Loss of Glis2/NPHP7 causes kidney epithelial cell senescence and suppresses cyst growth in the Kif3a mouse model of cystic kidney disease

Abstract

Enlargement of kidney tubules is a common feature of multiple cystic kidney diseases in humans and mice. However, while some of these pathologies are characterized by cyst expansion and organ enlargement, in others, progressive interstitial fibrosis and kidney atrophy prevail. The Kif3a knockout mouse is an established non-orthologous mouse model of cystic kidney disease. Conditional inactivation of Kif3a in kidney tubular cells results in loss of primary cilia and rapid cyst growth. Conversely, loss of function of the gene GLIS2/NPHP7 causes progressive kidney atrophy, interstitial inflammatory infiltration, and fibrosis. Kif3a null tubular cells have unrestrained proliferation and reduced stabilization of p53 resulting in a loss of cell cycle arrest in the presence of DNA damage. In contrast, loss of Glis2 is associated with activation of checkpoint kinase 1, stabilization of p53, and induction of cell senescence. Interestingly, the cystic phenotype of Kif3a knockout mice is partially rescued by genetic ablation of Glis2 and pharmacological stabilization of p53. Thus, Kif3a is required for cell cycle regulation and the DNA damage response, whereas cell senescence is significantly enhanced in Glis2 null cells. Hence, cell senescence is a central feature in nephronophthisis type 7 and Kif3a is unexpectedly required for efficient DNA damage response and cell cycle arrest.

The term cystic kidney disease refers to several conditions that have in common cystic expansion of renal tubules. In some cases, such as in polycystic kidney disease, cystic growth is the prevalent feature and is sustained by highly proliferating cells lining multiple fluid-filled cysts originating from tubular segments, a feature that caused this disease to be dubbed “neoplasia in disguise.”^1,2 On the other hand, kidneys in nephronophthisis (NPHP), a group of autosomal recessive cystic diseases, which represent the most frequent genetic cause of renal failure in the first 3 decades of life,³ are often decreased in volume, due to progressive tubular atrophy and interstitial fibrosis. Virtually, all the proteins encoded by the genes mutated in cystic kidney diseases are localized to the primary cilium of kidney epithelial cells.^4,5 The primary cilium is an important subcellular organelle and its integrity has been shown to be required for the correct functioning of several signaling pathways, including hedgehog (Hh) signaling.^6-9 Interestingly, recent evidence has shown that several ciliary proteins are unexpectedly related to the regulation of cellular response to DNA damage and DNA repair.^10-12 The kinesin Kif3a is among the many proteins necessary for the building of primary cilia, and kidney-specific inactivation of Kif3a in mice results in loss of cilia and rapid cyst formation in the kidneys.¹³ On the contrary, loss of function of the gene GLIS2/NPHP7, a repressor of Hh signaling,¹³ causes a NPHP-like phenotype in humans and mice, characterized by progressive kidney atrophy, interstitial inflammatory infiltration, and fibrosis.^5,13,14 This suggests that different mechanisms underlie the kidney phenotype of these 2 mouse models of cystic kidney disease, which possibly affect cell proliferation and apoptosis. The protein Glis2 is a paralog of another Hh inhibitor, Gli3, whose activity is dependent on primary cilia.⁹ To investigate a possible dependence of Glis2 on primary cilia and whether additional mechanisms are determining the kidney phenotype in Glis2 knockouts, we knocked out Glis2 in a mouse with kidney-specific (Ksp) inactivation of Kif3a (^Ksp-creKif3a^f/-). We found that genetic inactivation of Glis2 in kidney-specific Kif3a knockout mice, partially suppresses uncontrolled cell proliferation, cyst growth, and tubular apoptosis in this mouse model of cystic kidney disease. We show that immortalized tubular epithelial cells derived from Kif3a null kidneys display impaired stabilization of p53 in the presence of spontaneous DNA damage, defective activation of the G1/S checkpoint, ectopic cyclin B1 expression, and failure to arrest in the cell cycle, with consequent increased rates of cell duplication and apoptosis. Oppositely, stable Glis2 short hairpin RNA (shRNA)-mediated silencing is accompanied by activation of the serine-threonine-specific checkpoint kinase 1 (Chk1), stabilization of p53, and induction of cell senescence, a permanent cell cycle arrest, which reduces DNA damage and apoptosis in Kif3a null cells. Importantly, in vivo, pharmacological stabilization of p53, using an inhibitor of the protein mouse double minute 2 (MDM2), partially rescued cyst growth, tubular proliferation, and apoptosis in ^Ksp-creKif3a^f/- mice. Our findings indicate that Kif3a is required for cell cycle regulation and its loss results in impaired DNA damage response, whereas loss of Glis2 induces abnormal activation of Chk1 and promotes cell senescence. These results indicate that cell senescence is a central feature in NPHP type 7 and reveal an unexpected requirement of Kif3a for efficient DNA damage response and cell cycle arrest.

RESULTS

Glis2 inactivation in ^Ksp-creKif3a^f/- knockouts results in reduced cyst growth, improved renal function, and decreased tubular cell proliferation and apoptosis

Glis2, a repressor of Hh signaling is localized at the primary cilia of kidney epithelial cells, as its orthologs Gli1, Gli2, and Gli3.^5,13,14, Intact primary cilia are required for correct functioning of Gli3, another repressor of the Hh signaling pathway.⁹ To determine whether Glis2 functions are also dependent on primary cilia, we performed a standard epistasis experiment, by inactivating Glis2 in kidney-specific Kif3a^f/- knockout mice.^15,16 We generated ^Ksp-creKif3a^f/-;Glis2^mut/mut double knockout-transgenic mice (indicated in the figures simply as Kif3a^f/-;Glis2^-/- for graphical purposes) from Glis2^mut/mut null mice,¹⁴ Kif3a^+/f,¹⁶ Kif3a^+/-17 mice, and the Ksp-cre transgenic mouse.¹⁵ Kidneys of ^Ksp-creKif3a^f/- mice develop severe cystic kidney disease early in life and die at about 6 weeks of age.¹⁶ Glis2^mut/mut kidneys, instead, show increased expression of genes related to inflammation and tissue repair at young ages (extracellular matrix proteases, matrix chemokines, and cytokines) but develop tubular atrophy and interstitial fibrosis only later in life.^5,14 We evaluated the morphology of paraffin-embedded kidneys of ^Ksp-creKif3a^f/-;Glis2^mut/mut double knockout-transgenic mice at postnatal day P30 and compared it with ^Ksp-creKif3a^f/-;Glis2^+/mut mice of the same age (Figure 1a and Supplementary Figure S1). Kidneys of ^Ksp-creKif3a^f/-;Glis2^mut/mut double knockouts had delayed cyst growth, as shown by decreased kidney–total body weight ratio, reduced cystic area per total kidney area, and reduced average cyst size (Figure 1b). ^Ksp-creKif3a^f/-;Glis2^mut/mut double knockouts also had lower levels of blood urea nitrogen and serum creatinine than ^Ksp-creKif3a^f/-;Glis2^+/mut mice did (Figure 1c). To determine whether reduced cyst growth in kidneys of ^Ksp-creKif3a^f/-;Glis2^mut/mut double knockouts is associated with decreased cellular proliferation or increased apoptosis, we examined the rates of proliferation of tubular epithelial cells in ^Ksp-creKif3a^f/-;Glis2^mut/mut double knockouts and ^Ksp-creKif3a^f/-;Glis2^+/mut single mutants. This was accomplished by using antibodies against the cell cycle antigen Ki67 and an antibody against activated caspase 3 (aCasp3) that marks apoptotic cells. The proliferation index, expressed as percentage of Ki67-positive cells over total number of cells per optical field, was the highest in ^Ksp-creKif3a^f/-;Glis2^+/mut and significantly reduced in ^Ksp-creKif3a^f/-;Glis2^mut/mut double mutants (Figure 1d, upper panels, and e). Similarly, apoptosis in tubular cells was higher in ^Ksp-creKif3a^f/-;Glis2^+/mut single mutants and decreased in ^Ksp-creKif3a^f/-;Glis2^mut/mut double knockouts (Figure 1d, lower panels, and f). Primary cilia were absent in cysts of both ^Ksp-creKif3a^f/- and ^Ksp-creKif3a^f/-;- Glis2^mut/mut mice (Supplementary Figure S2A). Interestingly, expression of Gli1, one of the target genes of Hh that can be used as readout of the activation of this signaling pathway, was increased in ^Ksp-creKif3a^f/f immortalized cells⁸ (Supplementary Figure S2B). We conclude that Glis2 inactivation reduces kidney cyst growth and preserves renal function in the Kif3a mouse model of polycystic kidney disease by reducing tubular cell proliferation and not by inducing apoptosis.

