Multidisciplinary approaches for elucidating genetics and molecular pathogenesis of urinary tract malformations

Abstract

Advances in clinical diagnostics and molecular tools have improved our understanding of the genetically heterogeneous causes underlying congenital anomalies of kidney and urinary tract (CAKUT). However, despite a sharp incline of CAKUT reports in the literature within the past two decades, there remains a plateau in the genetic diagnostic yield that is disproportionate with the accelerated ability to generate robust genome-wide data. Explanations for this observation include: (1) diverse inheritance patterns with incomplete penetrance and variable expressivity; (2) rarity of single-gene drivers such that large sample sizes are required to meet the burden of proof; and (3) multi-gene interactions that might produce either intra- (e.g. copy number variants) or inter- (e.g. effects in trans) locus effects. These challenges present an opportunity for the community to implement innovative genetic and molecular avenues to explain the missing heritability and to better elucidate the mechanisms that underscore CAKUT. Here, we review recent multidisciplinary approaches at the intersection of genetics, genomics, in vivo modeling and in vitro systems toward refining a blueprint for overcoming the diagnostic hurdles that are pervasive in urinary tract malformation cohorts. These approaches will not only benefit clinical management by reducing age at molecular diagnosis and prompting early evaluation for comorbid features, but will also serve as a springboard for therapeutic development.

Overview

Congenital anomalies of the kidney and urinary tract (CAKUT) are a commonly occurring group of birth defects that constitute the primary cause of end-stage renal disease (ESRD) in children. These anomalies occur in ~5 per 1,000 live births and represent 40%−50% of pediatric ESRD cases worldwide ^1–6. CAKUT is also associated with increased incidence of chronic kidney disease (CKD), hypertension and cardiovascular disease with advanced age ⁷. We and others have shown that the risk of CKD increases over time among individuals with CAKUT and most cases show signs of renal impairment after the third decade of life ^8–17. Individuals with CKD are 5–10 times more likely to die before reaching ESRD, and ESRD cases have a high cardiovascular mortality ^18–22. Thus, understanding the pathobiology of CAKUT is paramount for developing new strategies to preserve kidney function, reduce cardiovascular-related morbidity, and ultimately diminish the burden that this population places on the health care system.

An entry point toward understanding the molecular underpinnings of CAKUT is the identification and functional assessment of genetic lesions contributing to the disorder (reviewed extensively elsewhere^23–26). To date, genetic studies on CAKUT have been based predominantly on small cohorts or pedigrees, are often studied using targeted approaches, and have historically been restricted to monogenic inheritance paradigms ²⁴. However, such approaches have been only partially successful toward explaining the genetic architecture hallmarked by CAKUT. Additionally, improved sequencing technologies and DNA array-based techniques have uncovered myriad of molecular insults ranging from single nucleotide variants (SNV) to copy number variants (CNV) encompassing tens to hundreds of genes. In this review, we summarize briefly the clinical and genetic complexity of CAKUT, and highlight recent combinatorial approaches which leverage large-scale genomic studies coupled to functional modeling in vertebrates and in vitro models. Together, these multidisciplinary approaches aim to overcome the limitations of traditional methodology and accelerate our understanding of CAKUT.

Introduction to CAKUT and the current state of the field

Clinical characteristics of CAKUT.

The clinical presentation of CAKUT is complex, from the standpoints of both multi-organ as well as multi-system involvement. More than 100 syndromes have been described in association with renal or urinary tract anomalies ²⁷. Although various forms of CAKUT may appear as part of a systemic condition involving multi-organ manifestations, the majority of cases present nonsyndromic forms confined to the kidney and urinary tract ²⁸. This observation suggests that aspects of genitourinary development are regulated by a repertoire of cell-type specific effectors. CAKUT can affect various parts of the kidney and urinary tract (Figure 1)^29–35, including the renal parenchyma (agenesis, dysplasia), ureter (duplication, dilation, obstruction, reflux), bladder (exstrophy, diverticulum), or urethra (posterior urethral valves, PUV), all of which are governed by tight spatio-temporal developmental regulation that is sensitive to the hormonal and metabolic environment ³⁶. Defects in urinary system morphogenesis range from alterations in the number, structure and/or position of the kidneys; obstructive or non-obstructive dilatation of the urinary tract; to dysplastic kidney lesions, including cystic disorders. Moreover, sex-specific differences in kidney development and, subsequently, in the prevalence of kidney disease and ESRD have been noted ^{9, 37–39}. Therefore, it is not surprising that multiple signaling cascades and pathways are required for normal kidney development in a complex yet interactive fashion.

