Loss of the Ciliary Kinase Nek8 Causes Left-Right Asymmetry Defects

Abstract

A missense mutation in mouse Nek8, which encodes a ciliary kinase, produces the juvenile cystic kidneys (jck) model of polycystic kidney disease, but the functions of Nek8 are incompletely understood. Here, we generated a Nek8-null allele and found that homozygous mutant mice die at birth and exhibit randomization of left-right asymmetry, cardiac anomalies, and glomerular kidney cysts. The requirement for Nek8 in left-right patterning is conserved, as knockdown of the zebrafish ortholog caused randomized heart looping. Ciliogenesis was intact in Nek8-deficient embryos and cells, but we observed misexpression of left-sided marker genes early in development, suggesting that nodal ciliary signaling was perturbed. We also generated jck/Nek8 compound heterozygotes; these mutants developed less severe cystic disease than jck homozygotes and provided genetic evidence that the jck allele may encode a gain-of-function protein. Notably, NEK8 and polycystin-2 (PC2) proteins interact, and we found that Nek8^−/− and Pkd2^−/− embryonic phenotypes are strikingly similar. Nek8-deficient embryos and cells did express PC2 normally, which localized properly to the cilia. However, similar to cells lacking PC2, NEK8-depleted inner medullary collecting duct cells exhibited a defective response to fluid shear, suggesting that NEK8 may play a role in mediating PC2-dependent signaling.

Cystic diseases of the kidney are prevalent in the human population and there is compelling interest in understanding the functions of the associated genes. The most common human cystic disease is autosomal dominant polycystic kidney disease (ADPKD) with an incidence of 1 in 750 individuals.¹ ADPKD is caused by mutations in either PKD1 or PKD2, which encode the polycystin proteins PC1 and PC2, respectively. PC1 is a G-protein coupled receptor that partners with PC2, a transient receptor potential polycystic family ion channel that regulates both calcium entry into and intracellular calcium release within renal epithelial cells.^2–7 The nephronophthisis (NPHP) types 1–11 and NPHPL1 comprise a less common group of autosomal recessive disorders, but they are collectively the leading genetic cause of ESRD within the first 3 decades of life.^8–10

We previously reported that the mouse juvenile cystic kidneys (jck) model of recessive polycystic kidney disease (PKD) is due to a missense mutation in the Nek8 gene; jck animals develop cysts in distal nephron segments and collecting ducts within the first week of life and the disease rapidly progresses, causing renal failure and death by 6 months of age.¹¹ Nek8 encodes a highly conserved member of the Nek family of serine/threonine kinases, characterized by an N-terminal kinase domain homologous to that of the Aspergillus nidulans NIMA protein that controls cell cycle entry during mitosis.¹² Mammalian Neks have divergent C termini, and the NEK8 C terminus contains multiple repeats that are homologous to RCC1, a guanine nucleotide exchange factor for Ran that is required for chromosome condensation,^13–15 although NEK8 guanine nucleotide exchange factor activity has not been demonstrated. The jck mutation in Nek8 results in a glycine to valine change in the second RCC repeat of the protein. Subsequent studies have reported missense mutations in the RCC domain-encoding regions of both rat Nek8 in the Lewis PKD model¹⁶ and human NEK8, in which three independent mutations were identified from 700 NPHP patients, making it a candidate for NPHP type 9.^11,17 Together, these data demonstrate that the NEK8 C terminus is critical for its function and the protein plays an important role in the maintenance of renal tubule integrity in the postnatal kidney. Of note, all of the reported mutations of Nek8 are missense mutations. We describe here the generation of a mouse carrying a null mutation of Nek8; in this line, mutant pups die shortly after birth and exhibit a variety of laterality defects and associated cardiac anomalies.

Mutations in several PKD-associated genes cause laterality defects in the developing mouse, and dysfunction of primary cilia is the critical link between the phenotypes. Kidney tubular epithelial cells each express a primary cilium that collectively act as mechanosensors of fluid flow through the tubules, transducing calcium-dependent signals hypothesized to keep the cells in a differentiated state.⁷ Early in development, the establishment of appropriate asymmetry in mammals requires the functions of two populations of cilia within the node, a pit in the ventral surface of the embryo. Motile cilia rotate and generate a leftward fluid flow, and primary cilia subsequently produce a left-sided calcium gradient.^18,19 Mutations in Pkd2, Kif3A, and Tg737(Ift88) illustrate how cilia are critical signaling centers in the kidney and the node. Loss of Pkd2 from the adult kidney results in PKD, whereas Pkd2-null embryos exhibit left-right defects.^20,21 Functionally, the PC1/PC2 complex mediates calcium influx in response to flow-induced sheer stress in cultured renal cells, and loss of either protein abolishes the response²²; similarly, Pkd2-null embryos do not exhibit a left-sided calcium signal.¹⁹ Heterotaxy also occurs with loss of the Pkd1-related gene, Pkd1l1; the gene product localizes to cilia and interacts with PC2, forming a node-specific complex required for proper asymmetry.^23,24

Alternatively, mutations that disrupt cilia assembly and/or structure cause both phenomena. PKD arises in the kidney-specific deletion of Kif3A, which encodes a kinesin subunit required for ciliogenesis, and in Tg737 (Ift88/polaris) hypomorphs that exhibit shorter cilia with bulbed tips.^25,26 Laterality defects occur upon targeted disruption of Kif3A, Kif3B, or Tg737 (Ift88); each of these knockouts lack nodal cilia and die mid-gestation with randomized heart looping.^18,27,28 Downstream of flow-induced calcium signals, left-sided determinants are expressed in the early embryo. The Nodal gene encodes a ligand of the TGF-β family and is expressed first in the node and subsequently in the left lateral plate mesoderm (LPM), and the transcription factor Pitx2 is later expressed in the left LPM; mutant alleles of each of these genes are embryonic lethal with laterality defects.^29–32