Figure 1. Glis2 inactivation in Kif3a^f/- knockouts results in reduced cyst growth, improved renal function, and decreased tubular cell proliferation and apoptosis.

(a) Macroscopic representative images of Kif3a^f/-;Glis2^+/-, Kif3a^f/-;Glis2^-/-, and Kif3a^f/+;Glis2^-/- mutant kidneys at P30 (upper panel) and the corresponding histology (hematoxylin and eosin staining) at low (middle panels) and high magnification (low panels). Kidneys of Kif3a^f/-;Glis2^-/- double mutants are smaller, have fewer and smaller cysts, and have better preserved parenchyma when compared with Kif3a^f/-;Glis2^+/- kidneys. Bars = 500 μm in the upper and middle panels and 100 μm in the lower panels. (b) Quantification of kidney weight to body weight ratio (KW/BW, g/g); cyst area (expressed as percentage of cyst area over total kidney area); and average cyst size in Kif3a^f/-;Glis2^+/- (red squares), Kif3a^f/-;Glis2^-/- (orange triangles), and Kif3a^f/+;Glis2^-/- (blue circles). The number of mice in each experimental group is reported on the x-axis. Results are mean ± SD. P values were obtained by Student t-test. (c) Renal function indexes in Kif3a^f/-;Glis2^+/-, Kif3a^f/-;Glis2^-/-, and Kif3a^+/f;Glis2^-/- mice. Serum creatinine (sCr) and blood urea nitrogen (BUN) are expressed in mg/dl. The number of mice in each experimental group is reported on the x-axis. Results are mean ± SD. P values were obtained by Student t-test. (d) Representative immunofluorescence microscopy images of Kif3a^f/-;Glis2^-/- double knockout kidneys compared with Kif3a^f/-;Glis2^+/-, Kif3a^+/f;Glis2^-/- single mutants using antibodies to Ki67, a marker of proliferating cells, and activated caspase-3 (aCasp3), a marker of apoptosis. Bars = 100 μm. (e) Quantification of tubular proliferation in Kif3a^f/-;Glis2^+/-, Kif3a^f/-;Glis2^-/-, and Kif3a^+/f;Glis2^-/- mice expressed as percentage of Ki67-positive cells over total number of cells (n = 3 mice per experimental group, 10 optical fields per mouse). Results are mean ± SEM. P values were obtained by Student t-test. (f) Quantification of the apoptotic tubular cells in Kif3a^f/-;Glis2^+/-, Kif3a^f/-;Glis2^-/-, and Kif3a^+/f;Glis2^-/- mice obtained using the percentage of positive aCasp3 cells over total number of cells (n = 3 mice per experimental group, 10 optical fields per mouse). Results are mean ± SD. P values were obtained by Student t-test.

Kif3a null kidney epithelial cells have accelerated cell cycle

To acquire more details about the causes of the high tubular proliferation rate observed in ^Ksp-creKif3a^f/- knockout mice in vivo, we examined the cell cycle of the previously described immortalized ^Ksp-creKif3a^f/f kidney epithelial cell line⁸ (reported shortly as Kif3a^f/f in the figures for graphical purposes) by flow cytometry and compared it with ^Ksp-creKif3a^+/f hemizygous cells as a control. Neither protein expression nor subcellular localization of Glis2 was significantly affected by the absence of Kif3a (Supplementary Figure S3A and B). We noticed that a higher proportion of ^Ksp-creKif3a^f/f cells than the control cells were in the S and G2/M phases of the cell cycle (Figure 2a and b). Then, we generated stable Glis2 knockdown cell lines from both ^Ksp-creKif3a^f/f null and ^Ksp-creKif3a^+/f hemizygous cells by lentivirus-mediated shRNA infection and assessed knock-down efficiency by Western blot (Supplementary Figure S3A). Interestingly, stable inactivation of Glis2 by shRNA-mediated silencing (indicated as Glis2^KD in the figures, KD for knock-down) in ^Ksp-creKif3a^f/f cells significantly decreased the proportion of cycling cells (Figure 2a and b). The accelerated cell cycle of the immortalized ^Ksp-creKif3a^f/f cells was further assessed by measuring cellular proliferation via a tetrazolium dye colorimetric assay (MTT) (Figure 2c) and by demonstrating increased bromodeoxyuridine (BrdU) incorporation (Figure 2d) in null cells compared with ^Ksp-creKif3a^+/f heterozygous control cells. Primary tubular cells obtained from ^Ksp-creKif3a^f/-;Glis2^mut/mut double knockouts had also higher proportion of cells in G1 and longer doubling time than did ^Ksp-creKif3a^f/- single mutant cells (Figure 2e). Although no differences in the cell cycle were noticed between ^Ksp-creKif3a^f/- and wild-type cells by flow cytometry cell cycle analysis, due to the overall slower proliferation rates of primary cells compared with the immortalized clones in which a residual activity of the SV40 antigen may still be present, increased proliferation of ^Ksp-creKif3a^f/- primary cells was clearly detectable by the MTT assay (Figure 2f). Collectively, these experiments indicate that the higher proliferation rate of Kif3a null kidney epithelial cells is cell-autonomous and their cell cycle anomaly is rescued by inactivation of Glis2.

Figure 2. Kif3a null kidney epithelial cells have accelerated cell cycle.

(a) Representative images of flow cytometry cell cycle analysis of Kif3a^f/f and Kif3a^+/f immortalized tubular cells stably transfected with nontargeting short hairpin RNA (nt-shRNA) or with Glis2-targeting shRNA (clones Glis2^KD-5.1 and Glis2^KD-7.2 for Kif3a^f/f cells, Glis2^KD-5.2 and Glis2^KD-7.1 for Kif3a^+/f cells; KDindicates knock-down).DNA content is represented on the x-axis; cell counts are shown on the y-axis. Counts of cellswith diploid (2n) DNA content (G1 phase) are highlighted in blue, tetraploid (4n) cells in pink (G2/M phase), and cells with intermediate DNA content (S phase) are highlighted in green. Insets report the relative cell cycle distribution (expressed as percentage) represented in each graphic. Percentages do not sumto 100%because debris and aggregated cells were excluded. (b) Quantification of the cell cycle distribution of Kif3a^f/f and Kif3a^+/f immortalized tubular cells with and without concomitant silencing of Glis2. Numbers in the pie chart correspond to average percentage of cells in each phase of the cell cycle (n=4 independent experiments). Percentages do not sumto 100% because they were obtained after gating out debris and aggregated cells. Cells in G1 are represented in blue, cells S in green, and cells in G2/Min pink. P values were obtained by Student t-test. (c) Assessment by colorimetric assay (MTT) of doubling times of Kif3a^f/f and Kif3a^+/f immortalized tubular cells with andwithout concomitant silencing of Glis2 at different time points. Values on the y-axis represent fold increase of optical density (OD). Doubling times are shorter in Kif3a^f/f null cells than in the heterozygous control cells and decrease after shRNA-mediated Glis2 silencing. n = 3 independent experiments. Results are mean ± SD. P values were obtained by Student t-test between Kif3a^f/f and Kif3a^+/f (blue brackets and values) and between Kif3a^f/f and Kif3a^f/f;Glis2^KD-5.2 knockout–knock-down cells (orange bracket and values). (d) Cumulative BrdU incorporation in Kif3a^f/f and Kif3a^+/f immortalized tubular cells with and without concomitant Glis2 silencing at different time points. Values on the y-axis represent the percentage of BrdU-positive cells at each time point. BrdU incorporation rates are increased in Kif3a^f/f null cells compared with the heterozygous control cells and decrease after shRNA-mediated Glis2 silencing. n = 3 independent experiments. Results are mean ± SD. P values were obtained by Student t-test between Kif3a^f/f and Kif3a^+/f (blue brackets and values) and between Kif3a^f/f and Kif3a^f/f;Glis2^KD-5.2 knockout–knock-down cells (orange bracket and values). (e) Representative images of flow cytometry cell cycle analysis of primary tubular cells obtained from Glis2^-/-, Kif3a^f/-, Kif3a^f/-;Glis2^-/- double knockouts, and wild-type mice as performed in (a). (f)MTT test onprimary tubular cells from Kif3a^f/-, Glis2^-/-, and Kif3a^f/-;Glis2^-/- mice. Values on the y-axis represent fold increase of optical density, timepoints are on the x-axis. Doubling times are shorter in Kif3a^f/- null cells than in Glis2^-/- and Kif3a^f/-;Glis2^-/- cells. Results are the average of 2 triplicate experiments and are mean ± SEM. P values were obtained by Student t-test.

