A Human Kidney Tubuloid Model of Repeated Cisplatin‐Induced Cellular Senescence and Fibrosis for Drug Screening

Abstract

In advancing pathophysiological models to assess renal drug responses, kidney organoids derived from human pluripotent stem cells mark notable progress. However, replicating aging‐ and senescence‐related pathologies remains a challenge. In this study, an alternative model is introduced using “tubuloids”—epithelial‐like structures generated from primary human renal proximal tubular epithelial cells (hRPTECs) isolated from resected human kidneys. Bulk RNA‐seq deconvolution confirmed that tubuloids are highly differentiated and predominantly composed of proximal tubule‐like cells. Exposure to cisplatin increased γH2AX, Kidney Injury Molecule‐1, and Cleaved Caspase‐3, markers for DNA damage response, epithelial damage, and apoptosis, respectively. Repeated cisplatin administration resulted in the upregulation of senescence markers and secretion of inflammatory cytokines, consistent with a senescence‐associated secretory phenotype (SASP). Supernatants from cisplatin‐treated tubuloids triggered myofibroblast activation, suggesting early fibrotic changes. A hRPTEC‐derived tubuloid model of cisplatin‐induced kidney injury is successfully developed that mimics senescence, SASP, and fibrosis—hallmarks of chronic kidney disease. This model offers a promising human‐relevant platform for studying renal epithelial responses and drug screening.

Replicating aging and senescence‐related pathophysiological responses in kidney organoids remains a significant challenge. Human adult renal tubular organoid, tubuloids, are successfully developed recapitulating cellular senescence that is the central pathophysiological mechanism of chronic kidney disease (CKD). The findings suggest that tubuloids represent a promising alternative model for studying CKD.

1. Introduction

Chronic kidney disease (CKD) is a progressive and irreversible condition characterized by glomerular sclerosis, tubular atrophy, interstitial fibrosis, and inflammatory cell infiltration.^{[ 1 ]} Despite its significant impact on global health, the underlying pathogenic mechanisms of CKD are poorly understood, and no definitive cure has been identified.

Recently, a shared mechanism underlying inflammation and fibrosis in CKD has emerged. This model suggests that DNA damage in tubular epithelial cells serves as the initial trigger, activating a cascade involving the DNA damage response, cell cycle arrest,^{[ 2 ]} cellular senescence,^{[ 3} , ^{4 ]} and the emergence of the senescence‐associated secretory phenotype (SASP).^{[ 5 ]} To explore these processes, researchers have used various experimental systems, including cell lines, animal models, and kidney organoids derived from human induced pluripotent stem cells (iPSCs)^{[ 6 ]} or human embryonic stem cells (ESCs). These models have advanced our understanding of CKD pathogenesis and therapeutic strategies.^{[ 7 ]} However, a major limitation remains: they inadequately replicate the aging process, particularly cellular senescence, which is prominent in humans with lifespans exceeding 80 years. Rodent models, with lifespans of just 2 years, cannot fully capture the spectrum of age‐related diseases. Furthermore, iPSC‐ and ESC‐derived kidney organoids mainly resemble fetal kidney stages, limiting their relevance for modeling age‐associated conditions, such as CKD.

In this context, “tubuloids,” 3D structures that mimic renal tubules, were first introduced in 2019, derived from adult primary human renal proximal tubular epithelial cells (hRPTECs).^{[ 8 ]} Schutgens et al. pioneered this approach by culturing primary hRPTECs from adult human kidneys and urine, successfully generating tubuloids that modeled viral infections in proximal tubules and hereditary diseases.^{[ 9 ]} Expanding on this foundation, we refined the tubuloid generation protocol using resected human kidneys.^{[ 10 ]} Our research has focused on evaluating tubuloid responses to both nephrotoxicant agents and potential therapeutics. Although these advances mark significant progress, the development of a robust pathophysiological model for aging‐ or senescence‐related kidney diseases using tubuloids remains a compelling and evolving area of investigation.

Cisplatin, a commonly used therapeutic agent for various malignancies, is known to cause nephrotoxicity, commonly referred to as cisplatin‐induced nephropathy.^{[ 11} , ^{12 ]} This condition is characterized by DNA damage primarily in the renal proximal tubules.^{[ 13 ]} In response, affected cells undergo cell cycle arrest to facilitate DNA repair; however, this often leads to cellular senescence and the development of a SASP. These processes are believed to be key contributors to the progression from acute kidney injury (AKI) to CKD.^{[ 14} , ^{15 ]} Therefore, modeling cisplatin‐induced nephropathy provides a valuable platform not only for studying nephrotoxic effects but also for replicating the pathophysiological transition from AKI to CKD.

The primary goal of this study was to assess the potential of tubuloids as a disease model by exposing them to cisplatin, aiming to replicate nephropathy, induce cellular senescence, and simulate the long‐term progression toward CKD.

2. Results

2.1. Primary hRPTECs form a Highly Polarized Structure

To generate tubuloids, hRPTECs were isolated from the kidney cortex, embedded in basement membrane matrix, and cultured with specific growth factors (Figure 1A). The process involved precise mincing of cortical tissue and selective culturing in a serum‐free medium enriched with EGF, yielding a population predominantly composed of proximal tubular epithelial cells (Figure 1B,C).

Figure 1.