Figure 1. Schematic of the human kidney system depicting proximity and prevalence based distribution of congenital anomalies of kidney and urinary tract.

Topological subdivision of the phenotypic complexity of CAKUT delineated at the anatomical structural level. This classification is commonly used for clinical purposes. The coloring scale reflects the estimated prevalence of conditions represented in the figure on the basis of large scale cohort and autopsy study findings ^{30, 32, 35}. The prevalence estimates of conditions range from rare (urethral atresia: 1 in 100,000)³¹, (bladder exstrophy: 1 in 50,000)³⁴, to common (vesicoureteral reflux and duplicated ureter: 1/100)^{29, 33}.

Variable penetrance and expressivity are reported within and across families who present with the condition. Phenotypic variability has been documented among family members with the same genetic defect, ranging from asymptomatic structural abnormalities to advanced kidney disease ^{40, 41}. Diverse forms of CAKUT have also been observed among members of the same family, emphasizing the phenotypic complexity of the disease ^{42, 43}. Moreover, monozygotic twins with phenotypically discordant CAKUT may serve as a model to study somatic mosaicism, stochastic effects, and epigenetic regulation ^{44, 45}. The clinical heterogeneity of CAKUT sub-phenotypes poses significant challenges for human genetics studies aimed at the identification of genes underlying these conditions. While certain disease phenotypes seem to be enriched for rare variants in specific genes ^{46, 47}, in the majority of cases the correlation between clinical presentation and the underlying genetic causal mutations is limited.

Genetic etiology of CAKUT: Twin, family, and cohort studies uncover novel candidates.

Genetic, environmental, or pharmacologic disruption of distinct stages in kidney and urinary tract development can cause the phenotypic spectrum observed in CAKUT ⁴⁸ (Figure 1). Although several environmental factors, including angiotensin converting enzyme (ACE)-inhibitors or maternal diabetes and obesity ^49–53 have links to CAKUT, a strong genetic predisposition to the disease is supported by the occurrence of familial forms ^54–56. While most individuals with CAKUT are sporadic cases, the recurrence risk of CAKUT among relatives (15–20%) in several family-based studies indicates a strong genetic contribution to the pathogenesis of the disease ^{55, 57–60}. One study reported CAKUT incidence as high as 51% among first-degree relatives, although this estimate is likely inflated due the high consanguinity of the cohort studied ⁵⁹. Further, there are high concordance rates in twin and sibling studies for CAKUT phenotypes; for instance, concordance rates of almost 80% and 35% in monozygotic and dizygotic twins, respectively, are observed in primary vesicoureteral reflux (VUR) ^{61, 62}.

Diverse inheritance patterns and single gene drivers.

Several lines of evidence support the role of single-gene disease drivers in both syndromic and nonsyndromic forms of CAKUT ⁵⁵. To date, ~50 single-gene causes for isolated CAKUT and >100 syndromic forms of CAKUT caused by mutations in single genes are described in the Online Mendelian Inheritance in Man (OMIM) database (https://www.omim.org) ^{63, 64}. Genes linked to Mendelian CAKUT were initially mapped using single or few, clinically ascertained families containing multiple affected individuals, which facilitated gene identification using positional cloning; however, a major caveat is that large families are rare, and the disease burden might affect the ability to procreate ^65–67.

