Target Protectors Reveal Dampening and Balancing of Nodal Agonist and Antagonist by miR-430

See allHide authors and affiliations

Science  12 Oct 2007:
Vol. 318, Issue 5848, pp. 271-274
DOI: 10.1126/science.1147535


MicroRNAs (miRNAs) repress hundreds of target messenger RNAs (mRNAs), but the physiological roles of specific miRNA-mRNA interactions remain largely elusive. We report that zebrafish microRNA-430 (miR-430) dampens and balances the expression of the transforming growth factor–β (TGF-β) Nodal agonist squint and the TGF-β Nodal antagonist lefty. To disrupt the interaction of specific miRNA-mRNA pairs, we developed target protector morpholinos complementary to miRNA binding sites in target mRNAs. Protection of squint or lefty mRNAs from miR-430 resulted in enhanced or reduced Nodal signaling, respectively. Simultaneous protection of squint and lefty or absence of miR-430 caused an imbalance and reduction in Nodal signaling. These findings establish an approach to analyze the in vivo roles of specific miRNA-mRNA pairs and reveal a requirement for miRNAs in dampening and balancing agonist/antagonist pairs.

MicroRNAs (miRNAs) are small RNA molecules ∼22 nucleotides long and function to block the translation and enhance the decay of target mRNAs (1). Recent studies have uncovered activities of specific miRNA families and have identified hundreds of putative target mRNAs (13). However, the physiological roles of specific miRNA-mRNA pairs are largely unknown (1, 2). To develop a method to disrupt specific miRNA-mRNA pairs, we focused on the zebrafish microRNA-430 (miR-430) family. This miRNA family is highly expressed during early zebrafish development, targets hundreds of mRNAs, and is required for embryonic morphogenesis and clearance of maternal mRNAs (4, 5). Analysis of 3′ untranslated regions (3′UTRs) with sites complementary to miR-430 identified squint (sqt), a member of the Nodal family of transforming growth factor–β (TGF-β) signals, and lft1 and lft2, members of the Lefty family of TGF-β signals (fig. S1). Nodals are the key regulators of mesendoderm induction and left-right axis formation, whereas Leftys act as antagonists of Nodal signaling (6, 7). The balance between Nodals and Leftys determines the extent of mesendoderm formation (68) (fig. S1). Given their potent and concentration-dependent effects, we hypothesized that miR-430 might be required to dampen these signals.

Four lines of evidence indicate that sqt, lft1, and lft2 are in vivo targets of miR-430. (i) Reporter mRNAs consisting of the green fluorescence protein (GFP) coding region and full-length sqt, lft1, or lft2 3′UTRs were repressed in the wild type but not in MZdicer mutants, which lack all mature miRNAs including miR-430. Derepression of reporter genes was most pronounced for lft2 and least marked for lft1, suggesting that lft2 is more strongly repressed by miR-430 than sqt and lft1 (Figs. 1A and 2D and figs. S2 and S5). (ii) Mutations of two nucleotides within the miR-430 target site (GCACUU to GGUCUU) abolished repression of reporter mRNAs (Fig. 1A and fig. S2). (iii) Endogenous expression of sqt, lft1, and lft2 mRNAs was increased in MZdicer mutants (Fig. 3, A and B, and fig. S2). (iv) Misexpression of sqt, lft1, or lft2 mRNAs containing mutated miR-430 binding sites (sqt mut-3′UTR, lft1mut-3′UTR, lft2mut-3′UTR) resulted in higher physiological activity (Fig. 1, B to F, and fig. S2). These results indicate that miR-430 represses sqt, lft1, and lft2 expression and activity.

Fig. 1.