A majority of proteins associated with human cystic diseases, including the polycystins and nephrocystins, localize to either the axoneme and/or the basal body of primary cilia.^8,33 Similarly, NEK8 localizes to the base of renal cilia and is thus far the only kinase that has been localized to the ciliary axoneme.^34,35 Cilia defects are indeed present in jck cystic renal epithelia and cells cultured from affected kidneys; mutant cilia are elongated and exhibit enhanced ciliary localization of PC1 and PC2 compared with wild-type cilia.^36,37 In addition, PC2 is hyperphosphorylated in jck kidneys and both wild-type and mutant NEK8 coimmunoprecipitate with PC2 from kidney lysates, suggesting that these proteins perform critical cooperative functions.³⁷

We originally hypothesized that the jck allele is hypomorphic because the only overt phenotype in homozygotes is PKD, mutant protein is expressed in the kidneys, and morpholino-mediated reduction of Nek8 in zebrafish causes pronephric cysts.¹¹ Here, we characterize the Nek8-null mutant and determine that ciliogenesis is intact, but left-right patterning is perturbed. We revisit morpholino knockdown in zebrafish and demonstrate that Nek8 depletion causes laterality defects, confirming a conserved role for Nek8 in left-right patterning. Furthermore, we generated jck/Nek8− compound heterozygotes; surprisingly, renal cystic disease in these mutants is much less severe than in jck homozygotes, suggesting the jck allele of Nek8 encodes a gain-of-function protein. Finally, there is marked similarity of Nek8−/− and Pkd2−/− embryonic phenotypes and we determine that PC2-associated activity is abrogated in cells lacking NEK8.

Results

Generation of Nek8-Null Mouse Model

To generate a Nek8-null allele, we obtained a C57BL/6 (B6) bacterial artificial chromosome containing the Nek8 genomic locus. Nek8 exon 3 was targeted by recombineering using a cassette composed of an internal ribosomal entry sequence (IRES), the lacZ gene, and the neomycin resistance gene flanked by FLP recombinase recognition sites (Figure 1A). Nek8:lacZ was designed to encode a truncated protein lacking the kinase domain and a β-galactosidase reporter to recapitulate endogenous NEK8 expression. Nek8:lacZ-positive B6/129 hybrid embryonic stem cell³⁸ clones were injected into B6 blastocysts, and the resulting chimeras were bred to obtain germline transmission of the Nek8:lacZ allele. Heterozygous Nek8:lacZ carriers are viable and fertile with no obvious phenotypes, whereas Nek8:lacZ homozygous pups die shortly after birth; carriers and homozygotes are obtained at expected 50% and 25% frequencies, respectively.

Figure 1.

Targeted disruption of Nek8 results in a true null allele. (A) Nek8 exon 3 is disrupted with an IRES-lacZ:neomycin cassette. FLP recombinase recognition sequences (frt) flank neomycin. (B) PCR genotyping distinguishes wild-type (+/+) embryos from heterozygous (+/−) and homozygous (−/−) lacZ-positive embryos. The wild-type band is amplified with F and R1 primers flanking exon 3. The mutant product is amplified with the F primer and R2 primer from the lacZ sequence. (C) RT-PCR analysis confirms that the lacZ reporter is spliced out of the mutant allele. rtF in exon 2 and rtR in exon 4 amplify a 400 bp product and a 150 bp product from wild-type and Nek8 mutant MEF cDNA, respectively. (D) No NEK8 is detected via Western blot analysis of MEF lysates.

Heterozygous- and homozygous-Nek8:lacZ embryos were analyzed for β-galactosidase activity, but enzyme activity was not detected in PCR-positive samples (Figure 1B and data not shown). We derived mouse embryo fibroblasts (MEFs) from E14.5 wild-type and homozygous embryos, analyzed the cDNA from the cells, and discovered the targeted exon 3 is spliced out of the mutant transcript, rendering the reporter ineffective (Figure 1C and data not shown). However, the aberrant splicing of exons 2–4 causes a frameshift and a creates premature stop codon resulting in a true null allele of Nek8, which we herein refer to as Nek8−, because no protein is detected in homozygotes by Western blot analysis (Figure 1D).

Nek8−/− Embryos Exhibit Randomization of Left-Right Asymmetry and Cardiac Defects

Nek8−/− mice die shortly after birth and mid-gestation mutants frequently exhibit edema and focal hemorrhages that are suggestive of cardiac defects (Figure 2, A–D). Upon autopsy, we observe randomization of left-right asymmetry in the mutants. The characteristic asymmetry of mouse body plan is defined in the thorax by a four-lobed right lung and single left lung lobe with a left-oriented heart apex; in the abdomen, the stomach and adjacent spleen reside on the left side (Figure 2, E–G). Approximately 15% of Nek8−/− mutants have normal body asymmetry, whereas 24% display situs inversus totalis, in which the body plan is a mirror image to that of a normal fetus (Figure 2, E, F, and H). However, a majority of Nek8−/− mutants (61%) exhibit right pulmonary isomerism (RPI), a “duplication” of the four-lobed right lung on the left side (Figure 2I); one-third of these mutants have normal abdominal situs, whereas two-thirds have inverted abdominal situs (Table 1).

Figure 2.

