The development of the kidney relies on the establishment and maintenance of a precise tubular diameter of its functional units, the nephrons. This process is disrupted in polycystic kidney disease (PKD), resulting in dilations of the nephron and renal cyst formation. In the course of exploring G-protein-coupled signaling in the Xenopus pronephric kidney, we discovered that loss of the G-protein α subunit, Gnas, results in a PKD phenotype. Polycystin 1, one of the genes mutated in human PKD, encodes a protein resembling a G-protein-coupled receptor. Furthermore, deletion of the G-protein-binding domain present in the intracellular C terminus of polycystin 1 impacts functionality. A comprehensive analysis of all the G-protein α subunits expressed in the Xenopus pronephric kidney demonstrates that polycystin 1 recruits a select subset of G-protein α subunits and that their knockdown – as in the case of Gnas – results in a PKD phenotype. Mechanistically, the phenotype is caused by increased endogenous G-protein β/γ signaling and can be reversed by pharmacological inhibitors as well as knocking down Gnb1. Together, our data support the hypothesis that G proteins are recruited to the intracellular domain of PKD1 and that this interaction is crucial for its function in the kidney.
The kidney is responsible for the excretion of waste products, but also for controlling water and solute homeostasis (Smith, 1953; Saxén, 1987; Vize et al., 2003). The performance and longevity of the kidney is contingent on the number and correct development of its functional units, the nephrons (Bertram et al., 2013). Nephrons form in a highly stereotypical fashion, generating several functionally distinct cell types along its proximal-distal axis (Desgrange and Cereghini, 2015; McMahon, 2016). In addition, nephrons are characterized by a precisely defined tubule diameter that maximizes channel, transporter and receptor-mediated water and solute reabsorption (Fischer et al., 2006; Karner et al., 2009; Lienkamp et al., 2012; Wessely et al., 2014).
The control of tubule diameter is disrupted in polycystic kidney diseases. This group of genetically inherited human disorders is characterized by the formation of fluid-filled kidney cysts as a result of a disrupted nephron diameter control, ultimately leading to end-stage renal disease (Harris and Torres, 2009; Grantham et al., 2011; Chebib and Torres, 2016). The genes mutated in its most common form, autosomal dominant polycystic kidney disease (ADPKD), are PKD1 and PKD2 (Hughes et al., 1995; The International Polycystic Kidney Disease Consortium, 1995; Mochizuki et al., 1996). Polycystin 1 (PKD1) encodes a large transmembrane protein and polycystin 2 (PKD2) is a member of the transient receptor potential (TRP) superfamily. The two proteins are believed to interact and regulate intracellular Ca2+ levels. However, the molecular mechanisms underlying the cellular changes that lead to cyst formation and expansion in response to loss of PKD1 are not well understood. Recent studies have focused on the intracellular C-terminal tail of polycystin 1, which binds multiple signaling proteins and is subjected to proteolytic cleavage, generating a fragment that can enter the nucleus (Chauvet et al., 2004; Low et al., 2006; Merrick et al., 2012). Another intriguing aspect of PKD1 is that it has hallmarks of a G-protein-coupled receptor (GPCR). Although it lacks the typical seven transmembrane domain configuration, the intracellular C-terminus harbors a G-protein binding domain (GBD), which binds to multiple G α subunits (Parnell et al., 1998, 2002; Delmas et al., 2002; Yu et al., 2010, 2011). Moreover, the extracellular N terminus contains an evolutionarily conserved GPCR auto-proteolysis site (GPS) involved in PKD1 sorting and function (Cai et al., 2014; Kurbegovic et al., 2014; Qian, 2015; Su et al., 2015; Trudel et al., 2016). In fact, the GPS motif is part of a larger domain, the GPCR autoproteolysis-inducing (GAIN) domain (Prömel et al., 2013; Trudel et al., 2016). This domain is one of the defining features of the adhesion GPCRs, a subgroup of at least 33 GPCRs in humans, the overarching function of which is still poorly described although they may act as autonomous adhesive and signaling units (Prömel et al., 2013). Despite these efforts, it is still unclear whether PKD1 indeed functions as a GPCR.
The molecular mechanisms underlying PKD are studied in many model organisms (Torres and Harris, 2007; Happé and Peters, 2014). The African clawed frog, Xenopus laevis, has emerged as a powerful alternative system due to the formation of functional kidneys 48 h post fertilization and the ease with which high-throughput knockdown studies can be performed (Wessely and Obara, 2008; White et al., 2010; Wessely and Tran, 2011). Here, we have used these advantages to investigate the function of polycystin 1 (Pkd1). We now demonstrate that the crucial feature of polycystin 1 with respect to its loss-of-function phenotype is its ability to bind trimeric G proteins. In a systematic approach examining the entire family of G-protein α subunits, we show that Pkd1 binds a subset of G-protein α subunits and restricts their signaling ability. In the absence of Pkd1, G-protein α- and β/γ-dependent signals become hyper-activated, contributing to the complex cellular phenotype seen in cystic renal epithelial cells. These data provide a novel angle towards treating ADPKD.
