Rapamycin treatment dose‐dependently improves the cystic kidney in a new ADPKD mouse model via the mTORC1 and cell‐cycle‐associated CDK1/cyclin axis

Abstract Although translational research into autosomal dominant polycystic kidney disease (ADPKD) and its pathogenesis has made considerable progress, there is presently lack of standardized animal model for preclinical trials. In this study, we developed an orthologous mouse model of human ADPKD by cross‐mating Pkd2 conditional‐knockout mice (Pkd2 f3) to Cre transgenic mice in which Cre is driven by a spectrum of kidney‐related promoters. By systematically characterizing the mouse model, we found that Pkd2 f3/f3 mice with a Cre transgene driven by the mouse villin‐1 promoter (Vil‐Cre;Pkd2 f3/f3) develop overt cysts in the kidney, liver and pancreas and die of end‐stage renal disease (ESRD) at 4–6 months of age. To determine whether these Vil‐Cre;Pkd2 f3/f3 mice were suitable for preclinical trials, we treated the mice with the high‐dose mammalian target of rapamycin (mTOR) inhibitor rapamycin. High‐dose rapamycin significantly increased the lifespan, lowered the cystic index and kidney/body weight ratio and improved renal function in Vil‐Cre;Pkd2 f3/f3 mice in a time‐ and dose‐dependent manner. In addition, we further found that rapamycin arrested aberrant epithelial‐cell proliferation in the ADPKD kidney by down‐regulating the cell‐cycle‐associated cyclin‐dependent kinase 1 (CDK1) and cyclins, namely cyclin A, cyclin B, cyclin D1 and cyclin E, demonstrating a direct link between mTOR signalling changes and the polycystin‐2 dysfunction in cystogenesis. Our newly developed ADPKD model provides a practical platform for translating in vivo preclinical results into ADPKD therapies. The newly defined molecular mechanism by which rapamycin suppresses proliferation via inhibiting abnormally elevated CDK1 and cyclins offers clues to new molecular targets for ADPKD treatment.


Introduction
Autosomal dominant polycystic kidney disease (ADPKD) is the most common of a group of inherited kidney disorders characterized by progressive cyst development and various extrarenal manifestations [1,2]. ADPKD, which has a worldwide prevalence estimated from 1:1000 to 1:4000, is the fourth most common single cause of endstage renal failure worldwide [3][4][5].
There has been considerable progress in elucidating the molecular mechanisms and pathogenesis of ADPKD [3,6,[22][23][24]. However, there is no standardized animal model of ADPKD, and some non-standardized ADPKD models are not systemically characterized. In addition, although several preclinical and clinical trials have reported that rapamycin has a beneficial effect on ADPKD [25,26], no direct link between polycystin dysfunction and mTOR signalling changes or ADPKD cystogenesis has been established. In this study, we developed and well characterized a newly established ADPKD model and demonstrated that rapamycin inhibits cystic progression in the ADPKD kidney by down-regulating the cell-cycle-associated CDK1 and cyclins (cyclin A, cyclin B, cyclin D1 and cyclin E), thereby arresting aberrant proliferation of the renal epithelia. The study not only generated a well-characterized, standardized mouse model for ADPKD, but also revealed the involvement of the mTORC1-CDK1/cyclin axis in ADPKD, which leads to new molecular targets for treating this disease.

Materials and methods Animals
This study was conducted using Vil-Cre;Pkd2 f3/f3 mice and their normal Pkd2 f3/f3 littermates (control). All of the mouse models used in this study were from a C57BL/6J inbred background. Pkd2 f3/f3 mice [27] were bred with various Cre transgenic mice (Table 1) to generate Pkd2 conditional-knockout mice. Mouse genotyping and rapamycin treatment are detailed in Supplementary Methods A in Appendix S1.

Cell lines, antibodies and reagents
The null-Pkd2 (E8) cell line and its maternally derived Pkd2 heterozygous (D3) cell line were described in our previous study [27]. Antibodies and reagents are listed in Supplementary Methods B in Appendix S1.

Microarray analysis, quantitative PCR, histology, immunofluorescence (IF) and Western blots
Histology and IF staining procedures were as previously described [27][28][29]. All other assays are described in Supplementary Methods C in Appendix S1.