Kif3a null kidney epithelial cells exhibit increased DNA damage and apoptosis

High cellular proliferation rates are often associated with increased DNA damage due to genotoxic stress (stalling of replication forks and incomplete DNA replication) and increased production of oxygen radicals, secondary to the alteration of the mitochondrial metabolism.¹⁸ Because of the high proliferation rates exhibited by ^Ksp-creKif3a^f/f cells, we tested the occurrence of DNA damage in ^Ksp-creKif3a^f/f cells versus ^Ksp-creKif3a^+/f control cells, using an antibody against the phosphorylated histone 2AX (γH2AX). Histone 2AX is recruited at sites of DNA breaks, where it is phosphorylated on Ser139 by the protein kinase ataxia-telangiectasia-mutated (ATM),¹⁹ and is commonly used as a marker of DNA damage. ^{20 Ksp-cre}Kif3a^f/f cells, compared with the ^Ksp-creKif3a^+/f control cells, had significantly increased DNA damage (measured by determining the average number of positive foci per nucleus in immunofluorescence confocal microscopy images) (Figure 3a and b) Similar results were also observed by Western blot (Figure 3c). Furthermore, the number of γH2AX foci in ^Ksp-creKif3a^f/f null cells dramatically decreased after shRNA-mediated Glis2 silencing (Figure 3a). With the exception of ^Ksp-creKif3a^+/f;Glis2^KD-5.2, which may be related to clonal variability, Glis2 knock-down, ^Ksp-creKif3a^+/f heterozygous cells had more DNA damage than did the nontargeted control, but was lower than ^Ksp-creKif3a^f/f knockout cells. Similarly, sections of ^Ksp-creKif3a^f/-;Glis2^+/mut kidneys showed higher proportion of cells with DNA damage than did ^Ksp-creKif3a^f/+;Glis2^mut/mut mice, and the number of tubular cells positive for γH2AX was significantly reduced in ^Ksp-creKif3a^f/-;Glis2^mut/mut double mutants (Figure 3b, e, and f). Collectively, these results suggest that both Kif3a and Glis2 null cells are subject to DNA damage, which is higher in Kif3a knockout cells than in Glis2 null cells. However, concomitant inactivation of Glis2 is associated with reduced DNA damage in Kif3a knockout cells.

Figure 3. Kif3a null kidney epithelial cells exhibit increased DNA damage and apoptosis.

(a) Representative immunofluorescence confocal microscopy images of antibody against phosphorylated histone 2AX (γH2AX) in Kif3a^f/f and Kif3a^+/f immortalized tubular cells with and without concomitant silencing of Glis2. (b) Quantification of DNA damage of (a). Values express average number of foci per nucleus (4 consecutive 3-μm confocal stacks from 10 consecutive images and 4 independent experiments). Kif3a^f/f and Glis2 knock-down Kif3a^+/f heterozygous cells show increased DNA damage compared with the respective control cells. Short hairpin RNA (shRNA)-mediated Glis2 silencing in Kif3a^f/f cells results in reduction of the number of foci per nucleus. Results are mean ± SD. P values were obtained by Student t-test. (c) Western blots of Kif3a^f/f and Kif3a^+/f immortalized tubular cell lysates with and without concomitant silencing of Glis2, probed with an antibody against γH2AX. (d) Representative immunofluorescence microscopy images of sections of Kif3a^f/-;Glis2^+/-, Kif3a^f/-;Glis2^-/- double mutants, Kif3a^+/f;Glis2^-/-, and wild-type kidneys using an antibody against γH2AX and showing cells with DNA damage. (e) Quantification of (d). Values express percentage of positive cells (n = 3 mice, 10 consecutive optical fields per mouse). Results are mean ± SD. P values were obtained by Student t-test. (f) Western blot of wild-type (n = 2), Glis2^-/- (n = 3), Kif3a^f/-;Glis2^+/- (n = 3), and Kif3af^/+;Glis2^-/- (n = 2) kidneys lysates probed with an antibody against γH2AX.

^Ksp-creKif3a^f/f kidney epithelial cells have defective G1/S cell cycle checkpoint in the presence of DNA damage

The coexistence of diffuse DNA damage and accelerated cell cycle in ^Ksp-creKif3a^f/f null cells was surprising, because diffuse DNA damage is expected to activate DNA damage response and cell cycle arrest, in order to allow DNA repair and avoid the transmission of compromised genetic material to daughter cells.²¹ In the presence of DNA damage, the DNA damage response blocks the cell cycle at the transition between the G1 and S phase (G1/S checkpoint), preventing initiation of replication. There is also a block between prophase and anaphase, before the onset of mitosis (G2/M checkpoint), that impedes progression through mitosis. We tested the competency of the G1/S and the G2/M checkpoints in the ^Ksp-creKif3a^f/f null cell line: cells were synchronized in G1 using double block with thymidine to prevent them from entering the S phase, or in G2 with nocodazole, which interferes with the organization of the spindle microtubules in prophase and prometaphase and prevents cells to undergo mitosis. To assess the state of the G1/S checkpoint, we measured by flow cytometry the fraction of G1-synchronized ^Ksp-creKif3a^f/f and ^Ksp-creKif3a^+/f cells that transit into the S phase of the cell cycle after thymidine washout. After release of the thymidine block (Figure 4a and b), the percentage of ^Ksp-creKif3a^f/f cells progressing into the S phase was higher than for the heterozygous control cells, indicating a decreased competency of the G1/S restriction point, which was restored after shRNA-mediated silencing of Glis2. The ratio of cells in G2/M 6 hours after release of prolonged nocodazole block was lower in ^Ksp-creKif3a^f/f cells than in ^Ksp-creKif3a^+/f control cells and was increased after shRNA-mediated Glis2 silencing, suggesting a possible activation of the G2/M checkpoint (Figure 4c and Supplementary Figure S4). Because this experiment only identifies cells with 4n DNA content, but does not distinguish between premitotic (G2) and mitotic (M) cells, we performed bivariate flow cytometry analysis, using an antibody against phosphorylated histone H3 (pHis3), a nuclear antigen present only in mitotic cells.^11,22 The fraction of cells in mitosis was about one-half in ^Ksp-creKif3a^f/f null cells compared with ^Ksp-creKif3a^+/ control cells (Figure 4d to f and Supplementary Figure S5). These results indicate that the G1/S checkpoint is defective in ^Ksp-creKif3a^f/f null cells and it is rescued by concomitant inactivation of Glis2. The G2/M checkpoint, in contrast, is effective in ^Ksp-creKif3a^f/f null cells, suggesting that the increased proportion of cells transiting through the G2/M phase after Glis2 shRNA silencing reported in Figure 4c is a secondary effect due to reduction of DNA damage.

Figure 4. Kif3a^f/f kidney epithelial cells have defective G1/S cell cycle checkpoint in the presence of DNA damage.

