Inhibition of microRNA suppression of Dishevelled results in Wnt pathway-associated developmental defects in sea urchin

Dishevelled is a key regulator in the highly conserved Wnt signaling pathway and plays the paramount role of directing signaling towards the different branches of Wnt signaling cascades (Gao and Chen, 2010). Components of the Wnt signaling pathways are expressed during early stages of embryonic development to specify cell fates, direct morphogenic movements, and maintain neuronal stem cells along with many other important functions (Gao and Chen, 2010; De, 2011). Upon Wnt ligand binding to the Frizzled (Fz) receptor, Dvl is recruited to the plasma membrane as it binds to Fz. Depending on which proteins interact with Dvl, it can activate either the canonical Wnt/β-catenin (cWnt) signaling cascades important for cell specification, the non-canonical Wnt planar cell polarity (ncWnt/PCP) or the non-canonical Wnt calcium pathway (ncWnt/Ca²⁺) to control cellular morphogenesis (Gao and Chen, 2010; Song et al., 2015).

The cWnt signaling pathway is a highly conserved mechanism for germ layer specification in all metazoans (Moon, 2005; Komiya and Habas, 2008). Dvl has been shown to be required for animal-vegetal axial patterning, β-catenin stabilization, and endodermal specification in bilaterians, including echinoderms and chordates, and non-bilaterians such as the starlet sea anemone in the Cnidaria phylum (Lee et al., 2007; Kumburegama et al., 2011). Previously, it has been shown that both Dvl (Weitzel et al., 2004; Peng and Wikramanayake, 2013) and β-catenin (Logan et al., 1999) localize to the vegetal pole of the sea urchin embryo, resulting in activation of endodermal and mesodermal cell specification.

The ncWnt pathways are β-catenin independent and consist of the ncWnt/PCP and the ncWnt/Ca²⁺ pathways. The ncWnt/PCP pathway regulates polarization of epithelial cells, enabling cell morphogenesis and tissue patterning. Mutation of ncWnt/PCP genes in vertebrates results in a broader and shorter body axis and neural tube closure defects due to blockage of cell movements (Simons and Mlodzik, 2008; Yin et al., 2009; Hayes et al., 2013). In Drosophila, PCP signaling is necessary for orientation of epithelial structures in the cuticle, wing and eye (Gubb and Garcia-Bellido, 1982; Adler, 2002; Sepich et al., 2011). The PCP pathway also consists of multiple genes required for ciliogenesis (Dale et al., 2009).

The ncWnt/Ca²⁺ pathway, mediated by G-protein signaling, stimulates release of intracellular Ca²⁺ and activation of protein kinase C to control actin polymerization (Gao and Chen, 2010). In Xenopus, embryos were found to have elevated Ca²⁺/CaMKII levels in the ventral side of the embryo (Kühl et al., 2000). In zebrafish, complete blockade of ncWnt/Ca²⁺ leads to a variety of developmental defects, such as gut abnormalities and tumor formation (Yoshida et al., 2004; De, 2011).

The Dvl protein has three highly conserved domains: an N-terminal DIX (Dishevelled, Axin) domain, a central PDZ (Postsynaptic density 95, Discs Large, Zonula occludens-1) domain, and a C-terminal DEP (Dvl, Egl-10, Pleckstrin) domain (Gao and Chen, 2010). In general, the DIX domain of Dvl is important for Wnt/β-catenin signaling, and the PDZ and DEP domains are both required for ncWnt/PCP and ncWnt/Ca²⁺ pathways (Axelrod et al., 1998; Tada and Smith, 2000). In addition, Dvl-DEP and the C-terminal region (DEP-C) interact with three discontinuous regions of Fz. DEP-C binding to Fz stabilizes the binding of the Dvl-DEP domain to Fz, which is required to drive cWnt/β-catenin activation in Xenopus and in HEK293T cells (Tauriello et al., 2012). Overexpression of Dvl-DIX in the sea urchin resulted in a lack of nuclear localization of β-catenin (Weitzel et al., 2004), in Drosophila resulted in a lack of β-catenin/Wnt activity (Axelrod et al., 1998), and in Nematostella resulted in a lack of archenteron and endodermal epithelium (Kumburegama et al., 2011). In Xenopus, overexpression of truncated Dvl possessing only PDZ and DEP domains relayed Wnt signaling predominantly via the ncWnt/Ca²⁺ cascade, as seen in elevated levels of intracellular Ca²⁺ and activation of Ca²⁺-dependent enzymes, such as PKC and Calcineurin (Komiya and Habas, 2008). Overexpression of Dvl-DEP in sea urchin, which blocks the ncWnt pathway through a dominant-negative mechanism, resulted in failed archenteron invagination (Byrum et al., 2009). Thus, depending on its interacting proteins, Dvl can direct various branches of the Wnt signaling pathways.

Relatively little is known of the impact of post-transcriptional regulation of Dvl on Wnt signaling pathways (He et al., 2015; Huang et al., 2018). MicroRNAs (miRNAs) are small non-coding RNAs that regulate post-transcriptional gene expression by binding to the 3′ untranslated region (3′UTR) of target mRNAs to repress their translation and/or induce mRNA degradation (Bartel, 2009). Similar to vertebrates, each sea urchin miRNA has many predicted targets (Song et al., 2012; Stepicheva et al., 2015; Stepicheva and Song, 2015). Previous studies indicated that the sea urchin embryo contains approximately 50 miRNAs, of which SpmiR-31 has been identified to target multiple genes important for primary mesenchyme cell (PMC) development and function (Song et al., 2012; Stepicheva and Song, 2015). Additionally, SpmiRDeep2-30364, and SpmiR2007 have been shown to suppress β-catenin (Stepicheva et al., 2015). The overarching hypothesis of the current study is that miRNAs modulate components of the Wnt signaling pathways to provide crucial regulation that impacts cell functions and embryonic structures. Specifically, this study examines miRNA regulation of Dvl. We found that the sea urchin embryo harbors four isoforms of Dvl, three of which have unique 3′UTRs. We demonstrated that miRNAs directly suppress Dvl5a and Dvl4a. By removal of miRNA suppression of these Dvl isoforms, we observed dose-dependent defects in gut morphology, skeletal length and PMC patterning. In addition, we observed profound ciliary defects in the Dvl miRNATP-injected larvae that resulted in swimming defects. Importantly, overexpression of Dvl isoforms recapitulated the Dvl miRNATP-induced phenotypes. These results indicate that miRNAs modulate the expression of Dvl isoforms thereby providing additional regulation to ensure proper embryonic development.

