Cilia are microtubule-based hair-like organelles that consist of two basic parts: the basal body and the axoneme (Satir and Christensen,2007). The basal body is derived from the mother centriole and serves to anchor the cilium to the plasma membrane, while the axoneme is the filamentous extension of the basal body that protrudes out of the cell surface into the extracellular environment. In vertebrates, some tissues have specialized motile cilia which are able to beat rhythmically due to the presence of dynein motor proteins in their axonemes (Satir and Christensen,2007). The vast majority of vertebrate cilia, however, are immotile (lack axonemal dynein) and are called primary cilia (Satir and Christensen,2007). They occur almost ubiquitously, are assembled by cells in the interphase stage, and disassembled at the beginning of the division cycle. In some differentiated cells, like photoreceptors and olfactory neurons, primary cilia have evolved specialized functions in sensory perception (Anholt,1993; Insinna and Besharse,2008). While the primary cilium on most other cell types have so long been regarded as a vestigial or atavistic structure, several studies have now shown that the organelle is in fact functionally active, and plays fundamental roles in the transduction of developmental signals like Hedgehog (Hh) and Wnt (Eggenschwiler and Anderson,2007).
The link between primary cilia and Hh signaling was first appreciated through the analysis of mice mutant for components of the intraflagellar transport (IFT) machinery (Huangfu et al.,2003). IFT is a kinesin and dynein motor protein-based intracellular cargo delivery system that is required to build cilia and flagella (Scholey,2003). Disruption of IFT results in the malformation or absence of primary cilia, and IFT mutant mice exhibit altered Hh signaling levels, such that cell types induced by high levels of Hh activity are lost, whereas ectopic signaling occurs in cells that normally do not respond to the Hh signal (Huangfu et al.,2003; Huangfu and Anderson,2005). Genetic and biochemical studies have now provided evidence that primary cilia are required for processing of the Gli transcription factors that regulate Hh target gene expression (Huangfu and Anderson,2005; Haycraft et al.,2005; May et al.,2005; Liu et al.,2005; Caspary et al.,2007). Loss of primary cilia induce an imbalance of the effective nuclear concentrations of Gli activator and repressor forms, that lead to the ambivalent loss- and gain-of-function effects on Hh pathway activation. While the mechanism by which the primary cilium regulates Gli processing is totally unclear at the moment, several reports have documented the localization of the Gli family members as well as the transmembrane proteins Patched (Ptc) and Smoothened (Smo; Ptc is the receptor for Hh ligands, while Smo transduces the signal intracellularly) to primary cilia of mammalian cells in a Hh-dependent manner (Haycraft et al.,2005; Corbit et al.,2005; Rohatgi et al.,2007,2009; Kovacs et al.,2008; Wang et al.,2009).
We and others have previously reported that zebrafish embryos mutant for the iguana (igu) gene have alterations in Hh signaling, and phenotypically exhibit a combination of both loss- and gain-of-function effects (Wolff et al.,2004; Sekimizu et al.,2004). For example, cells that are specified by maximal levels of Hh signaling, like the lateral floor plate cells of the ventral neural tube and muscle pioneer cells within the somites, are completely lost or reduced in igu homozygous embryos. By contrast, other cells, such as a population of muscle fibers that form in response to submaximal levels of Hh pathway activation are produced in over abundance. Moreover, extensive genetic epistasis analysis has shown that Igu functions downstream of Smo but upstream of the Gli proteins, and that the Gli activator and repressor ratio is disrupted in the absence of Igu activity (Wolff et al.,2004; Sekimizu et al.,2004). All of these changes in Hh signaling in igu mutants are highly reminiscent of Hh signaling defects manifest in mouse embryos with abnormalities in their primary cilia. The igu gene is widely transcribed during embryogenesis, and encodes a protein with a single zinc finger and coiled-coil domains—the zebrafish homolog of DZIP1 (DAZ interacting protein 1), a mammalian protein of yet unknown biological function that can bind to the germ cell factor DAZ (Deleted in Azoospermia; Moore et al.,2003,2004). Because the mechanism by which Igu/DZIP1 regulates Gli activity is not clear, the remarkable similarity between the altered levels of Hh signaling evident in igu mutant zebrafish and mice with defects in primary cilia led us to investigate whether Igu is required for the formation and/or function of primary cilia.
