Novel zebrafish polycystic kidney disease models reveal functions of the Hippo pathway in renal cystogenesis

The Hippo signaling pathway is a kinase cascade first characterized in Drosophila for its role in controlling organ size (Harvey et al., 2003; Huang et al., 2005; Justice et al., 1995). Hippo (Hpo) can bind to scaffold protein Salvador (Sav) and phosphorylate Warts (Wts). The phosphorylated Wts then binds to Mob as tumor suppressor (Mats) and phosphorylates the downstream effector Yorkie (Yki) to inhibit its function as a transcriptional coactivator. Orthologs of all Hippo pathway components have been found in mammals. Mammalian Ste20-like kinases 1/2 (MST1/2, the orthologs of Hpo, also known as STK4/3) binds to WW domain-containing protein 1 (SAV1, the ortholog of Sav), a member of the scaffold protein Salvador family. Then, phosphorylated LATS1/2 bind to MOB kinase activator 1A/B (MOB1A/B, orthologs of Mats) and phosphorylate two downstream effectors, Yes-associated protein 1 (YAP1, the ortholog of Yki) and WW domain-containing transcription regulator protein 1 (WWTR1, paralog of YAP1, also known as TAZ) (Sansores-Garcia et al., 2011; Yu et al., 2013; Zhang et al., 2010). YAP1/WWTR1 are transcriptional coactivators as they do not bind to DNA and activate transcription directly. Instead, they interact with other transcription factors to regulate gene expression. When the Hippo signaling pathway is not activated, YAP1/WWTR1 are mainly retained in the nucleus and serve as transcription coactivators. When the pathway is activated, YAP1/WWTR1 are phosphorylated and translocate to the cytoplasm, losing their functions as transcription coactivators (Yu et al., 2015). TEA domain transcription factors (TEADs) are thought to be the most important partners of YAP1/WWTR1 (Zhao et al., 2008). YAP1/WWTR1 can also interact with other transcription factors, including SMADs, p63 (also known as TP63), RUNXs and PAXs (Varelas, 2014). Together with these transcription factors, YAP1/WWTR1 can regulate the expression of a variety of downstream genes related to cell survival, proliferation, differentiation and migration.

Recently, the Hippo signaling pathway was found to play potential roles in polycystic kidney diseases (PKDs), which are a group of inherited disorders characterized by cyst formation within the kidney (Cornec-Le Gall et al., 2019). Autosomal-dominant PKD (ADPKD) is the most common type of PKD, which is caused by mutation of the PKD1 or PKD2 gene (Hughes et al., 1995; Mochizuki et al., 1996). The symptoms of ADPKD are enlarged kidney and formation of multiple fluid-filled cysts (Wilson, 2004). Analysis of clinical samples from ADPKD patients showed that the Hippo signaling pathway was altered and YAP1/WWTR1 were activated in renal cysts (Cai et al., 2018; Happé et al., 2011). Further studies revealed that loss of YAP1/WWTR1 reduced cyst formation in a Pkd1-deficient mouse model (Cai et al., 2018).

In zebrafish, the orthologs of all Hippo signaling pathway components have also been identified. Some studies on function of the Hippo signaling pathway members in zebrafish have recently been reported based on gene-knockout techniques. A recent study reported that serine/threonine kinase 3 (Stk3, ortholog of MST1/MST2 in zebrafish) and Sav1 played essential roles in the zebrafish biliary system (Brandt et al., 2020). Yap1/Wwtr1 were together found to be important for embryonic development (Kimelman et al., 2017), in particular the development of liver (Yi et al., 2018), eye (Miesfeld et al., 2015) and blood vessels (Astone et al., 2018; Nakajima et al., 2017). Wwtr1 alone was essential for micropyle formation (Dingare et al., 2018; Yi et al., 2019) and cardiac wall maturation (Lai et al., 2018). Interestingly, dysfunction of the Hippo signaling pathway was also implicated in zebrafish renal cyst formation (He et al., 2015; Skouloudaki et al., 2009; Xu et al., 2018; Zhang et al., 2015). However, existing studies on renal cyst formation were fragmentary and mostly based on morpholino-mediated gene knockdown. Further investigation with gene knockout is required to provide more conclusive evidence.

In the past decades, the zebrafish has become a popular model for studying human diseases (Pickart and Klee, 2014), including renal cyst formation (Gehrig et al., 2018; Poureetezadi and Wingert, 2016). However, owing to the lethality of the phenotype, most studies have been confined to pronephric cyst formation during embryogenesis. As there are only two nephrons in zebrafish pronephros, a similar phenotype to human PKD cannot be found in zebrafish embryos. The lack of an adult model has limited the application of zebrafish in the study of PKD.

Here, we report the generation of a zebrafish Stk3-deficient mutant line and a Wwtr1 overexpression line followed by phenotype analysis. Our findings provided critical clues on the function of the Hippo signaling pathway in zebrafish renal cyst formation. Furthermore, with the generation of a novel mesonephric PKD model in zebrafish, we have established a connection between pronephric phenotype in larval stage and polycystic kidney phenotype in adult stage, demonstrating the potential value of the zebrafish stk3 mutant and Wwtr1 overexpression line as models for PKD.

