miR-219 regulates neural progenitors by dampening apical Par protein-dependent Hedgehog signaling

Spatial and temporal information are integrated in developing nervous systems to produce a large array of distinct types of neurons and glia. In very broad terms, spatial cues specify formation of different kinds of neurons and glia at different positions whereas temporal information regulates the time at which particular dividing progenitor cells differentiate. One of the major spatial cues for developing vertebrate nervous systems is the secreted morphogen Sonic hedgehog (Shh). Shh binds to the cell surface transmembrane protein Patched (Ptch), relieving Ptch inhibition of the membrane spanning protein Smoothened (Smo). This initiates proteolytic processing of Gli transcription factors, which subsequently translocate to the nucleus to control expression of Hh pathway target genes (Aza-Blanc et al., 1997; Wang et al., 2000). Secretion of Shh by notochord, a mesodermal rod underlying the ventral neural tube and subsequently the floor plate – the ventral-most cells of the neural tube – creates a signaling gradient, to which neural progenitors arrayed on the ventral-to-dorsal axis respond by expressing position-specific transcription factors and producing distinct subsets of neurons and glia (Le Dréau and Martí, 2012; Jessell, 2000; Dessaud et al., 2008). For example, pMN progenitors, which are positioned close to the floor plate, express Olig2 and give rise to motor neurons and oligodendrocytes, whereas p2 progenitors slightly further from the Shh source, express Nkx6.1 and Irx3 and produce v2 interneurons and astrocytes.

Shh also functions as a mitogen, driving proliferation of neural progenitors (Ulloa and Briscoe, 2007). For instance, a recombinant N-terminal active fragment of Shh (Shh-N) elevated cell proliferation in cultured mouse retinas (Jensen and Wallace, 1997) and mouse cerebellar cell and slice cultures (Wechsler-Reya and Scott, 1999). Similarly, persistent expression of Shh in developing mouse spinal cord and cerebellum caused excess proliferation (Dahmane and Ruiz, 1999; Rowitch et al., 1999), and mice lacking functional Shh signaling inhibitors had larger neural tubes (Jeong and McMahon, 2005). Additionally, Shh mutant mice had significantly smaller neural tubes and general growth retardation (Litingtung and Chiang, 2000). Analysis of cell cycle regulation using cultured mouse cerebellar cells suggested that Shh can promote proliferation by upregulating signaling mediated by G1 cyclins (Kenney and Rowitch, 2000).

How do neural progenitors escape the mitogenic effect of Shh signaling to exit the cell cycle and differentiate? One possibility is that progenitors become less sensitive to Shh. During the early stages of neural development, progenitors have neuroepithelial characteristics, including apicobasal polarity. A prominent feature of this polarity is the assembly of protein complexes at the apical membrane. For example, the partioning-defective (Par) proteins Pard3, Pard6 and atypical protein kinase C iota (Prkci), form a complex that localizes to apical contact points between neuroepithelial cells (Afonso and Henrique, 2006; Chen and Zhang, 2013). In Drosophila, unequal distribution of apical Par polarity proteins maintains neuroblast fate during asymmetric divisions (Lee et al., 2006; Prehoda, 2009). Similarly, in mice, high levels of apical Par polarity proteins generally correlate with dividing neural progenitors and low levels with differentiation (Bultje et al., 2009; Costa et al., 2008; Kosodo et al., 2004). Notably, in vertebrate epithelial cells, Shh reception and signal processing occurs within primary cilia, which are cellular appendages positioned within the apical membrane (Ishikawa and Marshall, 2011). Recent evidence indicates that apical Par polarity proteins promote ciliogenesis (Fan et al., 2004; Krock and Perkins, 2014; Sfakianos et al., 2007). Together, these observations raise the possibility that apical Par proteins regulate the Shh signaling response in neural progenitors by regulating ciliogenesis.

We recently presented evidence that the microRNA miR-219 promotes differentiation of zebrafish neural progenitors at late stages of neural development by downregulating Pard3 and Prkci (Hudish et al., 2013). However, the mechanisms by which apical Par proteins maintain neuroepithelial progenitors and how their downregulation promotes differentiation have not been determined. In this study, we tested the hypothesis that apical Par proteins maintain neuroepithelial progenitors by promoting their ability to respond to Hh signals. Our data show that miR-219-deficient embryos have elevated Hh signaling, which is dependent upon Par protein functions. Embryos lacking Par protein functions have decreased Hh signaling, which is not rescued by depleting miR-219, indicating that miR-219 regulation of Hh signaling occurs through Par protein regulation. We also show that miR-219 and Par proteins regulate the length and rate of growth of primary cilia. We propose that miR-219 helps to initiate the transition of progenitors from proliferation to differentiation by dampening apical Par protein-dependent Hh signal transduction, which might occur, at least in part, by regulating ciliogenesis.

