Single-Cell and Spatial Transcriptomics in Renal Injury and Fibrosis Research

Abstract

Background

Kidney diseases, which are broadly classified into acute kidney injury (AKI) and chronic kidney disease (CKD), represent a significant and ongoing health burden in China and across the globe. AKI is a clinical syndrome marked by a rapid decline in renal function within 48 h due to diverse causes. Despite its high prevalence among hospitalized patients as a common complication, current therapeutic outcomes remain unsatisfactory. Therefore, a critical step toward resolving this issue is the precise identification of the specific cell types that drive renal regeneration during AKI repair. Renal fibrosis, pathologically characterized by excessive extracellular matrix (ECM) deposition, is the common final pathway of CKD and significantly impairs patient’s quality of life and prognosis. The focal nature of fibrotic lesions has prompted systematic investigations into the fibrotic microenvironment. Myofibroblasts are the central effector cells driving pathological ECM deposition. Nevertheless, their cellular origins remain elusive. A thorough understanding of myofibroblast origins, along with the composition and regulatory factors of the fibrotic microenvironment, is therefore crucial for developing effective treatments for renal fibrosis.

Summary

The kidney is a complex organ with intricate anatomical structures and diverse cellular composition. Traditional investigative methods which rely on conventional pathology and low-resolution molecular biology have been unable to capture cellular heterogeneity at single-cell resolution. This limitation has obscured functional distinctions among cell subpopulations and their critical spatial context, thereby leading to an inadequate understanding of intercellular communication. Fortunately, the advent of single-cell and spatial transcriptomics has revolutionized kidney research by enabling comprehensive profiling of functional signatures and intercellular crosstalk within the renal microenvironment.

Key Messages

This review summarizes the current applications of single-cell and spatial transcriptomics in renal regeneration and fibrosis. Furthermore, it introduces emerging technologies, such as proximity-dependent labeling, while rarely applied in kidney research to date, hold significant potential. Our aim is to provide researchers with insightful strategies for their future application in this field.

Introduction

Statement from the American Society of Nephrology, European Renal Association, and International Society of Nephrology indicated that as of 2021, more than 850 million people worldwide suffer from various forms of kidney disease, which is 20 times more than the prevalence of cancer worldwide (42 million). All chronic kidney disease (CKD) progresses irreversibly to end-stage renal disease (ESRD), at which point patients can only sustain normal physiological function through interventions such as kidney transplantation, hemodialysis, peritoneal dialysis, or supportive care [1]. Kidney diseases have placed a huge burden on society; the World Health Organization has recommended prioritizing kidney diseases as a major non-communicable disease of global concern.

Acute kidney injury (AKI) is a common and serious condition in hospitalized patients that can significantly increase the risk of death, but its diagnostic method relying on serum creatinine often leads to difficulties in early detection [2, 3]. AKI is not only an emergency, but also initiates a maladaptive repair process that is now recognized as a key driver in the pathogenesis of CKD [4]. CKD affects more than 10% of the global population, and renal fibrosis is a key common pathological pathway in its progression [5, 6]. The core driver of fibrosis is TGF-β1, which induces myofibroblast differentiation and epithelial-mesenchymal transition (EMT) through SMAD signaling, leading to excessive deposition of extracellular matrix (ECM) [7]. The Wnt/β-catenin pathway acts synergistically with TGF-β to aggravate fibrosis [8], while inflammatory mediators such as IL-11 and TWEAK further amplify the fibrotic response through JAK/STAT and Fn14/NF-κB pathways [8, 9]. Together, these signals act on a variety of cells in the kidney, ultimately leading to pathological ECM accumulation [7].

Myofibroblasts represent a repair-associated phenotype that emerges in response to tissue injury. Initially identified via transmission electron microscopy in rat granulation tissue, where they displayed fiber bundles resembling smooth muscle [10, 11]. Characterized by the expression of α-smooth muscle actin (aSMA), myofibroblasts could also be distinguished by vimentin, collagen type 1, CD73, platelet-derived growth factor receptor beta, and fibroblast specific protein-1/S100A4 [12, 13]. In renal fibrosis, myofibroblast activation is a central event, driving ECM deposition [12]. Multiple cellular origins have been proposed for myofibroblasts in kidney fibrosis. These include EMT, in which injury-induced nuclear translocation of SNAIL1 transmits mesenchymal signals to promote myofibroblast differentiation and sustain a pro-fibrotic microenvironment [14, 15]. Endothelial-to-mesenchymal transition also contributes to the myofibroblast pool while exacerbating capillary loss and tissue hypoxia. Upon kidney injury, pericytes can detach from capillaries and migrate into the interstitium, undergoing a transition from pericytes to myofibroblasts. This process is suggested to be a significant, albeit not the predominant, source of myofibroblasts [16]. Additionally, macrophages recruited after injury may undergo macrophage-to-myofibroblast transition, particularly M2-polarized macrophages, via the TGF-β/Smad3 pathway [17, 18]. The TGF-β pathway serves as a central regulator across these transitions. These maladaptive cellular interactions and signaling pathways represent potential therapeutic targets to inhibit fibrosis.