Four different isoforms of Dvl were identified from a previous RNA-Seq experiment in an effort to identify vegetal cortex-enriched mRNAs (L.W., W.W. and A.W., unpublished). All isoforms of Stronylocentrotus purpuratus Dvl (SpDvl5a, 1, 4a and 4b) share the exact same protein sequence except for the last exon. Only Dvl5a contains the highly conserved DEP-C domain (AMGNPSEFFVDVM) that is essential for interacting with Fz and its own Dvl-DEP domain to drive the cWnt/β-catenin pathway (Fig. 1A) (Tauriello et al., 2012). Dvl from another sea urchin species, Lytechinus variegatus, LvDvl, and SpDvl4a/4b have C-terminal amino acid residues unique to the sea urchin species (YFDDSSVTLL) (Fig. 1A) (Weitzel et al., 2004). Dvl1 lacks the last exon and contains a unique three amino acid sequence (NPS) in the most C-terminal region. Despite close identity of their protein sequences, these SpDvl isoforms have three unique 3′UTRs (Fig. 1B). Dvl4a differs from Dvl4b only in their 3′UTRs, where miRNA regulation typically occurs (Selbach et al., 2008). Dvl1 has the same 3′UTR as Dvl4a. Potential mRNA regulatory sites were bioinformatically identified by searching for inverse complementary miRNA seed sequences within the Dvl 3′UTRs (Song et al., 2012; Stepicheva et al., 2015; Stepicheva and Song, 2015). The precise regulation and function of each of these Dvl isoforms are not known.

We tested the expression of Dvl5a, Dvl4a/1 and Dvl4b isoforms, using primers designed against their unique 3′UTRs. It was not possible to design Dvl1-specific primers owing to its high similarity with Dvl4a (Fig. S1). The expression of these Dvl isoforms were assayed at various developmental stages [0, 6, 24, 30 and 72 h post-fertilization (hpf)], using real-time, quantitative PCR (QPCR). We observed that all Dvl transcripts were expressed throughout development and increased steadily from the egg stage up until the gastrula stage. At the larval stage, the expression of all Dvl isoforms decreased (normalized to that of egg) (Fig. 2A).

To examine the temporal and spatial expression of Dvl isoforms in early development, we constructed RNA in situ probes complementary to distinct 3′UTRs from Dvl5a, Dvl4a/Dvl1 and Dvl4b isoforms. Note that the in situ probe against Dvl4a will also recognize Dvl1 (Fig. S1). We observed that all Dvl isoforms were ubiquitously expressed during early stages of development. Interestingly, Dvl4a/Dvl1 is enriched in the ciliary band of the larval stage (Fig. 2B).

We also examined the localization of Dvl protein at blastula, gastrula and plutei stages, using a pan-Dvl antibody (Peng and Wikramanayake, 2013). At the blastula and gastrula stages, Dvl was ubiquitously expressed and enriched in the posterior of the embryo. Dvl was also enriched in the gut and PMCs of gastrulae. At the larval stage, Dvl protein was enriched in the gut, pyloric sphincter, PMCs and the ciliary band (Fig. 2C).

We bioinformatically identified 100% seed match of annotated sea urchin miRNAs within the 3′UTRs of all Dvl isoforms (Fig. 1B, Fig. 3A). To test the direct regulation of Dvl5a, Dvl4a and Dvl4b by these miRNAs, we cloned Dvl 3′UTRs downstream of Renilla luciferase (Rluc). Site-directed mutagenesis was used to alter the third and fifth base pairs of the seed sequences of SpmiRDeep2-30364 and SpmiR153* in Dvl5a; SpmiRDeep2-30364, SpmiR2007 and SpmiR2002 in Dvl4a; and SpmiR200 in Dvl4b (Fig. 3A). Mutated seed sites would abolish endogenous miRNA binding to the target sites in the 3′UTR of these reporter constructs (Staton and Giraldez, 2011; Stepicheva et al., 2015). In vitro-transcribed mRNAs of Rluc fused to the wild-type or mutated Dvl 3′UTRs and the firefly reporter construct (used as a normalization control for Rluc luciferase) were co-injected into newly fertilized eggs (Stepicheva et al., 2015; Stepicheva and Song, 2015).

We collected injected mesenchyme blastulae and used dual luciferase assays to test direct miRNA suppression of these Dvl isoforms (Fig. 3B). The Rluc with mutated SpmiRDeep2-30364 and SpmiR153* seed sequences resulted in a significant increase of normalized luciferase signals compared with the Rluc with wild-type seed sequences, indicating that at least one of these miRNAs directly suppress Dvl5a (Fig. 3B). For Dvl4a, Rluc fused with the 3′UTR with mutated seed sites for SpmiRDeep2-30364, SpmiR2007 and SpmiR2002 had significantly more luciferase signals compared with the Rluc fused with the wild-type 3′UTR, indicating that Dvl4a is directly suppressed by at least one of these miRNAs. We observed increased but not statistically significant Rluc signals from the Dvl4b construct with mutated SpmiR200 seed sequence compared with the Dvl4b construct with wild-type 3′UTR, indicating that this miRNA may not be bona fide or plays a weak regulatory role. The overall relatively small increase in luciferase reading caused by the removal of miRNA suppression is consistent with the previous studies (Selbach et al., 2008; Nicolas, 2011; Stepicheva et al., 2015). These results indicate that Dvl5a and Dvl4a are directly suppressed by miRNAs.

To test the impact of miRNA suppression of Dvl isoforms, we designed Dvl miRNATPs complementary to the miRNA regulatory binding sites of Dvl5a (SpmiR153* at +230 bp), Dvl4a (SpmiRDeep2-30364 and SpmiR2002) and Dvl4b (SpmiR200), as identified by bioinformatics analyses and luciferase assays (Fig. 1B, Fig. 3A). Zygotes were injected with a cocktail of three Dvl miRNATPs, each at 300 µM, or control TP at 900 µM. We then examined whether there were changes in the Dvl transcript and Dvl protein levels. Results indicated that Dvl miRNATP-injected embryos express Dvl isoforms and other components of the Wnt signaling pathway at similar levels compared with the controls (Fig. 4A). However, the transcript level of Rac1, which encodes a small GTPase that is downstream of the ncWnt/PCP pathway (Lindqvist et al., 2010), was significantly increased in Dvl miRNATP-injected embryos compared with control (Fig. 4A).

