Combined Deletion of Vhl and Kif3a Accelerates Renal Cyst Formation

Abstract

A subset of familial and sporadic clear cell renal cell carcinomas (ccRCCs) is believed to develop from cystic precursor lesions. Loss of function of the von Hippel-Lindau tumor suppressor gene (VHL) predisposes renal epithelial cells to loss of the primary cilium in response to specific signals. Because the primary cilium suppresses renal cyst formation, loss of the cilium may be an initiating event in the formation of ccRCC. To test this hypothesis, we analyzed the consequences of inducible renal epithelium–specific deletion of Vhl together with ablation of the primary cilium via deletion of the kinesin family member 3A (Kif3a) gene. We developed a microcomputed tomography–based imaging approach to allow quantitative longitudinal monitoring of cystic burden, revealing that combined loss of Vhl and Kif3a shortened the latency of cyst initiation, increased the number of cysts per kidney, and increased the total cystic burden. In contrast with findings in other cystic models, cysts in Kif3a mutant mice did not display accumulation of hypoxia-inducible factor 1-α (HIF1α), and deletion of both Hif1a and Kif3a did not affect cyst development or progression. Vhl/Kif3a double mutation also increased the frequency of cysts that displayed multilayered epithelial growth, which correlated with an increased frequency of misoriented cystic epithelial cell divisions. These results argue against the involvement of HIF1α in promoting renal cyst growth and suggest that the formation of simple and atypical renal cysts that resemble ccRCC precursor lesions is greatly accelerated by the combined loss of Vhl and the primary cilium.

von Hippel-Lindau (VHL) disease is an autosomal dominant genetic disorder caused by inheritance of a mutant allele of the VHL gene. Patients with VHL disease have an increased risk of developing renal cysts and bilateral, multifocal solid or cystic clear cell renal cell carcinomas (ccRCCs), in which the wild-type VHL allele is invariably lost due to somatic mutation or gene silencing.¹ Kidneys of patients with VHL disease develop a spectrum of VHL-deficient lesions that likely represent precursor lesions of solid and cystic ccRCCs, including small foci of micro-ccRCC with a solid appearance, simple cysts lined by a single layer of proliferating epithelial cells, and atypical cysts containing regions of multilayered epithelial growth resembling foci of ccRCC.² These findings suggest that ccRCC forms via cyst-dependent and cyst-independent pathways in patients with VHL disease.³ The VHL gene is also biallelically inactivated in >90% of sporadic ccRCC cases.⁴ Although most ccRCCs have a solid morphology, approximately 5% are cystic with small nodular aggregates of clear cell tumor cells within and between the walls of the cysts,^5,6 suggesting that sporadic ccRCC also arises via cyst-dependent and cyst-independent pathways.

pVHL mediates numerous biologic activities, including targeting the α-subunits of the hypoxia-inducible transcription factors (HIF1α, HIF2α, and HIF3α) for proteolytic degradation⁷ and stabilizing the microtubule network.⁸ This latter function of pVHL is important for regulation of the primary cilium, a microtubule-based structure that functions as a sensory organelle for numerous chemical and mechanical stimuli.⁹ Genetic mutations that compromise the signaling functions or the structure of the primary cilium cause polycystic kidney diseases.¹⁰ Human ccRCC tumors exhibit extremely low frequencies of ciliated cells¹¹ and loss of function of VHL in immortalized renal epithelial cells or ccRCC cell lines impairs the formation of primary cilia.^12–14 However, deletion of Vhl in renal epithelial cells in the mouse kidney or in cultured primary cells does not cause loss of the primary cilium, but sensitizes cells to lose the primary cilium in response to stimulation of the phosphoinositide 3-kinase (PI3K) signaling pathway and inactivation of glycogen synthase kinase 3β.^15,16VHL mutant cystic epithelial cells in kidneys of patients with VHL disease frequently lack a primary cilium and display hyperactivation of the PI3K signaling pathway and elevated levels of phosphorylated, inactive glycogen synthase kinase 3β.^15,16 Loss of the cilium is believed to be a secondary event that occurs in some VHL mutant cells and acts as the trigger for initiation of cyst formation. Combined deletion of Vhl and Pten together in mouse renal epithelial cells, but not of either gene alone, caused renal epithelial cysts with reduced frequencies of ciliated epithelial cells, supporting this hypothesis.¹⁵ However, it was not possible to conclude whether cyst formation results solely from loss of the cilium, from cooperative effects of loss of the cilium plus loss of Vhl, or from cooperative effects of loss of the cilium plus loss of Vhl plus activation of the PI3K pathway. Because Vhl/Pten mice developed diverse benign malignancies in the genital tract that necessitated early euthanasia,¹⁷ it was also not possible to age the mice to determine whether Vhl mutant cystic lesions are able to progress to form ccRCCs. Here we address several of these limitations by generating a mouse model allowing inducible deletion of Vhl together with genetic ablation of primary cilia through deletion of Kif3a, which encodes a protein component of the kinesin-II microtubule motor complex.¹⁸

