A Complex of BBS1 and NPHP7 Is Required for Cilia Motility in Zebrafish

Bardet-Biedl syndrome (BBS) and nephronophthisis (NPH) are hereditary autosomal recessive disorders, encoded by two families of diverse genes. BBS and NPH display several overlapping phenotypes including cystic kidney disease, retinitis pigmentosa, liver fibrosis, situs inversus and cerebellar defects. Since most of the BBS and NPH proteins localize to cilia and/or their appendages, BBS and NPH are considered ciliopathies. In this study, we characterized the function of the transcription factor Nphp7 in zebrafish, and addressed the molecular connection between BBS and NPH. The knockdown of zebrafish bbs1 and nphp7.2 caused similar phenotypic changes including convergent extension defects, curvature of the body axis, hydrocephalus, abnormal heart looping and cystic pronephros, all consistent with an altered ciliary function. Immunoprecipitation assays revealed a physical interaction between BBS1 and NPHP7, and the simultaneous knockdown of zbbs1 and znphp7.2 enhanced the cystic pronephros phenotype synergistically, suggesting a genetic interaction between zbbs1 and znphp7.2 in vivo. Deletion of zBbs1 or zNphp7.2 did not compromise cilia formation, but disrupted cilia motility. Although NPHP7 has been shown to act as transcriptional repressor, our studies suggest a crosstalk between BBS1 and NPHP7 in regulating normal function of the cilium.


Introduction
Autosomal recessive cystic kidneys are typically part of complex syndromes involving multiple organs. Since most gene products mutated in these syndromes localize to the cilium, an abnormal function of the cilium has been implicated in their pathogenesis. Cilia are specialized microtubule-based organelles attached to most vertebrate cells. The primary cilium is thought to function as a mechano-and/or chemosensor, which detects signals from extracellular environment and transmits them to the inside of the cell. Mutations of numerous ciliary proteins lead to structurally or functionally abnormal cilia [1]. The resulting defects are closely related to developmental and degenerative disorders, collectively termed 'ciliopathies' [2]. In kidney epithelial cells, the primary cilium is involved in important signalling cascades such as Hedgehog, Wnt, planar cell polarity, and calcium-dependant signalling pathways [3]. Cilia-related cystic kidney diseases include autosomal dominant as well as recessive polycystic kidney disease, including the Bardet-Biedl syndrome (BBS) and nephronophthisis (NPH) [4].
Bardet-Biedl syndrome (BBS) is a genetically heterogeneous autosomal recessive disorder (BBS 1-17) [5], symptoms of which include obesity, retinal degeneration, anosmia, post-axial polydactyly, cognitive impairment, hypogenitalism, renal and cardiovas-cular anomalies, diabetes mellitus and hypertension [6,7]. The BBS proteins have been reported to form a stable multiprotein complex called BBSome (BBS1, 2, 4, 5, 7, 8 and 9), and to play a role in ciliogenesis and cilia maintenance [8,9,10]. NPH is an autosomal recessive cystic kidney disease associated with retinal degeneration, cerebellar hypoplasia, liver fibrosis, situs inversus, and mental retardation [11,12,13]. Mutations in 15 genes have been linked to NPHP so far (OMIM 614845). The proteins encoded by NPH genes are called nephrocystins, and are highly conserved among species. They are involved in various cellular signalling events, employing the non-canonical Wnt, sonic hedgehog and planar cell polarity signalling pathways [11,14]. Recent studies indicate that most of BBS and NPHP proteins localize to the cilia/ basal body complex [15,16,17,18], providing a structural basis for the overlap of the symptoms exhibited by BBS and NPH patients, including renal cysts and retinitis pigmentosa, liver fibrosis, situs inversus and mental problems, and supporting the notion that BBS and NPHP protein family share functional pathways [19,20].
During the last decade zebrafish has emerged as a powerful animal system for studying ciliopathies [21,22]. In the present study, we analyzed the biochemical interaction between BBS1 and NPHP7, and investigated this interaction in zebrafish. BBS1 is one of the seven members of the BBSome, which regulates vesicular trafficking of proteins to the ciliary membrane [10,23]. Mutations of BBS1 in humans have been reported as one of the most frequent causes of BBS, implying a significant function of BBS1 [6,24,25,26]. BBS1 is highly expressed in the kidney [24]. NPHP7/GLIS2 (Gli-similar 2) is a member of the Gli-related Krüppel-like zinc-finger (ZF) transcription factor subfamily [27,28]. Glis2 is essential for maintaining renal functions by regulating genes which are involved in epithelial-to-mesenchymal transition, fibrosis and apoptosis [29,30]. In adult mouse kidneys, Glis2 was detected in epithelial cells of the renal tubule and Bowman's capsule [29]. Glis2 lacZ/lacZ mutant mice developed renal atrophy, fibrosis and glomerular cysts, thus resembling some of the key features of NPH [29,30]. Both BBS1 and GLIS2 are present in cilia [23,29,31].
Morpholino oligonucleotide (MO)-mediated knockdown of zbbs1 and znphp7.2 in zebrafish embryos caused phenotypic changes characteristic for ''ciliopathies''. Detailed analysis of cilia revealed that despite normal morphology, the cilia in the pronephric tubule and the nasal pit of zBbs1-and zNphp7.2depleted embryos showed aberrant motility, suggesting that the observed phenotypes are due to the disruption of the normal ciliary beating pattern.