To test if blocking miRNA suppression of Dvl translation resulted in a change in Dvl, we immunolabeled control TP- and Dvl miRNATP-injected embryos at the 32-cell stage, when Dvl protein is enriched at the vegetal cortex (Peng and Wikramanayake, 2013). We took confocal z-stack images of these embryos and measured the total level of fluorescence pixels from Dvl immunolabeling. Results indicated that the Dvl miRNATP-injected embryos displayed an overall 1.5-fold increase of Dvl compared with the control TP-injected embryos (Fig. 4B,C). Interestingly, removal of specific miRNA suppression of Dvl isoforms resulted in an increase of ubiquitous Dvl protein throughout the embryo, as well as increased Dvl in the vegetal cortex.

Because the cWnt pathway is involved in endomesoderm specification, we examined the effect of blocking miRNA suppression of Dvl on the embryonic gut (Fig. 5). The sea urchin embryonic gut consists of a muscular esophagus (foregut), a large stomach (midgut) and a tubular intestine (hindgut) (Burke, 1981; Burke and Alvarez, 1988). Newly fertilized eggs were microinjected with Dvl miRNATPs against all three Dvl isoforms at various concentrations. We used the Endo1 antibody, which recognizes antigens of the mid and hindgut (Wessel and McClay, 1985), to examine the epithelial cell lining of the gut. Results indicated that 58% of Dvl miRNATP-injected embryos expressed less Endo1 in the gut epithelium (the less severe phenotype) and 16% of the embryo failed to gastrulate (the more severe phenotype) (Fig. 5A).

Because blocking miRNA suppression of Dvl isoforms resulted in increased Dvl and various aberrant phenotypes, we overexpressed sea urchin Dvl isoforms to identify which phenotypes could be attributed to which Dvl isoform. We utilized the previously studied LvDvl, which has the sea urchin-specific C-terminus residues (YFDDSVSVTLL), to examine the effect of SpDvl4a/4b overexpression. The LvDvl was modified to contain the conserved DEP-C residues that are shared by Dvl of other species, which we denoted as LvDvl*, to resemble the SpDvl5a form (CEFFVDVM). These Dvl mRNAs were injected into zygotes. We found that overexpression of each isoform alone or in combination resulted in reduced Endo1 expression, in comparison with mCherry mRNA-injected control embryos (Fig. 5B), similar to the less severe gut phenotype of the Dvl miRNATP-injected embryos (Fig. 5A).

To evaluate gut morphology of the less severe gut phenotype of the Dvl miRNATP-injected embryos, we measured the width of the midgut and blastopore and calculated their ratios. Results indicated that in the highest dose of 300 µM, the width of the midgut of Dvl miRNATP-injected embryos was significantly narrower and the blastopore was wider than the embryos injected with the control TP (Fig. 5C). Results indicated that overexpression with LvDvl/SpDvl4a/4b, LvDvl*/SpDvl5a, or LvDvl/SpDvl4a/4b plus LvDvl*/SpDvl5a, led to significantly wider blastopore width and narrower midgut, similar to the Dvl miRNATP-injected embryos (Fig. 5D).

Because we observed gut morphological differences in Dvl miRNATP-treated embryos, we tested whether endodermal differentiation was affected by examining expression of alkaline phosphatase (AP), which is an enzyme in the gut that is used as a differentiated endodermal marker (Fig. 5E) (Kumano and Nishida, 1998; Drawbridge, 2003). A significant number of Dvl miRNATP-treated larvae had little or no alkaline phosphatase activity present in the intestine (hindgut) compared with the control (Fig. 5E). These results suggest that removing suppression of Dvl miRNAs resulted in Dvl protein increase that may negatively affect the function of the larval gut.

We investigated what might be the underlying molecular cause of these phenotypes in Dvl miRNATP-injected embryos by examining the spatial and quantitative expression of key transcription factors involved in endodermal specification. Krl and FoxA are expressed in the Veg2 cell lineage important for specifying the fore/midgut, and Eve and Bra are expressed in the Veg1 cell lineage important for hindgut specification (Peter and Davidson, 2011). The overall transcript levels of Krl, FoxA, Eve and Bra did not change in Dvl miRNATP- or combined Dvl mRNA (LvDvl/SpDvl4a/4b plus LvDvl*/SpDvl5a)-injected embryos compared with the control (Fig. 6A). However, the spatial expression domain of Krl was significantly decreased in Dvl miRNATP- or Dvl mRNA-injected embryos compared with controls (Fig. 6C). In addition, Eve exhibited a significant expansion of its spatial expression domain in Dvl miRNATP- or Dvl mRNA-injected embryos compared with controls (Fig. 6D). The spatial expression changes of key transcription factors involved in endodermal specification could result in changes in the gut morphology and potentially function of the gut.

To test whether the removal of miRNA suppression of Dvl had an impact on skeletogenesis, newly fertilized eggs were microinjected with Dvl miRNATPs against all three Dvl isoforms at various concentrations. We observed a dose-dependent decrease of the dorsoventral connecting rods (DVCs) at gastrula stage (Fig. 7A). Injection of Dvl miRNATP at 300 µM resulted in the most significant decrease of DVCs. In addition, results indicated that overexpression of LvDvl/SpDvl4a/4b, LvDvl*/SpDvl5a, or both together, resulted in a significant decrease in DVCs compared with the control, similar to Dvl miRNATP-injected embryos (Fig. 7B).

We also tested the PMC positioning in control TP- and Dvl miRNATP-injected embryos. We observed that embryos injected with 300 µM of Dvl miRNATP had the most severe PMC defects with clustered PMCs at the sub-equatorial ring and a complete lack of anterior migration (Fig. 7C). The lower concentrations of Dvl miRNATP also resulted in some embryos with clustered PMCs at the sub-equatorial ring, but PMCs were able to partially migrate anteriorly. Similarly, LvDvl/SpDvl4a/4b or LvDvl/SpDvl4a/4b plus LvDvl*/SpDvl5a co-injection resulted in scattered and clustered PMCs (Fig. 7D). No significant PMC patterning defects were observed in embryos injected with the LvDvl*/SpDvl5a or in the mCherry-LifeAct mRNA-injected embryos, indicating that the LvDvl/SpDvl4a/4b isoform overexpression resulted in PMC patterning defects (Fig. 7D).

Importantly, the PMC patterning defects persisted into the larval stage 5 days post-fertilization (dpf), when the Dvl miRNATP-injected and Dvl mRNA (LvDvl/SpDvl4a/4b plus LvDvl*/SpDvl5a)-injected larvae exhibited clustered PMCs and branched filopodia (Fig. 7E). Interestingly, we found that Dvl is expressed in the PMCs, suggesting that PMCs themselves may be responsive to Wnt signaling activation (Fig. 7F).

