Background
Nephrotic syndrome is associated with heavy proteinuria and peripheral edema [
1]. Idiopathic membranous nephropathy (IMN) is among the most common etiologies of primary nephrotic syndrome and its incidence has been rising in recent years [
2]. Most patients with IMN or idiopathic nephrotic syndrome (INS) present with marked proteinuria and peripheral edema [
1,
3]. Many observations support that a circulating factor may be responsible for some forms of nephrotic syndrome, including IMN [
3‐
5]. Although renal biopsy is the gold standard for differentiating glomerulonephritis from nephrotic syndrome, it is an invasive procedure and can sometimes be dangerous in patients taking antiplatelet agents or anticoagulation medications. Therefore, in cases where there is a relative contraindication to renal biopsy, a serology-based approach to diagnosing IMN has been suggested in a previous study [
6]. Studies have shown that the serum anti-phospholipase A2 receptor (PLA2R) antibody can be a potential biomarker for diagnosing, measuring disease activity, and predicting response to treatment in IMN [
7‐
9]. Lower anti-PLA2R titers are associated with high rates of spontaneous remission in IMN, thus favoring conservative therapy, while declining anti-PLA2R titer levels after treatment predict clinical response to rituximab treatment [
7]. However, the sensitivity of anti-PLA2R titer for the detection of IMN is low (between 52 and 78%) [
10‐
13], and 20–30% of patients test negative for serum anti-PLA2R antibodies [
14‐
16].
Extracellular vesicles (EVs) contain various molecules, including proteins, lipids, DNA, mRNA, and microRNA (miRNA), originating from the cell, and among these molecules, miRNAs have attracted the most attention since they can stably exist in various body fluids and play regulatory roles in gene expression [
17,
18]. EV-miRNAs appear to be more stable than free miRNAs, as EVs seem to protect and increase the stability of miRNAs [
19]. Recently, EV-miRNAs were reported to be more useful than free miRNAs for the detection of acute kidney injury (AKI) [
20]. In addition to AKI, some studies have shown the potential of EV-miRNAs to server as biomarkers in glomerulonephritis, such as for lupus nephritis [
21,
22]. Previous EV studies in renal diseases have usually focused on urinary EVs [
23,
24], whereas some recent reports have indicated a unique profile of circulating EV-miRNAs for nephrotic syndrome [
25‐
27]. These results suggest that circulating miRNAs could be used as biomarkers for nephrotic syndrome
. We also found that EV-miRNA profiles differ between the patients with diabetic nephropathy and those without diabetic nephropathy [
28]. However, there is limited data showing the potential role of miRNAs as diagnostic, prognostic, and therapeutic biomarkers for IMN.
In this study, to identify IMN-specific miRNAs at the time of kidney biopsy, we compared the EV-miRNA profiles of patients with IMN and other cohorts, including healthy volunteers (HVs) and patients with INS, using RNA sequencing. Furthermore, we attempted to identify circulating EV-miRNAs predictive of treatment response in patients with IMN by comparing EV-miRNA profiles in patients with and without clinical remission during treatment.
Materials and methods
Participants and data collection
We included 40 age- and sex-matched patients with IMN and INS and 20 HVs from a prospective glomerular disease cohort. In the present study, INS included focal segmental glomerular sclerosis and minimal change disease, which were defined by the association of the clinical features of nephrotic syndrome with renal biopsy findings of diffuse foot process effacement using electron microscopy [
29]. Patients with glomerulonephritis, who were diagnosed between January 2015 and Jun 2020, were enrolled from Soonchunhyang University Hospital and Presbyterian Medical Center. Patients with secondary causes of MN, such as lupus or malignancy, were excluded. This study was approved by the institutional review board of Soonchunhyang University (IRB No. 2016-01-002-007). Written informed consent was obtained from all the participants.
Remissions were defined according to the 2012 Kidney Disease: Improving Global Outcomes guidelines. Among patients with IMN, complete remission was defined as a reduction in proteinuria to 0.3 g/day. Partial remission was defined as a reduction in proteinuria to between 0.3 and 3.5 g/day (with at least 50% reduction versus baseline). Composite remission included either complete remission within 1 year after renal biopsy or partial remission with less than 2.5 g of proteinuria for 2 years following pathologic diagnosis. A refractory response was defined as the absence of composite remission during the follow-up period. Therefore, a total of 19 patients with IMN were divided into two groups: a well- responding (IMN-W) and refractory (IMN-R) group, based on the achievement of composite remission. Anti-PLA2R antibody was measured by ELISA method (EUROIMMUN AG, Lubeck, Germany) using serum sample collected on kidney biopsy.