Gnas knockdown results in a PKD phenotype
The Xenopus PKD phenotype is characterized by the formation of edema due to impaired kidney function, dilated pronephric tubules and ducts, and the loss of the expression of the sodium bicarbonate co-transporter Nbc1 in the late distal tubule (Tran et al., 2007, 2010). In the course of examining the effect of cholera toxin on trimeric G-protein signaling during proximal tubular growth (Zhang et al., 2013), we discovered that knockdown of Gnas, but not Gnal, resulted in a PKD phenotype very reminiscent of the one observed when the PKD genes polycystin 1 or polycystin 2 were knocked down (Tran et al., 2007, 2010; Xu et al., 2015; Kim et al., 2016). Microinjection of an antisense morpholino oligomer targeting Gnas (Gnas-MO) caused edema, the dilation of the pronephric tubules and ducts and the loss of the expression of the sodium bicarbonate co-transporter Nbc1 in the late distal tubule (Figs 1A-F, 2A-C and Fig. S1). Interestingly, one of the many hypotheses for the mechanism of action of polycystin 1 (PKD1) is that it acts as an atypical G-protein-coupled receptor (GPCR) (Parnell et al., 1998; Yuasa et al., 2004; Hama and Park, 2016). A sequence comparison between Xenopus and human PKD1 revealed that the previously described G-protein-binding domain (GBD) is more than 90% conserved at the amino acid level between human and frog (Fig. 1G and Fig. S2A). Thus, we decided to address whether the Gnas morphant phenotype could be explained by a direct link between trimeric G-proteins and PKD1. To assess this, we used a paradigm previously reported in zebrafish (Merrick et al., 2012), i.e. the rescue of the Pkd1 morphant phenotype by the Pkd1 C-terminal tail. As we have shown (Xu et al., 2015; Kim et al., 2016), eliminating Pkd1 in Xenopus embryos using an antisense morpholino oligomer (Pkd1-sMO) results in a PKD phenotype (Fig. 1H,I,L and data not shown). As expected, co-injection of synthetic mRNA encoding the intracellular C-terminal tail of Xenopus Pkd1 (Pkd1-CTT) was sufficient to rescue the late distal tubular expression of Nbc1 (Fig. 1J,L). To address, whether the GBD is involved, we next deleted the five core amino acids of the Pkd1 GBD, RKKVR (Fig. S2A), and co-injected this construct (Pkd1-CTTΔGBD) with the Pkd1-sMO. As shown in Fig. 1K,L, this mRNA could not restore the staining of Nbc1 in the late distal tubule. The sequence of the Pkd1 GBD overlaps with nuclear localization sequence (NLS) described by Chauvet et al. (2004) (Fig. S2A). Thus, we wondered whether Pkd1-CTTΔGBD exhibits changes in the subcellular localization. HEK293T cells were transiently transfected with a Myc-tagged version of Pkd1-CTT and Pkd1-CTTΔGBD, and processed for immunofluorescence analyses. As shown in Fig. S2B,C, both proteins localized to the nucleus. This implies that the G-protein-binding activity and not the nuclear localization of Pkd1-CTT is crucial for its rescue of the Pkd1 morphants. Together, these data suggest that an interaction between Gnas and Pkd1 may be a crucial aspect of Pkd1 function in the kidney.
Comprehensive analysis of G-protein α subunits
Previous studies have demonstrated that PKD1 can bind multiple G-protein α subunits (Parnell et al., 1998, 2002; Delmas et al., 2002; Yu et al., 2010, 2011). To better understand how Pkd1 could modulate trimeric G-protein signaling in the Xenopus kidney, we decided to identify which G-protein α subunits are expressed in the pronephric kidney. To achieve this, we systematically analyze the expression of all G-protein α subunits. RNAseq data indicated that the entire G-protein α family is expressed in stage 39 Xenopus embryos (Fig. S3A). The only exception was Gna12, which we could not identify in the Xenopus genome. Next, we performed whole-mount in situ hybridization covering all phases of Xenopus kidney development (Fig. S3B-O′ and data not shown). As expected, the G-protein α family exhibited very diverse expression patterns, with some members appearing rather ubiquitous, while others were much more restricted (e.g. Gnat2 staining in the eye and the pineal gland). However, with respect to the kidney, only Gnao and Gnat2 could not be detected in the developing pronephros.