(a) Representative flow cytometry cell cycle analysis of Kif3a^f/f and Kif3a^+/f immortalized tubular cells with and without concomitant silencing of Glis2 after release of double thymidine block, which synchronizes cells in G1. DNA content is represented on the x-axis, cell counts on the y-axis. Samples were assayed at time 0 (in the presence of thymidine) and 3, 6, and 8 hours (after thymidine washout). (b) Quantification of (a). Values on the y-axis represent percentage of cells that exited G1 at each time point. n = 3 independent experiments. Results are mean ± SEM. P values were obtained by Student t-test between Kif3a^f/f and Kif3a^+/f (blue brackets and values) and between Kif3a^f/f and Kif3a^f/f;Glis2^KD-5.1 knockout–knockdown cells (orange bracket and values). (c) Quantification of Kif3a^f/f and Kif3a^+/f immortalized tubular cells with and without concomitant silencing of Glis2, after release of the nocodazole block (n = 3 independent samples). Values on the y-axis represent percentage of cells in G2/M, 6 hours after nocodazole washout. Results are mean ± SD. P values were obtained by Student t-test. (d) Flow cytometry bivariate analysis of Kif3a^f/f and Kif3a^+/f cells in mitosis, using an antibody against phosphorylated histone H3 (pHis3). Samples were enriched of mitotic cells by being exposed to nocodazole for 6 hours. Numbers are the percentage of pHis3-positive cells in G2/M. (e) Representative graphic of the same experiment, showing the cumulative distribution of Kif3a^f/f compared with Kif3a^+/f pHis3-positive cells. Cell counts are on the y-axis, fluorescence intensity is in the x-axis. (f) Graphic reporting the statistics of a triplicate experiment. Results are mean ± SD. P value was obtained by Student t-test. nt-shRNA, nontargeting short hairpin RNA.

Kif3a and Glis2 null kidney epithelial cells have defective and excessive activation of p53, respectively

The presence of numerous nuclear γH2AX foci in ^Ksp-creKif3a^f/f cells indicated that detection of DNA damage and activation of the early stages of the DNA repair cascade are not compromised, suggesting that downstream mechanisms are implicated in the lack of G1/S competency in these cells. In mammals, DNA damage is principally sensed by the protein kinases, ATM, ataxia telangiectasia, and Rad3-related (ATR), which propagate the signal through the phosphorylation of 2 downstream kinases, Chk1 and Chk2. To explore the state of these kinases, we examined levels of Chk1 and Chk2 and their activated phosphorylated forms (Ser345 of Chk1, Ser19 of Chk2) in ^Ksp-creKif3a^f/f and ^Ksp-creKif3a^+/f cells, with and without concomitant Glis2 silencing, by Western blot. We observed more Chk2 activation in ^Ksp-creKif3a^f/f cells than in ^Ksp-creKif3a^+/f control cells, independent of Glis2 silencing in all the tested cell lines (Figure 5a, upper panels). In contrast, no differences in Chk1 phosphorylation were evident between ^Ksp-creKif3a^f/f and ^Ksp-creKif3a^+/f, but this kinase was strongly activated in both cell lines after shRNAmediated silencing of Glis2, suggesting that the normalizing effect of Glis2^KD on ^Ksp-creKif3a^f/f cells’ cycle may be associated with the activation of Chk1 (Figure 5a, middle panels). Chk2 and Chk1 are nonrelated kinases with partially overlapping functions that phosphorylate common substrates to promote cell cycle arrest.²³ The tumor suppressor p53 is one of the principal effectors downstream of Chk1 and Chk2, and it is phosphorylated and stabilized by these 2 kinases.^19,23,24 By Western blot, we found that compared with ^Ksp-creKif3a^+/f heterozygous control cells, ^Ksp-creKif3a^f/f cells have slightly higher content of p53 that increased after Glis2 silencing (Figure 5a, lower panels). To verify whether this pattern occurred in vivo, we first examined by immunofluorescence microscopy sections of ^Ksp-creKif3a^f/-;Glis2^+/mut kidneys, using an antibody against p53 and noticed that numerous interstitial cells were positive for p53 (Figure 5b). For this reason, in order to test the status of p53 exclusively in tubular epithelial cells, we obtained primary tubular cell cultures from kidneys of ^Ksp-creKif3a^f/-;Glis2^+/mut, ^Ksp-creKif3a^f/-;Glis2^mut/mut, Glis2^mut/mut, and wild-type mice and assayed their p53 content by Western blot. p53 was undetectable in wild-type and ^Ksp-creKif3a^f/-;Glis2^+/mut lysates, but it was increased in ^Ksp-creKif3a^f/-;Glis2^mut/mut, Glis2^mut/mut mice (Figure 5c, upper panels). In the same way, the ratio of activated Chk1 to total Chk1 was low in ^Ksp-creKif3a^f/- primary cells, increased in Glis2^mut/mut cells, and highest in^Ksp-creKif3a^f/-;Glis2^mut/mut (Figure 5c, lower panels). These results indicate that Kif3a and Glis2 null kidney epithelial cells have defective and excessive activation of Chk1 and p53, respectively, and suggest that the decreased proliferation in ^Ksp-creKif3a^f/-;Glis2^mut/mut double mutants compared with ^Ksp-creKif3a^f/-;Glis2^+/mut mice is related to activation of p53.

Figure 5. Kif3a and Glis2 null kidney epithelial cells have defective and excessive activation of p53, respectively.

(a) Western blot of Kif3a^+/f and Kif3a^f/f immortalized tubular cells with and without concomitant silencing of Glis2 of total and Ser19 phosphorylated checkpoint kinase 2 (Chk2) (upper panels), total and Ser345 phosphorylated Chk1 (medium panels), and of p53 (lower panels). Chk2 is activated in Kif3a^f/f cells compared with Kif3a^+/f control cells independent of Glis2 silencing. Chk1 is phosphorylated in the absence of Glis2 in Kif3a^+/f and Kif3a^f/f cells. Stabilization of p53 is minimal in Kif3a^f/f cells and is strongly induced by concomitant Glis2 silencing. pSer19Chk2/Chk2 and pSer345Chk1/Chk1 densitometry ratios were calculated by digital analysis and are reported in correspondence of each lane. (b) Immunofluorescence microscopy image of Kif3a^f/-;Glis2^+/- kidney probed with an antibody against p53 showing accumulation of p53 in numerous interstitial cells (white arrows). (c) Western blot of lysates of primary tubular cells obtained from wild-type, Glis2^-/-, Kif3a^f/-;Glis2^+/-, and Kif3a^f/-;Glis2^-/- kidneys probed with an antibody against p53 (upper panels) and total and S345 phosphorylated Chk1 (lower panels). In the p53 Western blot, the dash corresponds to 50 knock-down (KD), the arrowhead indicates 53 KD, and the asterisk points to aspecific bands. The ratios of pSer345Chk1 (arrowhead) to Chk1 (lower panels) were calculated by digital analysis and are reported above each lane.

Kif3 null kidney epithelial cells present ectopic accumulation of cyclin B1 and genomic instability