Conventional methods in kidney research face inherent limitations in accurately capturing the dynamic evolution of the renal tissue microenvironment and the heterogeneity at the boundary between disease and health. Current studies have primarily focused on the functions and molecular mechanisms of individual cell types. However, the cellular interaction networks involved in kidney injury and chronic progression remain inadequately characterized [7, 19, 20]. This gap hinders a comprehensive understanding of the role and underlying mechanisms of the renal tissue microenvironment in the progression of CKD. Nevertheless, advances in single-cell RNA (scRNA-seq) sequencing have opened new avenues for in-depth exploration of the kidney’s cellular composition, while spatial transcriptomics provides robust support for the precise localization of cell types, the assessment of local gene expression changes, and the elucidation of cell-cell interactions. In this review, we summarize the technologies and applications of single-cell sequencing, spatial transcriptomics, and other integrative multi-omics approaches in kidney disease research, aiming to facilitate mechanistic exploration at a microscopic level for this field.

Single-Cell RNA Sequencing in Nephrology

The kidney is a structurally complex organ comprising over 20 specialized cell types, which is not accurately portrayed using traditional sequencing technology. Single-cell genomics, built on second-generation sequencing, enables transcriptomic profiling of hundreds of thousands of individual cells, substantially enhancing resolution, depth, and accuracy. This approach provides a powerful methodology for characterizing renal cell types and states across nephrogenesis, tissue homeostasis, and disease progression [21, 22]. Early applications of scRNA-seq in developmental biology demonstrated its superior sensitivity; for example, an mRNA-Seq assay on a single mouse blastomere detected 270 more genes than conventional methods using hundreds of cells, along with 1,753 previously unannotated splice junctions [23]. As a robust alternative to scRNA-seq, single-nucleus RNA sequencing (snRNA-seq) integrates 10X Genomics with sequencing platforms such as Oxford Nanopore Technologies (ONT) and Illumina. By utilizing strong dissociation methods that lyse cell membranes to release nuclei, snRNA-seq imposes less stringent requirements on initial sample quality compared to conventional single-cell approaches. This makes it particularly suitable for profiling cell types that are difficult to dissociate, such as neurons and podocytes, while maintaining transcriptomic coverage and data integrity [24] (Fig. 1). Microfluidic technology involves the construction of micron-scale channel and chamber structures through the generation of two immiscible fluids. This technique facilitates the isolation, manipulation, and analysis of individual target cells [25]. The adoption of this technology has enabled 10x Genomics to achieve higher throughput and more precise cell isolation and labeling, thereby significantly enhancing detection efficiency [26], as shown in Table 1 [24, 27–32].

Fig. 1.

Technical workflow and application scenarios of single-cell sequencing. Firstly, the renal tissue is digested to generate a suspension of single cells or single nuclei. Single-cell sorting was performed using a microfluidic chip system (e.g., 10x Genomics Chromium System) to capture individual cells. Next, library construction is performed to prepare the genetic material for sequencing. The prepared libraries then undergo sequencing using specialized equipment. Subsequent data analysis involves reconstruct the kidney cell atlas, dissecting the renal fibrosis microenvironment and discovery of cell subpopulations and biomarkers.

Table 1.

Single-cell RNA sequencing in nephrology

Methods	Technology	Sample type	Measurement flux	Ref.
scRNA-seq	Oligo-dT probes capture the poly(A) tail of mRNA	Fresh tissue	Cytoplasmic + nuclear	[27–29]
snRNA-seq	Permeabilize the nuclear membrane to capture poly(A) mRNA within the nucleus	Fresh or frozen tissue	Nuclear	[24, 30]
scATAC-seq	Tn5 transposase inserts adapters into open chromatin regions	Fresh or frozen tissue	Open regions of chromatin	[31, 32]

The highly condensed structure of chromatin must undergo localized unwinding during replication and transcription to expose DNA sequences for recognition by transcription factors and regulatory elements. This structural openness, which enables essential molecular interactions, is referred to as chromatin accessibility. Single-cell assay for transposase-accessible chromatin with sequencing (scATAC-seq) enables genome-wide profiling of chromatin accessibility at single-cell resolution. This technology helps uncover epigenetic mechanisms governing gene expression, refine cell identity annotation, and locate differentially accessible regulatory regions [33]. For instance, in a study of clear cell renal cell carcinoma, researchers used scATAC-seq to construct a chromatin accessibility atlas, which revealed distinct regulatory features among tumor cell subpopulations and identified two long non-coding RNAs that promote cancer invasion and migration [34].

A particularly transformative bioinformatic advancement is trajectory inference (pseudotime) analysis, which uses computational algorithms to order cells along a continuous trajectory based on transcriptional similarity, effectively reconstructing dynamic biological processes [35, 36]. For instance, in a study of metabolic cell fate during human kidney development, researchers combined metabolic gene set enrichment with pseudotime reconstruction to delineate three distinct developmental trajectories and identify associated lipid molecules, a finding that offers significant insight into the mechanisms of renal development [37]. Despite their transformative impact on cellular resolution and pseudotime analysis, single-cell sequencing technologies face several technical challenges when applied to kidney research. These include: (1) the inability of current dissociation protocols to capture all renal cell types, as they may damage fragile populations while failing to isolate those embedded in the collagenous matrix [21]; (2) the induction of intracellular stress responses by the enzymatic and mechanical isolation procedures, which can alter transcriptional profiles and compromise sequencing accuracy; and (3) the lack of optimal tissue preservation methods, which severely limits the retrospective analysis of frozen archival specimens [24, 38, 39].

