Pkd2 Deficiency in Embryonic Aqp2⁺ Progenitor Cells Is Sufficient to Cause Severe Polycystic Kidney Disease

Visual Abstract

Abstract

Significance Statement

Autosomal dominant polycystic kidney disease (ADPKD) is a devastating disorder caused by mutations in polycystin 1 (PKD1) and polycystin 2 (PKD2). Currently, the mechanism for renal cyst formation remains unclear. Here, we provide convincing and conclusive data in mice demonstrating that Pkd2 deletion in embryonic Aqp2⁺ progenitor cells (AP), but not in neonate or adult Aqp2⁺ cells, is sufficient to cause severe polycystic kidney disease (PKD) with progressive loss of intercalated cells and complete elimination of α-intercalated cells, accurately recapitulating a newly identified cellular phenotype of patients with ADPKD. Hence, Pkd2 is a new potential regulator critical for balanced AP differentiation into, proliferation, and/or maintenance of various cell types, particularly α-intercalated cells. The Pkd2 conditional knockout mice developed in this study are valuable tools for further studies on collecting duct development and early steps in cyst formation. The finding that Pkd2 loss triggers the loss of intercalated cells is a suitable topic for further mechanistic studies.

Background

Most cases of autosomal dominant polycystic kidney disease (ADPKD) are caused by mutations in PKD1 or PKD2. Currently, the mechanism for renal cyst formation remains unclear. Aqp2⁺ progenitor cells (AP) (re)generate ≥5 cell types, including principal cells and intercalated cells in the late distal convoluted tubules (DCT2), connecting tubules, and collecting ducts.

Methods

Here, we tested whether Pkd2 deletion in AP and their derivatives at different developmental stages is sufficient to induce PKD. Aqp2Cre Pkd2^f/f (Pkd2^AC) mice were generated to disrupt Pkd2 in embryonic AP. Aqp2^ECE/+ Pkd2^f/f (Pkd2^ECE) mice were tamoxifen-inducted at P1 or P60 to inactivate Pkd2 in neonate or adult AP and their derivatives, respectively. All induced mice were sacrificed at P300. Immunofluorescence staining was performed to categorize and quantify cyst-lining cell types. Four other PKD mouse models and patients with ADPKD were similarly analyzed.

Results

Pkd2 was highly expressed in all connecting tubules/collecting duct cell types and weakly in all other tubular segments. Pkd2^AC mice had obvious cysts by P6 and developed severe PKD and died by P17. The kidneys had reduced intercalated cells and increased transitional cells. Transitional cells were negative for principal cell and intercalated cell markers examined. A complete loss of α-intercalated cells occurred by P12. Cysts extended from the distal renal segments to DCT1 and possibly to the loop of Henle, but not to the proximal tubules. The induced Pkd2^ECE mice developed mild PKD. Cystic α-intercalated cells were found in the other PKD models. AQP2⁺ cells were found in cysts of only 13/27 ADPKD samples, which had the same cellular phenotype as Pkd2^AC mice.

Conclusions

Hence, Pkd2 deletion in embryonic AP, but unlikely in neonate or adult Aqp2⁺ cells (principal cells and AP), was sufficient to cause severe PKD with progressive elimination of α-intercalated cells, recapitulating a newly identified cellular phenotype of patients with ADPKD. We proposed that Pkd2 is critical for balanced AP differentiation into, proliferation, and/or maintenance of cystic intercalated cells, particularly α-intercalated cells.

Introduction

Autosomal dominant polycystic kidney disease (ADPKD) is the most prevalent monogenic renal condition that occurs in approximately one in 1000 people.¹ ADPKD is characterized by high tubular epithelial cell proliferation and apoptosis, which results in kidney enlargement and development of multiple fluid-filled cysts. This process is accompanied by a significant reduction in functional nephrons, heightened inflammation, and interstitial fibrosis.^2,3 ADPKD is primarily caused by the mutations in polycystin 1 (PKD1) and polycystin 2 (PKD2). PKD1 mutations account for 80%–85% of cases, while PKD2 mutations account for 15%–20% of cases.¹ Patients with PKD2 mutations often present with less severe phenotypes than PKD1 in which cysts occur later in life, delaying ESKD.^4,5 Despite the severity and prevalence of ADPKD, tolvaptan, a selective vasopressin V2 receptor antagonist, is the only approved medication for ADPKD management. However, this treatment leads to severe aquaretic side effects.⁶ The lack of understanding of ADPKD mechanisms and effective treatments creates an urgent need to improve our understanding about ADPKD pathogenesis.

PKD1 and PKD2 are transmembrane proteins that form a heterodimeric complex through their cytoplasmic domains.⁷ This complex functions as a mechanosensor for maintaining the differentiation states of renal epithelial cells and controlling tubule diameter.⁸ PKD2 is a member of the transient receptor potential family and functions as a Ca2⁺ channel.⁹ The maintenance of appropriate Ca2⁺ gradient by PKD2 is essential, as its disruption is thought to be critical for the development of ADPKD.¹⁰ Hogan et al. found that there were significantly reduced levels of PKD2 in the exosomes of patients with ADPKD with PKD1 germline mutations.¹¹ This raises the possibility that there is a rigid stoichiometry of the polycystin complex, and if one of the members of the complex, such as PKD1, is downregulated, then the other members will be as well. It is unclear how this would be regulated. Although cysts in ADPKD can be derived from any nephron segment, the cysts derived from the distal renal segments, particularly collecting ducts, are more prominent and larger than those derived from other segments.^12,13

Because both Pkd1^−/− and Pkd2^−/− showed embryonic lethality, conditional knockout mice of Pkd1 and Pkd2 using Cre-LoxP system have been developed.^14–16 In these studies, different constitutive or inducible Cre drivers were used to inactivate Pkd1 and Pkd2 globally or in some organs and tissues, including the kidney. These efforts have illustrated the critical role of disrupted Pkd1 and Pkd2 in the cyst formation. However, in most, if not all, cases, Cre was not expressed in a tissue-specific or cell-specific manner, complicating data analyses, and making it hard to evaluate the contribution of a particular renal segment or cell lineage in polycystic kidney disease (PKD) development.

