Metformin Inhibits Cyst Formation in a Zebrafish Model of Polycystin-2 Deficiency

Autosomal dominant polycystic kidney disease (ADPKD) is a common kidney disease caused by mutations in PKD1 or PKD2. Metformin reduces cyst growth in mouse models of PKD1. However, metformin has not been studied in animal models of PKD2, and the cellular mechanism underlying its effectiveness is not entirely clear. This study investigated the effects of metformin on cyst formation in a zebrafish model of polycystin-2 deficiency resulting from morpholino knockdown of pkd2. We added metformin (2.5 to 20 mM) to the embryo media between 4 and 48 hours post fertilisation and observed pronephric cyst formation by using the wt1b promoter-driven GFP signal in Tg(wt1b:GFP) pkd2 morphants. Metformin inhibited pronephric cyst formation by 42–61% compared with the untreated controls. Metformin also reduced the number of proliferating cells in the pronephric ducts, the degree of dorsal body curvature, and the infiltration of leukocytes surrounding the pronephros. Moreover, metformin treatment increased the phosphorylation of adenosine monophosphate-activated protein kinase (AMPK) and enhanced autophagy in the pronephros. Our data suggest that metformin reduces cyst formation through activation of the AMPK pathway and modulation of defective cellular events such as proliferation and autophagy. These results also imply that metformin could have therapeutic potential for ADPKD treatment.

and adenosine monophosphate-activated protein kinase (AMPK) 16,17 . In recent years, the therapeutic potential of metformin in cancer and polycystic ovary disease has drawn increasing attention 18,19 . For example, data from retrospective studies indicate that patients treated with metformin have a lower risk of prostate cancer and other urologic malignancies 20 , suggesting that metformin could have clinical applications in addition to glycemic control in type 2 diabetes mellitus 21 .
MacCarty et al. hypothesised that metformin can reduce cyst formation on the basis of its abilities to activate AMPK and suppress cystic fibrosis transmembrane conductance regulator (CFTR) and mTOR 22 . Accordingly, a previous study demonstrated that metformin inhibited cyst growth in Madin-Darby canine kidney (MDCK) cells and Pkd1 conditional knockout mice 23 . However, the exact in vivo mechanisms underlying the effect of metformin on cystogenesis are not entirely understood. Furthermore, recent studies have suggested that polycystin-1-deficent and polycystin-2-deficent cells might be different in the degree of AMPK inhibition 24 . Hence, whether metformin can inhibit cyst growth in the setting of polycystin-2 deficiency and the possible mechanisms of its action remain to be determined.
In this study, we investigated the effects of metformin on the initiation of pronephric cysts using a zebrafish pkd2 model. We observed that metformin reduces pronephric cyst formation through the activation of AMPK and restores autophagy activity. These findings indicate that metformin could play a role in the treatment of early-stage ADPKD.