To study the function of the Hippo signaling pathway in zebrafish, we applied the CRISPR/Cas9 system to target the 6th exon of the stk3 gene (Fig. 1A). One mutant with a 5-bp insertion was generated, which resulted in nonsense frameshifting and a premature stop codon.

The homozygous stk3 mutant (stk3^−/−) was generated by in-crossing of the heterozygote (stk3^+/−). All homozygous mutant larvae (stk3^−/−) exhibited an edema phenotype of ascites during the period from 12 days post-fertilization (dpf) to 18 dpf (Fig. 1B). The Kaplan–Meier plot for edema onset showed that the median time of onset of edema was 14.5 dpf in the stk3^−/− mutant (Fig. 1C). After the onset of edema, the larvae died within 1 or 2 days. The survival curve showed that the median survival time was 15 dpf in the stk3^−/− mutant (Fig. 1D). To further investigate the phenotypes of the stk3^−/− mutant, we analyzed serial cross sections of the larvae at 15 dpf (Fig. 1E,F). Histological analysis revealed glomerular cyst formation and pronephric tubule dilation in the stk3^−/− larvae compared to the wild-type (WT) siblings. The areas of renal corpuscle in the stk3^−/− mutant were much larger than those in the WT siblings due to enlarged Bowman's spaces.

To verify the role of the Hippo pathway in zebrafish renal development and understand its signaling mechanism, we further generated two mutant lines for zebrafish yap1 and wwtr1, two downstream effectors of the Hippo pathway.

For yap1 knockout, we targeted the 1st exon of yap1 gene and obtained a mutant allele with a 4-bp deletion (yap1^−/−) (Fig. S1A). The yap1^−/− mutant suffered from growth retardation during the juvenile stage but gradually caught up with yap1^+/+ siblings at the adult stage (Fig. S1B-D). During development, yap1^−/− mutants did not exhibit edema or other kidney-related phenotypes. Histological analysis of adult yap1^−/− mutants showed no obvious dysfunction in the kidney compared to WT siblings (Fig. S1E).

For wwtr1 knockout, we generated a zebrafish wwtr1 mutant line by targeting the 1st exon of the gene. The mutant line carried a 7-bp deletion and a 293-bp insertion (Fig. S2A). Similar to the yap1 mutant, the wwtr1^−/− mutants did not exhibit any signs of edema formation or other kidney-related phenotypes (Fig. S2B). Histological analysis of adult wwtr1^−/− mutants also showed no obvious dysfunction in the kidney (Fig. S2C).

Our data on yap1^−/− and wwtr1^−/− mutants differed from previous knockdown studies with morpholino (morphant), which showed that yap1 and wwtr1 both played roles in zebrafish pronephros development (He et al., 2015; Zhang et al., 2015). To further confirm this, we also performed an experiment to knock down the expression of yap1 and wwtr1 in F0 zebrafish embryos using a CRISPR/Cas9 approach (crispant). Different from the yap1 knockout mutant (yap1^−/−), the yap1 crispant exhibited multiple defects, including pericardial edema, pronephric cysts and curved body (Fig. S3A). These phenotypes could be rescued with full-length yap1 mRNA. Furthermore, wwtr1 mRNA could also rescue the phenotypes of the yap1 crispant, albeit to a lesser extent (Fig. S3B). Similarly, the wwtr1 crispant exhibited multiple defects, including pronephric cysts and a curved body (Fig. S3C). These phenotypes could also be rescued with both wwtr1 and yap1 mRNA (Fig. S3D).

The discrepancy between null mutants and crispants suggests potential functional compensation in the knockout, but not knockdown, mutants (El-Brolosy et al., 2019; Ma et al., 2019; Rossi et al., 2015). To confirm this hypothesis, we generated a yap1 and wwtr1 double mutant (yap1^−/−;wwtr1^−/−). As reported previously (Kimelman et al., 2017), the yap1^−/−;wwtr1^−/− double mutants turned out to be embryonically lethal. The mutant embryos exhibited no tail extension and ceased development around 20 h post-fertilization (hpf) (Fig. 2A). The lethality could be rescued by yap1 (yap1^+/+;wwtr1^−/−, yap1^+/−;wwtr1^−/−), as shown by normal embryonic development at 3 dpf. Interestingly, although wwtr1 could also rescue the embryos, the restoration of normal development required both alleles of wwtr1 (yap1^−/−;wwtr1^+/+). The yap1^−/−;wwtr1^+/− fish exhibited multiple defects including pericardial edema, pronephric cysts and curved body, similar to those of the yap1 and wwtr1 crispants (Fig. 2B). We then performed whole-mount immunostaining with anti-acetylated tubulin (ace-Tub) to label cilia in pronephros, and maximal intensity projections of z-stack series were obtained to analyze the cilia. At 48 hpf, yap1^−/−;wwtr1^+/− fish exhibited pronephric tubule dilation and abnormal cilia arrangement (Fig. 2C).