Distal convoluted tubules (DCT1 and DCT2, marked by sodium chloride cotransporter [NCC]), connecting tubules, and collecting ducts are critical for regulating fluid, electrolyte, and acid–base homeostasis and innate immune defense against urinary tract infection. These roles are played by specific types of cells, including principal cells and intercalated cells. Principal cells produce water channel Aqp2 for water reabsorption. Intercalated cells express V-ATPase, which harbors ≥13 subunits, including B1 and B2 (B1B2), to regulate acid–base balance. Intercalated cells can be further divided into α-intercalated cells and β-intercalated cells, marked by AE1 and Pendrin, respectively.

After extensively validating the fidelity of a constitutive Aqp2Cre bacterial artificial chromosome transgene^17,18 and a tamoxifen-inducible Aqp2^ECE knock-in allele¹⁹ in recapitulating the activation of the endogenous Aqp2 promoter, we performed in vivo lineage tracing using these two Cre drivers and identified Aqp2⁺ progenitor cells (AP). During development, embryonic AP become detectable first at E15.5 and differentiate in a progressive manner into five or more cell types to create molecularly distinct DCT2, CNT1, CNT2, and collecting duct segments.²⁰ This differentiation process requires either loss of Aqp2 or B1B2 and gain of NCC (a well-established marker for DCT1 and DCT2) at E15.5, CAII (another marker for intercalated cells) at E16.5, or AE1 (an α-intercalated cell marker) and Pendrin (a β-intercalated cell marker) at P1.²⁰ AP-derived segments are characterized by NCC⁺ Aqp2⁻ RFP⁺ (DCT2), NCC⁻ Aqp2⁻ RFP⁺ (CNT1), and NCC⁻ Aqp2⁺ RFP⁺ (CNT2 and collecting duct), respectively, in Aqp2Cre RFP/+ kidneys. RFP⁺ indicates that these segments are derived from AP, although DCT2 and CNT1 have lost expression of Aqp2 during development.²⁰

Neonate and adult AP generate daughter cells either preserving the property of the AP (self-renewal) or evolving into DCT2/connecting tubules/collecting duct cells (multipotentiality), forming single-cell–derived multiple-cell clones (clonogenicity) during development, tissue maintenance, and/or injury repair.²¹ Hence, adult AP are identified as the first potential candidate of adult renal progenitor cells that satisfy the strictest definition of requiring the in vivo demonstration of (1) self-renewal, (2) clonogenicity, (3) multipotency, (4) participation in the tissue maintenance, and (5) injury repair.^20–22 While not directly tested, embryonic AP are believed to possess these capacities as well. Nevertheless, the role of AP in PKD pathogenesis remains unaddressed.

Here, we report that inactivation of Pkd2 in embryonic AP, but doubtfully in neonate or adult AP, results in severe PKD with progressive loss of cystic α-intercalated cells, recapitulating a newly identified prominent cellular phenotype of patients with ADPKD. Supplemental Table 1 lists all abbreviations.

Methods

Reagents

The primary antibodies used in immunofluorescence (IF) studies consisted of mouse anti–V-ATPase B1 and B2 (Santa Cruz, sc-55544), mouse anti-Aqp2 (Santa Cruz, sc-515770), mouse antipendrin (Santa Cruz, sc-518830), rat anti-AE1 (Alpha Diagnostic, AE11-A), rabbit anti-RFP (Clontech, 632496), rabbit anti-Aqp2 (Santa Cruz, sc-28629), rabbit anti-Aqp1 (Millipore, AB2219), rabbit anti-NCC (Millipore, AB3553), rabbit anti-V-ATPase B1 and B2 (Santa Cruz, sc-20943), rabbit anti-Pkd2 (Proteintech, 19126-1-AP), rabbit anti-Ki67 (Abcam, AB15580), mouse anti-proliferating cell nuclear antigen (PCNA) (Abcam, AB29), normal rabbit IgG (Santa Cruz, sc-2027), goat anti-Aqp2 (Santa Cruz, sc-9882), goat antiuromodulin (Santa Cruz, sc-19554), goat anticalbindin (Santa Cruz, sc-7691), goat antimegalin (Santa Cruz, sc-16478), and goat antipendrin (Santa Cruz, sc-16894). The secondary antibodies used were Invitrogen Alexa Fluor 568 donkey anti-rabbit IgG (A10042), Alexa Fluor 647 donkey anti-goat IgG (A21447), Alexa Fluor 647 donkey anti-goat IgG (A21447), and Jackson Immunoresearch Alexa Fluor 594 donkey anti-goat IgG (705-585-147), Alexa Fluor 488 donkey anti-mouse IgG (715-545-150), Alexa Fluor 647 donkey anti-mouse IgG (715-605-151), and 488-Fab fragment donkey anti-rabbit IgG (711-547-003).

Generation, Genotyping, and Induction of Aqp2Cre Pkd2^f/f and Aqp2^ECE/+ Pkd2^f/f Mice

Aqp2Cre^17,18 and Aqp2^ECE/+19–21 mice were bred with Pkd2^f/f (Jackson Laboratory stock 017292) to generate Aqp2Cre Pkd2^f/f (Pkd2^AC) and Aqp2^ECE/+ Pkd2^f/f (Pkd2^ECE) mice, respectively. PCR-based genotyping was conducted to identify the genotypes of Aqp2Cre²³ and Aqp2^ECE/+19 mice, as previously described. Pkd2^f/f mice were genotyped according to the manufacturer's recommendations. All mice were generated in a highly pure C57BL/6 background. Mice of both sexes were used with free access to water and a normal diet unless otherwise indicated. Aqp2Cre PKD2^f/f and their Aqp2Cre PKD2^f/+ littermates were sacrificed at P1, P6, P12, or P17. Aqp2^ECE/+ PKD2^f/f and their Aqp2^ECE/+ PKD2^f/+ littermates were induced with tamoxifen diet (Envigo, TD. 130857, 500 mg/kg) at P1 for 48 hours or a singular intraperitoneal injection of tamoxifen (1×2 mg, Sigma, T5648) at P60. All induced mice were sacrificed at P300.

