Introduction

Genetic kidney diseases are a major cause of chronic kidney disease (CKD) development and account for the majority of pediatric cases and ~ 10% of adult cases in need of kidney replacement therapy [1]. With the latest technological advances, like next generation sequencing and genome-wide association studies, it is expected that even more genetic cases will be identified. This is good news as in a majority of the cases a genetic diagnosis helps patients and doctors to gain insight into disease prognosis, treatment, or transplant decisions [2]. However, in many cases, there is still no therapy available to halt the disease progression and patients rely on kidney replacement therapy for their survival.

The kidney is a complex organ, home to multiple cell types with specific roles and functionalities. Their coordinated activity enables the clearance of metabolic waste products and balancing fluid and blood electrolytes. The nephron, the functional unit of the kidney, is comprised of different structural segments, each with its own assigned function. Every segment is susceptible to genetic mutations, with the majority of the genetic kidney disorders manifesting as glomerulopathies and tubulopathies. These disorders result in an overall malfunction of the kidney with life-threatening consequences. Although next-generation sequencing technologies can promote the clinical understanding of these genetic kidney diseases by providing guidance for molecular diagnostics and personalized treatment, this alone will not suffice. The combination of next-generation sequencing data and appropriate disease modeling will most likely provide a detailed vision of the complex genetic etiology. This was previously shown in a study where researchers found a variant of the PKHD1 gene, of which mutations cause polycystic kidney disease (PKD), and through in vitro testing of patient-derived cells were able to find ciliary defects in these cells, which were not detected earlier [3].

The development of disease models to understand the mechanisms of human genetic kidney disorders is of great interest. However, meeting the specificities of each disorder when proposing a model is not a trivial task. A plethora of disease models, ranging from zebrafish and C. elegans to rodents and in vitro patient-derived cell culturing, have contributed to the identification of novel genetic causes, new therapeutic targets, and to the development of new treatments. For a better overview of these models, we guide the reader to the work of Molinari et al. [4].

Although the ease of manipulation of in vitro cellular models allows the straightforward dissection of disease molecular mechanisms, the complexity of in vivo (animal) models supports the study of multiple cell interactions and tissue homeostasis in a (patho)physiologic context. Nevertheless, animal models do not fully genocopy or phenocopy human disease due to interspecies differences. They are too physiologically complex for a minimal reconstruction approach. The heterogeneity and the low frequency of certain kidney disorders prompt for the quest of more accurate and human patient-specific disease models [5]. With the current developments in (bio)manufacturing and the robust isolation and generation of disease-specific cells and organoids, in vitro disease models based on organs-on-chip (OOC) technology have been pushed to the forefront of scientific discoveries. OOC technology refers to microfluidic cell culture systems with controlled, dynamic conditions that directly replicate the microenvironment of tissues in the human body. OOC-based approaches are beyond traditional, flat, 2D cultures, as they exhibit tissue- and organ- level functions that are not found in other, more simple, in vitro models. Due to their physiologically relevant read-outs, OOC are used for preclinical drug testing and, more recently, as controlled, miniature representation of specific patients and diseases [5, 6]. In this review, we will first consider the major traits of genetic kidney diseases and discuss the cell-based in vitro models for their study. In this context, we will provide an overview of the current advances in the development of OOC-based in vitro models towards studying genetic kidney disease and discuss their potential use as tools to facilitate further insights into disease pathomechanisms and development of new therapeutical interventions.

