Clusterin elaborated by renal tubular epithelial cells under high oxalate stress serves as a matrix protein to facilitate kidney stone formation

Abstract

Calcium oxalate (CaOx) kidney stone is prevalent while the molecular mechanisms are not yet fully understood. The organic matrix of stones is rich in protein and has been considered to play a important role in stone aggregation and growth. Through multi-omics approaches, we identified that renal tubular epithelial cell (RTECs)-derived clusterin (CLU), as a matrix protein, promotes renal CaOx stone formation. Proteomics revealed significant enrichment of CLU in CaOx stones. Transcriptomic and single-cell transcriptome data from human kidney specimens confirmed marked upregulation of CLU in RTECs of kidney stone patients. In vitro, CLU accelerated renal CaOx crystal growth. Specifically, conditionally knock out of CLU in RTECs significantly alleviated CaOx crystal deposition in mouse kidneys. Furthermore, hyperoxaluria upregulated CLU in RTECs via the TGF-β1-SMAD2/3-Twist1 pathway, thereby facilitating stone formation. Our study highlights the role of the TWIST1-CLU regulatory axis in renal CaOx stone formation, suggesting CLU as a potential therapeutic target for CaOx kidney stones.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-026-06105-4.

Introduction

Kidney stone disease is among the most prevalent renal disorders, with prevalence rates of 5–10% in Europe, 8.8% in the United States, 4% in South America, and 1–19% in Asia [1, 2]. Surgical therapies for kidney stones are typically beneficial, but they may still cause problems such as loss of kidney function, severe urinary tract infections, ureteral strictures, and death [3–5]. Thus, enhancing our knowledge of kidney stones and investigating novel treatments or early diagnostic alternatives are critical. Calcium oxalate (CaOx) nephrolithiasis accounts for almost 80% of all kidney stone cases; nevertheless, its pathophysiology is not entirely understood, and early warning and diagnostic procedures are still debated [6–8].

Numerous previous studies have explored the origins of oxalate and calcium salts in calcium oxalate nephrolithiasis, as well as the process of precipitation. Studies have shown that protein components are implicated in CaOx nephrolithiasis, suggesting that protein components in urine and stone matrix may play important roles in the formation of CaOx nephrolithiasis [9]. Several proteins have been identified in the organic matrix of stones and are considered closely associated with stone formation; however, the specific mechanisms remain to be further explored.

Therefore, in this study, we performed proteomic profiling of kidney stones and, combined with multi-omics approaches, identified that RTEC-derived Clusterin (CLU), as a stone matrix protein, promotes renal stone formation. When exposed to a hyperoxaluric environment, RTECs upregulate CLU expression by activating the TGF-β1-SMAD2/3-Twist1 pathway to promote CLU transcription. Collectively, our study highlights that targeting the TWIST1-CLU regulatory axis could be a potential therapeutic strategy for CaOx nephrolithiasis.

Results

Combined multiomics analysis results revealed a significant correlation and potential causal relationship between CLU and the formation of CaOx kidney stones

We collected 5 calcium oxalate kidney stone samples and 20 urine samples from the Department of Urology at West China Hospital. Label-free mass spectrometry identified a total of 428 proteins present in all samples of CaOx stones. A KEGG enrichment analysis of these proteins showed that they are mostly linked to the complement system and the processes of clotting (Figure S1A). An analysis of the protein interaction network for the top 50 identified proteins indicated that CLU held a central position (Figure S1B). Using a Venn diagram of differentially expressed genes (DEGs) in human samples and label-free mass spectrometry results, we identified 17 proteins that may be secreted by renal tissues and subsequently enriched in kidney stones (Fig. 1A). The KEGG enrichment analysis of the 17 proteins linked to possible renal secretion showed that they play a part in activating the complement system, transporting proteins, and secretion (Figure S1C). The CLU protein was found in renal CaOx stones at levels significantly exceeding the average protein level (Fig. 1B). Besides, enzyme-linked immunosorbent assays (ELISAs) also showed that people with kidney stones had significantly higher levels of CLU protein in their urine (Fig. 1C).

Fig. 1.

CLU is essential for CaOx kidney stone formation. A Venn analysis between DEGs from Randall’s plaque mRNA sequencing and label-free MS DPs in calcium oxalate kidney stones. B A significantly greater CLU protein level was detected in the label-free MS than in the LFQ intensity of all identified proteins (the Kruskal‒Wallis test was applied since the amount of all mean proteins was skewedly distributed). The SPP1, CRYAB and APCS protein levels were omitted, as each has severe outlier values that result in severely distorted plots. C Significant increase of urine CLU protein in patients with kidney stones. D CLU is a DEG in human sample transcriptome sequencing. Screening threshold: FC > 1.5, P < 0.05. E The expression of the classical markers in each cluster from the scRNA-seq data of human Randall ‘s plaques. F Cell clustering of the included living cells in Fig. 2A. G The cell annotation of Fig. 2B. H The expression of CLU in each cell type from the scRNA-seq data of human Randall’s plaques. The expression of the CLU gene in Randall’s plaque group was significantly upregulated in epithelial cells. I The meta-analysis results of Mendelian randomization analysis based on the FinnGen and UKBioBank databases suggest that elevated CLU gene expression can lead to kidney stone formation. The OR provided represents the risk of stones per unit of elevated CLU expression. The data are presented as the means ± SEMs; *P < 0.05, ****P < 0.0001

Furthermore, we obtained transcriptomic data from the GEO public database to verify our findings [10]. After removing samples from people with tumors and injuries, principal component analysis (PCA) was used to exclude outliers. This led to 23 control samples and 24 kidney stone tissue samples being included in the analysis (Figure S1D). CLU was found to be significantly upregulated in kidney samples from patients with stones (Fig. 1D and Figure S2A). Besides, we also analyzed scRNA sequencing data from human kidney samples to consolidate the findings from conventional mRNA sequencing data. Initially, we obtained the scRNA-seq data of human samples from the GEO database (GSE176155) [11]. After PCA, cell clustering, and cell annotation, 15 separate groups of cells were found, with markers that showed what kind of cell they were (Fig. 1E and F). Figure 1G presents the annotation results, which encompass B cells, dendritic cells, endothelial cells, epithelial cells, fibroblasts, macrophages, mast cells, neutrophils, plasma cells, smooth muscle cells, T cells, and other cell types. There was significantly more CLU expression in epithelial cells in the Randall’s plaque group (RP group) compared to the control group (log2FC = 2.85, adj. P value < 0.0001, Fig. 1H). Additionally, we employed public genome-wide association study data from large sample populations to investigate the causal association between CLU protein secretion and the formation of renal CaOx stones and found that a unit increase in CLU gene expression causally elevated the risk of stone formation (OR = 1.09, P < 0.05, Fig. 1I), which indicated an association between CLU expression and kidney stone susceptibility.