Established PKD mouse models cpk, Pkd1^RC/RC,²⁴ and Thm1CKO²⁵ were used. The cpk model contains a homozygous mutation in cys1, which encodes the cilia-associated cystin. Pkd1^RC/RC mice carry a knock-in of a naturally occurring disease variant, PKD1p.R3277C (RC). Cysts appear as early as P6 and macroscopically visible thereafter in Pkd1^RC/RC mice. The Thm1CKO is an inducible ciliary PKD model that disrupts ciliary protein trafficking and cilia structure and manifests renal cystic disease at P42. Cpk, Pkd1^RC/RC, and Thm1CKO mutant mice were housed in microisolator cages on a high-efficiency particulate air-filtered ventilated rack under pathogen-free conditions. Cpk at P15 and Pkd1^RC/RC at the age of 5 months were used for this study. Gene inactivation in Thm1CKO mice carrying the ROSA26-CreERT recombinase was induced by intraperitoneal injections of tamoxifen (9 mg/40 g) into nursing dams at P0 and P3. The resulting progeny were euthanized at P42, and kidneys were harvested. We used littermates that were either negative for Cre or did not have two mutant Thm1 alleles as controls.

Human Tissues

Human ADPKD biomaterials were obtained at the time of nephrectomy from the Surgery Department at University of Kansas Medical Center or hospitals participating in the Tissue Donation Program at the PKD Foundation (Kansas City, MO). The average age of patients with ADPKD who have donated their kidneys for research was approximately 53 years (29–73 years) and were at or near ESKD. More specifically, the nine patients with ADPKD who were subject to cell counting analyses had an average age of 55 years (48–62 years). It is reasonable to assume that most, if not all, of the tissue samples were from PKD1 patients since approximately 80% of clinical cases are caused by PKD1 mutations, and individuals with PKD2 mutations have a milder phenotype with a later onset of ESKD (54 years for PKD1 versus 74 years for PKD2 mutations). Normal human kidney sections were obtained from cadavers whose kidneys were unsuitable for transplantation either because of abnormalities in vasculature or poor perfusion characteristics. The average age for these individuals was 44 years.

IF Studies

Mouse kidneys were perfused with PBS to remove blood cells and fixed in 4% paraformaldehyde for 48 hours at 4°C, embedded in paraffin and cut into 5 μm sections. Kidney sections were deparaffinized in three changes of xylene for 5 minutes each and rehydrated with 100%, 95%, 75%, and 50% ethanol washes, followed by boiling in antigen retrieval buffer (0.01 M sodium citrate, pH 6.0). Next, the sections were blocked with 5% BSA/2% normal donkey serum/0.5% Triton X-100 in PBS, followed by primary antibody staining at 4°C overnight. Primary antibodies were diluted in 5% BSA in PBS at 1:100–200. On the next day, the slides were washed in PBS for 5×5 minutes, incubated with the corresponding combination of secondary antibodies (1:200–1000), and mounted using Prolong Gold antifade mounting media (Invitrogen). Costaining of Aqp2/AE1/Ki67 was performed according to our previously published sequential staining protocol.^20,21

All slides were examined under an epifluorescence microscope (Olympus IX71) or confocal microscope (Zeiss LSM 880 NLO confocal microscope with Airyscan).^26,27 Entire sections were scanned with either 20× or 40× lens. Each marker was assigned an arbitrary color using Adobe Photoshop CS4 for consistency between experiments. Three or more mice per group were examined. Cell counting was conducted using Image J software and restricted to connecting tubules/collecting duct tubules or cysts, which were identified by the presence of at least one Aqp2⁺ cell. In each case, >1000 connecting tubules/collecting duct cells were counted per mouse. 4',6-diamidino-2-phenylindole was used to aid the counting. Total cell counts from each mouse or human kidney were presented.

Histologic Analyses

Paraffin-embedded kidney sections were subjected to hematoxylin and eosin staining according to standard protocol. Stained sections were scanned using NanoZoomer (Hamamatsu, Bridgewater, NJ). Slides from Pkd2^AC and Pkd2^ECE were blindly examined by a pathologist (A.R. Lightle) at Albany Medical Center.

Statistical Analyses

Unpaired t test was used for comparisons between two means unless otherwise specified. Multiple comparisons between three or more means were assessed via one-way ANOVA followed by Tukey's multiple comparison tests. Statistical significance was set at P < 0.05.

Results

Pkd2 Is Apparently Expressed in All Segments from Proximal Tubule to Collecting Ducts

To profile the expression of Pkd2 throughout the kidney, coexpression of Pkd2 with distinct segment-specific markers was assessed by double IF staining. Pkd2 was detected in tubules labeled by Aqp1, megalin, uromodulin, calbindin, or NCC (Supplemental Figures 1–5). Replacing the Pkd2 antibody with normal rabbit IgG yielded no signal in wild-type (WT) mice (Supplemental Figure 6, A–C). In Pkd2^AC mice, Pkd2 was barely detectable in Aqp2⁺ cells and in some Aqp2⁻ segments, which could be DCT2 and CNT1 cells (Supplemental Figure 6D). These DCT2 and CNT1 cells were presumably derived from AP, which expressed Aqp2 and Cre to inactivate Pkd2. Hence, the loss of Pkd2 in these cells was anticipated, although they had lost Aqp2 during the development. Taken together, our data (1) validate the specificity of the Pkd2 antibody, (2) imply that Pkd2 is eliminated equally in the tubule segments derived from AP, and (3) suggest that Pkd2 is likely synthesized in epithelial cells from the proximal tubule to collecting duct throughout the kidney. It should be noted that collecting duct, partially marked by calbindin, expressed higher Pkd2 than other segments.

Pkd2 Is Expressed in Embryonic, Neonate, and Adult AP and Their Derivatives throughout the Development

Our recent studies showed that embryonic, neonate, and adult cells expressing Aqp2 and V-ATPase subunits B1, B2, or both (Aqp2⁺ B1B2⁺ cells) are potential AP.^20,21 To determine whether Pkd2 is produced in AP and their derivatives, we performed Aqp2/B1B2/Pkd2 triple IF first in E15.5 kidneys. Robust Pkd2 expression was found in principal cells (Aqp2⁺ B1B2⁻), intercalated cells (Aqp2⁻ B1B2⁺), AP (Aqp2⁺ B1B2⁺), and transitional cells (TC1, Aqp2⁻ B1B2⁻) and detected in the cytoplasm, membrane, and cilia (Supplemental Figure 7A). The expression of Pkd2 in these cell types apparently continues throughout the developmental stages. This is evidenced by similar analyses of kidneys from postnatal day 1 (P1) and adult kidneys (P60) (Supplemental Figure 7, B and C).

