Modelling a ciliopathy: Ahi1 knockdown in model systems reveals an essential role in brain, retinal, and renal development

Abstract

Joubert syndrome and related diseases (JSRD) are cerebello-oculo-renal syndromes with phenotypes including cerebellar hypoplasia, retinal dystrophy, and nephronophthisis (a cystic kidney disease). Mutations in AHI1 are the most common genetic cause of JSRD, with developmental hindbrain anomalies and retinal degeneration being prominent features. We demonstrate that Ahi1, a WD40 domain-containing protein, is highly conserved throughout evolution and its expression associates with ciliated organisms. In zebrafish ahi1 morphants, the phenotypic spectrum of JSRD is modeled, with embryos showing brain, eye, and ear abnormalities, together with renal cysts and cloacal dilatation. Following ahi1 knockdown in zebrafish, we demonstrate loss of cilia at Kupffer’s vesicle and subsequently defects in cardiac left–right asymmetry. Finally, using siRNA in renal epithelial cells we demonstrate a role for Ahi1 in both ciliogenesis and cell–cell junction formation. These data support a role for Ahi1 in epithelial cell organization and ciliary formation and explain the ciliopathy phenotype of AHI1 mutations in man.

Electronic supplementary material

The online version of this article (doi:10.1007/s00018-011-0826-z) contains supplementary material, which is available to authorized users.

Introduction

Joubert syndrome and related diseases (JSRD) are autosomal recessive disorders, characterized by a developmental mid-hindbrain malformation. “Molar tooth sign” is the pathognomonic feature on cerebral MRI, which defines the cerebellar vermis hypoplasia, abnormally deep interpeduncular fossa, and elongated superior cerebellar peduncles diagnostic of Joubert syndrome (JBTS) [1]. Consequently, neurological anomalies can include hypotonia, ataxia, gaze palsy including oculomotor apraxia, learning difficulties, and abnormal neonatal breathing patterns. JSRD is considered a multisystem disease, with extra-neurological features including retinal degeneration [1–3], coloboma, cystic kidney disease (nephronophthisis [4] and multicystic renal dysplasia [5]), polydactyly and hepatic fibrosis [6, 7].

In addition to the clinical heterogeneity of JSRD, 11 causal genes have been identified in patients. These include JBTS types 1-11: INPP5E [8], TMEM216 [9], AHI1 [4, 9], NPHP1 [11], CEP290 [12–14], TMEM67 [15, 16], RPGRIP1L [5], ARL13B [17], CC2D2A [18], CXORF5 [19] and TCTN2 [22]. The protein products encoded by all of these genes have been localized in the primary cilium [20], a highly conserved cellular organelle, central to the regulation of cellular signaling pathways [21]. TCTN2 is a gene that has recently been associated with JBTS in three families [22]. Defects in ciliogenesis were observed follow Tctn2 knockdown in vivo [22]. Acknowledgment of the fact that the protein products of all of these genes have a role in the primary cilium, explains why JBTS, nephronophthisis, and other disorders caused by mutations in these genes are referred to as ciliopathies [21].

Mutations in AHI1 (Abelson-helper integration site-1) are a common genetic cause of JBTS, accounting for 12% of cases, and 20% of individuals with JBTS and Leber’s congenital amaurosis [9, 23–26]. AHI1 is highly conserved throughout evolution and encodes the Ahi1 protein (also known as Jouberin) [9]. By performing a detailed in silico search of proteomes of eukaryotic organisms, we have now identified additional homologues, confirming a conserved role for this gene between ciliated protozoa and man.

We have previously demonstrated that Ahi1 localizes to centrosomes/basal bodies of renal epithelial cells, and that it interacts with the protein product of NPHP1, nephrocystin-1 [27]. At a genomic level, there is evidence for AHI1 mutations or polymorphisms having an oligogenic affect and modulating the phenotype. Five patients with homozygous NPHP1 mutations combined with a R830W mutation in AHI1 have been identified [28]. This triallelism resulted in a more severe CNS defect [28]. Similarly, the relative risk (RR) of a retinal defect (such as retinal dystrophy) associated with nephronophthisis was significantly increased (RR = 7.5; 95% CI 4.0–11.2) in the presence of the R830W AHI1 mutation [26]. Early expression (from embryonic day 10.5) of Ahi1 is seen in murine brain, which persists into adulthood [9]. Previous work using zebrafish embryos aged 2.5–5 days post-fertilization has documented an expression pattern for ahi1 in the brain (olfactory bulb, telencephalon, diencephalon, tectum, and cerebellum) and retina [29].

Given this evidence implicating an important role for AHI1 in modulating brain, eye, and other phenotypes in JSRD, we investigated the effect of ahi1 knockdown in both cell and animal models. Zebrafish embryos transiently express a ciliated structure called Kupffer’s vesicle (KV) [30]. Cilia in KV generate a leftward fluid flow, which is fundamental for the development of left–right patterning and asymmetry during organogenesis [31]. Since the Ahi1 protein Jouberin localizes to the primary cilium/basal body complex [27], we were interested in evaluating the potential role of ahi1 in KV formation and the subsequent development of left–right asymmetry [32]. ahi1 knockdown in zebrafish led to a ciliopathy phenotype and multisystem disease, consistent with the human Joubert syndrome phenotype. Here we reveal the first evidence that ahi1 knockdown in zebrafish embryos leads to loss of cilia from KV and subsequently altered cardiac left–right asymmetry. ahi1 morphants also develop pronephric duct dilatation and cloacal abnormalities, in association with loss of cilia.

To extend these studies, we used a murine epithelial cell model to evaluate the function of Ahi1. Knockdown of Ahi1 in murine renal epithelial cells prevented development of primary cilia from basal bodies and reduced the ability of cells to form adherens junctions with neighboring cells, leading to a disordered epithelium.

Together, this novel data emphasizes the important and extensive role of AHI1 in development, from essential cellular signaling organelles, such as the primary cilium, to complex organ systems including the brain and the kidneys.

