miR-34b regulates multiciliogenesis during organ formation in zebrafish

Fluid flow and the resulting shear forces are important for specific aspects of organogenesis and tissue development, such as the establishment of left-right asymmetry, the formation of heart chambers, angiogenesis, hematopoietic stem cell emergence and neural progenitor cell migration (Freund et al., 2012; Lehtinen et al., 2011; Poelmann et al., 2008). In a dynamic fluid environment driven by cilia, the proper distribution of cilia along channel and cavity ducts, together with coordinated ciliary beating, control the force and direction of fluid flow. However, the principles that define cilia distribution and the resulting physiological implications remain unclear. Multiciliated cells (MCCs) feature multiple motile cilia on the cell surface and are widely distributed throughout the vertebrate body to regulate fluid movement in the intercellular space (Fliegauf et al., 2007; Hildebrandt et al., 2011). These cells remove mucus from the surface of the respiratory system, drive cerebrospinal fluid in the brain, perform olfactory functions and move ova in the oviducts. A recent report showed that Notch signaling is essential for the progression of multiciliogenesis (Marcet et al., 2011). However, research on the cellular and molecular mechanisms of multiciliogenesis is still in the early stages.

During zebrafish development, MCCs are abundantly distributed throughout the pronephric ducts and olfactory placode; thus, these structures serve as good models with which to study multiciliogenesis and its function in organ formation. A crucial process during organ formation is cellular rearrangement via coordinated cell migration, as is observed in the developing heart, liver, mammary gland and kidney (Abu-Issa and Kirby, 2007; Dressler, 2006; Ewald et al., 2008; Si-Tayeb et al., 2010). The vertebrate kidney is essential for blood homeostasis and the clearance of body waste. It is formed via the differentiation of multiple cell types and subsequent morphogenic cell movements (Dressler, 2006). Abnormalities in kidney and urinary tract development can lead to childhood kidney diseases, such as aplasia/dysplasia, vesicoureteral reflux and renal tubular dysgenesis.

The zebrafish has become an excellent model organism for kidney research due to the conservation of cellular and molecular regulatory pathways among vertebrates and the structural simplicity of the embryonic kidney (Drummond, 2005; Ebarasi et al., 2011). In zebrafish, morphogenic cell movement within the kidney begins at 24 hours post-fertilization (hpf), as a means of collective cell migration. The distal tubule cells then migrate proximally and manifest the position of each nephron segment. Subsequently, the proximal cells become convoluted by 72 hpf. Fluid flow, which is driven by blood pressure and cilia beating, is required for this process. The zebrafish pronephric ducts are filled with scattered single cilia prior to 30 hpf. Thereafter, organized cilia bundles replace the single cilia on the surface of cells in the middle pronephric ducts, whereas the single cilia at the proximal and distal ducts are preserved. The cilia in the middle pronephric ducts beat in a synchronized manner to drive the fluid flow from the anterior to posterior region of the duct (Kramer-Zucker et al., 2005; Liu et al., 2007; Ma and Jiang, 2007; Sullivan-Brown et al., 2008).

Although the relationship between ciliary defects and cystic kidney disease has been well established, little is known about the function of the cilia bundles in kidney development or about the molecular mechanisms that control multiciliogenesis. Disruption of fluid flow by a mechanical or genetic method can block both the migration of tubule cells and proximal tubule convolution (Vasilyev et al., 2009). However, the molecular mechanisms that regulate these processes are poorly understood.

A variety of microRNA species modulate organogenesis in vertebrates; however, few have been shown to participate in kidney morphogenesis and olfactory organ development. The miR-34 family, which is highly conserved in vertebrates, is downregulated in many cancer types and plays an important role in carcinogenesis. By functioning downstream of p53 signaling, miR-34 family members regulate cell proliferation, migration and apoptosis (Dressler, 2006; He et al., 2007; Hermeking, 2010). Under normal physiological conditions, miR-34a is involved in neuronal development and miR-34c is functionally required for spermatogenesis (Agostini et al., 2011a; Agostini et al., 2011b; Bouhallier et al., 2010; Liu et al., 2012). However, the role of miR-34b in development is largely unknown. Here, we report that the expression of miR-34b is enriched in both kidney MCCs and the olfactory placode and that miR-34b is required for normal kidney morphogenesis and olfactory organ development. miR-34b regulates multiciliogenesis, which is essential for the proper migration of kidney cells. In addition, we demonstrate that cmyb is a key mediator of this process.