Nek8−/− mice exhibit laterality and cardiac defects. (A–D) Lateral views of littermates. E14.5 wild-type embryo (A) compared with Nek8−/− embryo (B) with edema (arrows) and focal hemorrhages. E16.5 wild-type embryo (C) is larger than the mutant (D). (E–H) E18.5 wild-type compared with a mutant with SI. (E) Lateral views of a wild-type embryo (left) and a null embryos (right) illustrate abdominal situs defect in the mutant. (F) Frontal views of the embryos in E, in which internal organs mirror each other. (G and H) Hearts and lung lobes dissected out of the thoracic cavities of embryos in E and F. (G) Wild-type lungs contain four right lobes and one left lobe (R4, L1). (H) The mutant with SI has one right lobe and four left lobes (R1, L4). (I) Heart and lungs dissected out of a mutant with RPI; right and left lungs are four-lobed (R4, L4). (J–M) Nek8−/− embryos exhibit cardiac defects. (J and K) MRI images through thoraces of a wild-type and a null littermate with RPI. Sagittal views show that wild-type pulmonary artery (J) and aorta (J’) are present in distinct images. (K) Mutant pulmonary artery and aorta run parallel in the same image. (L and M) H&E-stained transverse cardiac sections. (L) The wild-type heart has intact atrial (*) and ventricular septa. (M) The SI mutant heart exhibits atrial (*) and ventricular (arrow) septal defects. (N) Cardiac looping occurs in Nek8-ATG MO zebrafish embryos in a dose-dependent manner. Scale bars, 1.0 mm in A–I; 100 μm in J–M. SI, situs inversus; s, stomach; sp, spleen; R, right; L, left; h, heart; pa, pulmonary artery; a, aorta; vs, ventricular septa.

Table 1.

Summary of laterality defects in Nek8−/− embryos

Abdominal Situs			Thoracic Situs
			SS	RPI	SI
SS	20 (39)	8 (15)	12 (24)	0
SI	31 (61)	0	19 (37)	12 (24)
Total	51 (100)	8 (15)	31 (61)	12 (24)

Data are shown as the number of embryos (% of total). SS, situs solitus; SI, situs inversus.

Mutants with laterality defects often have major structural cardiac anomalies.^29,39–41 Therefore, magnetic resonance imaging (MRI) and histologic analyses were performed to assess the hearts of Nek8−/− embryos. In three mutants with RPI we identified double outlet right ventricle (DORV), a defect in which both the pulmonary artery and aorta exit the right ventricle of the heart, accompanied by atrial and ventricular septal defects (A/VSD) (Figure 2, J and K and data not shown). DORV is best exemplified in lateral MRI images: the pulmonary artery and aorta are captured in separate images in a wild-type embryo, but the structures run parallel to each other in the same image in the Nek8−/− mice. Although mice with situs inversus totalis can be viable, as in inversin mutant mice,⁴² no Nek8−/− mice survive the perinatal period and we found A/VSD in mutants with complete situs inversus (Figure 2, L and M).

To further examine the requirement for Nek8 in establishing normal patterning in a vertebrate system, we utilized morpholino knockdown of Nek8 in zebrafish. We previously used this approach to evaluate cystogenesis in adult fish,¹¹ but laterality was not assessed. In this analysis, randomization of heart looping occurs in a dose-dependent manner in zebrafish embryos injected with a morpholino targeted to the initiator methionine codon of Nek8 (P<0.001; Figure 2N).

Early Laterality Markers Are Misexpressed in Nek8−/− Embryos but Nodal Cilia Are Intact

Left-right asymmetry mutants have defects in nodal signaling that are assessed via in situ hybridization of Nodal and Pitx2 between E8.0 and E8.5.^43–45 In wild-type presomite embryos, Nodal expression is symmetrical in the node. By the 3–4 somite stage, it becomes asymmetrically expressed in the node, staining stronger on the left side than the right, and expression is restricted to the left LPM (Figure 3A). In contrast, in Nek8−/− embryos at the 3–4 somite stage, Nodal expression within the nodes was always symmetrical (n=9) and was absent from the LPM in seven of nine embryos, bilateral in one embryo, and expressed in the left LPM in one embryo (Figure 3, B–D and Supplemental Figure 1). We examined embryos at the 6–8 somite stage for Pitx2 expression, which is normally apparent in the headfolds and a discrete region of the left LPM of wild-type embryos (Figure 3E). Pitx2 was appropriately expressed in the headfolds of stage-matched Nek8−/− embryos (n=6), but was absent from the LPM in four of six mutants (Figure 3F), right-sided in one mutant, and correctly expressed in one mutant.

Figure 3.

Laterality marker genes are misexpressed in Nek8−/− embryos but nodal cilia are intact. (A–F) In situ hybridization of laterality markers confirms signaling defects in Nek8−/− embryos (arrows indicate LPM). (A–D) Nodal is expressed in the node proper in 4–5 somite wild-type and null embryos, as well as in the wild-type left LPM shown in A. Mutant embryos lack LPM staining (B) or have bilateral (C) or left LPM expression (D). Pitx2 is expressed in 6 somite wild-type and null headfolds and in the wild-type left LPM (E), whereas mutant embryos usually lack LPM expression (F). Anti-acetylated tubulin immunofluorescence of E8.0 embryos shows that wild-type nodal cilia (G) (n=10) and null nodal cilia (n=5) (H) appear similar. High-magnification view of wild-type nodal cilia (I) (arrows) stained with anti-acetylated tubulin (green) and anti-NEK8 (red); NEK8 localizes to all node monocilia (merge; inset, NEK8 at base of cilia) and is not detected in Nek8−/− nodal cilia (J). Scale bars, 200 μm in A–I. R, right; L, left.