Formation of highly differentiated tubuloid structures from primary hRPTECs. A) Schematic of the establishment of tubuloids from hRPTECs isolated from adult human kidneys. B) Image of the human resected kidney used during the establishment of hRPTECs. C) Representative image showing hRPTECs. Scale bars: 50 µm. D) Two‐week protocol for generating tubuloids from 2D‐cultured hRPTECs. E) Representative image demonstrating the seeding of hRPTECs on an ultralow attachment plate. Scale bars: 25 µm. F) Representative images of tubuloids at different magnifications (×10 and ×40). Scale bars: 100 and 25 µm. G) Representative image of tubuloids at ×40 magnification, showing multiple layers along the Z‐axis. Scale bars: 25 µm. H) Immunostaining of makers for renal proximal tubular epithelial cells (LRP2/Megalin and LTL). Scale bars: 25 µm. I) Schematic workflow illustrating the deconvolution of bulk RNA‐seq data from tubuloids using a scRNA‐seq reference dataset derived from primary human kidney samples. J) UMAP plot of the reanalyzed primary kidney scRNA‐seq dataset (GEO accession number: GSE118184) used in the deconvolution analysis. K) Dot plot showing the expression of top three marker genes for each identified cell type. L) Deconvolution results showing the estimated proportions of constituent cell types in four individual tubuloid samples.

Tubuloid formation occurred over ≈2 weeks (Figure 1D). In particular, hRPTECs were seeded into ultralow attachment plates (Figure 1E), followed by Matrigel embedding and supplementation with FBS, EGF, FGF2, and HGF. Multiple tubuloids were formed in a single well, with nearly uniform sizes. The epithelial cells formed tubular‐like structures that encircled the basement membrane gel (Figure 1F). The 3D architecture was confirmed by manipulating the microscope along the Z‐axis direction (Figure 1G). Notably, the cells formed a monolayer structure similar to tubules or renal cysts, rather than a spheroid‐like structure. Furthermore, tubuloids expressed differentiation markers associated with proximal tubular epithelial cells, such as lotus tetragonolobus lectin (LTL) and LDL Receptor‐Related Protein 2 (LRP2/Megalin), confirming their highly differentiated and polarized structure (Figure 1H).^{[ 16 ]}

Furthermore, we sought to identify the cellular composition of these highly differentiated tubuloids. Total RNA was extracted from the tubuloids and subjected to bulk RNA‐seq. Deconvolution analysis was subsequently performed using a previously published single‐cell RNA‐seq dataset of primary human kidneys.^{[ 17} , ^{18 ]} This analysis revealed that the majority of cells within the tubuloids (≈60–65%) exhibited gene expression profiles characteristic of proximal tubular epithelial‐like cells (PT‐like cells) (Figure 1I–L).

2.2. Cisplatin Exposure Induces Tubular Structure Collapse and Elicits Acute DNA Damage Response (DDR) in Tubuloids

To evaluate the responsiveness of tubuloids to nephrotoxic agents, we exposed them to cisplatin. Cisplatin induces cytotoxicity in proximal tubular epithelial cells primarily via Organic Cation Transporter 2 (OCT2)–mediated uptake,^{[ 19 ]} resulting in direct DNA damage (Figure 2A). In line with prior studies demonstrating DDR activation, we observed significant upregulation of γH2AX—a marker of DNA double‐strand breaks—in vivo and in vitro.^{[ 20 ]} Initially, tubuloids were treated with cisplatin at three concentrations (0.2, 2.0, and 20.0 µg mL⁻¹) to determine dose‐dependent responses. Notably, high‐dose cisplatin treatment collapsed the 3D tubuloid architecture (Figure 2B), indicating extensive DNA damage (Figure 2C–E). Furthermore, Immunostaining showed decreased expression of differentiation markers such as LRP2 and Na+/K+‐ATPase (Figure 2D,G–J), suggesting cellular dedifferentiation in response to injury.

Figure 2.

Effects of cisplatin treatment on tubuloid structure, DDR, differentiation loss, cell death, and EMT. A) Schematic of the pathway by which cisplatin induces DNA damage in proximal tubular epithelial cells. B) Representative images of tubuloids exposed to cisplatin (0.2, 2.0, and 20.0 µg mL⁻¹) or control. C) Immunostaining of marker for DNA damage (γH2AX) in tubuloids exposed to cisplatin (0.2, 2.0, and 20.0 µg mL⁻¹) or control. Scale bars: 100 µm. D) Western blotting of γH2AX and Na⁺/K⁺‐ATPase in tubuloids exposed to cisplatin (20.0 µg mL⁻¹) or control. E) Statistical analysis of γH2AX expression (n = 4). F) Immunostaining of a differentiation marker for epithelial cells, Na⁺/K⁺‐ATPase, in tubuloids exposed to cisplatin (20.0 µg mL⁻¹) or control. Scale bars: 25 µm. G) Statistical analysis of Na⁺/K⁺‐ATPase expression (n = 4). H) Immunostaining of a differentiation marker for epithelial cells, LRP2/Megalin, in tubuloids exposed to cisplatin (20.0 µg mL⁻¹) or control. Scale bars: 25 µm. I) Representative plots of flowcytometry measuring LRP2/Megalin expression in tubuloids exposed to cisplatin (20.0 µg mL⁻¹) or control. J) Statistical analysis of LRP2/Megalin expression (n = 3). K) Immunostaining of an apoptosis marker, Cleaved Caspase‐3, in tubuloids exposed to cisplatin (20.0 µg mL⁻¹) or control. Scale bars: 25 µm. L) Representative plots of flowcytometry measuring Cleaved Caspase‐3 expression in tubuloids exposed to cisplatin (20.0 µg mL⁻¹) or control. M) Statistical analysis of Cleaved Caspase‐3 expression (n = 4). N) Immunostaining of an epithelial injury marker, Kidney Injury Molecule‐1 (KIM‐1), in tubuloids exposed to cisplatin (20.0 µg mL⁻¹) or control. Scale bars: 25 µm. O) Western blotting of KIM‐1 and Vimentin in tubuloids exposed to cisplatin (20.0 µg mL⁻¹) or control. P) Statistical analysis of KIM‐1 expression (n = 4). Q) Immunostaining of an apoptosis marker, Cleaved Caspase‐3, in tubuloids exposed to cisplatin (20.0 µg mL⁻¹) or control. Scale bars: 25 µm. R) Statistical analysis of Vimentin expression (n = 4). All data are presented as means ± standard errors (SEs). Statistical significance is presented by ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001.