Loss of pVHL function causes stabilization of HIF1α and HIF2α in epithelial cells of cystic lesions in patients with VHL disease.² Recent studies suggest that HIF1α might play a more general role in cystic lesions that arise in genetic conditions other than VHL disease. HIF1α, but not HIF2α, is stabilized in cystic epithelial cells due to microenvironmental hypoxia in a rat model of polycystic kidney disease and in human polycystic kidney disease tissues.¹⁹ Based on two in vitro models of cyst formation, it was proposed that Hif1α activation promotes cyst formation.²⁰ However, whether HIF1α contributes to the formation or progression of cystic lesions has not been tested in a physiologically relevant cystic model. We tested this idea by deleting Hif1a in the Kif3a-deletion model of kidney cyst formation.

Results

We generated mouse lines allowing inducible renal epithelium–specific deletion of Kif3a (Ksp1.3-Cre^{ERT2 Tg/+};Kif3a^fl/fl), Kif3a/Vhl (Ksp1.3-Cre^{ERT2 Tg/+};Kif3a^fl/fl;Vhl^fl/fl), and Kif3a/Hif1a (Ksp1.3-Cre^{ERT2 Tg/+};Kif3a^fl/fl;Hif1a^fl/fl). Nontransgenic (Ksp1.3-Cre^{ERT2 +/+}) littermates served as controls. Six-week-old mice were fed tamoxifen-containing food for 2 weeks to induce activation of Cre recombinase. Treated animals are hereafter referred to as Kif3a^Δ/Δ, Kif3a^Δ/Δ;Vhl^Δ/Δ, or Kif3a^Δ/Δ;Hif1a^Δ/Δ and controls as Kif3a^fl/fl, Kif3a^fl/fl;Vhl^fl/fl, or Kif3a^fl/fl;Hif1a^fl/fl, respectively. Recombination-specific PCR of genomic DNA isolated from kidneys revealed that all genes were deleted as expected (Figure 1A). Although this method of Cre activation leads to some variation between animals in the extent of gene deletion (Supplemental Figure 1K), the Kif3a locus was recombined at similar frequencies on average in Kif3a^Δ/Δ, Kif3a^Δ/Δ;Vhl^Δ/Δ, and Kif3a^Δ/Δ;Hif1a^Δ/Δ kidneys (Figure 1B). Renal epithelium–specific deletion of Vhl causes nuclear accumulation of HIF1α¹⁵ and Kif3a deletion causes loss of primary cilia as assessed by acetylated tubulin staining,¹⁸ providing two readouts that allow identification of gene-deleted cells. Epithelial cells lining cysts in Kif3a^Δ/Δ, Kif3a^Δ/Δ;Vhl^Δ/Δ, or Kif3a^Δ/Δ;Hif1a^Δ/Δ mice invariably lacked primary cilia (Figure 1C). HIF1α nuclear staining was observed in all cystic epithelial cells in Kif3a^Δ/Δ;Vhl^Δ/Δ but not in Kif3a^Δ/Δ mutant mice (Figure 1C) verifying deletion of Vhl and showing that cysts in Kif3a^Δ/Δ mice are not hypoxic, in contrast with results obtained in a rat polycystic kidney model.¹⁹ Tubules in Kif3a^Δ/Δ;Vhl^Δ/Δ mutant kidneys in which single or multiple epithelial cells were positive for Hif1α staining (Supplemental Figure 1B) were observed at a frequency approximately 10-fold higher than the number of cysts. Similarly, tubules were frequently seen in which some cells displayed elevated staining for Glut-1, a HIF1α-inducible protein and marker of Vhl deletion, whereas other cells were negative for Glut-1. In these tubules, 78% of Glut-1–negative cells but only 4% of Glut-1–positive cells displayed a primary cilium, demonstrating codeletion of Vhl and Kif3a in these cells (Supplemental Figure 1, H and J). These tubules harboring Vhl/Kif3a mutant cells exhibited no apparent morphologic abnormalities, meaning that cysts were not initiated within 9 months after gene deletion. We conclude that the frequency of gene deletion is not the limiting factor in determining the extent of cyst formation. Staining with markers of different nephron segments revealed that cysts develop from proximal tubules, thick ascending loops of Henle, distal convoluted tubules, collecting ducts, and the urinary pole of the glomerulus (Supplemental Figure 2), consistent with the expression pattern of the Ksp1.3-Cre^ERT2 transgene.²¹