Knock-out mouse models manifesting a phenotypic spectrum that recapitulates human CAKUT have been developed ^68–70, supporting the notion that a substantial proportion of both syndromic and non-syndromic CAKUT may be caused by single-gene defects in humans. However, monogenic forms of the disease seem to account for a small portion of cases and accurate heritability estimates in humans remain poorly deduced ^{53, 71–73}. In family-based linkage analysis studies, multigenerational disease segregation suggested multifactorial or dominant inheritance with variable expressivity and reduced penetrance in most kindreds ^74–77. Autosomal and X-linked recessive inheritance paradigms have also been implicated in human CAKUT ^78–81.

However, the observation that a majority of CAKUT cases occur sporadically, suggests a role for either de novo mutations, recessive inheritance, or complex disease determination ^{53, 82}.

The reduction in cost, rapid turnaround time, and advancements in bioinformatics approaches have fueled whole exome or whole genome sequencing (WES/WGS) efforts toward the discovery of new genes in sporadic and familial CAKUT ^83–86. Next-generation sequencing studies have also influenced the revision of risk estimates and prevalence of genes previously thought to contribute to CAKUT incidence ^{5, 87}. For example, in a study from the Netherlands an unexpectedly low incidence of PAX2 and HNF1B mutations was reported, previously thought to explain up to 5–15% of CAKUT cases ^{64, 82, 88}. Nicolaou et al. analyzed 453 CAKUT patients, PAX2 was found to explain 8% and 0.7% of all cases of kidney dysplasia/hypoplasia and overall CAKUT, respectively, while HNF1B was only identified in 3% and 0.2% of multicystic dysplastic kidney (MCDK) and overall CAKUT, respectively ⁸⁷. It is possible that the decrease from the original estimates of attributable risk of CAKUT genes is due to: (a) the “regression to the mean” phenomenon where, due to the selection of extreme phenotypes and random effect in discovery studies, there is over-estimation of effect size, and the subsequent risk measures are predicted to regress to the underlying population mean over time; (b) the expansion of sequencing efforts to a broad clinical phenotypic gamut of CAKUT, which is likely more complex than once thought; or (c) the estimates might be diluted by mild CAKUT cases that do not have an underlying genetic etiology. Hence, the attempts to establish population prevalence of mutations based on small and selected cohorts must be taken cautiously.

Multi-gene effects in CAKUT.

Case-control studies have unmasked unappreciated pleiotropic effects, genotype-phenotype correlations, and multi-locus interactions that would be challenging to uncover in single pedigrees. Recently, such approaches have been used to investigate the role of rare CNVs and SNVs in sporadic cases of CAKUT with complex genetic etiologies. Initial studies illustrated that up to 10–15% of kidney malformations are attributable to large rare CNVs ^89–93. For example, evaluation of a modestly sized renal hypodysplasia (RHD) cohort offered insights into hitherto unknown pleiotropic effects. We detected pathogenic CNVs in 22.5% of individuals with syndromic malformations and 14.5% of individuals with isolated urinary-tract defects ⁹⁰. Notably, the majority of known pathogenic CNVs detected in this cohort have reported involvement in developmental delay or neurocognitive disease. This observation not only suggests common molecular pathways in renal and neuronal development, but also bears clinical relevance since most kidney malformations are identified in utero, while cognitive deficits become apparent later in childhood. Thus, identification of a causal CNV in a child with CAKUT can also potentially improve the clinical management of neurocognitive defects. These data have now been validated in several additional studies ^{89, 92, 94–96}.

Assembly of large cohorts has offered further resolution of CAKUT phenotype by CNV type. In a study designed to define the CNV landscape of CAKUT ⁹⁶, we identified loci that were associated previously with a genomic disorder (GD-CNV) in a significantly enriched fraction of CAKUT patients (4% of 2,824 cases vs 0.6% of 21,498 controls). Our analysis showed that six well-known GD-CNVs account for 65% of patients with a GD-CNV, thus identifying major susceptibility loci for CAKUT. When we investigated the distribution of CNVs among different CAKUT phenotypes, we discovered that upper tract conditions (e.g. kidney agenesis or dysplasia) were associated predominantly with large deletions, whereas lower urinary tract phenotypes such as duplex kidney, VUR, or posterior urethral valves were enriched for duplications.

