MicroRNA-20a is essential for normal embryogenesis by targeting vsx1 mRNA in fish

Abstract

MicroRNAs are major post-transcriptional regulators of gene expression and have essential roles in diverse developmental processes. In vertebrates, some regulatory genes play different roles at different developmental stages. These genes are initially transcribed in a wide embryonic region but restricted within distinct cell types at subsequent stages during development. Therefore, post-transcriptional regulation is required for the transition from one developmental stage to the next and the establishment of different cell identities. However, the regulation of many multiple functional genes at post-transcription level during development remains unknown. Here we show that miR-20a can target the mRNA of vsx1, a multiple functional gene, at the 3′-UTR and inhibit protein expression in both goldfish and zebrafish. The expression of miR-20a is initiated ubiquitously at late gastrula stage and exhibits a tissue-specific pattern in the developing retina. Inhibition of vsx1 3′-UTR mediated protein expression occurs when and where miR-20a is expressed. Decoying miR-20a resulted in severely impaired head, eye and trunk formation in association with excessive generation of vsx1 marked neurons in the spinal cord and defects of somites in the mesoderm region. These results demonstrate that miR-20a is essential for normal embryogenesis by restricting Vsx1 expression in goldfish and zebrafish, and that post-transcriptional regulation is an essential mechanism for Vsx1 playing different roles in diverse developmental processes.

Introduction

MicroRNAs (miRNAs) are a group of endogenous noncoding small RNAs, with 22–24 nucleotides in length, first discovered in C.elegans.^1-3 MiRNAs have been validated in almost all examined plants and animals and are highly conservative in sequence.^3-11 It is well known that miRNAs can induce the breakdown of downstream mRNAs (mRNAs) and/or the inhibition of translation by binding to their target sites, mainly within the 3′ untranslated region (3′-UTR), of downstream target mRNA by base pairing,^12,13 and have been recognized as a major post-transcriptional regulators of gene expression. A number of studies indicate that miRNAs play important roles in diverse developmental processes.^14-17

In vertebrates, some regulatory genes play different roles at different developmental stages. These genes are initially transcribed in a wide embryonic region but restricted within distinct cell types at subsequent stages during development. Therefore, post-transcriptional regulation is required for the transition from one developmental stage to the next and the establishment of different cell identities. However, the developmental regulation of many multiple functional genes at post-transcription level remains unknown.

visual system homeobox-1 (vsx1) plays different roles at different developmental stages according to a precise spatiotemporal pattern. At blastula and gastrula stages, vsx1 is ubiquitously expressed at a low level in zebrafish, goldfish and mouse,^18-21 and crucial for axial-paraxial mesoderm patterning²¹ and prechordal mesendoderm formation²² in zebrafish. During neurogenesis, vsx1 is expressed in presumptive neurons of hindbrain and spinal cord in zebrafish and Xenopus,^23,24 suggesting that it might play a role in neurons formation in the central neural system. During retinogenesis of examined vertebrates, vsx1 is expressed weakly (in zebrafish and mouse) or strongly (in goldfish, chick and Xenopus) in undifferentiated, presumptive neural retina, then up-regulated selectively in presumptive bipolar cells,^18,23,25-27 and has been demonstrated to play a role in regulating the proliferation and differentiation of retinal progenitors as well as the functional maintenance of bipolar cells.^18,25-27 Therefore, post-transcriptional regulation might be required for the transition from one developmental stage to the next and the establishment of different cell identities.

In Xenopus, the 3′-UTR of Xvsx1 mRNA is sufficient to drive tissue-specific translation during retinogenesis.²⁸ In goldfish, vsx1 3′-UTR mediates ubiquitous translation during early embryogenesis and tissue-specific translation at late developmental stages.²⁹ It has been revealed that a set of miRNAs, miR-129, miR-155, miR-214 and miR-222, are ubiquitously and highly expressed in Xenopus retina at early developmental stages but the amounts of these miRNAs are decreased at the generation stage of polar cells.¹⁵ These miRNAs can inhibit translation by binding to the 3′-UTR of Xvsx1 mRNAs and the functional inactivation of these miRNAs could elicit the generation of additional bipolar cells.¹⁵ These results provide evidences that miRNAs are used to control the generation timing of bipolar neurons by inhibiting vsx1 mRNA translation in the retinal progenitors at early developmental stages. Recently, diverging duplicate vsx1 loci, named vsx1A1 and vsx1A2, have been identified in goldfish. Both vsx1A1 and vsx1A2 are transcriptional but their sequences at the 3′-UTRs are significantly diverged. vsx1 A1 3′-UTR-linked gene exhibits a ubiquitous translation pattern at blastula and gastrula stages but a tissue-specific translation pattern at later developmental stages. vsx1 A2 3′-UTR has lost the capability of mediating tissue-specific translation at later development stages.²⁰ It is reasonable that, at vsx1 A1 3′-UTR, some binding sites for miRNAs are crucial for tissue-specific protein expression during embryogenesis and retinogenesis. By screening different potential miRNA binding sites at the 3′-UTR of goldfish vsx1A1 and vsx1A2, we identified that miRNA 20a (miR-20a) can target vsx1 mRNA by binding to the 3′-UTR and effectively induce degradation or/and translational inhibition of vsx1 mRNA. Decoying miR-20a resulted in severe defects of embryogenesis in goldfish and zebrafish. The role of miR-20a in regulating vsx1 expression explains why transcription factor Vsx1 plays different roles at different developmental stages.