Ablation of Pkd2 in the Embryonic AP Results in the Development of Severe PKD

The widely distributed Pkd2⁺ cells along the renal segments from the proximal tubules to the collecting ducts prompted us to determine whether Pkd2 deficiency in the embryonic AP is sufficient to cause PKD. To this end, we generated Pkd2^f/f Aqp2Cre mice to inactivate Pkd2 in the embryonic AP and thus their derivatives. Aqp2Cre faithfully recapitulates the activation of the endogenous Aqp2 promoter and effectively drives Cre-mediated recombination as early as E15.5, the earliest time point when Aqp2 mRNA becomes detectable, and when AP emerge.²⁰ For simplicity, Pkd2^f/f Aqp2Cre mice are referred as Pkd2^AC hereafter. Pkd2^+/+ Aqp2Cre, Pkd2^+/f Aqp2Cre, and Pkd2^f/f mice were used as WT controls. While WT mice were normal, the Pkd2^AC mice developed severe cystic kidneys with very little renal parenchyma and died around postnatal day 17 (P17) (Figure 1, A–C). Pkd2^AC mice showed significantly reduced body weight and increased kidney weight and kidney weight/body weight ratio (Figure 1, D–F). Cysts were absent at P1 and became obvious at P6 and rapidly enlarged (Supplemental Figure 8). At P6, cortical cysts consisted of tubules and glomerular capsules dilated more than 2× the normal diameter, measuring up to 229 µm, and accounted for approximately 5% area of the kidney (Supplemental Figure 8). Diffuse tubular epithelial attenuation indicative of acute tubular injury was also noted in one kidney sample. At P12 and P17, cortical cysts enlarged up to 1004 and 1910 µm and occupied approximately 70% and approximately 90% space of the kidneys, respectively (Supplemental Figure 8). Hence, Pkd2 deficiency in embryonic AP is sufficient to induce cyst formation.

Figure 1.

Pkd2 deficiency in embryonic AP leads to severe PKD. (A) Breeding scheme. (B) Images of mice at P17 showing kidney enlargement and the reduction in body size of Pkd2^AC mice versus WT. (C) H&E images of WT and Pkd2^AC kidneys at P17 demonstrating profound cyst formation in Pkd2^AC mice. Magnified images are shown in Supplemental Figure 8. Scale bar: 1 mm. (D–F) BW, KW, and KW/BW ratio of WT and Pkd2^AC mice at P17. n=8. *P < 0.05. AP, Aqp2⁺ progenitor cells; BW, body weight; H&E, hematoxylin and eosin; KW, kidney weight; PKD, polycystic kidney disease; WT, wild-type.

The Cysts in Pkd2^AC Mice Extend from the Distal Renal Segments to the Adjacent DCT1

Because AP give rise to most, if not all, epithelial cells in the DCT2/connecting tubules/collecting duct, these cells are presumably deficient in Pkd2 and responsible for the initiation of cyst formation. To verify this and to determine whether cysts expanded into the adjacent segments, we examined the expression of various segment-specific markers in the cells lining the cysts. Kidneys from Pkd2^AC mice at P17 were subject to double IF in which Aqp2 was combined with one of the other segment-specific markers. Cysts were defined as dilated tubules whose width was ≥2× larger than normal tubules. Cystic cells positive for calbindin, NCC (a DCT marker), parvalbumin (a DCT1 marker), or uromodulin were observed. By contrast, cystic Aqp1⁺ and megalin⁺ cells were not noted (Supplemental Figure 9). These data suggest that cysts induced by deletion of Pkd2 in embryonic AP, which generate DCT2/connecting tubules/collecting duct, proximately extended to the adjacent DCT1 and possibly to the loop of Henle, but not to the proximal tubule.

Pkd2^AC Cysts Progressively Increase TCs Expressing Neither Principal Cell nor Intercalated Cells Markers Primarily in the Expense of Intercalated Cells and Completely Lost α-Intercalated Cells by P12

To determine whether changes in the relative abundance of various cell types occur in the cysts, we conducted Aqp2/B1B2, Aqp2/Pendrin, and Aqp2/AE1 double IF in both WT and Pkd2^AC mice at three stages: P6, P12, and P17. For each IF (n=3 mice/genotype/stage), we counted 1623–3125 cells in Aqp2⁺ tubules in WT or in Aqp2⁺ cysts in Pkd2^AC mice at each stage, categorized the counted cells into four groups on the basis of the expression of the markers examined (Supplemental Tables 2 and 3) and compared the percentages of each cell type between the two genotypes at each stage (see Methods). For Aqp2/B1B2 staining, the differences in the percentages of principal cells, intercalated cells, AP, and TC1 between WT and Pkd2^AC mice were significant at all three stages, with three exceptions. Pkd2^AC versus WT mice had comparable percentages of principal cells at both P6 and P12 and AP at P6 (Supplemental Figure 10). Specifically, the percentage of intercalated cells reduced from 27.4% (543/1983) in WT to 16.4% (464/2828) in P6 Pkd2^AC cysts. The reduction in the intercalated cells percentage was more pronounced at P12 (6.8%, 158/2334 versus 30.5%, 541/1775) and P17 (3.2%, 102/3125 versus 27.9%, 713/2552). Like intercalated cells, AP also significantly accounted for a lower portion in the cysts than in WT at both P12 and P17. By contrast, the abundance of TC1 was significantly higher in Pkd2^AC versus WT mice at each stage. The increase in cystic TC1 percentage became more dramatic as the disease progressed (Supplemental Figure 10).

