CaMK-II is a PKD2 target that promotes pronephric kidney development and stabilizes cilia

Although roles for Ca²⁺ signaling during early development have been supported by a rich and influential history of inquiry (Whitaker, 2006), the identification of the responsible Ca²⁺-dependent molecular targets has remained elusive. For example, cellular behaviors during gastrulation, somitogenesis and trunk, eye, brain and organ formation have been associated with Ca²⁺ flux through specific channels (Porter et al., 2003; Webb and Miller, 2003; Webb and Miller, 2007; Whitaker, 2006), but have not been attributed to specific Ca²⁺-dependent effectors.

CaMK-II, the Ca²⁺/calmodulin-dependent protein kinase type II, is best known as an enzyme enriched in the central nervous system and important for long-term potentiation (Hudmon and Schulman, 2002). CaMK-II has also been implicated in embryonic cell migration (Easley et al., 2008), cell cycle progression (Rasmussen and Rasmussen, 1995; Tombes et al., 1995) and convergent cell movement in response to non-canonical Wnt signals (Kühl et al., 2000). CaMK-II has recently been identified as an essential Ca²⁺ target in the zebrafish Kupffer's vesicle (KV), the ciliated organ necessary for establishment of left-right asymmetry (Francescatto et al., 2010). Cilia in developing tissues have emerged as signaling centers, whose functions hold important clues to human ciliopathies (Fliegauf et al., 2007; Lancaster and Gleeson, 2009; Veland et al., 2009).

Polycystic kidney disease is a ciliopathy that can be linked to defects in Ca²⁺ signaling. Autosomal dominant polycystic kidney disease (ADPKD) is the most common monogenetic kidney disease affecting one in 1000 births, typified by the development of massive kidney cysts and caused by mutations in the polycystins, PKD1 or PKD2 (Wilson, 2004). PKD1 is a large receptor-like integral membrane protein with an extracellular N-terminus that contains motifs for cell-cell and cell-matrix interactions (Torres et al., 2007). PKD2 is a member of the TRP superfamily of nonselective Ca²⁺-permeable ion channels (Clapham, 2003) found in the plasma membrane, the endoplasmic reticulum and in the primary cilium (Cai et al., 1999; Tsiokas, 2009). Both PKD1 and PKD2 contain coiled-coil domains in their intracellular C-terminus that enables the formation of a receptor-channel complex (Casuscelli et al., 2009; Giamarchi et al., 2010; Qian et al., 1997; Tsiokas et al., 1997). This complex is thought to play a role in regulating levels of intracellular Ca²⁺, where activation of PKD1 induces Ca²⁺ entry via PKD2 (Hanaoka et al., 2000; Vandorpe et al., 2001). Mice homozygous for mutations in Pkd2 develop cardiac abnormalities, situs inversus and cysts in the kidney and pancreas and die before E16.5 (Pennekamp et al., 2002; Wu et al., 2000). Suppression of pkd2 in zebrafish results in hydrocephaly, loss of left-right asymmetry, body curvature, pronephric occlusions and pronephric cysts (Bisgrove et al., 2005; Obara et al., 2006; Schottenfeld et al., 2007; Sun et al., 2004). Dominant-negative PKD2 causes zebrafish and rat kidney cysts (Feng et al., 2008; Gallagher et al., 2006). Interestingly, Ca²⁺-channel blockade increases cystic development in a PKD rat model, whereas a Ca²⁺ mimetic inhibits late stage cystogenesis (Gattone et al., 2009; Nagao et al., 2008). In summary, diminished activation of both intracellular Ca²⁺ signals and presumptive downstream Ca²⁺ targets leads to kidney cyst formation.

In this study, activated CaMK-II was detected during early zebrafish development in specific ciliated tissues including cells of the nervous system, the inner ear and pronephric kidney. In the developing kidney, CaMK-II activation was found to be dependent on PKD2 Ca²⁺ and was capable of restoring proper kidney development in PKD2-deficient embryos. These findings indicate that CaMK-II is a natural transducer of PKD2 Ca²⁺ to enable both anterior ductal cell migration and cloacal cilia stability. These findings have implications to other ciliated tissues and identify a potential new therapeutic target for ADPKD.

