Can Kidney Organoid Xenografts Accelerate Therapeutic Development for Genetic Kidney Disorders?

Abstract

A number of genetic kidney diseases can now be replicated experimentally, using kidney organoids generated from human pluripotent stem cells. This methodology holds great potential for drug discovery. Under in vitro conditions, however, kidney organoids remain developmentally immature, develop scarce vasculature, and may contain undesired off-target cell types. Those critical deficiencies limit their potential as disease-modeling tools. Orthotopic transplantation under the kidney capsule improves the anatomic maturity and vascularization of kidney organoids, while reducing off-target cell content. The improvements can translate into more accurate representations of disease phenotypes and mechanisms in vivo. Recent studies using kidney organoid xenografts highlighted the unique potential of this novel methodology for elucidating molecular mechanisms driving monogenic kidney disorders and for the development ofnovel pharmacotherapies.

A lack of mechanism-based drugs for most genetic kidney disorders highlights the need for therapeutic innovation. Drug development relies on the use of appropriate experimental models to assess compound safety and efficacy preclinically. Historically, genetically engineered mice (GEM) and kidney cell cultures have been the most widely used methodologies to study disease mechanisms and to test drugs. Although both approaches can provide valuable biologic insight, they fail to reproduce the clinical heterogeneity observed across patient populations, and often present limitations associated with the genetic background of the animals or cells used. Additionally, interspecies differences in molecules and signaling mechanisms of interest can significantly lessen the value of both methods as preclinical tools for the development of pharmacotherapies.

Recently, the derivation of human kidney organoids from human pluripotent stem cells (hPSCs) has emerged as a new experimental tool for renal disease modeling.¹ This is achieved in vitro by directing the nephric differentiation of hPSCs through the induction of metanephric mesenchyme, mimicking the embryonic mechanisms of kidney development.^2–8 Under three-dimensional differentiation conditions, the resulting kidney organoids contain developmental-stage nephron structures composed of podocyte cell clusters resembling glomeruli that are connected to tubular structures sequentially organized into segments expressing proximal convoluted tubule, the loop of Henle, and distal convoluted tubule markers.^5,6 Despite some limitations, the degree of resemblance to the human kidney, indicated by similarities in gene expression profiles and anatomic organization, as well as in the degree of cell type representation, means that hPSC-derived kidney tissues can recapitulate certain phenotypes and molecular mechanisms associated with specific genetic kidney disorders.

Kidney Organoids Can Replicate Genetic Kidney Disease Phenotypes and Mechanisms

Kidney organoids offer unique advantages over standard experimental methods for the study of genetic kidney diseases. First, given the organoids can be generated from patient-derived induced PSCs (iPSCs), gene mutations and alleles of interest can be studied in the genetic makeup of the patient. Preservation of the patient’s genetic background makes iPSC-derived kidney organoids a particularly effective approach to study genetic renal disorders over the use of GEM, in which strain background can affect the phenotype and underlying molecular mechanisms.^9,10 Logistically too, derivation of iPSCs carrying the causal alleles in the patient is comparatively less time consuming and laborious than generating GEM.¹¹

Second, as an alternative strategy mutations can be introduced into wild-type human embryonic stem cells (hESCs) using gene-editing tools such as clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated protein 9, which introduces DNA modifications with high precision at any desired site of the genome.¹² Both patient-derived iPSCs and gene-edited hESCs allow the study of phenotypes driven by single-gene defects, provided the mutations are dispensable for maintaining pluripotency. Critically, the use of hPSC-derived kidney organoids can accelerate the study of causal effects associated with mutations identified by means of linkage analysis or genome-wide association studies.¹²

Third, in vitro kidney organoids are a valuable tool for distinguishing the contribution to a specific disease phenotype of kidney cell–autonomous processes from non–cell autonomous processes driven by immune cells, and by hormones, cytokines, and other circulating molecules. The methodology also offers several advantages over traditional experimental approaches by permitting rapid and scalable generation of three-dimensional human kidney tissues for histologic, cellular, and molecular interrogation.