We previously showed that termination of neural progenitor divisions and subsequent neuronal and glial differentiation in zebrafish embryos is driven, in part, by negative regulation of Pard3 and Prkci by miR-219 (Hudish et al., 2013). Because Hh signaling is important for neural progenitor proliferation (Jensen and Wallace, 1997; Rowitch et al., 1999; Wechsler-Reya and Scott, 1999), we postulated that apical Par polarity proteins maintain neural progenitors by promoting responsiveness to Hh signals. To establish a baseline of Hh expression and response in zebrafish embryonic spinal cord, we used in situ RNA hybridization to detect expression of sonic hedgehog a (shha) and patched2 (ptch2), which encodes a Hh receptor. Transcription of ptch2 is increased with Hh signaling and ptch2 RNA can therefore be a readout for Hh pathway activity (Concordet et al., 1996). Floor plate cells expressed shha at 1, 2 and 3 days post fertilization (dpf) (Fig. 1A-C). At 1 and 2 dpf, cells surrounding the primitive lumen, which extends across the dorsoventral axis of the neural tube, expressed ptch2 (Fig. 1G,H). At 3 dpf, by which time the primitive lumen had transformed into a small, ventrally positioned central canal, only cells surrounding the central canal expressed ptch2 (Fig. 1I). These observations indicate that during early neural development, spinal cord cells located distant from the floor plate are responsive to Hh signals, whereas later in development, Hh signaling is limited to cells in the ventral spinal cord. Notably, this transformation of the Hh response coincides with a loss of neuroepithelial characteristics, reduction in neural progenitor number and neuronal and glial differentiation in dorsal spinal cord (Hudish et al., 2013).

If miR-219 promotes neural progenitor differentiation by reducing apical Par polarity proteins and, consequently, Hh signaling, then miR-219-deficient embryos should have an altered Hh response. To test this prediction, we injected embryos with an antisense morpholino oligonucleotide (MO) designed to block the function of miR-219, which is expressed specifically in the neural tube (Zhao et al., 2010; Hudish et al., 2013). At 1, 2 and 3 dpf, spinal cord shha expression, revealed by in situ RNA hybridization, was similar in miR-219 MO-injected embryos (Fig. 1D-F) and control embryos (Fig. 1A-C). Similarly, ptch2 expression appeared to be mostly unchanged by miR-219 MO injection at 1 and 2 dpf (Fig. 1J,K) compared with controls (Fig. 1G,H). However, at 3 dfp, ptch2 RNA was evident more dorsally in miR-219 MO-injected embryos (Fig. 1L) compared with control embryos (Fig. 1I). To quantify the Hh signaling response, we used quantitative RT-PCR (qPCR) with a probe that measures the levels of ptch2 RNA. This revealed that ptch2 transcripts were more abundant in 3 dpf miR-219-deficient embryos than in control embryos (Fig. 1M). Additionally, we assessed the levels of RNA encoded by gli1, which is also a transcriptional target of Hh signaling activity (Rowitch et al., 1999). Embryos lacking miR-219 function had more gli1 transcripts than control embryos (Fig. 1M). Finally, to measure the spatial distribution of ptch2 expression within living embryos, we quantified the fluorescence localization profile of Tg(ptch2:Kaede) embryos, which express Kaede fluorescent protein under control of ptch2 regulatory DNA (Huang et al., 2012). At 1 dpf, the shapes of the normalized fluorescence profiles were similar across the spinal cords of control and miR-219 MO-injected embryos (Fig. 1N,O,R,S,V). By contrast, the fluorescence profile was expanded in dorsal spinal cords of 3 dpf miR-219 MO-injected embryos relative to control embryos (Fig. 1P,Q,T,U,W). Together, these data indicate that miR-219-deficient embryos retain elevated Hh signaling in dorsal spinal cord into a late stage of embryonic development, consistent with the possibility that miR-219 reduces the Hh signaling response of neural progenitors.

We previously showed that reduction in miR-219 function resulted in maintenance of apical Par protein expression in the dorsal spinal cord and elevated numbers of proliferating neural progenitors (Hudish et al., 2013). Do these features of the miR-219 loss of function phenotype result from elevated Hh signaling? To test this possibility, we treated miR-219 MO-injected and control embryos with cyclopamine, a molecule that inhibits Hh pathway activation by binding to Smoothened (Chen et al., 2002). As expected, cyclopamine applied at 24 hpf nearly eliminated ptch2 transcripts by 3 dpf (Fig. 3A). Treatment of wild-type embryos with cyclopamine beginning at 24 hpf did not change the localization of apical proteins ZO-1 (Fig. 2A,B,R) or Prkci (Fig. 2E,F,S), which both outlined the small, ventrally positioned central canal. Consistent with our previous data (Hudish et al., 2013), miR-219-deficient embryos had enlarged lumens that extended into the dorsal spinal cord and were outlined by apical ZO-1 (Fig. 2C) and Prkci localization (Fig. 2G). By contrast, lumen size and apical protein localization in miR-219 MO-injected embryos treated with cyclopamine were more similar to that of wild-type control embryos (Fig. 2D,H). To quantify these changes, we categorized the dorsal extent of ZO-1 and Prkci lumenal localization as severe, mild and normal. Similar to our previous findings (Hudish et al., 2013), ZO-1 and Prkci lumenal localization extended into the dorsal spinal cords of miR-219 MO-injected embryos, which we categorized as severe phenotypes (Fig. 2R,S). By contrast, miR-219 MO-injected embryos treated with Cyclopamine had milder ZO-1 and Prkci localization phenotypes (Fig. 2R,S). These data raise the possibility that high levels of Hh signaling maintain neuroepithelial characteristics, including apical membrane protein localization, in the dorsal spinal cord.