To investigate further whether the failure of PMCs to migrate anteriorly might be due to defective motility, we co-injected mCherry-LifeAct mRNA with control TPs or Dvl miRNATPs into newly fertilized eggs to examine actin distribution. We observed that presumptive PMCs around the tri-radiate spicules of Dvl miRNATP-injected embryos have punctate actin localization, whereas the presumptive PMCs along the long skeletal spicules in the control TP-injected embryos had more even actin distribution within these cells (Fig. 7G). These results suggest that possible actin-remodeling defects might hinder proper PMC migration.

To identify the underlying molecular mechanism that led to Dvl miRNATP- and Dvl mRNA-induced skeletal defects, we examined expressions of the following gene categories: (1) Alx1, which is essential for PMC specification (Ettensohn et al., 2003); (2) biomineralization genes SM29, SM49, SM50 (Adomako-Ankomah and Ettensohn, 2013); (3) Hnf6, which regulates SM50 (Otim et al., 2004; Otim, 2017); (4) factors that have been shown to be important for PMC positioning, including Nodal, BMP2/4 (Yaguchi et al., 2008; Duboc et al., 2010), Pax2/5/8 (Rottinger et al., 2008), Vegf3 (Duloquin et al., 2007; Adomako-Ankomah and Ettensohn, 2013), Alk2/4/7, Slc26a5 and TGF-βrtII (Piacentino et al., 2015, 2016; Sun and Ettensohn, 2017); and (5) Kirrel, which is involved in PMC fusion (Ettensohn and Dey, 2017) (Fig. 8A). Results indicated no significant expression difference for most genes tested, except for Vegf3, which had 50% of the transcripts in both Dvl miRNATP- and Dvl mRNA-injected embryos compared with control embryos. Previous literature indicated that the Vegf3 signaling pathway provides guidance and differentiation cues to PMCs, as well as being crucial for forming skeletal elements (Duloquin et al., 2007; Adomako-Ankomah and Ettensohn, 2013).

We also examined the spatial expression of Wnt5 and Vegf3. Wnt5 acts as a short-range signal from the endoderm that activates the border ectoderm specification, which provides positioning cues for the PMCs (McIntyre et al., 2013). Results indicated that Dvl miRNATP-injected embryos did not have a significant change in the Wnt5 expression domain, compared with the control TP-injected embryos. The spatial expression domain of Vegf3 was significantly decreased in the Dvl miRNATP-injected or Dvl mRNA-injected embryos compared with controls (Fig. 8C). The expression domain of VegfR10, which encodes the cognate receptor for Vegf3, was unchanged and expressed in all PMCs in the Dvl miRNATP-injected, Dvl mRNA-injected, and control embryos (Fig. 8D).

We noticed that Dvl miRNATP- and Dvl mRNA-injected larvae had swimming defects compared with the controls. The ciliary band arises after gastrulation and the ncWnt/PCP pathway has been implicated in regulation of sea urchin ciliogenesis. Cilia are made up of microtubules consisting of tubulin subunits that enable the embryo to swim and feed (Yaguchi et al., 2010; Burke et al., 2014). To examine this swimming defect, we immunolabeled larvae with β-tubulin antibody to visualize cilia. The control TP-injected embryos had long, well projected cilia, whereas the Dvl miRNATP-injected larvae displayed aberrant apical and body cilia (Fig. 9A-C). The body cilia of Dvl miRNATP-injected embryos had kinks, whereas control TP-injected embryos cilia were long and curved (Fig. 9B). The apical cilia of the ciliary band in the Dvl miRNATP-injected larvae tend to be short, bent, sparse and disorganized compared with controls (Fig. 9A,C). To ensure that the aberrant morphology of the cilia in Dvl miRNATP-injected embryos was not due to fixation artifacts, we also examined cilia in living larvae. We observed that both apical and body cilia in Dvl mRNATP-injected embryos were bent with irregular beating compared with the control TP-injected embryos, which had long cilia with synchronous beating (Movies 1-4).

Similar to the Dvl miRNATP-injected embryos, overexpression of LvDvl/SpDvl4a/4b, LvDvl*/SpDvl5a, or both, resulted in bent ciliary defects compared with the long, protruding cilia of mCherry mRNA-injected control larvae (Fig. 9D, Movies 5-8).

Dishevelled is an important protein in the Wnt signaling pathway that plays a significant role in relaying cellular information to various developmental pathways (Gao and Chen, 2010). We identified several miRNAs that suppress different isoforms of Dvl and revealed that post-transcriptional regulation of Dvl is crucial for proper development. Furthermore, blockade of miRNA suppression of Dvl isoforms induced defects in the gut, spiculogenesis and ciliogenesis. These developmental defects of Dvl miRNATP-injected embryos are mimicked by the overexpression of Dvl isoforms. Developmental defects in Dvl miRNATP- and Dvl mRNA-injected embryos are likely to be due to perturbation of Wnt signaling pathways, as well as other Dvl functions in the embryo.

Dvl is ubiquitously expressed and highly enriched at the vegetal pole of early embryos (Peng and Wikramanayake, 2013). By blocking miRNA suppression of Dvl isoforms, we observed a 1.5-fold increase of the Dvl protein (Fig. 4B,C). This may be an underestimate, because we did not block all potential miRNA regulatory sites. Of note is that each Dvl miRNATP is complementary to the specific miRNA's seed sequence (6 bp) and the unique Dvl 3′UTR sequences flanking the miRNA seed sequence (additional 19 bp). Thus, each Dvl miRNATP uniquely targets that specific miRNA seed site within the particular Dvl isoform. We found that Dvl is increased in the entire Dvl miRNATP-injected 32-cell embryos with enhanced enrichment in the vegetal pole of the embryo compared with control TP-injected embryos. This suggests that the miRNAs we identified play a role in suppressing Dvl throughout the embryo. Furthermore, the increase in Dvl protein induced by Dvl miRNATP was dose dependent, indicating specificity of the miRNATPs (Fig. 4B,C, Fig. S2A).

All miRNAs that are predicted to bind to Dvl isoforms are present throughout development (Fig. S3). Interestingly, miRDeep2-30364, one of the miRNAs that may suppress sea urchin Dvl5a and Dvl4a, also suppresses the sea urchin β-catenin (Stepicheva et al., 2015). Previously, we have shown that miRDeep2-30364 and miR2007 directly suppressed β-catenin and resulted in increased gene expression of downstream cWnt-responsive transcription factors, including Eve, Bra, Krl and FoxA (Stepicheva et al., 2015). This is typical of miRNA function as individual miRNAs often target multiple genes in the same developmental pathway (Stepicheva and Song, 2015).