Methods

Sequence analysis

Putative Ahi1 orthologues were identified using a combination of reciprocal best BLASTP and iterative BLASTP as well as simple BLAST searches [33]. This initial search generated a very large number of false positives, which were only similar in terms of WD40 repeats. To overcome this problem, we performed further searches using human or Naegleria gruberi Ahi1 amino acids 1–540. These protein sequences were used to query the non-redundant predicted proteomes of 44 organisms (33 flagellate, 11 non-flagellate, Supplementary Table 1) chosen to represent a wide evolutionary spread of eukaryotes. Searches were carried out at NCBI (http://www.ncbi.nlm.nih.gov/) or JGI (http://www.jgi.doe.gov/) depending on the organism. An alignment was generated using MAFFT (Multiple Alignment using Fast Fourier Transform), trimmed to exclude the WD40 repeats and the low complexity N-terminal region, and displayed using GeneDoc (http://www.nrbsc.org/gfx/genedoc/) with “similarity group” shading in Conservation Mode. Pattern and profile searches were carried out used SMART, [34, 35] and Coils2 [36].

Zebrafish husbandry and genetic manipulation

Wild-type AB or golden zebrafish (ZF) were maintained and raised using standard animal husbandry. For zebrafish studies, all procedures were performed under Home Office UK license regulations. Embryos were kept at 28.5°C and 0.003% PTU (1-phenyl-2-thiourea; Sigma) was used to suppress pigmentation when necessary. Embryos were staged according to somite number for studying KV or hours post-fertilization (hpf) for later time points [37]. The cldnb:Lyn-GFP transgenic line (expressing Lyn-GFP driven by the claudin-b promoter) was a gift from Dr S. Burtey. cldnb:Lyn-GFP embryos express GFP at epithelial cell tight junctions with strong expression within the telencephalon, otic placode and pronephros (data not shown). To visualize cardiomyocytes and phenotype for cardiac looping abnormalities, we used transgenic zebrafish expressing GFP under control of the cmlc2 (myl7) promoter (cmlc2:GFP) [38]. Splice (5′-CCACACTCTGAAAGGGAAAAACATT-3′) and translation (5′-GAGTCATTAGCAGCTTTGTTTTTCC-3′) blocking antisense morpholino oligonucleotides (MOs) were designed (Gene Tools, Philomath, Oregon, USA) to target zebrafish ahi1 (NM_001077561). Golden and cldnb:Lyn-GFP transgenic zebrafish embryos (1–4 cell stage) were microinjected with between 0.5 and 6 ng MO diluted in Danieau solution and 0.5% phenol red. Mismatched control (5′-CCTCTTACCTCAGTTACAATTTATA-3′) and p53 (5′-GCGCCATTGCTTTGCAAGAATTG-3′) MOs were used in control experiments. Phenotypic rescue was performed by co-injecting 100 pg of in vitro transcribed mouse Ahi1 mRNA (NM_026203) (mMessage Machine, Ambion) with ahi1 splice or translation blocking MOs. Zebrafish were anesthetized with Tricaine solution and phenotyped at 56 hpf (to assess cardiac looping) and 72 hpf using morphology tables and severity scores. Images were captured using a fluorescent stereomicroscope (Leica MZ16F).

Zebrafish RNA isolation and RT-PCR

Total RNA was isolated from single ZF embryos from each experimental group at 24, 48, and 72 hpf. Embryos were anesthetized in Tricaine and washed in phosphate-buffered saline (PBS). A TissueRuptor (Qiagen) was used to homogenize tissue and a standard 1 ml Trizol (Invitrogen) protocol performed. Isolated RNA was resuspended in 10 μl RNAse-free water and quantified (Nanodrop ND-800 Spectrophotometer, Labtech). An equal concentration of RNA was used for each experimental group in reverse transcription (RT) reactions. Superscript VILO cDNA synthesis kit (Invitrogen) was used for RT. PCR using gene-specific primer pairs was performed: ahi1 forward (5′-GAGGTCAGATGGGCTGTTTT-3′) and reverse (5′- AGACCCAGGCATAACTTTCG-3′) using an annealing temperature of 55°C and 35 cycles. Following electrophoresis, equal volumes of PCR products were visualized on 1.5% agarose gels. Bands were excised, purified (Qiagen) and directly sequenced (MWG Eurofins).

Whole-mount immunohistochemistry

Uninjected and ahi1 MO injected cldnb:Lyn-GFP and golden ZF embryos were fixed at either 8–10 somite stage (in order to study KV) or 72 hpf, using 4% paraformaldehyde in PBS at 4°C overnight. To permeabilize embryos they were washed in ddH20 then pre-chilled acetone (−20°C) for 7 min. Embryos were washed in ddH2O and blocked in 5% bovine serum albumin, with 1% DMSO (Sigma) and 0.1% Tween. For cilial staining, embryos were incubated in primary antibody (mouse anti-acetylated tubulin antibody, 1:500, Sigma T6793) overnight at 4°C and detected using a donkey anti-mouse AlexaFluor594 conjugated secondary antibody (1:200, Invitrogen). For identification of KV epithelium, antibodies directed towards aPKC [39] were used (rabbit anti-aPKC (1:500, Santa Cruz) detected with goat anti-rabbit Alexa Fluor 488 conjugated secondary antibody (1:200, Invitrogen). Embryos were washed into PBS, mounted and imaged using confocal microscopy (LSM 510Meta, Zeiss and A1R Confocal, Nikon).

In situ hybridization and histology in zebrafish embryos

Whole-mount in situ hybridization was performed according to standard procedures [40] using golden ZF embryos. PCR primers used for generating zebrafish-specific ahi1 in situ hybridization probes were: forward (5′-TCATTCGTGTACTGTGCCAAG-3′) and reverse (5′-CAACCTGGTCACCTGTCTCA-3′) (yielding a 544 bp PCR fragment). PCR fragments were cloned into pGEM-T Easy vector (Promega). Two additional ahi1 riboprobes were generated using PCR with T7 and SP6 labeled oligonucleotides. Primer pair sequences were (5′-CAAGAGGAAGCTCCCACAAC-3′) and (5′-GGCTTTCTTTCGTGTCGATT-3′); (5′-AGGATGATGTAGAGGATTCAAGG-3′) (5′-CTCATAGAAGGATGAAACATGACG-3′), respectively.