Several laterality mutants have short nodal cilia or lack them altogether, which disrupts asymmetric signaling downstream of nodal flow.^18,27,28 Because NEK8 localizes to renal epithelial cilia, we examined the nodal cilia of E8.0 embryos using immunofluorescent detection of the ciliary marker acetylated α-tubulin and the anti-NEK8 antibody. Wild-type and Nek8−/− nodes are indistinguishable (Figure 3, G and H) and NEK8 is expressed in the nodal cilia of wild-type E8.0 embryos (Figure 3I). Taken together, the in situ hybridization and immunofluorescence data suggest that Nek8 performs a critical function in laterality determination, but is not directly required for ciliogenesis.

Nek8−/− Kidneys Do Not Develop PKD

Nek8−/− pups die within hours of birth but they can produce urine, indicating that the kidney is functional. We examined renal development in Nek8−/− embryos and the mutant kidneys develop similarly to those of wild-type through E15.5; however, by E16.5 and through P0, glomerular cysts are apparent in periodic acid–Schiff (PAS)-stained samples (data not shown and Figure 4, A and B). In addition, P0 Nek8−/− proximal tubules appear thin-walled and slightly dilated compared with the wild-type proximal tubules, which are thicker due to intact brush borders on the apical surfaces. To further assess the tubular phenotype, we stained kidneys with the fluorescein-conjugated Lotus tetragonolobus lectin (LTL) and Dolichos biflorus agglutinin (DBA) to identify the proximal tubules and collecting ducts, respectively. LTL staining confirmed that the Nek8−/− proximal tubules are indeed dilated at P0 but not at E18.5, whereas the DBA staining showed that P0 mutant collecting ducts are largely unaffected (Figure 4, C–F and data not shown). Importantly, the abundance and appearance of cilia in wild-type and Nek8−/− tubules is similar (Figure 4, C–F, acetylated α-tubulin in red).

Figure 4.

Nek8−/− kidneys exhibit glomerular cysts and proximal tubule dilation. (A and B) PAS-stained kidneys from P0 pups. (A and A’) The wild-type kidney has normal glomeruli and proximal tubules stain pale purple (arrows). (B and B’) The mutant kidney develops glomerular cysts and proximal tubules are distended. FITC-conjugated LTL lectin marks proximal tubules (arrows) and confirms that the wild-type tubules are intact (C), whereas the mutant tubules adjacent to glomerular cysts are distended (D). (E and F) FITC-conjugated DBA lectin marks the collecting ducts (arrows). Wild-type tubules (E) and mutant tubules (F) are similar in size and shape and cilia are present in both (red puncta, anti-acetylated tubulin immunofluorescence). Cilia (arrows; red, anti-acetylated tubulin) of control (pLKO) (G) and Nek8-knockdown (N8KD10) (H) IMCD cells are of similar morphology and length. (I) Western analysis of IMCD lysates illustrates N8KD10 cells lack NEK8 (arrow) compared with the larger nonspecific band recognized by the antibody. Cilia of wild-type (J) and Nek8−/− MEFs (K) are of similar morphology and length. Scale bars, 100 μm in A–F; 10 μm in G, H, J, and K. Original magnification, ×10 in A and B; ×40 images of boxed region in A’ and B’. g, glomeruli; gc, glomerular cysts.

Although ciliogenesis appears normal in Nek8 mutant nodes and kidneys, it is difficult to identify overt length or structural defects via immunofluorescence in these tissues. Therefore, we analyzed cilia in control and Nek8-deficient mouse inner medullary collecting duct (IMCD) cells (Figure 4, G and H) as well as in wild-type and null MEFs (Figure 4, J and K). To generate Nek8-depleted IMCD cells, small hairpin RNA (shRNA) against Nek8 was introduced and individual clones were isolated and expanded; NEK8 is not detected via Western blot in line N8KD10 compared with the vector control clonal (pLKO) line (Figure 4I). There are no morphologic or length differences between the cilia of wild-type and Nek8-deficient cells in culture: pLKO and N8KD10 cilia lengths are 3.3±1 μm and 3.0±0.6 μm, respectively; wild-type and Nek8−/− MEF cilia lengths are 2.9±0.7 and 2.8±0.9 μm, respectively. Statistical analysis using a nonparametric ANOVA test reveals that these differences are not significant, suggesting that ciliogenesis can occur normally in the absence of NEK8.

jck/Nek8− Compound Heterozygotes Have Less Severe Cystic Disease Than jck Homozygotes

We hypothesized that the jck allele of Nek8 is hypomorphic, and to test this we intercrossed jck/+ and Nek8+/− mice to generate compound heterozygous jck/Nek8− animals. Wild-type, jck/+, jck/Nek8−, and jck/jck mice were sacrificed at 7 weeks of age, the kidneys were weighed and histology was performed to assess disease severity. Surprisingly, jck/Nek8− animals develop very mild renal cystic disease and the combined kidney weights are not much greater than those of wild-type mice (Figure 5, A–D).

Figure 5.

jck/Nek8-compound heterozygotes exhibit moderate PKD and Nek8−/− kidneys do not develop cysts. (A–C) H&E stained kidneys at 7 weeks. (A) jck/+ kidneys do not develop PKD. jck/Nek8− kidneys (B) develop moderate PKD compared with the severe PKD of jck/jck mutants (C). (D) Comparison of combined kidney weights between wild-type (n=7), jck/+(n=8), jck/Nek8− (n=8), and jck/jck (n=15) animals. (E and F) Embryonic kidney explant culture cyst formation assay results. Nek8−/− kidneys do not develop many cysts, jck/Nek8− kidneys develop moderate cysts, and jck/jck cystogenic potential is the greatest when cultured with 8-Br-cAMP. Error bars in E represent SEM. Scale bars, 0.5 mm in A–C and F.