During the follow-up period, treatment decisions for the enrolled patients were made by the treating nephrologist. The most common reasons for initiating immunosuppressive therapy were patient characteristics (proteinuria, renal function, etc.) that were not properly controlled, and nephrologist clinical judgement.
Serum EV RNA isolation and assessment
RNA sequencing was conducted as previously described [
30]. Briefly, circulating EVs were isolated from the serum (1000 μL) using the ExoQuick isolation agent (System Bioscience, Palo Alto, CA, USA), according to the manufacturer’s guidelines. Supernatants obtained after centrifugation (3000 ×
g for 15 min) of the serum samples were mixed with ExoQuick reagent and incubated for 30 min at 4 °C. After another centrifugation at 1500 ×
g for 30 min, the supernatant was aspirated, and the pellet was retained. After resuspension of the pellet in sterile phosphate-buffered saline (200 μL), RNA was extracted using the miRNeasy Mini Kit (Qiagen, Hilden, Germany). All processes involving the suspension of exosomes were conducted according to the manufacturer’s guidelines. After RNA extraction, purified RNA was eluted in RNase-free water (20 μL). The purified RNA was analyzed using an Agilent Bioanalyzer 2100 with an RNA Pico Chip and Small RNA Chip to examine the size distribution of EV RNAs (Agilent Technologies, Santa Clara, CA, USA).
Characterization of EVs by cluster of differentiation 63 (CD63) detection
CD63 levels in circulating EVs were measured using the Exo-enzyme-linked immunosorbent assay (ELISA)-ULTRA CD63 kit (System Biosciences, Palo Alto, CA, USA), according to the manufacturer’s protocol.
Western blot analysis
Each sample were electrophoresed on SDS-PAGE gels and were transferred to nitrocellulose membranes. The membranes were probed with specific antibodies as follows; anti-CD9 (Abcam, Cambridge, MA, USA) and anti-GM-130 (Abcam). The membranes were incubated with horseradish peroxidase-coupled secondary antibody (Sigma). Following washing with TBS-T, the bound antibody was detected by enhanced chemiluminescence (Amersham, Buckinghamshire, UK).
Transmission electron microscopy (TEM)
This protocol was performed as described by Thery et al. and Rikkert et al. [
31,
32]. A droplet of exosome solution was placed on Para film, and a Formva-carbon-coated nickel grid (200 meshes, TED PELLA, USA) was floated on the drop to absorb the sample at room temperature. After 10 min, the exosomes were fixed with 2.5% glutaraldehyde and stained with 1% uranyl acetate. The sample was washed with distilled water and dried in the dark. The grid was observed using an electron microscope operating at 75 kV (H-7000B; Hitachi, Tokyo, Japan).
Exosome physicochemical properties
A Nano-ZS Zetasizer (Malvern Inc., UK) was used to estimate the particle size. The samples were diluted ten times with distilled water and particle size was measured three times in a set of 50 repetitions using disposable cuvettes (DTS1070; Malvern Inc., Worcestershire, UK) and analyzed using the Zetasizer software (version 7.11).
cDNA library preparation and small RNA sequencing
The samples were processed to produce exosomal RNA (10 ng) as an input for each library. Small RNA libraries were constructed using a SMARTer smRNA-Seq Kit for Illumina (Takara Bio, Shiga, Japan), according to the manufacturer’s guidelines. Sequencing libraries were constructed by polyadenylation, cDNA synthesis, and polymerase chain reaction (PCR) amplification.
The libraries were gel-purified and validated by assessing their size, purity, and concentration using an Agilent Bioanalyzer. The libraries were quantified using quantitative PCR (qPCR), according to the qPCR Quantification Protocol Guide (KAPA Library Quantification Kits for Illumina® Sequencing Platforms). We assessed the quality of the libraries using TapeStation D1000 ScreenTape (Agilent Technologies, Waldbronn, Germany). Equimolar amounts of libraries were pooled and sequenced on an Illumina HiSeq 2500 instrument (Illumina, San Diego, CA, USA) to generate 51 base reads. Image decomposition and quality value calculations were performed using modules in the Illumina pipeline. All procedures for next-generation sequencing (NGS) analysis were performed at Macrogen (Seoul, Korea).