Genetic kidney diseases

A large number of human genetic diseases (> 150) that affect kidney development or kidney tissue maintenance lead to functional and structural defects [1], as observed in most cases of CKD [7]. The range of phenotypes associated with genetic aberrations is remarkably wide. The renal cells that are most involved in the pathogenesis of various inherited kidney diseases include the following: (1) podocytes, the main performers of the renal glomerular filtration barrier, and (2) the (proximal) tubular epithelial cells of the nephron responsible of secretion, reabsorption and ion exchange [8]. A comprehensive overview of genetic kidney aberrations, their molecular and biological implications and associated pathological presentation has been prepared by others and is beyond the scope of the current review [9, 10]. As a summary, Fig. 1 includes the most common diseases and their altered phenotype throughout the segments of the nephron. The increased acceptance of early genetic testing and the establishment of national and international registries, allow for a real-time overview of genetic, medical, and family history. In this direction, the web-based registry, established by European Rare Kidney Disease Reference Network (ERKNet) (https://www.erknet.org), covers patients with rare kidney diseases, provides epidemiological and therapeutic management information, and includes patient cohorts for clinical research [11]. The in-depth documentation of new phenotypes and the isolation of patient-derived cells, made available to the scientific community, would support the development of disease-specific in vitro representations of the disease toward personalized therapeutic discoveries, with the ultimate goal to delay the onset of early-stage CKD [12].

Fig. 1
figure 1

Overview of the different segments of the nephron and their associated kidney genetic diseases. Glomerulus: congenital nephrotic syndrome (CNS) groups the genetic disorders that affect key glomerular filtration barrier (GFB) structural proteins. Slit diaphragms are affected by mutations in the genes NPHS1 (Finnish type CNS), NPHS2 (non-Finnish CNS) and WT1 (Denys-Drash syndrome) [7][7]. Within the GBM, the genetic disorders alter the encoding of collagen IV isoforms (Alport Syndrome, AS) [37], laminin (Pierson syndrome) [38], and LMX1B (less known, nail patella syndrome) [35]. Proximal tubule: formed by ciliated epithelial cells that are in close contact with capillaries. On the one hand, most of the disorders affecting the proximal tubule derive from aberrations in genes encoding for transporters of calcium, sodium, and potassium [50]. For instance, mutations in the genes CLCN5, OCRL1, causing Dent disease 1 and 2, respectively. Additionally, mutations in OCRL develop into Lowe syndrome [51]. On the other hand, mutations in genes encoding for key components in metabolic pathways are also associated with proximal tubulopathies. Defects in the CTNS gene lead to an abnormal intra-lysosomal accumulation of cystine that causes cystinosis and acidosis, which could eventually lead to Fanconi Syndrome and kidney failure. Distal tubule: the distal tubule epithelial cells express epithelial sodium channel (ENaC) and potassium channels in the apical membrane. Thus, genetic diseases often affect the regulation of Na–K reabsorption [60]. Mutations in the genes SCNN1B or SNCN1G have been linked to Liddle syndrome, characterized by excessive Na + reabsorption [67]. distal renal tubular acidosis (dRTA) is rooted in mutations in SLC4A1 [68]. The dominant pseudohypoaldosteronism type 1 (PHA1) is associated with NR3C3 mutations whilst recessive refers to SCNN1A, 1B and 1G [69]. Magnesium transporters are also subjected to genetic diseases, such as Bartter syndrome and its variant, Gitelman syndrome. While the former appears in the ascending limb of Henle’s loop, the latter is localized in the distal convoluted tube [70]. Gitelman has been associated with mutations in the SLC12A3 gene, which codes for the sodium-chloride cotransporter, and with CLCNKB gene, coding for renal chloride channel ClC-kb [70]. Collecting duct: the last segment of the nephron is in charge of water and Na + reabsorption. Aquaporin channels (AQP2) are also susceptible to mutations leading to dysfunction [72]

Full size image

In vitro cell-based models of genetic kidney diseases

Several disease models (animal and cellular) for inherited kidney diseases have been developed to identify aberrant pathways and novel therapeutic targets, or as drug screening and testing platforms [4]. Deviating from animal models that are unable to fully recapitulate human pathophysiology, human models are becoming highly appealing [13]. In vitro cell cultures have been shown to be efficiently applied to dissect the molecular pathways of genetic diseases, while providing a fast and cheap platform for high-throughput drug screening. Special attention has been paid to identify new cell sources in such a way that patient-derived cells or cells with a predefined mutation can be tested. Hence, not only can we better understand the pathological manifestations of the disease, but also design targeted and personalized therapies towards the phenotype and/or repairing the defective gene [10]. The easy and non-invasive isolation of podocytes and proximal tubule epithelial cells (PTECs) from patient urine samples can generate primary kidney cell cultures that retain the genetic signature of the disease [10, 14]. However, these cells cannot be cultured indefinitely and/or lose their phenotypical signature and spontaneously dedifferentiate, mainly due to the lack of a kidney-like microenvironment. Via immortalization, their phenotype can be conserved. Similarly, cells can be obtained from patient’s kidney biopsies and subsequently immortalized; however, the invasive retrieval procedure and the lack of availability of kidney material make this approach challenging and less used in practice [10, 15]. Still, immortalized, patient-derived cell lines may lose their sensitivity to external stimuli, such as drugs, becoming a faux representation of an otherwise dynamic and highly responsive biological system. Moreover, since these cell lines cannot be generated from every single patient, they are solely a representation of one specific patient, and the extrapolation to a population of distinct and unique individuals is limited.