Wild-type (AB and WIK), Tg(α1 subunit Na⁺/K⁺-ATPase-GFP) and Tg(β-actin:CAAX-GFP) zebrafish (Danio rerio) embryos were obtained through natural matings and raised at 28.5°C as previously described (Kimmel et al., 1995).

Digoxigenin-labeled antisense riboprobes (0.5-1.5 kb) were synthesized using T3 or T7 RNA polymerase from cloned cDNAs and then hybridized with fixed embryos as previously described (Rothschild et al., 2007). For whole-mount in situ hybridization, embryos were developed using alkaline phosphatase-conjugated anti-digoxigenin. gata3, cdh17, wt1a, pax2a and ret1 probes were prepared as previously described (Wingert and Davidson, 2008; Wingert et al., 2007). A γ1c antisense probe was generated from 48 hours postfertilization (hpf) kidney cDNA, TOPO cloned, linearized with NotI and transcribed using the T3 RNA polymerase. Fluorescent in situ hybridization was conducted as previously described (Rothschild et al., 2007) using fluorescent anti-digoxigenin antibodies and images were acquired using a Nikon C1 laser scanning confocal microscope.

Immunolocalization using anti-phosphorylated CaMK-II (anti-P-CaMK-II; phosphorylated at Thr²⁸⁷), CaMK-II and total CaMK-II antibodies has previously been described by this laboratory (Easley et al., 2006; Francescatto et al., 2010). The anti-phospho-Thr²⁸⁷ (anti-P-T²⁸⁷) antibody detects CaMK-II proteins across species and the total CaMK-II antibody reacts with the C-terminal region of all CaMK-II proteins.

Na⁺/K⁺-ATPase-GFP transgenic embryos at 24, 48 and 72 hpf were anesthetized and the pronephros dissected as previously described (Liu et al., 2007). Dissected pronephroi were dissolved in TRIzol reagent (Invitrogen) to prepare RNA, or harvested for CaMK-II activity assays as previously described (Rothschild et al., 2009; Rothschild et al., 2007). For flow sorted kidney cells, embryos were incubated in collagenase (Kramer-Zucker et al., 2005), then placed into 0.25% trypsin-EDTA for 30 minutes at 28.5°C. Embryos were filtered through mesh before sorting on a Becton-Dickinson FACS Aria II high-speed analyzer/sorter. RNA was prepared from Na⁺/K⁺-ATPase-GFP-positive cells.

Whole embryos were lysed or GFP-expressing pronephroi were obtained and total CaMK-II activity was assessed by measuring phosphate incorporation into the substrate autocamtide-2, as previously described (Rothschild et al., 2007).

Embryos were fixed in 4% paraformaldehyde in PBS alone for F-actin staining, or in conjunction with methanol for P-CaMK-II, total CaMK-II and acetylated tubulin immunostaining. Dent's fixative (80:20 methanol:DMSO) was used for immunostaining the α1 subunit of Na⁺/K⁺-ATPase (α6F), atypical protein kinase C (aPKC) and γ-tubulin. Embryos were incubated with rabbit anti-P-T²⁸⁷ (Millipore; 1:20), mouse anti-total CaMK-II (BD Biosciences; 1:20), mouse anti-acetylated α-tubulin antibody (Sigma Chemical Co.; 1:500), mouse anti-α6F (Developmental Studies Hybridoma Bank, University of Iowa; 1:10), rabbit anti-aPKC (Santa Cruz; 1:10), or rabbit or mouse anti-γ-tubulin (Sigma Chemical Co.; 1:500) followed by either goat anti-mouse Alexa Fluor 488, goat anti-mouse Alexa Fluor 568, or goat anti-rabbit Alexa Fluor 568 (Invitrogen; 1:500). Embryos were permeabilized with NP-40 before staining F-actin with Alexa-Fluor-488-phalloidin (Invitrogen; 1:500). Embryos were imaged using confocal microscopy (Nikon C1 Plus two-laser) on a Nikon E-600 compound microscope using a 20× or 40× dry or 100× oil immersion objective.

The zebrafish pkd2 morpholino oligonucleotide (MO; 4 ng), the camk2g1 MO (1 ng) and the control mismatch MO (1 ng) were purchased from GeneTools and used as previously described (Francescatto et al., 2010), with a constant injection volume of 1 nl. Wild-type (WT) or T²⁸⁷D CaMK-II δ_E were co-injected with the MO as previously described (Rothschild et al., 2009). Kidney-targeted WT and K⁴³A CaMK-II δ_E CaMK-II using the Na⁺/K⁺-ATPase promoter (Liu et al., 2007) were generated using Gateway technology (Invitrogen) (Francescatto et al., 2010).