Because we introduced Dvl miRNATP against at least one miRNA in all Dvl isoforms, we expected that elevation of Dvl protein would lead to the disassembly of the destruction complex and increase levels of β-catenin, which is a key effector protein of the cWnt pathway. β-Catenin enters the nuclei of vegetal blastomeres where it regulates endomesoderm specification (Weitzel et al., 2004). Our laboratory previously found that blocking miRNA suppression of β-catenin resulted in increased β-catenin and induced aberrant gut morphology (Stepicheva et al., 2015). Similarly, here we observed that Dvl miRNATP- or Dvl mRNA-injected embryos had significantly narrower midguts compared with control embryos (Fig. 5A,B). This is consistent with our observation that Dvl is localized to the gut and in the pyloric sphincter of larvae (Figs 2 and 9).

Whereas cWnt is important for gut development, ncWnt may contribute to the wide blastopore phenotype in the Dvl miRNATP- or Dvl mRNA-injected embryos. In Xenopus, manipulations of Dvl through injections of dominant-negative mutant Xdsh (Xdd1), which has an internal deletion of the conserved PDZ domain, led to disruptions of ncWnt/PCP genes and resulted in failure of blastopore to close (Sokol, 1996; Ewald et al., 2004). Also in Xenopus, overexpression of a core PCP gene, Strabismus (Stbm), or deletion of a crucial regulator of ncWnt/PCP, NEDD4L, both resulted in delayed blastopore and neural tube closing (Darken et al., 2002; Zhang et al., 2014).

Another indication that Dvl miRNATP-injected embryos have potential gut function defects is that they have much less AP staining in the hindgut (Fig. 5E). In sea urchin larvae, AP is expressed only in differentiated endoderm (Whittaker, 1990; Kumano and Nishida, 1998; Drawbridge, 2003). These results indicate that both the morphology and potentially the function of the gut are negatively impacted by increased Dvl, induced by blocking miRNA-mediated post-transcriptional suppression of Dvl.

To reveal the molecular mechanism of Dvl miRNATP- or Dvl mRNA-induced gut phenotypes, we assayed for changes in transcription factors important for endodermal specification (Fig. 6). Canonical Wnt/β-catenin signaling activates endodermal regulatory genes, such as Krl and Eve, to give rise to the foregut, midgut and hindgut (Howard et al., 2001; Peter and Davidson, 2010, 2011). We observed a significant decrease in the spatial expression domain of Krl and a significant increase in the expression domain of Eve in Dvl miRNATP- and in Dvl mRNA-injected embryos compared with controls (Fig. 6C-F). These results provided independent evidence that increased Dvl leads to expression changes in Krl and Eve (Fig. 6). Krl knockdown in the sea urchin blocks endoderm differentiation and results in failed gastrulation (Howard et al., 2001; Yamazaki et al., 2008). Thus, a moderate decrease in Krl expression domain may impact endoderm differentiation and gastrulation. Eve is directly activated by Wnt/β-catenin, Hox11/13b and Eve itself (Peter and Davidson, 2011). In the intermediate germ cricket, Gryllus bimaculatus, Krl RNAi resulted in elimination of some Eve-regulated segmental stripes, suggesting that Krl is involved in the formation of those stripes by regulating Eve (Mito et al., 2006). However, Krl morpholino knockdown in the sea urchin had no effect on Eve expression in the early blastula stage (18 hpf). Conversely, Eve morpholino knockdown had a small effect on Krl expression, indicating that they are not likely to regulate each other (Peter and Davidson, 2010). Nonetheless, the decreased expression domain of Krl and the increased Eve expression domain induced by Dvl miRNATPs or Dvl mRNA overexpression may contribute to the gut morphology and developmental defects we observed.

Blockade of miRNA suppression of Dvl isoforms resulted in dose-dependent defects in skeletal spicules and PMC patterning (Fig. 7). We also observed Dvl miRNATP-or Dvl mRNA-induced embryos have 50% less Vegf3 expression and decreased Vegf3 expression domain compared with controls (Fig. 8A,C). Dvl mRNA-induced gene expression changes and PMC defects are likely to be specific, as a lower dosage of Dvl mRNA-injected embryos did not result in these changes (Fig. S2B). However, we cannot distinguish direct versus indirect effects of Dvl mRNA overexpression-induced phenotypes. PMCs undergo a series of directed movements whereby their patterning is in part in response to the Vegf signaling pathway (Duloquin et al., 2007; Adomako-Ankomah and Ettensohn, 2013). PMCs express the VegfR10 receptor; binding of its ligand, Vegf3, which is expressed in the ectoderm, guides the PMCs to migrate anteriorly, as well as providing differentiation cues to the PMCs. Vegf3 also regulates the expression of many biomineralization genes in the PMC gene regulatory network (Adomako-Ankomah and Ettensohn, 2013). Loss of the VegfR10 receptor or Vegf3 results in PMCs that fail to pattern and do not secrete a skeleton (Adomako-Ankomah and Ettensohn, 2013; Duloquin et al., 2007). We did not detect expression changes of VegfR10 in Dvl miRNATP- or Dvl mRNA-injected embryos (Fig. 8D). The decrease in Vegf3 level and expression domain may in part explain the Dvl miRNATP- and Dvl mRNA-induced skeletal and patterning defects of the PMCs.

Another potential contributor to the PMC patterning defect in Dvl miRNATP-injected embryos may be defects in PMC motility. As ncWnt mediates cell polarity and actin polymerization, increased Dvl may misregulate cell motility mediators, resulting in defective PMC motility. We found that Dvl protein is expressed in the PMCs, suggesting that Dvl may activate the ncWnt pathways or Wnt-independent functions in mediating cell motility of PMCs (Fig. 7G). To investigate this further, we examined the transcript levels of RhoA, Rac1 and Cdc42 downstream of the ncWnt/PCP pathways in Dvl miRNATP-injected and control embryos. We found that Rac1 transcript levels were elevated in Dvl miRNATP-injected embryos by almost 4-fold (Fig. 4A). Because Rac1 regulates actin polymerization (Chung et al., 2000), we used an mCherry-LifeAct reporter, which encodes a 17 amino acid peptide that binds to filamentous actin (Riedl et al., 2008), to examine actin dynamics in the early embryo. We observed more punctate actin structures in the Dvl miRNATP-injected presumptive PMCs compared with the control (Fig. 7H). In chicken embryo fibroblasts, overexpression of Rac proteins induced disassembly of existing F-actin-containing stress fibers, which consist of long, highly branched bundled actin-rich protrusions, that resulted in dramatic changes in cell morphology (Albertinazzi et al., 1999). The punctate actin structures in Dvl miRNATP-injected embryos may indicate defective actin polymerization, contributing to the inability of PMCs to migrate properly.