miR-430 represses sqt and lft2 expression and activity. (A) mRNAs for GFP reporters (green) containing the 3′UTRs of sqt or lft2 are co-injected with control DsRed (red) mRNA. Expression is analyzed at 25 to 30 hours postfertilization (hpf). Wild-type (wt) reporters are repressed in wild-type embryos as compared to MZdicer mutants. Repression is abolished by mutations in the predicted miR-430 target sites. Predicted pairings of the 3′ UTRs to miR-430 are shown. The lft2 reporter appears more strongly repressed by miR-430 than the sqt reporter. (B) Outline of activity assays; sqt or lft2 open reading frame (ORF) with either wild-type or mutated 3′UTR is injected at the one-cell stage. mRNA activity is assessed at 50% epiboly (∼5 hpf) by RNA in situ hybridization or at 25 to 30 hpf by morphology. (C) Embryos injected with 2 pg of wild-type lft2 mRNA appear similar to uninjected controls, whereas injection of 2 pg of lft2mut-3UTR mRNA causes cyclopia (arrow) and loss of trunk mesoderm (arrowhead), hallmarks of reduced Nodal signaling. (D) Physiological activity of sqt or lft2 mRNA assessed by fascin (fas) induction, a marker for Nodal signaling activity. lft2mut-3UTR mRNA (2 pg) causes a stronger decrease in fas induction than wild-type lft2. sqt mut-3UTR mRNA (2 pg) leads to greater ectopic induction of fas than wild-type sqt. (E) Percentage of embryos with decreased Nodal signaling (cyclopia and loss of trunk mesoderm) at increasing concentrations of wild-type lft2 or lft2mut-3UTR mRNA. (F) Percentage of embryos with increased Nodal signaling (ectopic gsc induction covering >50% of the animal pole) at increasing concentrations of wild-type sqt or sqtmut-3UTR mRNA.

Fig. 2.

miRNA target protectors (TPs) interfere with specific miRNA-mRNA interactions. (A) Experimental approach. Target protectors are co-injected with GFP-reporters (green) into wild-type embryos and prevent miR-430–induced target repression. Predicted pairings of sqt-TPmiR-430 and lft2-TPmiR-430 to sqt and lft2 3′UTRs are shown. DsRed mRNA (red) is injected as a control. (B) Wild-type reporter is repressed in wild-type embryos. (C and D) Co-injection of sqt-TPmiR-430 or mutation of miR-430 target site prevents GFP repression. (E) sqt-TPmiR-430 does not affect repression of lft2-GFP reporter. (F) sqt-GFP reporter with introduced miR-1 target site is repressed in wild-type embryos. miR-1 is not expressed during early zebrafish embryogenesis. (G) sqt-TPmiR-430 prevents GFP repression in the absence of miR-1. (H) sqt-TPmiR-430 does not interfere with activity of injected miR-1. (I) sqt-TPmir-1 does not interfere with miR-430 activity.

Fig. 3.

Target protection results in increased sqt and lft2 expression and activity. (A) sqt-TPmiR-430 injection results in elevated sqt expression, similar to the finding in MZdicer mutants. sqt-TPmiR-430 does not increase sqt expression in MZdicer. (B) lft2-TPmiR-430 injection results in elevated lft2 expression, similar to the finding in MZdicer mutants. lft2-TPmiR-430 does not increase lft2 expression in MZdicer. (C) sqt-TPmiR-430-injected embryos exhibit increased gsc expression (arrowheads) that is suppressed by co-injection of a sqt-AUG morpholino. (D) lft2-TPmiR-430–injected embryos display cyclopia (arrowheads) that is suppressed by co-injection of a lft2-AUG morpholino.

To study the physiological role of miR-430/sqt and miR-430/lft interactions, we developed a method to disrupt the interaction of miRNAs with target mRNAs. RNA-binding morpholino antisense oligonucleotides are commonly used in zebrafish to block the translation or splicing of target RNAs (911). We reasoned that morpholinos overlapping with miRNA target sites might interfere with miRNA-mRNA interactions, thus protecting the target from the miRNA (target protector, TP) (Fig. 2A). Specificity would be attained by the sequences unique to the 3′UTR. To test this strategy, we analyzed the effect of morpholinos complementary to the region of the miR-430 target sites in the sqt or lft 3′UTRs. Four lines of evidence indicate that TPs interfere with miR-430–mediated repression of specific 3′UTRs. (i) Injection of sqt-TPmiR-430 blocked miR-430– induced repression of the sqt-GFP reporter (Fig. 2, B to D, and fig. S3). (ii) sqt-TPmiR-430 did not block repression of the lft2-GFP reporter, suggesting that TPs do not induce cross-protection (Fig. 2E). (iii) Control morpholinos complementary to other regions of the sqt 3′UTR did not prevent sqt-GFP repression by miR-430 (fig. S4). (iv) Injection of sqt-TPs into MZdicer mutants did not affect the levels of sqt–GFP reporter or sqt gene expression, suggesting that TPs do not cause nonspecific stabilization of mRNAs (Fig. 3A and fig. S5). Corresponding results were obtained with lft2-TPmiR-430 (Fig. 3B and figs. S2 to S5). To test whether TPs specifically block the interaction with one target site without affecting the interaction with other target sites in the same 3′UTR, we placed a miR-1 target site into the sqt–GFP reporter (Fig. 2, F to I). Protection of the miR-430 target site did not prevent miR-1–mediated GFP repression (Fig. 2H), and protection of the miR-1 target site did not interfere with miR-430–mediated repression (Fig. 2I). Taken together, these results indicate that target protection provides a powerful in vivo method to specifically investigate the role of individual miRNA-mRNA target site interactions.