Furthermore, the expression of the apoptosis marker Cleaved Caspase‐3 (Figure 2K–M)^{[ 19 ]} and the renal injury marker Kidney Injury Molecule‐1 (KIM‐1 (Figure 2N–P)^{[ 16 ]} increased, reinforcing the tubuloids’ sensitivity to cisplatin and their utility in modeling AKI and DDR.

Interestingly, we found that high concentrations of cisplatin increased the expression of Vimentin, an intermediate filament specific to mesenchymal cells (Figure 2O,Q,R). This finding suggests that proximal tubular epithelial cells undergo epithelial‐mesenchymal transition (EMT) as part of a tissue repair response.^{[ 21} , ^{22 ]} Previous studies have reported Vimentin upregulation following activation of the DDR,^{[ 23 ]} and our results support this sequential activation in cisplatin‐treated tubuloids. These findings provide further insight into the molecular mechanisms driving cisplatin‐induced nephrotoxicity.

2.3. Tubuloids Exhibit Cellular Senescence and Release Inflammatory Cytokines Following Repeated Cisplatin Treatment

To evaluate the potential of tubuloids to replicate chronic cellular responses, we subjected them to repeated cisplatin exposure. Cisplatin was initially administered at concentrations of 0.2 and 2.0 µg mL⁻¹, followed by three medium changes to eliminate residual drug. Cells were then cultured under standard conditions for one day to complete a single cycle, which was repeated five times (Figure 3A). The highest dose (20.0 µg mL⁻¹), which caused extensive cell death and structural damage (Figure 2), was excluded to focus on long‐term effects.

Figure 3.

Tubuloids exhibit cellular senescence and release inflammatory cytokines following repeated cisplatin treatment. A) Experimental design for repeated cisplatin treatment. B) Representative images of SA‐β‐GAL staining of tubuloids exposed to cisplatin (0.2 µg mL⁻¹) 5 times repeatedly or control. Scale bars: 100 µm. C) Immunostaining of p16 in tubuloids exposed to cisplatin (2.0 µg mL⁻¹) 5 times repeatedly or control. Scale bars: 100 µm. D) Western blotting of p16 in tubuloids exposed to cisplatin (0.2 and 2.0 µg mL⁻¹) 5 times repeatedly or control. E) Statistical analysis of p16 expression (n = 4). F) Western blotting of p21 in tubuloids exposed to cisplatin (0.2 µg mL⁻¹) 5 times repeatedly or control. G) Statistical analysis of p21 expression (n = 5). H) Western blotting of p53 and Lamin B1 in tubuloids exposed to cisplatin (2.0 µg mL⁻¹) 5 times repeatedly or control. I) Statistical analysis of p53 expression (n = 3). J) Statistical analysis of Lamin B1 expression (n = 3). K) Cell Cycle Analysis in tubuloids exposed to cisplatin (0.2 µg mL⁻¹) 5 times repeatedly or control. L) Immunostaining of an inflammatory cytokine, IL‐1β, in tubuloids exposed to cisplatin (2.0 µg mL⁻¹) 5 times repeatedly or control. Scale bars: 100 µm. M,N) Quantitative PCR of IL‐1β (M) and IL‐6 (N) in tubuloids exposed to cisplatin (0.2 and 2.0 µg mL⁻¹) 5 times repeatedly or control (n = 3). Statical analysis was applied to delta‐delta Ct value, not fold increase. All data are presented as means ± standard errors (SEs). Statistical significance is presented by ^* p < 0.05, ^** p < 0.01.

Following repeated treatments, tubuloids exhibited elevated levels of senescence markers, including SA‐β‐Gal, p16, p21 and p53, and a decreased level of Lamin B1 (Figure 3B–J), consistent with prior studies.^{[ 24} , ²⁵ , ²⁶ , ²⁷ , ^{28 ]} Cell cycle analysis also revealed G2/M phase arrest (Figure 3K), a characteristic of senescent renal tubular cells.^{[ 2} , ²⁹ , ^{30 ]} Furthermore, IL‐1β expression was significantly increased (Figure 3L), with qPCR showing 50–100‐fold upregulation of IL‐1β and IL‐6 mRNA (Figure 3M,N), indicating activation of an SASP.

2.4. Secretions Released from Cisplatin‐Treated Tubuloids Strongly Promote Myofibroblast Differentiation In Vitro

We hypothesized that prolonged exposure of tubuloids to high concentrations of cisplatin would promote myofibroblast activation, potentially contributing to tissue fibrosis via the secretion of paracrine factors from tubuloids, as previously reported in vivo.^{[ 31 ]} To test this, we employed a previously established fibrosis bioassay^{[ 10 ]} (Figure 4A). Tubuloids were grouped into three conditions: untreated, treated with 0.2‐µg mL⁻¹ cisplatin, and treated with 2.0‐µg mL⁻¹ cisplatin, with repeated dosing across groups. Following treatment, residual cisplatin was removed via multiple medium changes, and the tubuloids were cultured for an additional 2 days in serum‐free media to generate CM enriched with secreted factors. Subsequently, the collected CM was administered to hLFs (Figure 4B) and MPKFs (Figure S4, Supporting Information). The activation of myofibroblasts was assessed by evaluating the expression of α‐smooth muscle actin (αSMA) and collagen type I.^{[ 32 ]}

Figure 4.

