Circulating RAC1 contributed to steroid-sensitive nephrotic syndrome: Mendelian randomization, single-cell RNA-sequencing, proteomic, and experimental evidence

Abstract

Objectives

The small GTPase Rac1 (RAC1) has been linked to podocyte disorders and steroid-sensitive nephrotic syndrome (SSNS). The aim of this study was to explore and validate the potential causal association between circulating RAC1 and SSNS.

Methods

The association between circulating RAC1 and SSNS at both gene expression and proteomic levels was investigated using Mendelian randomization analysis, and further validated by single-cell RNA-sequencing, proteomic analysis, and experimental studies. The genetic instruments comprised cis-expression quantitative trait loci (cis-eQTLs) associated with RAC1 gene expression and protein QTLs correlated with plasma RAC1 protein levels. Causal associations were estimated utilizing the inverse variance weighted and MR-PRESSO methods. Validation of RAC1 expression was conducted through single-cell RNA-sequencing of peripheral blood mononuclear cells from patients with SSNS and healthy controls. Proteomic analysis was performed among patients with minimal change nephrotic syndrome. Experimental validation was conducted using a puromycin aminonucleoside (PAN)-induced nephrosis model.

Results

Increased expression of RAC1 was associated with a higher risk of SSNS (gene expression level: odds ratio [OR], 1.53; 95% confidence interval [CI], 1.02–2.28; protein level: OR, 1.82; 95% CI, 1.05–3.17). The results of MR-PRESSO were consistent (gene expression level: OR, 1.49; 95% CI, 1.17–1.92; protein level: OR, 1.81; 95% CI, 1.16–2.85). Single-cell RNA sequencing and proteomic analysis confirmed elevated RAC1 expression in patients with SSNS compared to healthy controls. Experimental data further supported increased RAC1 expression in PAN-induced nephropathy.

Conclusions

Increased expression of RAC1 might be causally associated with SSNS, suggesting that targeting RAC1 might represent a potential therapeutic strategy for SSNS.

1. Introduction

Nephrotic syndrome, characterized by massive proteinuria and hypoalbuminemia, stands as the most prevalent glomerular disease affecting both children and adults [1]. Steroid-sensitive nephrotic syndrome (SSNS), typified by complete remission within several weeks of corticosteroid administration, is a common phenotype in most pediatric NS and adult podocytopathies [2]. SSNS is predominantly attributed to podocyte disorders, with the majority of patients exhibiting as minimal change disease (MCD) and less than 20% of cases presenting with more severe focal segmental glomerulosclerosis (FSGS) upon renal biopsy [3]. Various circulating factors and pathogenic molecules have been hypothesized to play roles in the onset and relapse of SSNS [3,4]. Dysregulation of immune cells and potential circulating factors have also been linked to podocyte injury by influencing the rearrangement of the podocyte cytoskeleton [5]. The formation and maintenance of the foot process actin network rely on various signaling molecules, including the Rho family of small guanosine triphosphatases (Rho-GTPases), which play a crucial role in regulating the podocyte actin cytoskeleton and mediating NS pathogenesis [6].

Ras-related C3 botulinum toxin substrate 1 (RAC1), a member of the Rho-GTPases, has been implicated for inducing persistent proteinuria in a dose-dependent manner, spanning from MCD to FSGS [7]. Furthermore, RAC1 was found to be overexpressed in kidney biopsies from patients with MCD and idiopathic FSGS, and sera from patients with idiopathic FSGS and MCD activated the RAC1 pathway in cultured podocytes, while RAC1 was not activated by serum from MCD patients in remission [7]. These findings supported the pathogenic role of circulating factors in podocytopathies. Interestingly, circulating plasma extracellular vesicles (EVs) from patients with relapse SSNS exhibited significantly higher RAC1 levels compared to those from healthy controls and patients in disease remission. Moreover, circulating EVs from patients with relapse SSNS were efficiently internalized by podocytes and induced significantly enhanced motility and albumin permeability [8]. These findings suggested a potential correlation between peripheral circulating RAC1 and podocyte injury in patients with SSNS. However, the causality of this relationship has yet to be determined.

Mendelian randomization (MR) design is an epidemiological approach that strengthens causal inference regarding the outcome of interest by utilizing genetic variants as instrumental variables for the exposure [9]. In this study, we aimed to explore the causal association between circulating RAC1 gene expression and protein levels and SSNS using the MR approach. Additionally, the results of the MR analysis were further validated by analyzing RAC1 expression in peripheral blood mononuclear cells (PBMCs) derived from pediatric patients with idiopathic SSNS and healthy volunteers using single-cell RNA-sequencing data. Given that MCD is the predominant pathological feature of SSNS, we also presented the proteomic characterization of RAC1 in patients with MCD using publicly available data. Additionally, RAC1 expression was assessed in a puromycin aminonucleoside (PAN)-induced nephrosis model, which exhibits histopathological features similar to MCD.

