Understanding Nephrotic Syndrome Using Kidney Transcriptome Profiling and Computational Studies

Nephrotic syndrome (NS) is a major clinical manifestation of glomerular diseases in children and adult patients that mainly affects the podocytes. Despite decades of research in NS, the underlying causes are still not fully understood, and effective therapeutic agents that promote complete remission are still limited.1 Thanks to computational studies and machine learning, kidney transcriptome profiling is evolving at an impressive rate and has great potential to provide new insights into the pathogenesis and mechanisms of kidney diseases from a molecular, cellular, and clinical point of view.2 Genome scale expression datasets, such as the Nephrotic Syndrome Study Network (NEPTUNE) and the European Renal cDNA Bank, encompass clinical, genetic, transcriptional, proteomic, and metabolomic information.3,4 Therefore, in the last decade, they have become valuable resources for nephrologists and researchers seeking to identify genes and mechanisms whose expression is related to kidney disorders. In this issue, Li et al. performed several computational analyses of a NEPTUNE Study database (GSE200828) to identify shared molecular pathways across patients with FSGS, membranous nephropathy (MN), and minimal change disease (MCD).5 FSGS, MN, and MCD are distinct forms of glomerular diseases that all manifest NS. The identification of common molecular pathways in NS will potentially establish known and novel mechanisms of podocyte injury that may translate into the development of novel therapeutic targets to arrest proteinuria and preserve kidney function. To attain the goal, the authors first identified glomerular differentially expressed genes (DEGs) in FSGS, MN, and MCD and compared with the healthy kidney by using GEO2R, an interactive web tool that allows users to compare and identify DEGs among two or more groups of samples. Next, R packages and machine learning were used to identify enriched pathways in the Kyoto Encyclopedia of Genes and Genomes database. At last, the authors implemented the Cytoscape, Database for Annotation, Visualization and Integrated Discovery (DAVID) Bioinformatic, and STRING tools to reveal the most significant shared pathways contributing to FSGS, MN, and MCD. Their results suggest that these three types of NS share more than a thousand DEGs. The enrichment analysis identified focal adhesion as the most significative shared dysregulated pathway and established zyxin (ZYX) as the most important contributor to these glomerulopathies, followed by actinin 4 (ACTN4), CRK, and integrins (ITGA). Finally, to determine which type of glomerular cells express these genes, the authors used a single-cell RNA-seq dataset, obtained from glomerular-enriched cells in normal kidney tissue from a patient who underwent nephrectomy. Single-cell RNA-seq was processed with the Leiden algorithm from the Phyton Scanpy package. The findings revealed that ZYX, ACTN4, CRK, and ITGA3 are expressed in most known kidney cell types, especially podocytes. These results are of great interest because they confirmed, through bioinformatics analysis, the knowledge that we have acquired through clinical and experimental data: Proteinuric kidney diseases are the result of damage to the glomerular filtration barrier, and podocyte injury plays a central role in the initiation and progression of glomerular diseases. They also strengthen the evidence that dysregulation of focal adhesion is a key pathway in these glomerulopathies. In fact, in the past 20 years, numerous studies, including our investigation, have associated proteinuria with dysregulated actin dynamics and altered cell signaling at focal adhesion sites on podocytes.6,7 These disorders are manifested by foot process effacement and podocyte depletion. Hence, therapeutic agents that target focal adhesion and actin cytoskeleton on podocytes have been postulated in numerous studies as potential treatments for proteinuric kidney diseases. Another interesting finding reported by Li et al. is the identification of zyxin as the central hub in the FSGS, MN, and MCD. In 2010, different research groups reported that zyxin mediates important mechanisms for the maintenance and repair of the actin cytoskeleton in different epithelial cell lines.8,9 In 2013, Tan and coworkers studied the function of zyxin in an