For Aqp2/Pendrin staining, the four groups were referred as principal cells (Aqp2⁺ Pendrin⁻), β-intercalated cells (Aqp2⁻ Pendrin⁺), TC2 (Aqp2⁺ Pendrin⁺), and TC3 (Aqp2⁻ Pendrin⁻). The changes in their percentages in Pkd2^AC versus WT displayed a similar pattern, as presented in Supplemental Figure 10. In brief, significant changes in the abundance of principal cells between the two genotypes over the time course were not found. However, the percentages of β-intercalated cells and TC2 decreased and the percentage of TC3 increased significantly in Pkd2^AC versus WT at each time point, except β-intercalated cells at P6 (Supplemental Figure 11).

More dramatic cellular changes were revealed by Aqp2/AE1 staining. At P6, while TC4 (Aqp2⁺ AE1⁺) were significantly less abundant in Pkd2^AC versus WT mice, the percentages of principal cells (Aqp2⁺ AE1⁻), α-intercalated cells (Aqp2⁻ AE1⁺), and TC5 (Aqp2⁻ AE1⁻) were comparable (Figure 2, A and B). At P12, the percentages of TC5 were significantly increased in the expense of both α-intercalated cells and TC4 (Figure 2, C and D). The percentages of α-intercalated cells decreased from 14.2% (310/2185) in WT to 0% (0/2480) in Pkd2^AC cysts. At P17, out of 1623 cells counted in WT, principal cells, α-intercalated cells, TC4, and TC5 accounted for 67.2%, 12.8%, 1.8%, and 18.2%, respectively. These numbers were significantly altered to 59.2%, 0%, 0%, and 40.2% when 2654 cells were analyzed in Pkd2^AC cysts (Figure 2, E and F). These data collectively suggest that Pkd2^AC cysts progressively increased double negative TCs (TC1, TC3, and TC5) primarily in the expense of intercalated cells and completely lost α-intercalated cells by P12.

Figure 2.

Pkd2^AC mice progressively lose α-IC cells. (A–F) IF images of Aqp2 (red), AE1 (green), and DAPI (blue) and their corresponding cell counting, as detailed below. (A) IF images of Aqp2 and AE1 showing the presence of α-IC at P6 in both normal tubules of WT and cystically dilated tubules of Pkd2^AC mice. (B) Percent of Aqp2⁺ AE1⁻ (PC), Aqp2⁻ AE1⁺ (α-IC), TCs Aqp2⁺ AE1⁺ (TC4), and TCs Aqp2⁻ AE1⁻ (TC5) in total CNT/CD at P6 in WT and Pkd2^AC mice. (C and E) IF images of WT and Pkd2^AC at P12 and P17, respectively, demonstrating the absence of AE1 in cystic tubules of Pkd2^AC. (D and F) Percent of PC, α-IC, TC4, and TC5 in WT and Pkd2^AC mice at P12 and P17, respectively. In each case, 1623–3125 CNT/CD cells were counted (n=3 mice/group/genotype), as presented in Supplemental Table 2. The actual numerical percentages for B, D, and F are listed in Supplemental Table 3. CNT/CDs were recognized by the presence of ≥1 Aqp2⁺ cells lining a normal tubule in WT or cystically dilated tubules in the Pkd2^AC mice. Arrow-pointed cells are amplified 3× in the insert. Scale bar: 50 µm. *P < 0.05 versus WT. CD, collecting ducts; CNT, connecting tubules; DAPI, 4',6-diamidino-2-phenylindole; IC, intercalated cells; IF, immunofluorescence; PC, principal cells; TC, transitional cell.

Cystic TCs Are Apparently More Proliferative than Other Cell Types in PKd2^AC at P6

Ki67 and PCNA are expressed from late G1 to mitosis and from late G1 to early G2, respectively,²⁸ and controversially used as proliferation markers.^20,21,29,30 To probe if the dynamic changes in the cysts is due to faster growth of some cell types compared with others, we did Aqp2/X/Ki67 and Aqp2/X/PCNA (X=B1B2, Pendrin or AE1) staining in Pkd2^AC and WT mice at P6 (Figure 3). The percentage of cells expressing Ki67 or PCNA in each cell type was estimated after approximately 1000 cells/genotype/marker were counted in the WT and Pkd2^AC cysts. For X=B1B2 shown in Figure 3, the percentages of Ki67⁺ and PCNA⁺ cells in each cell type were significantly increased in Pkd2^AC versus WT, with only three exceptions. Moreover, Ki67⁺TC1 and PCNA⁺TC1 in Pkd2^AC cysts versus WT showed the biggest increases from 6% to 47% and from 0% to 60%, respectively. By contrast, changes in the percentages of Ki67⁺ intercalated cells and PCNA⁺ AP were not significant. The percentage of Ki67⁺ AP was even decreased in PKd2^AC versus WT (Figure 3). Similar results were obtained when replacing B1B2 with either Pendrin or AE1 (Supplemental Figures 12 and 13). Ki67⁺ AE1 and PCNA⁺ AE1 cells were rarely observed in both WT tubules and Pkd2^AC cysts. On the basis of the expression of Ki67 and PCNA, the cystic double negative TCs (TC1, TC3, and TC5) seem to be more proliferative than other cell types, which may contribute to the dynamic cellular changes in the cysts.

Figure 3.

Expression of Ki67 and PCNA in cystic cells in Pkd2^AC mice at P6. (A and B) IF images of Aqp2 (blue), B1B2 (red), Ki67 (A) or PCNA (B) (green), and DAPI (white) in normal tubules of P6 WT and cystic tubules of P6 Pkd2^AC mice. Scale bar: 50 µm. (C and D) Percent of Ki67⁺ (C) and PCNA⁺ (D) PC, IC, AP, and TC1 in P6 WT and Pkd2^AC mice. N=3. *P = 0.05 versus WT. PCNA, proliferating cell nuclear antigen.

Cystic α-Intercalated Cells Were Detected in Other PKD Mouse Models

To determine whether the complete loss of cystic α-intercalated cells is a common phenotype, four other PKD mouse models and their respective WT controls were similarly analyzed. Both cpk and Six2creFrs2αKO models develop rapidly progressing PKD and die by P21.^31,32 In Pkd1^RC/RC mice, cysts are present as early as P6 and macroscopically visible thereafter.²⁴ The Thm1CKO is an inducible ciliary PKD model.²⁵ Cyst-lining α-intercalated cells (Aqp2⁻AE1⁺) were equally observed in each of these PKD models and in their respective controls (Supplemental Figure 14), indicating that these PKD models did not share the complete loss of cystic α-intercalated cells with Pkd2^AC mice and patients with ADPKD described below.