Statistics and measurements of cystic index, proliferation, apoptosis, blood urea nitrogen (BUN) and creatinine (Cr) Statistics and detailed procedures of cystic index calculation, proliferation and apoptosis are described in Supplementary Methods D in Appendix S1.

Results
Development of a Pkd2 mouse model that mimics human ADPKD Standardized animal models for ADPKD would contribute greatly to the investigation of polycystin function, cystic pathogenesis and novel therapeutic compounds for treating ADPKD. Although we recently generated Pkd2 conditional-knockout mice (Pkd2 f3/f3 ) [27], we still needed to find an appropriate Cre-recombination system by which the Pkd2 f3/f3 model could be used to represent the clinical manifestations of human ADPKD and to assess the effect of various therapeutic interventions. By cross-mating Pkd2 f3/f3 mice with Cre transgenic mice driven by various kidney-related promoters, including cGt-Cre, Ksp-Cre, Nestin-Cre and Vil-Cre mice (Table 1), we found that the Vil-Cre; Pkd2 f3/f3 mice developed disease phenotypes that were similar to human ADPKD (Fig. 1). Cre-recombinase expressed in the Vil-Cre transgenic mouse under the control of the mouse villin-1 promoter is turned on at E12.5, when the kidney is starting to develop [30]. Compared to control mice (i.e. Pkd2 f3/f3 mice without Vil-Cre transgene) ( Fig. 1A-C), Pkd2 f3/f3 mice with the Vil-Cre transgene (Vil-Cre;Pkd2 f3/f3 ) developed severe gross cysts in the kidneys, liver and pancreas ( Fig. 1D-F). Anatomic and histological analyses of the Vil-Cre;Pkd2 f3/f3 kidney ( Fig. 1A versus D and G versus H) showed that renal failure was the likely cause of death in these mice, most of which died between 4 and 6 months of age (Fig. 1I) ideally temporal cystic phenotype, Vil-Cre;Pkd2 f3/f3 mice can be used as an ADPKD model to study the pathogenesis of Pkd2-associated ADPKD and to assess therapeutic responses.
Characterization of cystic phenotypes in the Vil-Cre;Pkd2 f3/f3 mice, an orthologous mouse model of ADPKD To validate Vil-Cre;Pkd2 f3/f3 mice as a suitable model for ADPKD, we systematically characterized the kidney, liver and pancreas by the kidney or liver/body weight ratio, cystic index and hepatorenal functions.
In the kidneys, compared to control mice (i.e. Pkd2 f3/f3 ), the kidney/ body weight ratio and the cystic index were significantly increased in Pkd2 f3/f3 mice with the Vil-Cre transgene (Vil-Cre;Pkd2 f3/f3 ). These mice had a significantly larger kidney/body weight ratio and renal cystic index by the time they were a half-month old (P < 0.05) ( Fig. 2A and B), indicating that cysts were already forming in the diseased kidneys at this early age. Anatomic and histological analysis revealed that the Vil-Cre; Pkd2 f3/f3 kidneys exhibited gross and microscopic cysts (Fig. 2C), in correlation with the kidney/body weight ratio and cystic index. By 4 months of age, normal parenchyma was almost entirely lost in the Vil-Cre;Pkd2 f3/f3 kidney ( Fig. 2Ce and j), indicating that these mice could suffer renal failure at this stage. To test renal function, we measured the blood urea nitrogen (BUN) and creatinine (Cr) levels in Pkd2 f3/f3 mice with and without the Vil-Cre transgene. Compared to their Pkd2 f3/f3 littermates, the Vil-Cre;Pkd2 f3/f3 mice had significantly higher BUN level at 2 months of age, but Cr level at 1 month of age ( Fig. 2D and E). These results indicated that Cr might be more sensitive than BUN for detecting renal function damage in this disease model. In addition, we also used the nephron-specific markers to clarify the renal segment of origin of the cysts in the Vil-Cre;Pkd2 f3/f3 mice. Immunofluorescence staining revealed that the cyst-lining cells in 3-month-old Vil-Cre;Pkd2 f3/f3 kidneys were positive for all renal segments (Fig. S1), indicating that the renal cysts derived from all nephronic segments of the kidney.