Antisense and sense control probes were transcribed with SP6 or T7 RNA polymerase, respectively, and labeled using the DIG RNA labeling kit (Roche Diagnostics). The hybridization temperature was 65°C for all probes. A charon in situ hybridization probe (kind gift from Dr. Hibi) was used to localize KV epithelium [41]. Sonic hedgehog (shh) [42] and cardiac myosin light chain 2 (cmlc2) probes [43] were used as positive controls for zebrafish in situ hybridization experiments. For negative controls, a sense ahi1 probe, a sense charon probe and no probe were used. Embryos were imaged using an Axioplan microscope (Axiocam HRC, Zeiss). For pronephric duct identification, following in situ hybridization protocols, ZF embryos were embedded in resin and sectioned at 5–8-μm intervals as described below. Sections were mounted before imaging using an axioplan microscope (Axiocam HRC, Zeiss).

For histological examination, ZF embryos were fixed at 72 hpf in 4% PFA in PBS at 4°C overnight. They were washed in DEPC-PBS at RT. Embryos were dehydrated by washing in ethanol series with DEPC-PBS (30, 50, 70, 90%) up to 100% ethanol. A standard resin embedding (Technovit, Heraeus Kulzer GmbH) protocol was performed and embryos were positioned in plastic moulds and allowed to set at 4°C for 48 h prior to sectioning at 5 μm using a microtome. The sections were mounted onto slides and stained using a standard methylene blue and fuchsin protocol [44]. Images were obtained using an axioplan microscope (Axiocam HRC, Zeiss).

Cell culture and siRNA using IMCD3 cells

Mouse inner medullary collecting duct (IMCD3) cells were cultured in DMEM/Ham’s F12 supplemented with 10% fetal calf serum (Sigma-Aldrich, UK). For functional analyses, passage 13–20 IMCD3 cells were cultured on Transwell permeable supports (# 3431, Corning). For SEM studies, passage 13–20 cells were cultured on 13-mm glass cover slips.

Cells were transfected with a pool containing 100 pmol of each of four siRNA duplexes (OnTargetPlus SMARTpool, Dharmacon) against mouse Ahi1 at 60–70% confluency using Lipofectamine 2000 according to the manufacturer’s instructions. The medium GC non-targeting control (Invitrogen) was used as a negative control. siRNAs were as follows: oligonucleotide 1, 5′-GGUCAAAAGACGAUCGCUA-3′; oligonucleotide 2, 5′-UGAAGUUAGCCGCCGUGUAA-3′; oligonucleotide 3, 5′-GCUAAAUGUCGUCGAGGUU-3′; oligonucleotide 4, 5′-GUGAAACACUGUAUCGAGA-3′. As a further control, siRNA studies were repeated with two individual siRNA oligonucleotides (numbers 2 and 3, above). All assays were carried out 96 h after transfection.

The following primary antibodies were used: anti-acetylated alpha tubulin clone C3B9 [45], rabbit anti-gamma tubulin (Sigma-Aldrich), rat anti-ZO-1 (MAB1520, Chemicon) and mouse anti-E-cadherin clone 36 (BD Transduction Laboratories). Secondary antibodies were: goat anti-mouse AlexaFluor488, goat anti-rabbit AlexaFluor594 (Molecular Probes) and goat anti-rat FITC (Jackson Immunoresearch). Western blotting was performed on whole-cell extracts to confirm siRNA efficiency as previously described [46].

We used mouse anti-AHI1 antibody at 1 in 500 dilution (ab93386, Abcam) together with rabbit anti-histone H3 (1:200) as a loading control. HRP-conjugated secondary antibodies (Dako) were used at 1:20,000. Blots were developed using Western Lightening enhanced chemiluminescence (Pierce) and scanned using Multi Gauge V2.2 software (Fujifilm) to determine relative band intensities.

Statistical analysis

Means and standard errors of means (SEM) were calculated using GraphPad Prism software. Results are presented as means ± SEM. For comparisons of phenotypes, contingency tables were generated. Fisher’s exact test was used to compare two groups and Chi-squared test was used to compare more than two groups, with a level of significance at p < 0.0001.

Results

AHI1 encodes a ciliary/basal body protein which is highly conserved throughout evolution

The human AHI1 gene encodes an 1,196 amino acid protein AHI1 (alias Jouberin), which is expressed in primary cilia and centrosomes of renal epithelial cells [27]. SMART domain analysis [34, 35] predicts several known protein domains including an N-terminal coiled-coil domain, six WD40 domains and a Src homology domain (SH3) (Fig. 1 and Supplementary Figure 1), which has recently been characterized using crystallography [47]. There is significant conservation of amino acid structure between human, mouse and zebrafish Ahi1 [48], with the exception of the absence of a predicted N-terminal coiled-coil domain [29] and an additional predicted WD40 domain in the murine isoform (Supplementary Figure 1).

Fig. 1.

Partial protein sequence alignment of predicted Ahi1 homologues in eukaryotes. The human AHI1 protein is shown schematically with known protein domains indicated. The alignment was trimmed relative to the human sequence as shown to display only amino acids 344–520, which demonstrate the presence of clear homologues outside the WD40 repeat region. Homology was determined by examination of candidate sequences including reciprocal BLASTP. Shading indicates conservation of amino acids with similar R groups. Dark grey denotes conserved in 50% of organisms; mid-grey denotes conserved in 30% of organisms; light grey denotes conserved in 10% of organisms. For full organism common and Latin names, see Supplementary Table 1

Given the strong conservation of Ahi1 among vertebrates (Supplementary Figure 1) and its localization and likely role in primary cilia and basal bodies, we sought to identify additional homologues. We found that the nucleotide and protein sequences of Ahi1 are highly conserved throughout evolution (Fig. 1 and Supplementary Table 1). We searched the predicted proteomes of 44 eukaryotic organisms chosen to represent a wide evolutionary spread of organisms, including those that build a cilium or flagellum (33 organisms) and those that lack these structures (11 organisms). Despite having only been described in multi-cellular organisms to date, we report that Ahi1 evolved prior to the evolution of multicellularity and is conserved between ciliated protozoa and man.