Moreover, we established models of CaOx nephropathy in both mice and rats (Fig. 2A for the mouse model; Figure S2B for the rat model). Traditional bulk RNA sequencing of mouse and rat kidneys revealed significant upregulation of Clu gene expression under high-oxalate conditions (Fig. 2B and C for the mouse model; Figure S2C-S2D for the rat model). The increased amount of Clu protein in the kidneys of mice and rats was confirmed by Western blotting (WB) and immunohistochemistry (IHC) (Fig. 2D for mice and Figure S2E–S2F for rats). Similarly, scRNA-seq data of mouse kidneys was performed (Fig. 2E and 2 F). Ten cell types were identified: B cells, collecting duct cells, distal tubular cells, epithelial cells, fibroblasts, macrophages, neutrophils, proximal tubular cells, T cells, and vascular smooth muscle cells (Fig. 2G). Clu expression was notably enriched in collecting duct cells, distal tubular cells, and proximal tubular cells (Fig. 2H). An increase in Clu expression was observed in the 5-day kidney stone-model group relative to the control group (log2FC = 4.57, adj. P value < 0.0001 for proximal tubular cells; log2FC = 3.68, adj. P value < 0.0001 for distal tubular cells; log2FC = 2.10, adj. P value < 0.0001 for collecting duct cells; Fig. 2H). These data reinforced the critical involvement of CLU within the lithogenic microenvironments.

Fig. 2.

CLU is highly expressed in renal tubular epithelia cells. A An in vivo CaOx nephropathy model was established in mice. B CLU is a significant differential gene in the transcriptome comparison of the mouse model. Screening threshold: FC > 1.5, P < 0.05. C. CLU mRNA expression was significantly increased in the kidneys of the mouse model. D WB validation of CLU protein upregulation in mouse model kidneys. E The expression of the classical markers in each cluster from the scRNA-seq data of mouse kidneys. F Cell clustering of the included living cells in Fig. 2E. G Cell annotation of Fig. 2F. H The expression of Clu in each cell type from the scRNA-seq data of mouse kidneys. The expression of the Clu gene in renal tubular epithelial cells tended to be significantly upregulated in the kidneys of model mice. scRNA-seq: single-cell RNA sequencing; **P < 0.01, ***P < 0.001

CLU-targeted interventions alleviate CaOx crystal formation in vivo and in vitro

To determine whether CLU can bind to CaOx crystals and promote their growth, we performed crystal-pulldown assays (Fig. 3A). In the crystal-forming pull-down assay, recombinant CLU (reCLU) was added in advance to the calcium oxalate monohydrate (COM) formation system. The results showed that the added reCLU increased the diameter of the formed COM crystals in a concentration-dependent manner (Fig. 3B and S3A), and given that albumin is the most abundant protein in urine, we compared its effects with those of CLU on crystal formation. Albumin did not promote crystallization, underscoring the specificity of CLU's effect (Figure S3B). WB confirmed that the amount of CLU protein pulled down by the crystals gradually increased (Fig. 3C). In contrast, in the crystal pull-down assay where reCLU was added after crystal formation, the amount of CLU protein pulled down by the crystals did not increase with increasing reCLU concentration (Fig. 3C). These findings indicate that CLU primarily acts as a matrix protein to bind to CaOx monohydrate crystals, thereby promoting their aggregation and growth.

Fig. 3.

Targeting CLU mitigate CaOx crystal formation in vivo and in vitro. A Schematic drawing of the newly designed crystal pulldown and crystal-forming pulldown methods. B Crystal diameter increased after the addition of different concentrations of the recombinant CLU protein. C WB analysis of crystal pull-down and crystal-forming pull-down. D Knockdown of CLU in HK2 cells via Sh-siRNA reversed the trend of increased crystal adhesion; the black bar indicates 150 μm. E Expression profile of Cdh16 via single-cell RNA sequencing of the mouse kidney. H CaOx crystal deposition in mouse kidneys following conditional knockout of the Clu gene via the Cre‒LoxP system, I along with their statistic measurement. WB: Western blotting. The data are presented as the means ± SEMs; ns: not significant; **P < 0.01, ***P < 0.001, ****P < 0.0001

Subsequently, we performed both in vivo and in vitro intervention experiments. In the crystal adhesion assay, knockdown of Clu resulted in a significant reduction of adhered CaOx monohydrate crystals to HK2 cell monolayers (Fig. 3D and Figure S3C). To mitigate potential health risks linked to systemic knockdown, we used the mouse Cre-LoxP system for the conditional knockout (CKO) of Clu in renal tubular epithelial cells for in vivo animal experiments (Figure S3D). The scRNA sequencing results of mouse kidneys indicated that the differential expression of Clu was most pronounced in Cdh16 + renal tubular epithelial cells (Fig. 3E). Conditionally knockout of Clu in the renal tubular epithelia cells (Clu^tec−KO) (Figure S3E-S3I) significantly decreased renal CaOx crystal deposition of mice (Fig. 3H and I).

The aforementioned results indicate that CLU proteins are involved in kidney stone formation through their direct influence on crystal development. Currently, there is limited published research regarding the molecular biological mechanism through which renal tubular epithelial cells secrete the CLU protein in high oxalate conditions. Thus, the current study further examined the potential mechanisms involved.