2. Materials and methods

2.1. Ethics

This study involved a secondary analysis of publicly available data sourced from genome-wide association studies (GWAS), single-cell RNA-sequencing study, and proteomic research. The original studies had obtained all necessary ethical approvals and informed consents from the participants, and no patient identification information was used in the study. Therefore, no additional ethical approvals or informed consents were required for this study.

2.2. MR analysis

2.2.1. Aim and design

The MR study first investigated the causal association between peripheral RAC1 expression and SSNS using expression quantitative trait loci (eQTLs) as instrumental variables. eQTLs represent genetic variants with genotype-specific effects on gene expression levels, specifically, only cis-eQTLs, which are single nucleotide polymorphisms (SNPs) located at the gene of interest, influencing its expression levels, were utilized for the MR analysis [10]. The analysis strategy was grounded in the principle that alleles of eQTLs significantly correlated with the expression level of the gene can induce corresponding changes in the gene product. By employing RAC1 expression levels as the exposure, this study aimed to identify its causal associations with SSNS. As a validation for this potential association, we also conducted MR analysis using RAC1 protein QTLs (pQTLs) as instrumental variables to confirm the causality between circulating RAC1 protein level and SSNS. In order to clarify whether there is a bi-directional association, reverse MR analysis was also conducted to investigate the effect of SSNS on RAC1 expression. The MR analyses was conducted in adherence with the statement for strengthening the reporting of observational studies in epidemiology using Mendelian randomization (STROBE-MR) [11].

2.2.2. Data source and genetic instruments identification

We initiated our investigation by assessing the tissue-specific expression profile of RAC1 in humans through the Genotype-Tissue Expression Portal [12]. This exploration aimed to validate the broad expression of RAC1 across various tissues, including whole blood and major organs.

Genetic instruments used for MR analysis were selected strictly, adhering to the following three assumptions of MR design: 1) eQTLs and pQTLs are reliably associated with RAC1 gene expression and protein levels; 2) independent of potential confounders that might be associated with both RAC1 and SSNS; 3) not directly associated with SSNS [13]. The study process and genetic instruments identification procedure are illustrated in Figure 1.

Figure 1.

Mendelian randomization assumptions and study process. MR, Mendelian randomization; RAC1, ras-related C3 botulinum toxin substrate 1; IV, instrumental variable; SSNS, steroid-sensitive nephrotic syndrome; eQTL, expression quantitative trait loci; pQTL, protein quantitative trait loci; SNP, single nucleotide polymorphism; GWAS, genome-wide association study.

To identify cis-eQTLs associated with RAC1 expression, we utilized GWAS summary data from the eQTLGen Consortium. The eQTLGen Consortium conducted GWAS to examine the correlation between over 11 million genetic variants and the expression levels of 16,897 genes, sourced from 31,684 individuals predominantly of European ancestry (Table 1) [14]. In this study, only eQTLs located at the RAC1 gene (Chr 7: 6374527 - 6403967) and associated with its expression at genome-wide significance (p value < 5 × 10⁻⁸) were considered. Subsequently, we further refined our selection by excluding SNPs with genetic linkage disequilibrium (LD), utilizing a clumping cutoff of r² at 0.3 within a 10,000 kilobase window based on a reference panel of European ancestry from the 1000 Genomes Project (phase 3) [19]. Instrument strength was assessed using the F statistic, and SNPs with an F statistic of less than 10 were removed. In addition, eQTLs associated with potential confounding factors, such as RHoA, cdc42, Klotho, CLIC5, CRK, ARF6, PLAUR, Piezo1, FHL2, EXT1, KANK1, KANK2, PLCE1, and PLCG2, were excluded [20–24]. Additionally, SSNS related factors, including demographic characteristics, upper respiratory tract infection [25], and birth weight [26], were also removed (Supplementary Table 1). After harmonizing the exposure and outcome data, palindromic variants and those directly associated with SSNS were further eliminated (p > 0.05). Finally, the cis-eQTLs used to investigate the causality between RAC1 and SSNS were identified.

Table 1.

Data source for Mendelian randomization study, single-cell RNA-sequencing, and proteomics.