In the last few years, organoid models for a number of genetic kidney disorders have been developed using either patient-derived iPSCs or genome-edited wild-type hPSCs.¹³ Those models can replicate phenotypes and mechanisms of monogenic kidney diseases, including autosomal dominant polycystic kidney disease (ADPKD),^8,14–17 autosomal recessive PKD (ARPKD),^18,19 nephronophthisis-related ciliopathy,²⁰ congenital nephrotic syndrome,²¹ proteinopathies such as MUC1 kidney disease (MKD),²² nephropathic cystinosis,²³ Alport syndrome-associated basement membrane defects,²⁴ and tuberous sclerosis complex-associated cystic disease and renal angiomyolipoma (AML).²⁵ The rapid progression and wide usage of organoid models is a clear indication of how radically this methodology is transforming the study of human congenital kidney disorders.

Kidney organoids for genetic kidney disease drug screening

The remarkable capacity to recreate disease-associated phenotypes and molecular pathways makes hPSC-derived kidney organoids an excellent tool for target identification and drug discovery. Compared with traditional two-dimensional kidney cell cultures, kidney organoids allow the assessment of drug effects in multiple kidney cell types, and the detection of drug-induced phenotype reversion in three-dimensional tissue structures. In addition, the approach permits direct assessment of compounds engaging human molecules, such as proteins, RNA, or DNA, eliminating the molecular interspecies differences that can arise using rodent kidney cells.

Kidney organoid-based high-content image screening assays have been used to optimize differentiation conditions, quantify basic nephron injury parameters, and to detect cystogenesis in a model of PKD.²⁶ Large-scale image-based assays using customized plates containing 400 organoids were recently used to screen a library of protein kinase inhibitors, to identify molecules capable of blocking cyst formation in an organoid model of ADPKD.¹⁶ This approach identified protein kinase C inhibitors UCN-01 and UCN-02, an IKK inhibitor, and a quinazoline molecule as cystogenesis inhibitors in vitro.¹⁶ In both studies, a second technique confirmed the findings, validating the results of the screenings. Using a kidney organoid-on-a-chip model of ARPKD, Hiratsuka et al. screened and identified molecules T-5224, NSC23766, and 2-MeOE2, inhibiting the FOS/AP1 transcriptional complex, RAC1 and HIF-1, respectively, as Food and Drug Administration–approved compounds that reduced cyst formation in vitro.¹⁸

In addition to drug discovery, in vitro kidney organoid assays have been used to validate drugs identified by other approaches. For example, MKD kidney organoids validated the small molecule BRD4780, which inhibits the cargo receptor TMED9, and was identified in a drug screening of patient-derived epithelial cells, as capable of selectively clearing mutant MUC1 retention.²²

Taken together, those studies demonstrate the feasibility of using kidney organoids for drug discovery and validation. They also highlight the enormous versatility and scalability that modifications in organoid preparation protocols could confer in order to increase the throughput of this approach for drug screens.

Limitations of Kidney Organoid Biology In Vitro

Although kidney organoids can effectively mimic mutation-driven phenotypes, they also present drawbacks. These include a degree of variability and reproducibility that partly stems from variability in the differentiation protocols used, from stochastic cell culture conditions, and from intrinsic properties of the hPSC lines used. In addition, in vitro–cultured kidney organoids have limitations that are consistent across differentiation protocols, including the developmental immaturity of the tissues, scarce vascularization, limited capacity to form collecting ducts, and the presence of off-target cells.²⁷

Application of single-cell transcriptomics to interrogate wild-type kidney organoid composition has shown that both protocols yield fetal-like tissues resembling first and second trimester human kidney.^28,29 This kind of analysis additionally revealed a 10%–20% non-renal cell content, including muscle-, neuron-, and melanoma-like cells.^28–30 Interestingly, longer organoid incubation times in vitro do not reduce off-target cell content, or result in further cell differentiation. In fact, it appears to have a negative effect on the expression of markers of terminal renal differentiation.^30,31 The developmental immaturity suggests that kidney organoids may be best for modeling diseases with phenotypes appearing at early developmental stages than conditions with phenotypes appearing late in development or postnatally.