Because Nek8−/− pups die at birth, we analyzed the cystogenic potential of null, jck, and compound mutant embryonic kidneys in an explant culture assay.^46,47 Kidneys were removed from E14.5 embryos and cultured for 5 days with or without a cyst-inducing cAMP analog. In the presence of 8-Bromo-cAMP, jck/jck kidneys develop the greatest numbers of cysts, whereas Nek8−/− kidneys have the least cystogenic potential of all the genotypes (Figure 5, E and F). Taken together, the findings that jck/Nek8− adult and cultured embryonic kidneys develop less severe cystic disease than homozygous jck kidneys, and that the Nek8-null kidney explants do not develop tubular cysts, suggest the jck mutation results in a gain-of-function protein.

Nek8−/− Phenotypes Resemble Those Caused by Loss of PC2 Function

The Nek8−/− phenotypes are remarkably similar to those of Pkd2−/− embryos, including randomization of left-right asymmetry, RPI, cardiac malformations, and glomerular cysts in late-gestation kidneys,^21,39 and NEK8 and PC2 have been shown to interact.³⁷ Therefore, we sought to determine whether the Nek8−/− defects are due to loss of PC2 expression or a change in localization. PC2 localizes to all node monocilia¹⁹ and we observe its localization is identical in E7.75 wild-type and Nek8−/− embryo nodes (Figure 6, A and B). In addition, we do not detect gross differences in PC2 levels via Western blot between littermatched E18.5 wild-type and Nek8−/− kidney lysates (Figure 6C).

Figure 6.

PC2 is expressed in Nek8-deficient embryos and cells but flow sensing is perturbed. Whole-mount immunofluorescence of E7.75 embryos shows PC2 localization in both wild-type (A) and Nek8−/− nodal cilia (B). (C) Western blot analysis shows that PC2 is expressed at similar levels in E18.5 wild-type and Nek8−/− kidney lysates. (D–G) Effects of fluid shear stress exposed to cultured IMCD cells on PC2-dependant intracellular Ca²⁺ influx. The representative time traces of GCaMP3 fluorescence intensity (normalized by the initial fluorescence measured before introducing the sheer stress to the cells) are shown for Ca²⁺-depleted wild-type PC2-knockdown (E) and NEK8-knockdown (N8KD10) IMCD cells (F). The live cell media contains 1.25 mM CaCl₂ at the time of the flow stimulation. Error bars represent SEM. (G) The maximum (peak) values of normalized GCaMP3 fluorescence intensity were obtained from multiple single time traces and averaged over the cell population. Error bars represent the SD of multiple single cell measurements. Nonparametric t tests were performed on the selected data sets (***P<0.001, **P<0.01). All of the measurements were carried out under identical conditions. Scale bars, 10 μm in A and B; 5 μm in A’ and B’.

Because PC2 expression/localization is intact in Nek8-null tissues, we hypothesized that PC2 activity may be perturbed in NEK8-deficient cells. To investigate PC2 function, we subjected cultured wild-type and N8KD10 (knockdown) IMCD cells to fluid shear stress and measured changes in PC2-dependent intracellular Ca²⁺ influx across the membrane using live cell microscopy; PC2 is properly localized in wild-type and N8KD10 cilia under the conditions used in the assay (Supplemental Figure 2). The cells stably express a genetically encoded calcium ion indicator, GCaMP3, used to monitor changes in intracellular Ca²⁺ levels.^48,49 Live fluorescence image time series were acquired while Ca²⁺ depleted cells were exposed to fluid shear stress. As expected, wild-type IMCD cells exhibit a robust response to fluid shear stress (Figure 6, D and G). PC2-depleted IMCD cells⁵⁰ were used as a negative control and show significantly reduced fluid flow response (Figure 6, E and G). The N8KD10 cells also have a reduced response to flow, suggesting that PC2 function may be perturbed in the absence of Nek8 (Figure 6, F and G).

Discussion

The Nek8 gene, which is mutated in the jck mouse model of PKD, encodes the only kinase that to date has been localized to the ciliary axoneme. In this study we report that its targeted disruption in the mouse is lethal and causes laterality defects, further demonstrating a link between cystogenesis and left-right asymmetry. Importantly, there is a conserved requirement for the gene in the specification of vertebrate left-right asymmetry as knockdown of Nek8 expression in zebrafish causes randomization of heart looping, a result described both here and by an independent group.⁵¹

A small number of genes implicated in human renal cystic diseases cause both PKD and laterality defects when disrupted in the mouse. Three such examples include Pkd2, inversin, and Nphp3; PKD2 is mutated is 15% of human ADPKD cases and mutations in INVS (inversin) and NPHP3 cause NPHP types 2 and 3, respectively.^1,52,53 NEK8 has been identified as an interacting partner of each of the mouse protein counterparts: NEK8 and PC2 reciprocally coimmunoprecipitate from kidney and cell lysates,³⁷ and NEK8 has been shown to colocalize with inversin and NHPH3 in the base of the cilium in an inversin-dependent manner.³⁵ The association of NEK8 with each of these proteins is intriguing due to the similarities and some of the differences we identified in the jck and Nek8-null mouse models compared with Pkd2, inversin, and Nphp3 mutants.