Genome wide association studies (GWAS) have also contributed to the identification of candidate susceptibility loci in association with some CAKUT phenotypes, including VUR^97–99, hypospadias ¹⁰⁰, and bladder exstrophy ¹⁰¹. Interestingly, the effect sizes detected for common variants in association with bladder exstrophy, hypospadias, and VUR were unusually large and suggest that moderately penetrant common variants may play a significant role in the predisposition to CAKUT ¹⁰¹. Together, the observation of incomplete penetrance, variable expressivity of disease, and possible complex inheritance in CAKUT posit that focusing on simple models of disease determination is inadequate to resolve the genetic underpinnings of disease.

Second-site modification can also influence CAKUT-related phenotype^102–104. Humans with autosomal dominant polycystic kidney disease (ADPKD) exemplify this phenomenon, in which a transcriptional network, in addition to multiple causative genes, might explain onset and severity of polycystic kidney disease ^{105, 106}. ADPKD cases with co-occurrence of HNF1B mutations manifest earlier onset of disease than individuals harboring hallmark PKD1 or PKD2 pathogenic variants in isolation¹⁰⁶. Concordantly, a role for contributory modifier genes was demonstrated in mouse models of autosomal-recessive polycystic kidney disease (ARPKD)¹⁰⁷. In the mouse, Hnf1b has been shown to bind specifically to the Pkhd1 promoter to stimulate transcription, and as a result, a dominant-negative Hnf1b mutant allele results in the inhibition of Pkhd1 promoter activity ^{108, 109}. Thus, transgenic mice expressing the dominant-negative Hnf1b mutation under the control of a kidney-specific promoter develop kidney cysts ¹¹⁰. Collectively, these studies and similar exemplars provide a molecular basis for the observed variability in disease progression in cystic kidney disease ¹¹¹.

Mouse models of CAKUT have elucidated mechanisms of kidney development.

Understanding CAKUT has been facilitated by the use of in vivo models, in particular, the mouse. Initial studies involving mouse models were limited to forward genetic approaches, however gene targeting (reverse genetics) strategies have both accelerated the validation of CAKUT associated genes, and also informed the role of these genes in kidney development and disease ⁷⁰. Mouse models have proven to be a useful tool for identifying and investigating signaling pathways such as glial cell-line-derived neurotrophic factor (GDNF)-RET, renin-angiotensin system (RAS), Wnt, and fibroblast growth factor that mediate kidney and urinary tract development ^112–116. Moreover, these animal models provide a spatiotemporally accessible context for defining the genomic and transcriptomic space in which susceptibility and modifier genes impact kidney development and disease.

While many single-gene knockout mouse models result in phenotypes that recapitulate human CAKUT, studies have reported discordance in the effect of genetic mutations on kidney phenotypes between the two species. For instance, mutations in angiotensinogen (AGT), renin (REN), angiotensin converting enzyme (ACE), or angiotensin II receptor type 1 (AGTR1) in humans are associated with autosomal recessive kidney tubular dysgenesis ^{117, 118}. In mice however, mutations in RAS genes result in severe hydronephrosis, medullary hypoplasia and in some cases duplicated ureters, phenotypes not observed in humans with the same gene mutations ^{119, 120}. Alternatively, it is possible that the variable genetic background of inbred mouse strains protect against disease severity or expressivity; for example C57BL/6 are resistant to several kidney phenotypes ^{70, 121}.

Complementary approaches to elucidate the molecular pathogenesis of CAKUT

Current challenges to overcome.