To test whether or not the excess number of neural progenitors of miR-219-deficient embryos also results from elevated Hh signaling, we performed immunohistochemistry to detect expression of Sox2, a marker of neural progenitors (Ellis et al., 2004). In 3 dpf wild-type embryos, Sox2⁺ cells surrounded the central canal and occupied intermediate positions of the spinal cord, just dorsal to the central canal (Fig. 2I). Wild-type embryos treated with Cyclopamine from 24 hpf had similar numbers of Sox2⁺ cells (Fig. 2J,T). Similar to our previous report (Hudish et al., 2013), miR-219 MO-injected embryos had more Sox2⁺ cells and these were distributed across the dorsoventral axis of the spinal cord (Fig. 2K,T). Cyclopamine-treated miR-219 MO-injected embryos had fewer Sox2⁺ cells compared with untreated miR-219 MO-injected embryos, but slightly more than control and cyclopamine-treated wild-type embryos (Fig. 2L,T). Furthermore, immunohistochemistry to detect phosphohistone H3 (PH3), which reveals cells in mitosis, showed that whereas miR-219 MO-injected embryos had significantly more PH3⁺ cells than control embryos (Fig. 2M-O), cyclopamine-treated miR-219 MO-injected embryos had approximately the same number of PH3⁺ cells as control embryos (Fig. 2P,U). Thus, Hh inhibition suppressed the excess neural progenitor phenotype of miR-219-deficient embryos, consistent with the possibility that miR-219 promotes neural progenitor cell cycle exit and differentiation, at least in part, by dampening Hh signaling.

To further investigate the relationship between miR-219 function and Hh signaling, we tested the prediction that artificially elevating Shh expression would phenocopy the loss of miR-219 function. To do so, we used the transgenic line Tg(hsp701:Shha:EGFP) (Shen et al., 2013), which expresses Shh fused to EGFP under the control of heat-responsive DNA elements, to temporally regulate Hh signaling levels (Shh HS). We allowed the embryos to develop normally until 1 dpf after which they were placed twice daily at a higher temperature until 3 dpf to induce transgene expression. As expected, Shh overexpression resulted in elevated levels of ptch2 mRNA (Fig. 3A) and expansion of the ptch2:Kaede reporter along the dorsoventral spinal cord axis (Fig. 3B-D). At 3 dpf, Sox2⁺ cells lined the ventral central canal of control embryos (Fig. 3E). By contrast, Shh-overexpressing embryos had an excess of Sox2⁺ cells, which lined the entire dorsoventral axis of the medial spinal cord (Fig. 3F). Quantification of the Sox2⁺ cells revealed a twofold increase in Shh-overexpressing embryos when compared with the control (Fig. 3G). Additionally, spinal cord cells of embryos induced to express Shh-EGFP incorporated more EdU, a thymidine analog used to detect cells in S phase of the cell cycle, than control embryos did (Fig. 3H-J). The spinal cord lumens of embryos induced to express Shh-EGFP were enlarged and outlined by apical ZO-1 (Fig. 3K,L) and Prkci localization (Fig. 3M,N), indicating that more cells maintained apical membrane. Thus, similar to loss of miR-219 function, prolonged and elevated Shh expression maintained neuroepithelial characteristics, consistent with the possibility that miR-219 negatively regulates Hh signaling.

Our data are consistent with an interpretation that miR-219 drives cell cycle exit and loss of neural progenitor characteristics by dampening Hh signaling. Previously, we presented evidence the miR-219 negatively regulates the apical Par polarity proteins Pard3 and Prkci (Hudish et al., 2013). Together, these observations raise the possibility that miR-219 suppresses Hh signaling in the neural tube by suppressing Pard3 and Prkci. To investigate this possibility, we blocked access to the miR-219 target site on both pard3 and prkci 3′UTRs. To do this, we used target protector MOs (TP MOs), which bind to the miR-219 target sites on either pard3 or prkci mRNAs (Hudish et al., 2013), rendering miR-219 unable to bind to pard3 or prkci mRNAs. At 3 dpf, pard3 TP MO-injected embryos expressed ptch2 RNA in dorsal spinal cord, in contrast to control embryos where only cells lining the central canal expressed ptch2 RNA (Fig. 4A,B). Although prkci TP MO had little apparent effect on the distribution of ptch2 RNA assessed by RNA in situ hybridization (Fig. 4C), qPCR revealed that embryos injected with pard3 TP MO and prkci TP MO had significantly elevated levels of ptch2 and gli1 transcripts (Fig. 4D). Therefore, blocking interaction of miR-219 with its pard3 or prkci target sites elevated ptch2 and gli1 transcript levels, supporting the hypothesis that maintenance of apical Par proteins promotes Hh signaling levels.