BBS1 and NPHP7 interact with each other
To investigate the interconnection between BBS and NPH protein families, we tested the interaction between NPHP7 and the BBS family members BBS1-12, using tagged versions of human proteins overexpressed in human embryonic kidney (HEK) 293T cells. We took interest in NPHP7 for the reason of being a transcription factor not having been very well characterized so far. Precipitation of BBS1 immobilized NPHP7 most extensively, suggesting that this interaction may be relevant in vivo. Since BBS1 has been characterized as a component of the BBSome, involved in ciliary transport processes, we decided to clarify the role of the BBS1/NPHP7 complex. Fig. 1A shows the interaction between human BBS1 and NPHP7. While BBS1 lacks apparent conserved functional domains, NPHP7 is a transcription factor with 5 zinc finger (ZF) domains responsible for DNA binding and transcriptional regulation [32]. To identify the domain of NPHP7 responsible for the interaction with BBS1, we generated three truncation constructs, Flag-NPHP7 1-156, Flag-NPHP7 141-359 and Flag-NPHP7 346-525 (Fig. 1B). We found that the Nterminus (Flag-NPHP7 1-156) of NPHP7 interacts with V5-BBS1 (Fig. 1A). Unfortunately, the lack of suitable antibodies for NPHP7 prevented us from confirming this interaction in vitro. Therefore, we turned to zebrafish, a genetically tractable animal model, to search for a potential genetic interaction between these two gene products.

zbbs1 and znphp7 are expressed in various tissues in the zebrafish
The zebrafish homologue of human BBS1 has been previously described [33]. Homology blast searches with the human protein as a query identified 2 potential zebrafish Nphp7 homologues. The genes encoding these proteins are located on chromosome 22 and 3, which we named znphp7.1 (Genbank KF054060, Zv9: ENSDARG00000078388) and znphp7.2 (Genbank KF054061, Zv9: ENSDARG00000073861), respectively; subsequent genome annotations (genome assembly ZV9) confirmed our findings. Protein alignments showed that zNphp7.1 sequence had 43.9% identity and 50.8% similarity to the human NPHP7/GLIS2; zNphp7.2 was 51.4% identical and 60.2% similar to the human gene product (Fig. 2A). As for the Gli family, the domain with the highest identity was the ZF domain: zNphp7.1 and zNphp7.2 were 89.3% and 91.3% identical with the human ZF domain of human GLIS2, respectively. Previous studies showed ubiquitous expression of BBS1 in human organs including fetal tissue, testis, retina and adipose tissue with the highest expression in the kidney [24]. Glis2 mRNA was also most abundantly expressed in mouse kidney with low levels in heart, lung, placenta, prostate, colon, and brain [27,28]. In zebrafish whole mount in situ hybridization with various antisense RNA probes of zbbs1 and znphp7.1 and znphp7.2 did not show consistently reproducible results. Semi-quantitative RT-PCR revealed that znphp7.1 and znphp7.2 are expressed maternally, at 6 hpf and 24 hpf whereas zbbs1 is expressed at 24 hpf. A second maternal splice variant of znphp7.2 (Transcript 2) was observed and confirmed by sequencing ( Fig. 2B and Fig. S1). The sequence alignment showed that 1 amino acid is substituted (S99R) and 18 amino acids (aa 101-118) are not coded for in transcript 2 of znphp7.2 (Fig. S1). In addition, semi-quantitative RT-PCR from various organs of adult zebrafish showed that zbbs1, znphp7.1 and znphp7.2 are expressed in kidney, eye and testis. However, znphp7.1 and znphp7.2 are additionally expressed in the heart, gut and muscle (Fig. 2C).