We next focused on these G-protein α subunits and designed antisense morpholino oligomers (MOs) against each of them that targeted, if possible, both homeologs with a single MO (with the exception of Gna14, which required two, see Table S1). MOs were validated by in vitro transcription/translation (Fig. S4). As in the case of Gnas, the MOs were injected individually into Xenopus embryos and analyzed for the occurrence of a PKD phenotype. Moreover, we tested for potential functional redundancy. G-protein α subunits are classified into four families based on their mode of downstream signaling (Neves et al., 2002; Dorsam and Gutkind, 2007). To account for compensation within these families, we performed combinatorial knockdowns that eliminated two or three family members simultaneously. Surprisingly, among all the combinations tested, only a small subset fulfilled all three criteria for the PKD phenotype, i.e. the occurrence of edema, dilations of the kidney tubules and loss-of-expression of Nbc1 in the late distal tubule (Fig. 2). These were the knockdown of Gnas, the knockdown of Gna14 (a Gnaq family member) and the combined knockdown of the two Gnai family members, Gnai1 and Gnai2 (Table 1).
Based on this unexpected specificity, we wondered whether a correlation exists between the PKD knockdown phenotype of G-protein α subunits and their ability to bind to Pkd1. Several G-protein α subunits have been shown to bind to PKD1 (Parnell et al., 1998, 2002; Delmas et al., 2002; Yu et al., 2010, 2011). Yet these studies have never addressed all G-protein α subunits, nor have they determined binding affinities. To achieve this, we performed surface plasmon resonance (SPR) analysis using bacterially purified Pkd1-CTT and reticulocyte lysate-produced G-protein α subunits. Only four Xenopus α subunits (Gnas, Gna14, Gnai1 and Gnai2), as well as mouse Gna12 [which has been shown to interact with Pkd1 (Yuasa et al., 2004), but has yet to be identified in the Xenopus genome] exhibited high binding affinities (Fig. 3A,A′). This interaction was independent of which partner was immobilized on the chip (Fig. S5). However, it was dependent on the G-protein-binding domain, because high-affinity binding was abolished when the Pkd1-CTTΔGBD protein was used instead (Fig. 3B,B′). Importantly, the calculated binding affinities (Fig. 3A′,B′) were in a similar range to the binding of the G-protein α subunits to GPCRs (Komolov et al., 2006). In summary, these analyses demonstrate a striking correlation between G-protein α subunits that exhibit high-affinity binding to Pkd1 and those exhibiting the PKD phenotype upon knockdown in Xenopus.
Analysis of G-protein signaling
The results so far are in agreement with the hypothesis that Pkd1 functions as a GPCR that signals via a distinct set of G proteins and activates α- and/or β/γ-dependent signaling cascades (Fig. 4A). To test this, we aimed to restore G-protein signaling in Pkd1 morphants. Morphant embryos and untreated controls were incubated with low molecular weight compounds that activate or inhibit effector molecules downstream of trimeric G proteins. As Gnas activates cyclic AMP (cAMP), we raised cAMP levels using the adenylate cyclase activator forskolin or two cAMP analogs: the PKA-specific 6-Bnz-cAMP-AM (cAMP-PKA) and the Rapgef3/4-specific 8-pCPT-2′-O-Me-cAMP-AM (cAMP-Epac). However, none of the compounds could rescue the PKD phenotype (Fig. 4C and data not shown). Similarly, mimicking Gnai family activity by inhibiting cAMP signaling using the protein kinase A (PKA) inhibitor H89, the Rapgef3/4 inhibitor brefeldin A (BFA) or by injecting the Rapgef4-MO (Zhang et al., 2013) did not have any effect (Fig. 4D). Next, we mimicked Gna14 signaling by stimulating its effector phospholipase C β (PLCβ) using m-3M3FBS, but did not observe a reversal of the Pkd1 morphant phenotype either (Fig. 4E).
G-proteins signal not only via α subunit-dependent pathways but also via β/γ subunits (Smrcka, 2008; Hama and Park, 2016). To test whether inhibiting β/γ signaling, like the loss of G-protein α signaling, induced a PKD phenotype in Xenopus embryos, we used two compounds (Fig. 4A). We either exposed embryos to the G-protein β/γ ‘hot-spot’ inhibitor gallein (Lin and Smrcka, 2011) or blocked PLCβ, a direct target of G-protein β/γ signaling, with U73122. However, neither method impacted Nbc1 staining in the late distal tubule of the kidney (Fig. 4F,G,J).