Stk3 is the core factor in the Hippo signaling pathway, which inhibits the activities of its downstream effectors Yap1/Wwtr1 through a phosphorylation cascade. When phosphorylated, Yap1/Wwtr1 translocate from the nucleus to the cytoplasm, therefore losing their functions as transcription coactivators (Yu et al., 2015). Disruption of stk3 in the mutant could possibly result in the activation of Yap1/Wwtr1. To further understand the mechanisms of Stk3 action, we went on to demonstrate the involvement of Yap1/Wwtr1 in Stk3 signaling, and we examined Yap1 first.

The WT fish and stk3^−/− mutants at 12 dpf were sampled for cryostat sectioning and immunofluorescent staining with anti-YAP1 and anti-PKCζ antibodies. The anti-YAP1 antibody has been tested in a previous study (Brandt et al., 2019). PKCζ was used as a marker for apical regions of the renal tubules. In WT controls, Yap1 was mainly localized in the cytoplasm of tubular epithelial cells without showing nuclear accumulation (Fig. 3A). Surprisingly, there was no obvious difference in stk3^−/− mutants, suggesting that Yap1 was not activated. Then, we sampled the stk3^−/− mutants without edema at 15 dpf for the same analysis, and no obvious nuclear accumulation of Yap1 could be detected at this stage (Fig. S4A). These results indicated that activation of Yap1 was not the main cause of the pronephric phenotypes in stk3^−/− mutants and that Stk3-independent mechanisms exist to regulate the subcellular location of Yap1.

To confirm the above findings, we generated stk3^−/−;yap1^−/− double mutants to study the function of Yap1 in stk3^−/− mutants. During larval stage, the stk3^−/−;yap1^−/− double mutant also exhibited edema at the same time point as the stk3^−/− mutant (Fig. 3B). Histological analysis revealed that renal tubule dilation and glomerular cyst formation in the stk3^−/−;yap1^−/− double mutant were comparable to those in the stk3^−/− mutant (Fig. 3C). Our observations suggested that the loss of Yap1 did not rescue or alleviate the pronephric phenotypes of the stk3^−/− mutant.

To further investigate the function of Yap1 in the zebrafish kidney, we overexpressed Yap1 in zebrafish. We cloned the coding sequence (CDS) of yap1 from cDNA of zebrafish and introduced a mutation, which converted the 87th amino acid (conserved with mammalian YAP1 S127) from serine to alanine, which prevents potential phosphorylation. The expression of Yap1^S87A was driven by the cytomegalovirus (CMV) promoter (Fig. 3D). A stable transgenic line of CMV:Yap1^S87A was generated with the Tol2 transposon system and we named it as Yap1 OE in this study. To confirm the functionality of Yap1 OE, we crossed this line with yap1;wwtr1 double mutant. Yap1 OE was able to rescue the phenotypes of the yap1^−/−;wwtr1^+/− fish (Fig. S4F). Before 5 dpf, ∼10-20% Yap1 OE embryos exhibited multiple defects, including ventral body curvature and pericardial edema (Fig. S4B,C), suggesting that Yap1 might have functions in early development. After removing these abnormal embryos at 5 dpf, the remaining fish were raised in the standard conditions, and no fish exhibited edema from larval stage to adult stage. Yap1 OE fish were identified through GFP signal (Fig. S4E). Histological analysis of adult fish showed no obvious dysfunction in the kidney of Yap1 OE fish (Fig. 3E).

To further confirm this, we tested the effects of verteporfin, an inhibitor of the YAP1-TEAD complex, on the stk3^−/− mutant. The progeny of stk3^+/− in-cross were treated with different doses of verteporfin (1, 2 and 5 μM). As stk3^−/− mutants normally exhibited edema between 12 dpf and 18 dpf, the treatment was initiated at 8 dpf and ended at 20 dpf. The results showed that verteporfin treatment did not influence the edema phenotype in stk3^−/− mutants (Fig. 3F).

Together, these results suggested that Yap1 was not activated in the renal tubular epithelial cells of stk3^−/− mutants and was therefore not responsible for the pronephric phenotypes of the stk3^−/− mutants.

After excluding the involvement of Yap1, we then investigated the activity of Wwtr1 in stk3^−/− mutants. We first performed immunostaining to examine the subcellular location of Wwtr1. The stk3^−/− mutants and WT siblings at 12 dpf were sampled for cryostat sectioning and immunofluorescent staining with anti-YAP/TAZ antibody labeling Wwtr1, anti-PKCζ antibody labeling apical regions of the renal tubules and DAPI labeling cell nuclei. The anti-YAP/TAZ antibody was confirmed to be specifically detecting Wwtr1 in zebrafish in a previous study (Kimelman et al., 2017). In WT controls, Wwtr1 was mainly localized in the cytoplasm of tubular epithelial cells, showing little nuclear accumulation (Fig. 4A). However, in stk3^−/− mutants, Wwtr1 was strongly localized in the nuclei of tubular epithelial cells, suggesting activation of Wwtr1 in stk3^−/− mutants.

