs associated with these GO terms (Fig. 6F). LOX was the core gene of TGF-β signaling-related and fibrosis-related GO terms and was relevant to the GO terms such as extracellular matrix organization, collagen fibril organization, connective tissue development, response to transforming growth factor beta, and transforming growth factor beta receptor signaling pathway. This indicated that LOX might be a key gene in fibrosis and TGF-β signaling pathway. TGF-β1 is also a key modulator of EMT in many diseases [49, 50]. Besides, the PemtPT cluster also exhibited an increase in response to TGF-β (Fig. 2J). Furthermore, we used Gene Set Enrichment Analysis (GSEA) to investigate whether DEGs were enriched in gene sets related to the partial EMT program after TGF-β1 stimulation, and we found that epithelial-mesenchymal transition (EMT) and negative regulation of cell-matrix adhesion were upregulated (Fig. 6G, H). Meanwhile, cell-cell junction assembly and positive regulation of cell-cell adhesion were downregulated (Fig. 6I, J). These results highly corresponded with partial EMT, which reflected EMT-like transcription profile changes and dedifferentiation of tubular epithelial cells.

Fig. 6.

TGF-β1 upregulated LOX expression and triggered the partial EMT program in HK-2 cells. (A)Validation of TGF-β1 mRNA expression in kidney tissues by qRT-PCR analysis (n = 4). (B) Protein levels of TGF-β1 in kidney tissues by western blotting. (C) Quantification of TGF-β1 expression (n = 3). (D)Volcano plot displaying up-regulated (red) and down-regulated (blue) differential expression genes (DEGs) in TGF-β1 treated HK-2 cells compared with untreated HK-2 cells (LOX: Up; LOXL1-4: No significant). (E) Bubble chart of the top 10 GO terms with the smallest P values in the Biological Process (BP). (F) Network diagram presenting the GO terms and corresponding differential expression genes. (G-J) Gene Set Enrichment Analysis of differential expression genes. *p < 0.05, **p < 0.01

Inhibition of LOX ameliorated TGF-β1-induced partial EMT program in HK-2 cells

To further explore whether LOX played a crucial role in the partial EMT program of HK-2 cells induced by TGF-β signaling pathway, we further treated HK-2 cells with TGF-β1 to induce the partial EMT program and used LOX inhibitor BAPN to intervene. We found that TGF-β1 significantly increased the expression of mature LOX in HK-2 cells (p < 0.01 vs. NG group), while BAPN treatment alleviated the upregulation of mature LOX (p < 0.05 vs. TGF-β group) (Fig. 7A, C). Then, we detected the expression of Snail, and we found that inhibiting LOX significantly alleviated the upregulation of Snail caused by TGF-β1 (p < 0.05, Fig. 7A, D). Afterwards, we examined the biomarkers of partial EMT, and similarly, we found that TGF-β1 significantly reduced the expression of epithelial cell biomarker, E-cadherin, and increased the expression of mesenchymal cell biomarker, Vimentin, while inhibiting LOX partially alleviated this phenomenon (Fig. 7A, E, F). This indicated that the partial EMT program of tubular epithelial cells caused by the TGF-β signaling pathway was partially mediated by LOX. We also used immunofluorescence to verify the partial EMT process. Mesenchymal cell biomarkers, such as α-SMA and Vimentin, were upregulated in HK-2 cells exposed to TGF-β1. BAPN treatment reduced the upregulation of α-SMA and Vimentin (Fig. 7B).

Fig. 7.

Inhibition of LOX ameliorated TGF-β1-induced partial EMT program in HK-2 cells. (A) Protein levels of mature LOX, Snail, E-cadherin and Vimentin in HK-2 cells by western blotting. (B) Representative immunofluorescence staining images displaying α-SMA and Vimentin expression. (C-F) Quantification of mature LOX, Snail, E-cadherin and Vimentin expression (n = 3). *p < 0.05, **p < 0.01

Discussion

Diabetic nephropathy arises from diabetes-related impairments in glucose metabolism, leading to kidney injury through metabolic, hemodynamic, inflammatory, and fibrotic pathways [5]. Notably, tubulointerstitial fibrosis stands as a pivotal pathology in DN, and serves as the primary pathogenic mechanism culminating in ESRD and a strong predictor of kidney failure [47, 51]. The present study for the first time found that increased LOX prompted a partial EMT of RTECs by modulating the transcription factor Snail, ultimately contributing to kidney fibrosis in diabetic mice. Remarkably, inhibition of LOX effectively mitigated the partial EMT of RTECs in diabetic mice, thereby attenuating kidney fibrosis and enhancing renal function. Additionally, we identified the TGF-β signaling pathway as an upstream regulator of LOX, and inhibiting LOX partially reversed the partial EMT program in HK-2 cells induced by the TGF-β signaling pathway.

