Cystic renal‐epithelial derived induced pluripotent stem cells from polycystic kidney disease patients

Abstract Autosomal‐dominant polycystic kidney disease (ADPKD) is the most common inherited kidney disease, leading to kidney failure in most patients. In approximately 85% of cases, the disease is caused by mutations in PKD1. How dysregulation of PKD1 leads to cyst formation on a molecular level is unknown. Induced pluripotent stem cells (iPSCs) are a powerful tool for in vitro modeling of genetic disorders. Here, we established ADPKD patient‐specific iPSCs to study the function of PKD1 in kidney development and cyst formation in vitro. Somatic mutations are proposed to be the initiating event of cyst formation, and therefore, iPSCs were derived from cystic renal epithelial cells rather than fibroblasts. Mutation analysis of the ADPKD iPSCs revealed germline mutations in PKD1 but no additional somatic mutations in PKD1/PKD2. Although several somatic mutations in other genes implicated in ADPKD were identified in cystic renal epithelial cells, only few of these mutations were present in iPSCs, indicating a heterogeneous mutational landscape, and possibly in vitro cell selection before and during the reprogramming process. Whole‐genome DNA methylation analysis indicated that iPSCs derived from renal epithelial cells maintain a kidney‐specific DNA methylation memory. In addition, comparison of PKD1+/− and control iPSCs revealed differences in DNA methylation associated with the disease history. In conclusion, we generated and characterized iPSCs derived from cystic and healthy control renal epithelial cells, which can be used for in vitro modeling of kidney development in general and cystogenesis in particular.


| INTRODUCTION
Polycystic kidney disease (PKD) is a heterogeneous group of diseases that can be inherited or acquired. Autosomal dominant polycystic kidney disease (ADPKD) is the most common heritable form of PKD.
Over time, these patients gradually acquire numerous cysts in both kidneys, resulting in renal function decline. Symptomatic treatment consists of blood pressure control, pain, and infection management. In addition, a vasopressin receptor antagonist (Tolvaptan) has become available, slowing renal decline in ADPKD patients with rapid progressing disease. [1][2][3] However, most patients develop kidney failure and need a dialysis of a kidney transplantation before the age of 60. 4 ADPKD is caused by a heterozygous germline mutation in PKD1 (~85%), PKD2 (~15%), or GANAB (~0.3%). 5-7 PKD1 encodes for polycystin-1, a transmembrane protein, which structurally looks like a receptor or adhesion molecule and forms a complex with polycystin-2, a calcium channel encoded by PKD2. GANAB, the alpha subunit of glucosidase II (GIIα), plays a role in glycosylation and quality control of polycystin-1 in the endoplasmic reticulum. 7 Expression of polycystin-1 is high in the fetal kidney and essential for kidney development. 8,9 After nephron formation has completed, PKD1 expression is reduced.
In the adult kidney, the exact function of PKD1 is unclear, but it is required in the renal epithelium to prevent cyst formation.
Cysts arise focally. The so-called "second hit model" refers to the observation that all renal epithelial cells harbor a heterozygous mutation, but only a small proportion of the cells will form a cyst. In this model, somatic mutations affecting the remaining healthy PKD1 allele are proposed to precede cyst initiation. This hypothesis is supported by the observation that heterozygous Pkd1 mice develop only a few cyst, whereas (kidney specific) inducible knock out of both Pkd1 alleles results in a severe cystic phenotype including renal failure, thus recapitulating the human phenotype. 10 Further evidence supporting this second hit model came from mutational studies on DNA from cyst lining epithelium, isolated from human kidney tissue samples, which displayed small somatic mutations or loss of heterozygosity (LOH) in PKD1 or PKD2. [11][12][13][14][15] Moreover, the second hit might also be present in genes other than the one affected in the germline. Evidence for this trans-heterozygous hypothesis is the identification of somatic mutations in PKD2 in cyst DNA from patients with a PKD1 germline mutation and vice versa. 15,16 Also copy number variations (CNVs) and small pathogenic somatic mutations at various loci in the genome of cyst lining cells have been reported. 17,18 However, the contribution of these mutations to cyst initiation has not been proven.
Conversely, there is also evidence against the second hit model. The second hit model does not explain cyst formation in autosomal recessive PKD, in which patients harbor a trans-heterozygous mutation in PKHD1. Nor can it explain the rare patients who are transheterozygous for an incompletely penetrant PKD1 allele and a pathogenic PKD1 allele. 19 In these cases, patients already have both alleles mutated and still exhibit focal cyst formation. Moreover, Pkd1+/− mice develop cysts shortly after induction of renal injury, indicating Pkd1 is haploinsufficient and a second hit in Pkd1 is not required for cystogenesis. 20 Finally, cystogenesis can also be provoked in normal kidneys-without a germline mutation in a PKD gene-by applying renal injury through drugs or ischemia. [21][22][23][24] Therefore, another mechanism for cyst formation has been proposed; the gene dosage model. 25 This model hypothesizes that a variation in PKD1 dosage is the underlying cause of cystogenesis.
Reduction of PKD1 expression levels could be the result of stochastic