To determine the role of miR-430 repression of sqt, we analyzed sqt-TPmiR-430–injected embryos. Similar to MZdicer mutants, sqt expression was elevated (Fig. 3A). Protection of sqt increased the induction of mesodermal marker genes such as goosecoid (gsc), indicative of higher Nodal signaling during blastula stages (6, 8, 12) (Figs. 3C and 4B and fig. S6). The increased gsc induction resulted from the protection of zygotically transcribed but not maternally loaded sqt (fig. S7). Ectopic gsc induction in sqt-TPmiR-430–injected embryos was suppressed by a morpholino blocking sqt translation, indicating that sqt-TPmiR-430 specifically increased sqt activity (Fig. 3C and fig. S6). To quantify the effects of increased Nodal signaling, we analyzed the number of sox17-expressing endoderm progenitors and dorsal forerunner cells during gastrulation (6, 8). Forerunner cells are induced by Nodal signaling at the dorsal margin and form Kuppfer'svesicle, an embryonic organ that functions during left-right axis formation (8, 13). Cell counting revealed an increase in the number of endodermal and forerunner cells in sqt-TPmiR-430 embryos (Fig. 4, C to E). These results suggest that miR-430 can dampen Nodal signaling by repressing sqt.

Fig. 4.

miR-430 maintains the balance between sqt and lft2. (A) Schematics of miR-430 regulation of sqt (S) and lft2 (L) in wild-type, wild-type + sqt-TPmiR-430, wild-type + lft2-TPmiR-430, wild-type + sqt-TPmiR-430 + lft2-TPmiR-430, MZdicer, and MZdicer+miR-430 embryos. Removal of miR-430 regulation in each case results in increased sqt and/or lft2 expression. (B) gsc expression is increased in sqt-TPmiR-430–injected embryos and decreased in lft2-TPmiR-430–injected embryos. gsc induction is similar in wild-type, wild-type + sqt-TPmiR-430 +lft2-TPmiR-430, MZdicer, and MZdicer+miR-430 embryos at 50% epiboly. (C) sox17 expression is reduced in wild-type + sqt-TPmiR-430 +lft2-TPmiR-430 and MZdicer embryos as compared to uninjected wild-types at 75% epiboly. sox17 labels endodermal cells (bracket) and forerunner cells (arrowhead). (D) Quantification of sox17–expressing endodermal cells (n = 5 to 10 embryos for each genotype per injection). (E) Quantification of sox17–expressing forerunner cells (n = 12 to 35 embryos for each genotype per injection). (D and E) Endodermal and forerunner cell numbers vary from embryo to embryo. Bars represent mean ± SEM, which are significantly different between wild-type and wild-type + sqt-TPmiR-430 (P < 0.0005 by two-tailed Student's t test), wild-type and wild-type + lft2-TPmiR-430 (P <10–12), wild-type and wild-type + sqt-TPmiR-430 +lft2-TPmiR-430 (P < 10–7), wild-type and MZdicer (P < 10–8), wild-type + lft2-TPmiR-430 and wild-type + sqt-TPmiR-430 + lft2-TPmiR-430 (P < 0.02), and MZdicer and MZdicer+miR-430 (P < 10–5) embryos. (F) Model for miRNA-mediated balancing of an agonist and an antagonist.