Figure 1: Generation of mice allowing inducible deletion of Kif3a, Kif3a/Vhl, or Kif3a/Hif1a. (A) PCR of genomic DNA from kidneys of Kif3a^{Δ / Δ} and Kif3a^fl/fl (top), Kif3a^{Δ / Δ};Vhl^{Δ / Δ} and Kif3a^fl/fl;Vhl^fl/fl (middle), and Kif3a^{Δ / Δ}; Hif1a^{Δ / Δ} and Kif3a^fl/fl;Hif1a^fl/fl (bottom) mice. PCR reactions using primers specific for the floxed (fl) and recombined (Δ) alleles were performed and the positions of the expected products are shown. The promoter region of ribosomal protein S6 (S6) serves as the internal control. (B) Quantitative real-time PCR analyses of the ratio of the Kif3a recombined and floxed alleles in kidneys from the indicated genotypes. Mean±SD (n=3–7). (C) Representative H&E, anti-HIF1α, and anti–acetylated tubulin staining of histologically normal and cystic (Cy) regions of Kif3a^{Δ / Δ}, Kif3a^{Δ / Δ};Vhl^{Δ / Δ}, and Kif3a^{Δ / Δ}; Hif1a^{Δ / Δ} kidneys. H&E, hematoxylin and eosin. Bar, 20 µM.

To quantitatively monitor cyst formation and progression, we established a microcomputed tomography (µCT)–based imaging technique involving intravascular injection of a contrast agent that is concentrated via the renal tubular system, allowing visualization of the kidney structure.²² Cysts either do not receive the contrast agent or fail to concentrate it and therefore appear as dark regions in µCT imaging. Mice were imaged at 8, 16, 20, 24, 28, 32, and 36 weeks after administration of tamoxifen food (Figure 2). Kidneys were isolated and blood plasma was taken after the last time point. µCT imaging revealed no cyst formation in control animals but cyst development in all mutant genotypes. Comparison of µCT images with histologic sections of the same kidneys showed that the µCT imaging approach accurately reflects the cystic burden (Figure 2). Kif3a^Δ/Δ and Kif3a^Δ/Δ;Hif1a^Δ/Δ animals exhibited a moderate cystic phenotype characterized by the development of 2–20 small to medium cysts per kidney. Although some Kif3a^Δ/Δ;Vhl^Δ/Δ mice exhibited an apparent moderate cystic phenotype, many developed a severe phenotype and displayed many more larger cysts. Total cystic burden was quantified by three-dimensional reconstruction of the volumes of all of the cystic lesions in the entire kidney (Figure 3A). Compared with Kif3a^Δ/Δ and Kif3a^Δ/Δ;Hif1a^Δ/Δ mice, Kif3a^Δ/Δ;Vhl^Δ/Δ mice developed cysts earlier and with higher penetrance at all time points (Figure 3B) and showed a dramatically increased total cystic burden (Figure 3C). Histologic analyses revealed that Kif3a^Δ/Δ;Vhl^Δ/Δ kidneys exhibited a 4-fold increase in the number of cysts per section compared with Kif3a^Δ/Δ and Kif3a^Δ/Δ;Hif1a^Δ/Δ kidneys (Figure 3D). Total kidney volumes determined using µCT correlated excellently with kidney weights (Supplemental Figure 3, A–C) further validating the accuracy of the µCT approach. Creatinine (Supplemental Figure 3D) and urea (Supplemental Figure 3E) in blood plasma showed nonsignificant trends toward increased levels in Kif3a^Δ/Δ;Vhl^Δ/Δ mice. Because the values for all control animals were in the normal ranges, performing monthly rounds of µCT imaging over the life of the animal does not compromise kidney function. Moreover, Kif3a^Δ/Δ;Hif1a^Δ/Δ and Kif3a^Δ/Δ;Vhl^Δ/Δ mice that were fed tamoxifen but not subjected to monthly imaging developed cysts similarly to those that were repeatedly imaged (Supplemental Figure 4), showing that the repeated injection of the contrast agent and µCT imaging does not contribute to the cystic phenotype.