With regard to left-right asymmetry, Pkd2−/− embryos die in utero and 90% of the mutants exhibit RPI,^39,21,54 whereas a majority of inversin mutants exhibit situs inversus.^55–57 The observation that Pkd2−/− early embryo nodes fail to elicit an asymmetric calcium signal in response to flow¹⁹ likely accounts for the absence of Nodal and Pitx2 expression in the LPM, causing the high percentage of RPI in mutant embryos. Interestingly, inversin nodal cilia produce a weak net leftward flow much slower than wild type,⁵⁸ which may cause right-sided Nodal expression,⁴⁴ more frequently lead to inversion of the body plan, and preclude the development of RPI in the thorax. A majority of Nek8-null embryos exhibit RPI whereas only 24% have situs inversus, and both Nodal and Pitx2 are frequently absent from the left LPM of mutants; thus, Nek8−/− mutants are more similar to Pkd2−/− than to the inversin mutant mice. We determined that Nek8-null phenotypes are not due to loss of PC2, because the protein is expressed and properly localized in mutant embryos, tissues, and cells. Therefore, we utilized a fluid flow assay to measure PC2-dependent mechanosensitive activity in cells lacking NEK8. IMCD cells stably knocked down for Pkd2 or Nek8 exhibit decreased calcium influx in response to shear stress compared with wild-type cells, suggesting PC2 is not fully functional in the absence of NEK8. Given that PC2 is required for both the IMCD cell responses to flow and the calcium gradients detected in the mouse embryo node, we hypothesize that the ability of PC2 to “sense” nodal flow and elicit a left-sided signal is diminished in nodes lacking NEK8. Perhaps NEK8 is a critical component of the PC2 channel complex that is required to mediate the signaling processes downstream of nodal calcium flux that are indispensable for the establishment of appropriate left-right asymmetry.

Similarly to other heterotaxy mutants, Nek8−/− embryos exhibit severe cardiac malformations that include DORV coupled with A/VSD.⁵⁹ It is not surprising to find such abnormalities in animals with RPI, because aberrant heart development accompanies the inappropriate symmetry in the thoracic cavity. However, we identified cardiac septal defects in mutants with complete situs inversus, a condition that is not lethal if the inverted body plan is appropriately established, suggesting that Nek8 may play a role in cardiac morphogenesis independent of the initial left-right asymmetry defect, as is the case for inversin mutants. It was originally reported that nearly 100% of inversin mutants display situs inversus totalis; therefore, the thoracic situs was “normal reversed” compared with wild-type littermates. However, a later study of inversin homozygotes found that 90% of the pups did have situs inversus, but 10% were identified with situs solitus; furthermore, cardiac malformations including outflow tract and septal defects were found in 37% of the pups, including those few with normal thoracic situs.⁶⁰ Perhaps Nek8 and inversin are required in the cardiac cilia present in mid-gestation mouse embryos,⁶¹ a biologic question best addressed by conditional deletion of the gene.

Our original observations that PKD develops in jck homozygotes, affected animals express the mutant protein, and morpholino knockdown of zebrafish Nek8 causes pronepephric cysts all suggested the jck allele represented a partial loss of function. This hypothesis was consistent with analysis of the Ift88 gene, which initially established the critical link between renal cysts and left-right asymmetry, because the hypomorphic Tg737 allele causes PKD, whereas the null mutation causes laterality defects.^26,28 Similarly, the pcy mouse carries a missense mutation in the Nphp3 gene and develops slowly progressive PKD, whereas null mutants die mid-gestation with laterality and severe, complex cardiac defects.^52,62

Of note, pcy/Nphp3− compound heterozygous mice are viable and develop severely cystic kidneys by week 12. This intermediate phenotype indicates that the pcy allele of the gene is hypomorphic compared with the null allele.⁶² We crossed the Nek8-null allele with jck carriers anticipating a similar result; that is, compound heterozygous mice would have a PKD phenotype more severe than that of jck homozygous mice. On the contrary, jck/Nek8− compound heterozygotes develop less severe PKD than the homozygous jck mutants. The null pups die shortly after birth and the kidneys do have glomerular cysts and dilated proximal tubules, but the defect is likely due to impaired glomerular function rather than early signs of NEK8-dependent PKD. Supporting this notion is that jck mutant glomeruli and proximal tubules do not develop cysts; it is the distal DBA-positive collecting ducts that are affected.^36,37,63 Therefore, we utilized embryonic kidney explant culture to further explore the cystic potential of the mutant kidneys.⁴⁶ This in vitro system recapitulates what we observe in the mice: Nek8−/− kidneys do not develop tubular cysts, whereas jck/jck homozygous kidneys are dramatically affected and jck/− kidneys have an intermediate phenotype.

The data therefore suggest that the jck allele encodes a gain-of-function protein, a hypothesis not initially considered because jck/+ adults do not exhibit PKD. However, Natoli et al. performed explant cultures with kidneys harvested a day earlier than in our study, and observed a significant difference between wild-type and jck/+ kidneys, reporting cyst percentages of 2.5% for wild-type and 7.5% for jck/+ compared with 15% for jck/jck kidneys.⁴⁷ In the adult jck/+ kidney, perhaps the presence of wild-type NEK8 is protective against the deleterious effect of the mutant protein. It is currently unclear how the jck allele may result in a gain-of-function protein. A recent study demonstrated that wild-type NEK8 and the jck-equivalent NEK8^G442V variant were similarly active in in vitro kinase assays toward β-casein and histone, but it remains unknown whether NEK8^G442V exhibits altered protein interactions and/or substrate specificity in vivo that would affect its function.⁶⁴ Another potential consequence of the mutation is the mislocalization of jck-specific NEK8 proteins; both Smith et al. and Zalli et al. suggest the mutant NEK8 fails to localize to cilia and centrioles in cells cultured from jck/jck kidneys and RPE cells, respectively.^36,64 However, Sohara et al. demonstrate that in sections of intact jck homozygous kidneys, ciliary expression of the mutant protein is detected along the entire axoneme, and is not confined to the proximal region in which it is normally localized.³⁷ Further studies are therefore required to understand the consequences of the jck missense mutation.