	Participants	No. participants/samples	Characteristics (age; sex)	Location/ethnicity	Sequencing sample	Data source
RAC1 eQTLs [14]	Population based	31,684	37 datasets without detailed information of age and sex	Majority European (89.45%)	Whole blood (80.4%) or PBMC (19.6%)	The eQTLGen Consortium, dataset: Full cis-eQTL summary statistics
RAC1 pQTLs [15]	Population based	35,559	55 ± 17 years; female (57%)	Iceland	Plasma	The deCODE Genetics, dataset: Proteomics_PC0_2870_29_RAC1
SSNS GWAS [16]	Cases with MCD or non-biopsied SSNS, and controls	2440 cases and 36023 controls	12 datasets without detailed information of age and sex	Multi-Ancestry population	Whole blood	GWAS Catalog, dataset: GCST90258619
Single-cell RNA sequencing [17]	Pediatric idiopathic SSNS and healthy controls	3 cases and 4 controls	Cases: 10.1 ± 4.5 years, female (33%); Controls: 11.8 ± 3.0 years, female (25%)	Canada	PBMC	Gene Expression Omnibus, dataSet, GSE233275
Proteomics analysis [18]	Patients with biopsy-proven MCD and healthy controls	15 cases and 20 controls	Cases: median age, 44 years, female (60%); Controls: median age, 41 years, female (35%)	France	Serum	ht tps:/ /w w w.ncbi.nlm.nih.gov/pmc/articles/PMC9263421/: Supplementary Table 1

RAC1, ras-related C3 botulinum toxin substrate 1; SSNS, steroid-sensitive nephrotic syndrome; eQTL, expression quantitative trait loci; pQTL, protein quantitative trait loci; PBMC, peripheral blood mononuclear cells, GWAS, genome-wide association study, MCD, minimal change disease.

The pQTLs data of RAC1 were sourced from a large-scale GWAS study for plasma proteome among 35,559 Icelanders conducted by the deCODE Genetics, involving 27.2 million sequence variants and 4,907 circulating proteins [15]. Considering that few SNPs were associated with plasma RAC1 protein levels at genome-wide significance, we initially extracted pQTLs at a locus-wide significance threshold (p value <1 × 10⁻⁵) and then excluded SNPs with LD using the same parameters as for eQTL selection. Subsequently, we ensured that the remaining SNPs met MR assumptions by verifying instrument strength (F > 10), independence from confounding factors and outcomes, and non-palindromic characteristics.

For genetic instruments used for reverse MR analysis, we firstly isolated SNPs that were significantly correlated with SSNS using a cutoff p value of 5 × 10⁻⁸. The next steps were to exclude genetic variations with LD (r² <0.001, kb = 10000), palindromic SNPs, variants with low instrument strength, and those directly related to confounding factors or RAC1. Finally, the instrumental variables for reverse MR analysis that preliminarily met the three assumptions of MR were selected.

2.2.3. Outcome

GWAS summary data for SSNS were acquired from a recent multi-population GWAS study for SSNS, encompassing 2,440 cases and 36,023 controls [16]. This extensive GWAS study aimed to discover and more fully characterize disease-associated genetic variations. The case cohort was defined as either biopsy-proven MCD or non-biopsied SSNS, with proteinuria >0.25 g/mmol or urinary protein to creatinine ratio (uPCR) ≥2.0, serum albumin <25 g/L, complete remission within 4–6 weeks after steroid administration, and an age of onset <21 years old. Detailed clinical information for the enrolled subjects is shown in Table 1.

2.2.4. MR statistical analysis

We conducted two-sample MR analysis to evaluate the causality between RAC1 and SSNS. The Wald ratio was utilized to calculate the effect of each set of exposure on outcome for each instrumental variable, and then the effect size was combined using the inverse variance weighted (IVW) method to generate an overall estimation of the causal effect, when two or more instrumental variables were available [27]. Cochran’s Q test was employed to assess heterogeneity among multiple SNPs. If heterogeneity was present, the random-effects model of the IVW method was used, otherwise IVW method with fixed effects was applied [28]. The estimates from IVW method were considered as the primary results of the causal effects in the absence of both heterogeneity and horizontal pleiotropy [29]. The odds ratios (OR) and 95% confidence intervals (CI) were provided as the results of MR analyses. A causal effect was deemed significant when the p value was less than 0.05. To address potential pleiotropy, we conducted MR-Egger regression to test for any pleiotropic effect. The MR pleiotropy residual sum and outlier (MR-PRESSO) global test were also utilized to identify and rectify horizontal pleiotropic outliers, thereby ensuring a precise estimate [30]. Accordingly, the results of the main estimates of MR-PRESSO were considered as part of the sensitivity analysis. As another part of sensitivity analysis, the leave-one-out method was adopted to assess the impact of each SNP on the primary MR results. We also conducted multivariate MR analysis (MVMR) by incorporating demographic characteristics, upper respiratory tract infection, and birth weight as adjustments, to minimize the impact of confounding factors.

2.2.5. MR software and packages

All statistical analyses and visualizations were conducted using R software (version 4.3.3). Two-sample MR analysis was performed using the ‘TwoSampleMR’ package, and MR-PRESSO analysis was conducted with the ‘MRPRESSO’ package. Scatter plots, single SNP effect plots, and leave-one-out plots were generated using the ‘TwoSampleMR’ package. MVMR was carried out using the ‘MendelianRandomization’ and ‘MVMR’ packages. The forest plots were created using the ‘forestploter’ package.