Patients with ADPKD Have a Similar Cellular Phenotype to Pkd2^AC Mice

To investigate whether the cellular phenotype of Pkd2^AC mice is reflective of human ADPKD patients, AQP2/B1B2, AQP2/PENDRIN, and AQP2/AE1 IF were done on kidney sections of 27 patients with ADPKD and nine normal controls. While these markers were detectable in all controls, cyst-lining AQP2⁺ cells were found only in 13 ADPKD samples. Cell counting was then focused on seven normal and nine ADPKD samples that were randomly selected from the 13 patients. For each IF, 3898–10146 cells were assessed (Supplemental Table 4). Compared with normal controls, patients with ADPKD had significantly lower percentages of principal cells, intercalated cells, α-intercalated cells, β-intercalated cells, AP, and TC4 and significantly higher percentages of TCs (TC1, TC3, and TC5). More strikingly, there were no AE1⁺ cells lining cysts (Figure 4). Our data strongly suggest that like in Pkd2^AC mice, α-intercalated cells were eliminated in ADPKD cysts as well.

Figure 4.

Patients with ADPKD have reduced IC and β-IC and completely lack α-IC. (A) IF images of AQP2 (red), B1B2 (green), and DAPI (blue) showing the reduction of IC in patients with ADPKD versus normal controls. (B) Percent of AQP2⁺ B1B2⁻ (PC), AQP2⁻ B1B2⁺ (IC), AQP2⁺ B1B2⁺ (AP), and TCs AQP2⁻ B1B2⁻ (TC1) in normal controls and patients with ADPKD. (C) IF images of AQP2 (red) and PENDRIN (green) showing the reduction of AQP2⁻ PENDRIN⁺ cells (β-IC) in patients with ADPKD versus normal controls. (D) Percent of AQP2⁺ PENDRIN⁻ (PC), AQP2⁻ PENDRIN⁺ (β-IC), TCs AQP2⁺ PENDRIN⁺ (TC2), and TCs AQP2⁻ PENDRIN⁻ (TC3) in normal controls and patients with ADPKD. (E) IF images of AQP2 (red) and AE1 (green) showing the presence of AQP2⁻ AE1⁺ cells (α-IC) in normal controls, but not in patients with ADPKD. (F) Percent of AQP2⁺ AE1⁻ (PC), AQP2⁻ AE1⁺ (α-IC), TCs AQP2⁺ AE1⁺ (TC4), and TCs AQP2⁻ AE1⁻ (TC5) in normal controls and patients with ADPKD. CNT/CD were identified by the existence of ≥1 AQP2⁺ cells lining a normal tubule in normal controls or cysts in patients with ADPKD. Rare apparently normal CNT/CD in patients with ADPKD were excluded from cell counting. 3896–10,146 CNT/CD cells were counted in normal individuals (n=7) and patients with ADPKD (n=9), as presented in Supplemental Table 4. Arrow-pointed cells are amplified 3x in the insert. Scale bar: 50 µm. *P < 0.05 versus normal control. ADPKD, autosomal dominant polycystic kidney disease.

The Percentage of Intercalated Cells Is Negatively Correlated with PKD2 Expression in ADPKD Cysts

To establish the relationship of the changes in the cellular composition with PKD2, AQP2/B1B2/PKD2 staining was conducted with four normal controls and the nine ADPKD samples, as assessed in Figure 4. The percentages of each cell type in the cysts and the percentage of cells expressing PKD2 in each cell type were estimated after 2762 and 8136 cells were counted in the normal and ADPKD kidneys (Supplemental Table 4). PKD2 was expressed in almost all cells in AQP2⁺ tubules in normal samples (Figure 5A). By contrast, PKD2 expression in ADPKD samples was mosaic. In patients with ADPKD, the percentage of intercalated cells in cysts positively correlated with the percentage of PKD2⁺ intercalated cells in cystic intercalated cells, while TC1 showed an opposite pattern. These correlations were significant. Such correlation was not significant for principal cells and AP (Figure 5, B–E).

Figure 5.

The percentage of cystic IC correlates with PKD2 expression in patients with ADPKD. (A) IF images of AQP2 (blue), B1B2 (red), and PKD2 (green) merged with DAPI (white) demonstrating PKD2 expression in most, if not all, CNT/CD and other epithelial cells in normal controls and in rare cells lining AQP2⁺ cysts in patients with ADPKD. Scale bar: 50 µm. (B–E) Correlation analyses between the percentages of each cell type in the cysts and the percentages of each cell type expressing PKD2 in each cell type as shown. A total of 8136 cells lining cysts harboring ≥1 AQP2⁺ cells in patients with ADPKD (n=9) were analyzed, as presented in Supplemental Table 4. Two thousand seven hundred and sixty-two CNT/CD cells were counted in normal controls (n=4) and excluded from the correlation analyses.

Effective Deletion of Pkd2 in Neonate and Adult AP and Their Derivatives Results in Mild PKD