Renal filtration assays used tetramethylrhodamine dextran (70,000 MW; Invitrogen) injected into the pericardial sac of transiently anesthetized [0.003% Tricaine (MS-222); Sigma] 56 hpf zebrafish embryos as previously described (Tobin and Beales, 2008). Embryos were imaged within 20 minutes of injection and then again at 72 hpf. Six of the seven control embryos, but none of the eight camk2g1 morphants, showed >50% reduction in rhodamine-dextran from the pericardial sac and anterior embryo.

Zebrafish embryos at 3dpf were fixed in 4% paraformaldehyde for 4 hours at room temperature and washed several times in PBT. The yolks were punctured and embryos were embedded in 5% low melting point agarose (Sigma). 100 μm sections were obtained using a Leica VT1000P vibratome and then labeled with 10 μg/ml propidium iodide (Sigma) and Alexa-Fluor-488-phalloidin (1:500). Sections were mounted in 50% glycerol in PBT between coverslips and imaged using confocal microscopy.

Live embryos were imaged using differential interference contrast optics after transient anesthesia and immobilization in low melting point agarose. Ciliary motility was imaged in the pronephric duct and cloaca in anesthetized embryos using a Nikon 60× water immersion Plan APO objective with DIC optics and 20 frames per second acquisitions. Lengths of cilia in fixed embryos were determined from anti-acetylated α-tubulin whole mounts using quantitative length algorithms in Nikon Elements software. For each condition, two to five experimental replicates were examined with total embryo numbers between 20 and 40. Statistical analyses were performed using the paired t-test. Statistical significance was set at P<0.005.

CaMK-II is most often linked to central nervous system function, where it can constitute as much as 1% of the total protein (Hudmon and Schulman, 2002). However, CaMK-II is expressed in every adult mammalian tissue (Tobimatsu and Fujisawa, 1989) and across all metazoan species (Tombes et al., 2003). CaMK-II is different from other CaM-dependent kinases because it is able to oligomerize and autophosphorylate at T²⁸⁷ upon prolonged Ca²⁺/CaM stimulation. T²⁸⁷ autophosphorylation converts CaMK-II into a Ca²⁺-independent state. Activated (P-T²⁸⁷) CaMK-II has been detected in zebrafish embryos as early as the 10-somite stage (Francescatto et al., 2010), but it is also found in the forebrain, in a region corresponding to the olfactory placode, in the pronephric duct and cloaca, and at the base of inner ear hair cell kinocilia (Fig. 1). These locations are a subset of total CaMK-II expression in the embryonic forebrain, olfactory placode, spinal cord, somites, pectoral fins, ear and pronephric kidney, as determined immunologically (Fig. 1) and by in situ hybridization (Rothschild et al., 2009; Rothschild et al., 2007). At 60 hpf, CaMK-II remains activated in kidney and ear cells and becomes activated in retina, anterior pituitary and neuromasts of the lateral line (Fig. 2).

The preferential activation of CaMK-II in the developing kidney was confirmed in dissected 3 dpf pronephric kidneys compared with 6 dpf whole embryos (Fig. 3A). In whole mounts, CaMK-II activation varied along the pronephros (Fig. 3B). The pronephric kidney becomes functional as early as 48 hpf and consists of the glomerulus, which empties into two pronephric ducts. These ducts connect at the cloaca for waste excretion (Drummond, 2005; Drummond et al., 1998). This simplified kidney nomenclature is used throughout this study, even though subregions have been recently characterized and named (Wingert et al., 2007). Within the pronephros, activated CaMK-II is first detected at 24 hpf in distal regions, and peaks at 30 hpf in both anterior and posterior regions of the pronephric ducts and cloacal cells (Fig. 3B). Activated CaMK-II persists until 72 hpf in the distal pronephric duct, cloaca and cloacal cilia. Although CaMK-II is expressed uniformly along the entire duct (Fig. 1I), CaMK-II activation is not uniform along the pronephros (Fig. 3B). In particular, a region of the pronephric duct in the distal half persistently lacks activated CaMK-II (Fig. 3B).