Results
Metformin prevents cyst initiation in pkd2 morphants. In this study, we used a zebrafish model of PKD2 for evaluating the effects of metformin on cyst initiation [25][26][27][28][29] . A translation-blocking morpholino oligos (MO) against pkd2 was injected in a transgenic line, in which green fluorescent protein (GFP) expression was driven by the pronephros-specific wt1b promoter. We treated the Tg(wt1b:GFP) pkd2 morphants with different concentrations of metformin in the E3 buffer between 4 and 48 hours post fertilisation (hpf). As illustrated in Fig. 1A, metformin treatment resulted in a significant reduction (42% to 61%) of cyst formation in the Tg(wt1b:GFP) pkd2 morphants compared with untreated controls at 48 hpf (P < 0.01). The frequency of glomerular cyst formation decreased dose-dependently from 71.4% ± 4.6% in the untreated pkd2 morphants to 41.5% ± 6.8% with 2.5 mM metformin and 33.4% ± 5.1%, 28.2% ± 4.6%, and 31.8% ± 4.5% with 5 mM, 10 mM, and 20 mM metformin, respectively (Fig. 1B). In transverse histological sections, we confirmed the rescue effect of metformin treatment on the pronephric cysts in the pkd2 morphants (Fig. 1C). These results indicate that metformin could inhibit the early stage of cyst formation in the zebrafish model of PKD2. Metformin reduces dorsal body curvature and cloaca malformation in pkd2 morphants. We examined the effect of metformin on the curvature phenotype of pkd2 morphants ( Fig. 2A-C). Dorsal axis curvature results from an overproduction of type II collagen in the notochord sheath and has been used as a surrogate readout for cystogenesis in pkd2 morphants 26 . As illustrated in Fig. 2G, metformin partially reduced the frequencies of the overall curvature and severe dorsal curvature (>90°) phenotypes by 14% (114/116 vs. 81/96, P < 0.001) and 15% (67/116 vs. 41/96, P < 0.05), respectively, compared with untreated controls. These data indicate that metformin also ameliorated the defects in extracellular matrix formation involved in the pathogenesis of ADPKD. For comparison, we treated pkd1a/b morphants with 10 mM metformin and the curvature phenotype was also partially suppressed, similar to that observed in pkd2 morphants ( Supplementary Fig. S1). However, the small percentages of pronephric cysts in the pkd1a/b morphants preclude us from further studies using this model 30 .
The cloaca malformation has been shown to correlate the formation of pronephric cysts in several zebrafish models of ciliopathies 6,[31][32][33] . In agreement with these studies, we found that the frequency of aberrant cloaca increased in pkd2 morphants and metformin significantly improved the phenotype by 21% (65/70 vs. 52/72, P < 0.01) ( Fig. 2D-F,H). These data indicate that metformin consistently suppressed the different phenotypes of pkd2 morphants. However, whether the improvement in cloacal morphology can lead to increased urine flow in metformin-treated pkd2 morphants requires further study of kidney function using rhodamine-dextran filtration assays 33-36 . Metformin reduces tubular cell proliferation. We next sought to determine whether metformin affects cell proliferation in the pronephric kidney. We performed double immunostaining using antibodies against phosphohistone H3 (PH3) and the Na/K-ATPase α-1 subunit (α6 F) to mark proliferating cells in the pronephric ducts (Fig. 3A). Significantly fewer proliferating cells were observed in the pronephric ducts of the pkd2 morphants treated with metformin than in those of the untreated pkd2 morphants (Fig. 3B). These data suggest that metformin reduced pronephric cyst formation in the pkd2 morphants through inhibition of epithelial cell proliferation. Metformin reduces leukocyte accumulation. Previous studies have revealed that inflammation contributes to cell proliferation and cyst formation in ADPKD 10 . To explore the mechanisms by which metformin inhibits cyst growth, we analysed the effect of metformin on leukocyte infiltration in the pkd2 morphants. Whole-mount in situ hybridisation revealed significantly increased l-plastin and mpx expression in the trunk area surrounding the pronephric ducts in the pkd2 morphants compared with mismatched MO controls (Fig. 4A). The pan-leukocyte marker l-plastin is an actin-binding protein preferentially expressed in monocytes/macrophages and mpx is a neutrophil marker 37 . Metformin treatment caused a significant reduction of leukocyte infiltration in the pronephric area in the pkd2 morphants ( Fig. 4B and C). These data indicate that metformin could have an additional anti-inflammatory role that contributes to its therapeutic effects on polycystic kidney disease.

Activation of AMPK by metformin.