The zebrafish facility and zebrafish study were approved by the Institutional Review Board of the Institute of Health Sciences, Shanghai Institutes of Biological Sciences, Chinese Academy of Sciences (Shanghai, China).

Wild-type AB and transgenic Tg(cmyb:EGFP) (North et al., 2007), cmyb^hkz and Tg(hsp70:cmyb) (Zhang et al., 2011) zebrafish were raised under standard conditions. Tg(hsp70:cmyb) embryos and siblings were heat shocked at 40°C for 1 hour every 3 hours from 18 hpf and fixed at 36, 48 or 72 hpf for analysis and identified by genomic PCR.

Morpholinos (Gene Tools) and miR-34b duplex (GenePharma, Shanghai) were injected in 1- to 2-cell stage embryos. Morpholinos and duplexes (including dosages) are detailed in supplementary material Table S1.

The miR-34b in situ hybridization was performed at 50°C using miRCURY detection probe according to the manufacturer’s protocol (Exiqon). For the in situ hybridization and immunofluorescence double staining, fluorescent in situ hybridization was performed first according to published protocols (Brend and Holley, 2009) using a Cy3 kit (Perkin Elmer); then, the antibody staining was performed using rabbit anti-GFP polyclonal serum (1:500; Invitrogen) and goat anti-rabbit Alexa Fluor 488 antibody (1:500; Invitrogen). For immunofluorescence double staining, first primary antibody against GFP was mixed with antibody against acetylated tubulin 6-11b-1, γ-tubulin (1:500; Sigma-Aldrich) or against PH3 (1:500; Santa Cruz), followed by staining with the secondary antibody (1:500; Alexa Fluor 488 or 546) the next day. TUNEL staining was performed as described (Fu et al., 2009).

A tissue block containing the kidney was dissected from Tg(cmyb:EGFP) embryos 3 days post-fertilization (dpf) (supplementary material Fig. S3A), digested with 0.25% trypsin (Invitrogen) for 1 hour and filtered through a 40 μm cell strainer (BD Falcon). About 20,000 GFP⁺ cells were sorted out using a FACSAria flow cytometer (BD) and used for RNA extraction. The percentage of GFP⁺ cells in the tissue block was 0.4±0.05%. Greater than 90% of cells were viable before and after cell sorting as counted by Trypan Blue staining.

For microarray analysis (performed by the Shanghai Biochip Company), ∼100 ng total RNA was used for cDNA generation, amplified and hybridized to Affymetrix Zebrafish Genome Arrays. Two independent sets of biological replicates (miR-34b MO and Ctrl MO) were used. The data were normalized using MAS 5.0 (Affymetrix). Student’s t-test was performed to identify genes with a significant change in expression (P<0.05). Genes with greater than 1.5-fold change in expression were used for gene ontology (GO) analysis by DAVID (Huang et al., 2009). The GOTERM_BP_FAT category was used and GO terms with P<0.05 were listed according to the fold enrichment score. The microarray data have been deposited in the Array Express database (www.ebi.ac.uk/arrayexpress) under accession number E-MEXP-3893.

Quantitative RT-PCR (QRT-PCR) reactions were performed using the Roche 480 system. Each sample was tested in triplicate and β-actin 1 was used to normalize signals for each queried transcript using the ΔΔCt method. Primers are listed in supplementary material Table S2.

Tg(cmyb:EGFP) embryos at 51-72 hpf were mounted in a self-made imaging chamber as previously reported (Kamei and Weinstein, 2005). Time-lapse imaging was performed using a Leica TCS SP5 confocal microscope with 10× objective every 10 minutes. Cell migration analysis was performed using Image-Pro Analyzer 7.0 (Media Cybernetics) image analysis software according to a published protocol (Hall et al., 2009); each embryo was captured independently and combined using Corel Videostudio Pro X2. Cells that were successfully tracked at all the time points were selected and their migration distances used for statistical analysis. Three independent experiments were performed for miR-34b MO- or Ctrl MO-injected Tg(cmyb:EGFP) embryos.

A 5% solution of tetramethylrhodamine-conjugated 70 kDa dextran (Molecular Probes) was injected into the common cardinal vein of 3.5-dpf embryos (Drummond and Davidson, 2010; Malicki et al., 2011) and urine excretion at the cloaca was recorded using a SteREO Discovery V20 fluorescence stereomicroscope (Carl Zeiss). The time from injection to the first visible urine excretion at the cloaca was used for statistical analysis.