Aqp2^ECE/+ mice harbor a knock-in ER^T2CreER^T2 (ECE) cassette at the position of the Aqp2 initiation codon to encode ECE.¹⁹ To determine whether the disruption of Pkd2 in neonate and adult Aqp2⁺ cells, which include principal cells and AP, can lead to PKD, Aqp2^ECE/+ Pkd2^f/f (Pkd2^ECE) and WT mice were induced by providing their mothers with the tamoxifen-containing diet from P1 to P2 (P1 induction) or with 1×2 mg tamoxifen at P60 (P60 induction) and sacrificed at P300, as we reported. These induction regimens were expected to ablate Pkd2 specifically and efficiently in principal cells and AP. Pkd2 disruption would also occur in intercalated cells, including α-intercalated cells and β-intercalated cells derived from the induced AP. To quantify the extent of Pkd2 ablation, Aqp2/B1B2/Pkd2 staining was conducted. The percentages of Pkd2⁺ cells in each cell type were computed separately in the cortex and outer and inner medulla. Analyses of 1466–3300 cells from six mice/group revealed that the percentages of Pkd2⁺ cells in each cell type throughout the kidney were significantly lower in Pkd2^ECE versus WT mice after either P1 or P60 induction, with one exception. The percentages of Pkd2⁺ intercalated cells in total intercalated cells were comparable between Pkd2^ECE and WT induced at P60 (Supplemental Figures 15–17 and Supplemental Table 5). The recombination efficiency ([%Pkd2⁺ X in WT− %Pkd2⁺ X in Pkd2^ECE]/%Pkd2⁺ X in WT, where X=principal cells, intercalated cells, AP or TC1) ranged from approximately 47% to approximately 81% for principal cells, AP, and TC1. Intercalated cells had lower recombination rates varying from 2% to 34% (Supplemental Table 6). In Pkd2^ECE mice induced at P1, scattered cortical cysts harboring tubules and glomerular capsules invaded approximately 5% of the area. The diameters of the largest cysts were only approximately 250 µm (Supplemental Figure 16). This phenotype became even milder in the Pkd2^ECE mice induced at P60. In these animals, sparse cortical cysts contained tubules with diameters enlarged up to 88 µm. The cystic area was <4% (Supplemental Figure 17). These results showed that effective deletion of Pkd2 in neonate and adult AP and principal cells only triggers the development of mild PKD.

Discussion

In this study, we demonstrate in mice and human ADPKD patients that (1) Pkd2 and its human counterpart is widely expressed in the kidneys; (2) specific disruption of Pkd2 in embryonic, but unlikely in neonate and adult Aqp2⁺ cells, causes severe PKD; (3) both Pkd2^AC and human ADPKD patients share the same cellular phenotype: increased various cystic TCs that express neither principal cells nor intercalated cell markers examined primarily in the expense of intercalated cells; (4) α-intercalated cells are completely lost as the disease progresses; (5) cystic α-intercalated cells exist in the other four PKD mouse models examined; and (6) the percentages of cystic intercalated cells and TC1 positively and negatively correlate with PKD2 expression in patients with ADPKD, respectively. Hence, mice with Pkd2 ablation in embryonic AP represent a new and unique ADPKD model that recapitulates the loss of α-intercalated cells in humans.

In mice, either global or organ-specific deletion of Pkd2 similarly causes cystic diseases of the kidney, liver, and pancreas.^33–38 However, our utilization of Aqp2Cre driver together with a Pkd2 conditional allele presents the first demonstration of lineage-specific deletion of Pkd2. The use of a different Aqp2Cre has been previously described to inactivate Pkd1.³⁴ Typically, Pkd1 versus Pkd2 loss results in more severe PKD; however, the kidney phenotype ³⁴ is milder compared with the phenotype described here. The Aqp2Cre mice ³⁴ contain a transgene with 11 kb of the mouse Aqp2 promoter driving Cre expression. It has been used to disrupt several genes.^34,39–41 Whether the transgene drives Cre expression exclusively in Aqp2⁺ cells remain unknown. The recombination rate has not been clearly documented. Our Aqp2Cre was generated by modifying a mouse P1-derived artificial chromosome to harbor a codon-improved Cre flanked by 125-kb 5′ upstream region and a 31-kb 3′ downstream region of the mouse Aqp2.¹⁷ Its high fidelity and high recombination rate have been extensively demonstrated.^18–21 Our more severe phenotype may result from a higher recombination rate, fidelity, or both to disrupt Pkd2, compared with what previously described.³⁴ Other novelties of our study include (1) the progressive loss of intercalated cells and eventual elimination of α-intercalated cells in Pkd2^AC cysts, a newly identified cellular phenotype recapitulated in human ADPKD patients, and (2) the unlikeness to disrupt a target gene in a principal cell-specific or intercalated cell-specific manner because of the existence of AP expressing both principal cell and intercalated cell markers. Aqp2^ECE/+ should no longer be considered as principal cell specific.

Multiple potential mechanisms may be responsible for the observed cellular changes. (1) Pkd2 loss could favor double negative TCs (TC1, TC3, and TC5) and restrict intercalated cell differentiation from AP. Consistently, TC1 had high recombination rates (50%–81% for P1 induction and 47%–73% for P60 induction), indicating that most TC1 were derived from the induced AP. By contrast, intercalated cell showed relatively low recombination rates (19%–35% for P1 induction and 2%–8% for P60 induction), suggesting that intercalated cell differentiated from the induced AP accounted for a small percentage in the total intercalated cell population. Correlation analyses further revealed positive and negative correlations of the percentages of intercalated cells and TC1 in the cysts with the percentages of Pkd2⁺ intercalated cells and Pkd2⁺ TC1, respectively. (2) Pkd2 depletion could lead to faster growth of some cell types compared with others in the cysts. Consistently, the increases in the percentages of TC1, TC3, and TC5 expressing Ki67 and PCNA were more pronounced than other cell types (Figure 3 and Supplemental Figures 13 and 14). (3) The Aqp2⁺ cysts examined in both patients with ADPKD and Pkd2^AC mice might contain DCT1 and even the loop of Henle. Inclusion of such cysts would increase the percentages of TC1, TC3, and TC5 in the cost of other cell types, particularly in large cysts. The significantly changed percentage of each cell type (except AQP2⁺ PENDRIN⁺) in patients with ADPKD and in P17 Pkd2^AC mice might be partially contributed by cyst expansion to the adjacent segments. However, such contribution in Pkd2^AC mice seems limited, as evidenced by the high recombination rates of TC1, as discussed above. (4) Cystic α-intercalated cells appeared at P6 and disappeared by P12, correlating loss of α-intercalated cells with the severity of the disease in Pkd2^AC mice. We think that such correlation cannot be simply applied to the other PKD models. Two of four other models showed rapidly progressing PKD and died by P21,^31,32 similar to Pkd2^AC mice. Nevertheless, cystic α-intercalated cells were found in each of them. In brief, these data aggregately indicate that while Pkd2 is apparently not required for the generation and maintenance of AP, Pkd2 is likely critical for a balanced differentiation of AP into, proliferation, and/or maintenance of various cell types, particularly α-intercalated cells, in both mice and humans. Because intercalated cells lack a primary cilium, Pkd2 may regulate intercalated cells abundance independent of its function in the primary cilium.