Our data indicate that elevated expression of Pard3 and Prkci in miR-219-deficient embryos elevates Hh signaling in the spinal cord. To investigate whether or not apical Par protein functions are necessary for Hh signaling, we first compared ptch2 RNA expression in wild-type and prkci mutant (Horne-Badovinac et al., 2001) embryos. Cells along the entire dorsoventral axis expressed ptch2 RNA in wild-type embryos at 2 dpf (Fig. 4E), whereas only a few cells located in the ventral spinal cords of 2 dpf prkci mutant embryos expressed ptch2 RNA (Fig. 4F). By 3 dpf, only cells in the ventral spinal cord of wild-type embryos expressed ptch2 RNA (Fig. 4G) and 3 dpf prkci mutant embryos expressed little or no ptch2 RNA (Fig. 4H). Thus, Prkci function is necessary for Hh signaling in the spinal cord, consistent with the possibility that modulation of apical Par protein levels by miR-219 modulates the Hh signaling level. To determine whether or not the elevated Hh signaling of miR-219-deficient embryos results from elevated apical Par protein levels, we injected miR-219 MO into prkci mutant embryos. In contrast to miR-219 MO-injected wild-type embryos, which expressed ptch2 at elevated levels (Fig. 1L), miR-219 MO-injected prkci mutant embryos expressed ptch2 similar to non-injected prkci mutant embryos (Fig. 4I). To further test this hypothesis we assessed transcript levels of ptch2 and gli1 by qPCR. Reduction of miR-219 in prkci mutant embryos had little to no impact on ptch2 or gli1 mRNA levels (Fig. 4J), supporting the hypothesis that miR-219 modulates Hh signaling levels through its regulation of apical Par proteins. Together with the data presented above, these results provide evidence that miR-219 dampens Hh signaling in the zebrafish spinal cord by reducing apical Par protein levels.

Our data support the hypothesis that reduction of apical Par proteins near the end of embryogenesis reduces the Hh signaling response, thereby triggering exit of neural progenitors from the cell cycle. What remains unclear, however, is how apical Par proteins regulate Hh signaling. Because previous data indicated that Par proteins could promote ciliogenesis and primary cilia are known signaling centers for Hh signaling, we predicted that apical Par proteins promote neural tube Shh signaling by regulating ciliogenesis. To test this, we evaluated primary cilia length using transgenic Tg(βact::arl13b-GFP) zebrafish, which express the ciliary protein Arl13b fused to GFP under the control of actb1 (β-actin) regulatory DNA (Borovina et al., 2010). We injected Tg(βact::arl13b-GFP) embryos with miR-219 MO and found that primary cilia of the trunk spinal cord were significantly longer than cilia in control embryos (Fig. 5A,B,D). By contrast, cilia were significantly shorter in prkci mutant embryos than in control embryos (Fig. 5C,D). Are cilia longer in miR-219-deficient embryos because apical Par proteins are elevated? To test this possibility, we injected prkci mutant embryos with miR-219 MO. Cilia length in miR-219- and prkci-deficient embryos was similar to that of uninjected prkci mutant embryos (Fig. 5D), suggesting that miR-219 modulation of cilia length is mediated by apical Par protein function. To test whether or not miR-219 regulates cilia lengthening via its apical Par protein targets, we injected embryos with miR-219 target protector MOs and measured cilia length. Injections of both prkci TP MO and pard TP MO resulted in longer cilia (Fig. 5E), similar to loss of miR-219. These data indicate that miR-219 downregulation of apical Par polarity proteins reduces primary cilia length.

Because the rate or timing of cilia formation relative to cell division might influence Hh signaling, we performed live imaging to directly observe cilia growth over time. To do this, we created the constructs sox19a:arl13b-GFP and sox19a:mCherry-CaaX, which use sox19a regulatory DNA to express cilia and membrane markers in neural progenitors. We co-injected these constructs into newly fertilized zebrafish eggs and screened the embryos at 24 hpf for doubly labeled spinal cord cells, which we then followed using time-lapse confocal microscopy. As cells divided, cilia were retracted (Fig. 5G-I) and then quickly reformed (Fig. 5K). By measuring the rate of cilia regrowth following division, we found that cilia extended more rapidly in miR-219 MO-injected embryos than in control embryos (Fig. 5N). Most of the accelerated growth in miR-219 MO-injected embryos occurred in the first 30 min, raising the possibility that Par proteins promote initiation of ciliogenesis. Consistent with the static cilia length measurements above (Fig. 5A,D), the cilia were also longer in miR-219 MO-injected embryos at the end of the imaging period (Fig. 5N).