Using gene editing tools, such as clustered regularly interspaced short palindromic repeat (CRISPR/Cas) technology, a candidate mutation can be precisely introduced in healthy cells, allowing a fast and easy generation of mutation-specific diseased cells, paving the way toward precision medicine [16]. An example includes the recent application of CRISPR/Cas9 on a human PTECs line to generate a model for cystinosis that revealed novel insights into the molecular mechanisms of the disease and the potential therapeutic effect of a new combination treatment to alleviate the symptomatology of the disease [17]. Using the same technology, we can target and modify virtually any cell type. By introducing the correct version of the gene into the genome of diseased cells, reversing the disease is in sight. Despite the versatility and the specificity of CRISPR/Cas technology that make it a powerful tool for the generation of disease models in vitro, this gene-editing tool still harbors some limitations. The editing efficiency of the CRISPR/Cas system differs vastly among various cell types, which hampers the generation of disease models due to high optimization costs and time consuming experiments. Additionally, CRISPR/Cas editing presents relatively high off-target effects, which could result in a misleading disease phenotype if not properly evaluated. Lastly, some cells may be sensitive to the DNA damage induced by CRISPR/Cas which can trigger apoptosis, preventing the generation and further study of these disease models [18].

The discovery of human-induced pluripotent stem cells (hiPSCs) has been instrumental for the growth of the in vitro disease modeling field [19,20,21]. These indefinitely growing cells are suitable alternatives that overcome ethical concerns related to animal studies and reflect the donor’s genetic background. In short, pluripotency is induced in adult somatic cells using four retrovirally transfected transcription factors (the Yamanaka factors) [22]. Then, hiPSCs can be directed towards the desired cell type by a finely tuned combination of growth factors [23]. Using hiPSCs, it is possible to, theoretically, generate all cell types existing in the kidney, which otherwise cannot be obtained by employing classical isolation strategies. When hiPSCs are isolated from a donor that carries a specific gene mutation, this characteristic will be maintained in the differentiated cells. Furthermore, using healthy hiPSCs and employing gene editing tools, such as CRISPR/Cas, insertions and deletions can be performed, which result in ‘diseased’ kidney cells after differentiation. Creating isogenic models also avoids misinterpretation of the results derived from differences in genetic background among donors. Nevertheless, the fact that these cells possess phenotypic heterogeneity, tumorigenicity derived from undifferentiated iPSCs in the cell population, and above all, inefficient recapitulation of late-onset diseases due to the lack of maturity in iPSC-derived cells limits their application in personalised medicine [24]. To date, glomerular and proximal tubule cells have been successfully generated from hiPSCs [25,26,27].