Although the community has much to celebrate with regard to understanding the underpinnings of aberrant kidney development, substantial interpretive challenges remain in human genetic studies. (1) In Mendelian paradigms, private mutations in small pedigrees are insufficient to explain causality. (2) The limited size of many cohort-based association studies often prohibits genome-wide significance for candidate CAKUT loci. (3) For CNVs, driver genes or modulating loci within a CNV can only be assigned with genetic approaches through the identification of phenotypically similar cases with smaller overlapping CNVs or point mutations in single genes encompassed by the CNV. (4) Mutations in some loss-of-function intolerant genes are embryonic lethal in mice and thus limits their utility in studying kidney development and pathology. These challenges bring forth the opportunity to employ alternative research models.

An intermediate vertebrate model: the zebrafish.

The zebrafish (Danio rerio) is a robust vertebrate model for studying kidney development and modeling human kidney disease (reviewed elsewhere ^122–124). Anatomical simplicity and rapid development of zebrafish provides the opportunity to study organs early during development ¹²⁵. The zebrafish pronephros is formed by four days post fertilization (dpf) and can be visualized by either whole-mount immunostaining or with transgenic lines that elaborate kidney structures with fluorescent proteins driven by kidney-specific promoters ¹²⁶. Importantly, zebrafish larvae display cell type similarity and structural resemblance of pronephric components with the mammalian glomerulus and proximal convoluted tubule (Figure 2A–C) ^{123, 126–128}. Lower urinary tract structures are absent in zebrafish, however, surrogates in the distal anatomical regions can be used to model anomalies of the ureters, bladder, or urethra.

Figure 2. In-vivo and in-vitro models to study CAKUT.

A. Schematics of mouse (left) and zebrafish (right) urinary system. Mammals have two symmetrical kidneys, each of which are connected to the urinary bladder through a collecting ducting (ureter). The mammalian kidney is composed of millions of kidney structural units known as nephrons. The zebrafish urinary system is comprised of two functional pronephric tubules but share structural and functional similarity with mammalian urinary system. The absence of lower urinary tract organs is a limitation of the zebrafish model.

B. Schematic of the mammalian kidney structural unit is represented as a straightened nephron tube. The glomerulus capsule (blood filter) is drawn as a pink cup-like sac structure. The tubule can be subdivided in to three segments including proximal, intermediate and distal tubule. The proximal tubule consists of the neck (black), proximal convoluted tubule (yellow) and proximal straight tubule (blue). The intermediate tubule consists of descending thin limb (light orange) and ascending thin limb (gray). The distal tubule consists of thick ascending limb (green), macula densa (light blue), distal convoluted tubule (red) and connecting tubule (indigo). The last segment of nephron consists of collecting duct (orange).

C. The zebrafish pronephros is drawn as straight tubule. The glomerulus capsule is represented as pink cup-like sac structure. The zebrafish tubule lacks the intermediate segment and contains only the proximal segment (neck, proximal convoluted tubule and proximal straight tubule; black, yellow and blue, respectively) and distal portion (distal early, corpuscle of stannius and distal late tubule; green, light blue and red, respectively). The last segment of nephron is the duct (orange, also known as cloaca).

Physiological relevance and assays of variant pathogenicity.

Zebrafish studies have supported a role for certain genes in both kidney development and disease. Many loci are involved in pathways such as GDNF-RET and Wnt ⁸². For instance, the transcription factor PAX2 (zebrafish orthologue, pax2a), which is part of the GDNF-RET pathway, has been shown to have an important role in zebrafish pronephros development through the ablation of the orthologous zebrafish locus ¹²⁷, consistent with PAX2 mutation-bearing humans ¹²⁹. Additionally, mutations in BMP4 and SIX2 (also GDNF-RET pathway genes) are associated with kidney hypodysplasia; studies carried out in zebrafish provided further evidence of disease association for these BMP4 variants ¹³⁰.