The subcellular location of activated CaMK-II provides additional insight into its role. P-CaMK-II is found associated with apical and basolateral cell surfaces and in intracellular clusters throughout the pronephros (Fig. 3C-E, arrowheads). Such activated CaMK-II clusters were previously reported in cells surrounding the KV (Francescatto et al., 2010) and have also been observed in brain tissue (Hudmon et al., 2005). P-CaMK-II is preferentially apical in proximal and distal ducts and more evenly distributed between apical and basolateral surfaces in the cloaca, colocalizing with plasma membranes in transgenic (CAAX:GFP) embryos (Fig. 3F-H). P-CaMK-II is found uniformly along the length of cloacal cilia (Fig. 3I-K), but is undetectable in ductal cilia in the same embryos (Fig. 3D), identifying the short (∼3 μm) cloacal primary cilia as unique with respect to CaMK-II activation. Activated CaMK-II colocalizes with the α1 subunit of Na⁺/K⁺-ATPase, a marker for transporting epithelium (Liu et al., 2007), in basolateral focal sections (Fig. 3L-N), but is not exclusive to these cells.

Although CaMK-II is encoded by four genes in mice and humans (Camk2a, -2b, -2g, -2d), it is encoded by seven transcriptionally active genes (camk2a, -2b1, -2b2, -2g1, -2g2, -2d1, -2d2) in zebrafish to generate over two dozen splice variants (Rothschild et al., 2009; Rothschild et al., 2007). Only γ1 CaMK-II was identified in the zebrafish kidney by using RT-PCR of isolated kidney cells or tissue (see Fig. S1A in the supplementary material). The γ1 CaMK-II splice variants expressed in the kidney (γ1_C and γ1_F) are putative cytosolic variants as previously described (Rothschild et al., 2009; Rothschild et al., 2007). Whole-mount in situ hybridization detected transcripts of the camk2g1 gene throughout the kidney all the way through to the cloaca (see Fig. S1B, arrow in the supplementary material) and in pronephric duct cells when counterstained with the acetylated tubulin antibody (see Fig. S1C in the supplementary material) or the pax2 probe (see Fig. S1D-F in the supplementary material).

Furthermore, using previously validated gene-specific translation-blocking antisense MOs (Francescatto et al., 2010; Rothschild et al., 2009), only γ1 CaMK-II suppression interfered with pronephros development. This MO would suppress the expression of all γ1 splice variants. In particular, the injection of the camk2g1 MO (1 ng) induced hydrocephaly and pronephric cysts, and destabilized cloacal cilia (Figs 4, 5). Hydrocephaly was observed as early as 48 hpf (Fig. 4B, arrow) and became more pronounced over time. Bilateral pronephric cysts were detected at the third somite, just posterior to the pectoral fins (Fig. 4D, arrow and inset) in whole embryos and in cross sections adjacent to the glomerulus (Fig. 4F, asterisks). Renal filtration assays (Fig. 4G-J) confirmed kidney defects in camk2g1 morphants due to the persistence of Rhodamine-dextran (Fig. 4J) after pericardial injection.

Pronephric cysts have been correlated with defects in the collective anterior-directed migration of pronephric cells that form the convoluted proximal segment (Vasilyev et al., 2009). The hook-like convolution in the anterior region of the pronephros (Fig. 5A,C) is absent in camk2g1 morphants (Fig. 5B,D). At 72 hpf, 87% of camk2g1 morphant embryos (n=168) lacked the convolution. Defective convolution can be quantified by measuring the distance from the posterior edge of the ear to the anterior pronephric duct at 72 hpf (Vasilyev et al., 2009). MO suppression of camk2g1 increased this distance from 140 μm in control embryos to 490 μm in morphant embryos (Fig. 5E).

The distal region of the pronephros is also susceptible to CaMK-II suppression. The distal pronephros exhibits gross (Fig. 5F,G) and kidney-specific (insets) morphogenic defects in camk2g1 morphants at 72 hpf. These distal defects, which appeared as occlusions when examined one embryo at a time by through-focal microscopy, were observed in 73% of morphant embryos (n=143).