By examination of the cystic liver and pancreas in Vil-Cre;Pkd2 f3/f3 mice, we found no cysts in the liver of Vil-Cre;Pkd2 f3/f3 mice before 1 month of age (Fig. S2Aa), or in the pancreas before 3 months of age ( Fig. S2Ba-c). However, gross cysts were observed in the liver starting at 2 months ( Fig. S2Ab-d) and in the pancreas starting at 4 months of age (Fig. S2Bd). Histological analyses revealed similar findings ( Fig. S2Af-h, Bh), but the cystic liver worsened rapidly with age. As Vil-Cre;Pkd2 f3/f3 mice developed severe cysts in the liver, we also examined the liver/body weight ratio and measured the liver function using alanine aminotransferase (ALT). The liver/body weight ratio and liver function did not differ significantly between Pkd2 f3/f3 mice with or without the Vil-Cre transgene (Fig. S2C-D), indicating that as with human ADPKD patients, liver disease is not a life-threatening factor in the mouse model. Our results showed that Vil-Cre; Pkd2 f3/f3 mice not only exhibited a spatially and temporally appropriate cystic kidney phenotype for ADPKD, but also developed liver cysts as is often seen in human ADPKD patients.
Effect of gender on disease severity in the Vil-Cre;Pkd2 f3/f3 mice Recent clinical studies have demonstrated that the progression of renal cysts in ADPKD may be influenced by gender [31][32][33][34]; the cystic disease appears to progress more quickly in males than in females. Male patients often develop end-stage renal failure and require haemodialysis earlier than female patients, suggesting that the prognosis for women is better than that for men [4,35,36]. To determine whether this gender difference was also present in the ADPKD mouse model, we analysed the Kaplan-Meier survival rate, kidney/body weight ratio and renal histology and function in male and female Vil-Cre;Pkd2 f3/f3 mice. Notably, more than 50% of the female Vil-Cre; Pkd2 f3/f3 mice survived to 5 months of age, approximately 1 month longer than their male counterparts (P < 0.01) (Fig. S3A), indicating a better survival rate and lifespan for females. In addition, histological analysis showed that the renal cystogenesis was more severe in the kidneys from male Vil-Cre;Pkd2 f3/f3 mice (Fig. S3B). The renal cystic index and kidney/body weight ratio differed significantly between male and female Vil-Cre;Pkd2 f3/f3 mice after 3 months of age (P < 0.05) ( Fig. S3C-D). Furthermore, the BUN and Cr levels in 4-month-old Vil-Cre;Pkd2 f3/f3 males were significantly elevated over those in age-matched females (P < 0.05) ( Fig. S3E-F). Thus, gender might influence cystogenesis in the ADPKD kidney, with female gender being a protective factor both in the animal model and in ADPKD patients [34,37,38].
Validation of the Vil-Cre;Pkd2 f3/f3 mice as an ADPKD animal model for drug treatment Rapamycin is reported to significantly ameliorate the cystic phenotypes of diverse ADPKD animal models [10,[39][40][41][42][43][44]. To ensure that the ADPKD mouse model could be obtained therapeutic response, we The kidney/body weight ratio for Vil-Cre;Pkd2 f3/f3 newborns and for mice at ½, 1, 2, 3 and 4 months of age, compared to that of age-matched control mice (Pkd2 f3/f3 ). (B) The renal cystic index of Pkd2 f3/f3 mice with or without the Vil-Cre transgene allele at the ages shown. (C) The Vil-Cre;Pkd2 f3/f3 kidney showed gross cysts (arrows) at 1 month of age (b) and massive gross cysts at 2-4 months (c-e); histological results (f-j) of the same kidneys were consistent with the gross examination. (D) BUN levels of Vil-Cre;Pkd2 f3/f3 mice at ½, 1, 2, 3 and 4 months of age: BUN values rose significantly above those in control (Pkd2 f3/f3 ) mice at 2 months of age and continued to worsen with age. (E) Cr levels showed a similar pattern, starting at 1 month of age. *P < 0.05; **P < 0.01; ***P < 0.001. Bar: 3 mm in Ca-e; 600 lm in Cf-j; 100 lm in boxed areas. M: months. 1623 treated Vil-Cre;Pkd2 f3/f3 mice with rapamycin (50 mg/kg/day i.p.) (Protocol I, Fig. 3Aa). The Protocol I-treated mice were given daily injections from postnatal day (P) 10 to 60; the control group was treated with the same injection and volume of DMSO solvent (placebo). The mice were killed 5 days (P65) after the end of treatment (ETR). Kaplan-Meier survival analysis showed that the Vil-Cre;Pkd2 f3/f3 mice treated with rapamycin had a significantly longer lifespan than those treated with the placebo (P < 0.001) (Fig. 3Ab). As seen also in untreated mice (Fig. S3), the female rapamycin-treated mice had a significantly higher survival rate than their male counterparts, with a median survival of about 7 months for females and 6 months for males (P < 0.05) (Fig. 3Ac).