An important effect of p53 activation on the cell cycle is mediated by repressing the transcription and stability of cyclin B1.^25,26 The cellular concentration of cyclin B1 increases during the G2 phase of the cell cycle and abruptly falls after its ubiquitination by anaphase promoting complex (APC) at the onset of mitosis. We determined the levels of cyclin B1 by flow cytometry using a specific fluorescein isothiocyanate-conjugated antibody against cyclin B1 in ^Ksp-creKif3a^f/f and ^Ksp-creKif3a^+/f cells, with or without concomitant Glis2 shRNA-mediated silencing. Cyclin B1 expression was higher in ^Ksp-creKif3a^f/f cells than in heterozygous control cells and was reduced in both cell lines after shRNA-mediated inactivation of Glis2 (Figure 6a and b and Supplementary Figure S6). In addition, we noticed that the distribution of cyclin B1 content along the cell cycle was also abnormal, being present not only in G2 but also in S and late G1 phases (Figure 6a, upper left panel, and Supplementary Figure S6), suggesting the possibility of premature mitosis in ^Ksp-creKif3a^f/f cells. To test this hypothesis, we probed ^Ksp-creKif3a^f/f cells and ^Ksp-creKif3a^+/f control cells with an antibody against the phosphorylated threonine 187 of p27^Kip1 (^T187p27^Kip1). T187 phosphorylation of p27^Kip1 occurs only in late G1, when it triggers p27^Kip1 ubiquitination, allowing cells to exit G1 and progress into the S phase of the cell cycle, and it is thus a marker of early phases of the cell cycle.²⁷ Since ^T187p27^Kip1 is only found in late G1 phase and not in premitotic cells, we tested whether metaphase ^Ksp-creKif3a^f/f and ^Ksp-creKif3a^+/f cells, recognized by the presence of condensed chromosomes, were positive for ^T187p27^Kip1 (Figure 6c). Seven of 7 ^Ksp-creKif3a^f/f mitotic cells and 1 of 12 ^Ksp-creKif3a^+/f hemizygous cells were positive for ^T187p27^Kip1 (P < 0.001 by χ-square statistics, n = 3 slides), indicating that ^Ksp-creKif3a^f/f cells have an uncoordinated cell cycle that leads to anticipated chromosome condensation and mitosis in the presence of still uncompleted DNA duplication. Such a cell cycle defect is consistent with the reduced DNA amount of the ^Ksp-creKif3a^f/f cells in G2/M (Figure 4a) and it is expected to be associated with centrosome aberrations and polyploidy. In fact, we frequently detected polyploidy by flow cytometry (Figure 6d) by immunofluorescence confocal microscopy in ^Ksp-creKif3a^f/f but not in ^Ksp-creKif3a^+/f immortalized epithelial cells (Figure 6e). This was also associated with centrosomal fragmentation (86% of ^Ksp-creKif3a^f/f and 4% ^Ksp-creKif3a^+/f, P < 0.0001 by χ-square statistics, n = 3 slides), as previously reported for Kif3a null fibroblasts²⁸ (Figure 6e). Finally, we also observed polyploid cells and megakarya by bright field microscopy in hematoxylin and eosin–stained kidney sections of ^Ksp-creKif3a^f/- mice, but not in ^Ksp-creKif3a^f/-;Glis2^mut/mut double knockouts (n = 3 mice) (Figure 6f). These results suggest that the cell cycle abnormality in the absence of Kif3a results in repeated DNA endoreduplication, DNA damage, and apoptosis in ^Ksp-creKif3a^f/- kidneys, and that Glis2 silencing-dependent activation of p53 limits these defects.

Figure 6. Kif3 null kidney epithelial cells present ectopic accumulation of cyclin B1, premature mitosis, and centrosome amplification.

(a) Flow cytometry of cyclin B1 in Kif3a^f/f and Kif3a^+/f immortalized tubular cells stably transfected with nontargeting short hairpin RNA (nt-shRNA) or with Glis2-targeting shRNA (clones Glis2^KD-5.1, Glis2^KD-7.2, Glis2^KD-5.2, and Glis2^KD-7.1; KD indicates knock-down), as performed in Figure 2a. Gates were established based on background fluorescence (nonimmune isotype antibody) for each sample. (b) Quantification of cyclin B1 content obtained from 4 independent replicates of the previous experiment. Values on the y-axis express percentage of positive cells for each cell group. Results are mean ± SEM. P values were obtained by Student t-test. (c) Immunofluorescence confocal microscopy images of Kif3a^f/f and Kif3a^+/f cells obtained with an antibody against phosphorylated threonine 87 of p27 (green channel) and the centromere marker Aurora kinase B. Condensed chromosomes with kinetocores (blue and red channels) indicate that the cells are in metaphase. Bars = 5 μm. (d) Flow cytometry cell cycle analysis of Kif3af/f immortalized tubular cells. Cell aggregates were not gated out for the analysis, to display polyploid cells with >4n DNA content. (e) Immunofluorescence microscopy images of Kif3a^f/f and Kif3a^+/f cells obtained with antibodies against the centrosomal marker γ–tubulin (red channel) and α-tubulin (green channel). DNA was stained with 4′,6-diamidino-2-phenylindole (DAPI) (blue channel). Arrows indicate fragmented (left panel) and normal centrosomes (right panel) in Kif3a^f/f and Kif3a^+/f cells. An abnormal megakaryon is also visible in the Kif3a^f/f cell (delineated by the arrowheads). Bars = 5 μm. (f) Representative images of Kif3a^f/-, Kif3a^f/-;Glis2^-/-, and Kif3a^+/f kidneys (hematoxylin and eosin). Arrows indicate tubular cells with megakarya or polyploidy. A polyploid cell is also highlighted at higher magnification in the inset (left panel). No nuclear abnormalities are apparent in Kif3a^f/-;Glis2^-/-, and Kif3a^+/f kidneys. Bars = 20 μm.

Pharmacological stabilization of p53 reduces cyst growth in ^Ksp-creKif3a^f/- kidneys

To further substantiate the role of p53 stabilization in partially reducing cyst growth in the ^Ksp-creKif3a^f/-;Glis2^mut/mut double mutants, we treated 10- to 12-day-old ^Ksp-creKif3a^f/- mice with the small-molecule nutlin-3, an inhibitor of MDM2,²⁹ at the dose of 40 μg/g on alternated days for 10 days.³⁰ MDM2 is an E3 ubiquitin protein ligase and is a major inducer of p53 degradation.³¹ Compared with vehicle-treated control mice of the same age, nutlin-3 treatment resulted in decreased kidney–total body weight ratio, cyst area over total kidney area, average cyst size, and improved renal function (Figure 7a and b and Supplementary Figure S7). Similar to what was observed in ^Ksp-creKif3a^f/-;Glis2^mut/mut double mutants, kidneys of nutlin-3-treated mice also showed reduction of proliferation and apoptosis of tubular cells (Figure 7c and d). These results indicate that stabilization of p53 is sufficient to slow cystic progression and prevent cell proliferation and apoptosis in Kif3a null kidneys.

Figure 7. Pharmacological stabilization of p53 reduces cyst growth in Kif3a^f/-;Glis2^+/- kidneys.

(a) Kidneys of Kif3a^f/-;Glis2^+/- treated with vehicle or with nutlin-3 for 10 days (hematoxylin and eosin). Kidneys of Kif3a^f/-;Glis2^+/- nutlin-3-treated mice are smaller, have fewer and smaller cysts, and have better preserved parenchyma than vehicle-treated control kidneys. Bars = 1000 μm. (b) Quantification of kidney weight to body weight ratio (KW/BW, g/g), cyst area (expressed as percentage of cyst area over total kidney area), average cyst size, and renal function of Kif3a^f/-;Glis2^+/- mice treated with vehicle or with nutlin-3. Serum creatinine (sCr) and blood urea nitrogen (BUN) are expressed in mg/dl. The number of mice in each experimental group is reported on the x-axis. Results correspond to mean ± SD. P values were obtained by Student t-test. (c) Representative immunofluorescence microscopy images of sections of Kif3a^f/-;Glis2^+/- mice treated with vehicle or with nutlin-3, probed with an antibody against Ki67 and activated caspase 3 (aCasp3) to detect cycling and apoptotic cells, respectively. Bars = 50 μm. (d) Quantification of the previous experiment. The number of mice in each experimental group is reported on the x-axis. Values express percentage of positive cells (10 consecutive optical fields per mouse). Results are mean ± SD. P values were obtained by Student t-test.