2.3. Single-cell RNA-sequencing

2.3.1. Participants and samples collection

The single-cell RNA-sequencing data of PBMCs derived from pediatric idiopathic SSNS and age-matched healthy controls were obtained from the Gene Expression Omnibus (Dataset: GSE233277) (Table 1). The original study was conducted by Al-Aubodah TA et al. to characterize the transcriptional signature of the nephrotic immune cell response and identify the contributing pathogenic cells [17]. In this study, one patient with a first presentation of SSNS and two children with active SSNS relapse were selected as the SSNS group. This group had an average age of 10.1 years and an average uPCR of 0.49 g/mmol. Four age- and sex-matched healthy volunteers, with an average age of 11.8 years and an average uPCR of 0.01 g/mmol, were identified as the control group (Supplementary Table 2). Blood samples were collected during hospital visits.

2.3.2. Single-cell and RNA preparation

The process of preparing single-cell RNA from PBMCs was as follows: Cryopreserved PBMCs were thawed and rested for 2 h at 37 °C in 5% CO₂. The cells were then washed and resuspended in PBS with 2% FBS before undergoing fluorescence-activated cell sorting (FACS) using the BD FACSAria Fusion. Live cells were isolated based on size and granularity. Sorted cells were washed twice in PBS with 0.04% BSA and adjusted to a concentration of 1,000 cells/µl. Samples were then processed according to the 10x Genomics Single Cell 5′ v1.1 user guide. Complementary DNA (cDNA) and 5′ gene expression libraries were generated following the standard 10x Genomics protocol. Libraries were sequenced using the NovaSeq 6000 (Illumina). Detailed methodologies can be found in the original study [17].

2.3.3. scRNA-seq data preprocessing and quality control

Data analysis was performed using the ‘Seurat’ package of R software. For quality control, doublet and nonviable cells were removed by excluding cells expressing <300 or >5000 genes, >20% mitochondrial genes, and >5% hemoglobin genes.

2.3.4. Generating scRNA-seq clusters

The scRNA-seq data were normalized and harmonized, and variable features were identified for each sample separately after quality control. To generate clusters consistently present across all samples, principal component analysis was used for integrated clustering in Seurat. Differential gene expression analysis between individual subclusters was performed to determine cell type subclusters, utilizing the FindMarkers function in the Seurat package. Cell types were then identified by their expression of the top ten marker genes (Supplementary Table 3). UMAP clustering was then completed using a resolution of 1.0.

2.3.5. RAC1 expression in single-cell RNA-seq

RAC1 expression in each group and cell type was assessed using the FindMarkers function and visualized with the DotPlot function. Adjusted p values for differential gene expression analyses were calculated using Wilcoxon rank-sum tests with Bonferroni correction, and a p value < 0.05 was considered significant and depicted on the graphs.

2.4. Serum protein signature of MCD

Serum protein signatures were identified using aptamer-based proteomics in 15 patients with biopsy-MCD (6 males and 9 females; mean age, 43 years) from Hôpital Tenon, Paris, France, and 20 healthy controls (13 males and 7 females; mean age, 41 years) from the Etablissement Français du Sang in Lyon, France. All participants with MCD presented with nephrotic syndrome (median serum albumin, 17.1 g/L; median urinary protein, 6.6 g/24 h). The original study was conducted by Muruve et al. [18]. SOMAscan proteomics analysis of serum samples was performed at the BIDMC Genomics, Proteomics, Bioinformatics, and Systems Biology Center (Boston, MA). This analysis measures the expression of 1,305 human proteins using highly selective, single-stranded, modified Slow Off-rate Modified DNA Aptamers. Detailed protocols for proteomics, data quality control, signal calibration, and hybridization control normalization were described in the original study [18]. Differential protein expression was assessed using t-tests with Benjamini and Hochberg correction for multiple hypothesis testing.

2.5. Experimental validation

2.5.1. Experimental design

Building on existing evidence that links RAC1 overactivity or overexpression to nephrotic syndrome [7,31], we employed a classic model of MCD using PAN rats, with glucocorticoid intervention to validate RAC1’s role in podocyte injury. The animal experiments were conducted in accordance with the Ethics Committee of Shanghai Yueyang Hospital of Integrated Traditional Chinese and Western Medicine (NO. YYLAC-2023-218-1). Male Sprague–Dawley rats weighing 120–150 g were purchased from Shanghai SLAC Laboratory Animal Co., Ltd (Shanghai, China). The animals were acclimatized for seven days under standard conditions (temperature 24–26 °C, humidity 50–60%, 12-h light/dark cycle) with free access to water and standard chow. Rats were randomly assigned to three groups: control (n = 6), PAN (n = 6), and PAN with glucocorticoid treatment (PAN + GC, n = 6). PAN-induced nephrosis was established by a single intravenous injection of PAN (50 mg/kg, #HY-15695, MedChemExpress LLC, New Jersey, USA), while the control group received saline intravenous injections. Following modeling, the GC group was administered 5 mg/kg prednisone by gavage for one week, while the control and PAN groups received sterile distilled water. Urine samples were collected every 24 h for urinary protein and creatinine measurements on days 0, 4, and 7, using metabolic cages. All rats were euthanized on day 8, and blood and renal cortex samples were collected.