Significance statement
Autosomal dominant polycystic kidney disease (ADPKD) is the most common inherited kidney disease, leading to kidney failure in most patients. In approximately 85% of cases, the disease is caused by mutations in PKD1. How dysregulation of PKD1 leads to cyst formation on a molecular level is unknown. The present study has generated induced pluripotent stem cells (iPSCs) of ADPKD patients to study the function of PKD1 in kidney development and cyst formation in vitro. The iPSCs revealed germline and autosomal mutations implicated in ADPKD and displayed an epigenetic memory of kidney epithelial cells, providing powerful models to study ADPKD in vitro.
transcription fluctuations or inactivation of the PKD1 gene by DNA methylation. Indeed, it was shown in mice that lowering Pkd1 expression to approximately 10% of the original level results in a cystic phenotype. 19,26 Interestingly, also an increase in Pkd1 expression was found to result in a cystic phenotype, confirming that regulation of proper PKD1 levels is crucial. 27,28 In the last decade, induced pluripotent stem cells (iPSCs) have proven to be a powerful in vitro system for studying human genetic disorders. 29,30 The advantage of these iPSCs is their self-renewing capacity, allowing indefinite expansion. This enables the use of a wellcharacterized cell line for longer periods of time, reducing variation between experiments and allowing genome editing. Moreover, iPSCs are monoclonal. Importantly, recently developed protocols to differentiate iPSCs into kidney organoids make it a suitable system to study kidney development. [31][32][33] Previously, iPSCs cells have been established from ADPKD patients heterozygous for a PKD1 mutation. [34][35][36][37] Since these iPSCs were derived from fibroblasts, somatic mutations that might have contributed to cystogenesis will be missed. Second, several studies have shown that iPSCs retain an epigenetic signature of the tissue of origin. [38][39][40] This residual epigenetic memory could contribute to a more efficient, directed differentiation back to the tissue of origin. 41,42 In this study, we established iPSCs derived from ADPKD patient cystic epithelial cells and from normal control kidney epithelial cells. Whole-genome mutational analysis revealed heterozygous germline mutations in PKD1 in all patients but no second hit in PKD1 or PKD2. Genome-wide DNA methylation analyses showed little differences between PKD1+/− and normal kidney-derived iPSCs, but did reveal a kidney-specific DNA methylation memory in renal epithelial derived iPSCs, not present in ESCs. These ADPKD iPSCs may provide a powerful model to study PKD1 function and the involvement of the second hit in cyst formation and kidney development in vitro.

| Generation and characterization of normal and cystic epithelial primary cells
To generate human iPSC models, we established primary renal tubular epithelial cell (TEC) cultures from ADPKD kidney explants ( Figure 1A).
Each cell line was derived from a unique cyst, by using the inner epithelial monolayer of individual cysts. As controls, normal TECs were isolated from unaffected regions of kidneys that were resected because of a malignancy. In total, eight TEC lines were derived from two ADPKD patients and two normal individuals (Table 1). Both cyst-derived as well as healthy control TECs displayed a typical epithelial morphology and no difference in karyotype stability ( Figure 1B, Figure S1). To further confirm the epithelial origin of the TECs, we applied immunocytochemistry staining for epithelial junction markers (β-catenin and ZO-1), which showed an epithelial-like honeycomb pattern, similar to an immortalized renal epithelial cell line (RPTEC/hTERT; Figure 1B). In addition, TECs were positive for KRT7, a renal epithelial marker, and negative for fibronectin, a mesenchymal marker, which is highly expressed in primary human fibroblasts ( Figure 1B). These findings were supported by quantitative real-time PCR (qRT-PCR), revealing expression of epithelial junction markers (OCLN, Occludin and CDH1, E-cadherin) and renal epithelial markers (SLC2A1 and L1CAM) in all TEC cell lines ( Figure 1C). In contrast, these cell lines did not express SNAI2/Slug, a mesenchymal marker ( Figure 1C). These results confirm that the TEC lines are of epithelial origin.