Loss of Glis2 induces cell senescence in kidney epithelial cells

Cell cycle arrest after p53 stabilization allows cells to repair damaged DNA. In cases of irreparable damage, cells may either undergo apoptosis or, alternatively, permanently exit the cell cycle and become senescent. Senescence can be mediated by p53 and is actuated through epigenetic silencing of genes implicated in cell proliferation, and results in condensed nuclear areas called senescence-associated heterochromatin foci.^32-34 To test for signs of senescence in ^Ksp-creKif3a^f/f immortalized kidney epithelial cells, we quantified the number of senescence-associated heterochromatin foci using an antibody against tri-methylated lysine 9 of histone 3 (H3K9me3), a marker of heterochromatin found in senescent cells.^{35 Ksp-cre}Kif3a^f/f cells had significantly less foci than the hemizygous ^Ksp-creKif3a^+/f control cells, whereas the number of foci was significantly increased in both ^Ksp-creKif3a^f/f and ^Ksp-creKif3a^+/f cells after stable shRNA-mediated silencing of Glis2 (Figure 8a and b). Western blots also showed increased amount of H3K9me3 in lysates of both ^Ksp-creKif3a^f/f and ^Ksp-creKif3a^+/f cells after Glis2 silencing, but not in cells infected with nontargeting shRNA (Figure 8c). In vivo, we found more H3K9me3-positive cells in kidneys of ^Ksp-creKif3a^f/-;Glis2^mut/mut mice than in ^Ksp-creKif3a^f/-;Glis2^+/mut mice (Figure 8d and e), suggesting that genetic inactivation of Glis2 induces cell cycle arrest by triggering cellular senescence in ^Ksp-creKif3a^f/- knockout kidneys. To further test this possibility, we detected the activity of the senescence-associated lysosomal β-galactosidase, a marker that is widely used to identify senescent cells,³⁶ on fresh kidney sections. Concordant with H3K9me3 content, senescence-associated lysosomal β-galactosidase activity was virtually absent in ^Ksp-creKif3a^f/-;Glis2^+/mut kidneys but was clearly detected in ^Ksp-creKif3a^f/-;Glis2^mut/mut double mutants and more intensely in Glis2^mut/mut kidneys (Figure 8f). Finally, p16, another marker of senescent cells, was upregulated both at RNA and protein level in primary cultures of proximal tubular cells of Glis2^mut/mut knockouts and of ^Ksp-creKif3a^f/-;Glis2^mut/mut double mutants but not in ^Ksp-creKif3a^f/-;Glis2^+/mut and wild-type kidneys (Figure 8g and h and Supplementary Figure S8). Collectively, these results indicate that genetic inactivation of Glis2 induces senescence in both wild-type and Kif3a null kidney tubular cells.

Figure 8. Loss of Glis2 induces cellular senescence in kidney epithelial cells.

(a) Representative immunofluorescence confocal microscopy images of Kif3a^f/f and Kif3a^+/f immortalized tubular cells with and without concomitant silencing of Glis2, obtained using an antibody against H3K9me3. Bars = 20 μm. (b) Quantification of the previous experiment, 3 independent samples per group (10 optical fields per sample, 4 stacks of 2 μm per field). Error bars are SD. P values were obtained by Student t-test. (c) Western blots of Kif3a^f/f and Kif3a^+/f immortalized tubular cells with and without concomitant silencing of Glis2, probed with an antibody against H3K9me3. H3K9me3 is increased in both cell lines after Glis2 silencing. (d) Representative immunofluorescence microscopy images of kidney sections of Kif3a^f/-;Glis2^+/-, and Kif3a^f/-;Glis2^-/- double mutant kidneys probed with an antibody against H3K9me3 at lower (upper panels, bars = 50 μm) and higher (lower panels, bars = 20 μm) magnification. (e) Quantification of H3K9me3 positive tubular cells in vivo. Ratio of positive cells was obtained dividing the number of positive nuclei by the total number of nuclei in 5 consecutive optical fields. n = 3 mice per group. Results are mean ± SEM. P values were obtained by Student t-test. (f) Bright field microscopy images of kidney sections of Kif3a^+/f;Glis2^+/+, Kif3a^f/-;Glis2^+/-, Kif3a^f/-;Glis2^-/-, and Kif3a^+/;Glis2^-/- mice after 5-bromo-4-chloro-3-indolyl β-D-galactosidase (X-Gal) staining. Senescence-associated lysosomal β-galactosidase (SA-β-gal) activity (blue) was virtually absent in Kif3a^f/-;Glis2^+/- kidneys but was clearly detected in Kif3a^f/-;Glis2^-/- double mutants and more intensely in Kif3a^+/f;Glis2^-/- kidneys. Bars = 50 μm. (g) Quantification by real-time polymerase chain reaction of p16 expression in primary tubular cells from Kif3a^f/-;Glis2^+/-, and Kif3a^f/-;Glis2^-/-, Kif3a^+/f;Glis2^-/-, and wild-type (WT) mice. Error bars are calculated from 3 experimental replicates and are SEM. (h) Western blot of lysates of primary tubular cells obtained from wild-type, Glis2^-/-, Kif3a^f/-;Glis2^+/-, and Kif3a^f/-;Glis2^-/- kidneys, probed with an antibody against p16. The p16 protein levels are increased in Kif3a^f/-;Glis2^-/- and Glis2^-/-, compared with Kif3a^f/-;Glis2^+/- and wild-type cells, respectively.

DISCUSSION

Tubule dilation and cyst formation, accompanied by a certain degree of tubular cell proliferation and apoptosis, are characteristics shared by most, if not all, cystic kidney diseases, including NPHP and polycystic kidney disease. Nonetheless, although pronounced cyst growth is prevalent in some forms of cystic kidney disease, such as polycystic kidney disease, progressive organ atrophy and fibrosis are typical features of NPHP. In this study we have examined the ^Ksp-creKif3a^f/- kidney-specific knockout mouse, in which kidneys are characterized by rapid cyst growth, and the Glis2^mut/mut mouse model of NPHP type 7, which reproduces virtually all the features of human NPHP, including diffuse tubular atrophy, interstitial inflammatory infiltration, fibrosis, and progressive decrease in kidney size.^5,14,16 The data that we have presented indicate that 2 opposite molecular mechanisms underlie the phenotype in these 2 mouse models of cystic kidney disease. On one side, we unexpectedly found that the kinesin Kif3a is required for regulation of the cell cycle, activation of Chk1, stabilization of p53, and cell cycle arrest, which are all compromised in its absence. Loss of Glis2, on the other hand, induces sustained Chk1 activation, execution of the DNA damage response and cell senescence, which is the principal cause of the NPHP phenotype in the Glis2 knockout mouse. The exact mechanism by which senescence is evoked in Glis2 null cells is still unclear, but the abnormal activation of Chk1 that we have observed, is suggestive that replicative stress during the S phase of the cell cycle may be the primary cause of senescence in these cells.³⁷ Interestingly, other ciliary proteins associated with ciliopathy phenotypes have recently been shown to be required for correct activation and functioning of the pathways that control the cellular response to DNA damage,^10-12,38-41 suggesting that permanent cell cycle arrest and senescence could be a common feature of these diseases. In this regard, it is worth noticing that expression of many secreted factors (extracellular matrix proteases, matrix components, chemokines, and cytokines), recognized as part of the senescence secretory–associated phenotype⁴² are upregulated in Glis2 null kidneys.^5,14 This correlation provides a possible explanation of the progressive kidney interstitial inflammation typical of NPHP. Irrespective of the upstream mechanisms of Chk1 activation, it is very likely that the consequent stabilization of p53 is central in determining the actuation of senescence in Glis2 null cells. The strong Chk1 activation and p53 stabilization in the absence of Glis2 is also likely to drive the partial rescue of the Kif3a knockout mice cystic phenotype. Our experiments suggest that the absence of Kif3a extensively perturbs the cell cycle, with consequent genomic instability and DNA damage, but at the same time results in defective activation of Chk1 and DNA repair. Loss of Glis2, instead, by inducing Chk1-mediated DNA damage response, stabilizes p53 with consequent decreased proliferation and DNA damage in the Kif3a null cells. This model would account for the apparent paradox that loss of Glis2 increases DNA damage in Kif3a hemizygous cells but has opposite effects in Kif3a null cells that have higher levels of DNA damage. Less clear is why loss of Glis2 does not seem to significantly affect the cell cycle of Kif3a hemizygous immortalized tubular cells, at least in vitro. One possibility is that the coexistence of other mechanisms, such as the epithelial-to-mesenchymal transition that is characteristic of the Glis2 knockout cells,^5,13 may prevail on the cell cycle arrest in Kif3a hemizygous cells. Another question that will need to be addressed is whether the unrestrained cell cycle and the defective activation of Chk1 in Kif3a null cells is consequent to loss of cilia or centrosome-related functions of this kinesin, considering that cilia have been widely related to the cell cycle and many of the proteins that regulate DNA damage response (such as several cyclins and cyclindependent kinases, p53, Chk1, and Chk2) are found in the centrosomes.^43-45 Understanding this point may help to better define the relationship between the ciliary dependent procystogenic signal that is activated in the absence of polycystins and the one that acts in the Kif3a knockouts and is dependent on loss of primary cilia.⁴⁶ It is interesting, in any case, that anomalies of DNA damage and repair, defective activation of p53, and genomic instability seem to be common to multiple cystic kidney diseases, including autosomal dominant polycystic kidney disease, pointing to an association between hyperproliferation and cell cycle dysregulation in these diseases.^47-49 In conclusion, we have shown that Kif3a null cells have an accelerated cell cycle, impaired stabilization of p53, and an inability to trigger DNA damage response–dependent cell cycle arrest. Contrarily, loss of Glis2 results in excessive activation of Chk1 and cell senescence, which is a central feature in NPHP type 7 and one that partially rescues the hyperproliferation of Kif3 defective cells and the cystic phenotype of Kif3a knockout kidneys.