A previous study revealed that Pkd1 −/− mouse embryonic fibroblasts exhibit a lower level of AMPK phosphorylation than do wild-type cells 9 . Metformin has been shown to stimulate AMPK in cell and mouse models of ADPKD 23,24 . Therefore, we evaluated the effect of metformin on the activation of the AMPK signalling pathway. Western blot analysis using an antibody against p-AMPKα at Thr172 indicated that the p-AMPK/AMPK ratio was significantly higher in the metformin-treated pkd2 morphants than in the untreated controls ( Fig. 5A and B, Supplementary Fig. S2). These data are consistent with the hypothesis that AMPK activation inhibits polycystic kidney disease. However, we did not observe a lower baseline expression level of p-AMPK in the pkd2 morphants than in the wild-type controls; this effect could be partly due to the presence of residual maternal pAMPK in the zebrafish embryos 38 . Metformin cannot prevent cyst formation in tsc1a morphants. A study indicated that the AMPK-dependent activation of tuberous sclerosis complex (TSC) 1 and 2 proteins plays a major role in the inhibition of the mTOR signalling pathway during energetic demands 39 . Therefore, we tested whether metformin prevented cyst initiation in the absence of TSC proteins. Morpholino knockdown of tsc1a induced severe glomerular cysts in zebrafish embryos at 48 hpf, as reported previously (Fig. 6A) 40 . Metformin apparently did not affect cyst formation in the tsc1a morphants (Fig. 6B). This observation is consistent with an AMPK/TSC-dependent mechanism underlying the prevention of cyst formation by metformin. Metformin enhances autophagy in the pronephros. Defective autophagy, an AMPK downstream cellular process that degrades cytoplasmic components for energy production, has been associated with the pathogenesis of ADPKD in recent studies 41 . Therefore, we examined whether metformin increased the autophagy activities in pkd2 morphants. Autophagy activity was assessed by detecting the intracellular levels of microtubule-associated protein light chain 3 (LC3) through immunofluorescence and Western blot analysis 42 . As illustrated in Fig. 7A, the pkd2 morphants exhibited lower LC3 staining in the pronephric ducts than did the control embryos, and the deficiency was rectified by metformin treatment. Consistent with these findings, the Western blot analysis indicated that the conversion of cytoplasmic LC3-I into LC3-II was significantly lower in the pkd2 morphants than in the control embryos, and metformin treatment restored the autophagy activities ( Fig. 7B and Supplementary Fig. S2). Furthermore, we observed that simultaneous knockdown of the autophagy gene atg5 significantly increased the cystic phenotype in the pkd2 morphants ( Fig. 7C and D). Although the magnitudes of these differences were moderate and further studies are needed to confirm our findings, these results suggest that the activation of AMPK and restoration of autophagy constitute a plausible mechanism for the inhibition of cyst growth by metformin 43 .

Discussion
The zebrafish has become an increasingly recognised model for identifying drug candidates for ADPKD treatment [44][45][46][47] . Chemical compound screening in zebrafish pkd2 models indicated that histone deacetylase inhibitors inhibited cyst growth 48 . Candidate drugs that are effective in other PKD animal models, including rapamycin, roscovitine, and pasireotide, have been successfully validated using the zebrafish system 34,49 . In the current study, we demonstrated that metformin activated the phosphorylation of AMPK and prevented cyst formation in a zebrafish pkd2 model. In particular, we demonstrated that metformin inhibited epithelial cell proliferation and restored normal autophagy activity in the pronephric kidney. Furthermore, we demonstrated that metformin did not prevent cyst formation in tsc1a morphants resulting from knocking down the AMPK downstream target TSC1.
Metformin was shown to suppress cyst growth and fluid secretion through the inhibition of mTOR and CFTR in a previous study using MDCK cell cysts, mouse embryonic kidney explant cultures, and Pkd1 mouse models 23 . In accordance with this study, we found that metformin inhibited early cystogenesis in zebrafish pkd2 morphant embryos, indicating that metformin can inhibit cyst growth in both PKD1 and PKD2 animal models. These results support the hypothesis that metformin reduces cyst formation in the early stages of ADPKD. Interestingly, a previous study suggests that polycystin-2 deficient cell lines are less amenable to metformin and rapamycin than polycystin-1 deficient cell lines due to the less activation of mTOR pathway 24 . However, we observed a similar suppressive effect of metformin on the curvature phenotype in both pkd1a/b and pkd2 morphants. The discrepancy could be explained by the differences between in vivo and in vitro cystic models, or metformin could have pleiotropic effects on multiple signalling pathways related to the PKD2 deficiency. Further studies will be required to clarify our findings.