Secretion released from cisplatin‐treated tubuloids strongly promotes fibrosis in vitro. A) Experimental design for fibrosis bioassay. B) A representative image of human lung fibroblasts (hLFs). Scale bar: 100 µm. C) Representative images of fibrosis bioassay stained with anti‐α‐SMA antibody on hLFs cultured with CM harvested from tubuloids exposed to cisplatin (0.2 and 2.0 µg mL⁻¹) or control or directly with cisplatin‐containing media. Scale bars: 100 µm. D) Quantification of αSMA‐positive area in hLFs cultured with CM or directly with cisplatin‐containing media measured using Image J (n = 8–10 in each condition). E) Representative images of fibrosis bioassay stained with anti‐collagen type I antibody on hLFs cultured with CM harvested from tubuloids exposed to cisplatin (0.2 and 2.0 µg mL⁻¹) or control or directly with cisplatin‐containing media. Scale bars: 100 µm. F) Quantification of collagen type I‐positive area in hLFs cultured with CM or directly with cisplatin‐containing media measured using Image J (n = 8–10 in each condition). G) Experimental design of the fibroblast contraction assay using collagen gel. H) Representative images of contraction assay. Each image represents a single well of a 24‐well plate. I) Quantification of area of collagen gel (n = 2). J) Experimental design of the fibroblast migration assay. K) OD 540‐nm fluorescence measurement results (n = 2). All data are presented as means ± standard errors (SEs). Statistical significance is presented by ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001.

CM obtained from tubuloids repeatedly treated with 2.0 µg mL⁻¹ cisplatin exhibited the most significant promotion of fibrosis (Figure 4C–F; Figure S4, Supporting Information). This concentration of cisplatin, as shown in Figure 3, corresponds to the maximal secretion levels of IL‐1β and IL‐6 from tubuloids, indicating consistency. Notably, direct cisplatin exposure to hLFs did not trigger fibrosis, confirming that tubuloid‐secreted paracrine signals, not the drug itself, mediated the fibrogenic effect.

We also assessed the functional impact of the CM on fibroblasts. In a collagen gel contraction assay, CM from tubuloids repeatedly treated with 2.0‐µg mL⁻¹ cisplatin triggered the strongest fibroblast‐mediated gel contraction, indicating heightened contractile activity (Figure 4G–I). Similarly, a migration assay demonstrated increased fibroblast motility in response to CM from the same treatment group (Figure 4J,K). Together, these findings suggest that soluble factors released by senescent tubuloids exhibiting a SASP profile exert paracrine effects that enhance fibroblast activation and function.

3. Discussion

In this study, we succeeded to characterize patient‐derived tubuloids and recapitulate almost all maladaptive responses of renal epithelial cells against cisplatin‐induced toxicity by them, not animal models or iPS kidney organoids (Figure 5 ). By using multiple patient‐derived tubuloids, our strategy has a potential to use tubuloids as a semipersonalized toxicity or drug efficacy platform.

Figure 5.

Tubuloids recapitulates the pathophysiological responses observed during the progression from acute to chronic phases of cisplatin‐induced nephrotoxicity.

As previously reported by Schutgens et al., tubuloids can be generated from both patient‐derived kidney tissue and urine samples.^{[ 8 ]} Building on this, Yousef Yengej et al. successfully differentiated tubuloids into distal nephron and collecting duct lineages,^{[ 17 ]} demonstrating the potential of these models to represent multiple nephron segments beyond the PT. Similarly, Kramann et al. modeled autosomal dominant polycystic kidney disease using patient‐derived CD24⁺ kidney cells,^{[ 33 ]} further highlighting the clinical relevance of tubuloids in genetic disease modeling. Like Yengej's model, our tubuloids predominantly consist of PT‐like cells, though not exclusively, and currently do not replicate the distal nephron or collecting duct compartments. Nonetheless, some distal and collecting duct cell populations are present (Figure 1L), and future studies may focus on refining differentiation protocols to target these segments more effectively. Notably, our simple and accessible protocol allows for the generation of tubuloids from a range of patient‐derived hRPTECs, offering scalability despite incomplete differentiation.

Importantly, our tubuloids respond robustly to cisplatin exposure, exhibiting hallmark features of senescence such as SA‐β‐Gal staining, increased expression of p16, p21, and p53, and decreased expression of Lamin B1, as well as activation of the SASP. This highlights their utility beyond nephrotoxicity screening—as a pathophysiological model for the transition from AKI to CKD, or for CKD itself. Our findings clearly illustrate the deleterious effects of cisplatin on tubuloids, revealing a cascade triggered by DNA damage. A dose‐dependent increase in γH2AX confirmed genotoxic stress, whereas the upregulation of KIM‐1 and Cleaved Caspase‐3 indicated acute injury and apoptosis, respectively—consistent with DDR activation. Additionally, elevated Vimentin levels suggest EMT, a potential component of maladaptive repair.