Nephrotic syndrome (NS) is a major clinical manifestation of glomerular diseases in children and adult patients that mainly affects the podocytes. Despite decades of research in NS, the underlying causes are still not fully understood, and effective therapeutic agents that promote complete remission are still limited. 1 Thanks to computational studies and machine learning, kidney transcriptome profiling is evolving at an impressive rate and has great potential to provide new insights into the pathogenesis and mechanisms of kidney diseases from a molecular, cellular, and clinical point of view. 2 Genome scale expression datasets, such as the Nephrotic Syndrome Study Network (NEPTUNE) and the European Renal cDNA Bank, encompass clinical, genetic, transcriptional, proteomic, and metabolomic information. 3,4 Therefore, in the last decade, they have become valuable resources for nephrologists and researchers seeking to identify genes and mechanisms whose expression is related to kidney disorders.
In this issue, Li et al. performed several computational analyses of a NEPTUNE Study database (GSE200828) to identify shared molecular pathways across patients with FSGS, membranous nephropathy (MN), and minimal change disease (MCD). 5 FSGS, MN, and MCD are distinct forms of glomerular diseases that all manifest NS. The identification of common molecular pathways in NS will potentially establish known and novel mechanisms of podocyte injury that may translate into the development of novel therapeutic targets to arrest proteinuria and preserve kidney function.
To attain the goal, the authors first identified glomerular differentially expressed genes (DEGs) in FSGS, MN, and MCD and compared with the healthy kidney by using GEO2R, an interactive web tool that allows users to compare and identify DEGs among two or more groups of samples. Next, R packages and machine learning were used to identify enriched pathways in the Kyoto Encyclopedia of Genes and Genomes database. At last, the authors implemented the Cytoscape, Database for Annotation, Visualization and Integrated Discovery (DAVID) Bioinformatic, and STRING tools to reveal the most significant shared pathways contributing to FSGS, MN, and MCD.
Their results suggest that these three types of NS share more than a thousand DEGs. The enrichment analysis identified focal adhesion as the most significative shared dysregulated pathway and established zyxin (ZYX) as the most important contributor to these glomerulopathies, followed by actinin 4 (ACTN4), CRK, and integrins (ITGA). Finally, to determine which type of glomerular cells express these genes, the authors used a single-cell RNA-seq dataset, obtained from glomerular-enriched cells in normal kidney tissue from a patient who underwent nephrectomy. Single-cell RNA-seq was processed with the Leiden algorithm from the Phyton Scanpy package. The findings revealed that ZYX, ACTN4, CRK, and ITGA3 are expressed in most known kidney cell types, especially podocytes.
These results are of great interest because they confirmed, through bioinformatics analysis, the knowledge that we have acquired through clinical and experimental data: Proteinuric kidney diseases are the result of damage to the glomerular filtration barrier, and podocyte injury plays a central role in the initiation and progression of glomerular diseases. They also strengthen the evidence that dysregulation of focal adhesion is a key pathway in these glomerulopathies. In fact, in the past 20 years, numerous studies, including our investigation, have associated proteinuria with dysregulated actin dynamics and altered cell signaling at focal adhesion sites on podocytes. 6,7 These disorders are manifested by foot process effacement and podocyte depletion. Hence, therapeutic agents that target focal adhesion and actin cytoskeleton on podocytes have been postulated in numerous studies as potential treatments for proteinuric kidney diseases.
Another interesting finding reported by Li et al. is the identification of zyxin as the central hub in the FSGS, MN, and MCD. In 2010, different research groups reported that zyxin mediates important mechanisms for the maintenance and repair of the actin cytoskeleton in different epithelial cell lines. 8,9 In 2013, Tan and coworkers studied the function of zyxin in an immortalized human podocyte cell line. Their findings suggested that zyxin plays a key role in podocyte structure and function using Nef interaction and that zyxin expression is increased to repair actin stress fibers and maintain podocyte phenotype despite the loss of cytoskeletal integrity. 10 Therefore, it is not surprising that computational studies identified ZYX among the most dysregulated genes in NS. However, altered zyxin expression may only be a consequence of the disease state, rather than being responsible for the development of the NS.
A limitation of the study presented by Li et al. is the lack of experimental data that validates their bioinformatic findings. Certainly, future directions for this research will be (1) to confirm whether zyxin expression in glomerular podocytes in patients with NS is increased, (2) to correlate zyxin expression with disease status and progression, and (3) to create strategies that modulate zyxin function on kidney injury. Studies investigating zyxin as a potential therapeutic target will be highly valuable.
Although it is very useful to identify shared pathways for proteinuric kidney diseases, several investigations have also attempted to determine, at the cellular level, the distinctive dysregulated pathway in each glomerular ailment. Sealfon et al. applied a machine learning framework to determine the molecular characterization of primary MN that distinguished it from FSGS and MCD. The authors found that MN-specific genes were enriched in the nuclear factor kappa B signaling pathway and were predominantly expressed in podocytes. 11 In a similar study, Zambrano et al. performed single-cell sequencing and observed that endothelial cells played an important role in the recruitment of immune cells in the pathogenesis of IgA nephropathy. 12 Furthermore, urinary single-cell transcriptomics have revealed that urine samples from patients with FSGS contains increased numbers of immune cells and a higher expression of epithelial-mesenchymal transition-associated genes in numerous kidney epithelial cells when compared with the MCD counterparts. 13 The presence of all kidney cell types and immune cells in urine has also been demonstrated by single-cell transcriptomic analysis in patients with diabetic kidney disease. 14 As illustrated in the previous paragraphs, the immense potential of kidney transcriptomics and bioinformatics has attracted researchers and nephrologists seeking to identify shared and distinct dysregulated molecular pathways in proteinuric kidney diseases. During the last decade, kidney transcriptomics using human and mouse models has become progressively more feasible and accessible to the scientific community. To take full advantage of these rich datasets, efficient gene sequencing and computational methods are required. Hence, we recommend following guidelines developed by experts, such as the one published by Luecken and Theis, 15 where they provide workflow recommendations throughout the entire gene sequencing and raw data processing (quality control, quantification, normalization, clustering, identification and validation of DEGs, etc.). Making use of these guidelines will have a substantial positive effect on the outcome.
In sum, kidney transcriptomics has been increasingly used to understand genome-wide transcriptomic variations in different conditions such as disease states. Tissue (bulk) RNA-seq, widely used in human patients and animal models, generates an output for the average expression of genes, reflected by cell type-specific gene expression differences and changes in cell composition. The emerging technology, single-cell RNA-seq, has enabled cell type-specific transcriptome profiling and identified the specific cell types involved in disease development and progression. However, its application in clinical studies still has some limitations to overcome, such as high cost and issues with cell dissociation in solid tissues. Combining tissue RNA-seq with single-cell profiling methods, as Li et al. performed in this issue, is a promising and more approachable strategy for understanding cell type-specific and cell heterogeneity changes in NS. The knowledge gained will greatly facilitate future research and clinical work.

Disclosures
The authors have nothing to disclose.

Funding
None.