Depletion of zBbs1 and zNphp7 in zebrafish embryos causes ciliopathic phenotypes including pronephric cyst formation
To address the role of zBbs1 and zNphp7 in zebrafish kidney development, we performed knockdown experiments by injection of MO, targeting zbbs1 translational start codon (AUG MO) and exon 2 splice donor site (SP MO) of zbbs1 to deplete zBbs1 [33] (Fig. S2A). To test the efficacy of zbbs1 AUG MO, we performed an in vitro transcription and translation (TNT) assay. This showed that zbbs1 AUG MO efficiently interfered with zBbs1 protein expression (Fig. S2B). Both MOs showed similar phenotypes, which implies that the observed phenotypes were knockdownspecific. However, the AUG MO caused less general developmental side effects, and was therefore used in all subsequent experiments. To examine the effect of depletion of zBbs1 on kidney development in vivo, we used the transgenic fish line Tg(wt1b:EGFP) [34]. At 55 hpf, zbbs1 morphants displayed cysts with distension of the neck segment of the pronephric tubule ( Fig. 3E and G) indicating that zBbs1 is required for normal zebrafish pronephros development. The overall appearance of the morphant embryos was categorized by the level of dysmorphy ranging from wild-type-like (WT) to increasingly severe dysmorphy (D1 to D3) ( Fig. 3G and Fig. S3). The grade of dysmorphy and the number of cystic morphants showed a MO dose dependency (Fig. 3G). In addition, we observed the previously described effects of zBbs1-depletion manifesting in hydrocephalus and curvature of the body axis ( Fig. 3B and Fig. S3), in convergent extension (CE) defects (Fig. 4), and in defects of left-right (L-R) asymmetry, i.e. organ laterality, such as heart looping ( Fig. 5A and B; Fig. S5A and B), and situs of liver and pancreas ( Fig. S5C and D) [35,36,37]. Moreover, zbbs1 deficient embryos showed defects in eye development displaying reduced eye size and impaired layer formation of the retina (Fig. S8) similar to the knockdown of the orthologue of retinitis pigmentosa gene 2 [38,39]. Altogether, the phenotypes detected in zbbs1 morphants were reminiscent of ciliopathy-associated phenotypic changes previously described in zebrafish [33,35].
Depletion of zNphp7.1 by either translational initiation site blocking AUG MO, or by a MO binding to the exon 5 splice donor site did not result in any specific phenotype readily distinguishable from wild-type (data not shown). For knockdown of znphp7.2, 2 different SP MOs, one targeting the exon 3 splice donor site (SP1) and the other targeting the exon 6 splice acceptor site (SP2) of znphp7.2 gene, were used (Fig. S4A). Both MOs caused similar phenotypes, confirming the specificity of the knockdown. However, the SP1 MO targeting exon 3 showed higher efficiency at lower concentrations and was used subsequently. znphp7.2 MO injections in 1-cell embryos caused aberrant splicing of intron 3 in znphp7.2 pre-mRNA leading to 2 splice products both of which cause an early truncation within the first ZF domain eliminating all ZF domains together with the C-terminus of zNphp7.2 ( Fig. S4B-D).
The phenotype of znphp7.2 depletion includes curvature of the body axis and hydrocephalus (Fig. 3), retinal degeneration with disturbed retinal layering and reduced eye size (Fig. S8) and cystic pronephros (Fig. 3C, F and H), which are very similar to the phenotype of zbbs1 morphants pointing to defects in cilia function. Co-injection of in vitro transcribed zbbs1 or znphp7.2 mRNA together with the corresponding MO partially rescued the morphant phenotype (Fig. S6), confirming the specificity of the MO-mediated knockdown. To exclude that pronephric cyst formation was attributable to cloaca malformation, FITC-conjugated fluorescent dextran (70 kD) was injected into the common cardinal vein of control and morphant embryos with pronephric cysts at 96 hpf. The excretion of fluorescent dextran with the urine was observed shortly after the injection in 22 control embryos, 16 zbbs1 and 21 znphp7.2 morphants (Fig. S7). Only, 2 znphp7.2 morphants showed cloaca blockage. Therefore, we conclude that pronephric cyst 2) consist of 446 and 489 amino acids respectively. Amino acid sequence alignment showed that zNphp7.1 shares 43.9% identity and 50.8% similarity with the human NPHP7/GLIS2 (hNPHP7); zNphp7.2 was 51.4% identical and 60.2% similar to the human homologue. The ZF domains of zNphp7.1 and zNphp7.2 were 89.3% and 91.3% identical with those of the human homologue, respectively. (*, completely conserved; ., identical in 2 sequences or belonging to same type of amino acid group in 2 or 3 sequences) (B) Semi-quantitative RT-PCR reveals maternal transcript expression for znphp7.1 and znphp7.2 whereas zbbs1 is not expressed maternally nor at 6 hpf. 2 maternal splice products were identified for znphp7.2 (open arrowhead: Transcript 1; filled arrowhead: Transcript 2). The transcript 2 of znphp7.2 is expressed only maternally. Sequencing of the lower splice product revealed an excision of 18 bp corresponding to amino acid (aa 101-118) (Fig. S1). (C) Semi-quantitative RT-PCR with organ specific cDNA from adult zebrafish indicated that zbbs1 is expressed in kidney, eye and testis. znphp7.1 and znphp7.2 are expressed in other organs including kidney, eye, heart, testis, gut and muscle. doi:10.1371/journal.pone.0072549.g002 The Function of NPHP7 in Zebrafish Kidney PLOS ONE | www.plosone.org formation of zbbs1 and znphp7.2 morphants was not caused by disruption of cloaca development (Fig. S7) [40].
Knockdown of znphp7.2 also caused CE defects judged by a wider body gap angle in relation to the stage of somitogenesis which translates into a shorter body axis and broader somites on in situ hybridization with probes for krx20 and myod1 (Fig. 4).
Cilia-driven fluid flow in the Kupffer's vesicle (KV) triggers symmetry breaking events during zebrafish development [37,41,42,43,44]. Defects in L-R asymmetry were observed in znphp7.2 morphant embryos for early laterality markers like lefty2 ( Fig. S5A and B) and in MO dose dependent manner for heart looping ( Fig. 5A and C) and the situs of liver and pancreas ( Fig.  S5C and D). This shows the involvement of zbbs1 and znphp7.2 in establishing organ laterality most likely via cilia function in KV.