2.5.2. Serum enzyme-linked immunosorbent assay (ELISA)

Serum levels of RAC1 were measured using commercially available rat ELISA kits (#JL15640, JONLNBIO, Co., Ltd, Shanghai, China).

2.5.3. Real-time quantitative polymerase chain reaction (RT-qPCR)

RT-qPCR analysis was performed to detect the amplification of serum RAC1. The total RNA was extracted using Trizol reagent and reverse-transcribed into cDNA (#G3013 and #G3330, Servicebio Biotechnology Co., Ltd, Wuhan, China)) according to the manufacturer’s instructions. The CFX Connect fluorescent quantitative PCR system (Bio-Rad, USA) was utilized for RT-qPCR amplification. Cel-miR-39-3p was used as the internal reference. The primers were synthesized by Servicebio Biotechnology (Wuhan, China) and are shown in Table 2. We quantified the relative mRNA expression using the 2^−ΔΔCT method. The amplification and melting curve of RT-qPCR are shown in Supplementary Figure 1.

Table 2.

Sequences of primers for RT-qPCR analysis.

Gene name	Forward primer (5′-3′)	Reserve primer (5′-3′)
RAC1	GGGATGATAAGGACACGATTGA	GCTGAGCACTCCAGGTATTTGAC
cel-miR-39-3p	ACACTCCAGCTGGGTCACCGGGTGTAAATC	TGGTGTCGTGGAGTCG

RAC1, ras-related C3 botulinum toxin substrate 1.

2.5.4. Histopathological examination

Renal cortex samples were fixed in 4% paraformaldehyde and sent to Servicebio Biotechnology Co., Ltd (Wuhan, China) for periodic acid–Schiff (PAS) staining. Glomerular changes were observed using a conventional light microscope (Olympus, Tokyo, Japan).

2.5.5. Transmission electron microscopy

Tissue blocks with a size of 1mm³ were fixed in pre-cooled 2.5% glutaraldehyde and sent to Servicebio Biotechnology Co., Ltd (Wuhan, China) for processing and slicing. Images were acquired using a transmission electron microscope (HITACHI TEM system 7800, Tokyo, Japan). Podocyte foot processes were analyzed at a magnification of ×6000. The number of filtration slits per 20 μm length of the glomerular basement membrane was calculated for each group, to evaluate the morphological injury of podocytes [32].

2.5.6. Statistical analysis for experimental data

All values are presented as means with standard errors. Group differences were analyzed using one-way ANOVA with post hoc Holm-Sidak multiple comparisons correction. Time-dependent data were analyzed using repeated-measures ANOVA followed by multiple comparison tests. All data were analyzed and presented using GraphPad Prism 9.0 software (GraphPad Software Inc., San Diego, CA, USA), with p values <0.05 considered statistically significant.

3. Results

3.1. Identification of genetic instruments

Before constructing genetic instruments for RAC1 expression, we confirmed the ubiquitous expression of RAC1, with similar expression levels observed in the kidney, liver, heart, and whole peripheral blood (Figure 2).

Figure 2.

Tissue-specific expression profile of RAC1 in humans including in the whole blood.

A total of 89 SNPs, located at the RAC1 gene and associated with RAC1 expression at genome-wide significance, were initially chosen as potential genetic instruments for evaluating the causality between RAC1 and SSNS risk. After excluding 83 SNPs with LD and 1 variant with palindromic characteristic, 5 cis-eQTLs emerged as the final instrumental variables for investigating the causal association between RAC1 and SSNS. These cis-eQTLs showed sufficient validity and were not associated with potential confounders and outcome (Supplementary Table 4).

Due to limited genetic variants associated with plasma RAC1 protein levels at the genome-wide significance, 265 trans-pQTLs with a p value of less than 1 × 10⁻⁵ were initially selected. After removing 239 SNPs with LD and 7 palindromic variants, 19 pQTLs were ultimately identified for assessing the causal correlation between plasma RAC1 protein and the risk of SSNS. All the selected SNPs exhibited satisfactory validity, with an F statistic greater than 10, and no association between the selected SNPs and potential confounders was found. Additionally, all the 19 p-QTLs were not directly associated with SSNS (p > 0.05). Detailed characteristics of these selected genetic instruments are shown in Supplementary Table 5.

A total of 5,179 SNPs were identified as being associated with SSNS using a cutoff p value of 5 × 10⁻⁸. Of these, 5,166 SNPs were found to be in LD, and 2 SNPs were palindromic variants. The remaining 11 SNPs had F statistics greater than 10 and were not associated with confounders or RAC1 pQTL. These SNPs were identified as suitable instrumental variables for reverse MR analysis (Supplementary Table 6). No eQTL data were available for reverse MR analysis between SSNS and RAC1 gene expression.