Although Wwtr1 was activated in stk3^−/− mutants, we could not conclude that its activation was responsible for the pronephric phenotypes of the stk3^−/− mutant. To provide further evidence, we generated stk3^−/−;wwtr1^−/− double mutants to study the involvement of Wwtr1 in the stk3^−/− mutants. During larval stage, the stk3^−/−;wwtr1^−/− double mutants exhibited a significantly delayed onset of edema compared to stk3^−/− single mutants (Fig. 4B). The median times of onset of edema were 21 dpf in the stk3^−/−;wwtr1^−/− double mutants, 17 dpf in the stk3^−/−;wwtr1^+/− mutants and 14.5 dpf in the stk3^−/−;wwtr1^+/+ mutants, showing a clear dose-dependent pattern (Fig. 4C). Histological analysis of the stk3^−/−;wwtr1^−/− double mutant at 21 dpf revealed mild renal tubule dilation and glomerular cyst formation (Fig. 4D). Quantification of the renal corpuscle areas showed significantly smaller renal corpuscles in stk3^−/−;wwtr1^−/− double mutants than in stk3^−/−;wwtr1^+/+ single mutants (Fig. 4E). As the yap1^−/−;wwtr1^−/− double mutants exhibited severe defects during embryonic development (Fig. 2A), we were unable to obtain a stk3^−/−; yap1^−/−;wwtr1^−/− triple mutant for further study.

Our observations suggested that the absence of Wwtr1 can significantly alleviate the renal abnormalities of stk3^−/− mutants, suggesting its involvement in the phenotypic development of the stk3^−/− mutants.

To further investigate the function of Wwtr1 in zebrafish kidney development, we generated a transgenic line overexpressing constitutively active Wwtr1 (Wwtr1^S79A). The CDS of wwtr1 was cloned with a mutation introduced, which converted the 79th amino acid (conserved with mammalian WWTR1 S89) from serine to alanine to prevent potential phosphorylation. The expression of Wwtr1^S79A was driven by the CMV promoter (Fig. 5A), and a stable transgenic line of CMV:Wwtr1^S79A was generated with the Tol2 transposon system, which we named Wwtr1 OE. To confirm the function of Wwtr1 OE, we crossed this line with the yap1;wwtr1 double mutant. Wwtr1 OE could rescue the phenotype of the yap1^−/−;wwtr1^+/− fish (Fig. S5A).

Before 5 dpf, ∼10-40% of Wwtr1 OE embryos exhibited multiple defects including ventral body curvature and pericardial edema (Fig. 5B,C). After removing these abnormal embryos at 5 dpf, the remaining fish were raised under standard conditions. During their development, some fish exhibited edema at different ages from larval to adult stage, with the median time of onset of edema being 18 dpf (Fig. 5D,E). The survival curve showed that the median survival time was 21 dpf (Fig. 5F). Some of the Wwtr1 OE fish could survive to adulthood.

During larval stage, the fish with edema were sampled for phenotype analysis. Hematoxylin and Eosin (H&E) staining revealed that Wwtr1 OE larvae exhibited renal tubule dilation and glomerular cyst formation with enlarged Bowman's space, which were nearly identical to the phenotypes of stk3^−/− mutants (Fig. 5G; Fig. S5C). These findings provided further evidence that the activation of Wwtr1 was responsible for the renal phenotypes in stk3^−/− mutants.

Then, we further investigated the phenotypes in surviving adults. Wwtr1 OE fish were identified through GFP signal (Fig. S5D). Dissection of adult Wwtr1 OE fish revealed abnormal kidneys (Fig. 6A). The normal kidneys in WT fish are flattened organs attached tightly to the dorsal side of the body cavity and covered by abundant pigment cells (melanocytes). In comparison, the kidneys of Wwtr1 OE fish were much larger and thicker with bulges on the surface, making it much easier to separate them from the body wall. They also showed much fewer melanocytes (Fig. 6A). Reno-somatic index (RSI) analysis showed almost three times higher RSI in the Wwtr1 OE fish than in the WT control (Fig. 6B). Histological analysis revealed that Wwtr1 OE fish exhibited severe renal tubule dilation and massive cyst formation in the kidney, which is similar to the mammalian PKD phenotype (Fig. 6D; Fig. S6A,B). Quantitative analysis showed an average ∼40% cystic index in Wwtr1 OE fish, much higher than that in the WT control (Fig. 6C). Both glomerular cysts and renal tubular cysts could be detected (Fig. 6D). Thus, overexpression of active Wwtr1 induced cyst formation in both pronephros and mesonephros of zebrafish.

The Hippo signaling pathway has long been studied for its roles in organ size control through regulating cell proliferation. In mammals, cyst formation has been found to be associated with increased cell proliferation in PKD individuals (Nadasdy et al., 1995). To address this issue, we examined cell proliferation rate in Wwtr1 OE fish. Kidneys of Wwtr1 OE and WT fish were sampled for cryostat sectioning and immunostaining with anti-PH3 labeling phospho-histone H3 (a mitosis marker) to demonstrate cell proliferation (Fig. 7A). A significantly higher proliferation rate in tubular epithelial cells was observed in Wwtr1 OE fish compared to WT controls at both 60 dpf and 90 dpf (Fig. 7B).