Despite effective ablation of Pkd2 in the induced Pkd2^ECE mice, the number of Pkd2⁺ cells within the DCT2, connecting tubules, and collecting ducts is presumably higher in Pkd2^ECE versus Pkd2^AC mice. This may contribute to the difference in the PKD severity.

In summary, Pkd2 inactivation in embryonic AP, but unlikely in neonate and adult AP, is adequate to induce severe PKD with progressive loss of α-intercalated cells, recapitulating a newly identified cellular phenotype of patients with ADPKD. The shared cellular phenotype is characterized by increased TCs expressing neither principal cell nor intercalated cell markers mostly in the cost of intercalated cells, particularly α-intercalated cells.

Future studies are required to determine and/or validate (1) if the PKD severity is affected by (a) the number of DCT2, connecting tubules, and collecting duct cells being knocked-out for Pkd2 and (b) the known kidney developmental switch that modulates PKD severity (pre <P12–14=severe PKD versus >P14=mild PKD), rather than a new function of Pkd2 in AP. This could be further tested using Pkd2^f/− Aqp2^ECE/+ mice induced with tamoxifen at various time points. Since one Pkd2 allele is already ablated and the other one is floxed, the phenotype of the induced Pkd2^f/− Aqp2^ECE/+ mice should be more severe than the similarly induced Pkd2^ECE mice. (2) If loss of cystic α-intercalated cells is truly Pkd2 dependent, additional Pkd2 models such as Pkd2WS25 could be examined. (3) If and how Pkd2 ablation in AP leads to differences in principal cell/intercalated cell specification. To address this question, clonal analyses of Pkd2^f/f Brainbow/Brainbow Aqp2^ECE/+ can be conducted as we reported.²¹ (4) If and how the abundance change of intercalated cell contributes to the observed PKD phenotype. (5) What is a functional role of Pkd2 in AP and its relevance to driving cyst initiation/growth. (6) Single cell profiling to reveal the molecular mechanisms by which Pkd2 regulates AP function.

Supplementary Material

jasn-35-398-s001.pdf ^{(29.4MB, pdf)}

Disclosures

D.P. Wallace reports research funding from Calico Labs and roles on the American Journal of Physiology Editorial Board and the Steering Committee for the PKD Research Resource Consortium. All remaining authors have nothing to disclose.

Funding

W. Zhang: National Institute of Diabetes and Digestive and Kidney Diseases (R01DK080236, 1R01DK136554-01) and Capital Region Medical Research Institute (AMC-6579). P.V. Tran: National Institute of Diabetes and Digestive and Kidney Diseases (DK103033). D.P. Wallace: National Institute of Diabetes and Digestive and Kidney Diseases (DK106912).

Author Contributions

Conceptualization: Wenzheng Zhang.

Data curation: Chao Gao, Akaki Tsilosani, Wenzheng Zhang.

Formal analysis: Akaki Tsilosani.

Funding acquisition: Pamela V. Tran, Darren P. Wallace, Wenzheng Zhang.

Investigation: Chao Gao, Andrea R. Lightle, Sana Shehzad, Akaki Tsilosani, Wenzheng Zhang.

Methodology: Enuo Chen, Chao Gao, Akaki Tsilosani, Wenzheng Zhang.

Project administration: Chao Gao, Akaki Tsilosani, Wenzheng Zhang.

Resources: Carlton M. Bates, Enuo Chen, Chao Gao, Madhulika Sharma, Pamela V. Tran, Akaki Tsilosani, Darren P. Wallace, Wenzheng Zhang.

Supervision: Wenzheng Zhang.

Writing – original draft: Wenzheng Zhang.

Writing – review & editing: Carlton M. Bates, Chao Gao, Andrea R. Lightle, Madhulika Sharma, Sana Shehzad, Pamela V. Tran, Akaki Tsilosani, Darren P. Wallace, Wenzheng Zhang.

Data Sharing Statement

All data are included in the manuscript and/or supporting information.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/JSN/E580.

Supplemental Table 1. Abbreviations.

Supplemental Table 2. Cell counting in Pkd2^AC mice.

Supplemental Table 3. The actual numerical percentages for Figure 2, B, D, and F.

Supplemental Table 4. Cell counting in patients with ADPKD.

Supplemental Table 5. Cell counting in Pkd2^ECE mice induced at P1 and P60.

Supplemental Table 6. Recombination efficiency (%) induced at P1 and P60.

Supplemental Figure 1. Coexpression of Pkd2 with Aqp1.

Supplemental Figure 2. Coexpression of Pkd2 with megalin.

Supplemental Figure 3. Coexpression of Pkd2 with uromodulin.

Supplemental Figure 4. Coexpression of Pkd2 with calbindin.

Supplemental Figure 5. Coexpression of Pkd2 with NCC.

Supplemental Figure 6. Validation of the specificity of the rabbit Pkd2 antibody.

Supplemental Figure 7. Pkd2 is expressed in all types of CNT/CD cells during development.

Supplemental Figure 8. Pkd2^AC mice progressively develop severe PKD.

Supplemental Figure 9. Cysts in Pkd2^AC mice extend from DCT2/CNT/CD to DCT1 and possibly even to the thick ascending loop of Henle.

Supplemental Figure 10. Pkd2^AC mice progressively lose PC, IC, and AP.

Supplemental Figure 11. Pkd2^AC mice progressively lose β-IC.

Supplemental Figure 12. Expression of Ki67 and PCNA in cystic cells in Pkd2^AC mice at P6.

Supplemental Figure 13. Expression of Ki67 and PCNA in cystic cells in Pkd2^AC mice at P6.

Supplemental Figure 14. Other PKD mouse models contain cystic α-IC.

Supplemental Figure 15. Pkd2 is effectively inactivated in neonate AP and their derivatives.

Supplemental Figure 16. Inactivation of Pkd2 in neonate AP and their derivatives leads to minor cyst formation.

Supplemental Figure 17. Inactivation of Pkd2 in adult AP leads to minor cyst formation.