| Cyst-derived TECs harbor somatic mutations in various genes but not in PKD1
Both patients were diagnosed with ADPKD based on established clinical criteria. 43 To investigate whether the patients carried a germline mutation in PKD1 and to test if additional somatic mutations were present in PKD1 or in other genes, we performed whole exome sequencing (WES) on TEC lines derived from three unique cysts, for each patient. We found a heterozygous, pathogenic (truncating/frame shift) mutation in PKD1, in exon 41 and 15 in patients 6 and 9, respectively ( Figure 2A). We did not detect additional somatic mutations in To test whether de novo DNA methylation was present at the remaining wild-type allele of PKD1, which could lead to gene silencing, we applied MeD-seq. This technique utilizes the methylationdependent restriction enzyme LpnPI to detect DNA methylation changes. MeD-seq analysis did not reveal increased promoter methylation of the unaffected PKD1 allele or changes in DNA methylation in the transcription start site (TSS, ± 1 kb), the gene body (starting 1 kb downstream of TSS until the transcription end sequence), as well as in gene proximal or distal regions ( Figure 2B and data not shown), nor did we find increased DNA methylation of the PKD2 or PKHD1 alleles suggesting that these genes have not been affected by epigenetic silencing mechanisms ( Figure S2A,B). To test whether the PKD1 or PKD2 mRNA expression level was affected in the ADPKD patientderived TECs, we performed qRT-PCR, showing variation in expression level between samples, but no differences between ADPKD and normal TECs ( Figure 2C). The lack of a second mutation in either PKD1 or PKD2 prompted us to test for the presence of other somatic mutations that might explain cyst formation. Somatic mutations were called through inter cyst comparisons (within each patient) only considering exonic regions and excluding synonymous mutations, identifying a total of 3 to 15 somatic mutations per cyst ( Figure 2D). All mutations were heterozygous, or present in a fraction of the TEC cells, and except for MUC2 were unique for one cyst. One cysts contained a pathogenic somatic mutation in IFT140, a ciliopathy gene that causes a cystic kidney phenotype, 44 suggesting that this second hit could have had played a role in cyst initiation. For the other mutations, no established relationship with PKD has been reported yet.
Thus, our analysis identified germline mutations in PKD1 but no somatic mutations in PKD1 or PKD2. Nonetheless, somatic mutations unique for individual cysts were found in various genes which may have contributed to cyst initiation in a trans-heterozygous manner.

| Cystic and normal renal epithelial cells can be reprogrammed to iPSCs
Next, we established iPSCs of the TEC lines obtained from patients 6, 9, and controls. Of each primary TEC culture, one subclone was used to establish either patient-derived-cyst-iPSC or control renalepithelium-iPSCs. Early passage TECs were transduced with a F I G U R E 1 Generation and validation of normal and PKD-patient derived tubular epithelial cells (TECs). A, Experimental setup: autosomal dominant polycystic kidney disease explants were used to isolate primary TECs, which were reprogrammed into induced pluripotent stem cells (iPSCs). B, Phase contrast microscopy and immunocytochemistry staining of junction markers ZO-1 (tight junction) and β-catenin (adherens junction), renal epithelial marker Keratin-7, and mesenchymal marker fibronectin (scale bar = 50 μm for all panels). C, qRT-PCR to determine expression of epithelial markers OCLN/Occludin (tight junction) and CDH1/E-cadherin (adherens junction), renal tubular markers SLC2A1 and L1CAM, and a mesenchymal marker SNAI2/Slug. RPTEC/hTERT cells and primary human fibroblasts were used as a positive and negative control, respectively. Ct values were normalized for GAPDH. The experiments were performed in triplicate twice; error bars represent the SD of both experiments polycistronic lentiviral vector expressing OCT4, SOX2, KLF4, and MYC, and a tdTomato reporter, under the control of a retroviral promoter (SFFV) that is rapidly silenced during the reprogramming process. 45 Although an equal number of (TEC) cells was plated for transduction, cystic TECs were growing notably slower, resulting in a lower confluence at the time of transduction reducing iPSC colony formation efficiency. However, we did establish over 10 iPSC colonies for each of the original TEC lines. TdTomato-negative iPSC colonies emerged from day 19 post-transduction onward. Morphologically, no differences between PKD and normal iPSC colonies were observed.