In addition to the two limitations discussed, organoids maintained under in vitro culture conditions lack most of the specialized vasculature of the kidney.³² In contrast with the extensive vascular network that surrounds the tubules and the blood capillaries that perfuse natural glomeruli in situ, organoids harbor vestigial vascular structures and overall fewer endothelial cells.³² Considering the critical role that kidney vasculature plays in blood filtering, this is a major anatomic deficiency limiting the use of kidney organoids to study mechanisms of renal physiology.

Another important limitation of kidney organoids is the restricted induction of ureteric bud (UB), due to the preferential specification of posterior intermediate mesoderm by protocols designed to induce metanephric mesenchyme. This results in limited representation of collecting duct tissue. Because collecting ducts and nephrons originate from adjacent, yet distinct, regions of the intermediate mesoderm, collecting duct–like structures can be generated independently through the induction of UB epithelium from hPSCs.^19,33–35 Kuraoka et al. showed that UB organoids from PKD1^−/− iPSCs recapitulated the mechanisms of cystogenesis that is observed in the collecting ducts of patients with ADPKD.¹⁷ Using an alternative differentiation strategy, Howden et al. showed that ureteric epithelium can be induced from primitive distal nephron cells present in kidney organoids.¹⁹ Ureteric epithelial structures generated from iPSCs carrying loss-of-function mutations in PKHD1 spontaneously formed cyst-like structures as the primitive ureteric epithelium transitioned toward ureteric stalk.¹⁹ Disease modeling with hPSC-derived UB organoids therefore emerges as a promising methodology to study genetic disorders associated with collecting duct anomalies. Additionally, the combination of hPSC-derived UB cells with nephron progenitor cells³³ offers unique potential to study genetic conditions, using a more complex representation of human kidney epithelial structures.

In vitro approaches have tackled some of the kidney organoid limitations discussed. These include inhibiting specific signaling pathways during differentiation to lessen off-target cell type content²⁸ and incorporation of fluid flow to improve organoid vascularization, resulting in increased endothelial cell content and formation of vascular networks.³⁶ Vascularization furthered nephron development, as a gene expression profile that more closely resembled that of the adult kidney indicated.³⁶ In addition to vascularization, fluid flow enhanced cyst formation in a kidney organoid model of ARPKD by activating cilia signaling, indicating this approach can improve the reproduction recapitulation of genetic disease phenotypes in vitro.¹⁸ Taken together, the results indicated that incorporation of biochemical and biophysical cues mimicking the kidney microenvironment promote renal organoid maturation and expression of late differentiation markers that may be critical for the recapitulation of disease-associated phenotypes in vitro.

Orthotopic Transplantation Improves Kidney Organoids for Disease Modeling In Vivo

The improvements observed on introduction of factors that mimic the kidney microenvironment, such as liquid flow, are consistent with the effects of xenotransplantation on kidney organoid tissue maturation and vascularization. Taguchi et al. first transplanted reconstructed metanephric mesenchyme derived from mouse PSCs beneath the kidney capsule of immunodeficient mice and harvested the resulting organoids after 1 week.⁴ Extensive tubulogenesis concomitant with vascularization was observed in the organoid xenografts with several vascular growth factors detected.⁴ Another key observation from those experiments was the donor’s blood vessels had integrated into the ESC-derived glomeruli, which, as discussed above, is an essential requirement for assembly of the glomerular filtration apparatus during development.⁴ Collectively, those results were consistent with studies by Sharmin et al. showing that xenotransplanted hPSC-derived metanephric mesenchymal progenitors generated vascularized glomeruli 10 days postimplantation.³⁷ Interestingly, the blood vessels were of mouse origin, showing the differentiating human renal tissues possessed the capacity to attract endothelium from the host niche.