Analysis of RNA sequencing data and proportions of miRNAs
Following sequence alignment, known and novel miRNAs were identified using the miRDeep2 algorithm. Prior to sequence alignment, we retrieved the Homo sapiens reference genome release hg19 from the UCSC Genome Browser, which was indexed using Bowtie (1.1.2), a program for aligning experimental and reference sequences. The reads were then aligned to the mature and precursor H. sapiens miRNAs obtained using miRBase 21. Uniquely clustered reads were sequentially aligned to the reference genome using miRBase 21 and the non-coding RNA database Rfam 9.1 to identify known miRNAs and other types of RNAs, respectively.
Analysis of miRNA expression levels
The raw data (reads for each miRNA) were normalized to the relative log expression using DESeq2. For preprocessing, miRNAs absent from more than 50% of all samples were excluded, leaving only mature miRNAs for analysis. We added 1 to the normalized read count of the filtered miRNAs to facilitate log
2 transformation and draw a correlation plot. For each miRNA, the base mean and log-fold changes were calculated between the groups. We conducted a statistical hypothesis test to compare the groups using the negative binomial Wald test in DESeq2. miRNAs differentially expressed between the two groups were defined as having a |fold change|≥ 2 and a false discovery rate (FDR)-adjusted p-value of < 0.05. We also performed hierarchical clustering analysis using complete linkage and Euclidean distance as measures of similarity to display the expression patterns of the differentially expressed miRNAs that satisfied the criteria of a |fold change|≥ 2 and an FDR-adjusted p-value of < 0.05. All data analyses and visualization of the differentially expressed genes were performed using R 3.3.1 (
www.r-project.org).
Identification of miRNA target genes and their molecular pathways
We uploaded miRNAs that were differentially expressed in the HVs and patients with IMN-W and IMN-R into commonly used analysis programs, such as DIANA-miRPath and miRSystem, for further analyses. The DIANA-miRPath v.3.0 database used DIANA-microT-CDS and TargetScan 6.2 to analyze miRNA-gene interactions. The database schema incorporated the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, Gene Ontology (GO), and GO slim annotations. Gene and miRNA annotations were derived from the Ensembl and miRBase databases, respectively.
Statistical analysis
Continuous variables with normal distributions are expressed as the mean ± standard deviation; variables without a normal distribution are expressed as medians with interquartile ranges. The t-test was used to analyze the statistical significance of the differences between continuous variables, and the chi-square test was used for categorical variables to compare the baseline characteristics between HVs and patients with IMN. The IMN group was further divided into two groups, and continuous variables were compared among the three subgroups (HVs vs. IMN-W vs. IMN-R) using the Kruskal–Wallis multiple comparison test. Receiver operating characteristics (ROC) analysis was used to calculate the area under the curve (AUC) for each miRNA for the diagnosis and prediction of treatment of response in IMN patients. Statistical significance was set at P < 0.05. Statistical analysis was performed using SPSS (version 22.0; IBM Corp., Armonk, NY, USA).
Discussion
In this study, we conducted RNA sequencing to examine the circulating EV-miRNA profiles of patients with IMN and compared them with those of patients with INS and HVs. We identified IMN-specific EV-miRNAs compared to those in HVs or INS subjects. Furthermore, we found that EVs-miRNAs were associated with treatment response in patients with IMN, which will be helpful for clinicians to predict the prognosis of these patients.
Differentially-expressed circulating miRNAs have been found in patients with various glomerular diseases, such as IgA nephropathy and lupus nephritis [
34,
35]. Significant differences in the expression profiles of urinary and circulating exosomal miRNAs have also been observed between HVs and patients with IMN [
36,
37]. A previous study suggested there is miRNA dysregulation in IMN, in which the differential expression of six miRNAs (upregulation of miR-152 and -15 and downregulation of miR-82, -98, -89, and -84) was observed in patients with IMN when compared to HVs [
38]. These miRNAs did not overlap with those identified in the present study. This difference might be due to the source of miRNAs, since the authors of the previous study analyzed miRNAs from peripheral blood lymphocytes. In the present study, we identified three IMN-specific EVs-miRNAs, which might be helpful in discriminating patients with IMN from those with INS or without nephrotic syndrome. Therefore, this study demonstrates the potential of EV-miRNAs as biomarkers for IMN diagnosis.