Studies have shown that cilia play important roles in renal cyst formation, and dysfunction of cilia has been associated with the pathology of PKD (Ma et al., 2017). The MST1/2-SAV1 complex was reported to promote ciliogenesis, and the loss of MST1/MST2 or activation of YAP1/WWTR1 mediated cilium disassembly in mammalian cell lines (Kim et al., 2015, 2014). To investigate whether Wwtr1 OE fish exhibited abnormalities in renal cilia, we performed cryostat sectioning and immunostaining with anti-ace-Tub to label cilia in the kidney at 60 dpf and 90 dpf. Maximal intensity projections of z-stack series were obtained to analyze the cilia in the kidney. In the kidney of WT controls, the cilia in the kidney were derived from epithelial cells and stretched to the lumen of renal tubules. As zebrafish renal tubules have both single ciliated cells and multiciliated cells, we counted the total numbers of cilia in renal tubules and divided them by total cell numbers. Significantly fewer cilia per cell were observed in the renal cysts of Wwtr1 OE fish compared to those of WT controls (Fig. 7C,D). However, the cilia length did not show obvious difference between Wwtr1 OE fish and WT controls (Fig. 7E).

Zebrafish is widely used as a vertebrate model in drug development. In this study, we found that stk3^−/− mutants exhibited renal tubule dilation and glomerular cyst formation, which was mediated through Wwtr1 activation, at least partially. Furthermore, Wwtr1 OE fish exhibited PKD phenotypes, which were similar to those in mammals. To evaluate whether stk3^−/− mutants could be used as a PKD model for drug development, we tested the sensitivity of stk3^−/− mutants to different drugs. As stk3^−/− mutants exhibited edema between 12 dpf and 18 dpf, all treatments were initiated at 8 dpf and ended at 20 dpf. None of the drugs in our study had obvious effects on edema onset in stk3^+/+ controls.

Previous studies have reported that corticosteroids can induce PKD in mammalian models (McDonald et al., 1990). Based on this, we tested whether stk3^−/− mutants were sensitive to treatments with corticosteroid agonists, dexamethasone and prednisolone, and the doses of treatments were as used in a previous study (Jobst-Schwan et al., 2019). The results showed that dexamethasone treatment aggravated the edema phenotype in stk3^−/− mutants in a dose-dependent manner. The median time of onset of edema was 15.0 dpf in the dimethyl sulfoxide (DMSO) group, and 14.5, 14.0 and 13.5 dpf in the 100, 500 and 1000 nM dexamethasone treatment groups, respectively (Fig. 8A). Similarly, prednisolone treatment also aggravated the edema phenotype in stk3^−/− mutants. The median time of onset of edema was 15.3 dpf in the DMSO group, and 14.0, 13.5 and 13.0 dpf in the 10, 50 and 100 μM prednisolone treatment groups, respectively (Fig. 8B). These results suggested that stk3^−/− mutants were sensitive to corticosteroid treatments.

The mTOR pathway has been reported to participate in cyst growth in mammals, and inhibition of the mTOR signaling reversed the renal cystogenesis (Shillingford et al., 2006; Tao et al., 2005; Wahl et al., 2006). To test whether inhibition of the mTOR pathway has similar effects on renal cystogenesis in the zebrafish, we treated fish with rapamycin, the mTOR pathway inhibitor, and the doses of treatments were as used in a previous study (Yin et al., 2016). As expected, rapamycin treatment significantly delayed the onset of edema in the stk3^−/− mutants. The median time of onset of edema was 14.5 dpf in the DMSO group, and 15.5, 15.75 and 16.5 dpf in the 10, 20 and 50 nM rapamycin treatment groups, respectively (Fig. 8C).

Histone deacetylase 6 (HDAC6) has also been reported to be involved in renal cyst formation in mammals (Ke et al., 2018). To test whether inhibition of HDAC6 reduces cystic cell proliferation and cyst growth, we treated fish with tubastatin A, an HDAC6 inhibitor. As expected, tubastatin A treatment significantly delayed the onset of edema in the stk3^−/− mutants. The median time of onset of edema was 15 dpf in the DMSO group, and 15.5, 16 and 18.5 dpf in the 100, 500 and 1000 nM tubastatin A treatment groups, respectively (Fig. 8D).

In the past decade, gene-editing tools, especially transcription activator-like effector nucleases (TALENs) and CRISPR/Cas9, have been widely used for studying gene functions in zebrafish. Several groups have generated mutants related to the Hippo signaling pathway focusing on different functions, including early embryonic development (Kimelman et al., 2017), liver development (Yi et al., 2018), eye development (Miesfeld et al., 2015), blood vessel development (Astone et al., 2018; Nakajima et al., 2017), micropyle formation (Dingare et al., 2018; Yi et al., 2019) and cardiac wall maturation (Lai et al., 2018). However, the function of the Hippo signaling pathway in the zebrafish renal system has not been studied using a gene-knockout approach.

In Drosophila, disruption of the Hippo pathway mainly led to hyperproliferation and enlarged organ size, with little information available on renal development (Harvey et al., 2003; Huang et al., 2005; Justice et al., 1995). In a mouse model, podocyte-specific knockout of Yap1 induced proteinuria and focal segmental glomerulosclerosis (Schwartzman et al., 2016). Deletion of Yap1 in mouse cap mesenchyme led to impaired nephron induction and neonatal death (Reginensi et al., 2013). Furthermore, mice with nephric duct-specific deletion of Yap1 exhibited defects in ureter-bladder junction (Reginensi et al., 2015), and renal tubule-specific Mst1/2 deletion induced hyperproliferation of renal tubular epithelial cells and renal fibrosis (Xu et al., 2020).