The TWIST1 protein can initiate CLU transcription under high-oxalate conditions

In order to indentify the transcription factor of Clu in renal tubular epithelial cells under hyperoxaluric conditions, we conducted batch correlation analysis using human renal tissue mRNA transcriptome data. The correlation coefficient between 314 genes and CLU gene expression was determined to be r > 0.5 and P < 0.05 (Fig. 4A). A Venn analysis was performed utilizing a list of transcriptional regulators obtained from the ChIP Atlas database in conjunction with 314 genes significantly associated with CLU. This analysis identified 16 genes that may regulate CLU transcription in kidney stones (Fig. 4B). Following screening, we analyzed the expression of each candidate transcriptional regulator in the human kidney tissue transcriptome data and identified that 9 transcriptional regulators were significantly upregulated in Randall’s plaque samples (Fig. 4C and D). Among the nine significantly upregulated transcriptional regulators, CLU was identified as the primary gene regulated by the TWIST1 transcription factor in stones (Fig. 4E). We analyzed potential transcriptional regulators of CLU using the ChIP-Atlas database and identified a total of 57 transcription factors (Table S1). The Venn diagram analysis of CLU-associated transcription factors identified in the ChIP-Atlas database, in conjunction with previously identified significantly CLU-associated genes, indicated that TWIST1 was the sole candidate (Fig. 4F). The findings provide additional evidence that the TWIST1 protein serves as a crucial regulatory transcription factor of the CLU gene in kidney stone formation.

Fig. 4.

Positive correlation between CLU and TWIST1 expression and potential transcriptional regulation of CLU by TWIST1 in kidney stone patients. A Batch correlation analyses revealed that 314 genes were significantly correlated with the CLU gene. Threshold: Pearson-r > 0.5, P < 0.05. B Venn analysis of the genes significantly related to the identified TFs and CLU genes revealed that 16 genes possibly regulated CLU expression in kidney stone patients. The background TF list was obtained from the ChIP Atlas database. C TWIST1 was significantly upregulated in kidney stone patients according to human data (control and Randall’s plaque). D mRNA expression of 9 potential CLU-related genes that are significantly upregulated in human kidney stone tissues. E Batch correlation analyses of 9 candidate CLU-related genes revealed that the CLU gene was the top target gene of TWIST1 in the human data. F Venn analysis between DEGs obtained from human samples and TFs related to CLU (predicted via the ChIP Atlas database); only TWIST1 was identified. TFs: Transcription factors, Normal: Healthy control, KS: Kidney Stone. The data are presented as the means ± SEMs; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Based on human Randall's plaque bulk RNA-seq data, we compared four widely reported transcription factors that can regulate CLU gene expression and found no significant differences (Fig. 5A) [12]. The TWIST1 protein has been confirmed to be significantly upregulated in the kidney tissues of rat models (Figure S4A). We inhibited TWIST1 expression in HK2 cells, resulting in a significant reduction of calcium salt adhesion induced by the high oxalate environment (Fig. 5B and C). The WB results demonstrated that the upregulation of the CLU protein was suppressed (Figure S4B). According to publicly available ChIP-seq data that were provided by the Gene Transcription Regulation Database [13], the TWIST1 protein appears to be significantly enriched in the CLU promoter region (Fig. 5D). We conducted DNA pulldown experiments in vitro to preliminarily confirm the binding of the TWIST1 protein to the promoter sequence of the CLU gene (Fig. 5E). Based on the evidence presented and the potential sequences predicted from the JASPAR database, we performed subsequent chromatin immunoprecipitation (ChIP) assays (Fig. 5F). ChIP‒PCR revealed visible bands at the predicted BS4 and BS7 regions (Fig. 5G). ChIP-qPCR experiments targeting these two binding sites revealed BS4 as a potential binding site (Fig. 5H). Our findings indicate an increased binding of TWIST1 proteins to BS4 sequences under the specified modeling conditions (Fig. 5I). Following the mutation of BS4, a connection between the TWIST1 protein and the CLU promoter region was associated with transcriptional initiation activity, as demonstrated in a dual-luciferase reporter gene assay (Fig. 5J).

Fig. 5.

CLU is mainly transcriptionally regulated by the TWIST1 protein under high-oxalate conditions. A The other four widely reported important TFs involved in EMT were not significantly altered in Randall ‘s plaques. B Crystal-adhesion assay and corresponding C. statistic measurements showed that TWIST1 knockdown in HK2 cells reversed the trend toward increased crystal adhesion caused by high oxalate concentrations; the black bar indicates 150 μm. D ChIP-seq results indicating that the TWIST1 protein can initiate the transcription of the CLU gene. Data were obtained from the Gene Transcription Regulation Database (http://gtrd.biouml.org/). E The TWIST1 protein can bind to the CLU promoter sequence according to the DNA pull-down results. E Schematic drawing showing the predicted binding sequence results based on the JASPAR database 2022 (https://jaspar.elixir.no/). F Schematic drawing showing the predicted binding sequence results based on the JASPAR database 2022 (https://jaspar.elixir.no/). G ChIP‒PCR revealed that BS4 and BS7 may be positive binding sites between the TWIST1 protein and the CLU gene. H ChIP‒qPCR confirmed that BS4 is the possible binding site between the TWIST1 protein and the CLU promoter sequence. I ChIP‒qPCR results revealed that significantly more TWIST1 proteins bind to the CLU gene promoter sequence under high-oxalate conditions. J A dual-luciferase reporter experiment confirmed that the TWIST1 protein can initiate the transcription of the CLU gene. TSS: Transcription start site. ns: not significant, *P < 0.05, **P < 0.01, ***P < 0.001

The TWIST1-CLU transcriptional axis is activated by high oxalate induced EMT

TWIST1 initiates the transcription of the CLU gene in high-oxalate conditions; however, the signaling pathway governing this regulatory axis is yet to be elucidated. GSEA demonstrated a significant activation of epithelial–mesenchymal transition (EMT) in both human Randall’s plaque samples and mouse model kidneys (Fig. 6A-C). The TWIST1 protein is well known for its role in EMT [14]. Due to the absence of rat-related datasets in the MSigDB database, we employed WGCNA and KEGG pathway enrichment for rat data. The WGCNA and subsequent KEGG analysis indicated that the gene blocks significantly associated with the kidney stone model were closely linked to the apical region of the epithelium (Figure S4C). The findings provide additional evidence for the occurrence of EMT in the rat model. The IHC analysis results for E-cadherin and vimentin support the hypothesis that elevated oxalate concentrations induce EMT in renal tubular epithelial cells within rat kidneys (Figure S4D).