The miR-34b target site (MRE) was predicted by MicroCosm (EBI, EMBL) and KeyTar (L. H. Lai group, East China Normal University) (supplementary material Table S3). 2×MRE or 3′ UTR was cloned and fused to EGFP in the pCS2⁺ plasmid (Fu et al., 2009). Capped mRNAs (with DsRed as an internal control) were synthesized with the mMESSAGE mMACHINE kit (Ambion) and injected with miR-34b duplex or Ctrl duplex as described previously (Fu et al., 2009).

All statistics were performed using an unpaired Student’s t-test.

A large-scale microRNA expression study previously showed that miR-34b is expressed in a punctate pattern along the zebrafish pronephric ducts at 3 and 5 dpf (Wienholds et al., 2005). We further characterized the expression of miR-34b during early development by whole-mount in situ hybridization (WISH). There was no maternal miR-34b expression; enriched zygotic expression along the pronephric ducts was detected beginning at 24 hpf and expression in the olfactory placode was detectable from 36 hpf (Fig. 1E, arrows; supplementary material Fig. S1). This expression pattern resembled the pattern of cmyb-EGFP-labeled kidney cells along the pronephric ducts in Tg(cmyb:EGFP) embryos (Bertrand et al., 2008; North et al., 2007) (Fig. 1A). These cells were thought to be MCCs and to possess hematopoietic potential (Bertrand et al., 2008; Vasilyev et al., 2009). Double fluorescent staining using an anti-EGFP antibody and a miR-34b probe showed that miR-34b is co-expressed with the cmyb-EGFP transgene (Fig. 1B). Surprisingly, we observed a ring-like pattern of miR-34b expression, consistent with its location in the cytoplasm, indicating that miR-34b functions mainly in the cytoplasm.

To further clarify whether the kidney cells labeled in the Tg(cmyb:EGFP) embryos are MCCs, we performed double fluorescent staining using anti-EGFP antibodies and RNA probes for rfx2, a known molecular marker of MCCs (Liu et al., 2007; Ma and Jiang, 2007). As shown in Fig. 1C, EGFP⁺ cells in the pronephric duct colocalized with rfx2. These results confirmed that cmyb-EGFP⁺ cells in the Tg(cmyb:EGFP) embryos are multiciliated kidney cells. Immunostaining with anti-EGFP and anti-acetylated tubulin antibodies showed that each EGFP⁺ cell had a cilia bundle consisting of multiple cilia (Fig. 1D). We also found expression of miR-34b, rfx2 and cmyb in the olfactory placode (Fig. 1E), where multiciliated sensory neurons reside. These results demonstrate that miR-34b expression is enriched in rfx2⁺ and cmyb⁺ MCCs in the kidney and olfactory placode. In the following experiments, we focused our efforts on the kidney to study the function of miR-34b in MCCs.

To study the in vivo function of miR-34b, two morpholinos (MOs) (miR-34b MO1 and MO2) were designed to block the maturation of miR-34b (supplementary material Fig. S2A). These MOs inhibited the maturation of miR-34b without causing circulation defects or global developmental abnormalities in the morphants, such as developmental retardation or body symmetry defects. These observations were consistent with the restricted expression pattern of miR-34b. However, kidney morphogenesis was clearly disrupted in these embryos at 3 dpf (Fig. 2). Both MOs led to similar kidney defects (supplementary material Fig. S2B), although miR-34b MO1 gave stronger and more consistent phenotypes and was therefore used in subsequent experiments.

In the miR-34b morphants, there was a strong accumulation of rfx2⁺ cmyb⁺ MCCs in the middle region of the pronephric ducts (Fig. 2A-F). Using the pronephric epithelial marker pax2, we found that proximal tubule convolution was blocked (Fig. 2G-I). The distribution of the distal trpm7⁺ transporting epithelial cells was also abnormal; these cells were shorter than normal and failed to extend towards the head kidney (Fig. 2J-L). The tubule diameter was severely expanded compared with that of control animals [41±8 μm (n=12) versus 25±5 μm (n=10), mean ± s.d.; P<0.0001, Fig. 2M,N].