Cisplatin exposure led to the collapse of the tubuloids’ well‐maintained 3D architecture, alongside signs of cellular dedifferentiation, evidenced by reduced expression levels of Na⁺/K⁺‐ATPase and LRP2/Megalin. Prolonged, repeated cisplatin treatment further induced cellular senescence, marked by increased levels of SA‐β‐Gal, p16, p21, and p53, along with decreased Lamin B1 expression. The senescent tubuloids also developed a SASP, releasing inflammatory cytokines, such as IL‐1β and IL‐6, contributing to chronic inflammation in the surrounding tissue. These secretions significantly promoted fibrosis in adjacent interstitial areas, as demonstrated by the fibrosis bioassay.

CKD is challenging to study due to its complex etiology and strong links to aging and cellular senescence. Current animal models—primarily mice—are limited by their short lifespan and genetic homogeneity, which do not reflect the decades‐long progression or diversity seen in human CKD. This highlights the urgent need for more accurate, human‐based disease models to advance CKD research.

Kidney organoids derived from induced pluripotent stem cells (iPSCs) or ESCs have recently gained attention as potential models for studying kidney diseases.^{[ 34} , ³⁵ , ^{36 ]} However, these organoids typically resemble fetal kidneys, limiting their ability to replicate age‐related characteristics of CKD seen in older individuals.^{[ 37 ]} Furthermore, generating iPSC‐derived kidney organoids requires complex differentiation protocols and specialized expertise, which can hinder widespread application. Meanwhile, primary human kidney cells cultured in 2D systems lack the structural and functional complexity of native tissue, including proper polarity and 3D architecture. Conventional CKD models—whether mouse models, iPSC/ESC‐derived organoids, or 2D primary cultures—face major challenges in replicating aging, senescence, and practical scalability. In contrast, the kidney‐derived tubuloids developed in our study offer a distinct advantage. Formed from patient‐derived cells, these tubuloids inherently reflect donor age and exhibit baseline senescence even without cisplatin exposure (Figure 3B–J), helping bridge a key gap in CKD modeling.

The formation of tubuloids follows a remarkably simple protocol, requiring no specialized techniques. Culture is performed in standard 24‐well ultralow attachment plates, and as shown in Figure 1F, multiple tubuloids develop within each well. During sample collection for western blotting, qPCR, and immunofluorescence, we harvested one sample per well as an independent condition, resulting in multiple tubuloids in each sample, effectively reducing variability across individual tubuloids.

Expanding our study to include tubuloids from multiple patients shows promise for capturing individual differences in kidney responses to various stimuli. Notably, our tubuloid production protocol allows us to derive these structures from different patients simultaneously, under uniform conditions.^{[ 38 ]} These characteristics point to tubuloids’ potential as a novel and useful screening tool for renal toxicity in drug discovery efforts.

Tubuloids have a highly differentiated 3D structure^{[ 17 ]} and mimic physiological responses to nephrotoxicant agents.^{[ 39} , ^{40 ]} Previous research has shown that these well‐differentiated tubuloids express various channels and transporters, providing them with greater physiological functionality than cells cultured in two dimensions. This increased physiological fidelity is likely to explain the observed sensitivity to cisplatin in our study, facilitating the replication of cellular senescence and other key responses. Importantly, tubuloids may capture biological reactions that are difficult to detect in conventional 2D cell cultures.

While this study centers on cisplatin‐induced nephropathy, our broader aim is to establish a robust human kidney‐derived tubuloid model for exploring the pathophysiology of CKD. Although the mechanisms linking cisplatin treatment to long‐term CKD remain poorly understood, even low‐dose cisplatin has been shown to cause persistent kidney damage resembling CKD.^{[ 41 ]} Cisplatin enters proximal tubular epithelial cells through the OCT2 transporter, inducing nuclear DNA damage that activates the DDR. Some affected cells undergo apoptosis, while others attempt repair through cell cycle re‐entry. When damage exceeds repair capacity, cells arrest in the G2/M phase,^{[ 2 ]} a hallmark of cellular senescence.

Senescent cells develop SASP, releasing inflammatory and fibrotic cytokines that trigger chronic inflammation and tissue fibrosis. These cells resist apoptosis and contribute to a self‐perpetuating injury cycle. This senescence‐driven mechanism plays a pivotal role not only in cisplatin‐induced nephropathy but also in the broader progression from AKI to CKD.

In our tubuloid model, we found not only a highly differentiated 3D structure but also pathophysiological responses to cisplatin, such as DDR, structural collapse, dedifferentiation, cell death, EMT, cellular senescence, SASP activation, and fibrotic changes. Our findings suggest that tubuloids can be a precise disease model for cisplatin‐induced nephropathy, faithfully replicating most pathological mechanisms. This highlights their commitment to developing a comprehensive CKD model.

Moving forward, we aim to enhance the complexity and physiological relevance of our disease models by integrating coculture systems and microfluidic technology. In particular, we plan to coculture tubuloids with key cell types such as vascular endothelial cells and fibroblasts. Furthermore, microfluidic devices will be employed to replicate urine flow dynamics, allowing us to better simulate the in vivo renal microenvironment.^{[ 42 ]}

Our successful development of a human kidney‐derived tubuloid model for cisplatin‐induced nephropathy marks a significant step forward in kidney disease research. These tubuloids, capable of recapitulating cellular senescence, SASP, and fibrosis, hold strong promise as a versatile platform for studying various kidney pathologies. Considering that fibrosis is the final common pathway in CKD progression, tubuloids offer a valuable and scalable model for investigating CKD mechanisms and testing therapeutic interventions.

In conclusion, the establishment of tubuloids as a precise disease model for cisplatin‐induced nephropathy represents a major advancement in kidney disease research. Unlike animal models or iPSC‐derived organoids, this platform is based on patient‐derived tissues, offering greater physiological relevance. With ongoing refinement, tubuloids hold strong potential for semipersonalized drug toxicity screening and for deepening our understanding of CKD pathophysiology.