Figure 2: Longitudinal imaging of cyst progression. Representative µCT images of wild-type (A), Kif3a^{Δ / Δ} (B), Kif3a^{Δ / Δ};Vhl^{Δ / Δ} (C), and Kif3a^{Δ / Δ}; Hif1a^{Δ / Δ} (D) mice taken at 8, 16, 28, and 36 weeks after treatment and corresponding histologic section of one kidney (36 weeks) from each of the imaged mice. Mice were scored visually as having a moderate or severe cystic phenotype and the number of animals of each genotype with each score is shown. H&E, hematoxylin and eosin. Bar, 1 mm in H&E–stained images.

Figure 3: Quantitative assessment of cystic burden reveals increased cyst formation in Kif3a^{Δ / Δ};Vhl^{Δ / Δ} kidneys. (A) Coronal µCT image and three-dimensional reconstruction of representative Kif3a^{Δ / Δ}, Kif3a^{Δ / Δ};Vhl^{Δ / Δ}, and Kif3a^{Δ / Δ};Hif1a^{Δ / Δ} kidneys at 36 weeks after treatment. Kidneys (yellow) and cysts (red) are shown on top of the x-ray image. (B) Percentage of animals displaying kidney cysts in µCT images at different time points after gene deletion. Kif3a^{Δ / Δ} (n=10), Kif3a^{Δ / Δ};Vhl^{Δ / Δ} (n=16), and Kif3a^{Δ / Δ};Hif1a^{Δ / Δ} (n=9). (C) Quantification of total cystic volume per kidney. Data are presented as the mean±SEM for Kif3a^{Δ / Δ} (n=14), Kif3a^{Δ / Δ};Vhl^{Δ / Δ} (n=30), and Kif3a^{Δ / Δ};Hif1a^{Δ / Δ} (n=16). (D) Number of cysts per kidney section determined by analyses of histologic sections through the longitudinal midline of kidneys. Data are presented as the mean±SEM for Kif3a^{Δ / Δ} (n=24), Kif3a^{Δ / Δ};Vhl^{Δ / Δ} (n=27), and Kif3a^{Δ / Δ};Hif1a^{Δ / Δ} (n=17). *P<0.05; ***P<0.001 (unpaired t test).

To determine whether increased cyst formation in Kif3a^Δ/Δ;Vhl^Δ/Δ mice results from increased cellular proliferation, we used bromodeoxyuridine (BrdU) to label proliferating cells in Kif3a^Δ/Δ;Vhl^Δ/Δ kidneys 3 weeks after initiation of tamoxifen feeding and analyzed tubules displaying mosaic gene deletion using Glut-1 immunostaining as a marker for Vhl deletion (Figure 4, A and B). Glut-1–positive and Glut-1–negative cells in these tubules exhibited very low rates of proliferation (<0.5%) with Glut-1–positive cells showing a slightly lower rate of proliferation (Figure 4C), indicating that increased proliferation does not underlie cyst formation in Kif3a^Δ/Δ;Vhl^Δ/Δ mice. To compare the proliferative status of epithelial cells in cystic lesions from the different genotypes, we selected cohorts of mice that displayed a comparable moderate cystic phenotype. Ki67 staining showed that some cysts contained proliferating cells, whereas others did not (Figure 4D). Quantifications revealed no differences between Kif3a^Δ/Δ, Kif3a^Δ/Δ;Hif1a^Δ/Δ, and Kif3a^Δ/Δ;Vhl^Δ/Δ mice in terms of the frequency of cysts that contained proliferating cells (Figure 4E) or the frequency of Ki67-positive cells within the proliferating cysts (Figure 4F). The increased cystic number and total cystic burden in Kif3a^Δ/Δ;Vhl^Δ/Δ mice appears to result from an increased frequency of cyst initiation at earlier time points, rather than an increased rate of proliferation of cystic cells.