These PMC patterning defects are not likely to be acting on the cWnt signaling pathway, as our previous study indicated that removing miRNA suppression of β-catenin had no impact on PMC patterning (Stepicheva et al., 2015). In addition, the overexpression of the LvDvl/SpDvl4a/4b resulted in significantly shorter DVCs and in PMC patterning defects (Fig. 7B,D). These results indicated that the skeletal and PMC defects observed in the Dvl miRNATP-injected embryos are likely to be due to the increased SpDvl4a/4b isoform. The molecular mechanism of how Dvl4a/4b regulates skeletogenesis remains unclear.

An unexpected and interesting finding is that Dvl miRNATP-, LvDvl/SpDvl4a/4b- and LvDvl*/SpDvl5a-injected embryos all had swimming and ciliary defects (Fig. 9, Movies 1-8). Sea urchins have motile monocilia on their ectodermal cells that allow the embryos and larvae to swim with coordinated ciliary beating and to feed (Kinukawa and Vacquier, 2007; Mizuno et al., 2017). A previous study indicated that Dvl-DEP overexpression resulted in embryos that failed to swim (Byrum et al., 2009). The swimming defect observed in Dvl miRNATP- and Dvl mRNA-injected embryos may result from improper ciliary coordination; the direction of ciliary movement depends on the orientation of the basal body, which is primarily determined by the PCP pathway (Marshall and Kintner, 2008; Kunimoto et al., 2012; Mizuno et al., 2017). Dvl in Xenopus is found at the base of the cilia and is essential for apical positioning of the basal bodies and positioning of cilia (Park et al., 2008; Mlodzik, 2016). Knockdown of Dvl1 and Dvl3 in Xenopus and knockdown of Stbm in Clytia resulted in phenotypes of fewer and shorter cilia (Park et al., 2008; Momose et al., 2012), indicating that Dvl plays an evolutionarily conserved role in proper cilia formation. PCP signaling plays a crucial role in ciliogenesis; however, the Wnt/β-catenin also regulates ciliogenesis via foxj1a expression in zebrafish Kupffer's vesicle. Downregulation of cWnt/β-catenin leads to depletion of Lef1 and Tcf7, inhibiting foxj1a transcription. This results in shorter and fewer cilia with loss of cilia motility that can be rescued with overexpression foxj1a (Lin and Xu, 2009; Caron et al., 2012; Zhu et al., 2015). Dvl has also been shown to interact with DCDC2, which binds to tubulin to enhance microtubule polymerization and localizes to the ciliary axoneme (Schueler et al., 2015). Wnt inhibitor treatment rescued the effects of dcdc2 knockdown-induced ciliopathy in renal spheroid cells and zebrafish embryos, suggesting that cWnt signaling is involved in the process of ciliation (Schueler et al., 2015). Thus, the ciliary defect observed in Dvl miRNATP-injected embryos may result from perturbations of both the ncWnt/PCP and the cWnt pathways.

This study demonstrated that miRNAs suppress various Dvl isoforms in the sea urchin embryo. We have shown that Wnt signaling pathway components are controlled post-transcriptionally by miRNAs (Fig. 10). Importantly, the level of increased Dvl protein, induced by blocking miRNA-mediated suppression of Dvl isoforms, is sufficient to induce various defects in the gut, spiculogenesis and ciliogenesis during development.

Adult Strongylocentrotus purpuratus were obtained from Point Loma Marine Invertebrate Lab, Lakeside, CA, USA. Adult males and females were given 0.5 M KCl intracoelomic injections for obtaining sperm and eggs. Filtered natural sea water (collected from Indian River Inlet; University of Delaware, DE, USA) or artificial seawater was used for embryo cultures incubated at 15°C.

Krl, Eve, Wnt5 and Vegf3 were cloned into Blunt-TOPO vectors (Thermo Fisher Scientific) (Stepicheva et al., 2015; Stepicheva and Song, 2015). Krl, Eve, Wnt5 and Vegf3 were linearized with restriction enzymes EcoRI, EcoRI, NotI and BamHI, respectively. All were in vitro transcribed with Sp6 RNA polymerase, except for Vegf3 which was in vitro transcribed with T7 RNA polymerase using the DIG RNA Labeling Kit (Sigma-Aldrich). Dvl5a and Dvl4b 3′UTRs were synthesized as gBlock DNA fragments from IDTdna.com (Integrated DNA Technologies) and cloned into Blunt-TOPO vectors. Dvl4a 3′UTR was PCR amplified with Dvl4a (For 5′-GGCCTCGAGAGTGCGGAAATTTTGAAATCAT-3′; Rev 5′-GGCGCGGCCGCCAATAATGACCGCTCAATTTTT-3′) and cloned into Blunt-TOPO vectors. XhoI and NotI cut sites are underlined. All Dvl probes are linearized with NotI and in vitro transcribed with Sp6 RNA polymerase. All Dvl in situ hybridizations were examined in 180-300 embryos for each stage in three to eight experimental replicates.

For generating Dvl 3′UTR luciferase reporter constructs, Dvl5a and Dvl4b 3′UTR was synthesized as a gBlock DNA fragment (Integrated DNA Technologies). Dvl4a 3′UTR in Blunt-TOPO was subcloned into the Renilla luciferase construct. All seed sequences were mutated at positions 3 and 5 within the miRNA seed sequences (Gregory et al., 2008; Stepicheva et al., 2015). SpmiRDeep2-30364 was modified from 5′-GUGCAAU-3′ to 5′-GUACGAU-3′; spu-miR-153* was modified from 5′-AAAAAT-3′ to 5′-AAGAGT-3′; SpmiR2002 was modified from 5′-CTGAAAT-3′ to 5′-CTTACAT-3′; SpmiR2007 was modified from 5′-CTGAAAT-3′ to 5′-CTTACAT-3′; and SpmiR200 was modified from 5′-GTATGAT-3′ to 5′-GTGTAAT-3′. Mutated nucleotides are underlined. All positive clones were identified by DNA sequencing (Genewiz). Firefly luciferase was used as loading control as previously described (Stepicheva et al., 2015). Luciferase constructs containing the Dvl 3′UTRs were linearized with EcoRI and in vitro transcribed using the mMessage machine kit with either T7 (for Dvl mRNAs) or Sp6 (for firefly luciferase mRNAs) RNA polymerases (Ambion) according to the manufacturer's instructions. mRNAs were purified by using the Macherey-Nagel Nucleospin RNA Clean-up kit according to the manufacturer's instructions. In vitro-transcribed mRNAs were loaded onto the Millipore spin columns to further clean the mRNAs prior to injections.