We found a clear correlation between the presence of a predicted Ahi1 protein in genomes of organisms that build cilia or flagella, and absence in the genomes of organisms that do not build cilia or flagella (Supplementary Table 1). Moreover, putative Ahi1 orthologues were present in the genomes of organisms that build only sensory cilia (Daphnia, Tribolium) as well as in the genomes of organisms that build either motile cilia/flagella only, or both motile and sensory cilia. Interestingly, Ahi1 homologues were only found in the genomes of organisms that build a canonical nine triplet centriole (9 + 0) and were invariably absent from the genomes of organisms that have a specialized singlet or doublet centriole architecture. This implies a possible role for Ahi1 in building or maintaining a cilium from a triplet centriole.

ahi1 expression patterns during early zebrafish development reveal prominent retinal and brain expression

We examined ahi1 expression in developing zebrafish using in situ hybridization in whole mount embryos. Identical patterns of ahi1 expression were seen with 2 additional ahi1 gene-specific riboprobes (data not shown). Negative controls omitting antisense riboprobe or using ahi1 sense riboprobes revealed very low levels of background staining (Supplementary Figure 2).

At the 8–10 somite stage, ahi1 expression was seen at KV (Fig. 2a, b). As a positive control for the localization of KV, we show expression of charon (Fig. 2c, d), a known marker of KV [41]. Patterns of expression within KV of ahi1 and charon are comparable. At the 8–10 somite stage ahi1 expression is also seen in the developing retina (Fig. 2e). There is no prominent notochord expression of ahi1, especially when compared to expression of shh (Supplementary Figure 2J). At 24 hpf, the forebrain (telecephalon, diencephalon), mesencephalon (tectum) and hindbrain (cerebellum) show ahi1 expression (Fig. 2f) in addition to strong retinal expression of ahi1 (Fig. 2h). At 48 hpf, ahi1 was expressed in the olfactory bulbs, retinal layers and inner ear (Fig. 2h, i). At 72 hpf, ahi1 expression continues to be prominent in the inner ear (Fig. 2j), functional retina (ganglion cell layer, inner nuclear layer and outer nuclear layer) and olfactory bulbs (Fig. 2j, k). In order to establish whether ahi1 was expressed in the developing pronephros, we used ahi1 in situ hybridization in whole mount 48 hpf embryos followed by resin embedding and tissue sectioning. We demonstrate ahi1 expression within the pronephric ducts (Fig. 2l), with prominent expression seen at the apical (luminal) surface.

Fig. 2.

ahi1 expression during embryogenesis in zebrafish. ahi1 expression studies in the developing zebrafish. Whole-mount in situ hybridization was performed using zebrafish embryos at a–e 12–14 hpf (8–10 somites), f, g 24 hpf, h, i 48 hpf and j, k 72 hpf. Scale bar 100 μm. Expression of ahi1 is seen at KV, near the tailbud in lateral (a) and dorsal (b) views. Expression of charon, as a known marker of KV is shown for comparison (c, d). Retinal expression of ahi1 is seen at 12 hpf (e, white arrow) while notochord expression is weak. At 24 hpf, the forebrain (telencephalon, diencephalon), mesencephalon (tectum, arrow in f) and hindbrain (cerebellum, arrow head in f) all show ahi1 expression, together with retinal expression (arrowed in g). At 48 hpf ahi1 is expressed in the hypothalamus and hindbrain (arrows in h) and inner ear (arrowheads in i). At 72 hpf ahi1 expression continues to be seen in the inner ear (arrowed in j), olfactory bulbs and retina including the ganglion cell layer (GCL), inner nuclear layer (INL) and outer nuclear layer (ONL) (arrowed in k). l A transverse section of a 48 hpf embryo following in situ hybridization with ahi1 probe, demonstrates expression of ahi1 in the distal pronephros (white arrows). NC notochord; S somites

Knockdown of ahi1 in zebrafish reveals phenotypes typical of a ciliopathy

Antisense morpholino oligonucleotide injections targeting the ahi1 ATG initiation codon (ATG MO) and an internal splice donor sequence of exon 8 (SPL8 MO) both caused defects in convergent extension, eye, hindbrain and ear development (Fig. 3e–o), as well as cyst formation in the pronephric kidney tubules (Fig. 3g, h, i, r). A variety of otic vesicle abnormalities were observed, with either absent, single or triple otoliths (Fig. 3j–o). These abnormal developmental phenotypes are comparable to the clinical features seen in humans with JSRD caused by AHI1 mutations, with a preponderance towards central nervous system defects, and milder renal phenotypes [4, 9, 23–26]. A mismatch control morpholino (mmMO) had no effect on nervous system development or renal cyst formation (Fig. 3b). Co-injection of murine Ahi1 mRNA with the ahi1 SPL8 MO completely rescued the morphant phenotype (Fig. 3d), suggesting specificity of the ahi1 knockdown. RT-PCR of 24–72 hpf embryos following ahi1 SPL8 MO injection demonstrated a splicing defect (Fig. 3p) with a PCR product of 453 bp reduced from 807 bp, with maximal knockdown occurring after 48 hpf. Direct sequencing confirmed skipping of exon 8 and introduction of a premature stop codon (data not shown). A modest dilatation of the proximal pronephros was seen in ahi1 SPL8 MO-injected embryos, with lumen diameter 5–10 μm, when compared to WT proximal pronephros diameter of 3–5 μm (Fig. 3q, r).