In two additional independent studies, van den Berg et al. and Tran et al. showed that orthotopic xenotransplantation of hPSC-derived kidney organoids alone promoted maturation of organoid nephron segments.^16,38 In those experiments, fenestrated and specialized endothelium, and the initial deposition of a glomerular basement membrane, were observed in the transplanted human kidney organoids after 7 days of implantation. Confirming Sharmin’s findings, blood perfusion of both glomerular and tubular structures was observed after 28 days of implantation.³⁸ Nam et al. found similar results.³⁹ They identified elongated tubular structures and areas with stroma cells in the transplanted organoids after 14 days of perfusion.³⁹ This study indicated that besides inducing maturation and vascularization, in vivo transplantation could also increase consistency and reproducibility of kidney organoids, regardless of the hPSC line used.³⁹ The notion was further supported by findings from Subramanian et al., showing infiltration of human organoid tissues by Plvap-expressing mouse renal microvasculature, opposed to NPHS1⁺ human podocyte clusters.³⁰ Most strikingly, scRNAseq analysis further revealed fewer neuronal precursors and melanoma cells in the organoid xenografts, although muscle-like cells persisted, indicating that orthotopic transplantation reduces off-target organoid cell content.³⁰

Three studies have used orthotopic xenotransplantation of hPSC-derived kidney organoids to model human kidney disease in vivo. In the first study, Tanigawa et al. investigated slit diaphragm defects in podocytes from iPSC-derived organoids expressing a mutant isoform of NEPHRIN.²¹ The authors exploited the increased podocyte maturity of transplanted kidney organoids compared with organoids cultured in vitro^30,37,38 to show that 20 days after orthotopic implantation, mutant podocytes lacked the filtration slits separating foot processes that were in podocytes from transplanted control wild-type kidney organoids.²¹ Also, in contrast to wild-type organoids, NEPHRIN was not detected in the junctions between foot processes, instead the protein was retained in the podocyte cytoplasm.²¹ Taken together, those results indicated that transplanted kidney organoids better recapitulated the initial stages of congenital nephrotic syndrome, potentially allowing the study of developmental mechanisms impairing slit diaphragm formation in human podocytes in vivo.

Two other studies used kidney organoid transplantation to model the renal lesions that commonly occur in patients with Tuberous Sclerosis Complex (TSC). Hernandez et al. and Pietrobon et al. independently showed that TSC2-deficient kidney organoids generated from TSC patient-derived iPSCs, or TSC1/2-deficient kidney organoids generated from hESCs, respectively, replicated major phenotypic features of renal AML.^25,40 The fact that similar results were obtained with patient-derived iPSCs or genetically engineered wild-type hESCs also suggested the approaches are equally effective for recreating monogenic disorders with high penetrance.

The TSC studies showed that TSC1/2-deficient organoids developed epithelial cysts that resembled the cystic kidney phenotype of some patients with TSC, and was driven by cell-autonomous and non–cell-autonomous mechanisms.^25,40 In both studies too, the renal manifestations were more prominent on orthotopic transplantation of the organoids into the kidney capsule of immunodeficient rats. Hernandez et al. showed that 2 weeks after implantation, TSC2^−/− kidney organoid xenografts were significantly larger compared with isogenic TSC2^+/− and TSC2^+/+ organoids, mimicking the growth of both AML tumors and cysts that is observed in patients with TSC.²⁵ Faster tissue growth was concomitant with increased levels of mTORC1 activation in TSC2^−/− AML cells and cyst lining epithelium, likely due to amino acids and growth factors not present in vitro.²⁵ The authors further exploited this in vivo model to elucidate a molecular mechanism of tumor survival that could preserve AMLs from complete ablation induced by rapalog therapy, allowing the tumors to regrow when treatment is interrupted.⁴¹