The cmyb-EGFP⁺ cells along the pronephric duct have hematopoietic potential (Bertrand et al., 2008). We analyzed the development of different blood lineages in the miR-34b morphants by WISH at 30 and 36 hpf and at 2, 3 and 5 dpf using scl (tal1 - Zebrafish Information Network) to monitor progenitor cells, mpo (mpx) to mark the myeloid lineage, gata1 and hemoglobin α embryonic 1 (hbae1) to label the erythroid lineage, CD41 (itga2b) to mark thrombocytes and rag1 to indicate T-cell development. There were no differences in overall cell numbers or cell lineage types detected at any of the hematopoietic sites investigated, including the head kidney (supplementary material Fig. S2D). We conclude that the miR-34b-expressing cells in the pronephric duct region were not hematopoietic stem cells prior to 5 dpf.

To understand the cellular basis of the observed kidney phenotypes, we first focused on the MCCs, in which miR-34b is highly expressed. MCCs may accumulate in the middle region of the pronephric ducts at 3 dpf for a number of reasons. It could be attributed to overproliferation or decreased apoptosis of MCCs in this region. Alternatively, there may be biased differentiation of bipotential precursor cells toward an MCC fate at the expense of trpm7⁺ transporting epithelial cells (Liu et al., 2007; Ma and Jiang, 2007). It is also possible that cell migration is blocked in these embryos. To discriminate between these possibilities, we first manually counted the number of MCCs and found that this value was normal from 36 to 48 hpf in the miR-34b morphants (supplementary material Fig. S2C). Next, we analyzed cell proliferation and apoptosis by phosphohistone H3 (PH3) and TUNEL staining, respectively, at 36, 48 and 72 hpf. In both the controls and miR-34b morphants, only a few MCCs were PH3⁺, and no increase in PH3 staining was found in the MCCs at any of the developmental stages analyzed (Fig. 3A). There was also no change in TUNEL staining of the pronephric duct region in the miR-34b morphants (Fig. 3B). These results demonstrated that the accumulation of MCCs that is observed when the levels of miR-34b are reduced is not due to alterations in either cell proliferation or apoptosis. We then analyzed the balance of bipotential cell differentiation by trpm7 WISH at 36 hpf. In the miR-34b morphants, both the distribution of trpm7⁺ cells and the number of MCCs were normal in comparison to the control animals (Fig. 3C,D; supplementary material Fig. S2C), suggesting that there is no biased differentiation of precursor cells toward MCCs at the expense of trpm7⁺ transporting epithelial cells.

Finally, we tracked the migration of the MCCs by confocal time-lapse imaging from 51 to 72 hpf using Tg(cmyb:EGFP) fish. We found that the migration distance of the MCCs in miR-34b morphants was significantly decreased compared with that of control animals [66±11 μm (n=10) versus 110±23 μm (n=10); Fig. 3E-G]. The migration of the cells in the proximal ducts was substantially blocked, whereas migration in the distal ducts was almost normal (Fig. 3E,F; supplementary material Movie 1), possibly owing to the previously reported slow migration of distal cells (Vasilyev et al., 2009). The inhibited migration of the proximal MCCs towards the head kidney section might explain why the MCCs accumulated in the middle region of the pronephric ducts and suggests that miR-34b is important for the proper migration of MCCs during kidney morphogenesis.

To determine how miR-34b regulates the migration of MCCs and kidney morphogenesis, we isolated MCCs from the body section of Tg(cmyb:EGFP) fish at 3 dpf by FACS and performed microarray analysis on these cells (supplementary material Fig. S3A). The expression of 456 genes (of the 8296 examined) was changed more than 1.5-fold in MCCs of the miR-34b morphants (supplementary material Table S4), with 381 genes upregulated and 75 genes downregulated, when compared with MCCs from embryos injected with control MO. A GO enrichment analysis of these genes showed that cilium assembly and morphogenesis were among the processes most significantly affected in the miR-34b knockdown embryos (supplementary material Table S5). These include ciliary structural genes and genes involved in intraflagellar transport (Fig. 4A). The increased expression of these genes, along with key transcription factors for ciliogenesis such as foxj1a, foxj1b and rfx2 (Bisgrove et al., 2012; Chung et al., 2012; Stubbs et al., 2008; Yu et al., 2008), was confirmed by QRT-PCR (Fig. 4B) and WISH (supplementary material Fig. S3B).