Recent advances in snRNA-seq have significantly advanced our understanding of AKI pathogenesis by delineating key cellular players and their interactions. Relevant snRNA-seq study identified a pro-inflammatory S100a9^hiLy6c^hi macrophage subset that initiates the injury cascade [40], while also characterizing functionally distinct lymphatic endothelial cells (LECs) subpopulations with enhanced lymphangiogenic potential and immune regulatory functions post-AKI, which also uncovered the significant involvement of LEC-mediated intercellular interactions in key immune regulatory pathways [41]. Sc-sequencing analysis identified a cascade of pathways activated during the injury transition period, based on early and late-stage post-ischemia/reperfusion injury (IRI) samples. Specifically, VCAM1+/CCL2+ proximal tubule cells (PTCs) significantly activated in late-stage injury were traceable to early repair-phase cells, suggesting the existence of delayed-onset secondary injury foci in PTCs [42]. Animal models are essential for AKI research, but cytopathological heterogeneity across etiologies remains unclear. SnRNA sequencing of five AKI models showed crystal-induced injury (folic acid/sodium oxalate) mimics obstructive damage, with Havcr1+ tubule cells but not novel Krt20 PTCs driving fibrosis and maladaptive subtypes controlling inflammation in different AKI model [43].

Myofibroblasts represent a protective cell state that emerges post-injury, defined by ECM secretion and actomyosin-based fiber contraction; however, their persistent activation can lead to pathological fibrosis [44]. Unraveling the pathogenic mechanisms of myofibroblast-mediated fibrosis requires precise identification of their cellular origins. Earlier studies using differentiation marker expression and lineage tracing have suggested that bone marrow-derived fibroblasts, tubular epithelial cells, endothelial cells, pericytes, and interstitial fibroblasts can all differentiate into myofibroblasts, though their relative contributions remain unclear [5]. Fortunately, single-cell transcriptomic sequencing has provided powerful tools to address this challenge. A single-cell RNA sequencing study of human fibrotic kidneys identified resident mesenchymal cells, including Notch3+/RGS5+/PDGFRα-pericytes, MEG3+/PDGFRα+ fibroblasts, and COLEC11+/CXCL12+ fibroblasts, which are the major cellular sources of renal myofibroblasts [16]. Additional evidence supporting endothelial cells as a source of myofibroblasts comes from single-cell sequencing of 3 patients with renal fibrosis and three healthy controls, combined with spatial transcriptomics of an interstitial fibrotic kidney sample. The integrated analysis revealed that endothelial cells can transition into myofibroblasts in a TGF-β dependent process, which can be promoted by midkine [45]. A study based on single-cell sequencing technology, targeted the renal fibrosis model induced by unilateral ureteral obstruction, confirmed the transition of macrophages to myofibroblasts during the process of renal fibrosis [46].

ScRNA-seq analysis of samples from patients with chronic kidney transplant rejection (CKTR) and healthy controls has overcome the limitations of conventional bulk transcriptomic approaches. By delineating gene expression profiles at single-cell resolution, scRNA-seq revealed an increased proportion of CD8⁺ T cells, cytotoxic T lymphocytes, B lymphocytes, and myofibroblasts in CKTR samples, highlighting microenvironmental alterations associated with allograft rejection and fibrotic progression [47]. Recent scRNA-seq studies have further identified two fundamental patterns of cellular reprogramming in CKD. In early-stage CKD, PTCs exhibit a protective metabolic adaptation characterized by upregulated fatty acid β-oxidation. As disease progresses to chronic stages, this adaptive response transitions to suppressed lipid metabolism, representing a maladaptive shift [48].

Diabetic kidney disease (DKD) is a chronic kidney disorder pathologically dominated by fibrosis. scRNA-seq of diabetic rat kidneys revealed that mesangial cells transdifferentiate into myofibroblasts under hyperglycemic conditions, thereby driving the progression of renal interstitial fibrosis. Further experimental validation confirmed that within the DKD microenvironment, a fibrosis-associated shift in macrophage polarization toward the M2 phenotype precedes the mesangial-to-myofibroblast transition. This process is activated through the TGF-β1/Smad2/3/YAP signaling axis and collagen deposition [49].These findings establish that while early CKD involves metabolic adaptation of tubular cells, DKD progression is dominated by macrophage-mediated fibrotic transformation of mesangial cells, highlighting disease-specific cellular reprogramming pathways in CKD.