The studies discussed indicate that orthotopic transplantation is critical to improve kidney organoid tissue maturity and cellular composition. Although the approach has yet to be validated for most organoid models of kidney disease, and despite caveats that include limited renal tissue representation, such as the absence of collecting ducts, initial studies indicate that transplantation enhances expression of nephron disease phenotypes, while preserving molecular mechanisms and pathways driven by signals in the kidney microenvironment. In the future, transplantation of renal organoids combining hPSC-derived nephrons and collecting ducts, capable of urine formation, would be critical to investigate how disease mechanisms affect physiology.

Implications for Mechanism-Based Therapeutics

The ability of hPSC-derived kidney organoids to recapitulate human genetic kidney disease in vivo opens new paths for the use of kidney organoid xenografts in therapeutic development. Combined with pharmacogenomics, the approach is particularly suitable for testing patient-tailored therapies to treat monogenic disorders, including ciliopathies or proteinopathies, . Different causal genes may require the use of drugs targeting different molecular mechanisms to treat patient subpopulations⁴². Similar to human cancer xenografts, patient-derived kidney organoid xenografts also address the effect of genetic variability, most critically in drug response genes.^42,43 Incorporation of the individual’s genetic makeup will increase the accuracy of efficacy assessments for targeted therapies. Over time, wide implementation of patient-derived kidney organoid xenografts could accelerate precision drug development and lead to expansion of drug development pipelines, potentially translating into a spectrum of tailored treatment options.

Kidney organoid-bearing animal models can be instrumental for investigating compound pharmacodynamics.⁴⁴ Westerling-Bui et al. used a model of reversible podocyte injury caused by protamine sulfate in athymic rats carrying wild-type iPSC-derived kidney organoids.⁴⁴ They tested a small molecule, GFB-887, which inhibits the transient receptor potential canonical 5 and therefore blocks actin reorganization driven by calcineurin-synaptopodin-Rac1 signaling.⁴⁴ Their results showed that delivery of 10 mg/kg GFB-887 either orally or via the renal artery for 3 consecutive days partially preserved synaptopodin levels, and reduced foot process effacement in podocytes of organoid xenografts exposed to protamine sulfate, indicating the drug could prevent podocyte remodeling caused by protamine sulfate.⁴⁴ Assessment of drug pharmacodynamics in organoids, rat kidneys, and rat plasma samples by means of liquid chromatography indicated that GFB-887 orally dosed for 3 consecutive days resulted in equivalent drug exposure in human organoids compared with rat plasma.⁴⁴ In addition, quantification of GFB-887 accumulation in several tissues, including the liver, adrenal gland, pancreas, and kidney, by means of whole-body autoradiography showed that kidney concentrations of GFB-887 were two- to three-fold higher than plasma concentrations.⁴⁴ As a whole, the study showcased the effective use of kidney organoid xenografts for the preclinical development of new therapies targeting transient receptor potential canonical 5 (TPC5) in podocytopathies.

Xenograft studies also allow the assessment of treatment toxicity and tolerability indicated by changes in body weight and by physiologic and behavioral cues. Off-target drug effects can be assessed in both nondiseased human and host kidney tissues, for example by measuring nephrotoxicity indicators. Detection of biomarker proteins, such as KIM1 or NGAL, in blood samples and in kidney sections can determine the degree of tubule stress or injury the tested compound caused.^45,46 Furthermore, detecting human and rodent isoforms of those proteins allows simultaneous assessment of the degree of damage in organoid and host kidney nephrons. Additionally, finding human isoforms of cytochrome c and histones H1, H2A, and H2B in serum samples can reveal human cell death and tissue necropsy resulting from the treatment tested. Such readouts are particularly valuable when testing therapies targeting organoid xenografts recapitulating kidney tumors.²⁵