4. Experimental Section

Cell Culture Experiment

Human kidney samples were collected from patients undergoing clinically indicated nephrectomy at the Institute of Science Tokyo Hospital, Tokyo, Japan. The study protocol was approved by the Institutional Review Board of the Institute of Science, Tokyo (M2022‐005). hRPTECs were isolated from tumor‐free regions of kidneys resected for renal cell carcinoma or urinary tract malignancies, using a modified version of a previously established method.^{[ 43 ]} In brief, renal cortex tissue was minced and enzymatically digested with collagenase type II (1.0 mg mL⁻¹) (Worthington Biochemical, NJ, USA). The enzyme reaction was terminated with fetal bovine serum (FBS), and the digested tissue was resuspended in hRPTEC growth medium, consisting of DMEM/F‐12 (Nacalai Tesque, Kyoto, Japan), supplemented with bovine serum albumin (BSA) (Nacalai Tesque), antibiotic–antimycotic (Thermo‐Fisher Scientific, MA, USA), hydrocortisone (Thermo‐Fisher Scientific, MA, USA), ITS liquid media supplement (Sigma‐Aldrich, MO, USA), and human recombinant epidermal growth factor (EGF) (Thermo‐Fisher Scientific, MA, USA). Cells were cultured for 7–10 days at 37°C in a humidified 5% CO₂ incubator. All animal procedures followed institutional guidelines and were approved by the Animal Care and Use Committee (approval no. A2023‐109C).

Human Renal Tubuloid Culture

hRPTECs were seeded onto ultralow attachment plates at a density of 5.0 × 10⁵ cells per well using Advanced RPMI 1640 medium (Thermo‐Fisher Scientific), supplemented with 5% FBS. After a 2‐day incubation period, Matrigel (Corning, NY, USA) was added to stimulate tubuloid formation. The following day, the culture medium was changed to Advanced RPMI 1640 medium with 5% FBS, EGF, fibroblast growth factor 2 (FGF2), and hepatocyte growth factor (HGF) (tubuloid medium). The medium was changed once or twice weekly to maintain optimal cell growth conditions. Tubuloids were usually ready for experimentation within 1–2 weeks of starting the culture.

Bulk RNA‐Sequencing (RNA‐Seq) Analysis

Total RNA was extracted by initially homogenizing tubuloids in Sepasol‐RNA I Super G (Nacalai Tesque, Japan), followed by purification using the Direct‐zol RNA Miniprep Plus Kit (Zymo Research, R2070) according to the manufacturer's protocol. Given the limited amount of RNA, quality control was conducted using an Agilent TapeStation. Samples with DV500 values greater than 75% were validated by Rhelixa, Inc., Japan. RNA‐seq library preparation and sequencing were also outsourced to Rhelixa Inc. Libraries were prepared using the SMART‐Seq mRNA HT LP kit (Takara Bio, Z4792N), and sequencing was performed on an Illumina NovaSeq X Plus system.

Raw reads were trimmed and filtered using fastp v0.22.0^{[ 44 ]} with the options: −q 20 −u 30 −n 10 −l 30 −w 8. Quality control was performed before and after trimming using FastQC v0.12.1, and consolidated quality reports were generated using MultiQC v1.28. The trimmed reads were aligned to the human reference genome (GRCh38 release 45), and transcript‐level quantification was performed using Salmon v1.10.3.^{[ 45 ]}

A log‐transformed TPM count table was generated and subsequently deconvoluted using a primary kidney single‐cell RNA‐seq reference dataset derived from GSE118184,^{[ 46 ]} applying the omnideconv v0.1.1 package^{[ 47 ]} with the DWLS v0.5.2^{[ 48 ]} in R 4.4.2.

To construct the primary kidney single‐cell RNA‐seq reference dataset, single‐cell RNA‐seq data from primary kidney and renal tubuloid cells deposited in NCBI Gene Expression Omnibus (accession numbers: GSE118184^{[ 46 ]} and GSE206538^{[ 17 ]}) were reanalyzed using the Seurat v5.3.0 package^{[ 49 ]} in R 4.4.2. During quality control, cells with a mitochondrial gene proportion exceeding 20% were excluded. Log‐normalized gene counts were calculated using the NormalizeData function, and highly variable genes were identified using FindVariableFeatures. Principal component analysis was performed on the variable genes. The datasets from GSE118184 and GSE206538 were integrated using the IntegrateLayers function with the canonical correlation analysis (CCA) method. Cell clustering and dimensionality reduction were performed using Uniform Manifold Approximation and Projection (UMAP) at a resolution of 0.5. Cell types were annotated based on previously published reports.^{[ 17 ]} The gene expression matrix for the primary kidney dataset was extracted from the integrated Seurat object and subsampled to retain a maximum of 200 cells per cell type. A signature matrix was then constructed from the subsampled expression matrix using the build_model function in omnideconv v0.1.1^{[ 47 ]} with the DWLS v0.5.2^{[ 48 ]} method. Finally, the log‐transformed bulk RNA‐seq expression matrix was deconvoluted using the deconvolution function in omnideconv,^{[ 47 ]} also with the DWLS v0.5.2 method.^{[ 48 ]}

Immunofluorescence Staining

The tubuloids were fixed with 4% paraformaldehyde (PFA) in PBS and permeabilized with 0.1% Triton X‐100 in PBS. Blocking was performed with 3% BSA in PBS, followed by incubation with primary antibodies for 1 h. After thorough PBS washes, the tubuloids were incubated with secondary antibodies for 30 min, washed with PBS again, and mounted on glass slides. Slides were covered with ProLong Glass Antifade Mountant containing NucBlue (Thermo‐Fisher Scientific) and sealed with coverslips for fluorescence microscopy.