zbbs1 and znphp7.2 show a genetic interaction in cystogenesis
The co-immunoprecipitation suggested a link between BBS1 and NPHP7 (Fig. 1). To confirm this interaction in vivo, we Knockdown of zbbs1 and znphp7.2 influences cilia motility As indicated above, knockdown of either zbbs1 or znphp7.2 led to defects in cardiac looping in zebrafish embryos. One potential cause of this defect could be the defective ''anlage'' of the KV [45]. Indeed, our examinations revealed that depletion of zBbs1 causes defects in KV formation in a substantial number of embryos consistent with previous findings [33], while morphological examination of znphp7.2-depleted zebrafish embryos showed normal formation of KV ( Fig. 7A and B). Further examinations of the ciliogenesis in KV of znphp7.2 knockdown embryos were performed using immuno-staining for acetylated tubulin. This result indicated that development of cilia in the KV appeared shorter compared to control, however, the overall distribution of cilia in KV was unchanged ( Fig. 7C and D). In addition, cilia formation in the pronephric tubule of zBbs1-and zNphp7.2depleted embryos was not disturbed even though the cilia in zBbs1-depleted embryos were longer, and the cilia in zNPHP7.2depleted embryos were shorter at significant level compared to control ( Fig. 7E and F). The ciliary morphology and length in the pronephric tubule were also assessed in the combined knockdown condition of both zbbs1 and znphp7.2 which showed the synergistic enhancement of cystic pronephros phenotype. In this condition, the experimental results showed that single and simultaneous knockdown of zbbs1 and znphp7.2 does not affect cilia length in pronephric tubule ( Fig. 7G and H). This data raised the question, whether the motility of cilia in morphant zebrafish embryos was impaired. Therefore, we assessed cilia motility by live embryo confocal imaging using the Tg(bact:Arl13b-GFP) transgenic line which labels cilia with GFP in various organs [46]. We used this fish line to be able to detect subtle changes in motility within the population of cilia of one tubule. The movies were obtained along the whole pronephric tubule. For better comparison, three regions were defined as anterior, middle, and posterior (Fig. S9A). Regular beating movements of cilia in the lumen of the pronephric tubule in downstream direction, towards the cloaca, were observed in control zebrafish embryos (Movie S1, S2 and S3) as described previously [37]. The movies were recorded from morphant embryos with clear pronephric cysts. Strikingly, the cilia in zbbs1 (0.5 mM)-and znphp7.2 (0.2 mM)-depleted zebrafish embryos exhibited greatly reduced motility and abnormal beating pattern (Movie S4 and S5, S6 for zbbs1 MO; Movie S7 and S8, S9 for znphp7.2 MO). These results suggest that zBbs1 and zNphp7.2 are required for motility of cilia in pronephric tubule of zebrafish.
Furthermore, disturbed cilia movement was detected in the nasal pit of zbbs1 and znphp7.2 morphants at 72 hpf, confirming that zBbs1 and zNphp7.2 play a role in cilia motility. This was monitored by ordinary videomicroscopy with a high-speed camera system (Movie S10, S11 and S12). Immuno-staining of cilia in the nasal pit with acetylated tubulin antibody shows that formation of cilia in the nasal pit is unimpaired in zbbs1 and znphp7.2 morphant embryos (Fig. S9B).
To scrutinize the cause of the impaired motility of cilia, electron microscopy was performed to examine ultrastructure of cilia in the pronephric tubule of zbbs1 and znphp7.2 morphants: ultrastructural defects, e.g. deficiency of dynein arms, could not be detected in pronephric cilia of both morphant embryos (Fig. 7I). Hence, we conclude that the disturbed motility of zbbs1 and znphp7.2 morphant cilia is not necessarily associated with the axonemal ultrastructure of cilia [47].
To examine the interaction between zbbs1 and znphp7.2, cilia beating in the pronephric tubule of morphants with combined knockdown condition was analyzed. Interestingly, the simultaneously zBbs1-and zNphp7.2-depleted embryos displayed severely impaired motility of cilia compared to zBbs1-or zNphp7.2-deficient embryos (Movie S13, S14, S15, S16, S17, S18, S19, S20, S21, S22). In the movies of combined knockdown embryos, the arrows point to the individual cilia showing impaired movement. The movies were recorded from the single knockdown morphant embryos without pronephric cysts, whereas the movies were recorded from the combined injected morphant embryos showing clear pronephric cysts. This result supports the genetic interaction between zbbs1 and znphp7.2 which was shown for pronephric cystogenesis in this study. Therefore, we conclude that the cystic pronephros phenotype in the knockdown of zbbs1 and znphp7.2 is due to the aberrant motility of cilia and disturbed fluid flow in the pronephric tubule [37,48].