One possible mechanism by which metformin prevents cyst formation is through the activation of AMPK and inhibition of cell proliferation. The AMPK signalling pathway plays a crucial role in maintaining normal kidney structure during nephron morphogenesis by affecting cell proliferation and migration 50,51 . Activated AMPK restores the cell energy balance by shutting down the ATP-consuming synthesis pathways, and thus causes cells to switch from an anabolic to a catabolic state 18 . Metformin reduced cell proliferation in an AMPK-dependent manner in MDCK cysts grown in 3D collagen gels 23 . Similarly, intraperitoneal injection of metformin resulted in a reduction in the number of Ki67-positive epithelial cells in the cystic kidneys of Pkd1 flox/− Ksp-Cre mice at postnatal day 7 23 . Furthermore, forced activation of AMPK by 2-deoxyglucose (2DG) restored normal extracellular signal-regulated kinase (ERK) activity, inhibited glycolysis, and reduced the cystic index and proliferation rate in Pkd1 conditional knockout mice 9 . Our findings are consistent with these results and suggest that metformin may inhibit cystogenesis through an AMPK-dependent pathway.
Growing evidence substantiates the role of macrophages in promoting cyst growth in ADPKD. Macrophages were demonstrated to stimulate cell proliferation and cyst expansion in coculture experiments within a collagen matrix 52 . The recruitment and retention of renal macrophages contributes to the increased proliferation during cyst growth in ADPKD 10 . A previous study showed that metformin treatment reduced inflammatory cytokine production in peripheral blood mononuclear cells obtained from healthy volunteers 53 . Metformin also suppressed inflammation through activation of AMPK and phosphatase and tensin homolog (PTEN) in vascular smooth muscle cells 54 . In accordance with these findings, we observed that metformin reduced the recruitment of macrophages in pkd2 morphants, which could be an alternative mechanism through which metformin reduces cell proliferation and cyst formation. Another significant finding in this study is that metformin corrects the defective autophagy observed in pkd2 morphants. Autophagy is a highly regulated cellular process that degrades and recycles intracellular proteins and organelles in lysosomes during metabolic stress or nutritional deprivation 55 . Our finding that pkd2 morphants had defective autophagy is consistent with a previous study showing that Pkd1 −/− mouse embryonic fibroblasts had insufficient autophagy activity upon glucose deprivation 9 . Furthermore, our results suggested that metformin might reduce cyst formation through the enhancement of autophagy in ADPKD 56 . Autophagy suppression could reduce the senescence of cyst-lining cells and lead to increased proliferation, apoptosis, and cyst growth 57 . Autophagy has also been shown to regulate the formation and function of primary cilia 58 . The detailed mechanisms that link autophagy and cyst growth require further study 59 .
The similarities in the cystic phenotype of pkd2 and tsc1a morphants prompted us to investigate whether metformin could also inhibit cyst formation in this previously described model of tuberous sclerosis 40 . However, our results demonstrated that metformin could not rescue the cystic phenotype of the tsc1a morphants. This observation is consistent with an AMPK-dependent effect of metformin because TSC1 is a known downstream target of AMPK. Long-term treatment with metformin also failed to suppress renal tumours in a Tsc1 +/− mouse model 60 .
We were unable to further determine the relative contribution of AMPK activation in the inhibition of cystogenesis because of a lack of a proper morpholino or specific antagonist to reduce AMPK activity without interfering with the development of zebrafish embryos 38 . Therefore, we cannot exclude the possibility that metformin inhibits cystogenesis through an AMPK-independent effect 18,61 .