In addition to these miRNAs, we identified EVs-miRNAs that are helpful in predicting treatment response in patients with IMN. To our knowledge, this is the first study of EV-miRNAs for predicting treatment response in patients with IMN. The prediction of clinical remission in patients with IMN is important for nephrologists because complete or partial remission is associated with good kidney survival [
39,
40]. When a refractory case is anticipated with conventional therapy, the clinician may be able to come up with alternative treatment options to avoid the adverse effects of immunosuppressive drugs. In our study, we identified 23 miRNAs that were differentially expressed in patients with IMN-R which were distinct from those in patients with IMN-W. Interestingly, such pathways seem to be associated with cancers. Although the causality link between malignancy and MN remain unsolved, MN is the most common type of malignancy-associated glomerular lesions. Recently, von Haxthausen et al. reported no differences in antigen-specific IgG subclasses between IMN and malignancy-associated MN [
41], suggesting a similar pathogenesis between idiopathic and malignancy-associated MN. Therefore, we believe that circulating EVs-miRNAs could be used as non-invasive markers for predicting clinical remission in patients with IMN during treatment.
Of the possible pathways associated with treatment response in patients with IMN in the present study, the transforming growth factor-β (TGF-β) signaling pathway seems to be associated with predicting the treatment response based on our data. Renal fibrosis is the usual final outcome of excessive accumulation of extracellular matrix, and TGF-β plays an important role in tissue fibrosis by upregulating matrix protein synthesis and inhibiting matrix degradation [
42]. Previous studies have shown that urinary TGF-β1 levels are elevated during active disease and correlate with histological characteristics and proteinuria [
43‐
45]. In patients with MN, higher initial urinary TGF-β1 levels are associated with persistent nephrotic syndrome and kidney function decline at 12 months [
44,
46]. Of the 15 miRNAs associated with the TGF-β pathway in our study, the levels of let-7f and miR-23a were associated with baseline proteinuria on kidney biopsy, which is known to be a predictor of IMN [
47,
48]. In addition, EV-miR-23a levels directly correlated with anti-PLA2R antibody concentrations. These findings are consistent with previous experimental results. Overexpression of miR-23a has been reported to repress renal cell viability and proliferation by suppressing the heat shock protein 70 [
49]. In addition, upregulation of let-7f in diabetic mice is helpful in relieving podocyte injury in diabetic nephropathy [
50]. Interestingly, proteoglycans, which are a family of highly glycosylated proteins mainly involved in tissue organization, seems to be associated with differentially expressed miRNAs in patients with IMN-R in this study. Such finding might be due to the inhibitory influence on TGF-β signaling pathway [
51].
Besides the TGF-β pathway, based on the identified miRNAs in our study, the Hippo signaling, Forkhead homobox type O (FoxO), and Rap1 pathways are analyzed to be associated with renal fibrosis in patients with IMN-R. The Hippo signaling pathway not only participates in the crosstalk with other signaling pathways, including the TGF-β and Wnt/β-catenin signaling pathways, but also plays a role in the pathogenesis of renal tubulointerstitial fibrosis [
52,
53]. The Rap1 pathway influences renal fibrosis via regulation of the cyclic adenosine 3′,5'-monophosphate signaling pathway [
54]. In this study, some miRNAs, such as let-7f and miR-23a, were also found to be associated with the FoxO and Hippo pathways, similar to the TGF-β pathway. Previous studies have also reported such a relationship in other diseases [
55,
56]. Intriguingly, PCA of serum-derived EVs from HVs and the IMN, and INS groups revealed a relatively poor demarcation. However, PCA revealed some overlap between the serum-derived EVs from HVs and those from patients with IMN-W, with a fully distinct miRNA profile for serum-derived EVs from patients with IMN-R. Although these results might imply that fibrotic changes in IMN-R are important factors affecting treatment response, further studies are needed to investigate the pathological mechanism for predicting the clinical responses of patients with IMN. Our study had some limitations. First, the number of enrolled participants was relatively small; therefore, larger prospective randomized controlled trials are needed to confirm our results. Next, EV RNA was isolated using commercial kits; however, isolation using other methods may yield different results. To date, there is no gold standard method for EV isolation. Finally, we could not validate our findings in different cohorts, and this can be taken up in future studies.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.