Discussion
Multiple ciliopathy-associated genes have been identified that are involved in syndromic cystic kidney disease. Despite phenotypic overlap, the functional relationship and interactions are not well understood yet [4]. The clinical features and the severity of phenotypic changes vary with different identified mutations within one gene, but also with the mutational load and the combination of mutations in two or more genes of ciliopathy-associated gene families [49,50,51,52]. Based on the clinical overlap, we speculated that the similarity in phenotypes is reflected by functional interaction(s) between BBS and NPH gene products. Immunoprecipitation assays identified an interaction between BBS1 and NPHP7. The interacting domain was located to the N-terminus of NPHP7. To examine the in vivo relevance of this interaction, we identified the zebrafish homologues of NPHP7, zNphp7.1 and zNphp7.2, likely the result of zebrafish genome duplication events [53]. The protein sequence identity between zebrafish and human is high within the zinc finger region as expected for the GLI/GLIS protein families (human sequence aa168-317, http://smart.emblheidelberg.de) [27,28]. Since the knockdown of znphp7.1 was phenotypically identical to control embryos, the biological function of znphp7.1 remains unclear. The phenotypic changes (curved body axis, hydrocephalus, retinal degeneration and pronephric cyst formation) induced by knockdown of znphp7.2 were compatible with depletion of other ciliopathy gene products [37]. We excluded cyst formation on the grounds of cloaca malformation by fluorescent dye injection und urinary excretion experiments. There is no detailed description of the clinical phenotype available in the Canadian Oji-Cree kindred with the first known human NPHP7/GLIS2 mutation (homozygous splicesite mutation, c.755+1G.T) and recently found second mutation in the Turkish patient (c.523T.C, p.C175R) [29,54]. In zebrafish, knockdown of znphp7.2 induced defects in left-right asymmetry of organ situs consistent with the concept that mutations of NPHP7 belong to the growing family of ciliopathies. Kupffer's vesicle revealed a normal size and shape in znphp7 MO-injected zebrafish embryos, whereas defects in formation of Kupffer's vesicles were described in zbbs1 MO-injected zebrafish embryos [33], suggesting that the laterality defects in znphp7 morphant embryos were due to a ciliary or post-ciliary event. Both zBbs1-and zNphp7.2-depleted embryos displayed defects in CE movements, previously described for the knockdown of zbbs1 [36]. Knockdown of zbbs1 expands the expression of axin2, a downstream target of canonical Wnt signalling [36], and Glis2 negatively modulates Wnt/beta-catenin signalling [55]. Thus, both zBbs1 and zNphp7 appear to facilitate non-canonical Wnt while antagonizing canonical Wnt signalling.
The synergistic effect of znphp7.2 and zbbs1 MOs on cyst formation further supports the notion that BBS1 and NPHP7 (GLIS2) act within the same or overlapping signalling pathways.
The phenotypic changes observed in either znphp7or zbbs1depleted zebrafish embryos clearly suggested a defect of motile cilia or cilia connected signalling pathways. We first excluded defects in ciliary axoneme formation: Although the ciliary length, measured in the Kupffer's vesicle or pronephric tubule, was significantly different in znphp7.2-and zbbs1-depleted zebrafish embryos compared to control, there was no change in ciliary length observed in the set of experiments with combined MO injection. Furthermore, the general distribution and electron microscopic ultrastructure revealed no difference between morphant and control. However, our results of video-microscopy and confocal microscopy revealed an astonishingly impaired ciliary motility in pronephric tubule and nasal pit of znphp7.2 and zbbs1 morphants, suggesting that the ciliopathy-associated phenotypes caused by the knockdown of either zbbs1 or znphp7.2 are due to the disruption of ciliary motility [37].