These results raised the possibility that multiciliogenesis might be affected in MCCs by miR-34b knockdown. We analyzed multiciliogenesis in MCCs by immunostaining with antibodies against acetylated tubulin, which is a major component of cilia, and EGFP, which marks cmyb-EGFP⁺ cells. At 30 hpf, the pronephric ducts in both the control animals and miR-34b morphants were filled with scattered single cilia (supplementary material Fig. S3C). However, at 36 hpf, when the cilia bundle was formed in each MCC in the control embryo, few cilia bundles were observed in the MCCs of the miR-34b morphants (Fig. 4C,D, arrows). By 2-3 dpf, the cilia bundles of the MCCs were aligned with the direction of fluid flow in the control embryo, whereas the thin and disorganized cilia bundles in the MCCs of miR-34b morphants were perpendicular to the fluid flow (supplementary material Fig. S3D, arrows). The single cilia in the proximal and distal ducts were, however, normal (supplementary material Fig. S3E, arrows). Furthermore, the cilia bundles in the olfactory placode did not form in the miR-34b morphants (Fig. 4E). On average, cilia were shorter in miR-34b morphants (2.3±0.3 μm) than in control animals (5.8±0.6 μm). These observations indicate that miR-34b plays essential roles in the formation and organization of cilia bundles in MCCs.

A previous report showed that the beating of cilia in the pronephric ducts drives directional fluid flow within the ducts (Kramer-Zucker et al., 2005). This led us to speculate that fluid flow might also be affected in miR-34b morphants. We analyzed fluid flow using a dye excretion assay (Fig. 4F). In accordance with the previous report, the time needed for excretion at the cloaca was 3.7±2.5 minutes in the control embryos. However, in miR-34b morphants the time needed more than doubled to 9.9±6.3 minutes (Fig. 4G), suggesting that fluid flow was severely affected. Another report showed that fluid flow is required for proximal tubule convolution and cell migration; blockage or irregularity of fluid flow results in defects in convolution and cell migration in the pronephric ducts (Vasilyev et al., 2009). Taken together, our results suggest that miR-34b regulates cilia bundles on MCCs, which are important for maintaining a steady fluid flow during kidney morphogenesis.

We next searched in silico prediction databases for putative miR-34b targets based on microarray data and the published literature (Hermeking, 2010). Ultimately, 24 genes (supplementary material Table S3) were tested for miR-34b regulation using an in vivo reporter assay (Fu et al., 2009), in which either the predicted microRNA recognition sites (MREs) or the 3′ UTR was fused to EGFP and injected along with either a miR-34b duplex or control duplex. Transcripts of cmyb were upregulated 2-fold according to the microarray data, and the upregulation was confirmed both by QRT-PCR in FACS-isolated MCCs and by WISH analysis of whole embryos (Fig. 5A,B,D). We found two conserved miR-34b target sites in the 3′ UTR of cmyb. However, our reporter assay suggested that cmyb might not be a direct target of miR-34b (supplementary material Table S3). Enhanced fluorescence in the MCCs of Tg(cmyb:EGFP) embryos indicated that cmyb expression was regulated at the transcriptional level (Fig. 5C).

A previous report showed that cmyb function is required for multiciliogenesis in the subventricular zone of the brain (Malaterre et al., 2008). During normal kidney development, expression of cmyb in MCCs begins at 24 hpf and decreases after 36 hpf, as monitored in the Tg(cmyb:EGFP) line (data not shown). We hypothesized that the upregulation of cmyb might play an important role in the defects caused by the inhibition of miR-34b function and predicted that normalization of cmyb expression levels in miR-34b morphants might rescue the cilia and kidney morphogenesis defects, whereas overexpression of cmyb alone should mimic the defects caused by reduced levels of miR-34b. When a series of 0.075 mM or 0.125 mM doses of cmyb MO were injected in combination with the miR-34b MO into Tg(cmyb:EGFP) embryos these treatments significantly rescued multiciliogenesis, MCC distribution and proximal tubule convolution (Fig. 5E-G). Overexpression of cmyb in Tg(hsp70:cmyb) zebrafish (Zhang et al., 2011) blocked cilia bundle formation and induced the accumulation of MCCs (Fig. 5H,I), similar to what was observed in the miR-34b morphants (Fig. 2; Fig. 4B). These results confirm that increased cmyb expression is directly responsible for the defective multiciliogenesis and abnormal kidney morphogenesis observed in miR-34b morphants.