Our data provide evidence that downregulation of apical Par proteins by miR-219 reduces cilia length and growth rate, dampens the Hh signaling response and promotes the loss of neural progenitor characteristics. Does modulation of cilia dynamics account for modulation of Hh signaling and cell cycle exit? Mouse embryos lacking cilia are nearly devoid of Hh signaling (Huangfu et al., 2003; May et al., 2005) and mouse mutations that alter cilia formation also disrupt Hh signaling (Gray et al., 2009; Heydeck et al., 2009; Zeng et al., 2010) suggesting that regulation of ciliogenesis could have corresponding effects on Hh signaling. Zebrafish embryos that lack functions of Ift88 or Talpid3 have no primary cilia and are deficient for high-level Hh signaling but have expanded low-level signaling (Ben et al., 2011; Huang and Schier, 2009), suggesting that in zebrafish, cilia promote efficient Hh signaling. To test whether or not primary cilia, independent of apical Par protein functions, are necessary for neural progenitor maintenance, we examined dzip1 mutant embryos, which lack primary cilia and have a Hh signaling profile similar to that of Ift88 and Talpid3-deficient embryos (Wolff et al., 2004; Sekimizu et al., 2004; Kim et al., 2010; Huang and Schier, 2009; Tay et al., 2010). At 3 and 5 dpf, dzip1 mutant animals had similar numbers of Sox2⁺ and PH3⁺ spinal cord cells as the wild type (Fig. 6A-H), indicating that the lack of primary cilia does not cause depletion of neural progenitors. Thus, the low level of Hh signaling in dzip1 mutant embryos is sufficient to maintain neural progenitors. These results raise the possibility that apical Par proteins maintain Hh signaling and neural progenitors via cilia-dependent and -independent mechanisms.

A key element of brain size and complexity is the regulated transition of dividing neural progenitors to differentiated neurons and glia. Neuroepithelial progenitors have apicobasal polarity, which can determine how progenitors respond to extracellular signals by subcellularly localizing signal reception and processing machinery. Here, we provide evidence that miR-219-mediated downregulation of apical Par polarity proteins helps trigger the transition from proliferation to differentiation by diminishing the progenitor cell response to Shh.

Because Shh reception and signal processing occur at apically positioned primary cilia of vertebrate epithelial cells, we reasoned that one way the proliferation to differentiation transition could be controlled is through regulation of apical membrane characteristics. Consistent with this, apical Par proteins localized to and promoted formation of cilia in cultured cells (Fan et al., 2004; Sfakianos et al., 2007) and reduction of Pard3 function reduced the length of photoreceptor cell cilia in zebrafish embryos (Krock and Perkins, 2014). In zebrafish embryos, Prkci and Pard3 localize to the apical membranes of dividing spinal cord progenitors but become depleted as the progenitors differentiate (Hudish et al., 2013; Ravanelli and Appel, 2015). These observations raise the possibility that modulation of the abundance and localization of apical Par protein can tune the Hh signaling response by modulating ciliogenesis or other apical membrane characteristics.

Following our previous work showing that miR-219 promotes differentiation of spinal cord progenitors by downregulating Pard3 and Prkci (Hudish et al., 2013), we began this study by investigating whether or not miR-219 function regulates Hh signaling. Indeed, loss of miR-219 function elevated the Hh signaling response, but only if Par protein functions were intact. Although miR-219 potentially targets many RNAs, the results of our pard3 and prkci target protector MO experiments provide compelling evidence that miR-219-mediated Hh signaling is determined primarily by regulation of Pard3 and Prkci levels. miR-219-deficient embryos maintained excess spinal cord progenitors, but only if Hh signaling was intact, and forced expression of Shh phenocopied miR-219 loss of function. Our data also revealed that inhibition of Hh signaling depleted Prkci from apical membranes of dorsal spinal cord cells, raising the possibility that apical Par proteins and Hh signaling engage in a positive regulatory loop to maintain neural progenitors. Consistent with this possibility, recent data show that Gli1 promotes prkci transcription by directly binding DNA regulatory elements (Atwood et al., 2013). Altogether, our observations suggest that miR-219 breaks the positive feedback loop between apical Par proteins and Hh signaling to initiate neural progenitor cell cycle exit. In the chick neural tube, apical abscission separates apical Par proteins from progenitors prior to neuronal differentiation (Das and Storey, 2014), suggesting that several distinct mechanisms operate to drive neuroepithelial progenitors into a differentiation pathway.

