Multicellular organisms such as zebrafish use miRNAs to regulate gene expression in a tissue- or time-specific manner, guiding developmental decisions^1,2. To identify target genes regulated by miRNAs, we first developed a microarray to examine temporal miRNA expression patterns during the first 5 d post-fertilization (dpf) of zebrafish development (unpublished data). To understand the function of a subset of these miRNAs, we performed loss-of-function experiments using antisense morpholino oligonucleotides complementary to mature miRNAs. Morpholinos have been used extensively in zebrafish as antisense inhibitors of mRNA translation and splicing³ but are also capable of interfering with miRNA function (Supplementary Fig. 1 online). Injection of morpholinos designed to block the function of miR-214 (214^MO) yielded embryos with U-shaped somites at 1dpf (1 dpf) (Fig. 1a–d). Expression of miR-214 begins during early somitogenesis and continues throughout embryogenesis (Fig. 1e). In situ hybridization showed that miR-214 is expressed in somites at 1 dpf (Fig. 1f,g; see also ref. 2).

Somites are transient embryonic structures derived from paraxial mesoderm that give rise to muscle and skeleton⁴. Presomitic mesodermal cells immediately adjacent to the notochord (adaxial cells) are highly influenced by Hedgehog and give rise to the slow-twitch muscle lineage^5,6. Lateral presomitic cells give rise to fast-twitch muscle fibers and experience little stimulation by Hedgehog initially, whereas later-developing fast-muscle fates are dependent on Hedgehog signaling⁷. There are two slow muscle cell types that require precise Hedgehog signals for proper development: superficial slow fibers (SSFs), which migrate from the midline to populate the surface of the myotome, and slow muscle pioneers that remain close to the midline^6,8. Muscle pioneers require higher levels of and longer exposure to Hedgehog for proper specification than SSFs and can be distinguished from slow muscle fibers by the expression of the transcription factor Engrailed (Eng)^4,6. In situ hybridization demonstrated that inhibition of miR-214 function resulted in a loss of eng2a-positive cells during early segmentation (Fig. 2a,b), consistent with the U-shaped somite defects observed in 214^MO-injected embryos (Fig. 1a–d). Reduction of this marker suggested an overall decrease in Hedgehog signaling in the adaxial cells of 214^MO-injected embryos. To test this, we analyzed expression of patched1 (ptc1), which encodes a Hedgehog ligand receptor whose transcription is activated by Hedgehog⁹. Embryos injected with the 214^MO showed decreased expression of ptc1 in adaxial cells and upregulation in lateral cells. (Fig. 2c–f and Supplementary Fig. 2 online). This suggested that miR-214 might regulate the level of Hedgehog signaling during somite differentiation.

To further characterize defects associated with inhibition of miR-214 function, we monitored several markers of adaxial cell derivatives at later stages. All slow fibers express the homeodomain protein Prox1 as well as a myosin heavy-chain isoform specific to slow muscle (slow myosin HC)¹⁰. Codetection of slow myosin HC and Prox1 by immunohistochemistry showed fewer slow muscles in miR-214 morphants at 1 dpf (Table 1 and Supplementary Fig. 2). Additionally, we monitored Eng proteins, which are expressed in both muscle pioneers and medial fast fibers, a fast-twitch muscle cell type that arises after SSF migration and is dependent on Hedgehog signaling but does not express Prox1 (ref. 6). Injection of 214^MO reduced the number of muscle pioneers during late somitogenesis (Fig. 2g and Table 1). To verify that the altered somite differentiation in miR-214 morphant embryos is due to perturbation of Hedgehog signaling, we sought to rescue the 214^MO phenotype by coinjection of synthetic sonic Hedgehog (shh) RNA (Fig. 2g–i, Table 1 and Supplementary Fig. 2). Notably, miR-214 morphant defects were completely suppressed by Shh misexpression.

Next, we analyzed the effect of miR-214 misexpression on Hedgehog-mediated cell fate specification (Fig. 2j–m and Supplementary Fig. 2). Unlike downregulation of miR-214 function, injection of synthetic miR-214 into one-cell embryos did not yield appreciable alterations of ptc1 expression in somites (Supplementary Fig. 2). However, ectopic expression of excess miR-214 did result in perturbation of Hedgehog-regulated markers in the ventral neural tube (Fig. 2j–m and Supplementary Fig. 2). nkx2.2a is expressed in the lateral floor plate of the neural tube, and its expression is stimulated by the highest levels of Hedgehog activity, whereas olig2 expression is adjacent and dorsal to the nkx2.2a domain^11–13. Misexpression of miR-214 resulted in expanded expression of nkx2.2a and a dorsal shift of olig2 expression, further supporting a role for miR-214 in the modulation of Hedgehog signaling.