In zebrafish, some studies have been published concerning the roles of the Hippo signaling pathway in pronephric development. Morpholino-mediated knockdown of wwtr1 in zebrafish induced cystic dilatation of the pronephric tubules and changed their proximodistal patterning (Zhang et al., 2015). Also, Scrib was found to regulate renal cyst formation through Yap1, and either knockdown or overexpression of Yap1 induced cyst formation in zebrafish (Skouloudaki et al., 2009; Xu et al., 2018). In another study, morpholino knockdown of Yap1 in zebrafish resulted in pronephric cyst formation with ciliary defects (He et al., 2015). Although these studies provided some information concerning the involvement of the Hippo signaling pathway in renal development, they were mostly based on a morpholino-mediated knockdown approach. Morpholino knockdown was previously the most commonly used reverse genetics approach in zebrafish, owing to its efficiency and convenience, prior to the emergence of the genome-editing approaches (i.e. TALEN and CRISPR/Cas9). However, this approach has some shortages: the effects of morpholinos cannot last for longer than 5 days and are not heritable. Furthermore, recent studies have reported that ∼80% of morphant phenotypes were not observed in knockout mutants generated with a TALEN or CRISPR/Cas9 system (Kok et al., 2015). Despite this, morpholino-based gene knockdown is still being used as an ancillary tool for knockout approaches (El-Brolosy et al., 2019; Ma et al., 2019; Rossi et al., 2015).

In this study, we induced both gene knockout and gene knockdown through the CRISPR/Cas9 system. Our knockdown results in yap1 and wwtr1 crispants were consistent with previous studies (He et al., 2015; Zhang et al., 2015). Because the phenotypes of yap1 and wwtr1 crispants could be rescued by both yap1 and wwtr1 mRNAs, we believe that the CRISPR/Cas9-mediated knockdown was specific, without off-target effects. Interestingly, although the double mutant of yap1 and wwtr1 (yap1^−/−;wwtr1^−/−) showed embryonic lethality, demonstrating functional importance of the Hippo pathway in development, single knockout mutants of yap1 and wwtr1 did not exhibit pronephros defects as seen in their crispants. Examination of yap1 and wwtr1 double mutants suggested functional compensation between Yap1 and Wwtr1 in pronephros development. The loss of Wwtr1 could be fully compensated by both yap1^+/+ and yap1^+/−, whereas the loss of Yap1 could be rescued fully by wwtr1^+/+ but only partially by wwtr1^+/−, suggesting a gene-dose effect for wwtr1. The yap1^−/−;wwtr1^+/− fish exhibited formation of pronephric cysts, which is similar to observations in yap1 and wwtr1 crispants. The functional compensation or redundancy revealed by yap1 and wwtr1 double mutants can well explain the lack of phenotypes in yap1 and wwtr1 single mutants, but not the phenotypes of their crispants. Such discrepancy between knockdowns and knockouts has been reported in a variety of organisms especially zebrafish (Rossi et al., 2015), but there is no clear explanation for the mechanism at this moment. Genetic compensation in response to gene mutations has been proposed for the reduced penetrance of mutant genes, which is a widespread phenomenon observed in a variety of organisms from yeast to humans (El-Brolosy and Stainier, 2017). Although the exact mechanism underlying genetic compensation remains poorly understood, two possibilities have been proposed: DNA damage in the mutant gene and mutated mRNA transcribed from the mutant gene. The DNA damage or lesion in the mutant may induce global chromatin reorganization, and the mutant mRNA may trigger mRNA decay pathways to produce RNA fragments, both of which may induce transcriptional adaptation response that results in compensation of the lost gene functions. Because the triggers of both mechanisms work upstream of protein translation, such genetic compensation occurs in knockouts, but not knockdowns (El-Brolosy et al., 2019). A recent study in zebrafish showed that the mutant of the slc25a46 gene showed no phenotype whereas its crispant displayed severe disease symptoms. The discrepancy was due to genetic compensation that did not involve mRNA decay (Buglo et al., 2020). This observation was similar to our results on yap1 and wwtr1 mutants and crispants. Although we do not have evidence for the exact mechanism underlying genetic compensation in yap1 and wwtr1 single knockouts, changed expression of yap1 and wwtr1 is likely part of the transcriptional adaptation response for each other, as suggested by the double mutants.

For renal cystogenesis, we found that stk3^−/− mutants exhibited pronephric dysfunctions including renal glomerular cyst formation and tubule dilation, suggesting an important role for Stk3 in zebrafish renal development. Further evidence showed that Wwtr1, but not Yap1, was involved in phenotypic development of the stk3^−/− mutants. Wwtr1 OE fish phenocopied stk3^−/− mutants at larval stage and developed PKD symptoms at adult stage. Together, our findings suggest that loss of Stk3 in zebrafish induced pronephric dysfunctions through activation of Wwtr1 instead of Yap1, and these dysfunctions resulted in the symptoms of early-stage PKD.