Besides improving survival, intensive rapamycin treatment also significantly reduced the growth of renal cysts (Fig. 3B), the renal cystic index (Fig. 3C) and the kidney/body weight ratio (Fig. 3D) in Vil-Cre;Pkd2 f3/f3 mice. The treatment response to rapamycin was significantly better in females ( Fig. 3C and D), consistent with the survival analyses (Fig. 3Ac). We also investigated renal function and found that BUN and Cr levels were significantly improved in rapamycin-treated as compared to placebo-treated Vil-Cre;Pkd2 f3/f3 mice (P < 0.01 and P < 0.05, respectively) ( Fig. 3E and F). Improvements in BUN and Cr were similar in males and females. Interestingly, although rapamycin rescued the BUN level in Vil-Cre;Pkd2 f3/f3 mice to that in control Pkd2 f3/f3 mice (Fig. 3E), rapamycin significantly reduced Cr (P < 0.05) but could not bring it down to the normal control levels (P < 0.001) (Fig. 3F). Thus, the intensive rapamycin protocol (Protocol I, Fig. 3Aa) could significantly improve the cystic phenotype of the Vil-Cre;Pkd2 f3/f3 kidney, but might not fully rescue renal function.
Kaplan-Meier survival analyses showed that the lifespans were significantly improved in Vil-Cre;Pkd2 f3/f3 mice treated with protocols II-IV compared to those treated with the placebo (P < 0.001) (Fig. 4B-D). Rapamycin also significantly reduced the renal cyst growth (Fig. 4E) and kidney/body weight ratios (Fig. 4F) and improved BUN (Fig. 4G) and Cr values (Fig. 4H). Statistical analyses showed that Protocol II, with the longest duration of rapamycin treatment (100 days), yielded better results than did protocols III and IV, in which mice received rapamycin for the first or second 50 treatment days, respectively. Protocol II significantly reduced renal cyst growth (P < 0.01) (Fig. 4E), lowered the kidney/body weight ratios (P < 0.05) (Fig. 4F) and improved the BUN (P < 0.01) (Fig. 4G) and Cr (P < 0.001) (Fig. 4H) levels compared to protocols III and IV. We also compared the survival rates in the Protocol II group with those of other protocols (I, III and IV) (Figs 3Aa and 4A). Protocol II prolonged the lifespan in Vil-Cre;Pkd2 f3/f3 mice to a median survival of about 8 months, compared to 6.5 months (P < 0.05) for Protocol I (Fig. 4I), and about 7 months for protocols III and IV (P < 0.001) (data not shown). Thus, extending the duration of rapamycin treatment significantly improved the prognosis of Vil-Cre;Pkd2 f3/f3 mice.