3.2. RAC1 and the risk of SSNS

Using five cis-eQTLs as genetic instruments, MR results suggested that elevated gene expression of RAC1 was associated with an increased risk of SSNS (OR, 1.53; 95% CI, 1.02–2.28, p = 0.040) (Figure 3, Supplementary Figures 2(a) and 3(a)). Consistently, a higher level of plasma RAC1 protein was found to be associated with an increased risk of SSNS using 19 selected pQTLs (OR, 1.82; 95% CI, 1.05–3.17, p = 0.032) (Figure 3, Supplementary Figures 2(b) and 3(b)).

Figure 3.

Mendelian randomization results of the association between RAC1 and SSNS. MR, Mendelian randomization; RAC1, ras-related C3 botulinum toxin substrate 1; SSNS, steroid-sensitive nephrotic syndrome; OR, odds ratio; CI, confidence interval; MR-PRESSO, Mendelian randomization pleiotropy residual sum and outlier.

3.3. Reverse MR analysis between SSNS and RAC1

A reverse MR analysis was conducted to explore the potential bi-directional causal correlation between SSNS and RAC1. Due to the lack of available eQTL data, reverse MR was only performed between SSNS and the proteomic level of RAC1. Using the 11 selected SNPs to conduct reverse MR analysis, we did not find any association between SSNS and plasma RAC1 protein level (OR, 0.99, 95% CI, 0.98–1.01, p = 0.088) (Figure 3, Supplementary Figures 2(c) and 3(c)).

3.4. Sensitivity, heterogeneity, and pleiotropy analysis

The results of MR-PRESSO were consistent with the estimates from the IVW method, indicating a positive causal association between RAC1 and SSNS at the gene and protein levels (eQTL: OR, 1.49; 95% CI, 1.17–1.92, p = 0.022; pQTL, OR, 1.81; 95% CI, 1.16–2.85, p = 0.016) (Figure 3). The leave-one-out sensitivity analysis also demonstrated a positive trend in the association between RAC1 and the risk of SSNS (Supplementary Figure 2(d–f)).

Cochrane’s Q test, MR Egger regression, and MR-PRESSO global test were employed to evaluate the heterogeneity and pleiotropy of the MR analyses. No evidence of heterogeneity and pleiotropy was found in all the performed MR analyses between RAC1 and SSNS (Figure 3).

3.5. Multivariate MR analyses

The MVMR results showed a positive correlation between circulating RAC1 and SSNS at both mRNA level (OR, 1.16; 95% CI, 0.98–1.37, p = 0.083) and protein level (OR, 1.02; 95% CI, 0.92–1.13, p = 0.653); however, neither association reached statistical significance (Supplementary Table 7). Additionally, reverse MVMR analysis did not demonstrate a causal effect of SSNS on circulating RAC1 levels (OR, 1.00; 95% CI, 0.99–1.01, p = 0.821) (Supplementary Table 7).

3.6. Single-cell RNA-sequencing of PBMC from SSNS

After several quality control steps (Supplementary Figure 4), the transcriptomes of 27,892 immune cells in PBMCs from participants with active SSNS and 37,526 immune cells from healthy controls were analyzed. Following integrated clustering, 10 distinct immune cell types were identified, including T cells, B cells, natural killer cells, monocytes, and their subtypes (Figure 4(a)), and RAC1 was found to be widely expressed in these immune cells (Figure 4(b)). Compared with healthy controls, the expression of RAC1 in peripheral immune cells of SSNS patients was significantly increased (Figure 4(c)). Considering the potential pathological roles of B cells and T cells, particularly CD8+ T cells, in the pathogenesis of SSNS [33–35], we compared RAC1 expression in these immune cells. Our analysis revealed that RAC1 expression was significantly elevated in peripheral blood B cells, as well as CD4+ and CD8+ T cells, among patients with SSNS compared to healthy individuals (Figure 4(d)).

Figure 4.

Single-cel RNA sequencing analysis of peripheral blood mononuclear cells from SSNS and healthy controls. (a) UMAP of peripheral blood mononuclear cells from SSNS and healthy volunteers; (b) RAC1 expression in peripheral blood mononuclear cells; (c) Dotplot for differential expression of RAC1 in peripheral blood mononuclear cells between SSNS and healthy controls; (d) Violin plot for differential expression of RAC1 in different immune cell subtypes between SSNS and healthy controls. SSNS, steroid-sensitive nephrotic syndrome; HC, healthy controls.

3.7. Serum RAC1 expression using proteomic analysis

SOMAscan analysis of 1,305 proteins was performed using serum samples from patients with minimal change nephrotic syndrome and healthy controls. The differential expression analysis revealed a significant increase of serum RAC1 protein in patients with MCD compared to healthy individuals (fold change: 2.28, p = 1e − 05) (Figure 5(a)).