Since the disruption of cilia also induces kidney cysts in mouse and zebrafish embryos (Kramer-Zucker et al., 2005; Sullivan-Brown et al., 2008; Veland et al., 2009; Wilson, 2008), cilia were evaluated in camk2g1 morphants. Ductal cilia formed and were motile in camk2g1 morphants (Fig. 5H,I; see Movie 1 in the supplementary material). By contrast, cloacal cilia formed normally at 24 hpf in morphants, but then began to disassemble at 48 hpf (Fig. 5J,K) and by 72 hpf (Fig. 5L,M) had completely disassembled. Basal bodies persisted on the apical surface (Fig. 5M). Any remaining cloacal cilia in camk2g1 morphants were immotile (see Movies 2 and 3 in the supplementary material). On average, thirteen cloacal cilia form at 72 hpf in control embryos, but less than one remains in camk2g1 morphants (Fig. 5N). Of the remaining cloacal cilia, their length had decreased from 3 to 1.5 μm (Fig. 5O). These results point to a role for activated CaMK-II in stabilizing cloacal cilia.

Since cystogenic morphants and mutants have been linked to alterations in cellular polarity, aPKC, γ-tubulin and F-actin were examined in camk2g1 morphants but showed normal polarity (data not shown). Cystogenic morphants and mutants could also affect specification of important pronephric genes, but these were also unaffected in camk2g1 morphants (Fig. 6). For instance, the expression of pax2a in intermediate mesoderm was unaltered at the 14-somite stage (Fig. 6A,B). At 24 hpf in both control and morphant embryos, cdh17 demarcates the entire pronephric duct (Fig. 6C,D), whereas gata3 and ret1 appear in the distal regions (Fig. 6E-H). At 24 and 48 hpf, the anterior glomerular marker, wt1a, labels the podocyte precursor fields in both control and morphant embryos (Fig. 6I-L). The incomplete convergence of wt1a cells in camk2g1 morphants at 48 hpf might also reflect morphogenic defects.

Kidney development is also disrupted by a targeted dominant-negative CaMK-II mutant. The K⁴³A point mutant lacks phosphotransferase activity, but can still hetero-oligomerize with endogenous CaMK-II and has previously been determined to act in a dominant-negative fashion (Francescatto et al., 2010). WT or dominant negative (DN) GFP-CaMK-II was targeted to pronephric cells. Embryos expressing DN, but not WT CaMK-II, developed pronephric cysts (Fig. 7C) and failed to undergo ductal convolution (Fig. 7F). Pronephric ducts that expressed DN, but not WT GFP-CaMK-II, were often shortened or branched at their anterior end (Fig. 7F,F′), suggestive of deficient or misdirected migration. Even though expression was mosaic (arrowheads), DN CaMK-II interfered with anterior convolution in a majority (76%; n=64) of embryos. However, neither distal pronephric duct development nor cloacal cilia beating were blocked (see Movie 4 in the supplementary material). The dominant inhibitory effect of this CaMK-II mutant on endogenous P-CaMK-II was seen in cells expressing DN, but not WT CaMK-II (Fig. 7G-L). For example, all five cells expressing GFP-DN CaMK-II in the field of view shown in Fig. 7J-L had diminished P-CaMK-II, whereas both WT cells had unaltered P-CaMK-II (Fig. 7G-I).

CaMK-II can be activated by Ca²⁺ release from channels that are found in the endoplasmic reticulum, the plasma membrane or ciliary membrane, such as PKD2. The suppression of pkd2, a Ca²⁺-conducting channel found in all these locations, causes a reduction of P-CaMK-II in cells surrounding the Kupffer's vesicle, leading to a loss of organ asymmetry (Francescatto et al., 2010). PKD2 is also expressed in the pronephros and pkd2 morphants have cysts (Obara et al., 2006) that are similar in size and location to those observed in camk2g1 morphants (Fig. 4F). As a result, we postulate that CaMK-II acts downstream of PKD2 during kidney development.

To determine if PKD2 Ca²⁺ activates CaMK-II, P-CaMK-II was assessed in pkd2 morphants along the pronephric kidney in whole-mount embryos at 30 hpf (Fig. 8A-F) and in dissected 72 hpf pronephroi. In these morphants, P-CaMK-II at apical cell surfaces and at cytosolic clusters (Fig. 8C, arrowheads) was markedly reduced (Fig. 8D). Cloacal P-CaMK-II was reduced, but not eliminated, at both basolateral and apical surfaces (Fig. 8F), but completely eliminated from cilia (Fig. 8F, inset, asterisk). P-CaMK-II levels in other locations not linked to PKD2 function, such as the ear, forebrain and spinal cord cell bodies, was not diminished in pkd2 morphants (data not shown).