As re-activation of downstream G-protein signaling was insufficient to restore the Pkd1 morphant phenotype, we envisioned an alternative explanation: a model based on competition for G-protein β/γ subunits (Fig. 4B). It is based on the fact that trimeric G-proteins function as mixed pools of α and β/γ subunits, where the β/γ subunits are shared between the different α subunits (Smrcka, 2008). We hypothesize that the absence of Pkd1 increases the pool of free G-protein β/γ subunits. They are then available to signal either in complex with other α subunits or even by themselves. To test this hypothesis, we inhibited β/γ-dependent signaling in Pkd1 morphants and observed a very dramatic effect (Fig. 4F-J). Both gallein and U73122 restored Nbc1 expression in Pkd1-sMO-injected embryos. As gallein efficacy is quite batch dependent (see Materials and Methods) and PLCβ signaling is also regulated by Gna14 (Fig. 4A), we decided to verify the observation by directly eliminating G-protein β/γ signaling. As in the case of the G-protein α subunits, we surveyed the five known G-protein β subunits for expression in the developing pronephros (Fig. S6A,C-G′ and data not shown). Among those, Gnb1 displayed the most prominent pronephric kidney expression. An antisense morpholino oligomer targeting Gnb1 (Gnb1-MO) efficiently blocked Gnb1 mRNA translation. It did not result in any overt phenotypes when injected on its own (Figs 4J, 5E and Fig. S6B). However, when co-injected with the Pkd1-sMO, the double morphants exhibited restored Nbc1 expression and did not develop edema (Figs 4J and 5E), thus supporting our competition model.
To further strengthen this, we tested another paradigm associated with renal cyst formation: the randomization of the angle of cell division in renal tubules (Fischer et al., 2006; Nishio et al., 2010; Fedeles and Gallagher, 2013). Unfortunately, direct imaging of mitotic figures in the Xenopus pronephric kidney was impractical owing to the low number of cell divisions (Romaker et al., 2014; data not shown). Thus, we opted for a pulse-labeling approach that followed the fate of cells post cell division, analyzing whether dividing cells are found along the length of the tubule or circumferential to it. To achieve this, embryos were injected with 5-ethynyl-2′-deoxyuridine (EdU) at stage 40, fixed at stage 42 and analyzed for the presence of labeled cell doublets in transverse or sagittal sections (Fig. 5F,G). In line with data in mouse and zebrafish (Fischer et al., 2006; Bubenshchikova et al., 2012), control embryos displayed a preference of dividing cells ending up aligned along the length of the nephron (Fig. 5H). Conversely, Pkd1 or Gnas morphants displayed the opposite distribution with cells integrating rather circumferentially. Treating Pkd1-sMO or Gnas-MO-injected embryos with gallein or co-injecting the Gnb1-MO reversed the phenotype and the cell division profile was similar to the one observed for control embryos.
Based on these data, we propose that the absence of Pkd1 in Xenopus results in unbalanced G-protein signaling, which is causally involved in the formation of the amphibian PKD phenotype. In humans, ADPKD is caused by mutations in either PKD1 or PKD2, and it is generally thought that the two proteins form an interdependent complex (Torres et al., 2007). Thus, we wondered whether hyperactivation of G-proteins is a general consequence of a PKD phenotype and also underlies the Pkd2 loss-of-function phenotype. However, when we analyzed Xenopus embryos injected with a Pkd2-MO treated with or without gallein, edema formation was not rescued upon gallein exposure (Fig. 5E). Therefore, the competition model is only observed for Pkd1, but not Pkd2, suggesting that Pkd2 functions downstream or parallel of Pkd1.
Loss of Pkd1 results in hyperactivation of G-protein signaling
All experiments so far were performed using Xenopus. To test the wider applicability of our model, we decided to use a mammalian cell culture system, two isogenic proximal tubular cell lines generated from Pkd1 floxed mice (Pkd1Fl/− and Pkd1−/−) (Shibazaki et al., 2008). As previously reported (Merrick et al., 2012), these cells mimic the cystogenic potential seen in vivo. When grown under 3D culture conditions in matrigel at the non-permissive temperature of 37°C, the Pkd1Fl/− cells form tubular structures and the Pkd1−/− cells form cysts (Fig. 6A,E). In agreement with the Xenopus data, treating the Pkd1 mutant cells with the β/γ inhibitor gallein interfered with the cyst formation and resulted in a conversion to more tubular structures (Fig. 6B,F).
Upregulation of cyclic AMP (cAMP) is a central aspect of the pathogenesis of PKD and is believed to contribute both to increased fluid secretion and proliferation in cystic epithelia (Calvet, 2015). Although it is generally thought to be an indirect mechanism, our competition model (Fig. 4B) suggests that not only β/γ, but also α signaling should be activated by the absence of Pkd1. To achieve this, we exposed cells to the PKA inhibitor H89 (Fig. 6C,G). Although H89 had no discernible effect on the control cells, it reduced cyst size of the Pkd1-null cells. Moreover, the combination of both H89 and gallein was at least additive (Fig. 6D,H). Interestingly, an upregulation of PKA signaling would have been missed in Xenopus, as the inhibition of PKA does not impact kidney development (Zhang et al., 2013).