Figure 5.

Serum protein signature of RAC1 in minimal change nephrotic syndrome and experimental verification. (a) Serum RAC1 expression in patients with minimal change nephrotic syndrome versus healthy controls; (b) changes in urinary protein creatinine ratio of experimental animals; (c) periodic acid–schiff staining of renal tissue of experimental animals; (d) ultrastructure of podocytes and the number of filtration slits per 20 μm length of glomerular basement membrane; (e) serum RAC1 protein levels using ELISA; (f) serum RAC1 expression using RT-qPCR. RAC1, ras-related C3 botulinum toxin substrate 1; MCNS, minimal change nephrotic syndrome; HC, healthy control; PAN, puromycin aminonucleoside; GC, glucocorticoid.

3.8. Serum RAC1 expression in PAN induced nephrosis

A significant increase in uPCR was observed at 8 days after a single intravenous injection of PAN in male Sprague–Dawley rats (Figure 5(b)). Previous studies have shown that PAN-induced nephrosis did not cause apparent changes in glomerular structure under light microscopy; however, ultrastructural analyses have demonstrated podocyte injury [32]. Consistent with these findings, no significant differences were observed in glomeruli in PAS-stained kidney sections among the three groups (Figure 5(c)). However, podocyte foot process effacement, the typical histopathological feature of MCD, was observed in the PAN models, and glucocorticoid treatment alleviated this pathological injury (Figure 5(d)). ELISA results revealed that peripheral blood RAC1 expression was elevated in the PAN group compared to the control and glucocorticoid-treated groups (Figure 5(e)), and RT-qPCR results similarly showed a significant increase in circulating RAC1 expression in PAN-induced nephrosis (Figure 5(f)).

4. Discussion

SSNS, while often responsive to glucocorticoids with complete remission achievable within weeks of treatment initiation, remains a clinical challenge due to associated adverse effects and frequent relapses [2,36]. Understanding the underlying pathological mechanisms and identifying therapeutic targets are essential in addressing this challenge, particularly considering the pivotal role of podocytes in SSNS pathogenesis [37]. Podocytes, terminally differentiated epithelial cells characterized by actin-based projections, are crucial for maintaining glomerular function [38].

Among the various signaling molecules regulating podocyte actin cytoskeletal dynamics, RAC1 has emerged as a key player in SSNS pathogenesis [39]. Accumulated evidence have suggested that RAC1 hyperactivation exacerbates podocyte injury, while RAC1 inhibition protects against kidney injury and glomerular sclerosis [7,40,41]. Additionally, RAC1 activation has been associated with decreased expression of podocyte slit diaphragm proteins nephrin and podocin in a dosage-dependent manner, highlighting its multifaceted role in podocyte function and injury [42]. Beyond podocyte actin cytoskeleton reorganization, RAC1 activation influences diverse cellular processes, including inflammation, fibrosis, apoptosis, and mineralocorticoid receptor activation in kidney disease [43,44]. Interestingly, RAC1 activity has also been implicated in tumor cell migration, invasion, and metastasis, positioning it as a potential therapeutic target for cancer treatment [45]. Potent and selective RAC1 inhibitors, such as GYS32661 and MBQ-167, were under preclinical development for advanced solid tumors [46]. Inhibition of RAC1 activity has shown promise in improving cognitive deficits, synaptic plasticity, and Alzheimer’s disease-related memory loss [47]. Additionally, RAC1 has also been found to play a pathogenic role in atherogenesis, making it a potential target for anti-atherosclerotic therapy [48]. These findings suggested that RAC1 has pathogenic effects in a range of diseases, and its inhibition may represent a potential therapeutic strategy for kidney disease, cancer, neurodegenerative disorders, and cardiovascular conditions.

While previous studies primarily focused on RAC1’s role in podocyte injury using animal models or cultured podocytes, recent findings suggest a potential pathogenic role for circulating RAC1 in SSNS [8]. Elevated RAC1 levels and p38-MAPK activation in plasma EVs from SSNS patients during relapse suggested a selective transmission of protein signals involved in cytoskeletal rearrangement during disease exacerbation [8]. Moreover, treatment with RAC1-enriched relapse EVs induced RAC1-GTP (activated RAC1) expression in podocytes, enhancing motility and albumin permeability [8]. The RAC1 pathway was activated in cultured podocytes when exposed to sera from patients with idiopathic FSGS and MCD, but not when exposed to serum from patients with MCD in remission [7]. This evidence suggested a correlation between circulating RAC1 and podocyte dysfunction, implicating RAC1 as a potential circulating factor contributing to disease pathogenesis.