Lysyl oxidases are traditionally regarded as important secreted extracellular matrix modifying enzymes [52]. Recent studies increasingly reveal their intracellular localization and activity, implicating them in various cellular processes [27, 28, 53, 54]. A review synthesized the findings on LOX localization, highlighting its presence not only in extracellular space but also within cytoplasm and nucleus [55]. Consequently, LOX exhibits diverse functions beyond its role in catalyzing ECM crosslinking. Its intracellular activity influences gene transcription, promotes tumor metastasis, and modulates cell adhesion and motility [22, 23, 56–59]. Initially released as a proenzyme into the extracellular space, LOX undergoes processing by procollagen C-proteinases, including BMP-1, mTLL1, or mTLL2, to attain maturity [55, 60]. It was reported that mature LOX can be translocated from the extracellular space to the cytoplasm and even the nucleus, but the precise mechanism of this transport remains unclear [24, 53, 61, 62].

Importantly, studies have shown that partial EMT plays an important role in diabetic kidney fibrosis [18]. Snail is an important transcription factor for EMT. Snail activation is both necessary and adequate for the development of partial EMT in kidney fibrosis [14]. It was reported that the catalytic domain of LOX can bind to the N-terminus part of Snail intracellularly, and LOX inhibition can revert the EMT program by suppressing Snail in breast cancer, as LOX can stabilize Snail protein and hinder its degradation [23, 28]. In the present investigation, we found that inhibiting LOX effectively reduced Snail expression and alleviated partial EMT in diabetic kidneys. Consequently, this intervention led to a decrease in α-SMA positive myofibroblast in the kidney interstitium, resulting in mitigating fibrosis and improving renal function. These findings underscore the therapeutic potential of targeting the LOX-Snail axis to counteract kidney fibrosis in diabetic nephropathy.

The TGF-β signaling pathway is an important driver of diabetic kidney fibrosis and a key modulator of EMT [47, 63]. Notably, in the nephrons of diabetic rats, there is an increase in LOX and LOXL2 expression, whereas LOXL1 or LOXL3 levels remain unaffected [29]. Intriguingly, while the TGF-β signaling pathway has been reported to upregulate the LOX transcription but not LOXL2 in RTECs [64, 65], these results are consistent with our bioinformatics analysis. In this study, we demonstrated that TGF-β induces the expression of LOX, which subsequently stabilizes Snail protein and mediates partial EMT. Furthermore, inhibiting LOX effectively mitigated the TGF-β1-induced partial EMT program by reducing Snail levels. These findings underscore the intricate interplay between TGF-β, LOX, and Snail in orchestrating the partial EMT process in diabetic kidney fibrosis.

However, Our study encountered certain limitations that warrant acknowledgment. Firstly, due to the lethality associated with LOX knockout [66], we used BAPN as a LOX inhibitor, but BAPN also inhibited other LOX family members due to the high homology at the C-terminal regions of the LOX family, which contain the catalytic domains. This broad inhibition potentially may affect the interpretation of our results. Furthermore, our study did not delve into investigating the inflammatory response and cell cycle arrest, both of which are known consequences of partial EMT. Addressing these aspects could provide further insights into the comprehensive mechanisms underlying diabetic nephropathy.

Despite these limitations, we confirmed for the first time that inhibiting LOX effectively reduced diabetic kidney fibrosis and improved renal function, underscoring the significance of LOX’s intracellular function in diabetic kidney fibrosis. Moreover, our study suggested the possible synergistic effects of LOX on partial EMT and collagen crosslinking, enriching our comprehension of the mechanisms underlying LOX-mediated fibrosis. Our findings shed light on the intricate interplay among LOX, partial EMT, and diabetic kidney fibrosis (Fig. 8), laying the groundwork for future research endeavors aiming to unravel the complexities of this pathological process.

Fig. 8.

A schematic diagram showing LOX promoting partial EMT in diabetic kidney fibrosis. LOX expression is elevated by high glucose via the TGF-β signaling pathway. LOX, in turn, stabilizes Snail protein, leading to partial EMT of renal tubular epithelial cells, ultimately contributing to diabetic kidney fibrosis

Conclusions

In summary, we demonstrate that hyperglycemia upregulates LOX expression via the TGF-β signaling pathway. Subsequently, LOX stabilizes Snail protein, initiating partial EMT in RTECs, ultimately contributing to diabetic kidney fibrosis. Importantly, inhibiting LOX effectively mitigates partial EMT in RTECs, attenuates diabetic kidney fibrosis, and leads to an improvement in renal function. The present results may offer a promising therapeutic potential of targeting LOX for addressing diabetic nephropathy.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Yicheng Lu contributed to conceptualization, investigation, methodology, formal analysis, visualization, data curation and writing-original draft. Heyangzi Li contributed to investigation, methodology and data curation. Mohan Chen contributed to formal analysis and visualization. Yicheng Lin contributed to formal analysis and visualization. Xiaoming Zhang contributed to conceptualization, funding acquisition, project administration, supervision, validation, writing-review and editing. All authors read and approved the final manuscript.

Funding

This study was supported by the National College Students’ innovation and entrepreneurship training program (202310335067), and Zhejiang Xinmiao Talents Program (2022R401072).

Data availability

Data will be made available on request.

Declarations

Ethical approval

The animal studies were reviewed and approved in accordance with the policies instituted by the Committee of Animal Experiment Center of Zhejiang University (Approval Number 20210192).

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.