Spatial Transcriptomics Technologies in Nephrology

Although scRNA-seq continues to expand the catalog of cell states across diverse organisms and tissues, it inherently lacks spatial context, failing to capture the microenvironments in which these cellular identities reside [50]. Resolving the spatial distribution of RNA within tissues has long been a major methodological challenge. While in situ hybridization has been a conventional approach for RNA localization, its technical complexity and low throughput limit its utility for the simultaneous analysis of multiple targets [51]. Spatial transcriptomics has emerged to address this gap by enabling direct, in situ imaging of RNA transcripts or by computationally integrating spatial coordinates with sequencing data from tissue sections. This technology provides powerful support for discovering novel cell types and states, deciphering cell-cell interactions, characterizing disease niches, and elucidating ligand–receptor communication networks [52]. Spatial omics methodologies can be broadly classified into three categories based on their underlying principles: in situ capture-based sequencing, imaging-based approaches, and region-of-interest-based detection. In situ capture-based methods enable near-complete profiling of the protein-coding transcriptome, with spatial resolution ranging from subcellular (a few hundred nanometers) to supracellular (hundreds of micrometers), depending on the platform. Imaging-based techniques offer the highest resolution, allowing precise localization of individual RNA molecules within tissue sections. In 2016, the 10x Genomics Visium platform was developed and applied to spatial transcriptomics. Subsequent advances have dramatically increased precision: μST achieved resolution below 1 μm, while microfluidic-assisted DBiT-seq reached 10 μm [53]. More recently, Visium HD has enabled high-quality spatial transcriptomics at 2 μm resolution. Notably, NGS-based platforms such as Seq-Scope and Stereo-seq have pushed resolution to 0.5–0.7 μm, permitting distinct gene expression profiling at the subcellular level [54, 55] (Fig. 2). With the development of spatial transcriptome technology, the trade-off between spatial resolution and detection sensitivity has become prominent. Improving spatial resolution typically requires denser probe arrays or thinner sections, but this limits the number of mRNA molecules captured per detection unit, leading to increased data sparsity and missed detection of low-abundance genes. At the same time, insufficient sequencing depth will reduce the sensitivity of gene detection, while excessive sequencing depth will not only greatly increase the cost. Thus, technology selection requires a balance between resolution, gene flux, sample compatibility, and cost. Previous studies have successfully dissected the renal fibrosis microenvironment using Visium platform (spot size: 55 μm) [56]. Higher resolution techniques such as Stereo-seq (Spot size: 0.6 μm) can better capture various types of cells and describe the renal microenvironment [57], as shown in Table 2 [54, 58–63]. However, the cost is significantly higher, and gene throughput is often limited. Thus, a moderate level of resolution may be most cost-effective for studying the renal fibrotic microenvironment.

Fig. 2.

Technical characteristics and workflow of spatial transcriptomics. a Spatial transcriptomics enables high-resolution gene expression mapping within intact renal tissue sections. This technology combines histological imaging with transcriptome-wide sequencing, preserving cellular spatial relationships to reveal niche-specific molecular mechanisms in kidney injury and fibrosis. b Workflow of spatial transcriptomics. The first step is library preparation, tissue mRNA is captured on a spatially barcoded array, followed by NGS library synthesis. After that, in situ sequencing and alignment is conducted, sequencing coupled with imaging maps gene expression to precise tissue coordinates, generating spatial transcriptomic profiles. Finally, key gene is validated targeted by in situ hybridization.

Table 2.

Spatial transcriptomics technologies in nephrology

Methods	Technology	Sample type	Resolution, μm	Ref.
10x Genomics Visium	Spatial Barcode	Frozen or FFPE tissue sections	55	[58]
10x Genomics Visium HD	Spatial Barcode	Frozen or FFPE tissue sections	2	[59, 60]
Seq-scope	Seq-Scope HDMI-Array	Frozen tissue sections	0.5–0.7	[61, 62]
Stereo-seq	DNA nanoballs	Frozen or FFPE tissue sections	0.5–0.7	[55, 63]

Spatial transcriptomics provides high-resolution insights into AKI pathogenesis by delineating specific cell-cell interactions and microenvironmental dynamics. Spatial transcriptomic analysis of distinct AKI mouse models revealed model-specific immune infiltration mechanisms: a study investigating different AKI mouse models revealed that neutrophil infiltration (mediated by tubular ATF3) predominates in ischemic AKI, while macrophage recruitment (driven by MDK) characterizes septic AKI, demonstrating etiology-specific immune patterns in kidney injury [64]. Spatial transcriptomics sequencing holds significant clinical implications for diagnostic and therapeutic strategies. Integrated spatial transcriptomic analysis identified the thick ascending limb as a metabolically active zone with high oxidative phosphorylation dependence, demonstrating unique vulnerability in both acute and chronic kidney injury models. Clinically, the widely used sedative propofol was found to impair thick ascending limb mitochondrial function, exacerbating tubular damage in ischemia-reperfusion injury models [65]. Regarding the specific roles of macrophages post-recruitment, another study revealed biphasic macrophage infiltration (day-1 and -14 post-AKI), with early co-localization to injured tubules and late proximity to fibroblasts during fibrosis progression. Pseudotime analysis distinguished a novel IGF-secreting EAM subset promoting fibroblast communication [66]. Another study based on spatial transcriptomics specifically focused on the fibrotic microenvironment co-established by injured renal tubule cells and fibroblasts, identifying the Clcf1-Crfl1 signaling axis as the critical molecular mediator sustaining this pathological niche [67]. Research focusing on the transition process from AKI to CKD has revealed a conserved fibro-inflammatory niche characterized by dysfunctional PTCs, fibroblasts, and immune cells in both murine and human samples. This niche is coordinately regulated by the transcription factor Runx2, along with the signaling pathways of PDGF and integrin β2 [68]. While male mice dominate AKI modeling due to female resistance, spatial transcriptomics of a female IRI-AKI model demonstrated conserved cell clustering with sexually dimorphic Cyp4a14/Cyp7b1/Slco1a6 expression, while identifying persistent proximal tubule-macrophage/lymphocyte interactions at 6-week post-injury that mechanistically illuminate the AKI-CKD transition [69].