We also found that cmyb was upregulated in the olfactory placode of miR-34b morphants (Fig. 5J). The injection of 0.125 mM cmyb MO together with the miR-34b MO rescued cilia bundle formation in the olfactory placode (Fig. 5K). These results show that the miR-34b-cmyb genetic pathway is required for multiciliogenesis in both the kidney and the olfactory organ during development.

Multiciliogenesis begins with centriole multiplication in the cytoplasm, which is followed by membrane docking of the centrioles (Dawe et al., 2007). Using γ-tubulin as a centriole marker, we found tens to hundreds of centrioles scattered throughout the cytoplasm of MCCs at 24 hpf. At 30 hpf, centrioles began to aggregate in the cytoplasm and migrate to the cell surface in a fraction of MCCs. Most of the centrioles reached the cell surface by 36 hpf. However, in miR-34b morphants, most of the centrioles were retained in the cytoplasm until at least 48 hpf (Fig. 6A), indicating that the centrioles were unable to dock at the membrane. The injection of 0.125 mM cmyb MO together with the miR-34b MO rescued the centriole distribution defect in the MCCs of the kidney (Fig. 6A). These results suggest that the miR-34b-cmyb genetic pathway is required for the proper membrane docking of centrioles.

Genes involved in centriole formation, such as plk4, stil, pcm1, nek2, poc1a and cep72, were also more highly expressed in the MCCs of miR-34b morphants compared with control animals (Fig. 6C). plk4 and stil play key roles in regulating centriole duplication (Brito et al., 2012). WISH analysis showed that plk4 and stil are first expressed at ∼24 hpf in MCCs, were downregulated from 30 hpf and were not detected at 36 hpf (Fig. 6D; Fig. 7F). This expression pattern is consistent with the timing of centriole multiplication, indicating that plk4 and stil might play important roles during centriole multiplication. However, in the miR-34b morphants, expression of plk4 and stil was maintained at a high level until at least 36 hpf (Fig. 6D,E). These results suggest that there is a prolonged centriole multiplication period in miR-34b morphants.

Next, we tested how cmyb is involved in miR-34b-regulated centriole multiplication by performing loss- and gain-of-function analyses using the cmyb^hkz3 mutant (Zhang et al., 2011) and the hsp70:cmyb transgenic line. In the middle section of the kidney duct fewer MCCs had membrane-docking centrioles in cmyb mutants than in wild-type embryos (Fig. 7A,C). Accordingly, fewer cilia bundles with organized structures were observed in the cmyb mutants (Fig. 7B,D). Loss of cmyb also resulted in the downregulation of plk4 and stil, but not of foxj1b or rfx2. Conversely, overexpression of cmyb blocked the membrane docking of centrioles and induced a profound and specific upregulation of plk4, stil, rfx2 and foxj1b in kidney MCCs, consistent with the results from the miR-34b morphants (Fig. 7E-G) and suggesting that cmyb participates in the regulation of multiciliogenesis by controlling the key transcriptional regulators of centriole multiplication. These results suggested that an optimal level of cmyb expression is required for both centriole multiplication and multiciliogenesis.

Proximal tubule convolution and distal cell migration are important for the formation of a functional kidney in zebrafish. A previous report showed that fluid flow driven by blood pressure and cilia is essential for this process (Vasilyev et al., 2009). Here, we show that proper multiciliogenesis of MCCs is required for kidney morphogenesis and identify the miR-34b-cmyb axis as an important genetic pathway in the regulation of this process. miR-34b was preferentially expressed in MCCs and may indirectly suppress cmyb expression, permitting normal multiciliogenesis, fluid flow and kidney cell migration. The miR-34b-cmyb genetic pathway controls the multiplication and membrane docking of centrioles during multiciliogenesis, providing new insight into the molecular regulation of this key process. The miR-34b-cmyb genetic pathway is also required for multiciliogenesis in the olfactory placode. Because miR-34b is specifically expressed in MCC-enriched tissues in human and Xenopus (Hsu et al., 2008; Liang, 2008; Marcet et al., 2011), it is likely that the miR-34b-cmyb genetic pathway is evolutionarily conserved and widely used for the control of multiciliogenesis. This finding has improved our understanding of fluid flow control, organ morphogenesis and multiciliogenesis.