We also measured another parameter of cystogenesis: cell growth. When switched to the non-permissive temperature of 37°C, control cells barely proliferated. Conversely, Pkd1 mutant cells continued to divide (Fig. 6I,J). Addition of gallein or H89 reduced cell proliferation, with gallein being more potent than H89. However, the combination of both was most potent, blunting proliferation to the baseline levels seen in the Pkd1-expressing isogenic controls.
Next, we directly assessed whether G-protein signaling is upregulated in the absence of Pkd1. To achieve this, we interrogated the activity of Gnas or Gnai using changes in cAMP levels as readout for these two G-protein families. Activation of the Gnai family by the specific agonist somatostatin caused a concentration-dependent decrease in cAMP levels (Fig. 6K). This dose-response curve was shifted in the Pkd1−/− cells. Mutant cells were about 10-fold more sensitive to somatostatin when compared with control cells. The same hypersensitivity was observed with parathyroid hormone (PTH), which activates Gnas. A given concentration of PTH had a more pronounced effect on cAMP levels in mutant than in control cells (Fig. 6L). To verify that the latter effect was indeed due to hyperactivation of Gnas, we immunoprecipitated Gnas from cells in the presence or absence of PTH, and analyzed it using an antibody specifically recognizing activated GTP-bound Gnas. Overall levels of Gnas were identical in both cell types (Fig. 6M,N). Yet upon PTH stimulation, higher levels of activated Gnas were detected in Pkd1−/− cells than in the isogenic control cells. Together our data support the model that loss of Pkd1 changes the cellular response to GPCR-mediated signaling.
Recent structural analyses of PKD1 have suggested that Pkd1 is an aGPCR (Trudel et al., 2016). This is based on the analysis of the GPCR auto-proteolysis site (GPS), which turned out to be part of a larger domain, the GPCR autoproteolysis-inducing (GAIN) domain characteristic of adhesion GPCRs (aGPCRs), the second largest subgroup of GPCRs (Langenhan et al., 2013; Prömel et al., 2013; Trudel et al., 2016). One interesting feature of aGPCRs is that only some of them seem to signal via G-proteins (Gupte et al., 2012), instead they use a range of other signaling modalities (reviewed by Langenhan et al., 2013). This is similar to Pkd1 itself. Early studies have focused on the connection between PKD1 and G proteins, which was then supplanted by the identification and characterization of many other signaling moieties (for a review, see Harris and Torres, 2014; Hama and Park, 2016). Yet consensus has not been reached over whether G-protein-mediated signaling by PKD1 is functionally relevant.
Using the pronephric kidney of Xenopus as a model system, we now provide new evidence for the interaction between Pkd1 and G proteins. This interpretation is based on five key observations: (1) in contrast to the complete intracellular C-terminal tail, a construct lacking the G-protein binding domain (GBD) is unable to rescue Pkd1 morphants; (2) the C-terminal tail binds a select class of G-protein α subunits (i.e. Gnas, Gnai1, Gnai2, Gna12 and Gna14) at affinities in a range similar to those observed for other GPCRs; (3) loss-of-function studies for these G-protein α subunits (but none of the others) display a PKD phenotype in Xenopus embryos that is highly similar to the one observed for Pkd1; (4) the phenotype of Pkd1 morphants can be reversed by inhibiting downstream G-protein β/γ signaling, demonstrating that the interaction between Pkd1 and G proteins is functionally relevant; (5) the mechanism is evolutionarily conserved as it is observed not only in Xenopus, but also in murine renal epithelial cells that lack Pkd1. These data demonstrate that a crucial functional domain of PKD1 resides within its GBD.
One conundrum of the data presented here is that the amount of cellular polycystin 1 is believed to be rather low in comparison with the G-proteins. Thus, the observed imbalance in G-protein signaling probably only occurs when polycystin 1 is in close proximity to other GPCRs. Two such spatially confined subcellular compartments are lipid rafts and cilia. In particular, the latter has been implicated in PKD (Avasthi et al., 2017; Ma et al., 2017; Pala et al., 2017). Both GPCRs and polycystin 1 are present in cilia (Yoder et al., 2002; Schou et al., 2015; Hilgendorf et al., 2016). Moreover, the cilia localization of polycystin 1 is crucial for the development of a fully established cystic phenotype (Ma et al., 2013). Interestingly, Ma et al. (2017) proposes a cilia-dependent cyst activation (CDCA) signaling pathway, but the nature of it is still unknown. In future, it will be interesting to see whether part of the CDCA is the crosstalk between polycystin 1 and GPCRs in cilia.