A potential mechanistic link between apical Par protein function and Hh signaling was uncovered by our time-lapse imaging studies of ciliogenesis. In particular, reducing miR-219 function and blocking access of miR-219 to its target sites on pard3 and prkci mRNAs increased the length of spinal cord primary cilia and the rate at which spinal cord progenitors elongated cilia. Thus, longer and faster-forming primary cilia correlated with elevated Hh signaling, consistent with recent observations of cilia length and Hh signaling in cultured fibroblasts and mouse neural tube (Stasiulewicz et al., 2015). By contrast, prkci mutant embryos had shorter cilia, correlating with reduced Hh signaling. These observations suggested the intriguing possibility that cilia dynamics, regulated by apical Par proteins, tunes Hh signaling strength and thereby determines progenitor fate. However, we found that dzip1 mutant embryos, which lack primary cilia (Glazer et al., 2010; Huang and Schier, 2009; Tay et al., 2010), had apparently normal numbers of spinal cord progenitors. Whereas Hh signaling is active in dzip1 mutant embryos, albeit at reduced levels, our data indicate that Hh signaling is substantially reduced in prkci mutant embryos. Thus, the absence of apical Par protein functions might have a more dramatic effect on Hh signaling than the lack of primary cilia does, suggesting that Par protein-mediated modulation of Hh signaling occurs both by modulating cilia dynamics and via cilia-independent mechanisms.

How else might apical Par proteins promote Hh signaling? One possibility is that they do so by regulating Notch signaling, which can maintain neural progenitor fate. In support of this, depletion of Pard3 from developing mouse neocortex reduced Notch signaling and caused dividing cells to undergo neuronal differentiation (Bultje et al., 2009). Pard3 binds Numb (Nishimura and Kaibuchi, 2007), which negatively regulates Notch activity (Knoblich et al., 1995; Rhyu et al., 1994), and Numb and Numb-like are required for Pard3 regulation of Notch in the mouse neocortex (Bultje et al., 2009). Thus, Pard3 might promote Notch signaling and progenitor fate by preventing Numb and Numb-like from blocking Notch activity. Notably, recent data indicate that Notch signaling promotes Hh signaling in the neural tube by promoting localization of Ptch1 and Smo to cilia and cilia length (Kong et al., 2015; Stasiulewicz et al., 2015). Additionally, studies of basal cell carcinomas revealed that Prkci can phosphorylate and activate Gli1, thereby potentiating Hh signaling (Atwood et al., 2013). Similarly, in flies, aPKC, a Prkci homolog, can phosphorylate and activate Smo and the Gli homolog Ci, leading to expression of Hh target genes, including aPKC (Jiang et al., 2014b). Thus, the apical Par polarity protein complex might regulate Hh signaling directly, via Gli1 activation, and indirectly, via modulation of cilia dynamics. The role of cilia in mediating Hh signaling appears to differ in mice and zebrafish in that defective ciliogenesis significantly abrogates Hh signaling in mice (Huangfu et al., 2003; May et al., 2005) whereas it dampens but also expands Hh signaling in zebrafish (Ben et al., 2011; Huang and Schier, 2009). Perhaps the relative roles of direct and indirect modulation of Hh signaling by apical Par proteins differs in mice and zebrafish, accounting for distinct cilia defect-associated phenotypes.

Could miR-219 also regulate Hh signaling by targeting mRNAs that encode pathway components? Numerous studies have shown that various Hh pathway genes are targets of microRNAs in development and cancer (Akhtar et al., 2015; Flynt et al., 2007; Gao et al., 2013a,b; Jiang et al., 2014a; Ketley et al., 2013; Li et al., 2012; Wen et al., 2015; Wu et al., 2012; Yu et al., 2015; Zhang et al., 2015). Target prediction analysis identified a single potential miR-219 binding site within the 3′ UTR sequence of zebrafish smoothened (smo) mRNA, but not in other pathway members. Although we do not yet know whether smo is a functional target, this raises the possibility that miR-219 regulates the Hh signaling response by coordinately regulating transcripts encoding apical Par polarity proteins and Smo.

Altogether, our data suggest that apical Par polarity proteins and Hh signaling engage in positive feedback regulation to maintain neural progenitors in a dividing, undifferentiated state. Initiation of miR-219 expression during the late stages of spinal cord development (Hudish et al., 2013) appears to break the feedback loop cycle by depleting apical Par proteins, which dampens Par protein-dependent Hh signaling activity. Some childhood medulloblastomas express miR-219 at abnormally low levels relative to levels in neural stem cells (Ferretti et al., 2009; Genovesi et al., 2011), raising the possibility that failure to break a Par protein-Hh signaling feedback loop can contribute to tumor formation.