MATERIALS AND METHODS

Animal studies

^Ksp-creKif^f/- mice express Cre recombinase in kidney epithelial cells and were previously described.¹⁶ The phenotypes of ^Ksp-creKif^f/-;Glis2^+/mut and ^Ksp-creKif^f/-;Glis2^+/mut mice were compared with those of ^Ksp-creKif^f/-;Glis2^mut/mut double knockouts and ^Ksp-creKif^f/+;Glis2^mut/mut mice. For pharmacological stabilization of p53, the small molecule nutlin-3 was diluted in dimethylsulfoxide and injected intraperitoneally on alternate days for 2 weeks starting from P10 to P12 at the dose of 40 μg/g. All animal studies were approved by the Institutional Animal Care and Use Committees from the University of Texas Southwestern Medical Center.

Blood urea nitrogen and serum creatinine

Blood samples were obtained by periorbital capillary puncture. Creatinine concentrations were measured using a P/ACE MDQ Capillary Electrophoresis System (Beckman Coulter, Brea, CA) at 214 nm.⁵⁰ Blood urea nitrogen measurements were obtained using the Vitros 250 chemistry analyzer (GMI Inc., Ramsey, MN).

Digital image analysis

Whole kidneys stained with hematoxylin and eosin were imaged using Zeiss Axioscan Z1 microscope (Jena, Germany). Quantification of cyst number and cyst area were obtained by digital elaboration of microscope images using the ImageJ software (National Institutes of Health, Bethesda, MD; http://rsb.info.nih.gov/ij/).

Cell cultures

Kif3a^f/f and Kif3a^f/+ cells obtained by SV40 T-mediated immortalization as previously described⁸ were grown at permissive temperature (33 °C) in the presence of 0.1 unit/ml interferon-γ and allowed to differentiate at 37 °C without interferon for at least 3 days before the experiment.

Primary tubular cell cultures

Primary tubular cells were isolated from kidneys of experimental mice (21–30 days old) after digestion of cortex fragments with collagenase and cultured under sterile conditions, according to a previously described method.⁵¹ After 7 days, cell cultures formed a confluent monolayer and were used for the experiments.

Glis2 silencing by lentiviral shRNA-mediated vector cell infection

Glis2-targeting shRNA vectors packed for lentiviral infection were used to create stable cell clones as previously described.¹³ Clones were named based on the targeting shRNA used (for example, Kif3a^f/f;Glis2^KD-5.1 corresponds to clone 1 obtained by infecting Kif3a^f/f with the shRNA construct 5; Kif3a^f/f;Glis2^KD-7.2 corresponds to clone 2 obtained by infecting Kif3a^f/f with the shRNA construct 7; and so on). Clones were chosen based on the efficiency of the shRNA constructs.

Antibodies

Beta actin (A3854), alpha-tubulin (T6074), gamma-tubulin (SAB503045), and antiacetylated α-tubulin (6-11B-1) were purchased from Sigma (St. Louis, MO). Cleaved caspase-3 (AB3623MI) was from Fisher Scientific (Thermo Fisher Scientific, Waltham, MA). Ser139 phospho-histone H2A (9178) Chk1 (2360), Chk2 (2662) Ser345 phospho-Chk1 (2348), Ser19 phpspho-Chk2 (2666), and Ser10 pHis3 fluorescently conjugated pHis3 (9708) were from Cell Signaling (Danvers, MA). Ki67 (ab15580), p53 (ab26), H3K9Me3 (ab8898), and p16 (51243) were from Abcam (Cambridge, MA). Kip1 (phospho Thr187) antibody was from GeneTex (Irvine, CA). Anti-Aurora kinase B was purchased from BD Biosciences (611082; San Jose, CA), and the Glis2 polyclonal antibody was described previously.¹³

MTT proliferation assay

Life technologies stock solution was prepared using sterile phosphate-buffered saline (PBS; 5 mg MTT/ml). Cells were seeded in 96-well plates and allowed to attach for 24 hours and fresh medium was supplemented with 100 μl of MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) was added to each well. Cells were then incubated in the dark at 37 °C for 2 hours, medium with MTT was removed, and 100 μl of dimethylsulfoxide per well was added to stop the reaction. Absorbance was read at λ = 540 nM using an automatic plate reader.

Senescence-associated β-galactosidase staining

Fresh PBS perfused unfixed kidneys were obtained from experimental mice. Samples were placed in optimal cutting temperature compound on top of dry ice. Then, 10-μm cryosections were washed in PBS, fixed in 4% formaldehyde, and then incubated at 37 °C for 12 to 16 hours with fresh 5-bromo-4-chloro-3-indolyl β-D-galactosidase (X-Gal, 1 mg of per ml) solution (20 mg of dimethylformamide per ml, 40 mM citric acid/sodium phosphate pH 6.0, 5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, 150 mM NaCl₂ mM MgCl₂).

Immunofluorescence

Tissues were collected after perfusion with PBS and 4% paraformaldehyde, then fixed in 4% paraformaldehyde for 2 hours on ice, left in a solution of 30% sucrose in PBS at 4 °C overnight, and embedded in optimal cutting temperature compound. Tissue sections were incubated in a solution of 0.1% sodium borohydride (NaBH₄) for 30 minutes and incubated in blocking solution (10% goat serum, 0.1% bovine serum albumin [BSA] in PBS) for 1 hour at room temperature. Tissues were incubated overnight at 4 °C with primary antibody diluted in blocking solution and with fluorescently labeled secondary antibody for 1 hour at room temperature. Samples were mounted with ProLong (Life Technologies, Thermo Fisher Scientific) with 4′,6-diamidino-2-phenylindole. Images were acquired using a Zeiss Axioplan 2 deconvolution microscope or a Zeiss LSM 510 confocal microscope with constant acquisition parameters between samples and control cells.

Flow cytometry

Cell cycle analysis

Immortalized cells were allowed to differentiate for 3 days in the absence of γ-interferon at 37 °C, then they were replated and harvested at 50% confluence and fixed with 70% cold ethanol and stained with propidium iodide solution (50 μg/ml, Sigma). Cell cycle distribution was obtained by flow cytometry using a FACS Calibur dual-laser instrument (BD Biosciences). The software FlowJo (FlowJo, Ashland, OR) was used for the data analysis.

Double thymidine block

Cells were synchronized in the G1 phase of the cell cycle by adding thymidine (2.5 mM) to the culture medium for 16 hours in the absence of γ-interferon at 37 °C, allowed to grow in thymidine-free medium for 12 hours, and reexposed to thymidine for an additional 16 hours. An aliquot of cells (time 0) was fixed and used to record synchronization by flow cytometry. The remaining cells were allowed to grow and were harvested at 3, 6, and 8 hours in thymidine-free medium. Cells’ DNA was then stained by propidium iodide and the fraction of cells that progressed to the S phase was evaluated by flow cytometry analysis. The numbers here correspond to the ratio between the percentage of cells in G1 after thymidine washout at 3, 6, and 8 hours over the number of cells in G1 before washouts (time 0), multiplied by 100 ( $\mathrm{G}1\mathrm{ex}.=\frac{\mathrm{G}1.t0-\mathrm{G}1.t\mathrm{n}}{\mathrm{G}1.t0}•100$).