One major issue not resolved in this study is whether and how PKD1 itself activates trimeric G-protein signaling activity. aGPCRs can be activated by multiple means that mostly involve the cleavage of the protein at the GPS site and activities residing either in the N- or the C-terminal fragment (Langenhan et al., 2013; Prömel et al., 2013). In the kidney, fluid flow has long been thought to be the primary trigger for PKD1 activation, but other mechanisms, such as polycystin 1 acting as a WNT receptor, are emerging (Kim et al., 2016). Our experiments with pharmacological activators (Fig. 4) did not identify an active the GPCR signaling role for Pkd1. But as this is a negative result, we cannot exclude the possibility that a secondary messenger exists that is activated downstream of polycystin 1. Importantly, such a function would need to be a phenotype that we have not yet described for the Pkd1 morphants or is so subtle that it is overshadowed by the G-protein squelching aspect of the phenotype. Regardless, the most definitive answer will likely come from genetic studies in mouse. Unfortunately, mouse mutants of trimeric G proteins have been poorly studied for their function in the kidney. With respect to the G proteins identified in this study, Gnas, Gnai2 and Gnb1 are lethal or severely growth retarded (Yu et al., 1998; Fu et al., 2006; Okae and Iwakura, 2010), Gnai1 has not been analyzed for its kidney function or explored because compound mutants with Gnai2 (Pineda et al., 2004) and Gnai14 mutants are not available.
This study has direct implications for our understanding of autosomal dominant polycystic kidney disease (ADPKD). Mutations in polycystin 1 account for 70% of all individuals diagnosed with ADPKD. Cyst formation and expansion in ADPKD has long been correlated with an increase in GPCR signaling. In fact, the most promising therapeutic approach so far is the use of the vasopressin receptor (V2R) antagonist tolvaptan (Torres et al., 2012). Although it has been proposed that the increase in vasopressin signaling is a consequence of a defect in calcium homeostasis, we now propose a more direct model. Loss of polycystin 1 increases the pool of available G proteins and allows other GPCRs (such as V2R) to recruit them. As a consequence, these GPCRs can now be overactivated – obviously in the presence of appropriate ligands. It is also noteworthy that polycystin 1 does not bind all G-protein α subunits. Instead, it binds only a selected group (one or two from each family) and thus loss of polycystin 1 will not hyper-activate all GPCR signaling.
Our model also explains the pleiotropic nature of the PKD phenotype. Different organs have different assortments of GPCRs and although, for example, vasopressin/V2R signaling is abundant in the kidney, it is not present in the liver. This obviously also has implications for the development of future therapeutic agents. Reducing certain GPCR signaling pathways may be more easily achievable than interfering with ubiquitous second messengers such as Ca2+. The beneficial effect of vasopressin antagonism on kidney cysts (Torres et al., 2012) may just be the first successfully targeted GPCR to alleviate the multiple organ manifestations of ADPKD.
Another conclusion from our study is that G-protein β/γ signaling in PKD is more crucial than generally thought. In G-protein signaling, the contributions of G-β/γ-dependent signaling are often overlooked, even though they are involved in multiple aspects of GPCR-mediated signaling and regulation (Clapham and Neer, 1997; Smrcka, 2008; Dupre et al., 2009). Early studies have shown that PKD1 bound to Gnai/Gnao regulates Ca2+ and K+ channels via a G-protein β/γ-dependent mechanism (Delmas et al., 2002, 2004). More recently, another modulator of G-protein β/γ signaling, the activator of G-protein signaling 3 (AGS3) has been implicated in PKD (Kim et al., 1999; Nadella et al., 2010). AGS3 acts as a guanine dinucleotide dissociation inhibitor by binding to inactive Gnai and Gnao, and thereby liberates the G-protein β/γ subunits. These, in turn, regulate PKD2 channel activity and cell proliferation. Interestingly, as in the present study, the effects of AGS3 can be, at least in acute kidney injury, inhibited by addition of the G-protein β/γ signaling inhibitor gallein (Regner et al., 2011). In fact, there has been an emerging interest in targeting G-protein β/γ signaling for clinical intervention in multiple diseases such as heart failure or acute kidney injury (Lin and Smrcka, 2011; White et al., 2014; Bernardo and Blaxall, 2016). Especially as gallein and its derivatives do not cause obvious side effects in mice (Lin and Smrcka, 2011). Thus, inhibiting β/γ signaling in conjunction with other, for example, G-protein α inhibitors may be a future possibility for PKD therapy.