Figure 4: No differences between genotypes in rates of proliferation of cystic cells. (A and B) Examples of tubules from Kif3a^{Δ / Δ};Vhl^{Δ / Δ} mice in which a wild-type cell (B) and a Vhl-deficient cell marked by positive Glut-1 staining (green) (A) stain positively for BrdU (red). (C) Percentage of Glut-1–positive (n=2629) and Glut-1–negative (n=2798) cells that stain positively for BrdU. Data are derived from analysis of multiple sections from three Kif3a^{Δ / Δ};Vhl^{Δ / Δ} mice. (D) Examples of cysts from each genotype that either display no cells that stain for Ki67 (Ki67⁻) or that contain Ki67-labeled cells (Ki67⁺). (E) Percentage of cysts that display Ki67-labeled cells. Data are presented as the mean±SEM for Kif3a^{Δ / Δ} (n=54), Kif3a^{Δ / Δ};Vhl^{Δ / Δ} (n=56), and Kif3a^{Δ / Δ};Hif1a^{Δ / Δ} (n=62). (F) Percentage of Ki67-positive cells per Ki67-positive cyst. Data are presented as the mean±SEM for Kif3a^{Δ / Δ} (n=36), Kif3a^{Δ / Δ};Vhl^{Δ / Δ} (n=37), and Kif3a^{Δ / Δ};Hif1a^{Δ / Δ} (n=36). Bar, 10 µm in A and B; 20 µm in D.

The majority of cysts in all genotypes were simple cysts lined by a single epithelial layer (Figure 5, A–C). In Kif3a^Δ/Δ and Kif3a^Δ/Δ;Hif1a^Δ/Δ mutant mice, 1.9% (n=204) and 2.1% (n=282) of cysts, respectively, displayed small foci of multilayered or disorganized epithelial growth that projected into the lumen of the cyst (Figure 5, A and C). By contrast, in Kif3a^Δ/Δ;Vhl^Δ/Δ mice, the frequency of these atypical cysts was 4-fold higher (8.0%, n=914) (Figure 5B). Atypical cysts were also observed in mice that were not subjected to monthly imaging (Supplemental Figure 4D), indicating that this phenotype is not a secondary effect of the experimental protocol. No tumors were observed in these mice.

Figure 5: Kif3a^{Δ / Δ};Vhl^{Δ / Δ} mice display increased frequencies of misoriented epithelial cell division and atypical cysts. Representative H&E-stained sections of simple cysts (left) and atypical cysts (right) from Kif3a^{Δ / Δ} (A), Kif3a^{Δ / Δ};Vhl^{Δ / Δ} (B), and Kif3a^{Δ / Δ}; Hif1a^{Δ / Δ} (C) mice 36 weeks after gene deletion was induced in adult mice. (D) Anti-phosphohistone H3 (red) and DAPI (blue) staining of normal (left) and misoriented (right) anaphases. The local plane of the cystic epithelium was determined by drawing a line through the two adjacent nuclei and the anaphase plane was determined by drawing a line parallel to the midline of the dividing chromosomes to calculate the anaphase angle α. (E) Quantification of the angle α of anaphases in Kif3a^{Δ / Δ} (n=41) and Kif3a^{Δ / Δ};Vhl^{Δ / Δ} (n=42) cystic epithelia. *P<0.05, Mann–Whitney U test. (F) Examples of cleaved caspase 3–positive cystic epithelial cells in Kif3a^{Δ / Δ} and Kif3a^{Δ / Δ};Vhl^{Δ / Δ} kidneys. The arrowhead points to an apoptotic cell lying above the epithelial layer. (G) Frequency of cysts in Kif3a^{Δ / Δ} (n=518), Kif3a^{Δ / Δ};Vhl^{Δ / Δ} (n=582), and Kif3a^{Δ / Δ};Hif1a^{Δ / Δ} (n=542) kidneys that displayed one or more cleaved caspase 3–positive cystic epithelial cells. H&E, hematoxylin and eosin; DAPI, 4′,6-diamidino-2-phenylindole. Bar, 20 µm in A–C; 10 µm in D; 20 µm in F.