We used antibodies that recognize acetylated tubulin, a component of ciliary axonemes, to monitor the status of primary cilia in igu mutant zebrafish embryos. In an earlier study, we have reported that like in mammals, most cells of zebrafish embryos elaborate short primary cilia that typically measure 1–1.5 μm in length (Yu et al.,2008). Confocal scans of the myotome of wild-type zebrafish embryos at 24 hours postfertilization (hpf) showed numerous primary cilia decorating the surfaces of muscle cells (Fig. 1A). Contrary to the wild-type situation, no primary cilia could be detected in the myotome of igu mutant embryos at the same stage of development (Fig. 1B). We also examined other tissues in igu embryos for the presence of primary cilia, to ascertain whether the block in primary cilia formation was a global effect. Like in the myotome, primary cilia were absent from all other tissues examined, thus, pointing to a fundamental requirement of Igu activity for ciliogenesis (Fig. 1C,D, and data not shown). This phenotype was completely penetrant, and 100% of the homozygous igu mutant embryos at 24 hpf (identified by their curled body axis and reduction in the floor plate; Brand et al.,1996) exhibited a complete absence of primary cilia throughout their bodies (for a quantitative analysis of this and other phenotypes see Supp. Table S1, which is available online). To determine whether the loss of primary cilia was a developmental stage-specific effect, we stained embryos derived from matings of igu heterozygotes at the late gastrula stage (10 hpf). At this stage, wild-type and mutant siblings are morphologically indistinguishable, but the mutants can be unambiguously recognized by the aberrant expression of patched1 (ptc1), a direct target of Hh signaling. We found that all of the embryos with altered pattern of ptc1 transcription that were sampled showed a severe lack of primary cilia, whereas large numbers of primary cilia could be readily detected in the wild-type sibling embryos (Fig. 1E,F). To confirm that the ciliary defect in the igu mutants indeed arises from disruption of the igu gene, we used two strategies. We have previously shown that injection of morpholinos, which inhibit proper splicing of the igu transcript, can very closely phenocopy the igu mutant phenotype (Wolff et al.,2004). Using one of these morpholinos (splice morpholino 1), we found that primary cilia were strongly reduced in all of the injected embryos that we examined (Fig. 1G). In a converse experiment, we injected in vitro synthesized wild-type igu mRNA into eggs derived from igu heterozygote fish. Staining with anti-acetylated tubulin antibodies revealed that primary cilia were restored to a substantial degree in the mutant embryos (Fig. 1H).
The assembly of the ciliary axoneme occurs by IFT from the basal body, a derivative of the mother centriole. The basal body first migrates and docks with the apical membrane, which is then followed by the IFT-dependent elaboration of the axonemal filament. We next examined the basal bodies with antibodies that label γ-tubulin, a central constituent of centrioles and basal bodies. Compared with wild-type siblings, we could detect no obvious changes in the distribution or numbers of basal bodies in cells of the igu mutants (Fig. 2A,B). High resolution imaging of thin sections through the neural tube, cells within which project primary cilia in a polarized manner toward the central lumen, revealed that the basal bodies of the igu mutants, like those of wild-type embryos, were able to migrate to the apical surface (Fig. 2C,D). To verify whether the apically positioned basal bodies docked with the plasma membrane, we examined transmission electron microscope (TEM) sections. In wild-type embryos, we could readily detect the axonemes of primary cilia that arose from basal bodies anchored perpendicularly to apical membranes at the base of ciliary pits (Fig. 2E). By contrast, sections from mutant embryos were completely devoid of axonemes (Fig. 2F,G). Moreover, although the basal bodies appeared to dock perpendicularly with the apical membrane, formation of ciliary pits was not evident (Fig. 2F,G). All of the above observations allow us to conclude that in the absence of Igu function, there is an early and severe arrest of the primary ciliogenic program. Furthermore, Igu activity seems to be critically required for the formation of the ciliary pit and the subsequent step of axoneme growth, but not for the generation of basal bodies or for their migration to and positioning at the apical membrane.