Fig. 3.

ahi1 knockdown in zebrafish recapitulates a ciliopathy phenotype. ahi1 MO-injected embryos show a range of dysmorphology. Lateral views of 72 hpf embryos including a wild-type (WT) uninjected, b mismatched MO injected, c murine Ahi1 mRNA injected and d rescue of ahi1 MO-injected embryo with murine Ahi1 mRNA were phenotypically normal. e 4 ng of ahi1 translation blocking MO-injected embryo (ahi1 ATG MO) and f, g 1 ng ahi1 SPL8 MO-injected embryos demonstrate curved body axis, hydrocephalus (asterisk), h coloboma with abnormal eye development, pericardial edema and pronephric cysts (indicated by arrow, g–i). h, i Zoomed views of ahi1 SPL8 MO-injected embryos demonstrating pronephric cysts (indicated by arrow). Lateral (j, l, m) and dorsal (k, m, o) views of otic placode. j, k Wild-type otic placode containing two otoliths, l, m empty otic placode in ahi1 ATG MO-injected embryo, n, o single otolith in ahi1 SPL8 MO-injected embryo. p RT-PCR of mRNA from single WT (expected size 807 bp, arrowed) and ahi1 SPL8 MO-injected embryos at 24–72 hpf revealing abnormal splicing of mRNA (band seen at 453 bp, arrow head) with an in-frame deletion of exon 8 (354 bp). A maximal knockdown of ahi1 is seen at 48 hpf, with some recovery of wild-type RT-PCR product at 72 hpf. q, r Resin sections of WT, and ahi1 SPL8 MO-injected embryos at 72 hpf (black arrows, pronephros). q WT undilated pronephros. r Mildly dilated pronephros (black arrows) induced by ahi1 SPL8 MO. Scale bar 10 μm

Given the strong retinal expression seen with ahi1 in situ hybridization (Fig. 2) and the presence of abnormal eye shapes in ahi1 morphants (Fig. 3), we carried out histological analysis of the retina using control and ahi1 morphants at 72–120 hpf. We observed a defect in lamination of the cell layers of the retina including the ganglion cell layers and the inner and outer nuclear layers in ahi1 morphants. This was in contrast to control embryos, where normal lamination was seen (Supplementary Figure 3). The development of the retinal pigment epithelium (RPE) however appeared normal in ahi1 morphants.

To quantify the phenotypes of ahi1 morphants, we examined developing embryos at 72 hpf. The most common phenotype observed following ahi1 knockdown was body axis curvature (“curly tail”), occurring in a dose-dependent manner, in a mean of ~27% and ~12% of morphants injected with 6 and 3 ng of ahi1 ATG MO, respectively (Supplementary Figure 4A). Similarly, a curly tail phenotype was seen in ~43 and ~39% of morphants injected with 2 and 1 ng ahi1 SPL8 MO, respectively (Supplementary Figure 4B). We also observed dose related defects of the nervous system (hydrocephalus) in a mean of 12 and 3% of embryos with 6 and 3 ng ahi1 ATG MO injection, respectively (Supplementary Figure 4A). Similarly, hydrocephalus was seen in a mean of 31 and 22% of embryos with 2 and 1 ng ahi1 SPL8 MO injection, respectively (Supplementary Figure 4B). Generally, all the phenotypic anomalies were more prevalent with ahi1 SPL8 MO rather then ahi1 ATG MO injection, with a dose related response consistently evident (Supplementary Figure 4A,B). Higher doses of ahi1 SPL8 MO (up to 6 ng) were attempted in order to induce an increased frequency of abnormal phenotypes, particularly of pronephric cysts. However, such doses led to an excess of mortality at 24 hpf (Supplementary Figure 4C).

Complete rescue of the morphant phenotype (indistinguishable from uninjected wild-type) occurred in 89% of embryos (n = 62) embryos following co-injection of ahi1 MO with murine Ahi1 mRNA compared to 21% (n = 106) of embryos with a wild-type phenotype following injection with ahi1 MO (Supplementary Figure 4D, p < 0.0001, Fisher’s exact test). For each phenotype, mRNA rescue of ahi1 knockdown showed a significant reduction (p < 0.0001, Fisher’s exact test) of numbers of embryos displaying the abnormal phenotype (Supplementary Figure 4E).

Knockdown of ahi1 in zebrafish reveals a loss of cilia at Kupffer’s vesicle and a reversal of cardiac looping

Developing zebrafish transiently express a ciliated structure called KV [30], which is the equivalent of the mouse embryonic node and important in establishing left–right patterning. We looked for abnormalities in cilia in KV following ahi1 MO injection. Under light microscopy at the 8–10 somite stage, KV is visualized in uninjected control and ahi1 morphant embryos, evident in both lateral and dorsal views (Supplementary Figure 5A,B,D,E). KV remains intact and morphologically preserved. Using an in situ hybridization probe directed towards charon, a KV-specific marker [41], we confirm that there is no change in the expression pattern within KV in ahi1 morphants compared to control uninjected embryos (Supplementary Figure 5C,F).

We next performed immunostaining using an acetylated tubulin antibody to study cilia, on control and ahi1 morphants at the 8–10 somite stage. We demonstrated in cldnb:Lyn-GFP embryos, that KV is present and was ciliated in control embryos (Fig. 4a–c). However, in ahi1 morphants, while KV was identified using cldnb:Lyn-GFP fluorescence (Fig. 4d) there was some disorganization of the usual expression pattern of cldnb:Lyn-GFP, together with loss of cilia from KV (Fig. 4d–f). In repeated experiments using embryos of 8–10 somites, cilia were absent from KV in 23 out of 25 (92%) cldnb:Lyn-GFP embryos following ahi1 MO injection (Supplementary Figure 6A).

Fig. 4.

Loss of cilia at Kupffer’s vesicle in ahi1 morphant zebrafish embryos. a–f cldnb:Lyn-GFP transgenic zebrafish embryos were fixed at the 8–10 somite stage and labeled with anti-acetylated tubulin (red) for visualization of cilia at Kupffer’s vesicle (KV). a–c Uninjected embryos show cldnb:Lyn-GFP expression (green) at cell–cell junctions of epithelial cells within KV, with a ciliated (red) apical surface. d–f ahi1 SPL8 MO injected embryos demonstrate disorganized cldnb:Lyn-GFP expression at KV (green) and a loss of cilia. g–l Fluorescent immunostaining of golden zebrafish ahi1 morphant embryos fixed at 8–10 somite stage and labeled with aPKC antibodies (green) to identify KV epithelium and anti-acetylated tubulin (red) for visualization of cilia. g–i An intact KV epithelium, with normal labeling of aPKC and preserved cilia expression. j–l A disrupted KV epithelium with diffuse aPKC signal and almost complete absence of cilia. Scale bar 20 μm

In order to further assess this apparent disruption to the development of KV in ahi1 morphants, we injected golden embryos with ahi1 MO and used primary antibodies directed towards aPKC, a marker of the apical membranes of KV cells [49], in combination with the ciliary marker, acetylated tubulin. In 8–10 somite embryos we were able to visualize KV, using aPKC staining (Fig. 4g–l). Some ahi1 morphants had a preserved KV appearance with a normal aPKC staining pattern and preserved cilia (Fig. 4g–i). While in some other ahi1 morphants, there was a less distinct pattern of aPKC staining, associated with almost complete absence of KV cilia (Fig. 4j–l).