Rapamycin down-regulates renal tubule proliferation in Vil-Cre;Pkd2 f3/f3 mice by inhibiting cell-cycle-associated cyclins and cyclin-dependent kinase Previous studies demonstrated that rapamycin can effectively inhibit mTORC1 and its downstream factors [59,60]. Recent studies have shown that the mTOR pathway is dysregulated in ADPKD patients and in ADPKD animal models [45,61]. Rapamycin significantly inhibits cyst formation and greatly improves the prognosis for several ADPKD animal models [10,[40][41][42][43][44][46][47][48][49]. However, the molecular link between polycystin dysfunction and mTOR signalling changes has not been clarified. In this study, we initially investigated the mTORC1 downstream indicators S6K1, S6rp, 4E-BP1 and eIF4E in the Vil-Cre; Fig. 3 Rapamycin delayed cyst formation in Vil-Cre;Pkd2 f3/f3 mice. (A) A diagram showing the initial treatment schedule for rapamycin injection (Protocol I) (a) in this study. Kaplan-Meier survival curves showed that compared to a placebo treatment, rapamycin significantly prolonged the lifespan in Vil-Cre;Pkd2 f3/f3 mice (b) (P < 0.001), and that the lifespan of rapamycin-treated Vil-Cre;Pkd2 f3/f3 females was significantly longer than that of their male counterparts (c) (P < 0.05). (B) Histological examination of kidney sections from 2-month-old mice showed that cysts were reduced in the rapamycin-treated compared to DMSO (placebo)-treated Vil-Cre;Pkd2 f3/f3 mice, and that kidneys from the rapamycin-treated Vil-Cre;Pkd2 f3/f3 females appeared less cystic than those of their male counterparts (b versus e). (C) The cystic index was significantly lower in rapamycin-treated than placebo-treated Vil-Cre;Pkd2 f3/f3 mice (P < 0.001). Rapamycin produced a significantly stronger therapeutic response in the females than in the males (P < 0.05). (D) Kidney/body weight ratios for the mice in (C) were consistent with the cyst-index findings. (E) BUN levels were significantly suppressed in the rapamycin-treated compared to placebo-treated Vil-Cre;Pkd2 f3/f3 mice (P < 0.01). In both male and female Vil-Cre;Pkd2 f3/f3 mice, rapamycin reduced the BUN to the level found in control (Pkd2 f3/f3 ) mice. (F) In the same mice, rapamycin significantly suppressed the Cr level (P < 0.05) but did not rescue Cr to the control level (P < 0.001). Bars: 600 lm in B. M: Male; F: Female.
To explore the molecular mechanism by which rapamycin suppresses epithelial proliferation, we have searched our previously established cDNA microarrays data between our E8 cell line (Pkd2null, Pkd2 À/À ) and its maternally derived D3 cell line (Pkd2 f/À , Pkd2 +/À ) [27]. Of the signature genes, Ccnd1 mRNA level encoding cyclin D1, whose function closely associates cell proliferation, showed a fourfold higher elevation in E8 than in D3 cells. To validate the microarray data, we performed quantitative PCR using mRNAs of E8 and D3 cells, as well as renal tissue mRNA from Vil-Cre;Pkd2 f3/3 and its normal control Pkd2 f3/f3 . We found that down-regulating PC2 not only significantly increased Ccnd1 expression, but also elevated all cellcycle-associated cyclins and CDK1 (Fig. S5G-H).
To confirm the finding from these mRNA assays was also affected in the mouse kidney tissue or in their epithelial cells with and without rapamycin treatment, we performed Western blot analyses of kidneys from control (Pkd2 f3/f3 ) mice and from rapamycin-or placebo-treated Vil-Cre;Pkd2 f3/f3 mice. The results revealed that all of the cell-cycleassociated cyclins (A, B, D1 and E) were abnormally up-regulated in the Vil-Cre;Pkd2 f3/f3 kidney tissue (Fig. 5A). These aberrantly upregulated cyclins were still further significantly suppressed by rapamycin to levels approaching those in the control kidney (Fig. 5B-E), indicating that rapamycin suppressed epithelial proliferation in the Vil-Cre;Pkd2 f3/f3 kidney by inhibiting cell-cycle-associated cyclins. Interestingly, although rapamycin brought the cyclins A, B and E down almost to normal control levels (Fig. 5B,C and E), cyclin D1 was reduced to only approximately half of the excess level found in the untreated Vil-Cre;Pkd2 f3/f3 kidney (Fig. 5D). The result suggested that rapamycin could not fully rescue all of the abnormally cell-cycleassociated cyclins in the Vil-Cre;Pkd2 f3/f3 kidney.