Fig. 6.

The EMT process regulated by TGFβ1 may be the initiating factor that leads to the activation of the TWIST1-CLU transcriptional regulatory axis. A scRNA-seq in mice revealed differential expression gene enrichment analysis of Cdh16 + cells between the 5-day calculus model group and the control group, suggesting significant enrichment of the EMT pathway in the calculus model group. B GSEA revealed that EMT is significantly activated in kidney stone patients. C GSEA revealed that EMT is significantly activated in a mouse kidney stone model. D GSEA revealed that the TGFβ signaling pathway is activated in Randall plaque samples from humans. E GSEA revealed that the TGFβ signaling pathway is activated in the kidneys of model mice. F The expression of TGFβ1 is elevated in Randall’s plaque samples from humans. G The expression of TGFβ1 is elevated in the kidneys of model mice. H Calcium deposition in mouse kidneys was significantly reduced following the administration of a selective small molecule inhibitor of the TGFβ1 receptor, black bar indicates 500 μm. I WB verified altered EMT, TGFβ1 signaling pathways and Clusterin in the mouse model after receiving TGFβ1 receptor blockers. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

The TGFβ1 signaling pathway is also well known to trigger EMT, especially in the kidney [15, 16]. GSEA revealed significant positive associations between TGFβ pathway activation and both human disease and mouse models (Fig. 6D and E). Analysis of human bulk RNA-seq data indicated a significant increase of TGFβ1 in human kidney samples(Fig. 6F). Furthermore, in animal models, a significant positive association was observed between the expression levels of TGFβ1 and Clu (Figure S4E). The upregulation of TGFβ1 at the protein level under high-oxalate conditions has been confirmed in animal models (Figure S4F-4G). While TGFβ1 is the most extensively researched element of the TGFβ pathway, we assessed the expression levels of TGFβ2, TGFβ3, and other soluble factors linked to EMT in human samples and observed no significant differences (Figure S4H). In the mouse model, the selective TGFβ1 receptor inhibitor SB413542 was employed to inhibit the TGFβ1 signaling pathway, resulting in a significant reduction of calcium salt deposits in the model kidney (Fig. 6H). A WB panel demonstrated a significant reduction in the EMT phenotype within the mouse kidney following the inhibition of the TGFβ1 signaling pathway, which led to decreased upregulation of clusterin protein (Fig. 6I).

In the TGFβ1 signaling pathway, SMAD2 and SMAD3 serve as primary signal transducer molecules [17, 18]. Initially, nucleocytoplasmic protein isolation was employed to identify elevated levels of P-SMAD2, P-SMAD3, and SMAD4 within the cell nucleus under modeling conditions (Fig. 7A). Following 24 h of treatment of HK2 cells with 0.5 mM sodium oxalate, a significantly greater amount of P-SMAD2/3 proteins was observed to bind to SMAD4 in the experimental group compared to the control group. Similar findings were corroborated through reverse Co-IP experiments (Fig. 7B-D, Figure S5A-S5B). Biotinylated DNA probes were utilized to assess the binding of P-SMAD2/3 proteins to the promoter regions of the TWIST1 and CLU genes (Fig. 7E). To determine if the P-SMAD2/3 complex can activate the expression of TWIST1 and CLU genes upon nuclear entry, we examined publicly accessible ChIP-seq data from the Gene Transcription Regulation Database [13]. Our analysis revealed significant P-SMAD2/3 enrichment within 1000 bp upstream of the transcriptional start sites for both TWIST1 and CLU genes (Fig. 7F, Figure S5C). Based on the aforementioned evidence and the predicted binding sequences of P-SMAD2/3 to the promoter regions of TWIST1 and CLU, as indicated by the JASPAR database, we conducted subsequent chromatin immunoprecipitation (ChIP) assays (Fig. 7G, Figure S5D). ChIP‒PCR analysis revealed that P-SMAD2/3 binds to specific chromatin fragments within the TWIST1 and CLU promoter regions (Fig. 7H, Figure S5E). The ChIP‒qPCR experiments indicated that, under the specified modeling conditions, an increased binding of P-SMAD2/3 to the promoter regions of the TWIST1 and CLU genes was observed at BS2 and BS1, respectively (Fig. 7I and J, Figure S5F‒S5G). We finally validated the impact of TWIST1 and CLU knockdown on the TGFβ1 signaling pathway and EMT in HK2 cells (Figure S5H).

Fig. 7.

The TWIST1-CLU transcriptional axis can be initiated by the TGFβ1-SMAD2/3 pathway. A After 24 h of stimulation with 0.5 mM oxalate, SMAD2/3/4 translocation into the nucleus significantly increased in HK2 cells. B Significantly more P-SMAD3 and SMAD4 bound to SMAD2 after 24 h of treatment with 0.5 mM oxalate. C Significantly more P-SMAD2 and SMAD4 bound to SMAD3 after 24 h of treatment with 0.5 mM oxalate. D Significantly more P-SMAD2 and P-SMAD3 bound to SMAD4 after 24 h of treatment with 0.5 mM oxalate. The blots of B-D are displayed in Figure S5 A-B. E DNA pull-down revealed that P-SMAD2/3 proteins can bind to the TWIST1 promoter sequence. A similar relationship can also be found between the P-SMAD2/3 and CLU promoters. F ChIP-seq results indicated that the P-SMAD2/3 complex could initiate the transcription of the TWIST1 gene. Data were obtained from the Gene Transcription Regulation Database (http://gtrd.biouml.org/). G Schematic drawing showing the predicted binding sequence between the P-SMAD2/3 protein and the TWIST1 promoter on the basis of the JASPAR database 2022 (https://jaspar.elixir.no/). H ChIP‒PCR revealed that all three BSs are possible binding sites between P-SMAD2/3 and the TWIST1 promoter sequence. I ChIP‒qPCR revealed that P-SMAD2/3 can bind to the TWIST1 promoter at BS2. The ChIP‒qPCR threshold was set to FE = 4. J A significantly greater number of P-SMAD2/3 complexes bound to the TWIST1 gene promoter sequence at BS2 under high-oxalate conditions. K. Schematic pathway illustrating the role of TGFβ1-SMAD2/3-TWIST1-CLU in kidney stone formation. The data are presented as the means ± SEMs; *P < 0.05, **P < 0.01, ***P < 0.001