Cilia within KV help to generate a leftward fluid flow, termed nodal flow [31]. This initiates a set of molecular signals which determines laterality. Since we have shown that cilia are absent within KV in ahi1 morphants, we hypothesized that this would lead to a defect in organ laterality. We therefore examined developing embryos for abnormalities in cardiac asymmetry.

The normal zebrafish heart extends initially from the midline to the left side in approximately 97% of embryos, before looping places the ventricle closer to the midline and ultimately to the right of the atrium (D-loop) by approximately 48 hpf. Usually, in up to ~2% of embryos this is reversed (L-loop) and in 1% the heart remains in the midline (no loop) [50, 51]. Aberrations of left–right asymmetry are associated with an increased frequency of reversed loops (L-loop) or no loop (when the heart tube remains straight and often lies in the midline) [32].

In uninjected control zebrafish embryos examined at 56 hpf (n = 145), the heart was seen to loop towards the right in 100% of embryos (Fig. 5). Strikingly, in ahi1 injected embryos (n = 221), 68% of embryos demonstrated reversed cardiac looping, with the heart in a L-loop pattern, 6% of embryos showed no looping, while 26% of ahi1 morphants showed the normal phenotype of D-looping (Fig. 5 and Supplementary Figure 6B). Mismatch morpholino (mmMO) injection produced a normal pattern of heart looping (Fig. 5d).

Fig. 5.

Heart laterality defects after ahi1 knockdown. Paired brightfield and GFP fluorescence images are shown of cmlc2-GFP zebrafish embryos at 56 hfp. Examples of a normal D-looping of heart in uninjected embryos, b no loop, and c L-looping in ahi1 SPL8 MO injected embryos are shown. Ventricle, V and atrium, A are arrowed. d Summary table of cardiac looping defects

Knockdown of ahi1 in zebrafish is associated with pronephric duct and cloacal dilatation and loss of pronephric cilia

Using cldnb:Lyn-GFP transgenic zebrafish, using fluorescent microscopy, we visualized the epithelium of the pronephric ducts. In uninjected control embryos, ciliated pronephric ducts, with a non- dilated lumen and a uniform columnar epithelium were observed (Fig. 6a–c). In mismatched MO injected fish no cystic dilatations were observed (data not shown). However, ahi1 morphants displayed a range of phenotypes. Figure 6d–f demonstrate cloacal dilatation (Fig. 6e, asterisk) where the expression pattern of cldnb:Lyn-GFP in the pronephric ducts was preserved, together with preserved cilia within the pronephros (Fig. 6d, arrowed). In ahi1 morphant embryos with more severe phenotypes, cldnb:Lyn-GFP expression in the pronephros appeared disorganized, with dilatation of the pronephros and an absence of cilia within the pronephric duct (Fig. 6g–i).

Fig. 6.

Pronephric duct dilatation and absent cilia in ahi1 morphant zebrafish embryos. Fluorescent immunostaining of cldnb:Lyn-GFP transgenic zebrafish embryos at 72 hpf using anti-acetylated tubulin antibody (red), as a cilial marker. a–c Uninjected control zebrafish embryos demonstrate a ciliated pronephros (a, arrowed) with cldnb:Lyn-GFP expression seen in a columnar epithelium (b and overlay c). d–i ahi1 SPL8 MO-injected embryos demonstrate a spectrum of severity with dilatation of the pronephric ducts (e and h, asterisk), with disordered cldnb:Lyn-GFP expression and either a preservation of cilia (d and overlay f) or an absence of cilia (g and overlay i). Scale bar 50 μm

Knock down of Ahi1 in IMCD3 cells leads to a reduction in primary cilia and cell polarity defects

Given our data showing the requirement of ahi1 for ciliogenesis in KV and the pronephros of the zebrafish, we tested the effect of gene silencing of Ahi1 on ciliogenesis in murine collecting duct cells. We used a pool of four siRNA duplexes targeted against Ahi1, as well as two individual duplexes. siRNA was transiently transfected into IMCD3 renal epithelial cells, and the presence of cilia assessed 96 h later using anti-acetylated α-tubulin antibody as a ciliary marker. We used Western blotting to determine that siRNA had been successful (Fig. 7d). In contrast to wild-type (data not shown) and negative control siRNA-transfected cells, ciliary formation was reduced in Ahi1-silenced cells (Fig. 7a–c), although centrioles/basal bodies could still be observed with anti-γ-tubulin antibody (Fig. 7a). Ciliary formation was significantly reduced (p < 0.0001, Chi-square test) on silencing of Ahi1: 74% of negative control-transfected cells showed ciliary staining, while Ahi1-silenced cells using a pooled Ahi1 siRNA displayed cilia in only 24% (Fig. 7b). In addition to an absence of primary cilia, scanning electron microscopy revealed an indistinct pattern of cell–cell boundaries in Ahi1 siRNA-treated cells (Fig. 7c), suggesting a problem with formation of a polarized epithelium. The establishment of cell–cell contacts is essential for epithelial polarity and tissue morphogenesis [52, 53]. We therefore tested the effect of Ahi1-silencing on the distribution of the tight-junction marker ZO-1 and the adherens junction marker E-cadherin and found an abnormal appearance of E-cadherin in Ahi1 siRNA-treated cells compared to control cells (Fig. 7e). In contrast, there was no apparent difference in tight junction appearance (Fig. 7f). Only one nucleus was generally visible per ring of tight junctions (Fig. 7f), indicating that the lack of cell boundaries at the apical surface is not due to multinucleated cells, and suggesting that cytokinesis failure and cell fusion are not features of the phenotype.