Most cyclins couple with CDKs to form CDK/cyclin complexes which control kinase activity and substrate specificity [66,67]. CDK/ cyclin complexes exert important role in driving cell cycle transition and regulating cell cycle progression. Our mRNA data showed that only CDK1, but not other cell-cycle-associated CDK2/4/6, was significantly affected (Fig. S5G-H). We therefore tested all these CDKs in the same tested kidneys. Compared to the kidneys from control and rapamycin-or placebo-treated Vil-Cre;Pkd2 f3/f3 mice, the similar result to mRNA finding was obtained in the Vil-Cre;Pkd2 f3/f3 kidney ( Fig. 5F-J). This aberrantly up-regulated CDK1 could also be significantly suppressed by rapamycin to level of the control kidney (Fig. 5G). As the tested CDKs' function was usually activated by their phosphorylation sites, we therefore performed Western blot analyses to investigate all activated CDK forms, including CDK1pT161, CDK2pT160, CDK4pT172 and CDK6pY13. The same finding was seen in the tested renal tissues (Fig. 5K and L and Fig. S5I-L). Because CDK-activating kinase (CAK) phosphorylates CDK1pT161, we tested expressional level of a CAK catalytic subunit CDK7 in the same kidney. Similar results to CDK1pT161 were also seen in the tested renal tissues (Fig. 5K and M).
To further validate this finding, we conducted cell-based experiments to detect the cyclins (A, B, D1 and E) and CDKs (CDK1, CDK2, CDK4 and CDK6) in cells. We treated E8 (Pkd2-null) and its maternally derived D3 (Pkd2 +/À ) cells with rapamycin at different concentrations (non-treated and treated with 1 nM or 10 nM for 12 hrs) ( Fig. 6A and C-F and Fig. 7A-C) and durations (0, 6 and 12 hrs at 10 nM) ( Fig. 6B and G-J and Fig. 7D-F). Increasing the rapamycin concentration appeared to produce a trend towards the delayed expression of the cell-cycle-associated cyclins (A, B, D1 and E) and CDK1/CDK1pT161 in both D3 and E8 cells. The highest rapamycin concentration (10 nM) brought all of these cyclins and CDK1/ CDK1pT161 down to their normal control levels (Figs 6C-F and 7B-C). Extending the duration of rapamycin treatment also increased its effect (Figs 6B and 7D). Treating Pkd2-null cells (E8) with rapamycin at a concentration of 10 nM for the longest treatment duration (12 hrs) reduced the cyclins and CDK1/CDK1pT161 to the normal control levels (Figs 6G-J and 7E and F). In parallel, we also applied CDK1 inhibitor Dinaciclib to these cells with different concentrations and durations. The result showed that the highest concentration and the longest duration brought CDK1/CDK1pT161 down to their normal control levels (Fig. 7G-I and J-L). Thus, rapamycin significantly repressed the aberrant activation of cell-cycle-associated cyclins and CDK1/CDK1pT161 in Pkd2-null cells in a dose-and time-dependent manner.

Discussion
We used a Pkd2 conditional-knockout model [27] to screen a spectrum of Cre transgenic mice driven by various kidney-related promoters (Table 1) and identified the Vil-Cre;Pkd2 f3/f3 ADPKD mouse model, which spatially and temporally mimics human ADPKD phenotypes. To validate Vil-Cre;Pkd2 f3/f3 mice as a useful ADPKD model for preclinical trials, we used a rapamycin-based treatment that is known to be effective for ADPKD mice [40][41][42]. The response of the Vil-Cre; Pkd2 f3/f3 mice to rapamycin was consistent with that reported for other ADPKD models, and the progression and timeline of the cystic phenotype and of renal failure in the Vil-Cre;Pkd2 f3/f3 mice, along with the ease of evaluating the progression of the disease, make these mice a suitable model for evaluating and conducting preclinical trials of therapeutic interventions for ADPKD.