Our data contribute to a growing body of evidence that metabolic abnormalities are crucial in the pathogenesis of ADPKD 50,62 . Inhibition of aerobic glycolysis by using 2DG suppressed cell proliferation, leukocyte infiltration, and cyst formation in mouse models of PKD1 9 . Mild to moderate food restriction was demonstrated to activate AMPK and reduce cyst area, renal fibrosis, and inflammation 63 . These studies support a new therapeutic strategy for ADPKD, which involves reprogramming of cellular metabolism 64 .
Although our results support the hypothesis that metformin is a potential treatment for ADPKD, metformin could cause lactic acidosis in patients with advanced renal failure 18 . Therefore, the inhibitory effect of metformin on cyst growth in animal models requires confirmation in future clinical studies. A randomised, double-blind clinical trial evaluating metformin in the treatment of patients with ADPKD was started in 2016 and is expected to be completed in the next few years (ClinicalTrials.gov Identifier: NCT02656017).
In conclusion, metformin reduces cyst formation in pkd2-deficient zebrafish embryos. The data suggest that metformin may prevent cyst formation through activation of the AMPK pathway and modulation of defective cellular events such as proliferation and autophagy. The findings also indicate the therapeutic potential of metformin in treatment of ADPKD at the early stages. Further studies are required to assess whether metformin can improve clinical outcomes in patients with ADPKD.

Methods
Zebrafish maintenance. Animal experiments were approved by the Chang Gung University Institutional Animal Care and Use Committee. The investigation conformed to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Zebrafish and embryos were maintained according to standard procedures 65 . The embryos were staged according to hours post fertilisation. We used the Tg(wt1b:GFP) line (kindly provided by Prof. Christoph Englert, Fritz Lipmann Institute, Jena, Germany) for in vivo observation of pronephric cysts [25][26][27]66 .
Immunofluorescence. Embryos were fixed overnight in 4% paraformaldehyde at 4 °C. Immunostaining was performed in whole-mount embryos as described previously 68 . A rabbit anti-PH3 antibody (1:200, Millipore) was used to label proliferating cells, and a mouse anti-α6 F antibody (1:200, Developmental Studies Hybridoma Bank) was used to label pronephric epithelial cells. LC3, an autophagosome marker, was labelled using a rabbit anti-LC3B antibody (1:200, Novus Biologicals). Secondary antibodies used were the Alexa Fluor 594 goat antirabbit IgG and Alexa Fluor 488 goat antimouse IgG (1:500, Molecular Probes). Proliferating cells in the pronephric ducts were counted in anti-PH3 and anti-α6 F stained whole-mount embryos using fluorescence microscopy. For confocal imaging, embryos were flat-mounted and the alpha 6 F and PH3 signals were recorded in z-series stacks using a Zeiss LSM 510 confocal microscope. For analysis of LC3 staining, the immunostained embryos were embedded in the OTC medium and cryosectioned through the pronephros. The sections were mounted in Vectashield (Vector Laboratories) with DAPI and imaged using confocal microscopy.
Whole-mount in situ hybridisation. The embryos were fixed overnight in 4% paraformaldehyde at 4 °C. Whole-mount in situ hybridisation was performed according to published protocols 65 . Antisense digoxigenin-labelled RNA probes were synthesised from linearised plasmid templates containing l-plastin and mpx cDNAs.

Statistical analysis.
Values are expressed as the mean ± SEM. Comparisons between groups were performed using the Student's t test or ANOVA followed by Dunnett's multiple comparison test. Values that were not normally distributed were analysed using the Kruskal-Wallis test followed by Dunnett's multiple comparison test. Categorical variables were analysed using Fisher's exact test or the chi-square test. P values less than 0.05 were considered statistically significant. All analyses were performed using GraphPad Prism 5.0 (GraphPad, La Jolla, CA, USA).