Fig. 7.

siRNA-mediated knockdown of Ahi1 causes loss of cilia and cell polarity defects. Immunofluorescence (a, e, f), scanning electron microscopy (c) and Western blot (d) of IMCD3 monolayers transfected with negative control siRNA or siRNA against Ahi1. a, b Loss of primary cilia (acetylated alpha-tubulin staining, green) in Ahi1-silenced cells (right panels) compared to the cilia seen in control-transfected cells (left panel, white arrows). DNA is labeled with DAPI (blue) while the centrosome (gamma-tubulin staining) is shown in red. b Quantification of primary cilia in transfected cells. At least 1,500 cells were counted for each condition; n = 5 experiments. In control-treated cells, 74% were ciliated, while 24% of cells were ciliated following treatment with pooled AHI1 siRNA (*, p < 0.0001, Chi-square test). c As well as the loss of primary cilia seen in control cells (c, left panel, white arrows), Ahi1-silenced cells failed to properly differentiate the apical cell surface. Very few microvilli were present and cell–cell boundaries were indistinct compared to those in control cells (left panel, red arrowheads). d Western blot showing success of siRNA. The pooled siRNA sample showed an 87% reduction in band intensity relative to the control, while the individual siRNAs gave a reduction of 72 and 69% for AHI1 oligo 1 and AHI1 oligo 2, respectively. e, f Aberrant appearance of adherens junctions (e, E-cadherin staining) in Ahi1-silenced cells compared to the control; in contrast tight junctions (f, ZO-1 staining in green, DAPI in blue) appear normal. NB: all Ahi1 siRNA images show cells transfected with a pool of siRNAs. Scale bars 5 μm

Discussion

WD40 repeats are enriched in ciliary and basal body proteins [54] and although Ahi1/Jouberin was not identified as part of this predicted flagellar and basal body proteome [54], it has subsequently been identified as a member of the ciliary proteome [55], and a member of the photoreceptor sensory cilium complex [56].

Additional evidence for a ciliary role for Ahi1 was provided by the localization of Ahi1 to the mother centriole and the basal body [57]. Using a bioinformatics approach, we have confirmed a pattern between Ahi1 expression and ciliated organisms (Supplementary Table 1). Similar profiles have previously been observed for flagellar proteins [54, 58, 59], suggesting that the localization of Ahi1 to the mother centriole and the basal body [57] is functionally significant. These in silico data indicate that Ahi1 is an ancient protein that appeared prior to the evolution of multicellularity suggesting that Ahi1 may play a role in forming both motile and sensory cilia. The presence of a predicted Ahi1 homologue in the Aureococcus anophagefferens genome is interesting, as this organism is not known to have a flagellate life-cycle stage [60, 61]. No centrioles have been observed in Aureococcus to date; however, the A. anophagefferens genome does have components of the core centriole [62] and it has been proposed to have a flagellate stage in its life-cycle based on the presence of flagellar components in its genome [61, 63, 64]. Our genomic analysis showing the presence of an Ahi1 homolog in A. anophagefferens is entirely consistent with this idea. We also found a markedly divergent Ahi1-like sequence in Thalassiosira pseudonana. The centriole architecture for this organism is unknown; however other centric diatoms build a doublet centriole [65]. This might indicate that Thalassiosira is an exception to the restriction of Ahi1 to the genomes of organisms that build a triplet centriole. Ultrastructural analysis of centriole architecture in this organism will be needed to see if this is indeed the case.

Consistent with previous studies [57], we did not observe a complete abolition of ciliogenesis in siRNA-treated IMCD3 cells, but rather a significant reduction in the number of primary cilia observed. Similar experiments using siRNA knockdown of ciliopathy genes in IMCD3 were recently reported [22]. Here, siRNA targeting Ahi1 did not lead to a significant change in ciliogenesis, rather a modest defect in spheroid growth in 3D culture was observed, suggesting a role for Ahi1 in tissue apical organization as well as other yet to be defined roles [22]. The restriction of Ahi1 to the genomes of organisms that build a canonical nine triplet microtubule centriole might be indicative of a role for Ahi1 in building or maintaining a cilium from a triplet centriole and signaling pathways specific to this structure. Given the cilium phenotype described here and previously [57], it will be interesting to examine the ultrastructural appearance of centrioles in Ahi1-mutant cells to see if centriole maturation or acquisition of the accessory structures that characterize a mature basal body are compromised. Alternatively, there could be a problem with formation of the ciliary vesicle in Ahi1-mutant cells or Ahi1 could play a role in the construction of the axoneme itself.

In work dating back nearly 60 years, Sorokin described the formation of cilia, which required three stages [66]. This included an initial phase where a “primary vesicle” appears at the end of a centriole. Following attachment of this vesicle, a “ciliary bud” is formed. Secondly, the ciliary shaft develops and elongates; a postulated requirement for this is the addition of ciliary vesicles which fuse to the primary vesicle. Finally, there is “emergence” of the cilium to the cell surface and its protrusion out of the cell [66]. More recently, the depression in the plasma membrane in which the cilium sits, known as the ciliary pocket, has been elegantly described [67] in both renal and retinal epithelial cells. This structure may be important in regulating vesicular traffic and is intimately related to the actin cytoskeleton. Given that Ahi1 interacts with Rab8a, a small GTPase, and this interaction is required for ciliogenesis and vesicular trafficking [57], it seems that Ahi1 function might be critical to the ciliary pocket and the conserved protein domain structure of Ahi1 is consistent with a role as a scaffolding protein [9] at this location.