Deletion of cyst-suppressing genes early in postnatal development induces a more severe and rapid-onset cystic phenotype than gene deletion in adult kidneys.^21,23 We injected nursing dams with a dose of tamoxifen that induced a limited number of cysts in the feeding pups and followed these using µCT imaging for 9 months (Supplemental Figure 5). Deletion of Vhl in P2 mice did not cause the development of cysts (Supplemental Figure 5). Staining for HIF1α and acetylated tubulin confirmed the functional effects of deletion of Kif3a and Vhl in the relevant genotypes (Supplemental Figure 6). This method of deletion gave considerable variation within and between litters in the extent of cyst formation, precluding meaningful quantitative comparisons of cystic burden. Morphologic analyses of kidneys 9 months after gene deletion revealed similar phenotypes to adult gene deletion with respect to the formation of simple and atypical cysts and absence of tumors. In Kif3a^Δ/Δ and Kif3a^Δ/Δ;Hif1a^Δ/Δ mutant mice, 3.3% (n=333) and 3.5% (n=692) of cysts, respectively, contained regions of atypical epithelial growth (Supplemental Figure 7), whereas 9.3% (n=647) of cysts in Kif3a^Δ/Δ;Vhl^Δ/Δ mice exhibited atypical morphology (Supplemental Figure 7). These results collectively show that combined mutation of Vhl and Kif3a increases the frequency of transition to atypical cysts.

Cyst initiation in several animal models of loss of function of primary cilia is preceded by misorientation of tubular epithelial cell divisions, resulting in cell divisions across the plane of the tubule and tubular expansion.²⁴ The formation of multilayered cystic epithelia might also involve nonplanar epithelial cell division. Anaphase cells in cystic epithelia of Kif3a^Δ/Δ and Kif3a^Δ/Δ;Vhl^Δ/Δ mice were identified by staining for phosphohistone H3 and the plane of cell division determined by measuring the angle between the plane of the cystic epithelium and a line parallel to the midline of the separating mitotic chromosome arrays (Figure 5D). Anaphases in Kif3a^Δ/Δ;Vhl^Δ/Δ cysts displayed a significantly greater spread of angles than anaphases in Kif3a^Δ/Δ cysts (Figure 5E). Notably, 33% of Kif3a^Δ/Δ;Vhl^Δ/Δ anaphases were oriented more out of the plane of the epithelium than within the plane of the epithelium (angle <45°), whereas only 8% of Kif3a^Δ/Δ cystic anaphases were misaligned. We reasoned that misoriented cellular division may result in loss of cellular contact with the basement membrane, causing cell death. Consistent with this idea, Kif3a^Δ/Δ;Vhl^Δ/Δ cysts more frequently displayed one or more cells that stained positively for cleaved caspase 3, a marker of apoptosis, than cysts in Kif3a^Δ/Δ and Kif3a^Δ/Δ;Hif1a^Δ/Δ kidneys (Figure 5, F and G). In many cases, these cleaved caspase 3–positive cells were observed above the layer of the cystic epithelium, consistent with our model.

Discussion

We previously proposed that loss of the microtubule-stabilizing function of pVHL in combination with secondary signaling alterations in the PI3K pathway results in failure of epithelial cells to maintain the primary cilium, causing preneoplastic cyst formation in patients with VHL disease.^3,15,16 This study validates this hypothesis but also suggests a modification of the initial model of cyst formation. Combined deletion of Vhl and Kif3a shortens the latency of cyst initiation and increases the total number of cysts and total cystic burden compared with deletion of Kif3a alone, arguing that loss of the primary cilium is not the sole initiating event in cyst formation but that loss of the primary cilium acts cooperatively with loss of pVHL function to induce cyst formation. Because simple cysts are believed to progress to atypical cysts in the cyst-dependent pathway of ccRCC formation, the increased frequency of atypical cysts in Kif3a^Δ/Δ;Vhl^Δ/Δ mice indicates that Vhl loss plus cilium loss promotes the progression toward ccRCC. However, because these lesions persist for at least 6–9 months in mice without progressing to form tumors it is likely that additional mutations that occur frequently in ccRCC such as PBRM1, BAP1, SETD2, or TP53 or PI3K pathway alterations^4,25 are required to cause tumor formation from these atypical cysts. Consistent with the idea that cilia loss is an early event in ccRCC formation, sporadic human ccRCC tumors exhibit very low frequencies of ciliated cells.¹¹ Although genomic analyses of ccRCC did not reveal the mutation of genes that encode structural components of primary cilia,^4,25 activation of the PI3K pathway is a frequent event^4,25–29 and is known to cause loss of cilia in VHL null cells.^15,16 It is possible that other yet-to-be-identified genetic alterations may also cooperate with loss of VHL to cause loss of the cilium. Interestingly, homozygous deletion of Vhl combined with heterozygous deletion of Bap1 during mouse kidney development caused the formation of simple and atypical cysts as well as tumors.³⁰ The observation that these lesions exhibit high levels of PI3K pathway activation suggests that investigation of the status of primary cilia in these mice is warranted.