In conclusion, this study identifies CLU as a key mediator in CaOx kidney stone formation. CLU is highly expressed in kidney stone patients, animal models, and lithogenic environments, promoting CaOx crystal aggregation and growth. Targeted CLU inhibition reduces crystal deposition in vivo and in vitro. Mechanistically, high oxalate activates the TGFβ1-SMAD2/3 pathway, driving nuclear translocation of phosphorylated SMAD2/3, which binds to the TWIST1 promoter to upregulate TWIST1. TWIST1 then directly binds to the BS4 region of the CLU promoter, enhancing CLU transcription. This TGFβ1-SMAD2/3-TWIST1-CLU axis, linked to EMT under hyperoxaluric conditions, provides a novel regulatory mechanism underlying CaOx stone formation (Fig. 7K). These findings highlight CLU and its upstream regulatory pathway as potential therapeutic targets for kidney stone disease.

Discussion

Kidney stone disease results from the accumulation of solid material in the kidneys, constituting a significant healthcare challenge in numerous countries globally [19, 20]. Calcium oxalate stones constitute roughly 80% of all kidney stones [21]. Crystalline components in tubular fluids may induce ionic supersaturation, resulting in crystallization. Small crystals can adhere to the apical surface of renal epithelial cells and subsequently increase in size, ultimately leading to self-aggregation. The deposited crystals subsequently serve as nuclei for additional growth and stone formation [22]. Various molecules, including small chemical compounds and macromolecules such as proteins and glycosaminoglycans (GAGs), have been identified in urine and stone matrix. These molecules are believed to play a role in the stone formation process by binding to crystal surfaces, thereby promoting the aggregation and growth of stones [9].

CLU, referred to as lipoprotein J (ApoJ), SPG2, or TRPM-2, was initially identified and isolated from rat testes in 1979 [23]. This protein was also named as Clusterin due to its ability to aggregate blood cells in vitro [24]. Recent studies indicate that CLU exhibits contrasting roles across various diseases. CLU was initially identified as being associated with Alzheimer's disease, where it is crucial for the removal of misfolded proteins (e.g., β-amyloid) and cellular debris from tissues, leading to the perception of its neuroprotective effect [25, 26]. Clusterin is recognized as a marker of acute kidney injury, with some studies indicating a potential protective effect on renal function [27–31]. Nevertheless, the negative consequences linked to the aggregation of Clusterin proteins have been documented in atherosclerosis and age-related macular degeneration [32, 33]. This study proposes that, under hyperoxaluria stimulation, local kidney cells distribute significant amounts of Clusterin proteins to safeguard their functions. Excess Clusterin proteins present in the urine associate with small crystals within the urine. This process enhances crystal diameter and facilitates stone formation.

Knocking out the CLU gene in renal tubular epithelial cells of model mice using the Cre‒LoxP CKO system resulted in a notable decrease in calcium salt deposition. This experiment demonstrated the sequential relationship between the upregulation of Clusterin protein and stone formation. A clinical trial published in 2015 utilized ELISAs to quantify CLU levels in urine samples from 49 patients with kidney stones and 41 control subjects without kidney disease. The authors found that urine samples from the kidney stone group exhibited significantly elevated CLU levels compared to the control group [34]. Analysis of kidney stone samples through a proteomic technique revealed a substantial presence of CLU protein, indicating that Clusterin may play a direct role in both the kidney damage associated with stones and the formation of kidney stones [35]. Previous studies have established a significant association between Clusterin protein expression and kidney stones based solely on clinical data. This correlation has led to the introduction of a radiation-free kidney stone test that measures the amount of Clusterin protein in urine [36]. Therefore, the present work builds upon prior research while also complementing and refining existing incomplete theories.

In this work, we also analyzed the detailed mechanism of increased CLU protein expression in the renal tube epithelium in the presence of high oxalate concentrations by a variety of methods and identified the TGFβ1-SMAD2/3-TWIST1-CLU pathway. In the initial phase of pathway investigation, a significant correlation was observed between the TWIST1 and CLU genes. A 2012 study identified a transcriptional regulatory relationship between TWIST1 and CLU through ChIP and luciferase reporter gene assays in prostate cancer PC-3 cells [37]. Another study, also in the field of prostate cancer research, reported that TWIST1 could transcriptionally regulate CLU expression after insulin-like growth factor stimulation [38]. These two studies were well performed and provided encouraging results. Despite the existing evidence indicating a significant correlation between Clusterin and kidney stones, and the high expression of TWIST1 in the kidneys of affected patients, no studies have yet explored the role of the TWIST1-CLU transcriptional axis in stone formation under conditions that promote stone development [10]. TWIST1 has been widely shown to be associated with EMT, as many studies have demonstrated the relationship between EMT and stone formation; therefore, analyzing the role of EMT in stone formation was not our main aim in this study [14, 39]. However, we still used the data and knowledge accumulated in the field of EMT to confirm the importance of the TGFβ1 signaling pathway [40]. Many studies have analyzed the relationship between TGF1β family proteins and kidney stone formation, but most have focused on stone-induced renal fibrosis or renal EMT [41, 42]. In the present study, we did not focus on the effects of TGFβ1 on cell differentiation but rather on its effects on the downstream secretion of the Clusterin protein. Indeed, the TGFβ1 signaling pathway plays a crucial role in the expression of another important kidney stone protein component, namely, osteopontin [43, 44]. Therefore, targeting TGFβ1 may be an effective strategy if Clusterin-targeted drugs to treat the onset or recurrence of kidney stones cannot be developed [31]. It is noteworthy that the animal models employed in our study rely on acute chemical induction of hyperoxaluria, which does not fully recapitulate the slow, progressive nature of human nephrolithiasis. Nevertheless, the use of human stone samples and Randall's plaque tissues allowed us to identify mechanistic clues within a human context, which were subsequently functionally validated in the currently well-accepted animal model. The development of more physiologically relevant models in the future will help advance the exploration of pathogenesis in calcium oxalate kidney stones. Besides, kidney stones are more prevalent in men than in women, potentially due to hormonal or anatomical factors. To minimize the confounding effects of sex differences, the present study utilized only male mice in our experimental models.