However, it is possible that Ahi1 might have multiple roles in cells. Our data in IMCD3 cells suggest that Ahi1 is required for formation of adherens junctions and subsequent apical–basal polarity. In IMCD3 cells, the Ahi1 siRNA treatment dramatically decreased the staining of the cell–cell contact areas as demonstrated by anti-E-cadherin staining. In contrast, the tight junctions, by anti-ZO-1 staining showed no abnormality. Formation of apical–basal polarity and cell–cell junctions are both dependent on an intact actin cytoskeleton, and the disruption of adherens junctions is consistent with the idea that Ahi1 also plays a role in establishing or responding to cell polarity via the formation and organization of cytoskeletal networks [57]. Recent data suggests that Ahi1 may act as a potential bridging molecule between other nephrocystin proteins, given AHI1 interacts with NPHP2, MKS1, MKS6, and Tectonic-1 [22]. There is also increasing evidence for non-ciliary roles for proteins previously thought of as having a purely ciliary function [68]. Examination of the publicly available EST databases suggests expression of a number of different Ahi1 isoforms and it is possible that different isoforms fulfill different cellular roles. Alternatively, Ahi1 might play different roles in different cellular locations, as has been demonstrated for the “ciliary” cation channel polycystin-2 [69]. While the localization described for Ahi1 is to the mature centriole [57], many proteins undergo highly dynamic changes in their localization that are not observed by immunofluorescence. Live-cell imaging of tagged Ahi1 will be required to test if this is the case.

We have demonstrated prominent early embryonic expression of ahi1 in zebrafish where ahi1 is expressed in KV, a ciliated organizing structure, as well as in multiple tissues, including the developing brain and eye. We have also established that ahi1 is present in the developing pronephros. Previous expression studies in murine tissues showed Ahi1 expression in neurons of the developing hindbrain, midbrain and forebrain, pituitary, testis, and kidney [29]. Ahi1 expression in murine brain has been reported during early embryonic life (day 10.5) and persisted into adult life. A role for Ahi1 in neurons and neuronal function is implicated given the fact that all the major anatomical defects in JBTS involve brain structures [6, 9, 23, 29, 70, 71]. In contrast, in a recently described murine model (Ahi1 ^−/−), the brain morphology was grossly preserved [26] and neuronal-specific Nestin-Cre conditional Ahi1 knockout mice survived to weaning, suggesting a role for Ahi1 outside of the CNS, which affects survival [26]. Murine Ahi1 interacts with Huntingtin-associated protein 1 (Hap1) and consistent with a role for Ahi1 in cerebellar development, Hap1-knockout mice showed reduced Ahi1 levels, abnormal cerebellar development and abnormal axonal decussation [72]. There is evidence that AHI1 mutations and polymorphisms may modify the brain phenotype of JBTS patients [28].

Consistent with a role of ahi1 in the epithelium of the pronephric ducts, we have shown that using zebrafish to model JSRD, ahi1 morphants may develop renal cystic disease, pronephros dilatation and cloacal abnormalities. We have previously noted this pattern of cloacal disruption in nphp6 (alias cep290) morphants [12]. Given the zebrafish pronephros is segmented in a similar manner to the human nephron [73], a dilatation of the most distal pronephric segments is consistent with cortico-medullary cyst formation as seen in patients with nephronophthisis [74]. The absence of cilia within the pronephros of some ahi1 morphants is in keeping with the complete loss of cilia within KV of ahi1 morphants and the significant reduction of cilia in siRNA-treated IMCD3 cells. In addition, the irregular expression pattern of cldnb, suggests a disruption of the distal pronephros epithelium in ahi1 morphants. This, together with aberrant appearance of adherens junctions in siRNA-treated IMCD3 point towards Ahi1 mediating alterations in epithelial morphogenesis, cell adhesion and polarity.

The most common phenotype we observed in ahi1 morphants was a defect in body axis leading to curvature (“curly tail”), in association with defects of left–right asymmetry, edema and pronephric cysts. Typically, such phenotypes are seen in zebrafish with loss of intraflagellar transport proteins, which are required for cilia assembly [75]. Other models of ciliopathies in zebrafish also show “curly tails” alongside defects in cilia length or motility, left–right defects and pronephric cysts [76–79]. The ahi1 morphant “ciliopathy” phenotype we describe is consistent with the demonstrated defect in cilia formation at KV and in the pronephros of ahi1 morphants. Other similar zebrafish models of JBTS demonstrating combinations of hydrocephalus, body axis curvature and pronephric cysts have been reported including nphp6 [12], arl13b [76], cc2d2a [18] and cxorf5 [80].

We show that a likely consequence of the lack of cilia and abnormal KV in ahi1 morphants is the dramatic change in cardiac looping, with a significant reversal of the normal D-looping pattern (Fig. 5). Situs inversus has been described in a small number of patients with mutations in ciliopathy genes, including NPHP2 [81], NPHP3 [82] and NPHP6 [13]. Although we are not aware of patients with AHI1 mutations and situs inversus, other cardiac defects are often noted, such as atrial septal defects [23].

Our data confirms ahi1 expression during retinal development during zebrafish embryogenesis, with a disruption of ahi1 in the zebrafish leading to loss of retinal lamination. Consistent with the severe retinal phenotype we describe in ahi1 morphants, elegant studies in murine models have confirmed that Ahi1 is a modifier of retinal phenotype [26]. The Ahi1 ^−/− mice demonstrated a rapid loss of outer nuclear layers by 1 month of age. Electron microscopy revealed that photoreceptor ciliary axonemes were intact, with a preserved 9 + 0 microtubular doublet configuration [26]. This suggests that Ahi1 is not involved in the structural development of the photoreceptor axoneme, but rather may have functional consequences and affects the photoreceptor cell maintenance and survival.

In conclusion, we have demonstrated in early zebrafish development an expression pattern for ahi1 which includes KV, neuronal and retinal structures, and the developing pronephric ducts. ahi1 morphant zebrafish recapitulate the human JSRD phenotype with structural brain defects, retinal dystrophy and renal cysts. We provide novel evidence of disruption of the ciliated organizing center KV and consequently altered left–right patterning in ZF embryos. In both the zebrafish pronephros and renal epithelial cells we confirm a requirement for Ahi1 in ciliogenesis and cell–cell junction morphology suggesting roles for Ahi1 in the development and maintenance of these structures.

Electronic supplementary material

Below is the link to the electronic supplementary material.