| DISCUSSION
Here, we have generated iPSCs from cystic renal epithelial cells. iPSCs from healthy renal epithelial cells have been established previously, 46,47 but this was not yet done from cyst cells. We found that our cyst-iPSCs contain somatic mutations, but only a few of these mutations were also present in TECs they were derived from. In addition, we show that these renal-iPSCs retain residual epigenetic kidney memory, which can be beneficial in directed differentiation to kidney organoids.
Using WES and additional mutation analysis of PKD1 specifically (LR-PCR and MLPA), we found that both ADPKD patients have germline mutations in PKD1, but we did not find somatic mutations in or LOH of PKD1 in TECs derived from cysts of these patients. In addition, we did not detect reduced PKD1 mRNA levels. Off note, we also did not detect increased methylation in the promotors of PKD1, PKD2, or PKHD1, suggesting that epigenetic silencing of these genes did not lead to cystogenesis. The fact that we did not find genetic mutations nor epigenetic alterations in PKD1 could mean that a second hit did not occur in PKD1 or PKD2, but that merely the germline mutation leads to haploinsufficiency which is in line with previous findings that Pkd1+/− mice develop a cystic phenotype when renal injury is induced. 20 Alternatively, a second hit in PKD1 may have occurred in the cyst but was lost during culture of the primary TECs used in this study. This could be explained by polyclonality of the cysts, as reported previously, 48 or a growth advantage in cell culture of cells with a single germline mutation over cells that were PKD1 null, as were reported for other systems, were cancer cells, or were outgrown by wild-type cells in standard culture conditions. 49 Remarkably, we found several somatic mutations in genes other than PKD1 or PKD2, ranging from 3 to 15 mutations per cyst. This is in line with a recent study showing that cyst cells contain somatic mutations in non-PKD1/PKD2, ciliopathy, or cancer-related genes. 18 In concordance with that, many of the genes affected by a somatic mutation identified in our analysis are also linked to the cilium or cancer. Moreover, in one TEC line, a somatic mutation was found in IFT140, a gene that causes a cystic kidney phenotype in mice. 44 In addition, some of the affected genes are known to function in pathways previously linked to cystogenesis; ITCH, as a negative regulator of Wnt signaling like PKD1 itself, 50  Between days 2 and 9, 2 mmol/L of valproic acid was added daily.
Around day 26 onward, the iPSC colonies were picked and expanded.

| In vitro differentiation of iPSCs to EBs
To induce EB formation, iPSCs were dissociated from the MEF feeder layer using collagenase IV 1 mg/mL, harvested by centrifugation at Antibodies are listed in the Appendix (Table 1). Images were acquired with a Leica SP5 confocal microscope and processed using Fiji.

| DNA isolation
Cells were collected by centrifugation after collagenase treatment and lysed overnight at 37 C in lysis buffer (0.2% sodium dodecyl sulfate and 1 mg/mL Proteinase K). The next day, phenol and chloroform extractions were performed and DNA was precipitated using isopropanol and washed with 70% ethanol. Finally, DNA was dissolved in 10 mM Tris buffer (pH 7.5).

| Whole exome sequencing and analysis
Genomic DNA (gDNA) was collected from TECs at passage <5, DNA was sheared in a Covaris S220 instrument, and prepared for sequencing using

CONFLICT OF INTEREST
The authors declare no potential conflicts of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request. The WES and MeDseq data from this study have been submitted to the Gene Expression Omnibus 67 database under the accession number PRJNA600136, except HuES-8, which is available under PRJNA375757 as these data were already published.