In primary cilia dyskinesia (PCD) the impaired ciliary motility often goes along with changes in ultrastructure, e.g. deficiency of dynein arms, central pair complex or radial spokes, but there have been reports of unchanged ciliary ultrastructure with DNAH11 [56]. With DNAH11 one possible explanation is that the mutant DNAH11 protein retains its N-terminal domain, which is required for the correct incorporation of outer dynein arms [57]. Another key mechanism for regulation of ciliary motility is phosphorylation, i.e. phosphorylation/dephosphorylation of the dynein arms, through kinases and phosphatases that are anchored immediately adjacent to their axonemal substrates [58]. These kinases and phosphatases could be candidates regulated by events downstream of znphp7.2 and zbbs1 without causing ultrastructural defects on electron microscopic level.
Previous reports localized BBS1 and GLIS2 to cilia [23,29,31]; this is particularly remarkable for the transcription factor NPHP7/ GLIS2 preferably found in the cell nucleus. Recently, the BBSome, structurally related to coat complexes, has been reported to mediate the transport of cargo to the ciliary compartment [23]. BBS1 interacts with other proteins such as BBS3, LepRb, Rabin8, and has been implicated in protein and membrane trafficking to cilia [10,59,60]. Thus, BBS1 might be involved in the transport of GLIS2 to the cilium, where posttranslational modifications of GLIS2 could take place similar to the mandatory processing of GLI transcription factors in the cilium [61,62]. Cleavage of GLIS2 has been described to be induced for example by p120 catenin [63]. Alternatively BBS1 could also facilitate the transport of GLIS2 from the cytosol to the nucleus. Curiously, other NPHP gene products such as NPHP2/Inversin [64] and NPHP15/ CEP164 [65] have been shown to localize to both the cilium as well as the nucleus.
In conclusion, we demonstrate that depletion of Nphp7/Glis2 leads to a NPH-typical phenotype in zebrafish including defects in left-right organ asymmetry, further validating Glis2 as a legitimate member of the NPH gene family in spite of the limited description of human mutations. We also show that Bbs1 and Nphp7 are both required for normal ciliary motility, a pathway responsible for normal body asymmetry. Although the presence of BBS1/NPHP7 in the cilium suggests a cilia-specific function, this module could also regulate the expression of other gene products required for ciliary motility.
Westerfield M: The Zebrafish Book. A Guide for the Laboratory Use of Zebrafish (Danio rerio), University of Oregon Press 1995.
The transgenic fish line Tg(wt1b:eGFP) was a kind gift of Christoph Englert [34]. The Tg(bact:Arl13b-GFP) transgenic line we received thankworthy from Brian Ciruna's lab [46]. . In order to decrease the nonspecific effects of the reagents to zebrafish embryos, all MOs were co-injected with p53 MO (0.1 mM). p53 MO (0.1 mM) was also injected into all wild-type embryos as controls [66]. For combined knockdown experiment with 2 different MOs simultaneously, all MO injection samples were balanced with Cont MO (standard control MO, Gene Tools LLC.) MO to have 0.3 mM in order to inject the same amount of MOs in total. The clone of zebrafish bbs1 homologue was obtained from ImaGenes GmbH. The full-length zbbs1 was isolated by PCR and cloned into the vector pCS2+ for in vitro transcription/ translation assay. To produce the zbbs1 mRNA for rescue experiment, we modified the sequence of zbbs1 with sequencedirected mutagenesis and generate a construct with 3 silent mutations in zbbs1 AUG MO binding site (original sequence from start codon, atg tca tct gtt; modified sequence, atg tct tca gta). The full-length of znphp7.2 obtained by PCR using zebrafish embryonic cDNA was also inserted into pCS2+ (znphp7.1 (Genbank KF054060, Zv9: ENSDARG00000078388) and znphp7.2 (Genbank KF054061, Zv9: ENSDARG00000073861)). Phenotype rescue experiments were performed with in vitro transcribed 20 pg of full-length MO-resistant zbbs1 mRNA or 10 pg of wildtype znphp7.2 mRNA (mMessage Machine kit, Ambion) which was co-injected with zbbs1 AUG MO (0.4 mM) or znphp7.2 SP1(0.3 mM) respectively.