The following primary antibodies were used: rabbit anti‐LRP2/Megalin antibody (1:100) (ab236244, Abcam, MA, USA), rabbit anti‐γH2AX antibody (1:200) (#9718, Cell Signaling Technology, MA, USA), mouse anti‐Na⁺/K⁺‐ATPase antibody (1:200) (05‐369, Sigma‐Aldrich), goat anti‐KIM‐1 antibody (1:200) (AF1750, R&D systems, MN, USA), rabbit anti‐caspase‐3 antibody (1:200) (#9664, Cell Signaling Technology), rabbit anti‐Vimentin antibody (1:200) (10366‐1‐AP, Proteintech, IL, USA), rabbit anti‐p16‐INK4A polyclonal antibody (1:200) (10883‐1‐AP, Proteintech), rabbit anti‐IL‐1β polyclonal antibody (1:200) (bs‐0812R, Bioss, MA, USA), rabbit anti‐Collagen type 1 antibody, rabbit anti‐αSMA antibody conjugated with Cy3 (1:400) (C6198, Sigma‐Aldrich), and rabbit anti‐collagen type 1 antibody (1:250) (600‐406‐103, Rockland). For brush border staining, Biotinylated LTL (1:200, Vector Labs B‐1325‐2) was used with Alexa Fluor 633 Streptavidin (1:500, Thermo‐Fisher S21375).

Images were acquired using standard or confocal fluorescence microscopy (Eclipse Ti2, Nikon), and analysis was performed using the NIS‐Elements Advanced Research software and ImageJ Fiji (https://imagej.net/software/fiji/downloads).

Western Blot Analysis

The tubuloids were lysed, and proteins were extracted and purified. WesternBlue (Promega, WI, USA) was used for band visualization. ERK2 (p44/42 MAPK) was detected using rabbit anti‐ERK2 antibody (1:200; Cell Signaling Technology 9102S) as a loading control. Primary antibodies were also used, including rabbit anti‐p16‐INK4A polyclonal antibody (1:200) (Proteintech), mouse anti‐p21 antibody (1:100) (sc‐6247, Santa Cruz Biotechnology, TX, USA), rabbit anti‐Lamin B1 polyclonal antibody (1:1000) (12987‐1‐AP, Proteintech), rabbit anti‐p53 polyclonal antibody (1:1000) (10442‐1‐AP, Proteintech), rabbit anti‐KIM‐1 antibody (1:400) (PA5‐20244, Thermo‐Fisher Scientific, MA, USA), rabbit anti‐caspase‐3 antibody (1:200) (Cell Signaling Technology), and rabbit anti‐Vimentin antibody (1:1000) (#5741, Cell Signaling Technology, MA, USA). Densitometry was performed using ImageJ Fiji (https://imagej.net/software/fiji/downloads).

Flow Cytometry

Tubuloids were digested with 2.5 g L⁻¹ Trypsin and 1 mmol L⁻¹ EDTA in phenol red at 37°C for 20 min. Single cells were fixed with 4% PFA in PBS containing 5% FBS, permeabilized with 0.1% Triton X‐100 in PBS, and stained with primary antibodies for 1 h, followed by secondary antibodies for 45 min. Rabbit anti‐LRP2/Megalin antibody (1:250) (Abcam) and rabbit anti‐caspase‐3 antibody (1:250) (Cell Signaling Technology) were used for primary antibodies. Cells were analyzed using BD FACSLyric (BD Biosciences, MA, USA). Data were analyzed using FlowJo (BD Biosciences, MA, USA).

Senescence‐Associated‐β‐Galactosidase (SA‐β‐Gal) Staining

SA‐β‐Gal staining was performed using a commercial kit (CBL, CA, USA) per the manufacturer's protocol. Four wells of organoids cultured in 24‐well p were stained per condition, and representative images of the entire well were captured.

Quantitative Reverse Transcription‐Polymerase Chain Reaction (qRT‐PCR)

qRT‐PCR was used to assess IL‐1β, IL‐6, and 18S mRNA expression following established protocols. The real‐time PCR detection system used was the Thermal Cycler Dice Real Time System Lite TP700 (Takara Bio, Shiga, Japan) along with TB green Premix Ex Taq II (Takara Bio, Shiga, Japan).^{[ 50 ]} The total RNA was extracted from frozen kidney tubuloids stored at −80°C using Sepasol‐RNAI Super G (Nacalai Tesque, Kyoto, Japan). cDNA was then synthesized from the isolated total RNA using ReverTra Ace (TOYOBO, Tokyo, Japan).^{[ 51 ]} The cDNA generated was then amplified through 40 PCR cycles, with denaturation at 95°C for 5 s and annealing/extension at 60°C for 30 s.

The nucleotide sequences of the primers used for qRT‐PCR analysis were as follows:

human18S.

Forward:5′‐GCAGAATCCACGCCAGTACAAG‐3′

Reverse:5′‐GCTTGTTGTCCAGACCATTGGC‐3′

humanIL‐1β.

Forward: 5′‐ CCACAGACCTTCCAGGAGAATG ‐3′

Reverse: 5′‐ GTGCAGTTCAGTGATCGTACAGG ‐3′

humanIL‐6.