Nocodazole block

Cells were first synchronized in the G1 phase by adding thymidine (2.5 mM) to the culture medium for 16 hours and grown in thymidine-free medium for 8 hours; then they were synchronized in late G2 phase by growing them in the presence of nocodazole (100 ng/ml) for 20 hours. After this time, all cells were collected, an aliquot was used as a control for achieved synchronization and the remaining cells were washed in PBS and grown in fresh medium without nocodazole for 6 hours. Cells’ DNA was then stained by propidium iodide and analysis for DNA content was performed by flow cytometry. The percentage of G2 exiting cells was expressed by cell numbers in G2 at 6 hours divided by the cell numbers in G2 at time 0.⁵²

Cumulative BrdU incorporation assay

Cells were seeded at 37 °C in the absence of γ-interferon at low density and allowed to grow for 16 hours. BrdU (10 μg/ml, Sigma) was added to a final concentration of 20 μg/ml. Cells were harvested at 2, 4, and 6 hours; fixed in 4% paraformaldehyde; and incubated with anti-BrdU fluorescently labeled antibody (Abcam) and propidium iodide. BrdU incorporation was determined by flow cytometry, data analysis was performed using the software FlowJo.

Cyclin B1 analysis

Immortalized cells were allowed to differentiate for 3 days in the absence of γ-interferon at 37 °C, then they were replated and harvested at 50% confluence and fixed with 70% cold ethanol. Cells were permeabilized in 0.25% Triton X-100 in PBS, and then resuspended in 100 μl of staining buffer (1% BSA in PBS) containing 0.5 μg of fluorescein isothiocyanate–conjugated cyclin B1 antibody (554108; BD Pharmingen, San Diego, CA) or isotype antibody for 1 hour at room temperature, followed by propidium iodide staining.

pHis3 expression bivariate analysis

Immortalized cells were allowed to differentiate for 3 days in the absence of γ-interferon at 37 °C, and then they were replated and treated with nocodazole (50 ng/ml) for 6 hours. After fixation with ice-cold 70% ethanol, cells were permeabilized with 0.25% Triton X-100/PBS for 15 minutes on ice, blocked with 2% BSA/PBS for 15 minutes, and incubated with anti-pHis3 antibody for 2 hours. Cells were then washed 3 times in PBS and stained with propidium iodide (50 μg/ml, Sigma). Cell cycle distribution was obtained by flow cytometry using a FACS Calibur dual-laser instrument (BD Biosciences). FlowJo software was used for the data analysis.

Quantitative real-time PCR

Total RNA was isolated using TRIzol (Invitrogen, Thermo Fisher Scientific) and purified with Qiagen RNeasy Mini Kit (Alameda, CA). First-strand reverse transcription reactions were performed on 1 μg of total RNA using the ThermoScript RT-PCR Kit (Invitrogen). Real-time polymerase chain reactions (PCRs) were performed using iQ SYBR Green Supermix (Bio-Rad, Berkeley, CA). The following real-time PCR primers were used: p16-Fw CGTACCCCGATTCAGGTGAT; p16-Rev:TTGAGCAGAAGAGCTGCTACGT; betaactin Fw: GGCTGTATTCCCCTCCATCG; beta-actin Rev CCAGTTGGTAACAATGCCATGT. All real-time PCR experiments were performed in triplicates.

Sodium dodecylsulfate–polyacrylamide gel electrophoresis/Western blotting

Protein lysates were mixed with 4× Laemmli sample buffer (161- 0737; Bio-Rad) containing 100 mm dithiothreitol and denatured at 95 °C for 10 minutes. Samples were run on polyacrylamide gels and transferred to polyvinylidene difluoride membranes (LC2002, Thermo Fisher Scientific, Florence, KY). Membranes were blocked in 5% BSA, probed with primary antibody in 1% BSA for 2 hours at room temperature or overnight at 4 °C and with secondary antibody in 1% BSA for 1 hour at room temperature. Antibody binding was visualized with luminol reagent (sc-2048; Santa Cruz Biotechnology, Santa Cruz, CA). When necessary, blots were stripped with 0.1 m tris–hydrochloride, pH 7.8; 10% sodium dodecylsulfate; and 0.70% β-mercaptoethanol for 20 to 30 minutes at 50 °C before reprobing.

Supplementary Material

Supplements

Figure S1. Images of kidneys of the paraffin-embedded Kif3a^f/-;Glis2^+/-, Kif3a^f/-;Glis2^-/-, and Kif3a^+/f;Glis2^-/- mice used in this work. Hematoxylin and eosin. Bars = 500 μm.

Figure S2. (A) Immunofluorescence microscopy images of Kif3a^f/-;Glis2^+/-, Kif3a^f/-;Glis2^-/- and Glis2^-/- mice kidney sections. An antibody against acetylated α-tubulin was used to identify primary cilia and 4′,6-diamidino-2-phenylindole (DAPI) to mark nuclei of tubular epithelial cells. Boxed areas are magnified in the insets at the upper-right corner of each image. Cilia are absent in Kif3a^f/-;Glis2^+/-, Kif3a^f/-;Glis2^-/- double knockouts and clearly visible in Glis2^-/- tubular cells (arrows). Bars = 20 μm. (B) Expression of Gli1, in Kif3a^f/f compared with Kif3a^+/f immortalized cells measured by quantitative real-time polymerase chain reaction (log scale). Gli1 expression is significantly increased in Kif3 knockout cells. Error bars are SEM (triplicate samples).

Figure S3. (A) Western blot of Kif3a^+/f and Kif3a^f/f immortalized tubular cells with and without concomitant silencing of Glis2, probed with an antibody against Glis2. Glis2 protein expression is comparable between Kif3a^+/f and Kif3a^f/f cells and is decreased in all 4 Glis2-silenced clones. (B) Representative immunofluorescence confocal microscopy images of Kif3a^+/f and Kif3a^f/f immortalized tubular cells obtained using an antibody against Glis2. Glis2 subcellular localization is not affected by the absence of cilia. Bars = 20 μm. nt-shRNA, nontargeting short hairpin RNA.

Figure S4. Representative images of flow cytometry cell cycle analysis of Kif3a^f/f and Kif3a^+/f immortalized tubular cells with and without concomitant silencing of Glis2 before release of the nocodazole block, showing the synchronization of the cells in G2/M and progression in the cell cycle after release. DNA content is represented on the x-axis; cell counts are shown on the y-axis. nt-shRNA, nontargeting short hairpin RNA.

Figure S5. (A) Flow cytometry bivariate analysis of Kif3a^f/f and Kif3a^+/f cells in mitosis, using an antibody against phosphorylated histone H3 (pHis3). Samples were enriched of mitotic cells by being exposed to nocodazole for 6 hours. Numbers are the percentage of pHis3- positive cells in G2/M. (B) Graphic reporting the statistics of a triplicate experiment for each experimental group. Results are mean ± SD. P value was obtained by Student t-test. nt-shRNA, nontargeting short hairpin RNA.

Figure S6. (A) Cell cycle distribution of Kif3a^+/f and Kif3a^f/f immortalized tubular cells. Cells in G1, S, and G2/M phases of the cell cycle were gated as indicated to determine the cyclin B1 distribution in each phase of the cell cycle. DNA content is represented on the x-axis; cell counts are shown on the y-axis. (B) Cumulative histograms of cyclin B1 concentration (red) compared with the isotype (blue) in the 3 phases of the cells cycle of Kif3a^+/f and Kif3a^f/f immortalized tubular cells. (C) Cumulative histograms of the cyclin B1 distribution of Kif3a^f/f (red) compared with Kif3a^+/f (blue) immortalized tubular cells in the 3 phases of the cells cycle. Immunofluorescence intensity is represented on the x-axis; cell counts are shown on the y-axis.

Figure S7. Images of paraffin-embedded kidneys of Kif3a^f/- mice treated with the MDM2 inhibitor nutlin-3 or with dimethylsulfoxide (DMSO) as a control. Hematoxylin and eosin. Bars = 500 μm.

Figure S8. Western blots of lysates of primary tubular cells obtained from wild-type (WT), Glis2^-/-, Kif3a^f/-;Glis2^+/-, and Kif3a^f/-;Glis2^-/- kidneys, probed with an antibody against p16. The p16 protein levels are increased in Kif3a^f/-;Glis2^-/- compared with Kif3a^f/-;Glis2^+/- and less markedly in Glis2^-/- compared with wild-type cells. Numbers report p16–β-actin density ratio determined by digital image analysis above each panel and the average result from all the panels at the bottom of the figure. Lysates in panels 1 and 2 (from top to bottom) were obtained from 2 different primary culture plates of kidney cells of the same experimental mice, as well as lysates in panels 4 and 5. n=3 mice.