For decades, the zebrafish molecular toolkit was restricted to labor-intensive forward genetics approaches ¹²⁷. However, reverse genetics methods including morpholino-based knockdown and CRISPR/Cas9 genome editing accelerated the use of zebrafish as a tractable tool for modeling genetic findings in CAKUT ^131–133. Accumulating studies have shown the utility of both stable mutant or transient suppression models for testing disease relevance and the direction of variant effect with in vivo complementation studies ^{85, 134–140}. In a recent example, we showed that CRISPR/Cas9 F0 mosaic mutants or transient knockdown of the candidate gene greb1l in zebrafish results in proximal convolution tubule area reduction, recapitulating human phenotype ⁸⁵. Additionally, we used this model to determine the pathogenicity of four missense variants identified in cases ⁸⁵.

Dissection of CNVs.

CNVs are a major contributor to CAKUT ^{81, 89, 90, 93, 141–143}. Recently, zebrafish have been utilized to identify both phenotype driver genes as well as intra-CNV genetic interactions. We reported a significant enrichment of heterozygous deletions of 22q11.2 (a.k.a. DiGeorge syndrome locus) in CAKUT subjects without overt clinical manifestations of DiGeorge syndrome, compared to controls ¹⁴⁴. A minimal 370 kb region containing 9 protein-coding genes was exclusive to kidney defects. Seven of nine transcripts had an orthologous locus in the zebrafish genome. For four genes (lztr1, pi4ka, serpind1, and slc7a4), gene suppression models were indistinguishable from controls; however, for crkl, aifm3 and snap29 morphants or F0 mutants, we observed altered kidney convolution and pronephric tubule length. We also tested genetic interaction; pair-wise suppression of subeffective doses of aifm3 with snap29 exacerbated kidney phenotype in zebrafish but loss of crkl in combination with either of the other two genes resulted in no detectable additive or epistatic effects. Together, this study highlights zebrafish as a tractable model for CNV dissection.

Anatomical limitations and surrogate models of CAKUT in zebrafish.

Lower urinary tract structures including the urinary bladder are not present in zebrafish, thus making it difficult to model certain human disorders including VUR, bladder obstruction and ureterovesical junction obstruction. Even so, there are reports in which the zebrafish pronephros terminus (a putative homolog to end of mammalian nephric duct which inserts into the bladder) has been used to model lower urinary tract malformations ^{139, 145, 146}. Recently, WNT3 was associated with severe human congenital urological malformations including bladder exstrophy-epispadias complex; in concordance with the human mutational data, knockdown of wnt3 in zebrafish resulted in cloaca abnormalities including expansion of the cloaca lumen at 3 dpf and 4 dpf ¹⁴⁶. In other instances, the gene driver might not be abundantly expressed in the pronephros limiting functional modeling in zebrafish. For example, our recent discovery of high burden of 16p11.2 microdeletions in CAKUT and its attribution to TBX6 gene dosage ^{96, 143} required direct functional modeling studies in the mouse, since Tbx6 is not readily detectable in the zebrafish pronephros and no reliable readouts were available.

In sum, zebrafish are not perfect anatomical models of CAKUT. However, the functional conservation of certain genes allows for the testing of a genetic hypothesis generated from human data of an imperfectly matched anatomical structure.

Humanized tools: In vitro models of kidney pathology.

In some instances, in vivo vertebrate models are hampered by a lack of genomic conservation, or they cannot provide precise human cellular contexts. Pluripotent Stem Cells (PSCs) are able to differentiate into multiple kidney lineages with self-renewal capabilities when grown under defined conditions ^147–150. PSCs constitute a wide variety of cell types including Embryonic Stem Cells (ESC) derived from the mammalian blastocyst stage and somatic cells reprogrammed to an embryonic-like-state (termed Induced Pluripotent Stem Cells; iPSCs) ^{151, 152}. Human kidney tubular cells present in urine can also be used to generate human iPSCs which retain the cell origin epigenetic memory ^{147, 153}.

Together these models can potentially provide a controlled environment to study kidney disease and explore underappreciated pathways involved in kidney development that might be intractable using in vivo models.