MATERIALS AND METHODS
Xenopus experiments were approved by the Institutional Animal Care and Use Committee (IACUC) and adhered to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All experiments in this study were performed using Xenopus laevis embryos obtained by in vitro fertilization, maintained in 0.1× modified Barth medium (Sive et al., 2000) and staged according to Nieuwkoop and Faber (1994). Antisense morpholino oligomers (MOs) were obtained from GeneTools. The sequences of the MOs are summarized in Table S1. MOs were diluted to a concentration of 1 mM. To target both of the homologs of Gna14, the two MOs were mixed at a 1:1 ratio. The targeting efficacy for newly designed MOs was verified using in vitro transcription/translation using the Retic Lysate IVT Kit (Ambion) in the presence of Transcend tRNA.
For all the injections, a total of 8 nl of morpholino oligomer solution was injected radially at the two- to four-cell stage into Xenopus laevis embryos. At the onset of our study, only Xenopus tropicalis polycystin 1 had been identified and cloned. Therefore, the pCS2-Pkd1-CTT construct was generated by PCR using a Xenopus tropicalis polycystin 1 EST (Image# 7650819); the pCS2-Pkd1-CTTΔGBD was generated using the QuikChange II Site-Directed Mutagenesis Kit (Agilent) deleting the core G-protein-binding domain (GBD), i.e. the five amino acids RHKVR. For synthetic mRNA, all plasmids were linearized with NotI and transcribed with SP6 RNA polymerase using the mMessage mMachine (Ambion). For the rescue experiments, the Pkd1-sMO was injected radially at the two- to four-cell stage followed by two injections of 2 ng synthetic mRNA into one vegetal blastomere at the eight-cell stage.
For the drug experiments, Xenopus embryos were cultured until stage 20, when polycystin 1 mRNA expression can first be detected. At this time point, injected and uninjected controls were treated with the indicated amounts of the chemical compounds by simply adding the compounds to the medium; drugs were exchanged every 24 h and embryos were fixed when the untreated controls reached stage 39 or 42. Brefeldin A (BFA), forskolin and H89 were obtained from Sigma, 6-Bnz-cAMP-AM (cAMP-PKA) and the 8-pCPT-2′-O-Me-cAMP-AM (cAMP-Epac) were obtained from Biolog, and gallein and m-3M3FBS were obtained from Calbiochem. For all the drug experiments, we carefully titrated the compounds using a wide concentration range, as the IC50 concentration is often not equivalent to the EC50. One important note of caution is that the quality of gallein is very batch/supplier-dependent and needs to be carefully titrated.
Whole-mount in situ hybridization and histology
In situ hybridization and its analysis by paraplast sectioning and several antisense probes have been described previously (Belo et al., 1997; Zhou and Vize, 2004; Tran et al., 2007). The information for the novel antisense probes is summarized in Table S1. For histological staining, Xenopus embryos were fixed in Bouin's fixative, dehydrated, embedded in paraplast, sectioned at 7 µm, dewaxed and stained with Hematoxylin and Eosin.
Cell culture studies
All cells are tested every 6 months for mycoplasma contamination. The nuclear localization studies were performed in HEK293T cells (ATCC) transiently transfected with a Myc-tagged version of Pkd1-CTT or Pkd1-CTTΔGBD using jetPRIME (Polyplus Transfections). 48 h after transfection, cells were processed for immunofluorescence using the mouse anti-Myc antibody (clone 9E10, DSHB, 1:50) and an Alexa Fluor-647-coupled secondary antibody (#A31571, Life Technologies, 1:2000). Images were analyzed by confocal microscopy and analyzed by LAS X (Leica Microsystems).
The two isogenic proximal tubular cell lines Pkd1Fl/− and Pkd1−/− were a kind gift from Dr S. Somlo (Shibazaki et al., 2008). Cells were propagated at 33°C, but all experiments were performed at the non-permissive temperature of 37°C. The 3D matrigel cultures were set up as previously described (Merrick et al., 2012). Drugs were added the next day, replaced every 24 h and cells were imaged after 7 days in culture. For the proliferation assay, 200,000 cells were seeded into a 10 cm plate, grown for 24 h, counted using the Beckman Z2 Coulter and 200,000 cells were then re-seeded; the growth curve was calculated based on the daily dilution factors.
To measure cAMP levels upon somatostatin treatment, cells were starved overnight in DMEM/F12, treated with 10 µM forskolin for 10 min and then stimulated with different concentrations of somatostatin (Calbiochem) for 30 min. cAMP levels were determined using the Direct cAMP ELISA kit (ENZO). For the parathyroid hormone (PTH) assay, cells were starved overnight in DMEM/F12, treated with different concentrations of PTH in the presence of 1 mM IBMX (to prevent cAMP degradation) for 15 min and analyzed as described above.