Kidney-derived cells from hiPSCs and adult stem cells (ASCs) can also be cultured as 3D self-organized multi-cellular structures, known as organoids, which offer a more in vivo-like representation of the cellular heterogeneity resulting in a superior disease model when compared to 2D cell culture. ASCs isolated from kidney tissue and/or urine were used to generate tubuloids, organoids recapitulating the adult kidney tubular epithelium. These tubuloids have then been thought to be included in a biobank in which different patient-derived tubuloids can be used for the study of inherited kidney tubulopathies [28, 29]. Increasing the repertoire of disease-specific organoids will allow us to unravel the role of genotype in the disease process and whether common or divergent disease mechanisms occur in patients with the same diagnosis but different genetic context. The use of high content analysis methodologies, such as RNA sequencing and proteomics, enable the identification of molecular networks that are altered [30, 31]. Nevertheless, organoid maturation is limited, comparable to embryonic kidney during the first trimester, even if maintained for long periods in culture [32]. Thus, their use to model fully developed kidney diseases is limited. Even more, organoid cultures lack physical directional cues that drive the appropriate self-organization for the cells within the organ. To improve the level of maturity of the organoid system, biomechanical stimulation (via flow) and the introduction of vascular networks (via flow or host implantation) have been proposed. These features are crucial for the complete study of phenotype and tissue- and organ- level manifestation of the disease [33]. The implementation of microfabrication and microfluidic devices, such as OOC, enables the introduction of flow in vitro, allowing a better representation of the in vivo situation [75]. The team studied the infection with pseudorabies virus on electrolyte transport, particularly the Na+–Cl– cotransporter and Na+K+–ATPase. A successful analysis of the Na+ reabsorption on the chip supported the evidence that the beforementioned genetic alterations could be investigated further in this setup. The results proved that cells exposed to FSS showed better Na+ reabsorption, tighter cell junctions, more acetylated tubulin expression, and larger cell heights with more microvilli on the apical side, when compared to the static controls [75]. Moreover, this report is the first one of its kind that included viral infections. This proof of concept opens the door to viral transductions on-chip, a field which remained unexplored and could potentially lead to the establishment of more precise in vitro models.

Collecting duct

The collecting duct (CD) is the final segment of the nephron and allows urine to be excreted as waste. The main process in this segment is the reabsorption of water and Na+ [76]. To date, the only available OOC-based model was proposed by Jang et al. by culturing rat inter-medullary CD cells on a single-channel chip [77]. They established that FSS, the addition of arginine-vasopressin and osmolar differences between channels regulate cytoskeletal organization and aquaporin migration towards the apical domain. Moreover, the migration process was reversed after terminating fluidic stimulation [77].

The step forward: a combined KOC model

In the last few years, a substantial number of attempts to develop KOC have been proposed, including conceptual approaches based on computational models [78]. Nevertheless, the tendency is to move towards a system of chips in which each segment can be separately replicated and then connected. The combination of glomerular filtration and tubular reabsorption processes was proposed by Sakolish and Mahler who combined a biotic filter, mimicking glomerular filtration, with a PT chip [79]. Even though the glomerular segment lacked podocytes, their data supports that the combination of segments is not only possible, but also suitable to sustain enhanced junctions, baso-apical polarization, cytoskeletal reorganization and up-regulation of transport proteins. KOC has been combined with remote organs, including the liver, to establish systemic models [80]. A “body-on-a-chip” including kidney, intestine, liver, heart, lung, skin, blood–brain barrier, and brain has been recently proposed [81]. The addition of a diseased KOC in an otherwise healthy line-up of organs would provide further insight into how those genetic diseases that are largely limited to the kidney affect the activity of remote organs. Many kidney diseases manifest in other organs as well and, thus, in the “patient-on-a-chip” approach in which cells carrying the specific mutation, but representing different organs, are loaded on tissue-specific chips, the systemic implications of the disease could be recapitulated and therapeutically targeted [54, 79, 82, 83].

Conclusion

Kidney-on-chip emerges as a powerful tool for the in vitro representation of kidney function as it provides a much-needed dynamic platform for disease modeling and drug testing. From glomerulus to distal tubule, researchers have successfully proposed models for the nephron’s segments focusing on the replication of exchange barriers. Despite promising results and a technology readily and commercially available, the current approaches fall short when mimicking genetic diseases. Novel protocols for harvesting primary cells and differentiation of hiPSCs into individual kidney cell types, combined with gene editing, would shape our understanding of the kidney (patho-)physiology and aid in the development of novel therapies. A step towards the personalized approach has been taken. Accelerating the process is a matter of unifying efforts and strengthening the dialogue between biologists, engineers, geneticists, clinicians, and patients, early in the development of the models. Taken together, the unique capabilities of OOC technology can be used to develop new in vitro models of human genetic disease models.