The fact that jck homozygotes do not exhibit laterality defects suggests that the missense mutation does not abrogate PC2 activity. However, the PKD phenotype of jck mice may still be a consequence of PC2 dysregulation, because PC2 is upregulated and mislocalized in jck cystic tissues.³⁷ Similar to loss of Pkd2, its transgenic overexpression in the kidney causes cysts⁶⁵; perhaps the excess PC2 in jck kidneys underlies the progressive cystogenesis we observe in these mice. Overall, our data provide in vivo support for the importance of the biochemical interaction identified between NEK8 and PC2 and compel additional studies to determine the nature of the relationship between these two proteins, both of which are critical for the establishment of left-right asymmetry and are required later in life for the maintenance of renal epithelial integrity.

Concise Methods

Generation of the Nek8:lacZ Allele, Mouse Husbandry, and Genotyping

EL350 E. coli were transformed with a C57BL/6 (B6) BAC (RP23–163A8) containing the Nek8 genomic locus and exon 3 was targeted with an IRES-lacZ cassette via gap repair.⁶⁶ Nek8:lacZ was electroporated into B6/129 hybrid embryonic stem cells^38,67 and a positive clone was injected at the Brigham and Women’s Hospital Transgenic Core. Chimeric males were crossed to B6 females and germline transmission of the Nek8:lacZ allele was confirmed by PCR. Male and female Nek8:lacZ carriers were intercrossed and homozygotes were not present at weaning. Matings were set, embryos were harvested at various gestational time points, and tails were used for genotyping to confirm homozygous offspring. A three-primer genotyping strategy was utilized as follows: a forward primer that anneals to both wild-type and mutant loci 5′-GACACCATTAGGCGCCTTCC-3′ was combined with a wild-type–specific reverse primer 5′-CTTCTCAAAGCAGGCCTTGG-3′ and a mutant-specific reverse primer 5′-GGGGATCCATATTATCATCG-3′. Maintenance and genotyping for the jck mouse line has been described previously,¹¹ and intercrosses were performed between jck/+ females and Nek8+/− males to obtain compound heterozygotes. All animals were housed in accordance with Harvard Medical School Animal Resources and Comparative Medicine regulations.

Zebrafish Morpholino Injection and Phenotype Analyses

For Nek8 morpholino knockdowns, wild-type strain TU-AB zebrafish embryos were injected at the one- to four-cell stage with 4.6 nl of a 0.125 mM, 0.25 mM, or 0.5mM solution of a Nek8 ATG translation blocking oligo: 5′-CTTCTCATACTTCTCCATGTTTTCG-3′. Heart looping was determined at 29 hours postfertilization. Uninjected embryos or embryos injected with a scrambled control morpholino 5′-CCTCTTACCTCAGTTACAATTTATA-3′ exhibited normal development.

Cell Lines, RT-PCR, and Western Blot Analyses

MEFs were derived from E14.5 embryos using standard procedures⁶⁷ and subsequently immortalized via retroviral infection of the large T antigen.⁶⁸ Transcript analysis was performed on RNA isolated with Trizol (Invitrogen) from wild-type and mutant MEF lines; Superscript One-step RT-PCR (Invitrogen) was used to amplify transcripts with primers specific to Nek8 exon 2 (5′-CAACCACCCCAACGTCATCG-3′) and exon 4 (5′-GCTACTCTTCTGGTTGTAGGG-3′).

Nek8 knockdown was performed in IMCD cells⁵⁰ by lentiviral infection of shRNA-expressing constructs that were generated by annealing complementary primers and cloning into the pLKO-puromycin vector (shRNA target sequences were obtained from the Broad Institute shRNA Consortium). A hairpin targeting nucleotide 1570 (sh1570) conferred the most robust knockdown when analyzed in a single infection; we subsequently isolated clonal lines and screened for the loss of NEK8 expression by Western blot and identified line N8KD10.

For fluid shear stress experiments, stable wild-type, N8KD10, and PC2-deficient IMCD cell lines expressing GCaMP3 were generated using a blasticidin-resistant retroviral vector as previously described.⁶⁹ GCaMP3-positive cells were grown in MatTek chambers (35-mm petri dishes with a bottom coverslip insert) to 40%–50% confluency before serum starvation (0.5% FBS) (MatTek, No. 1.5). Cells were serum starved for 96 hours before fluid shear stress experiments in order to reach their highest frequency of ciliation.

For Western blot analyses, cell and tissue lysates were subjected to SDS-PAGE. Blots were incubated with primary antibodies overnight at 4°C at the following concentrations: anti-NEK8 (1:5000¹¹), anti-PC2 (1:1000⁷⁰), and anti-tubulin (1:500; Sigma).

MRI

A multi-channel 7.0-T MRI scanner (Varian Inc.) with a 6-cm inner bore diameter insert gradient set was used to acquire anatomic images of E18.5 mouse embryos. Before imaging, samples were immersed in 2 mM ProHance in PBS (gadoteridol; Bracco Diagnostics Inc.) for a week, and were then placed into 2 mM ProHance in low-melting-point agarose (Fisher). Parameters used in the scans were optimized for contrast within the mature mouse embryo as follows: a T2-weighted, three-dimensional fast spin-echo sequence, with repetition time/echo time (TR/TE) of 325/30 milliseconds, eight averages, field of view of 14×14×25 mm, and matrix size of 348×348×624, giving an image with 40 μm isotropic voxels.