Both experimental models and clinical biopsies of renal fibrosis consistently reveal a localized distribution of fibrotic lesions [70], a pattern attributed to a “fibrogenic niche” shaped by spatially confined events such as ECM deposition, renal cell injury, inflammatory infiltration, myofibroblast activation, tubular atrophy, and microvascular rarefaction [71–73]. Within this niche, fibroblasts, tubular epithelial cells, and endothelial cells collectively contribute to fibrogenesis through the secretion of ECM components, extracellular vesicles, soluble factors, and metabolites [7]. The integration of single-cell RNA sequencing with spatial transcriptomics which combines gene expression profiles with spatial localization data enables the systematic reconstruction of such microenvironments. In practical workflows, our analytical approach integrates single-cell and spatial transcriptomics data through a multi-step framework. We typically begin by performing UMAP-based clustering on single-cell RNA sequencing data to identify distinct cell populations, analyze their dynamic changes, and resolve key subpopulations by identifying specific marker genes. This is followed by pseudotime analysis to trace cellular trajectories and weighted gene co-expression network analysis to delineate core signaling axes. Concurrently, spatial transcriptomics data are processed through spatial proximity profiling to quantify neighboring cell types surrounding key subpopulations. Additionally, we investigate cell-cell communication in functionally specialized regions by integrating single-cell interaction data, such as ligand-receptor pairs prioritized through NicheNet or CellPhoneDB, to reveal coordinated biological processes across spatial and temporal dimensions [16, 56, 66, 74]. For instance, integrative analysis of Visium (spot-level) and CosMx (single-cell-level) spatial data from fibrotic kidneys identified four distinct microenvironments: glomerular, immune, tubular, and fibrotic [56]. Notably, the immune microenvironment which composed of dendritic cells, plasma cells, and B and T lymphocytes resides within fibrotic regions and plays a pivotal role in fibrogenesis. A gene signature derived from the fibrotic niche was shown to predict future fibrosis progression in patients [56]. Similarly, in a rat model of hyperuricemic nephropathy, single-cell and spatial transcriptomics mapped a fibrotic microenvironment involving macrophages, epithelial cells, and endothelial cells, revealing significant epithelial expansion and macrophage–epithelial crosstalk [75]. Complement activation has also been implicated in renal fibrosis. C3 enrichment in the tubulointerstitium during fibrosis has been consistently observed in multiple scRNA-seq datasets [76]. Spatial transcriptomics of IRI and UUO mouse models further indicated that C1q is predominantly enriched in macrophages and stromal cells within the fibrotic niche [77].

Artificial Intelligence in Nephrology

Artificial intelligence (AI) refers to computational systems capable of simulating human cognitive functions, such as learning, reasoning, decision-making, and language processing [77]. The integration of AI into medical research dates to the late 20th century [78], and recent advances in sequencing technologies, including single-cell and spatial transcriptomics which have generated vast quantities of disease-associated digital data, creating significant opportunities for the application of deep learning and other machine learning algorithms [78]. For example, one study used a deep learning model to analyze spatial multi-omics data from patients with intrahepatic cholangiocarcinoma, identifying five spatially distinct features with differential prognostic outcomes. This approach also proposed subtype-specific therapeutic strategies, offering a theoretical foundation for personalized treatment and the potential to significantly improve patient prognosis [79]. AI has demonstrated significant clinical utility in pulmonary nodule management, where AI-driven three-dimensional CT reconstruction enables precise vascular watershed mapping for nodule localization. This approach, combined with real-time intraoperative navigation, facilitates watershed-based topographic resection and reduces surgical duration from 120 to 20 min while maintaining procedural precision [80]. In nephrology, a cost-effective machine learning model for early CKD prediction has been developed, incorporating six distinct algorithms (Decision Tree, Random Forest, Multilayer Perceptron, AdaBoost, XGBoost, and LightGBM). This model integrates routinely available clinical parameters including hematological indices, urinalysis results, and demographic characteristics. Validation across a cohort of 22,263 individuals (11,786 CKD patients and 10,477 controls) yielded robust performance metrics, with area under the curve values of 0.9235 and 0.8962 for internal and external validation, respectively [81]. The accelerating development of AI methodologies, coupled with expanding multi-omics data resources, presents transformative opportunities for renal research. We propose that AI-facilitated integration of multidimensional omics data will provide novel mechanistic insights into renal pathophysiology and advance precision medicine approaches in nephrology.