To identify the contribution of Dvl isoforms to Dvl miRNATP phenotypes, we injected mRNA coding for two Dvl isoforms into zygotes. LvDvl protein is 93% identical to SpDvl4a/4b. The sea urchin species L. variegatus diverged from S. purpuratus about 50 million years ago. These two sister species of sea urchin have many similarities at the genomic as well as at the gene regulatory levels (Crain and Bushman, 1983; Ettensohn et al., 2004). Because of the relatively close evolutionary distance of L. variegatus to S. purpuratus and their high protein identify especially at the C-terminus of the protein, we used the available LvDvl clone (Weitzel et al., 2004) to represent the SpDvl4a/4b protein (Fig. 1). Because of the difficulty of cloning the SpDvl5a isoform, we used PCR to replace the sequence coding for the amino acids YFDDSVSVTLL in LvDvl to a sequence coding for the highly conserved CEFFVDVM amino acid sequence. Because the goal was to test whether overexpression of SpDvl isoforms would induce phenotypes similar to Dvl miRNATPs-injected embryos, we did not add the endogenous 3′UTRs to these protein coding sequences. Thus, LvDvl was used to mimic SpDvl4a/4b (denoted as LvDvl/SpDvl4a/4b) and LvDvl+ CEFFVDVM was used to mimic SpDvl5a (denoted as LvDvl*/SpDvl5a). To obtainLvDvl*, the LvDvl:Flag construct was modified using Q5 Site-Directed Mutagenesis Kit (NEB). Primers used to change the amino acids YFDDSVSVTLL on the C terminus of LvDvl:Flag to CEFFVDVM were: For 5′-GTCGATGTCATGTGAATTCAAGGCCTCTCG-3′ and Rev 5′-AAAGAACTCACAAGGGTTTCCCATAGCCAT-3′. The Flag tag region was deleted after mutation. Plasmids were sequenced to verify that correct mutations were introduced. These plasmids were in vitro transcribed and mRNA coding for the two Dvl isoforms were injected into zygotes.

Microinjections were performed as previously described with modifications (Cheers and Ettensohn, 2004; Stepicheva and Song, 2014). For luciferase assays, 100 ng of Renilla luciferase and 60 ng of firefly luciferase were prepared in 2.5 µl of injection solution consisting of 0.5 µl 100% glycerol and 0.5 µl Texas Red dextran (Molecular Probes). Approximately 1-2 pl was injected into each newly fertilized egg. The stock injection solution contained 150, 250 or 300 µM of each Dvl miRNATP or 450, 750 or 900 µM of control TP in 20% sterile glycerol, 2 mg/ml 10,000 MW Texas Red lysine charged dextran (Molecular Probes) (as previously described; Stepicheva et al., 2015; Stepicheva and Song, 2015). For Dvl miRNATPs, we blasted each of the Dvl miRNATPs against the annotated sea urchin genome and identified them to be complementary to only the Dvl genes. Each Dvl miRNATP is complementary to the miRNA’s seed sequence (6 bp) and the unique Dvl 3′UTR sequences flanking the miRNA seed sequence. Thus, each Dvl miRNATP is uniquely targeting that specific miRNA seed site within the particular Dvl isoform. We microinjected zygotes with a cocktail of Dvl miRNATPs corresponding to miR153* at position +230 of Dvl5a (5′-CTAGCATTTTTTTTTTGAAAGCTGT-3′), to miR-2002 and miR-Deep2-35240 at positions +1198 and +1224 of Dvl4a (5′-TTGCACTTCATGATGCATAGAATAC-3′) and to miR-200 at position +219 of Dvl4b (5′-ATTAATATCATACCCAAAACATATT-3′) at 300 µM each or 900 µM of the negative control (5′-CCTCTTACCTCAGTTACAATTTATA-3′) that corrects a splicing error of the human β-globin pre-mRNA (Kang et al., 1998).

To examine the actin dynamics in control TP and Dvl miRNATP, we microinjected newly fertilized eggs with 500 ng mCherry-LifeAct reporter mRNA, 0.5 µl 100% glycerol and 0.5 µl of Texas Red in a 2.5 µl injection solution (Riedl et al., 2008; Stepicheva et al., 2017).

To identify the impact of individual Dvl isoforms, LvDvl/SpDvl4a/4b mRNA, encoding the sea urchin-specific form of Dvl ending with YFDDSSVTLL, and LvDvl*/SpDvl5a mRNA modified to contain the conserved DEP-C domain (CEFFVDVM), were injected into newly fertilized eggs. Stock injection solutions of 2.5 µl consisted of 0.5 µl 100% glycerol, 2.5 µg LvDvl/SpDvl4a/4b or LvDvl*/SpDvl5a or combinations of both (each with 1.25 µg of mRNA), with 500 ng of mCherry mRNA. The control solution contained 500 ng of mCherry without Dvl mRNA.

All dual luciferase quantification was performed using the Promega Dual-Luciferase Reporter (DLR) Assay Systems with the Promega GloMax 20/20 Luminometry System (Promega). Fifty embryos at the blastula stage were collected in 22 µl 1× lysis buffer and vortexed for 1 min. Embryonic lysates were either stored at −80°C or processed immediately. Prior to luciferase readings, 100 µl of the Luciferase Activating Reagent II (LAR-II) was added to each well of the 96-well plate. Then, 20 µl of the embryonic lysates was added and luciferase reading for the firefly was obtained. Subsequently, 100 µl of the Stop and Glow solution was added to quench firefly luciferase (FF) signal and the Renilla luciferase (Rluc) reading was obtained. Renilla luciferase readings were subtracted from the corresponding initial reading to obtain a firefly reading, which was used to calculate the RLuc/FF ratio. The values of Rluc-containing mutated seed sites were normalized to the corresponding Rluc-containing wild-type miRNA target sites.