Histology and Immunofluorescence

Embryos and kidneys were fixed in Bouin’s solution and were embedded in paraffin, and 6-μm sections were cut for hematoxylin and eosin (H&E) staining. Kidneys were fixed in 4% paraformaldehyde and were embedded in paraffin, and 8-μm sections were cut for PAS staining and immunofluorescence/lectin procedures. PAS staining (Sigma) was performed according to the manufacturer’s instructions. For immunofluorescence and lectin staining, slides were subjected to citrate buffer antigen retrieval and incubated with antiacetylated tubulin antibody (Sigma) at a 1:10,000 dilution combined with either FITC-conjugated LTL or DBA (Vector Labs) at 4°C overnight. Goat anti-mouse Alexa Fluor 594 (Invitrogen) was used to visualize the tubulin and slides were mounted with Vectashield (Vector Labs).

Whole-mount immunofluorescence was performed on E7.75–E8.0 embryos with anti-NEK8 (1:500) or anti-PC2 (1:1000⁷¹) antibodies overnight at 4°C. Embryos were then incubated with antiacetylated tubulin for 1 hour at room temperature followed by goat anti-rabbit Alexa Fluor 488 and anti-mouse Alexa Fluor 647 secondary antibodies (Invitrogen). Embryos were mounted under individual coverslips with Vectashield (Vector Labs) and processed for genotyping after imaging.

For cilia length analysis, IMCD cells and MEFs were plated on glass coverslips (VWR Scientific); at approximately 30% confluence, the cells were serum-deprived for 48 hours before incubation with antiacetylated tubulin and Alexa Fluor 647. A nonparametric ANOVA test was carried out to compare the cilium length distributions obtained.

Whole-Mount In Situ Hybridization and Fixed Tissue Genotyping

E8.0–E8.5 embryos were harvested and fixed overnight in 4% paraformaldehyde at 4°C, and were then dehydrated through a graded methanol series. For hybridization, embryos were rehydrated and taken through the whole-mount in situ procedure described by Wilkinson.⁷² Embryos were hybridized with digoxigenin-labeled mPitx2c⁷³ and nodal⁴⁴ probes overnight at 65°C, incubated with antidigoxigenin antibody (Roche) conjugated with alkaline phosphatase and probes were visualized with BM purple substrate (Roche). Whole litters were processed simultaneously; after images were obtained, DNA was prepared from the embryos for genotyping. The forward primer is the same as described above, whereas the reverse primers wild-type 5′-CAGAGCCAGCAGGATCTGCAC-3′ and mutant 5′-CGGCTTCGGCCAGTAACGTTAG-3′ efficiently amplify DNA target sequences from fixed tissue.

Renal Explant Culture

Renal explant culture was performed as described by Natoli et al.⁴⁷ Kidneys were harvested from E14.5 embryos and transferred to Corning transwell filters suspended in DMEM/F12/10% FBS. At 24 hours, media were supplemented with either DMSO or 100 μM 8-bromo-cAMP (Sigma) and changed daily for 4 days. Cyst percentage was calculated from images by obtaining the total area of the cysts divided by the area of the kidneys, as determined using ImageJ software and data were analyzed with Prism software. The SEM was calculated from the cyst percentages of pooled data from at least four independent experiments (n=10 kidneys per genotype), and the P values were determined using the Bonferroni multiple comparison test.

Microscopy and Imaging

Whole-mount embryos and renal explants were analyzed on a Leica DM12₅ dissecting microscope and images were captured with Leica FireCam software. Histologic and immunofluorescent experiments were analyzed on a Zeiss Imager.Z1 microscope and images were captured with AxioVision software; immunofluorescence slides were imaged with the ApoTome engaged for better resolution of cilia and nuclei.

Fluid Shear Stress Experiments and Statistical Analyses

GCaMP3-positive IMCD cells were introduced to CO₂ independent Hanks balanced salt solution media supplemented with HEPES (25 mM final concentration), nonessential amino acids, sodium pyruvate, glucose and GlutaMAX. Intracellular calcium stores were purged by challenging cells with 2 mM EGTA and 100 µM ATP before Ca²⁺ influx measurements. Cells were introduced back to EGTA-free Hanks balanced salt solution–based media and the cell chambers were mounted on the stage of a Nikon Eclipse Ti inverted microscope equipped with a Nikon Plan Apo 20×A 0.75 NA objective lens and a CoolSnap-HQ (Photometrics). The Nikon FITC cube was used to efficiently reflect 488 nm wavelength and pass the emission wavelengths into the charge-coupled device camera detection channel. While fluorescence image time series were acquired, cells were challenged with fluid shear stress via the controlled addition of Ca²⁺ media. The acquisition settings were kept constant for all samples so that valid comparisons could be made between measurements from different data sets. Acquisition parameters were set within the linear range of the charge-coupled device camera detection.

A custom MATLAB (The MathWorks Inc) subroutine was written to analyze acquired image time series. The mean intensity of the background noise was calculated from empty dark regions in the images and each frame was individually corrected. Regions of interest (ROIs) from individual cells were chosen for the subsequent analysis. The mean intensity values measured from the first 30 frames (before exposing cells to fluid shear stress) of acquired image time series defined ROIs and used to normalize the values of fluorescence intensity to unity. The SDs of the recovered mean values were obtained from the analysis of multiple ROIs in multiple independent experiments.

Disclosures

None.