As mentioned earlier, in addition to primary cilia, specific vertebrate tissues elaborate motile cilia; in zebrafish embryos, motile cilia occur, for instance, in Kupffer's vesicle (KV, an organ of laterality equivalent to the mammalian node), and the pronephric ducts (kidney tubules; Essner et al.,2005; Kramer-Zucker et al.,2005). Developmental abnormalities or improper motility of these cilia result in the disruption of left–right asymmetry and the formation of kidney cysts (Sun et al.,2004; Essner et al.,2005; Kramer-Zucker et al.,2005). The strong effect of the loss of igu function on the development of primary cilia led us to evaluate whether the gene is also critical for the assembly of motile cilia. We found that unlike the acute loss of primary cilia, igu mutant embryos nevertheless differentiated motile cilia in their KV and pronephric ducts (Fig. 3A–F). However, in KV, the cilia appeared variably reduced in length and sparser compared with wild-type embryos, whereas in the pronephric ducts cilia length was more or less normal, but their numbers were variably reduced, especially in the distal-most segment of the ducts (Fig. 3A–F). In keeping with these incompletely penetrant effects on the lengths and numbers of motile cilia, the determination of left–right asymmetry (as adduced by the chirality of heart tube rotation) was mildly affected in igu embryos, and they showed no evidence of cyst formation in their kidneys (Supp. Table S1). Therefore, in contrast to the primary cilia, the requirement of Igu function for motile cilia formation is less stringent. We propose that this disparity could reflect a redundant role of igu activity in cells that get specified to differentiate motile cilia.
Our recent work and that of Kintner and colleagues with the forkhead domain containing Foxj1 family of transcription factors has shown that they function as master regulators of the motile ciliogenic program (Yu et al.,2008; Stubbs et al.,2008). In the zebrafish embryo, two foxj1 genes, foxj1a and foxj1b, are expressed exclusively in tissues with motile cilia. In the absence of foxj1a function, formation of motile cilia in KV and pronephric ducts, but not of the more widely occurring primary cilia, is completely arrested. Moreover, provision of Foxj1 activity in cells that normally make primary cilia can re-program them instead to ectopically differentiate motile cilia-like cilia, at the expense of the primary cilia (Yu et al.,2008; Stubbs et al.,2008). We mis-expressed the Foxj1a protein in igu embryos using a heat inducible transgene to determine whether the mutant cells, like those of wild-type embryos, can be induced to make ectopic motile cilia in response to Foxj1a. Despite the fact that igu does not have an essential role in the generation of the endogenous motile cilia, strikingly, we found that embryos lacking igu function were completely refractory to Foxj1a activity, and showed no indication of any ectopic ciliation (Fig. 3G,H). This differential effect reinforces the notion that, in cells that normally differentiate motile cilia, Igu acts redundantly with another protein, whereas in those that make primary cilia, the requirement of Igu activity for ciliogenesis is indispensable (see the Discussion section on possible functional redundancy with the paralogous gene dzip1-like).
Much of our understanding of the mechanism of ciliary assembly and disassembly is centered on our knowledge of the IFT-dependent synthesis of the axoneme (Pedersen et al.,2008). In addition, many centrosomal proteins have now been shown to play essential roles in the control of primary ciliogenesis (Santos and Reiter,2008). This is perhaps unsurprising, because the basal body matures from the mother centriole. Consequently, in mice, protein products of almost all of the genes whose dysfunction leads to defects in ciliogenesis and Hh signaling are either components of IFT or the centrosome, and localize to the axoneme and/or the basal body (Wong and Reiter,2008). We have previously shown that the distribution of a green fluorescent protein (GFP) -tagged version of the Igu protein expressed in zebrafish embryos is predominantly cytoplasmic (Wolff et al.,2004); however, localization to specific cytoplasmic compartments was not investigated. Here, we first generated a heat inducible transgene to express GFP-tagged Igu transiently in zebrafish embryos. Induction of this transgene led to effective rescue of primary cilia formation in igu mutants, indicating that the GFP-Igu chimera is able to effectively substitute for the wild-type protein in promoting ciliogenesis (data not shown; see Supp. Table S1). However, such transient transgenesis, where injected DNA gets distributed in a mosaic manner, is not ideal for accurate determination of protein localization due to variability in the levels of transgene expression among cells within a particular embryo as well as between individual embryos. To circumvent this problem, we next generated a stable transgenic strain of zebrafish, Tg (hs::gfp-igu) (see the Experimental Procedures section), that reliably allows uniform levels of the GFP-tagged Igu protein expression in all cell-types, and in all transgenic embryos in response to heat induction. Using high resolution confocal microscopy, we then monitored the subcellular distribution of GFP-Igu in embryos derived from the Tg(hs::gfp-igu) fish to determine the localization of the protein with respect to the ciliary apparatus. Images gathered from the examination of several independent fields of cells drawn from several transgenic embryos showed that the protein consistently localized in a conspicuous punctate pattern to the bases of primary and motile cilia, but was excluded from their axonemes (Fig. 4A,B). Labeling with antibodies to γ-tubulin showed that the GFP-Igu punctae were always in very close proximity of the basal bodies, and often were found to colocalize with these structures (Fig. 4C,D).