The University of Colorado Anschutz Medical Campus Institutional Animal Care and Use Committee approved all zebrafish studies. Embryos were produced by pair-wise mating and kept at 28.5°C in egg water or embryo medium (15 mM NaCl, 0.5 mM KCl, 1 mM CaCl₂, 1 mM MgSO₄, 0.15 mM KH₂PO₄, 0.05 mM NH₂PO₄, 0.7 mM NaHCO₃). Embryos were staged to hpf or days post-fertilization (dpf) according to established zebrafish guidelines (Kimmel et al., 1995). The experiments conducted in this study used the following strains of zebrafish: AB, Tg(hsp701:Shha:EGFP) (Shen et al., 2013) and Tg(βact::Arl13b-GFP) (Borovina et al., 2010), dzip^+/− (iguana^ts294e) (Sekimizu et al., 2004), prkci^≠/− (has^m567) (Horne-Badovinac et al., 2001) and Tg(ptch2:Kaede) (Huang et al., 2012). Animals of either sex were used.

Embryos and larvae were fixed in 4% paraformaldehyde (PFA) overnight at 4°C. Fixed embryos were embedded in 1.5% agar with 5% sucrose and transferred to a 30% sucrose solution in scintillation vials and incubated at 4°C overnight. The blocks were then frozen and cut into 10-15 µm sections using a cryostat microtome. The sections were incubated with the following primary antibodies: rabbit anti-phosphohistone H3 (1:1000, 06-570, Millipore), rabbit anti-PkC (1:200, sc-216, Santa Cruz Biotechnology), mouse anti–ZO-1 (1:200, 33-9100, Invitrogen), rabbit anti-Sox2 (1:500, ab97959, Abcam). For fluorescent detection of antibody labeling, we used Alexa Fluor 568 and Alexa Fluor 647 goat anti-mouse and goat anti-rabbit secondary antibodies (1:200, Invitrogen). To detect EdU incorporation, we incubated the slides in 250 µl EdU Detection Reaction mix (Invitrogen) for 40 min at room temperature. Images were collected on a Zeiss Axio Observer microscope equipped with a PerkinElmer UltraVIEW VoX spinning disk confocal system and Volocity imaging software (PerkinElmer). Images were contrast enhanced using either Volocity or Photoshop (Adobe CS4).

Embryos and larvae were fixed in 4% PFA overnight at 4°C and stored in methanol at −20°C. After linearizing plasmids with the appropriate restriction enzymes, antisense cRNA was transcribed using Roche digoxigenin-labeling reagents and T3, T7 or SP6 RNA polymerases (New England Biolabs). After processing embryos for in situ RNA hybridization, they were embedded in agar and sectioned as described above. Sections were rehydrated in 1× PBS for 30 min then covered with 75% glycerol. Images were obtained on a Zeiss Axio Observer microscope equipped with DIC optics, a Retiga Exi color camera and Volocity imaging software. Some images were contrast enhanced using Photoshop (Adobe CS4).

Antisense morpholino oligonucleotides were purchased from Gene Tools, LLC. These included: miR-219 MO (5′-CAAGAATTGCGTTTGGACAATCA-3′) (Zhao et al., 2010), pard3 TP MO (5′-CTGATTGTCAGAGCATCTCTACTAC-3′), control pard3 TP MO (5′-ACAGAGTCAAAGTGACGGACTCC-3′) and prkci TP MO (AAGCGACCGTCACACACTCCTCCGC) (Hudish et al., 2013). Morpholino oligonucleotides were dissolved in water to create stock solutions of 1 mM and diluted in 2× injection buffer (5 mg ml⁻¹ Phenol Red, 40 mM HEPES and 240 mM KCl) to create a working injection concentration of 0.25 mM. All morpholinos were co-injected with 0.09 mM dose of p53 MO. We injected 1-2 nl into the yolk just below the single cell of fertilized embryos. All morpholino oligonucleotide injected embryos were raised in embryo medium at 28.5°C. We previously validated the miR-219 MO using pard3 and prkci 3′ UTR reporter assays and by showing that miR-219 MO phenotypes are dependent on Pard3 and Prkci functions (Hudish et al., 2013).

Neural progenitors were labeled using a conserved regulatory DNA element near zebrafish sox19a, which is expressed by neural cells (Okuda et al., 2006). The 1554 bp sox19a 3′ regulatory element was amplified using the primers AscI fwd 5′-ggcgcgccTGTAAGCACACGGCCATTTA-3′ and FseI rev 5′-aaaggccggccTCACCTAACCCCGAGAAGAA-3′ and cloned into the Tol2 Gateway entry vector p5E-FAbasal using AscI/FseI (NEB) restriction enzyme sites as described previously (Kwan et al., 2007). Four-part Gateway cloning was used to construct the final vector pEXPR-Tol2-sox19a:mCherry-CaaXpA-pA2 using LR Clonase II Plus Enzyme Mix (Invitrogen 12538-120) and pME-mCherry-CaaX, p3E-polyA, and pDEST-Tol2-pA2 vectors as described previously (Kwan et al., 2007). The pME-Arl13b-EGFP vector was created by subcloning Arl13b-EGFP from a pCS2-Arl13b-EGFP plasmid (gift from Brian Ciruna, The Hospital for Sick Children, Toronto, Ontario, Canada) by traditional cloning using the PstI and BamHI restriction sites. Multisite Gateway cloning was used to create pEXPR-Tol2-sox19a:Arl13b-EGFP-pA2 as described above. Embryos were injected at the single-cell stage with 1 nl of 50 ng/µl plasmid DNA in 0.2 M KCl and sorted for GFP⁺/mCherry-CaaX⁺ neural progenitors at 24 hpf.