A second method to evaluate whether activated kidney CaMK-II is dependent on PKD2 is through CaMK-II activity assays conducted in the presence and absence of Ca²⁺, as previously described (Francescatto et al., 2010). Prolonged Ca²⁺ stimulation leads to autophosphorylation (P-CaMK-II), yielding an enzyme that is active in the absence of Ca²⁺. The percentage of Ca²⁺-dependent activity that is Ca²⁺ independent (autonomous) can be determined from cell lysates and kinase assays; this `autonomy' rarely exceeds ∼50%. CaMK-II autonomy was determined in lysates prepared from excised pronephric ducts from control (Fig. 8G,H) and pkd2 morphant (not shown) embryos. Isolated kidneys contain some non-kidney cells, but are enriched in P-CaMK-II (Fig. 3A). CaMK-II autonomy was approximately 37% in control embryos and 23% in pkd2 morphant embryos, thus demonstrating a significant, but not complete reduction (∼40%) in CaMK-II autonomy (Fig. 8I). These results indicate that the high levels of activated CaMK-II in the zebrafish kidney are due in large part to PKD2 Ca²⁺.

Further evidence that CaMK-II acts downstream of PKD2 was obtained by rescuing pkd2 morphants with a constitutively active phosphomimetic (T²⁸⁷D) CaMK-II mutant (Fig. 9). The pkd2 MO was injected alone or with GFP-tagged T²⁸⁷D δ_E CaMK-II, as previously described (Rothschild et al., 2009). Like camk2g1 morphants, over 90% of pkd2 morphants exhibited hydrocephaly, tail curvature (Fig. 9B), pronephric occlusion (Fig. 9E), and lacked convolution (Fig. 9H,H′). When embryos were injected with T²⁸⁷D, but not when injected with WT CaMK-II, tail curvature and hydrocephaly were reduced (Fig. 9C) and pronephric occlusions (Fig. 9F) decreased from 96% to 52% (n=213) of embryos. Convolution defects were reduced from 91% (n=187) in pkd2 morphants to 52% (n=189) in morphants expressing T²⁸⁷D CaMK-II. Like DN CaMK-II (Fig. 7), ectopic T²⁸⁷D CaMK-II was enriched apically in pronephric epithelial cells (Fig. 9J,L), whereas Na⁺/K⁺-ATPase is more basolateral in the same cells. The ability of T²⁸⁷D CaMK-II to rescue both the posterior and the anterior defects of pkd2 morphant embryos was found to be statistically significant (Fig. 9M,N).

As seen with camk2g1 morphants, pkd2 morphants have cloacal cilia that become unstable during the second day of development (Fig. 9O-R). At 72 hpf, pkd2 morphants retained only 2.5 cilia, with an average length of 1.5 μm (Fig. 9O,P), whereas basal bodies persisted apically (Fig. 9Q,R). Similar to camk2g1 morphants, 83% of pkd2 morphants exhibited defective or no cloacal cilia beating (see Movie 5 in the supplementary material). Ciliary defects in pkd2 morphants could not be recovered with ectopic T²⁸⁷D CaMK-II, perhaps for the same reason that DN CaMK-II was incapable of inducing cloacal cilia defects. Ciliary instability was not observed in ductal cilia, indicating a special, yet undefined role for PKD2, Ca²⁺ and CaMK-II in cloacal cilia stability.

A linkage between PKD2 and CaMK-II was also demonstrated by co-injecting embryos with sub-effective doses of MOs (Fig. 10A-C). Such synergistic approaches using pairs of MOs have been previously used to link separate components in the same genetic pathway (Mably et al., 2006). Injection of one-quarter of the normal amount of pkd2 or camk2g1 MOs alone did not alter anterior pronephric development, as assessed by both morphology (Fig. 10D-F) and ear-kidney distance (Fig. 10H). However, the co-injection of these same sub-effective pkd2 and camk2g1 MO amounts inhibited convolution in 66% (n=74) of embryos (Fig. 10D-G) and reduced anterior migration (Fig. 10H) to the same extent as the higher levels of either MO alone (Figs 5, 9). Although occlusions were not evident in pkd2/camk2g1 co-morphants (Fig. 10G), cloacal cilia were destabilized (Fig. 10I-M) and beat improperly in 85% of injected embryos (see Movies 6-8 in the supplementary material). The findings from this study collectively demonstrate the relationship between PKD2 and CaMK-II in the developing zebrafish kidney and their role in cloacal cilia stability and proper pronephric morphogenesis.