The Hippo signaling pathway has been implicated in renal diseases including PKD. In human ADPKD patients, strong nuclear YAP1 accumulation was observed in the cyst-lining cells (Happé et al., 2011). Gene set enrichment analysis revealed upregulation of YAP1 target genes in human ADPKD samples (Cai et al., 2018). In mice, deletion of Wwtr1 resulted in PKD (Hossain et al., 2007; Makita et al., 2008). A recent study demonstrated that YAP1 was the direct downstream target of PKD1 mutations in ADPKD patients, and YAP1 was found to accumulate in nuclei of the tubular epithelial cells in a Pkd1 knockout mouse (Cai et al., 2018). Loss of Yap1/Wwtr1 dramatically reduced cyst formation induced by Pkd1 deletion (Cai et al., 2018; Lee et al., 2020). As an upstream scaffold protein, kidney-specific deletion of Sav1 in mice induced hyperproliferation of renal tubular epithelial cells and cyst formation through activating YAP1 (Kai et al., 2016). These studies provided solid evidence for important roles of the Hippo signaling pathway in renal cyst formation.

In the present study on zebrafish model, global deletion of stk3 also resulted in the development of PKD. We further demonstrated the involvement of Wwtr1, but not Yap1, in cystogenesis. Although global deletion of wwtr1 did not induce PKD in zebrafish, overexpression of its active form caused severe renal cyst formation. Although the Hippo signaling pathway is clearly involved in PKD of both mammals and zebrafish, the regulation of the pathway seemed different in these models. In mammals, YAP1 and WWTR1 seem to function redundantly in cystogenesis. However, it was the activation of Wwtr1, but not Yap1, that served as the main effector for cystogenesis in zebrafish. Despite this difference, evidence of the lost function of the Hippo signaling pathway in PKD development is obviously similar between mammals and zebrafish, including cyst formation in renal tubules, hyperproliferation of renal tubular epithelial cells and cilia dysfunction. Increased proliferation of renal tubular epithelial cells seems to be the major cause of renal cyst formation (Nadasdy et al., 1995). As many receptors are located on the cilia, cilia dysfunction can impair signal transduction and cell homeostasis, which can accelerate cyst formation (Ma et al., 2017). Together, our results provide additional evidence for the function of the Hippo signaling pathway in renal cyst formation in vertebrates.

The zebrafish has become a popular model for studying human genetic diseases. Zebrafish and human genomes have high orthology (Howe et al., 2013). High fecundity, small body size, ex utero development, transparent embryos and low maintenance cost make zebrafish an excellent model for high-throughput screening. Recently, the application of gene-editing tools in zebrafish has enabled us to target different genes for disease modeling (Zhu and Ge, 2018).

Studies so far on PKD modeling in zebrafish have adopted a few zebrafish lines that form renal cysts in pronephros at embryonic stage (Gehrig et al., 2018; Mangos et al., 2010; Obara et al., 2006; Poureetezadi and Wingert, 2016). In the present study, we generated a Stk3-deficient zebrafish line that could serve as a potential model for PKD. In particular, this fish mutant line could be used as a tool for drug development, as demonstrated by our pharmacological experiments. We demonstrated that the median time of onset of edema was a reliable criterion to assess drug effects. With this model, we can also investigate the possible pharmacological mechanisms underlying drug actions. Our results showed that treatment of the stk3^−/− mutant with corticosteroids aggravated the edema phenotype, which might be due to changed threshold of corticosteroid tolerance in the stk3 mutant. Previous studies reported that the mTOR pathway can be activated by YAP/TAZ in the kidney (Gui et al., 2018). We showed that inhibition of the mTOR pathway can alleviate the edema phenotype in the stk3^−/− mutant. Treatment with tubastatin A (an HDAC6 inhibitor) delayed the onset of edema in the stk3^−/− mutant. As the stk3^−/− mutant developed edema through the activation of Wwtr1, which led to cell proliferation, inhibition of proliferation by the HDAC6 inhibitor reduced cystic formation.

The Wwtr1 OE line we created was a novel zebrafish line that exhibited PKD phenotypes in mesonephros at adult stage. This transgenic line could serve as a model to investigate the progression of PKD. In the future, we may cross the Wwtr1 OE line with the transparent casper line (White et al., 2008) and lines with kidney-specific expression of fluorescent proteins (such as cdh17:mCherry) (Diep et al., 2011) to generate PKD models with transparent body and kidney-specific fluorescence, which may enable us to monitor the progression of PKD in vivo.

In summary, we have provided solid evidence, using a genetic approach, for involvement of the Hippo signaling pathway in zebrafish renal development. Furthermore, we established two novel zebrafish PKD models for both embryonic or larval stage and adults, which have great potential not only for investigating the disease and its progression during development, but also for drug screening and therapy development.

All the experiments in this study were performed on AB strain zebrafish. The larvae were first raised in a Forma Environmental Chamber (Model 3949; Thermo Fisher Scientific) and then maintained in a ZebTEC Multilinking Rack Zebrafish System (Tecniplast, Buguggiate, Italy), with the photocycle of 14 h light and 10 h dark. The temperature, pH and conductivity of the system water were 28±1°C, 7.5 and 400 µS/cm, respectively. All experiments were performed according to the protocols approved by the Research Ethics Panel of the University of Macau.