Because many normal tubules harboring deletion of Vhl and Kif3a do not form cysts within 9 months of gene deletion, it is evident that loss of Vhl together with loss of primary cilia does not automatically lead to cyst formation. Cellular proliferation after damage to nephrons triggers cyst formation in adult kidneys that lack a variety of cyst-predisposing genes.^21,23 We suggest that the low rate of continuous cell turnover in the adult kidney³¹ is the trigger for initiation of cyst formation by Kif3a- or Vhl/Kif3a-deficient cells, with the latter genotype being more sensitized than the former to initiate cyst formation. We observed no increase in proliferation rate of Vhl/Kif3a mutant cells at an early time point after gene deletion, arguing that the increased initiation of cyst formation is not due to alteration in the rate of cellular turnover. A likely contributing factor to increased cyst formation is dysregulation of planar epithelial cell division. During development or repair of a damaged adult kidney tubule, cellular divisions normally orient along the length of the tubule. Misoriented cellular divisions across the plane of the tubule increase the tubular diameter, an initial step in cyst formation.²⁴ The primary cilium controls oriented cell division via noncanonical WNT and planar cell polarity signaling pathways; loss of cilia, or of several important signaling components of cilia, causes misorientation of cellular division in normal tubules before cysts form.²⁴ pVHL also controls the plane of cellular division by regulating the stability of the astral microtubule network, thereby ensuring the correct orientation of the mitotic spindle.³² Ischemic damage to induce cellular proliferation in Vhl null kidney tubules increased the frequency of misoriented epithelial cell divisions, correlating with the formation of microcysts in these animals.³³ Consistent with the idea that Vhl/Kif3a mutation compromises two different mechanisms that ensure oriented cell division, anaphases in cystic epithelia of Kif3a^Δ/Δ;Vhl^Δ/Δ mice were four times more frequently misoriented out of the plane of the cystic epithelium than those in Kif3a^Δ/Δ mice. The observation that anaphases in Kif3a^Δ/Δ;Vhl^Δ/Δ cysts exhibit a larger scatter of angles than those in Kif3a^Δ/Δ mice is consistent with the spindle tumbling phenotype caused by loss of Vhl that results in cells frequently choosing a random plane of cell division.^32,33 Expansion of a simple cyst and maintenance of a single epithelial layer requires that cells continue to divide within the plane of the cystic epithelium. The transition to multilayered epithelial growth might occur if a cell division occurs out of the plane of the cystic epithelium, resulting in a daughter cell growing on top of the existing cystic epithelium. The fact that the frequency of misaligned anaphases (33%) is higher than the frequency of atypical cysts (8%–9%) in Kif3a^Δ/Δ;Vhl^Δ/Δ mice may potentially be reconciled by the failure of daughter cells to survive if they divide away from contacts with the basement membrane. Indeed, an enhanced rate of apoptosis was observed in cysts in Kif3a^Δ/Δ;Vhl^Δ/Δ mice. Although the extremely low rate of cellular division in adult kidneys prevented assessment of the orientation of cellular division in tubules before cyst initiation, it is plausible to suggest that an increased rate of misoriented cell division by Kif3a^Δ/Δ;Vhl^Δ/Δ tubular epithelial cells during normal cellular turnover might contribute to increased cyst initiation.