We conclude that the Ca²⁺/CaM-dependent protein kinase, CaMK-II, is an essential effector of the Ca²⁺ channel, PKD2, in the developing pronephros. We have reached this conclusion based on multiple lines of evidence. First, CaMK-II deficiencies caused by either CaMK-II translation-blocking MOs or dominant-negative constructs phenocopy pkd2 morphants. Second, pkd2 morphants have reduced levels of activated (autophosphorylated) CaMK-II. Third, constitutively active CaMK-II can rescue pkd2 morphants. Finally, CaMK-II and PKD2 suppression work synergistically, thus supporting their placement in the same genetic pathway. Our findings suggest that CaMK-II promotes pronephric development by enabling both anterior migration and posterior morphogenesis, including ciliary stability. This is the first report of a direct downstream Ca²⁺-dependent target of PKD2 in the kidney and is consistent with the finding that CaMK-II is activated by PKD2 in ciliated cells of the zebrafish KV (Francescatto et al., 2010).

Activated CaMK-II has now been detected in multiple ciliated tissues during development including the KV, the inner ear and neuromasts. These sites of CaMK-II activation are consistent with known locations of prolonged Ca²⁺ elevation in embryos of zebrafish and other species (Webb and Miller, 2003; Webb and Miller, 2007). These locations are also consistent with tissues that exhibit morphogenic defects after CaMK-II suppression (Francescatto et al., 2010; Rothschild et al., 2009). Many of these tissues express PKD2 (Obara et al., 2006), but other Ca²⁺ channels, such as the RyR, are also known to be required for full CaMK-II activation (Francescatto et al., 2010). This suggests that CaMK-II integrates Ca²⁺ signals from more than a single source and is consistent with the inability to completely inactivate pronephric CaMK-II in PKD2-deficient embryos.

There are many ways in which CaMK-II could enable development and function of ciliated tissues. In the kidney, anterior and posterior regions can be considered separately, but are interdependent. Zebrafish pronephros development requires anterior migration of the proximal pronephric epithelia to form pronephric tubules that link to the glomerulus and form the embryonic kidney. Collective migration of duct cells begins at 28.5 hpf, coincident with highly active CaMK-II in the proximal pronephric duct, where pronephric epithelial cells undergo a proximally directed migration toward the glomerulus. This migration occurs as a sheet, where apical cell connections are maintained and basal surfaces project lamellipodia at the leading edge. Pronephric migratory cells require dynamic adhesions to the basement membrane and focal adhesion turnover (Vasilyev et al., 2009). The inactivation of focal-adhesion proteins leads to the development of cysts in mice (Lo et al., 1997). Suppression of camk2g1 and pkd2 inhibited proper pronephric duct migration and anterior convolution, whereas pkd2 morphants, ectopically expressing constitutively active CaMK-II, partially recovered cellular migration.

Both CaMK-II and PKD1 have been identified at focal adhesions and have been independently linked to focal adhesion turnover (Easley et al., 2008; Wilson et al., 1999). PKD1 interacts with the extracellular matrix (ECM) and focal adhesion proteins (Wilson, 2004). Cells from ADPKD patients with mutations in PKD1 exhibit increased adhesiveness to the extracellular matrix and decreased motility (Wilson et al., 1999). This phenotype is identical to that observed when CaMK-II is acutely inhibited in mouse fibroblasts (Easley et al., 2008). Focal adhesion turnover requires the dephosphorylation of tyrosine residues on paxillin and FAK (Mitra et al., 2005). PKD1 colocalizes with paxillin in renal cell culture, suggesting a possible link between PKD1 and CaMK-II at focal adhesions (Israeli et al., 2010; Wilson, 2001). Therefore, the lack of migration in anterior pronephric cells in pkd2 and camk2g1 morphants could be due to the inhibition of CaMK-II at focal adhesions.