One-hundred Dvl miRNATP or control TP-injected blastulae were collected. Total RNA was extracted using the Macherey-Nagel Nucleospin RNA Clean-up XS kit according to the manufacturer's instructions. cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad). QPCR was performed using 2.5 or 7.5 embryo equivalent for each reaction with the Fast SYBER or PowerUp Green PCR Master Mix (Thermo Fisher Scientific) in the QuantStudio 6 Real-Time PCR cycler system (Thermo Fisher Scientific). Results were normalized to the mRNA expression of the housekeeping gene ubiquitin and shown as fold changes compared with control embryos that were injected with the control morpholino using the ΔΔCt method as previously described (Stepicheva et al., 2015). Primer sequences were designed using the Primer 3 Program (Rozen and Skaletsky, 2000) and are listed in Table S1. Three to six biological replicates were conducted. Statistical significance was calculated using two-tailed unpaired Student’s t-tests.

Activity of endogenous AP was used to assay for differentiated endoderm of the larvae as previously described (Hinman and Degnan, 1998; Drawbridge, 2003; Annunziata et al., 2013; Stepicheva et al., 2015). Because we observed a significant difference in the width of the midgut of control TP- and Dvl miRNATP-injected embryos, we used this assay to examine the differentiated endoderm. Five-day-old larvae were fixed in MOPS-paraformaldehyde based fixative (4% paraformaldehyde, 100 mM MOPS pH 7.0, 2 mM MgSO₄, 1 mM EGTA and 0.8 M NaCl) for 10 min at room temperature. Embryos were washed with alkaline phosphatase buffer three times (100 mM Tris pH 9.5, 100 mM NaCl, 50 mM MgCl₂, 0.1% Tween-20), followed by treatment with the staining solution (0.1 M Tris pH 9.5, 50 mM MgCl₂, 0.1 M NaCl, 1 mM levamisole, 10% dimethylformamide, 45 µl of 75 mg/ml NBT and 35 µl of 50 mg/ml BCIP per 10 ml of solution). Once the color development was observed, the staining was terminated with washes with MOPS buffer (0.1 M MOPS pH 7.0, 0.5 M NaCl, 0.1% Tween-20). Images were acquired with a Nikon D90 digital camera connected to a Zeiss Observer Z1 microscope.

Gastrula (48 hpf) and larval-stage (120 hpf) embryos were fixed in 4% paraformaldehyde (20% stock; EMS) in artificial sea water overnight at 4°C. Four 10 min PBS-Tween (0.05% Tween-20 in PBS) washes were performed, followed by 1 h of blocking with 4% sheep serum (Sigma-Aldrich). Primary antibody incubation was performed with Endo1 or ID5 antibodies (McClay et al., 1983) at 1:50 overnight at 4°C. Embryos were washed three times for 15 min each with PBS-Tween followed by goat anti-mouse Alexa 488-conjugated secondary antibody (A11001, Thermo Fisher Scientific) at 1:300. After three more PBS-Tween washes, embryos were incubated with Hoechst dye at 1:1000 for 5 min, followed by two more PBS-Tween washes.

To examine the level of Dvl protein changes in Dvl miRNATP-injected embryos, we injected the control TP and the Dvl miRNATP with Texas Red dextran. These 32-cell stage embryos (4-5 hpf) were fixed in 4% paraformaldehyde (20% stock; EMS) in PBS for 20 min, post-fixed with 100% ice cold methanol for 10 min, and washed with several washes of 1× PBS (Peng and Wikramanayake, 2013). Injected embryos were then incubated with primary rabbit anti-SUDdsh-C antibody (Peng and Wikramanayake, 2013; 1:400 overnight at 4°C). Following this incubation, the embryos were washed with PBS-Tween with 0.01% bovine serum albumin three times and then incubated with goat anti-rabbit Alexa 488-conjugated secondary antibody (A11034, Thermo Fisher Scientific) at 1:300. After three more PBS-Tween washes, embryos were incubated with Hoechst dye at 1:1000 for 2 min, followed by three more PBS-Tween washes.

Larval-stage (120 hpf) embryos were fixed in 4% paraformaldehyde (20% stock; EMS) in artificial sea water overnight at 4°C. Four 10 min PBS-Tween washes were performed, followed by 1 h of blocking with 4% sheep serum (Sigma-Aldrich). Double immunofluorescence was performed with primary antibodies overnight at 4°C, using the following dilutions: anti-SUDdsh-C antibody at 1:400 and β-tubulin (Developmental Studies Hybridoma Bank, E7) at 1:10,000. Embryos were washed three times for 15 min each with PBS-Tween followed by 1 h room temperature incubation with goat anti-mouse Alexa 647-conjugated secondary antibody (A-21237, Thermo Fisher Scientific), followed by goat anti-rabbit Cy3 at 1:300, sequentially. After three more PBS-Tween washes, embryos were incubated with Hoechst dye at 1:1000 for 2 min, followed by two more PBS-Tween washes. These immunolabeled embryos in PBS-Tween were mounted onto protamine sulfate-coated slides and imaged using an LSM 780 scanning confocal microscope (Zeiss Incorporation) and data analysis was performed using Zen software (Zeiss Incorporation).

To measure semi-quantitatively the amount of Dvl in control TP- and Dvl miRNATP-treated embryos, confocal z-stack images of 32-cell embryos were collected. The maximum projections of z-stacks of each embryo were collected. The amount of fluorescence pixels were quantified with Metamorph (Molecular Devices). The level of Dvl in Dvl miRNATP- injected embryos was normalized to that of the control TP-injected embryos.

To measure DVC length, we took a z-stack of differential interference contrast (DIC) and 1D5-immunolabeled images with a Zeiss Observer Z1 microscope. Normal PMC phenotypes have the following features: (1) sub-equatorial ring formation of PMCs at the vegetal pole of the embryo, (2) anterior migration of the PMCs, and (3) syncytial cables formed among the PMCs. Normal gut phenotypes fulfill the following criteria: (1) fully extended gut tube at gastrula stage and (2) Endo1 antibody staining in mid- and hindgut epithelia. To detect ciliary defects in live embryos, we acquired images using the time-lapse function of an LSM 780 scanning confocal microscope (Zeiss Incorporation).

For spicule length, gut/blastopore widths, and expression domains, Student’s t-tests were used to assess the statistical significance between control and experimental groups. For patterning of PMCs, the Cochran–Mantel–Haenszel test was used to assess the statistical significance of the percentage of normal embryos in the control TP-injected group compared with the Dvl miRNATP-injected group. In all graphs, error bars represent s.e.m.

We thank Dr David McClay (Duke University) for the 1D5 antibody; Dr Gary Wessel (Brown University) for the Endo1 antibody; Dr Charles Shuster (New Mexico State University) for the mCherry-LifeAct construct; and Dr Deni Galileo (University of Delaware) for the Metamorph program. We also thank the three anonymous reviewers for their valuable feedback.