To determine the levels of activated GTP-bound Gnas, cells were starved overnight in DMEM/F12 and treated with 10−6 M PTH. Cell lysates were used for immunoprecipitation using the activated Gnas antibody (#26906, NewEast Biosciences) and analyzed by western blot using a Gnas-specific antibody (#sc-823, Santa Cruz Biotechnologies, 1:500). The same antibody was used for determining total Gnas levels. Equal loading was confirmed using an α-tubulin antibody (#T9026, Sigma, 1:2000).
Determination of the orientation of cell division
As the amount of mitotic cells in the developing pronephric kidney (i.e. phospho Histone H3-positive) at a given time is rather low, we used pulse labeling with EdU to follow the fate of cells after cell division. Xenopus embryos were injected at stage 40 with 8 nl EdU stock solution into the belly and cultured until stage 42. Embryos were fixed in 4% PFA and processed for paraplast sectioning. To determine whether cells divided along the length of the pronephros or circumferentially, embryos were sectioned either transverse or sagittally. They were then de-waxed, rehydrated and re-fixed in 4% PFA for 15 min. Staining was performed as outlined in the instructions of the Click-iT EdU imaging kit (Thermo Fisher Scientific) using 500 µl per slide for 30 min. Wash the slides briefly with 3% BSA. To distinguish cells of the pronephric kidney from the surrounding interstitium, sections were then subjected to immunofluorescence staining using anti α-Na/K ATPase (clone #A5, DSHB, 1:50) and anti mouse Alexa Fluor-488 (#A11001, Life Technologies, 1:2000). Slides were mounted in Prolong Gold with DAPI (Thermo Fisher Scientific) and analyzed by fluorescence microscopy and analyzed by LAS X (Leica Microsystems). For all sections containing cells of the pronephros, the frequency of duplets of EdU-labeled daughter cells among the total number of Na/K ATPase and DAPI-positive cells were determined. Cells that were not present as EdU-labeled duplets were not considered to have divided in the plane of the section and were not included in the counts.
Surface plasmon resonance analysis
Binding affinities between the C-terminal tail of Pkd1 and different G-protein α subunits were determined using the BIAcore 3000. The different G-protein α subunits were generated using in vitro transcription/translation as outlined above and bound to a streptavidin chip. Protein quality was confirmed by western blot and equal binding to the chip was determined by comparing the baselines (Fig. S4). The Pkd1-CTT and Pkd1-CTTΔGBD proteins were produced as GST-fusion proteins, purified using glutathione beads and cleaved off from the GST-tag using ProTEV Plus (Promega). Protein integrity and amount were determined using the BSA standards of the BCA Protein Assay Kit (Thermo) and Coomassie Blue staining. To determine association and dissociation constants, different concentrations (0.1 ng/ml to 100 ng/ml) were used. In order to confirm the binding affinities, the experiment was also performed in reverse; GST-Pkd1-CTT and GST-Pkd1-CTTΔGBD were bound to a CM5 chip coupled with a GST antibody (#06-332, Millipore) and different concentrations of GST-purified Gnas (0.1 ng/ml to 100 ng/ml) were used.
Methodology and statistics
All experiments were performed a minimum of three times. All data from the Xenopus experiments were repeated using clutches of eggs from different females and independent fertilizations. The group size in each individual experiment was ≥25 embryos/condition and the sex of the embryos was undetermined. For the drug exposure experiments, injected and uninjected embryos were randomly divided into the experimental groups. As in all our Xenopus experiments, embryos that did not gastrulate properly or showed severe developmental abnormalities were discarded. All other embryos were used for the data analysis. The Xenopus data are presented as cumulative numbers with the number of embryos analyzed presented above the individual bars. Figures showing in situ hybridization (whole mount and sections thereof), as well as fluorescence and light-field microscopy are representative images.
We thank Drs M. Blum, J. Larraín, E. Pera and S. Piccolo, as well as all members of the laboratory for critically reviewing the manuscript and helpful discussions.
Conceptualization: O.W.; Formal analysis: B.Z., U.T., O.W.; Investigation: B.Z., U.T., O.W.; Writing - original draft: B.Z., U.T., O.W.; Writing - review & editing: B.Z., U.T., O.W.; Visualization: B.Z., U.T., O.W.; Supervision: O.W.; Project administration: O.W.; Funding acquisition: O.W.
The Molecular Biotechnology Core at the Lerner Research Institute/Cleveland Clinic is supported by a grant from National Institutes of Health/National Center for Research Resources to Dr Satya P. Yadav (1S10RR016789-01A1). This work was supported by a grant from National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases (5R01DK080745) and an Lerner Research Institute Chair's Innovative Research Fund to O.W. Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.