Forward 5′‐ AGACAGCCACTCACCTCTTCAG ‐3′

Reverse: 5′‐ TTCTGCCAGTGCCTCTTTGCTG ‐3′

Cell Cycle Analysis

Cells were adjusted to 2.0 × 10⁵ cells per sample, and a 300‐µL cell suspension was prepared. FxCycle Violet Stain was added to each flow cytometry sample, followed by incubation at room temperature for 30 min in the dark. The stained cells were analyzed using a BD FACSLyric flow cytometer (BD Biosciences, MA, USA), and the data were analyzed using FlowJo (BD Biosciences, MA, USA).

Fibrosis Bioassay

Two types of fibroblasts were used in this study. Human lung fibroblasts (hLFs) were obtained from Lonza (CC‐2512) and cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% FBS. Mouse primary kidney fibroblasts (MPKFs) were isolated from a 12‐week‐old male C57BL/6J mouse. The kidneys were minced and digested in a solution of 1.0‐mg mL⁻¹ Collagenase Type II (Sigma‐Aldrich). The enzymatic reaction was terminated by adding FBS, and the cell suspension was filtered and centrifuged. The resulting pellet was resuspended in DMEM (Nacalai Tesque, Kyoto, Japan) supplemented with 10% FBS, 5‐ng mL⁻¹ FGF‐basic (Thermo‐Fisher Scientific), and antibiotic–antimycotic. MPKFs were used for experiments between passages 3 and 5.

Tubuloids were treated repeatedly with or without cisplatin. After 48 h of exposure, the medium was replaced with serum‐free DMEM to generate conditioned media (CM). Following an additional 48 h of incubation, culture supernatants were collected as CM. hLFs and MPKFs were seeded at a density of 2.0 × 10⁴ cells per well in 8‐well chamber slides (Nunc Lab‐Tek Chamber Slide System, Thermo‐Fisher Scientific) and cultured in DMEM containing 10% FBS. On the following day, CM collected from the tubuloids was added to the fibroblasts and incubated for 48 h. After treatment, cells were washed twice with PBS. For immunofluorescence staining in the fibrosis bioassay, hLFs and MPKFs were fixed with 4% PFA in PBS after PBS washing. The cells were permeabilized using 0.1% Triton X‐100 in PBS and then blocked with 3% BSA.

Contraction Assay

The effects of each CM on the contractile activity of hLFs were assessed using the CytoSelect 24‐Well Cell Contraction Assay Kit (Cell Biolabs, CA, USA), according to the manufacturer's instructions. A cell suspension was prepared at a final density of 2.0 × 10⁶ cells mL⁻¹ and mixed with the Collagen Gel Working Solution at a 2:8 ratio. A total of 500 µL of the mixture was dispensed into each well of a 24‐well plate and incubated at 37°C in a 5% CO₂ atmosphere for 1 h to allow collagen polymerization.

Following polymerization, 1 mL of each CM was added to the corresponding well, and the plates were incubated for an additional 24 h under the same conditions. Gel contraction was assessed by measuring and analyzing the gel area through photographs of each well taken after 24 h of treatment.

Cell Migration Assay

The effects of each CM on the migration of hLFs were evaluated using the CytoSelect 24‐Well Cell Migration Assay (8‐µm pore size, colorimetric format; Cell Biolabs, CA, USA), following the manufacturer's protocol.

A cell suspension was prepared at a concentration of 5.0 × 10⁵ cells mL⁻¹, and 300 µL was added to the upper chamber (inner side of the insert). Subsequently, 500 µL of each CM was added to the lower chamber (well bottom), and the plate was incubated at 37 °C in a 5% CO₂ atmosphere for 24 h. After incubation, non‐migrated cells remaining on the upper surface of the insert were removed using a cotton swab. Migrated cells were stained with Cell Stain Solution for 10 min; the insert was then gently rinsed with water. Next, 200 µL of Extraction Solution was added to each insert and incubated for 10 min. A 100‐µL aliquot of the extracted solution was transferred to a 96‐well microplate, and absorbance was measured at 540 nm using a FilterMax F5 plate reader (Molecular Devices, CA, USA).

Quantification and Statistical Analysis

The number of samples analyzed in each experiment was indicated in the corresponding figure legends. Statistical significance between groups was assessed using one‐way analysis of variance. P‐values < 0.05 were used to denote statistical significance. All statistical analyses were performed using GraphPad Prism (GraphPad Software Inc., San Diego, CA, USA).

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

Y.N, and M.M. contributed equally to this work. Y.N., M.M., Y.S., I.O., R.S., and Y.M. performed the experiments, collected and analyzed data, and wrote the manuscript. Y.M., M.M., Y.S., I.O., and R.S. established the hRPTECs, and S.M. helped in performing the procedure. T.F., H.K., F.A., K.S., T.M., E.S., and S.U. supported the data analysis. K.M. performed the deconvolution analysis of the bulk RNA‐seq data. A.S., Y.N., and H.K. prepared fibroblasts and provided technical support for the fibrosis assay. Y.W., S.Y., and Y.F. resected the patients’ kidneys as standard treatment for malignant diseases. Y.M. developed the experimental strategy, supervised the project, and edited the manuscript. All authors discussed the results and implications and commented on the manuscript.

Supporting information

Supporting Information

Nakao Y., Mori M., Sekiguchi Y., et al. “A Human Kidney Tubuloid Model of Repeated Cisplatin‐Induced Cellular Senescence and Fibrosis for Drug Screening.” Adv. Healthcare Mater. 15, no. 7 (2026): e01795. 10.1002/adhm.202501795

Data Availability Statement

The bulk RNA‐seq data generated in this study have been deposited in Gene Expression Omnibus (GEO) under accession number GSE302756. Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact Yutaro Mori (y‐mori.kid@tmd.ac.jp).