Fluid flow and fluid shear forces are important during organogenesis and tissue development. In a cilia-driven fluid system, the proper distribution of cilia along the duct (channel) is essential for constant fluid flow. However, the underlying mechanisms and functional importance of cilia distribution during development are not well understood. In zebrafish, the pronephric duct is filled with both scattered single cilia (proximal and distal ducts) and organized cilia bundles (middle ducts) (Kramer-Zucker et al., 2005; Liu et al., 2007; Ma and Jiang, 2007; Sullivan-Brown et al., 2008). miR-34b was preferentially expressed in MCCs, and knockdown of miR-34b specifically blocked the formation and maintenance of the cilia bundles without disturbing the single cilia, thus revealing the essential function of the cilia bundles in the middle ducts.

The major difference between miR-34b morphants and other cilia mutants is the lack of cyst formation in the anterior kidney (Sullivan-Brown et al., 2008). This result is also consistent with a previous report on the cyst-inducing mutant kurly, which displays kidney cysts and impaired single cilia, but normal cilia bundles (Sullivan-Brown et al., 2008). These results suggest that single cilia play an important role in cyst formation at the anterior kidney, whereas cilia bundles do not. According to mammalian kidney cyst models, the sensory role of the primary cilia is important during cyst formation (Chapin and Caplan, 2010), raising the question of whether the single cilia in the zebrafish kidney duct have a sensory function. Our results show that the proper distribution of single cilia and cilia bundles during organogenesis is important for correct fluid control and organogenesis. The principles that regulate cilia formation and distribution merit further research.

Although many signaling pathways and molecules play key roles in regulating primary cilia or single motile cilia, few are known to be important in the molecular control of multiciliogenesis. In our studies, we found that an optimal level of cmyb expression is crucial for proper multiciliogenesis. Higher expression levels of cmyb block the membrane docking of centrioles, whereas loss of cmyb impairs centriole multiplication, and both result in defects in cilia bundle formation and organization. miR-34b serves as a guardian to maintain optimal levels of cmyb. Further studies will utilize RIP-ChIP to provide insight into the targets of miR-34b and discover how it is involved in maintaining the level of cmyb in MCCs (Nelson et al., 2010; Thomson et al., 2011). We found that plk4 and stil were preferentially expressed in MCCs during centriole multiplication and were upregulated after miR-34b knockdown. Loss- and gain-of-function studies showed that cmyb is genetically upstream of plk4 and stil, which thus participate in the process of centriole multiplication. How plk4 and stil participate in centriole multiplication deserves further research.

We observed the upregulation of ciliary genes, including rfx2 and foxj1a, in miR-34b morphants, which is consistent with previous reports on cilia mutants. This upregulation is correlated with pronephric duct dilation, which is induced by a stretching force (Hellman et al., 2010). In the miR-34b morphants we observed similar tubule dilation (Fig. 2I,J), consistent with the upregulation of these cilia genes. The upregulated ciliary genes could also reflect a state of incomplete multiciliogenesis resulting from the blockage of centriole membrane docking, suggesting that the processes of centriole multiplication and ciliogenesis are closely connected. Further research is required to illustrate how these two processes are related.

miR-34b regulates the proliferation, apoptosis and migration of cancer cells by intersecting with p53 signaling (He et al., 2007; Hermeking, 2010). In our analysis, we did not observe any changes in cell proliferation, apoptosis or in the components of the p53 pathway in the miR-34b morphants, which might be because MCCs were not in a proliferative state. However, we did discover a link between miR-34b and centriole multiplication. It will be interesting to determine whether this centriole duplication defect is involved in miR-34b-related cancer. In addition, a recent study showed that primary cilia play an important role in tumorigenesis (Hassounah et al., 2012). Our results also raise the question of whether miR-34b contributes to cilia-related cancers, especially in tissues with MCCs.

We thank Dr Len Zon at Harvard Medical School for providing the Tg(cmyb:EGFP) line. Dr Yiyue Zhang at Southern Medical University (China) provided the Tg(hsp70:cmyb) and cmyb^hkz3 lines; Song Guo and Haiyang Hu of Dr Philipp Khaitovich’s group at CAS-MPG Partner Institute helped with the microarray analysis; Dr Lihui Lai at East China Normal University helped with microRNA target identification; and Drs Anming Meng at Tsinghua University, Iain A. Drummond at Massachusetts General Hospital and Laure Bally-Cuif at CNRS provided WISH plasmids. We thank Dr Weijun Pan and all members of the laboratory for helpful discussions.