Proximity Labeling Technique in Nephrology

Proximity labeling (PL) technology represents a transformative approach, enabling nanoscale labeling of neighboring proteins, RNA, and cellular components in live cells through enzyme-catalyzed reactions. This powerful technique facilitates comprehensive analysis of molecular interaction networks, protein subcellular localization, and related characteristics [82, 83]. The PL workflow involves fusing an engineered enzyme (e.g., TurboID) to a protein of interest or targeting it to specific subcellular compartments. The enzyme generates reactive intermediates that covalently label proximal biomolecules, which are subsequently purified and analyzed via mass spectrometry or high-throughput sequencing. Crucially, labeling efficiency correlates with molecular proximity, ensuring preferential identification of genuine interactors [84].

DNA microscopy pioneers a chemistry-based imaging method that encodes biomolecule positions via molecular diffusion. This optics-free technique labels transcripts in situ with randomized nucleotides, concatenates adjacent molecules during amplification, and computationally reconstructs subcellular maps. It enables spatial analysis in frozen tissues where conventional microscopy fails.

PL technology has enabled transformative applications across multiple biological research fields. DNA microscopy pioneers a chemistry-based imaging method that encodes biomolecule positions via molecular diffusion. This optics-free technique labels transcripts in situ with randomized nucleotides, concatenates adjacent molecules during amplification, and computationally reconstructs subcellular maps, which enables spatial analysis in frozen tissues where conventional microscopy fails [85]. The LOV-domain-controlled translation localization ligase, developed from PL technology, enables codon-resolution monitoring of translation in specified subcellular regions under physiological conditions. Applied to mitochondrial studies, it revealed 20% of nuclear-encoded mitochondrial genes undergo outer mitochondrial membrane-localized translation, uncovering a regulatory hierarchy for outer mitochondrial membrane-preferential translation and establishing a novel framework for spatially resolved translation research [86]. Quinone methide-assisted identification of cell spatial organization pioneers a chemical PL strategy using quinone methide electrophiles to map cellular interactions across micrometer distances. This enzyme-activated system labels proximal cells without direct contact, revealing spatial gene regulation in tumor-macrophage cocultures and identifying distinct T cell immune niches in mouse spleen via scRNA-seq. The method enables tissue-scale analysis of cellular organization beyond physical contact [87].

PL has enabled proteomic characterization of subcellular regions inaccessible to conventional fractionation methods. For example, using split-TurboID, researchers identified over 100 endoplasmic reticulum-mitochondria contact proteins, including numerous previously unknown participants in this functional interface [88, 89]. In podocyte research, biotin labeling of NPHS2 coupled with mass spectrometry revealed 54 podocyte-specific proteins, including the newly characterized ILDR2. This protein, upregulated in both human and mouse podocytes, may play a role in maintaining podocyte structural integrity [90]. While PL has yet to be applied in renal regeneration and fibrosis studies, its capacity to resolve spatial protein organization under pathological conditions suggests it will provide novel mechanistic insights into these processes.

Other Spatial Multiomics in Nephrology

Spatial Proteomics

Proteins are universally recognized as the fundamental units of biological function. The critical role of post-translational modifications in modulating protein activity and cellular dynamics underscores the unique value of spatial proteomics over transcriptomics, such as phosphorylation and ubiquitination [91, 92]. Imaging mass cytometry (IMC), which uses metal-tagged antibodies coupled with high-resolution laser ablation and mass cytometry, enable the visualization of up to 50 proteins and post-translational modifications at subcellular resolution [93]. Moreover, IMC’s compatibility with formalin-fixed paraffin-embedded tissues significantly enhances its clinical utility.

A pharmacological intervention study in cancer cells revealed that single-cell proteomic profiles remained highly correlated across different cell cycle stages, whereas transcriptomic data exhibited substantial heterogeneity [92]. In breast cancer patient samples and adherent cell lines, IMC enabled precise delineation of cellular subpopulations, with results consistent with those from multiplex immunofluorescence and immunohistochemistry [93].

In renal injury and fibrosis research, the integration of IMC with single-cell sequencing mapped spatially resolved proteomic landscapes in post-transplant AKI samples. These analyses revealed reduced expression of proximal tubule markers (e.g., ECAD and AQP1) alongside upregulated fibrotic proteins in structurally damaged regions, thereby validating mechanisms of acute tubular injury [94]. Similarly, combining IMC with transcriptomics in a mouse model of immune checkpoint inhibitor-associated nephrotoxicity identified macrophages as key regulators of fibrotic progression, with CXCL9 and MMP12 emerging as critical molecular drivers [95]. However, current limitations in the sensitivity and throughput of mass spectrometry-based approaches, along with their relatively weaker integration with spatial context compared to deep-sequencing technologies, suggest that a multimodal strategy combining these methods may offer an optimal solution.