To determine the in vivo role of miR-430 repression of lft, we focused on lft2 because the repression of lft2 by miR-430 was more pronounced than it was for lft1 (Fig. 1A and fig. S2). lft2 target protection resulted in elevated lft2 expression, similar to the finding in MZdicer mutants (Fig. 3B). lft2-TPmiR-430–injected embryos displayed cyclopia, reduced gsc expression (Figs. 3D and 4B and fig. S6) (6, 8, 12), and fewer sox17-expressing endodermal and forerunner cells (Fig. 4, C to E). These results indicate that miR-430 can enhance Nodal signaling by dampening lft2.

To determine the role of miR-430 in simultaneously dampening both sqt and lft2, we co-injected sqt-TPmiR-430 and lft2-TPmiR-430. The induction of gsc was not strongly affected in sqt/lft2-TPmiR-430 embryos and MZdicer mutants (Fig. 4B), but the expression of sox17 revealed a reduced number of endodermal and dorsal forerunner cells in sqt/lft2-TPmiR-430 embryos and MZdicer mutants (Fig. 4, C to E). These results indicate that loss of miR-430 regulation leads to an imbalance of sqt and lft inputs and reduces some outputs of Nodal signaling.

Our study of miR-430 and Nodal signaling provides two major insights. First, the regulation of sqt and lft2 by miR-430 identifies a role for miRNAs as dampeners and balancers of agonist/antagonist pairs and reveals a previously unknown regulatory layer of Nodal signaling (Fig. 4, A and F, and fig. S1). miR-430 reduces the absolute levels of sqt and lft2 expression (dampening) and regulates their relative levels to achieve optimal activity of the Nodal pathway (balancing). The protection of sqt and lft2 from miR-430 does not appear to lead to major phenotypic changes during blastula stages (gsc expression) but reduces Nodal signaling during gastrulation (sox17 expression). Because Nodal and Lefty signals have complex regulatory interactions (6, 7), multiple mechanisms might contribute to this temporal difference. For example, stronger derepression (Figs. 1A and 3, A and B) and longer persistence of lft2 after loss of miR-430 regulation could inhibit Sqt and the related Nodal signal Cyclops during gastrulation (6, 8, 12). The regulation of Nodal signaling by miR-430 is likely to be conserved, because miR-430 is found in other vertebrates (miR-302, miR-372, and miR-519) and predicted miR-430 target sites are present in other Nodal and Lefty genes (fig. S1) (4). More generally, our results reveal a regulatory interaction in which a repressor (miR-430) dampens the expression of both an agonist (sqt) and an antagonist (lft) (Fig. 4, A and F, and fig. S1). Dampening might not only allow balancing of counteracting inputs but also add robustness (1418). For example, our overexpression experiments show that the embryo can tolerate increased expression of miR-430–regulated sqt or lft mRNA, whereas loss of miR-430–mediated regulation leads to gain-of-function phenotypes (Fig. 1C and fig. S2). miRNA-mediated balancing of agonist/antagonist pairs might also contribute to the evolution of phenotypic changes. The short region of sequence complementarity required for the recognition of miRNA target sites allows for the rapid acquisition, loss, or modulation of miRNA-mRNA target interactions (1921). Our results raise the possibility that target sequence variations could change the balance of agonist/antagonist expression and induce phenotypic changes such as the expansion or reduction of progenitor fields.

Second, our study introduces a method to test the role of specific miRNA-mRNA pairs in vivo (fig. S8). Thousands of miRNA-mRNA interactions have been predicted, but less than a dozen have been shown to have an in vivo function (2, 3). The sequence-selectivity of morpholino target protectors makes them excellent agents to disrupt specific miRNA-mRNA interactions. Other antisense oligonucleotides and small molecules that bind to miRNA target sites or their vicinities are also likely to serve as target protectors. Target protectors not only uncover the physiological role of miRNA-mRNA interactions, but also illustrate how miRNA phenotypes are a composite created by up-regulation of multiple targets (fig. S8). Additionally, target protectors might serve as therapeutic agents (fig. S8). More than 30% of all human genes are thought to be miRNA targets (13). By blocking the interaction of specific miRNA-mRNA pairs through the use of target protectors, the translation and stability of particular mRNAs could be increased and result in the suppression of hypomorphic mutations or the up-regulation of beneficial gene products such as tumor suppressors or peptide hormones (fig. S8).

Supporting Online Material

Materials and Methods

Figs. S1 to S8


References and Notes

View Abstract

Navigate This Article