Building upon these observations, our study investigated the causal relationship between increased circulating RAC1 and the elevated risk of SSNS using MR methodology based on large-scale GWAS data. Intriguingly, single-cell sequencing analysis of PBMCs from SSNS patients revealed significantly elevated expression of RAC1, particularly among B cell and T cell subtypes, compared to healthy controls. These findings, along with previous observations of immune cell dysregulation in SSNS, suggest a complex interplay between immune cells, circulating RAC1, and podocyte injury in the pathogenesis of SSNS [49]. Notably, patients with MCD exhibited a dominance of cytotoxic T cells and natural killer cells in renal tissue, along with an increase in circulating CD8+ T cells, further supporting the involvement of immune dysregulation in SSNS pathogenesis [50]. Based on these findings, we propose a hypothesis wherein circulating RAC1, extensively expressed by immune cells such as T cells, may access podocytes through EVs and activate positive feedback, leading to podocyte injury and subsequent albuminuria (Figure 6). This model highlights the potential role of circulating RAC1 as a mediator of podocyte injury in SSNS pathogenesis, warranting further investigation into its mechanistic underpinnings and therapeutic implications. Consistent with the single-cell sequencing results, serum proteomic analysis of patients with MCD revealed an increase in RAC1 protein levels. Similar findings were confirmed using a PAN-induced MCD-like rat model. Additionally, we observed that glucocorticoid treatment was associated with a decrease in serum RAC1 levels. Previous studies have also demonstrated that glucocorticoids can inhibit RAC1 activity, thereby protecting podocyte barrier function [51].

Figure 6.

Potential mechanism of circulating RAC1 mediating podocyte injury via extracellular vesicles.

To our knowledge, this is the first MR study investigating the causal relationship between circulating RAC1 and SSNS. The advantage of this study lies in the utilization of large-scale GWAS data for MR analyses, enhancing the credibility of our results. The rigorous selection process for instrumental variables, combined with comprehensive sensitivity analyses and pleiotropy tests, further strengthens the robustness of our findings. Moreover, our study uniquely employed both eQTLs and pQTLs to explore the causal correlation between RAC1 and SSNS, with validation using single-cell sequencing, proteomics and animal experimental data.

However, several limitations should be acknowledged. Firstly, the data used in this study were sourced predominantly from European populations, which may limit the generalizability of the findings to other ethnicities. Additionally, the sample sizes for single-cell sequencing and proteomic analysis were relatively small, potentially impacting the precision and reliability of the circulating RAC1 expression measurements. As with the results of the MVMR analysis, the observed association between RAC1 and SSNS may be influenced by various confounding factors or potential sources of bias. Therefore, further clinical studies involving multiple ethnic groups and larger cohorts are warranted to validate the causal relationship between circulating RAC1 and SSNS. Furthermore, using tools like MetaXcan could provide additional insights into the tissue-specific expression of RAC1, which might help to corroborate the MR results. Given the existing evidence linking circulating exosome RAC1 expression with SSNS [8], future basic research should be designed to explore the causal relationship between exosome RAC1 and SSNS more deeply.

5. Conclusions

In conclusion, this study suggests a potential positive association between increased expression of RAC1 and the risk of steroid-sensitive nephrotic syndrome (SSNS), emphasizing the significance of RAC1 as a potential therapeutic target for SSNS. However, further investigation is warranted to fully elucidate the underlying mechanisms and validate the potential of targeting RAC1 in SSNS treatment.

Funding Statement

This work was supported by the National Natural Science Foundation of China (Grant numbers [82274391]).

Ethics approval

This study entailed a secondary analysis of publicly available data that had already obtained ethical approvals and informed consents from the participants, and no patient identification information was used in the study. Therefore, no additional ethical approvals or informed consents were required. The animal experiments were performed in accordance with the Ethics Committee of Shanghai Yueyang hospital of integrated traditional Chinese and western medicine (No. YYLAC-2023-218-1).

Author contributions

Yi Zhang: Methodology, Software, Formal analysis, and Writing – Original Draft; Tianwen Yao: Validation, Data Curation, and Writing – Original Draft; Yanqiu Xu: Formal analysis and Visualization; Yi Wang: Supervision; Shisheng Han: Conceptualization and Project administration, and Funding acquisition.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data used in this study are available via online data repository:

• GWAS data for RAC1 eQTLs, from the eQTLGen Consortium, and is available at the following URL: https://www.eqtlgen.org/cis-eqtls.html; Dataset name: ‘Full cis-eQTL summary statistics’.

• GWAS data for RAC1 pQTLs, from the deCODE Genetics, and is available at the following URL: https://www.decode.com/summarydata/; Dataset name: ‘Ferkingstad, E. et al.’ (Registration required).

• GWAS data for SSNS: from GWAS Catalog, and is ­available at the following URL: http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/GCST90258001-GCST90259000/GCST90258619/; Dataset name: ‘GCST90258619.tsv’.

• Single cell sequencing data, from Gene Expression Omnibus, and is available at the following URL: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE233275; Dataset name: ‘GSE233275_RAW.tar’.

• -Proteomics data: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9263421/; Dataset name: Supplementary Table 1.