Similar to most miRNAs, miR-214 has many predicted targets, one of which is suppressor of fused (su(fu))¹⁴ (Fig. 3a). Su(fu) is a well characterized negative regulator of Hedgehog signaling essential for proper specification of muscle cell types during somitogenesis^6,15. To test whether miR-214 targets su(fu), we monitored GFP fluorescence in embryos microinjected with mRNAs derived from the following reporter constructs: the 3′ UTR of su(fu) cloned downstream of the GFP ORF (GFPsu(fu)), GFP fused to two perfect miR-214 recognition elements (GFP 2XMRE) or GFP lacking heterologous 3′ UTR sequences (GFP –UTR) (Fig. 3b–h). Consistent with the hypothesis that sequences in the su(fu) 3′ UTR contain bona fide recognition elements, GFP fluorescence was substantially lower in embryos coinjected with miR-214 RNA and GFPsu(fu) mRNA (Fig. 3e). We verified the fluorescence experiments by protein blots performed with antibodies against GFP on whole-embryo lysates (Fig. 3h).

To confirm that the decrease in slow muscle cells caused by miR-214 inhibition is directly due to su(fu) derepression, we sought to rescue the defect through simultaneous downregulation of both su(fu) and miR-214 (Fig. 4a–h) (Table 1). Injection of two independent morpholinos targeted to su(fu) (su(fu)^MO1,2) caused an increase in the number of slow muscle cells but did not have a significant effect on the number of muscle pioneers (Fig. 4g,h and Table 1)⁶. In contrast, injection of a mismatched morpholino (su(fu) mm^MO) did not have any effect (Fig. 4a,b and Table 1). Notably, coinjection of 214^MO along with su(fu)^MO1,2 resulted in a restoration of slow muscle cells (Fig. 4e,f and Table 1) compared with embryos coinjected with 214^MO and su(fu)-mm^MO (Fig. 4c,d and Table 1). Inhibition of su(fu) alone has been shown to increase the number of medial fast fibers while not affecting specification of muscle pioneers⁶. Colabeling of Eng and Prox1 demonstrated that the rescued Eng-expressing cells in 214^MO and su(fu)^MO1,2 coinjected embryos are muscle pioneers and not medial fast fibers (Fig. 4a,c,e,g). Together, our results indicate that su(fu) inhibition suppresses the deficiency of muscle pioneers and SFFs associated with loss of miR-214 function, supporting the hypothesis that miR-214 modulates Hedgehog signaling largely by regulation of su(fu).

Su(fu) participates in Hedgehog signaling by altering the function of the Gli family of transcription factors. Su(fu) tethers both the activator and repressor forms of Gli in the cytoplasm, resulting in downregulation of both activities^6,16. Inhibition of Su(fu) by miR-214 in adaxial cells, which experience high amounts of Hedgehog signaling, allows maximal activation of Gli-mediated transcription (Fig. 5). Regulation of su(fu) by miR-214 in lateral muscle cells, which are exposed to lower levels of Hedgehog signaling, results in increased repressor activity, ensuring a commitment to fast-muscle cell fate (Fig. 5). These distinct effects on Hedgehog signaling are demonstrated by changes in the expression of ptc1 in 214^MO-injected embryos. Inhibition of miR-214 function permits increased expression of Su(fu), resulting in decreased ptc1 expression in adaxial cells and increased expression in lateral presomitic mesoderm (Fig. 2d). The decreased Hedgehog response in adaxial cells causes loss (as in the case of muscle pioneers) or reduction (as in the case of SSFs) in cell types dependent on high levels of Hedgehog signaling, triggering the formation of U-shaped somites. In contrast, increased expression of ptc1 in lateral somites can be explained by the decreased activity of Gli repressor forms, which are also negatively regulated by Su(fu). Likewise, aberrant expression of Hedgehog-regulated genes in the neural tube caused by ectopic expression of miR-214 can be explained by disruption of Su(fu) expression. In the presence of ectopic miR-214, neural tube cells become more sensitive to Hedgehog signals, causing them to acquire more ventral fates, as demonstrated by the increase in nkx2.2a expression and a dorsal shift in olig2 expression. Together, these data support a role for miR-214-mediated regulation of su(fu) that is essential for specification of muscle cell types during somitogenesis by sharpening the response to different levels of Hedgehog signals.

Compared with their function in invertebrates, fungi and plants, the function of miRNAs in vertebrates has been proposed to be more limited, subtly modulating cell types or acting as a redundant mechanism for ensuring appropriate gene expression patterns¹⁷. We have shown a requirement for miR-214 in the specification of muscle cell fate during somitogenesis. This finding suggests that some vertebrate miRNAs have decisive roles during development, being required for the generation of specific cell types.