Zebrafish embryo manipulations
The body gap angle of embryos at 8-10 somite stage was measured to assess convergent extension movement [36]. KV areas of control and morphant embryos were measured at 8-10 somite stage using imageJ (http://rsb.info.nih.gov/ij/). The average area measured in control embryos was set as 1.00 to determine other measurement of morphants.
All animal studies were approved by the Committee on Research Animal Care, Regierungsprä sidium Freiburg .81/G-11/41).

Cell culture and transfection
For co-immunoprecipitations, HEK 293T cells (received from American Type Culture Collection, ATCC, Manassas, VA) were grown in DMEM with 10% FBS. The 70-90% confluent cells were split at a ratio of 1:5 and transfected the next day with plasmid DNA applying the calcium phosphate method as described previously [67]. The transfection was stopped after 6-8 h by replacing the media.

Co-immunoprecipitation
To construct Flag-tagged BBS1, NPHP7 and truncations of NPHP7/GLIS2, the cDNAs of full-length BBS1 and NPHP7 were amplified from the EST clone by PCR and cloned between MluI and NotI site of the Flag-pcDNA6. HEK (Human embryonic kidney) 293T cells were transiently transfected with plasmids by the calcium phosphate method, and lysed after incubation for 24 h in immunoprecipitation buffer containing 20 mM Tris-HCl (pH 7.5), 1% Triton-X 100, 25 mM NaF, 12.5 mM Na4P2O7, 0.1 mM EDTA, 50 mM NaCl, 2 mM Na3VO4, and protease inhibitors. After lysates were cleared by centrifugation at 15 000 g for 30 min at 4uC, the supernatants were incubated with 30 ml of anti-Flag-M2 beads for 3 h [68]. The beads were washed extensively with immunoprecipitation buffer, and the retained proteins were analyzed with western blot analysis. Antibodies used in this western blot analysis, included FLAGH monoclonal antibody produced in rabbit (Sigma, Hamburg, Germany), rabbit antibody to V5 (Serotec).

Eye size measurement and HE Staining for eye sections
Eye size measurement and HE Staining for eye sections were performed with 80 hpf and 96 hpf embryos, respectively. Image J (http://rsb.info.nih.gov/ij/) was used for measurement of the area of the eye. The embryos at 96 hpf that were sectioned for Hematoxylin and Eosin staining were fixed in BT-fix (4% PFA, 0.1 M Na 2 HPO 4 buffer (pH 7.3), 3% sucrose, 0.12 mM Ca 2 Cl) at 4uC overnight. After being washed in PBS and taken through an ethanol dehydration series, the embryos were embedded in JB-4 resin (Polyscience Inc.) and sectioned at 3-5 mm using a microtome (Leica). Slides were stained for Hematoxylin and Eosin.

In situ hybridisation and immunohistochemistry
In situ Hybridisation (ISH) was carried out as described previously [69]. Dig-labelled antisense RNA was transcribed with T3 or T7 RNA polymerase (Roche). The myl7 (myosin, light polypeptide 7) antisense probes were generated by linearizing template plasmids with NotI and transcription with T7 RNA polymerase. krx20 (krox20) antisense probe was produced by linearizing template plasmids (pBS KS vector) with PstI and transcription with T3 RNA polymerase. myod1 (myogenic differenti-ation 1) in pCRII-Topo was linearized by BamHI and transcribed by T7. lft2 (lefty2) in pCS2+ was linearized by ClaI and transcribed by T7. ins (preproinsulin) and foxa3 (forkhead box A3) probes in pBS SK vector were linearized with XbaI and XhoI respectively and transcribed with T7 and T3 RNA polymerase respectively.
For immunohistochemistry, primary mouse monoclonal antibodies, anti-acetylated tubulin (Sigma) were used in a 1:500 dilution. Secondary anti-mouse antibodies labeled with Cy3 (Jackson ImmunoResearch and Molecular Probes) were used. Embryos were fixed in 1% DMSO and 4% PFA in PBS and stained with primary and secondary antibodies by conventional whole-mount antibody staining method. The length of stained cilia was measured with imageJ.