Mice and flies, two organisms where the Hh pathway has been most extensively investigated at the genetic and cell biological levels, have completely opposing requirements for the primary cilium in signal transduction. In mice, the primary cilium is absolutely essential, whereas the organelle is completely dispensable for Hh activity in Drosophila. There is now evidence that primary cilia are also used for Hh signaling in amphibians and birds (Park et al.,2006; Yin et al.,2009). On the other hand, until very recently, there existed contradictory reports regarding the requirement of primary cilia for Hh signaling in the zebrafish. Mutations in several genes with important roles in ciliary development have been characterized, but none of them display any obvious effects on Hh signaling (Sun et al.,2004; Tsujikawa and Malicki,2004; Omori et al.,2008; Lunt et al.,2009). This list includes many genes encoding components of IFT; inactivation of the mammalian counterparts of these genes results in dysfunction of the Hh pathway. It has been shown that zebrafish ift genes are expressed maternally (Sun et al.,2004). Therefore, maternal mRNAs of the ciliary genes could be responsible for tempering the effects of the zygotic mutations on cilia formation. Against the view that cilia are not necessary for Hh pathway activity in the zebrafish are data which show that, in response to Hh, Smo translocates to primary cilia in zebrafish embryos just like it does in mammalian cells, and that transient “knock down” of certain ciliary genes causes a concomitant decrease in Hh signaling levels (Beales et al.,2007; Aanstad et al.,2009). A notable deficiency inherent in all of these studies—which seems to be the principal point behind the controversy—is that the extent to which the development of the primary cilia was affected was not scrutinized in substantial detail in any instance. While our manuscript was in review, Huang and Schier have demonstrated that zebrafish embryos lacking both maternal as well as zygotic expression of the IFT gene ift88, are unable to make any cilia (primary and motile), and exhibit defects in Hh signaling highly reminiscent of igu mutant embryos (Huang and Schier,2009). They have also independently discovered that igu mutant zebrafish have defects in ciliogenesis. This work, in combination with our earlier exhaustive analysis of anomalous Hh signaling in igu mutants, and the thorough characterization of the ciliogenic defects that we have presented in this report now provide incontrovertible genetic evidence that the role of the primary cilium as a hub for Hh signal transduction is indeed conserved in the zebrafish.
The findings of two other groups, which were published during the revision of our manuscript, have also shed light in evolutionary terms, on the role of Igu proteins in ciliogenesis and the function of cilia in Hh signaling. RNAi analysis of the planarian igu homolog and examination of zebrafish igu mutant embryos led Glazer and colleagues to conclude that Igu proteins are evolutionarily conserved components of ciliogenesis in metazoans (Glazer et al.,2009). They have further substantiated their claim through the combined knock down of the two human homologs of Igu—DZIP1 and DZIP1-like (these proteins, like zebrafish Igu, also localized to basal bodies)—which resulted in defective primary cilia formation in cultured cells. The zebrafish genome also contains the dzip1-like gene (our unpublished observation), and we propose that its functional redundancy with igu/dzip1 must account for the incompletely penetrant motile cilia phenotype apparent in the igu mutants. In a parallel study, Rink et al. have also demonstrated that the planarian igu homolog is essential for ciliogenesis (Rink et al.,2009). Strikingly, in contrast to the zebrafish, both of these studies show that the loss of igu and cilia in planarians has little effect on Hh signal transduction. Thus, regulation of ciliogenesis appears to be the ancestral function of the Igu/Dzip1 proteins, a function that has come to be of significance for the Hh pathway in vertebrates where the primary cilium is used for signal transduction.