Compared to in vivo models, human-derived iPSCs (hiPSCs) may provide increased accuracy for modeling species- and individual-specific aspects of pathophysiology, which are not recapitulated in animal models. This is due to their ability to retain the entire genomic background in which the mutation arose, which is integral for modeling intricate genetic pathways and networks in diseases of multifactorial origin such as CAKUT ¹⁴⁷. Genomic backgrounds affect the penetrance and expressivity of disease phenotypes. The use of hiPSC technology preserves such features and allows for the investigation of complex genetic mechanisms, which can be obscured by secondary effects of the genetic background in animal models ^{154, 155, 156}.

Advancement in 3D culture conditions have made it possible to generate mini 3D organ structures termed as “organoids”, multicellular structures that have the capacity to self-organize in-vitro ¹⁵⁷. Such methodology has overcome the limitations of 2D cultures, whereby branching of the ureteric bud was unattainable ¹⁵⁸. Although recapitulation of the kidney’s structural and cell-type complexity in vitro is challenging, evolving strategies that induce ureteric phenotypes could facilitate modeling of specific CAKUT defects. Human iPSCs have been differentiated into kidney organoids, nephron like structures with characteristics of podocytes, distal and proximal tubules, and loops of Henle (Figure 3) ^159–163. For example, induction of hiPSCs into either ureteric tip or nephron progenitors, followed by in vitro recombination, has been shown to generate distal nephron structures that fuse to collecting ducts ¹⁶¹. Additionally, genome editing in hiPSCs can target CAKUT-related genes and subsequently recapitulate human kidney phenotypes. For example, CRISPR/Cas9 disruption of polycystic kidney disease genes PKD1 or PKD2 in hiPSCs did not affect hiPSCs pluripotency or epithelial morphogenesis, and as a promising proof-of-principle, showed tubular expansion by cyst formation in kidney organoids ^{164, 165}. Still, challenges regarding the tractability of kidney organoids in modeling human kidney diseases remain. Furthermore, the structural differences of in vitro tubular organoids from in vivo organ systems have made it difficult to conduct a wide range of functional assays which restrict, at present, the pathobiological hypotheses that can be tested ^{166, 167}. However, efforts to optimize culture conditions are ongoing with the objective of expanding the use of these models ^168–170. In the future, complementary studies of 3D tissue architecture with single cell multimodal analysis could help unravel molecular etiology and structural consequences of mutations that predispose to CAKUT.

Figure 3. In vitro models to study CAKUT.

Kidney organoids can be generated from pluripotent stem cells derived from rodents or humans. A. Human fibroblasts or peripheral blood cells can be reprogrammed to generate human induced pluripotent stem cells (hiPSCs) that harbor patient-specific genetic variants. These can be differentiated into nephron/mesodermal progenitors or into ureteric epithelium. Recombination of these lineages in appropriate culture conditions generates cellularly complex 3D organoids. Far right: glomeruli are stained with podocalyxin (purple), proximal tubule epithelial cells are stained with Lotus tetragonolobus lectin (red), and ureteric epithelial cells are stained with E-cadherin (green). Scale bar: 100 μm. B. Similarly, rodent embryonic stem cells (ESCs) or Wolffian duct can be recombined with metanephric mesenchyme (MM) and cultured. Resultant hybrid organoids form segmented nephrons with functional structures. Far right: MM-derived epithelial cells are stained with peanut agglutinin lectin (red) and ureteric epithelial cells are stained with Dolichos biflorus (green). Scale bar: 50 μm; image courtesy of S.K. Nigam ^161–163.

Future perspectives

Capturing the “missing heritability” of CAKUT.