Cloacal cells undergo shape changes, exhibit apicobasal polarity and depend on migratory morphogenesis (Pyati et al., 2006; Slanchev et al., 2011; Vasilyev et al., 2009), but unlike proximal epithelial cells, do not migrate as collectives (Vasilyev et al., 2009). Activated CaMK-II is enriched at the apical surface of ductal cells and is known to directly interact with and stabilize F-actin (Easley et al., 2006; Lin and Redmond, 2008; Okamoto et al., 2007; Sanabria et al., 2009). Apical constriction mediated by actinomyosin is necessary for the alterations in cell shape that lead to proper tissue development (Daggett et al., 2007), including duct formation (Sawyer et al., 2010) and could explain a role for CaMK-II in cloacal cells.

This study has also provided some insight into the role of Ca²⁺ signaling in ciliary stability. Our findings indicate that the PKD2-dependent activation of CaMK-II is necessary to maintain primary cilia in the cloacal region. CaMK-II is activated along the entire length of these cilia and when inactivated destabilizes cilia. Activated CaMK-II might stabilize cilia through its inhibition of the microtubule destabilizing protein, kinesin-13 (MCAK) (Holmfeldt et al., 2005; Piao et al., 2009). Interestingly, cilium assembly is dependent on the Polo kinase, Plk4, acting on CEP 152 (Cizmecioglu et al., 2010; Hatch et al., 2010). CaMK-II is known to collaborate with the Plx1 Polo kinase to promote meiotic resumption (Liu and Maller, 2005).

Although the loss of cloacal cilia or the presence of ductal occlusions can cause improper anteriorly directed pronephric epithelial migration and cyst development, independently blocking anterior kidney development, as observed with the DN CaMK-II, can induce cysts without a loss of cilia. A loss of fluid flow is also seen when either ductal cilia fail to assemble or in morphants where blood flow is inhibited (Vasilyev et al., 2009). These separate, but inter-related, anterior and posterior morphogenic events are paralleled by the anterior and posterior activation of CaMK-II.

There is a significant amount of overlap between known CaMK-II substrates or binding partners and those implicated in kidney development or cystic disease. For example, histone deacetylases (HDACs) are known CaMK-II targets and have become a viable therapeutic target in treating PKD. CaMK-II is known to phosphorylate HDAC4 or HDAC5 at conserved residues (Backs et al., 2008; Zhang et al., 2007), leading to cytosolic HDAC retention and thus the upregulation of MEF2C target genes (Little et al., 2007; Zhang et al., 2007). One of these MEF2C target genes is missing in metastasis (Mim; Mtss1 – Mouse Genome Informatics), which is necessary for ciliogenesis (Bershteyn et al., 2010) and actin cytoskeletal organization (Saarikangas et al., 2011). Mef2c knockout mice and MIM-deficient mice develop tubule dilations and cysts, whereas inhibition of HDAC5 in Pkd2^–/– mice suppresses renal cyst formation (Xia et al., 2010). In zebrafish, HDAC inhibition reverses cystogenesis in pkd2 morphants (Cao et al., 2009). A plausible mechanism to partially explain kidney development and homeostasis would depend on the activation of CaMK-II by PKD2 Ca²⁺, which then leads to the phosphorylation and cytosolic retention of HDAC4 and/or HDAC5 and sustained mef2c gene expression.

In summary, our results are consistent with PKD2-dependent Ca²⁺ elevations, which activate CaMK-II. Activated CaMK-II could then balance apical polarity and constriction, cell migration, ciliary function and gene expression as the kidney forms, functions and grows (Fig. 11). Cysts are the result of mutations in components of this pathway at any location to cause incomplete ductal-glomerular connections, pronephric occlusions and cloacal ciliary disassembly. These findings have identified CaMK-II as a previously missing link in this pathway and raise the possibility that CaMK-II and its targets might be novel therapeutic targets for ADPKD and other ciliopathies.

The authors gratefully acknowledge Jamie Lahvic, Bennett Childs, Jamie McLeod, Alexandra Myers and Amritha Yelamilli for their assistance with this study; and Becky Burdine, Alan Davidson, Amanda Dickinson, James Lister, Lila Solnica-Krezel and Joe Yost for sharing useful reagents and advice. Supported by National Science Foundation grant IOS-0817658.