Wild-type and heterozygous igu mutant (iguts294e) (Brand et al.,1996) strains of zebrafish were maintained under standard conditions of fish husbandry. The heat inducible Tg(hs::gfp-igu) stable transgenic strain was generated by injecting wild-type zebrafish eggs with linearized plasmid (see below) containing the hs::gfp-igu transgene. The resulting embryos were raised to adulthood, and transgenic founder fish were identified by heat shocking progeny embryos followed by screening for GFP expression. Identical pattern of GFP-Igu localization was observed in embryos derived from two independent founder fish. All experiments with zebrafish embryos were approved by the Singapore National Advisory Committee on Laboratory Animal Research.
Morpholino and Igu Expression Constructs
The anti-igu morpholino (Wolff et al.,2004) was purchased from GeneTools and dissolved in sterile water at a concentration of 1 mM. The full-length zebrafish igu cDNA (with its endogenous 5′ untranslated region) was used for the rescue of igu mutants by mRNA injection. Capped igu mRNA was synthesized from linearized plasmid using the mMessage Machine Kit (Ambion). The 5′ in-frame fusion of the gfp cDNA with that of igu, generated in our earlier study (Wolff et al.,2004), was subcloned into the pHspIG heat shock vector to produce hs::gfp-igu.
mRNA encoding Igu (∼0.1 μg/μl), linearized plasmids containing the hs::gfp-igu and hs::foxj1a-HA transgenes (∼25 ng/μl) and anti-igu morpholinos (300 μM) were injected into zebrafish embryos at the one-cell stage. Embryos injected with the hs::gfp-igu construct were raised to adulthood and then screened for stable transgenic founders.
Induction of Tg(hs::gfp-igu) and hs::foxj1a-HA transgenes
For assaying ectopic cilia formation in response to Foxj1a mis-expression, embryos injected with the hs::foxj1a-HA transgene construct were heat shocked twice (11 hpf and 24 hpf), at 37°C for 30 min at each time point, as described previously (Yu et al.,2008). The embryos were subsequently fixed for immunostaining at 26 hpf. For rescue of primary cilia in igu mutants by transient transgenesis using the hs::gfp-igu transgene, injected embryos were also heat shocked two times (14 hpf and 20 hpf), at 37°C for 1 hr at each stage, and then fixed at 24 hpf for immunostaining. For subcellular localization of the GFP-Igu protein, embryos obtained from Tg(hs::gfp-igu) transgenic fish were heat shocked (37°C) at 12 hpf for 40 min and then fixed at around 14 hpf for immunolabeling with anti-GFP together with acetylated tubulin or γ-tubulin antibodies.
In Situ Hybridization and Antibody Staining
In situ hybridization with the ptc1 probe and antibody staining were done according to standard protocols. The following antibodies were used: rabbit anti-HA (1:200; Santa Cruz Biotech), rabbit anti-βcatenin (1:200; Abcam), rabbit anti-GFP (1:2,000; Abcam), mAb anti-acetylated tubulin (1:500; Sigma), and mAb anti–γ-tubulin (1:500; Sigma). For confocal microscopy, Alexa Fluor-conjugated secondary antibodies (1:500; Molecular Probes) were used for detection of signals. Embryos were counterstained with DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) to visualize nuclei.
Quantitative Analysis of Embryonic Phenotypes
Quantification of numbers of mutant and morpholino-injected embryos that were analyzed for specific phenotypes are presented in Supp. Table S1.
Microscopy and Figure Preparation
Stained embryos were cleared and mounted in 70% glycerol. A Zeiss compound microscope (Axio plan2) and an Olympus Fluoview laser scanning confocal microscope were used for high-resolution imaging. Confocal imaging was done using PlanApo ×60 and ×100 lenses with NA of 1.42 and 1.45, respectively. TEM analysis of primary cilia in wild-type and igu mutant embryos was done as described previously (Yu et al.,2008). Figures were assembled using Adobe Photoshop CS4.
We thank members of our laboratory for discussion. S.R. is an adjunct faculty member in the Department of Biological Sciences, National University of Singapore.