Vhl deletion in mouse proximal tubular epithelial cells caused cyst formation in a small fraction of older mice.³⁴ Because this phenotype could be rescued by codeletion of Arnt, encoding HIF1β (the dimerization partner of HIF1α and HIF2α that is necessary for transcriptional activation), but not by codeletion of Hif1a,³⁴ it is possible that the combined activities of HIF1α and HIF2α may promote cyst formation in Kif3a^Δ/Δ;Vhl^Δ/Δ mice. A more general role for HIF1α stabilization in cyst progression has been recently proposed.^19,20 However, in this study, we find that cystic lesions induced by Kif3a deletion are not characterized by HIF1α stabilization nor does the deletion of Hif1a have any effect on the formation of cysts in this model. HIF1α stabilization may be limited to more severe polycystic kidneys than achieved in this model, or cellular signaling changes induced by mutation of the PKD1/PKD2 genes in autosomal dominant polycystic kidney disease may contribute to HIF1α activation in cysts through constitutive mammalian target of rapamycin activation,^35–37 which causes enhanced translation of HIF1α.^38,39

Finally, these studies have developed and validated a noninvasive method of quantification of cystic burden in mice that should prove useful for future studies investigating the genetic basis of, and therapies for, polycystic kidney diseases and renal cancers.

Concise Methods

Animals

Mice expressing a tamoxifen-inducible kidney-specific Cre recombinase under the Ksp1.3 promoter⁴⁰ were crossed with Kif3a^fl/fl,¹⁸Vhl^fl/fl,⁴¹ and/or Hif1a^fl/fl42 animals. Gene deletion in 6-week-old animals was achieved by feeding with tamoxifen food (400 parts per million) from Harlan Laboratories for 2 weeks. For gene deletion in P2 animals, nursing dams were injected intraperitoneally with tamoxifen (0.1 mg/g body wt) from days P2–P4. BrdU (80 µg/g body wt) was injected intraperitoneally 18 hours before euthanasia. Mouse experiments were conducted under experimental license 131/2012 from the Veterinary Office of the Canton of Zurich. Sequences of genotyping primers are given in Supplemental Table 1.

µCT Imaging

Visipaque 270 (GE Healthcare) was injected intravenously at 8 µl/g body wt as described.²²µCT images were obtainedwith a Quantum FX microCT (Perkin Elmer) with the following settings: 100 µA, 90 kV, respiratory gating, and fine scanning. µCT data were exported as DICOM images to be analyzed manually with the Myrian software (intrasense) for volumetric measurements and three-dimensional reconstruction and in the Quantum FX data format for overview x-ray pictures.

Immunohistochemistry

Immunohistochemistry and immunofluorescence of formalin-fixed paraffin-embedded kidneys was performed as described⁴³ with the following antibodies: α-ATPV1B1 (gift from C. Wagner), α-Aqp2 (gift from J. Loffing), BrdU (MAP3510; EMD Millipore), cleaved caspase 3 (9661; Cell Signaling Technology), α-Glut1 (Ab14683; Abcam, Inc.), α-HIF1α (NB100-105; Novus Biologicals ), α-Ki67 (TEC-3; DakoCytomation), α-NaPi (gift from J. Biber), α-NCC (AB3553; EMD Millipore), α-THP (sc-20631; Santa Cruz Biotechnology) and α-acetylated tubulin (T6793; Sigma-Aldrich). Secondary Alexa 488– or Alexa 568–coupled antibodies were from Life Technologies. Horseradish peroxidase –coupled α-mouse secondary antibody was from Thermo Fisher Scientific.

Genomic DNA Quantitative PCR Analyses

Formalin-fixed, paraffin-embedded kidney sections were washed in Xylol for 1 minute twice and air dried. Genomic DNA was isolated from these sections or from powdered whole kidneys with the Arcturus PicoPure DNA Extraction Kit (Life Technologies). Real-time PCR analysis was performed with SYBR FAST Universal 2× qPCR Master Mix and the primers are listed in Supplemental Table 1.

Blood Plasma Analyses

Animals were anesthetized and blood was collected by puncture of the heart and treated with sodium-heparin. Creatinine and BUN were determined using the UniCel DxC 800 Synchron Clinical System.

Disclosures

None.

We thank Peter Igarashi for providing mice, Nadine Nägele for assistance with plasma measurements, Thi Dan Linh Nguyen-Kim and Olivio Donati for assistance with quantifications of µCT images, and Jürg Bieber, Carsten Wagner and Johannes Loffing for providing antibodies.

This work was supported by the Swiss National Science Foundation (SNF Professorship PP00P3_128257) and the European Research Council (Starting Grant 260316).