Spatial Metabolomics

Spatial metabolomics enables quantitative mapping of lipids and metabolites in tissue sections at single-cell resolution. Based on matrix-assisted laser desorption/ionization-mass spectrometry imaging, this approach allows spatially resolved analysis of metabolites such as glycogen and N-linked glycans in mammalian tissues [96, 97]. Mass spectrometry imaging delineates intratumoral metabolic heterogeneity by generating spatially resolved metabolite profiles that correlate with tumor progression, which simultaneously maps drug distribution, metabolic conversion and endogenous metabolites, providing mechanistic insights into drug action while guiding precision drug delivery system optimization to address therapeutic challenges posed by tumor heterogeneity [98]. Integrated spatial omics revealed pEMT in situ carcinoma cells remodel the tumor microenvironment via collagen secretion. Spatial metabolomics identified unique phospholipid metabolic signatures in these regions, confirming tumor-stroma metabolic symbiosis and providing novel insights into tumor-stromal metabolic crosstalk [99]. Spatial metabolomics also plays a pivotal role in elucidating specific mechanistic pathways underlying disease pathogenesis: In a study on metabolic dysfunction-associated steatotic liver disease, spatial multi-omics identified GPR35-dependent suppression of anti-inflammatory ELF4 and lipid regulator CIDEA with concurrent SAA upregulation in specific lobular regions. Deficiency reduces hepatoprotective metabolite 3′,5′-IMP in central veins, confirming GPR35’s role in phospholipid homeostasis and inflammation control [100].

Spatial metabolomics also shows promising potential in elucidating the pathogenesis of kidney diseases. Previous studies have shown that MALDI-MSI can facilitate early detection of AKI by identifying early lipid degradation products. For example, a spatial lipidomic atlas generated in a murine model of IRI-induced AKI revealed significant elevation of phosphatidylcholine O-38:1 (PC O-38:1) in proximal tubules, suggesting its potential as an AKI biomarker [101].

Stereo-cell: Spatial Enhanced-Resolution Single-Cell Sequencing with High-Density DNA Nanoball-Patterned Arrays

Recently, a high-throughput single-cell transcriptomics platform named Stereo-cell, based on DNA nanoball (DNB)-patterned arrays, has been introduced. The core of DNA nanoball arrays is the high-density, ordered arrangement of DNA nanoballs, each bearing unique spatial barcodes on a chip, with their coordinates pre-decoded through sequencing. Subsequently, capture probes that are functionalized with molecular barcodes and poly-T sequences are conjugated to the nanoballs. When a tissue section is applied to the chip, released RNA is captured in situ, followed by reverse transcription and sequencing. This process facilitates the precise mapping of each gene’s expression back to its specific spatial coordinate within the original tissue architecture [54].

This technique enables unbiased and precise cell capture, allowing highly accurate quantification of cellular components,such as monocytes in peripheral blood mononuclear cells. With an ultra-high throughput capacity (ranging from 10⁴ to 10⁶ cells), it also facilitates the identification of rare cell populations, including human hematopoietic stem and progenitor cells, which are often missed by conventional sequencing approaches. Stereo-cell combines fluorescence staining and antibody labeling to simultaneously capture cell morphology, transcriptomes, and surface proteins, thus enabling multimodal cell profiling. A key feature of the platform is its use of a silicon chip with a flat poly-L-lysine-coated surface, which supports in situ cell culture combined with transcriptome capture, thereby preserving native cellular states. Moreover, the method shows excellent compatibility with large cells such as skeletal muscle fibers cardiomyocytes, adipocytes, and even protists.

Despite its advantages, stereo-cell faces challenges including fabrication complexity, analytical pipelines, low-level transcript contamination in adjacent cells, and limited multi-omics coverage beyond transcriptomes and specified proteins. In summary, Stereo-cell technology overcomes key limitations of conventional spatial transcriptomics and substantially broadens the scope of transcriptomic analysis. While further optimization is needed for technical implementation and application scenarios, this platform holds strong potential to advance kidney disease research in the future.

Conclusion

Kidney diseases, whether manifesting as AKI characterized by rapid functional decline or progressing to fibrosis-the final common pathway to end-stage renal failure-all involve complex intercellular communication among multiple cell types. Emerging omics technologies, such as scRNA-seq, and spatial transcriptomics have afforded an unprecedented view into kidney microenvironment along with its composition and regulatory mechanisms. Promising strategies for multi-modal integration, such as the combination of single-cell and spatial transcriptomics with epigenomic, proteomic, or imaging data, enable a comprehensive understanding of kidney injury and fibrosis. By employing proximity-dependent labeling to map the spatial functional domains of key proteins and leveraging AI-driven high-throughput mining of multi-omics datasets to identify druggable targets, we can design novel small-molecule therapeutics. These candidates undergo rigorous validation through preclinical studies in murine and minipig models, ultimately advancing to clinical cohort trials for human translation [102].