Rapamycin has been very effective for treating ADPKD mice, but has been less successful in clinical trials with ADPKD patients [26,[52][53][54]57]. Several explanations have been suggested for the different responses of mice and humans to rapamycin [41,42], one being that to prevent adverse events, ADPKD patients have received lower doses (2-3 mg/kg) of rapamycin than those used in mouse models Fig. 7 Rapamycin suppresses CDK1 and its activated form CDK1pT161 in vitro. (A) The same cell lines (E8 and D3) were also incubated with 0, 1, or 10 nM rapamycin for 12 hrs. Western blots of the lysates of E8 and D3 cells showed that rapamycin decreased the expression of CDK1 and its activated form CDK1pT161 in a dose-dependent manner. (B-C) Normalized quantitative analysis using the densitometry values from the Western blots in (A). (D) The cell lines were also incubated with 10 nM rapamycin for 0, 6 or 12 hrs. Western blots of the cell lysates showed that rapamycin significantly decreased the expression of CDK1 and its activated forms CDK1pT161 in a manner dependent on the duration of incubation. (E-F) Normalized quantitative analysis using the densitometry values from the Western blots presented in (D). Other cell-cycle-associated CDKs (such as CDK2, CDK4 and CDK6) were not affected by PC2 expressional levels and rapamycin treatment (data not shown). (G) The same cell lines were cultured with CDK inhibitor Dinaciclib with 0, 5 and 10 lM for 24 hrs. Western blots of the cell lysates showed that Dinaciclib significantly inhibited the expression of CDK1 and its activated forms CDK1pT161 in a dose-dependent manner. (H-I) Normalized quantitative analysis using the densitometry values from the Western blots presented in (G). (J) The cell lines were also cultured with CDK inhibitor Dinaciclib with 10 lM for 0, 12 or 24 hrs. Similar results to different concentrations were seen in Western blot assays on the different durations of Dinaciclib incubation. (K-L) Normalized quantitative analysis using the densitometry values from the Western blots presented in (J). N.S. = No significance; *P < 0.05; **P < 0.01. Fig. 8 The molecular mechanism of the mTOR pathway in Vil-Cre;Pkd2 f3/f3 mice, with or without rapamycin treatment. Based on our study and other group finding, we deduced the molecular link between mTOR signalling changes and the polycystin-2 dysfunction in cystogenesis. (A) The loss of PC2 impairs normal PC1 function, which may in turn disrupt PC1's inhibitory effect on the mTORC1 pathway by disturbing interactions between PC1 and the tuberous sclerosis 2 gene product (TSC2). Activated mTORC1 promotes the phosphorylation of 4E-BP1 and eIF4E, which activates cyclin D1 gene expression. Cyclin D1 couples with CDK4/6 and phosphorylates retinoblastoma protein (RB), thereby releasing (E2F), which promotes the expression of the other cell-cycle-associated cyclins (A, B and E) and CDK1, eventually leading to increased cell proliferation.   [41,42]. Another possible reason for this discrepancy is that rapamycin has not been used as a primary treatment for humans in the early stages of ADPKD cyst development [52,55,68]. To explore whether intensive rapamycin treatment could significantly improve disease phenotypes in an orthologous mouse model of ADPKD under similar conditions as those in human trials, we designed four rapamycin treatment protocols that varied the duration and timing of rapamycin treatment. Our results indicated that a longer duration of rapamycin treatment (Protocol II, from P10 to P110) provided the strongest therapeutic response than any of the other protocols (protocols I, III and IV, each with a duration of 50 days) in the ADPKD animal model. Protocol II yielded the largest increase in lifespan, significantly reduced the growth of renal cysts and significantly improved kidney function. Western blot analyses showed that rapamycin inhibited the activated forms of p-S6K1, p-S6rp, p-4E-BP1 and p-eIF4E (indicators of mTORC1 activity; Fig S6) in a dose-dependent manner, and this inhibition coincided with a significant improvement in the renal cystic phenotype in the rapamycin-treated Vil-Cre;Pkd2 f3/ f3 mouse model. Thus, it is possible that an extended rapamycin treatment protocol would greatly benefit ADPKD patients with a PKD2 mutation.
Interestingly, our results showed that early (P10-60, Protocol I)-and late-treatment (P60-110, Protocol IV) protocols using the same rapamycin dosage provided similar improvements in survival, cystic phenotype and renal function. This finding indicated that starting rapamycin treatment during the early stages of cyst development might not be critical for a better prognosis in the ADPKD model. In addition, we also compared the effects of Protocol I with those of Protocol III, which used the same rapamycin dosage and treatment period as Protocol I (P10-60) but includes a non-treated observation period ending at P110. The responses to Protocol I and Protocol III did not differ significantly in the ADPKD model, indicating that rapamycin effect on cystogenesis of the animal model could be such sustained long after the last dose of rapamycin is administered.