Transmission electron microscopy
Zebrafish embryos at 55 hpf were fixed with PBS containing 1% PFA/2% glutaraldehyde for 1 h at RT and post-fixed with 1% osmium tetroxide (Polyscience) for 30 min at RT. The embryos were dehydrated in ethanol (50% and 60% for 10 min each) and transferred into 1% uranyl acetate (Polyscience) in 70% ethanol overnight at 4uC.
The embryos were subjected to further dehydration through a series of 80%, 90%, 98% and 100% ethanol. After washing the embryos with propylene oxide, they were embedded in Durcupan (Fluka). Ultra-thin sectioning was performed with a Leica EM UC6. Serial sagittal sections (60 nm) were conducted from rostral to caudal (from the beginning of the yolk extensions to copper-grids). Sections were observed and imaged in an electron microscope (LEO 906E; Carl Zeiss). All images were exported as TIFF files.
Confocal microscopy with Tg(bact:Arl13b-GFP) transgenic line All movies were taken with LSM-I-Live Duo ZEISS LSM 510 DUO using ZEN 2010 software (Carl Zeiss, Inc.) and laser diode 489. Image acquisition for the moving cilia was 30 frames per second and the recorded movies are replayed in real-time. The movement of cilia in pronephric tubule was imaged with 55 hpf embryos of the Tg(bact:Arl13b-GFP) transgenic line [43]. The embryos were incubated with 40 mM BDM (2,3-butanedione monoxime, Sigma-Aldrich) for 5 min to eliminate the interfering signal of blood flow, and then the mounted embryos in 1% low melting temperature agarose were covered with egg water (20 mM BDM) for imaging. The movies were taken with LD LCI Plan-Apochromat 256/0.8 (zoom: 2.0).

High-speed videomicroscopy
The 55 hpf phenylthiourea (PTU)-treated embryos were put in E3 egg water containing tricaine. The imaging of beating cilia in the nasal pit was conducted on a Zeiss Axioplan microscope (Carl Zeiss, Jena, Germany) using a 636/0.55 water immersion lens installed with a high-speed Photron FastCAM-PCI 500 videocamera (Photron LTD) [37]. Image acquisition was 250 frames per second by Photron FastCAM version 1.2.0.7 (Photron LTD). The movies are replayed in 30 frames per second.

Protein blotting and in vitro translation
1 mg of plasmids DNA (wild-type full-length zbbs1 in pCS2+) were mixed with or without MOs and in vitro transcription/ translation reactions were performed in the presence of biotinylated lysine according to manufacturer's manual (TNTH Quick Coupled Transcription/Translation Systems, Transcend TM Non-Radioactive Translation Detection Systems, Promega). 1 ml of the reaction was diluted in 14 ml of 16 protein sample loading buffer. The reaction mixture was analyzed by SDS-PAGE and protein blotting using HRP-conjugated streptavidin.

Statistical analyses
Statistical analyses, Mann-Whitney rank sum test and Student's t-test, were performed with SigmaStatH (Systat Software Inc.). Student's t-test was used for the data of ciliary length, eye size and KV area measurement. In the case of the comparison of body gap angle, t-test could not be applied. Therefore, the differences in the variance between control-and morphant-embryo body gap angle were analyzed using the Mann-Whitney Rank Sum Test. The probability value of P,0.05 was considered statistically significant (***, P#0.001; **, P#0.01; *, P#0.05). Movie S20 The anterior part of pronephric tubule in a combined morphant embryo. The pronephric cilia of morphant embryos with combined knockdown (zbbs1 0.1 mM+znphp7.2 0.2 mM) showed more severe defective movements compared to zbbs1 or znphp7.2 single knockdown morphants (Movie S16, S17, S18, S19). The movies were recorded from the combined injected morphant embryos showing clear pronephric cysts. (AVI) Movie S21 The posterior part of pronephric tubule in a combined morphant embryo. The pronephric cilia of morphant embryos with combined knockdown (zbbs1 0.1 mM+znphp7.2 0.2 mM) showed more severe defective movements compared to zbbs1 or znphp7.2 single knockdown morphants (Movie S16, S17, S18, S19). The movies were recorded from the combined injected morphant embryos showing clear pronephric cysts. The arrows point to the individual cilia showing impaired movement. (AVI) Movie S22 The cloaca area of a combined morphant embryo. The pronephric cilia of morphant embryos with combined knockdown (zbbs1 0.1 mM+znphp7.2 0.2 mM) showed more severe defective movements compared to zbbs1 or znphp7.2 single knockdown morphants (Movie S16, S17, S18, S19). The movies were recorded from the combined injected morphant embryos showing clear pronephric cysts. (AVI)