In AKI research, scRNA-seq has elucidated heterogeneous transcriptional profiles across renal cell populations and identified key cell subsets, such as the S100a9^hiLy6c^hi inflammatory monocyte population that initiates inflammatory cascades, LECs with potent lymphangiogenic capacity, and immune-regulatory LEC subtypes. In the context of renal fibrosis, scRNA-seq has revealed that resident mesenchymal cells, including Notch3+/RGS5+/Pdgfra–pericytes, Meg3+/Pdgfra+ fibroblasts, and Colec11+/Cxcl12+ fibroblasts, are potential sources of renal myofibroblasts. In addition, the use of genetic lineage tracing tools also revealed that renal myofibroblasts are the central cell type in fibrosis progression [103]. Spatial transcriptomics has further delineated the glomerular, immune, tubular, and fibrotic microenvironments that emerge during fibrosis, deepening our understanding of cell-cell interactions underlying disease pathogenesis. The deep integration of single-cell sequencing with spatial transcriptomics and other omics technologies holds promise for achieving cellular-level mapping of “cell state-molecular interaction-spatial localization.” Combined analysis of clinical renal biopsy samples using single-cell and spatial transcriptomics is expected to identify disease-specific molecular subtypes, trace their cellular origins, and pinpoint potential therapeutic targets. Furthermore, this approach enables systematic characterization of immune microenvironments in pathological specimens, allowing development of drug target screening models based on spatial microenvironmental signatures, thereby offering substantial clinical translation potential.

In the context of massive multi-omics data, the integration of AI with complex medical data and multi-omics technologies has accelerated the exploration of disease mechanisms, while demonstrating significant advantages in both diagnostic efficiency and precision in clinical practice. PL technique has advanced our understanding of interactions among neighboring proteins, RNAs, and cells in live-cell states. Other sequencing technologies, such as spatial proteomics, spatial Metabolomics, and stereo-cell, each have their specific focus, and have also achieved various breakthroughs compared to traditional techniques in terms of sequencing depth, accuracy, and sample compatibility. Integrated multi-omics analysis can synergistically combine their strengths and compensate for their weaknesses, holding significant potential for exploring the pathogenesis of kidney diseases in the future.

Fibrosis is a common pathological endpoint of chronic diseases affecting various organs, characterized by excessive ECM deposition resulting from the activation of myofibroblasts [104, 105]. Key convergent pathways driving this process include TGF-β, PDGF, FGF, Wnt/β-catenin, and YAP/TAZ signaling [106–109]. The application of single-cell and spatial transcriptomics in renal fibrosis has begun to elucidate the unprecedented cellular heterogeneity and intricate intercellular crosstalk within the injured microenvironment [6, 20, 56, 110, 111]. This approach is identifying novel pro-fibrotic fibroblast subpopulations and immune-stromal interactions that regulate disease progression. Importantly, the fundamental principles and cellular players uncovered in the kidney, such as conserved pathogenic fibroblast lineages and niche signals, are likely applicable to understanding fibrosis in other organs, including the liver, lungs, and heart. Additionally, immune cells play multifaceted roles in organ fibrosis beyond merely initiating inflammation [7, 19, 112]. They directly regulate fibroblast activation by secreting key cytokines and engage in complex cellular crosstalk within the fibrotic niche. Single-cell and spatial transcriptomics are pivotal for deconvoluting this complexity, as they precisely define pro-fibrotic immune cell subsets, reveal their dynamic transitions during disease progression, and map the location-specific immune-stromal interactions that drive pathological matrix deposition, thereby offering novel therapeutic targets [113, 114]. This cross-organ perspective, enabled by high-resolution technologies, paves the way for the development of targeted anti-fibrotic strategies.

While single-cell and spatial transcriptomics have provided unprecedented cellular-resolution landscapes of renal diseases, revealing microenvironmental heterogeneity, intercellular communication networks, and dynamic disease progression patterns, significant challenges remain. First, technological adaptations tailored for renal samples, such as spatial multi-omics methods compatible with low-input biopsy specimens. Second, the integration barrier across multi-omics platforms remains a significant challenge. The existing spatial metabolomic and epigenomic data lack true spatiotemporal synchronization, which hinders our ability to capture the intricate crosstalk between metabolic, epigenetic, and transcriptional processes. Third, dynamic process monitoring remains limited; current technologies inadequately capture transient cellular state transitions and rely on multi-timepoint sampling, necessitating live-cell dynamic tracing at single-cell resolution. Clinically, translating vast single-cell datasets into actionable biomarkers requires interdisciplinary collaboration, particularly integrating AI with pathology and diagnostics. New technological platforms and analytical frameworks are critically needed to achieve deeper interrogation of biological systems.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

This study was supported by grants from the National Nature Science Foundation of China (responsible for study conception, design, and planning; Grant No. 82370734) and Guangdong Provincial Science and Technology Project (responsible for manuscript writing and decision to publish; Grant No. KS0120240293).

Author Contributions

Wenbiao Wang decided on the topics. Yubing Chen and Zhiming Ye collected literature and drafted the review and figures. Yubing Chen and Wenbiao Wang wrote the manuscript.

Funding Statement

This study was supported by grants from the National Nature Science Foundation of China (responsible for study conception, design, and planning; Grant No. 82370734) and Guangdong Provincial Science and Technology Project (responsible for manuscript writing and decision to publish; Grant No. KS0120240293).