In conclusion, CLU is a protein synthesized by renal tubular epithelial cells in response to hyperaciduria, mediated through the TGFβ1-SMAD2/3-TWIST1 signaling pathway. It can directly interact with calcium oxalate crystals, thereby contributing to the formation of renal calcium oxalate stones. The concentration of CLU protein in urine may serve as a biomarker for the early detection or non-invasive identification of renal calcium oxalate stones.

Methods

Animal modeling and conditional knockout

The protocols for the animal studies were reviewed and approved by the Experimental Animal Ethics Committee of West China Hospital (20211712A, 2017063 A).This study analyzes the hypothesis using both mouse and rat models. All animals utilized in the current study were male. This study conducted in vivo knockout experiments in mice utilizing the Cre-LoxP system for conditional knockout (CKO) experiments [45]. The gRNA of the mouse Clu gene, a donor vector containing the loxP site, and Cas9 mRNA were co-injected into fertilized mouse eggs to generate targeted CKO offspring. F0 founder mice were identified by polymerase chain reaction (PCR) and sequence analysis and crossed with wild-type mice to test germline transmission and F1 mouse production. F1-targeted mice were crossed with tissue-specific Cdh16-Cre gene-deleted mice to produce F2 [46]. Heterozygous Cre + mice were bred to homozygous mice. The primers used to identify the genotypes are shown in Table S2. A total of 24 mice (6 weeks old) were included in this study. The blank control group (Group 1) was fed a normal diet without any additional treatment. In the renal CaOx stone model group (Group 2), 80 mg/kg/day of glyoxylic acid was intraperitoneally injected for 7 days [47]. For Group 3, a renal CaOx stone model was established in Clu-CKO mice. For Group 4, the flox/flox control group, a renal CaOx stone model was established in flox/flox control mice. There was also a total of 24 C57BL6J mouse (6-week-old) with the same body weight included in the work. After one week of the adaptive process before the experiment, 24 mice were randomly assigned into four groups. Group 1. Blank control group (normal diet without any treatment). Group 2. Renal CaOx stone model group: 80 mg/kg/day of glyoxylic acid was intraperitoneally injected for 7 days. Group 3. TGFβ1 receptor selective inhibitor (TGFβ1i) + model group: 80 mg/kg/day of glyoxylic acid was intraperitoneally injected for 7 days + 10 mg/kg TGFβ1i intraperitoneal injection intervention 3 times/week. SB413542 (supplied by Abmole, USA) was used as TGFβ1i in this study, and the dosage was determined by previously published similar studies [48, 49]. SB413542 was dissolved by 10% DMSO + 40% PEG300 + 5% Tween-80 + 45% saline. Group 4. Solvent control group: 80 mg/kg/day of glyoxylic acid was intraperitoneally injected for 7 days + same volume of solvent used for SSB413542 described above three times/week. All four groups of mice received one week of modelling and were euthanized under 4% chloral hydrate.

This research employed six-week-old male Sprague–Dawley (SD) rats, establishing the renal calcium oxalate stone model via a continuous diet of 1% ethylene glycol for four weeks. Both the right and left kidneys were collected for subsequent analysis. Renal tissue sections were stained using Von Kossa (VK) staining with a commercially available kit (Solarbio, Beijing, China). The percentage of area occupied by deposits was calculated using Image J software version 1.52. Immunohistochemical staining was conducted on formaldehyde-fixed, de-paraffinized tissue sections using DAB substrate, with positive signals assessed via Image J software v1.52.

Single-cell RNA sequencing of mouse models

Renal cell suspensions with a concentration of 1,000 cells/µL were prepared and sequenced using 10 × Genomics ChromiumTM (Single Cell 3′ library and Gel Bead Kit v3) by following the manufacturer’s protocol as described in previous study [50]. The samples were sequenced on an Illumina Novaseq6000 instrument using 150 base paired-end reads. Subsequently, the scRNA-seq data of mouse model was clustered and annotated utilizing the following signature genes: B cells (Ms4a1, Cd19, Cd79a, Cd79b), T cells (Cd3d, Cd3e, Cd3g, Trbc2, Trbc1), Macrophages (Cd68, Adgre1, Itgam, Cd14), Endothelial cells (Pecam1, Cdh5, Cldn5, Kdr, Emcn, Egfl7), Smooth muscle cells (Acta2, Tagln, Myh11, Myl9, Rgs5), Fibroblasts (Dcn, Col1a1, Col3a1, Pdgfra, Fn1, Postn), Epithelial cells (Krt18, Cdh1, Epcam), Neutrophils (S100a8, S100a9, Cxcr2, Csf3r, Lcn2), Proximal tubular cells (Slc34a1, Lrp2, Glyat, Acsm2, Akr1c21), Distal tubular cells (Slc12a3, Pvald, Calb1, Abca13), Principal cells (Aqp2, Aqp3, Avpr2, Fxyd4, Cav1) and intercalated cells (Atp6v0d2, Atp6v1g3, Fam13a, Mme).