Dechorionated embryos were labeled with 5-ethynyl-2′-deoxyuridine (EdU). Embryos were incubated in 2 mM EdU (Click-iT EdU AlexaFluor 555 detection kit; Invitrogen, c10338) in embryo medium with 10% DMSO for 30 min at room temperature. The fish were then fixed in 4% PFA in PBS with 116 mM sucrose and 150 µM CaCl₂ at 4°C overnight.

Tg(hsp701:Shha:EGFP) and non-transgenic control embryos were placed in a 50 ml beaker in approximately 10 ml of embryo medium and immersed in a 38°C water bath for 30 min after which they were allowed to recover at RT for 20 min and were placed back in the incubator afterwards. Heat shock was performed at 24, 32, 48 and 56 hpf to keep transgene expression high. Embryos were fixed at 72 hpf and processed for immunohistochemistry as described above.

Tg(hsp701:Shha:EGFP);Tg(ptch2:Kaede) embryos were heat shocked as described above. At 3 dpf, embryos were mounted in low-melting temperature agarose and illuminated with UV from a mercury bulb source for 1 min per embryo. Immediately afterward, photoconversion was confirmed by assessing both red and green fluorescence. z-stack images were obtained of the trunk spinal cord above the yolk extension using a 20× objective mounted on the confocal microscope described above. The fluorescence distribution of the ptch2:Kaede reporter was analyzed using ImageJ/Fiji. Briefly, maximum intensity projections for each spinal cord were manually straightened and aligned along the bottom edge of the major strip of Kaede fluorescence intensity in the ventral spinal cord. Next, each image was background subtracted and normalized to the maximum signal within the major strip of Kaede intensity to generate 8-bit images displaying the same intensity range. Finally, a 500 pixel wide (149 µm) line-scan was generated along the dorsoventral axis of each image. For each pixel position along the linescan, the average pixel intensity and the standard error of the mean were plotted for all averaged spinal cords within an experimental condition. The average stack projector plug-in was used to generate the average images, which were displayed using the ‘physics’ look-up table. All images are displayed within the range of 0-150 arbitrary intensity units.

RNA was isolated from 15-20 pooled larvae for each control or experimental condition. Samples for each condition were collected in triplicate. Reverse transcription was performed using the iScript Reverse Transcriptase Supermix for RT-PCR (Bio-Rad, 1708840). Real-time qPCR was performed in triplicate for each cDNA sample using an Applied Biosystems StepOne Plus machine and software v.2.1. TaqMan Assays were used to detect patched2 (Dr03118687_m1), gli1 (Dr03093669_m1) and endogenous control rpl13a (Dr03101115_g1).

No animals were excluded from analysis. For some experiments, only a subset of animals was used for quantification, but these were chosen at random. Animals were randomly assigned to control and experimental groups. Cell counts were obtained by direct observation of sections using the microscopes described above. For Sox2, EdU and PH-3, 10 sections per embryo from 15 embryos per group with two or three replicates were counted to produce the average number per section. P values were generated using an unpaired t-test using GraphPad Prism software.

All measurements were made using spinal cord primary cilia and not the longer motile cilia of the floor plate. Time-lapse imaging was performed on control and miR-219-injected embryos with frames captured at least every 17 min, for a total of 120 min. For both static and live cilia length measurements, primary cilia length was measured on maximum intensity projections in Volocity or in ImageJ using a custom script to automate use of the polyline tool. For live measurements, length was measured at every point after a cell divided, except for cases where the cilium reoriented along the z-axis, which is difficult to measure accurately. The cilia length was binned into 17 min intervals and averaged. The rate of cilia formation was calculated between each binned time interval and significance was assessed with an unpaired t-test.

Cyclopamine was reconstituted in 100% ethanol at a stock concentration of 10 mM and diluted fresh for every treatment. Approximately 20-25 embryos were placed in glass dishes and treated with 50 µM cyclopamine (Cayman Chemicals) or 0.5% ethanol starting at 24 hpf. Embryos were placed in the incubator in the cyclopamine or control solutions until 3 dpf at which point they were fixed and analyzed.

We thank members of the Appel laboratory for valuable discussions and Christina Kearns for technical assistance. We are grateful for the gifts of the transgenic line Tg(hsp701:Shha:EGFP) from Rolf Karlstrom, the Tg(βact::Arl13b-GFP) transgenic line from Thomas Schilling and the Arl13b-GFP plasmid from Brian Ciruna.