Although rapamycin has shown clear benefits in preclinical experiments and clinical trials for ADPKD, the precise molecular mechanism by which PC2 regulates cell growth via mTOR and how rapamycin inhibits cyst growth and epithelial-cell proliferation have not been clarified. To investigate the molecular mechanism by which rapamycin prevents proliferation, we examined the serial factors most likely to be involved in cell proliferation. Based on our microarray data and qPCR assay, we identified cyclin D1 was significantly up-regulated. Subsequently, we also examined other cyclinfamily members and their cofactors CDKs, by which cell cycling was driven. We found that all of the cell-cycle-associated cyclins (A, B, D1 and E), activated CDK1 (CDK1pT161) and CDK7 (which involves in the phosphorylation of CDK1pT161) were significantly up-regulated in Pkd2-null renal cells and in the tissue of our ADPKD model mice and that these cyclins and CDK7/CDK1 were downregulated by rapamycin treatment. This finding indicated that the increased epithelial proliferation observed in the ADPKD kidney might result from abnormal activation of the mTORC1 pathway and the subsequent up-regulation of cell-cycle-associated cyclins and activated CDK1.
Taking our findings together with those of other distinguished reports, we deduced the molecular mechanism of rapamycin effect on ADPKD as follows: PC2 dysfunction causes PC1 C-terminal region (Cterm) to mislocalize and significantly down-regulate PC1 expression (Fig. S7) [69][70][71]. The down-regulation of PC1 disrupts interactions between PC1 and tuberous sclerosis type 2 (TSC2) [45,[72][73][74]; this abolishes the ability of the TSC complex to inhibit the mTORC1 pathway and thereby activates the mTORC1 downstream factors S6K1, S6rp, 4E-BP1 and eIF4E (Figs S4 and S6) [75,76]. Activated eIF4E binds to the translation initiation site of the cyclin D1 gene, causing it to overexpress cyclin D1, a key regulator of cell growth (Figs 5D and 6E and I) [77][78][79]. Cyclin D1 forms a complex with CDK4 or CDK6 to further phosphorylate the retinoblastoma protein (RB) and release E2F from the RB/E2Fgene-suppressing domain [80] (Fig. S8). E2F is a transcriptional factor that can initiate and enhance the expression of all of the other cell-cycle-associated cyclins, cyclin E (Figs 5E and 6F and J), cyclin A (Figs 5B and 6C and G) and cyclin B (Figs 5C and 6D and H) [66,67,[81][82][83]. These activated cyclins cooperate with their corresponding CDKs to promote the renal epithelial proliferation and induce cystogenesis in the ADPKD model (Fig. 8A). When rapamycin was applied, the PC2 inactivation-induced overactivation of mTORC1 can be inhibited, and subsequently suppress all downstream factors of the mTORC1 pathway (Fig. 8B), thereby finally arresting abnormal cell proliferation in the orthologous mouse model of ADPKD.
In the light of our previous studies [27,28], we identified and established a new ADPKD mouse model bearing Pkd2 mutant alleles (Vil-Cre;Pkd2 f3/f3 ). We systematically characterized the Vil-Cre; Pkd2 f3/f3 mouse and verified its suitability as a model that mimics the phenotype and progression of human ADPKD. Using this new mouse system, we demonstrated that these animals responded to rapamycin in a time-and dose-dependent manner: increasing the dosage or duration of treatment improved the therapeutic effect. We also demonstrated the molecular mechanism by which the mTOR inhibitor rapamycin arrests the aberrant epithelial proliferation in the ADPKD kidney, which resulted de novo from PC2 dysfunction. This newly developed ADPKD model will promote translational medicine from in vivo preclinical trials to ADPKD therapies and will also provide a platform for identifying new molecular targets for treating ADPKD based on the newly defined mTORC1-CDK1/cyclin axis.

Author contributions
GW conceived and designed the experiments. AL, JM, XS and JM performed the experiments. AL, YX, LZ, XZ, DW, GM, CL and GW analysed and interpreted the data. AL, CL and GW wrote the manuscript.