The study of in vivo drug-delivery methods with the potential to increase drug efficacy can be effectively performed using kidney organoid xenografts. Experiments with TSC2^−/− iPSC-derived AML organoid xenografts showed the antitumor activity of locally delivered low-dose rapamycin-loaded nanoparticles (500 ng to 2 μg).²⁵ Nanoparticles delivered into the subcapsular space near the lesions effectively disrupted AML cell antiapoptotic mechanisms, activating caspase and promoting AML cell apoptosis in the organoids.²⁵ The efficacy of locally delivered small-dose nanoparticles contrasted with the effect of higher oral doses of rapamycin (0.5 mg/kg) tested separately in organoid-bearing rats.²⁵ Consistent with the limited efficacy of oral everolimus therapy in TSC patients, oral rapamycin had a cytostatic effect on TSC2^−/− organoid xenografts, but failed to significantly reduce organoid xenograft size.⁴⁷ Additional analysis of off-target effects showed that the cell death induced by the local rapamycin nanoparticles was confined to AML organoid cells, with no signs of tissue damage or cell stress in the rat kidney.²⁵ Taken together, the results highlighted the value of transplanted kidney organoids to elucidate disease mechanisms and to test novel therapeutic approaches for renal AML.

One key aspect of modeling disease with transplanted kidney organoids is the requirement for immunodeficient animals. Two strains commonly used in drug studies are nonobese diabetic severely compromised immunodeficient (NOD/SCID) mice and athymic RNU rats. Both strains lack mature T cells, B cells, and natural killer cells, which prevents xenograft rejection. Immunodeficiency, and particularly the absence of innate immune cells, can modify kidney disease progression and may alter the phenotype. In CKD, monocyte/macrophage infiltration and increased expression of proinflammatory cytokines are observed before, or at the time of, cyst initiation in murine models and human patients, and promote cyst growth.^48–50 Moreover, innate immunity is a key driver of tubulointerstitial inflammation and fibrosis associated with PKD,⁵¹ nephrotic cystinosis,⁵² MKD,⁵³ and Alport syndrome,⁵⁴ among other genetic disorders.⁵⁵ Therefore, it is important to consider that the degree of immunosuppression varies among immunodeficient animal strains. Some, including NOD/SCID mice and RNU rats, retain significant innate immunity and therefore better represent the inflammatory microenvironment of the kidney.^56–59

Orthotopically implanted hPSC-derived kidney organoids, building on their capabilities as drug-discovery tools in vitro, emerge as a potentially groundbreaking tool for preclinical therapeutic innovation. This is partly due to the fact that the kidney microenvironment improves organoid tissue maturity, vascularization, and anatomic organization of nephron structures, enhancing expression of disease phenotypes. In contrast, biologic processes that may alter tissue composition in vivo, including cell stress and fibrosis, must be carefully examined. Optimized xenografts generated from patient-derived iPSCs could make it possible to recreate disease mechanisms and phenotypic features in a personalized manner. In summary, as kidney organoid methodologies improve, use of patient-tailored in vivo models, both in academic and industry settings, will likely become instrumental for developing precision therapy pipelines for genetic kidney disorders.

Disclosures

D.R. Lemos reports having an ownership interest in Alphabet Class A, Lithium Americas, Moderna, Novo Nordisk, Pfizer Livent, Tesla, and Toyota; and reports receiving research funding from Brigham and Women's Hospital and Dialysis Clinic, Inc. All remaining authors have nothing to disclose.

Funding

None.

Author Contributions

D.R. Lemos conceptualized the study; D.N. Cabrera-Barragan, T.-C. Kuo, D.R. Lemos, D.O. Lopez-Cantu, and M. Lopez-Marfil wrote the original draft; and T.-C. Kuo and D.R. Lemos reviewed and edited the manuscript.