Re-analysis of GEO published Randall’s plaque RNA-sequencing dataset

The public GEO dataset was obtained through the NCBI website (dataset ID: GSE73680, www.ncbi.nlm.nih.gov/geo)[10]. By using primary offered matrix data to perform further analysis, we found that it was hard to obtain a perfect RNA expression comparison since the original matrix was in normalized signal form. Fortunately, original count files were offered for every single sample. Thus RNA-seq data re-analysis was based on raw data in this study. First of all, all control samples acquired from trauma and renal carcinoma patients were excluded to avoid potential sample selection bias [10]. After the primary sample selection, PCA analysis was performed to evaluate sequencing discrimination results. According to the primary PCA plot (Figure S1F), we found that some samples in both groups significantly deviated from the confidence ellipse, and we also excluded these samples from further analysis. In the differential expression gene selection, the threshold was set as fold change (FC) > 1.5 and P < 0.05, and all calculations were conducted with limma package in R. Gene Set Enrichment Analysis (GSEA) was performed with GSEA v4.1.0 software (http://www.gsea-msigdb.org/gsea/index.jsp), and hub-gene selection was performed with the help of String database (http://string-db.org) and cytoHubba plugin Cytospace v3.8.2 software. Additional weighted gene co-expression network analysis (WGCNA) was performed based on the WGCNA package according to the developer-offered protocol.

For the scRNA-seq data of human Randall’s plaque, the NCBI website was accessed through the dataset ID GSE176115. The scRNA-seq data underwent quality control (QC) using the Seurat package (version 5.0.3). Cells with gene expression levels below 200 or above ninety percent of the maximum gene count were excluded to ensure the data quality. Cells with mitochondrial gene content above 20% and those with erythrocyte-related gene content over 5% were also eliminated. To address batch effects, data from different sequencing batches were merged using the harmony algorithm. The scRNA-seq data was processed through a series of standard analytical steps, including data normalization, variable feature detection, scaling, PCA, and harmonized integration. Subsequently, the scRNA-seq data was clustered and annotated utilizing the following signature genes: B cells (MS4A1, CD19, CD79A), T cells (TRBC2, TRBC1, CD3E, CD2), Macrophages (ITGAM, CD14), Mast cells (TPSAB1, IL1RL1, CPA3, TPSB2), Endothelial cells (CDH5, EGFL7, CLEC14A), Smooth muscle cells (ACTA2, TAGLN, SMTN, TPM2), Fibroblasts (COL6A1, APCDD1, PTGDS, ELN), Epithelial cells (EPCAM, CDH1, KRT18, KRT19, KRT8, KRT7), Neutrophils (FCGR3B, CSF3R, S100A9, S100A8), Plasma cells (CD38, MZB1, JCHAIN), and Dendritic cells (CD1C, FCER1A, CLEC10A, CD86).

Protein extraction from kidney stones and proteomics based on mass spectrometry (MS)

After obtaining approval by the West China Hospital of Sichuan University Medical Research Ethics Committee (2018182) and signing Informed consent for clinical sample collection, five calcium oxalate monohydrate kidney stone samples were collected from West China Hospital, Department of Urology. All kidney stones were ground with a high-throughput tissue grinder in RIPA buffer containing protease and phosphatase inhibitors. In label-free MS analysis, the liquid chromatography-mass spectrometry analysis technique was applied. After the separation and ionization, different size of peptides was injected into the Orbitrap Exploris™ 480 Mass Spectrometry System for MS analysis. The MaxQuant was used to search all of the raw data thoroughly against the protein database (UniProt) [51].

Bulk RNA-sequencing

Total RNAs were extracted using Trizol Reagent (Invitrogen, CA, USA). The high-throughput sequencing and analysis for mRNA were carried out by Tsingke Biotechnology Co., Ltd (Beijing, China).

Protein Co-immunoprecipitation (Co-IP), DNA pulldown, and Chromatin immunoprecipitation (ChIP)

In this study, protein co-immunoprecipitation was conducted using the Thermo Pierce Co-Immunoprecipitation (Co-IP) Kit (26149, USA). After antibody immobilization, cell protein extraction, purification, and quantification using the BCA assay, the prepared cell lysate and antibody immobilized on resin were incubated at 4ºC overnight. Utilize the provided elution buffer to elute immunoprecipitated bait and prey proteins for subsequent analysis (Western Blot and Mass Spectrometry in this study).

The DNA pulldown was conducted using the BersinBio Bes5004 DNA pulldown kit from Guangzhou, China. Desthiobiotin-marked DNA probes, specifically the promoter sequences of CLU and TWIST1 (as detailed in Table S3), were synthesized by BersinBio and subsequently incorporated into the pre-extracted, purified, and quantified nuclear lysate. Following overnight incubation at 4ºC, the extracted proteins were eluted using the provided elution buffer and subsequently utilized for further Western blot analysis.

In this study, ChIP was performed with the help of with BersinBio Bes5001 ChIP kit (Guangzhou, China). The detailed ChIP protocol was the same as previously published literature [52]. Cells were cross-linked with 1% paraformaldehyde (Thermo, 28908, 16% w/v formaldehyde solution, USA) and subsequently quenched using a glycine solution. Chromatin fragmentation was accomplished through ultrasonication at 35% power, with a duration of 2 s of sonication followed by a 5-s pause, repeated over a period of 15 min. Immunoprecipitation was conducted overnight at 4ºC using ChIP grade anti-TWIST1 antibody, anti-SMAD2/3, and normal IgG (Millipore). Following elution, the primer sequences utilized for subsequent PCR are presented in Table S2. The candidate binding sequence between the specified transcription factor and the corresponding promoter sequences was predicted using the JASPAR database 2022 (https://jaspar.genereg.net/, Table S4).