The CRISPR/Cas9 system was applied to generate the zebrafish mutant lines according to the protocols reported previously (Lau et al., 2016). The targets were designed with ZIFIT Targeter (http://zifit.partners.org/zifit). Oligonucleotides synthesized were annealed and inserted into the pDR274 vector. The sgRNAs and Cas9 RNA were prepared using MEGAscript T7 and mMESSAGE mMACHINE SP6 kits (Life Technologies, Carlsbad, CA, USA), respectively. The mixture (4.6 nl) of sgRNA (20 ng/μl) and Cas9 mRNA (200 ng/μl) was co-injected into one-cell-stage embryos with a Nanoject system (Drummond, Broomall, PA, USA) to generate the F0 mutant fish. F0 fish were raised and those carrying mutations were outcrossed with WT fish to obtain F1 fish. The oligonucleotides used in this study are listed in Table S1.

The CRISPR/Cas9 system was applied to knock down the target genes. The target site design and mRNA preparation were performed as described above. The mixture (4.6 nl) of sgRNA (20 ng/μl) and Cas9 mRNA (200 ng/μl) was co-injected into one-cell-stage embryos with the Drummond Nanoject system to generate the knockdown crispants. To rescue the phenotypes of crispants, yap1 mRNA (100 pg) or wwtr1 mRNA (24 pg) was prepared using mMESSAGE mMACHINE SP6 kits and co-injected with sgRNA and Cas9 mRNA. The amounts of mRNA injected were based on previous studies (He et al., 2015; Zhang et al., 2015).

The CDS of yap1 was cloned from zebrafish cDNA and a mutation was introduced using overlap PCR, which converted the 87th amino acid from serine to alanine (Yap1^S87A). This sequence was inserted into the pSK-GFP (Tol2-CMV-GFP-pA-CMV-MCS-pA-Tol2) vectors using a Gibson Assembly kit (NEB, Ipswich, MA, USA). The mixture (4.6 nl) of Tol2 transposase mRNA (50 ng/μl) and Wwtr1 overexpression vector (20 ng/μl) was co-injected into one-cell-stage embryos to generate the Yap1 OE line. The positive embryos were screened by monitoring the GFP signals. The Wwtr1 OE line was generated with the same approach.

Zebrafish larvae and adult fish were anesthetized by MS222 (Sigma-Aldrich, St Louis, MO, USA) and imaged with a SMZ18 stereomicroscope (Nikon, Tokyo, Japan) or digital camera (Canon EOS 700D). Then, the whole fish or dissected kidneys were fixed in Bouin's fixative for 24 h before processing. After standard histological processing, the samples were embedded in paraffin and sectioned at 5 μm for H&E staining. The slides were viewed under an ECLIPSE Ni-U microscope (Nikon) and imaged with a Digit Sight DS-Fi2 digital camera (Nikon).

For whole-mount immunofluorescence, zebrafish larvae were fixed in 4% paraformaldehyde (PFA) overnight at 4°C and processed for immunostaining as described previously (Hammond-Weinberger and ZeRuth, 2020).

Zebrafish larvae or kidneys of adult fish were fixed in 4% PFA overnight at 4°C and then incubated in 30% sucrose overnight. The samples were embedded in optimal cutting temperature (OCT) medium and sectioned at 10 µm on a CM5030 cryostat (Leica, Wetzlar, Germany). After being air-dried overnight, the slides were washed with PBS for 3×5 min. The slides were placed in a humid chamber and incubated with block buffer (5% normal horse serum, 0.1% Triton X-100 in 1× PBS) overnight at 4°C. The following day, the slides were incubated with primary antibodies diluted with block buffer overnight at 4°C, incubated with secondary antibodies for 2 h and mounted with ProLong Gold Antifade Reagent (Invitrogen). Images were acquired using a Nikon A1MP+ fluorescence confocal microscope and analyzed with Fiji software (ImageJ). The antibodies used are listed in Table S2.

For sample preparation, zebrafish embryos at 5 dpf were homogenized and lysed in RIPA buffer (Thermo Fisher Scientific) on ice. The supernatant was collected and heated at 95°C for 10 min with loading buffer. The supernatant was used for western blot analysis according to the standard protocol (Li and Ge, 2011). The antibodies used are listed in Table S2.

Embryos were obtained with stk3^+/− in-cross and distributed randomly to different treatment groups. Drugs or vehicle control were added to fish water at 8 dpf and changed every other day. All treatments were initiated at 8 dpf and ended at 20 dpf. The larvae were monitored twice a day, and individuals with edema were recorded and genotyped. The remaining individuals were all genotyped at the end of the experiment.

Verteporfin, dexamethasone, prednisolone, rapamycin and tubastatin A were all purchased from Sigma-Aldrich and dissolved in DMSO as stocks. All treatment and vehicle control groups contained 0.001% DMSO.

Kaplan–Meier plots for the onset of edema and survival were analyzed using the log-rank (Mantel–Cox) test. All other values in this study were expressed as mean±s.e.m., and statistical significance was analyzed by one-way ANOVA or unpaired Student's t-test using Prism (GraphPad, San Diego, CA).

We thank the Histology Core of the Faculty of Health Sciences for technical support.