For a small proportion of CAKUT cases there is a high-risk actionable genotype underlying the disease. However, for the majority of individuals, genetic risk factors are either elusive, or when present, account for a minimal incremental risk for disease, thus limiting their utility for genetic counseling and management ^{24, 171}. Extending variant identification to systematic inclusion of a wider spectrum of variant types (e.g. protein coding and non-coding variants; single nucleotide variants (SNV) and structural variants (SV)) will address the issue of missing heritability as demonstrated in other complex disease studies ^172–176. We anticipate that reduced costs will facilitate a transition from WES to WGS; this will reduce current approaches to a single platform capable of detecting genomic lesions within the coding and non-coding genomic space. Still, complex rearrangements will require more sophisticated methodologies to detect large numbers of breakpoints in a single event (multi-breakpoint SVs) or structurally polymorphic loci with multiple allelic formations (multi-allelic SVs) ^177–182. Recently, two platforms that offer single molecule sequencing with long-fragment reads and improved throughput have emerged: (1) Single Molecule Real-Time sequencing (SMRT) from PacBio and, (2) Oxford Nanopore’s MinION, a real time nanopore-based DNA sequencing instrument ^183–186. The utility of these technologies has been demonstrated in ways including the reconstruction of long-range haplotypes and the enhanced mapping of complex SV ^186–191.

Investigation of non-coding regulatory variants has also contributed to greater understanding of the genetic landscape of several complex genetic diseases ^{181, 192–196}. Regulatory variants play crucial roles in transcriptional regulation by modulating transcriptional factor binding, chromatin states as well as epigenetic modifications ^197–202. Combined information from WGS and transcriptomic data from relevant cell types can inform pathogenicity and impact of non-coding variants on mRNA quantity and mRNA splicing ^{203, 204}. One persistent challenge for CAKUT is to obtain the pertinent urinary tract tissue for study since biopsies or surgical resection of kidney and urinary tract are rarely conducted in these patients. Even when a nephrectomy or surgery for correction of a ureteric or urethra defect are performed, the collected specimen is of limited utility due to the temporal window at which it is collected (post-natal rather than at the relevant in utero developmental timepoint). Further, tissue quality is often compromised by secondary changes like fibrosis or infection and inflammation. The use of patient-derived iPSCs in conjunction with genomic analysis will likely overcome these hurdles. Another challenge is to obtain the ideal cell type. Bulk RNA-sequencing is confined to describing the average transcriptional profile across a heterogeneous population of cells, which may obscure signals of interest pertaining to specific cell types. The application of single-cell profiling is anticipated to provide greater understanding of specific kidney cell profiles while revealing novel regulatory mechanisms that underlie cell functions ^205–210. These approaches offer specificity by avoiding biologically-relevant variability at the individual cell level and enable reduced cost by averaging across large and diverse cell populations, such as those present in the kidney.

Clinical and translational possibilities for CAKUT

While utilizing rapidly evolving technologies in clinical practice enhances diagnostic capabilities and precision medicine provision, it also creates opportunities for complications that have not been foreseen until now. The perceived benefits present a source for concern, particularly in relation to the paucity of knowledge related to clinical actionability of variants identified, particularly those of unknown significance that have not been reported previously or that are present at low frequencies in the general population and display incomplete penetrance. Evidence supporting clinical actionability for most genetic disorders including kidney pathologies varies ²¹¹. The need to devise and implement standardized, evidence-based approaches to accurately characterize the clinical actionability of genomic findings is of particular importance for CAKUT, because almost all patients are now diagnosed in utero while the long-term outcome may remain unpredictable for decades. CAKUT patients are currently at risk of under-detection of severe CKD as well as overtreatment of relatively mild kidney disease, which burdens an individual’s health and also society. The actionability profiles of CAKUT disorders is foreseen to facilitate genetic counseling and improvement of health care decisions. For instance, genetic variants potentially mediate gene functions related to drug absorption, distribution, metabolism, and excretion ^212–215. Consequently, genetic associations identified in CAKUT may suggest therapeutic development strategies. However, the direct translation of genomic findings into effective treatments has only been demonstrated for a small subset of genetic variants identified ^{216, 217}. Ultimately, development of multidisciplinary systems capable of integrating and converting genetic data into potential drug candidates may help researchers to uncover novel drug targets to improve kidney function survival of patients.