Western blotting (WB), nuclear and cytoplasmic protein extraction

Proteins were extracted from cells or tissues lysed in RIPA buffer with protease and phosphatase inhibitors (Thermo Fisher Scientific, CA, USA) and quantified using a BCA protein assay kit (Thermo Fisher Scientific). Proteins were isolated using SDS-PAGE and subsequently transferred to a polyvinylidene fluoride (PVDF) membrane. The membranes were blocked with 5% BSA or milk for 2 h and subsequently incubated with the specified primary antibodies overnight at 4 °C, followed by HRP-conjugated secondary antibodies. A comprehensive list of all antibodies utilized in this study is provided in Table S5. Immunoreactivity was visualized utilizing ECL reagents from 4 A Biotechnology, China. Densitometry analysis was conducted utilizing Image J (National Institutes of Health, Bethesda, MD, USA). The relative optical density of the bands was quantified and labeled above the corresponding Western blot bands. The subcellular fractionation was performed using the Nuclear and Cytoplasmic Protein Extraction Kit P0028 from Beyotime Biotechnology, China.

Cell lines and reagents

The HK2 cell line was the main cell line used in this study, and it was purchased from the Cell Bank of the Academy of Science (Shanghai, China) and cultivated in the recommended medium with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 μg/mL streptomycin. Short tandem repeat (STR) profiling was used to identify that the HK2 cell used in this work was not contaminated by other cells. Sodium oxalate was purchased from Aladdin Biochemical Tech (Shanghai, China. The time for treatment was 24 h unless indicated.

Crystal adhesion assay

Interventions or treatments were applied to cell monolayers at high confluence (70–80%). Subsequently, the cells were washed with PBS, and a medium containing 100 μg/mL CaOx crystals was added to the wells, followed by a 15-min incubation. The cells were thoroughly washed again with PBS. The final images were acquired using the Celigo Imaging Cytometer (USA, Nexcelom Bioscience) following the washing process. The analysis of crystal-adhesion results was conducted using Image J software v1.52.

Vector and transfection

This study involved the transfection of TWIST1-RNAi, TWIST1-overexpression, CLU-RNAi, and plasmids utilized in a dual-luciferase reporter gene assay (shRNA from Shanghai, Genechem), along with the corresponding control RNA (shRNA-NC), into cells during the logarithmic growth phase. Transfection was conducted utilizing the Lipofectamine 3000 transfection reagent (Invitrogen, USA) or a lentiviral vector (Shanghai, Genechem) in accordance with the manufacturer's protocol. Table S6 presents the transfected sequences.

Enzyme-linked immunosorbent assay (ELISA)

After obtaining approvel by the West China Hospital of Sichuan University Medical Research Ethics Committee (2018182) and signing Informed consent for clinical sample collection, 20 urine samples were collected from West China Hospital, Department of Urology. In this work, ELISA was applied to determine the concentration of CLU in the urine samples of kidney stone patients and healthy control people. After signing Informed consent for clinical sample collection, 20 urine samples were collected (10 for kidney stone patients and 10 for health control). Human CLU(Clusterin) ELISA Kit based on the double-antibody sandwich method was purchased from Elabscience Biotechnology (Wuhan, China), and we followed the producer’s protocols.

Crystal pulldown assay

The crystal pulldown experiment was a newly designed experiment based on previously published literature [53]. There were two types of crystal pulldown experiments in this work, and we offered a schematic drawing in Fig. 3A. The same volume of 1 mM Na₂C₂O₄ was added into 5 mM CaCl2 and incubated at room temperature for one hour to form calcium oxalate monohydrate crystals [54]. For crystal-forming pulldown, different amounts of recombinant Clusterin protein (reCLU) were added in CaCl2 before adding Na₂C₂O₄. In the crystal pulldown, reCLU should be added after the crystal formation and incubated for another one h. After the pulldown, fluid in the tube was discarded and the crystals were washed with Tris–HCl (PH 7.4) five times to clean uncombined reCLU.

Mendelian randomization (MR) analysis

MR is a data analysis technique for assessing etiological inference in epidemiological studies that uses genetic variants strongly correlated with exposure factors as instrumental variables to assess causal relationships between exposure factors and outcomes [55]. For CLU expression, genome-wide association study (GWAS) data were obtained from the eQTLGen Consortium [56]. For kidney stones, two GWAS datasets, FinnGen [57] and the UKBioBank [58], were used in this study. MR was performed based on the TwoSampleMR R package with P1 < 5*10^−8, P2 < 5*10^−5, r2 < 0.001, and kb = 10000 [59]. MR meta-analysis was further conducted with RevMan software. Detailed information can be found in Table S8.

Statistical analysis

The continuous data in this study are presented as the mean ± standard error (SEM) unless otherwise noted. As appropriate, comparisons were performed using Student’s t-test, one-way ANOVA, or the Mann‒Whitney U test. Pearson correlation analysis was used to evaluate the correlation between the two groups. All the statistical analyses were carried out with SPSS (version 23.0; SPSS, Inc., Chicago, IL, USA) and GraphPad Prism (version 8.0; La Jolla, CA, USA).

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CaOx

Calcium oxalate

RTEC

Renal tubular epithelial cell

FBS

Fetal bovine serum

CKO

Conditional knockout

scRNA-seq

Single-cell RNA sequencing

reCLU

Recombinant CLU protein

MR

Mendelian randomization

GWAS

Genome-wide association study

WGCNA

Weighted gene co-expression network analysis

SEM

Standard error of mean

ELISAs

Enzyme-linked immunosorbent assays

DEGs

Differentially expressed genes

WB

Western blot

PCR

Polymerase chain reaction

ChIP

Chromatin immunoprecipitation

EMT

Epithelial-mesenchymal transition

TF

Transcription factor

PCA

Principal component analysis

Author contributions

Conceptualization and Methodology: KJW, XJ; Data curation and Project administration: YCM, LHL, ZYJ, LLD, YL; Investigation and formal analysis: YCM, LHL, ZYJ, LLD, YL, YQY, XNL, DT, ZJY, JW, LYX; Manuscript Writing- Original draft: YCM, XJ; Manuscript editing and manuscript review: KJW.

Data availability

Raw data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Conflict of interest

All authors have declared that no conflict of interest exists.