Results

Selection of miRNAs predicted binding to the 3′-UTR of goldfish vsx1A1

The 3′-UTR of goldfish vsx1 A1 can mediate tissue-specific translation but the 3′-UTR of vsx1 A2 does not possess this capability.²⁰ To determine which miRNAs are involved in mediating tissue-specific translation of vsx1, we first analyzed the potential miRNA binding sites at the 3′-UTR of goldfish vsx1A1 and vsx1A2. A bioinformatics screening tool called PITA, which is able to compute the optimal sequence complements between a set of mature miRNAs and a given mRNA using a weighted dynamic programming³⁰ was utilized. Based on evolutionary conservation of miRNAs, mouse miRNA database was selected for identifying the potential miRNA target sequences by the standard parameter settings with 8 mer seed, a single mismatch and one G : U wobble allowed. Since the value is an energetic score, the lower the score is, the stronger the binding of miRNA to the given sequence. According to the rough rule of thumb,³¹ miRNAs having values below –5 were picked. 96 and 51 miRNAs were supposed to have potential binding sites at the 3′-UTR of vsx1 A1 and A2, respectively, and 13 miRNAs were supposed to have potential binding sites at both 3′-UTRs of vsx1 A1 and A2 (Fig. 1A; Table S1).

Figure 1.

MiRNAs with potential binding sites on 3′-UTR of goldfish vsx1A1 and A2. (A) Number of miRNAs which has potential binding sites on goldfish vsx1 A1 and A2 3′-UTRs predicted by PITA. (B) Alignment of 3′-UTR sequences of zebrafish vsx1, goldfish vsx1A1 and A2. Seed sequences and positions of 4 binding sites which might form better helixes with their corresponding miRNA candidates at the 3′-UTR of vsx1 A1 than that of vsx1 A2 are labeled with different colors and the stop codon is marked by a black solid box.

To exclude the false positives, we carried out a target prediction of miRNAs on zebrafish vsx1 3′-UTR via another approach TargetScanFish,³² which retrieves predicted regulatory targets of zebrafish miRNAs.³³ 11 miRNAs were predicted to have potential binding site at the 3′-UTR of zebrafish vsx1 mRNA (Table S2). Only four miRNAs that have potential binding sites at both zebrafish vsx1 and goldfish vsx1 A1 3′-UTRs (Fig. 1B) were screened for further analysis with the RNAhybrid method by examining the best fit and minimum free energy of miRNA-mRNA hybridization.³⁴ According to the RNAhybrid algorithm, miR-129-5p, miR-137-3p, miR-181c-5p and miR-20a-5p might form better helixes with their corresponding seed sequences at the 3′-UTR of vsx1 A1 than that at the 3′-UTR of vsx1 A2 (Fig. 1B; Fig. S1), Therefore, these miRNAs were selected as the competitive candidates for target recognition test.

MiR-20a can bind to vsx1 mRNA 3′-UTR

Recently, novel classes of chemically engineered oligonucleotides, termed “antagomirs” or “agomirs,” have been developed and proved to be efficient and specific silencers or enhancers of endogenous miRNAs.^35-38 In order to determine which of the 4 candidate miRNAs is able to inhibit the expression of vsx1 A1 or A2 3′-UTR-mediated green fluorescent protein (GFP) in vivo, each of the 4 miRNA agomirs was separately coinjected with vsx1 A1 3′-UTR-linked or vsx1 A2 3′-UTR-linked GFP reporter sensors at one cell stage. When 30 pg vsx1 A1 3′-UTR-linked GFP reporter sensor was coinjected with 688 pg agomir-20a, at early gastrula stage, dramatic decrease of GFP expression was detected, but no detectable change of GFP expression was observed when the vsx1 A2 3′-UTR-linked or vsx1 3′-UTR absent GFP reporter sensor was coinjected with agomir-20a at the same dosage (Fig. 2B-I). However, coinjection of 688 pg agomir-181c, –137 and –129 had no detectable effect on the expression of the vsx1 A1 or vsx1 A2 3′-UTR-linked GFP sensors (Fig. S2), suggesting that neither vsx1 A1 nor vsx1 A2 3'UTR is the target of these miRNAs.

Figure 2.

MiR-20a inhibits vsx1A1 3′-UTR mediated GFP expression. (A) Schematic diagram of pCS2-GFP, pCS2-GFP-vsx1A2 3′UTR, pCS2-GFP-vsx1A13′UTR and pCS2-GFP-vsx1A13′UTR-Mut reporter plasmids. (B–I) GFP expression driven by different 3′-UTRs at early-gastrula embryos. The GFP reporter sensors are indicated on the top of each column. The injected agomir is indicated at the left.

Sequence analysis showed that the 3′-UTR of vsx1 A1 mRNA contains only one potential binding site for miR-20a. To confirm that miRNA-20a can repress protein expression by base pairing the binding site of 3′-UTR of vsx1 A1, we further constructed a GFP reporter sensor driven by a mutant 3′-UTR of vsx1 A1 in which the core sequence GCACUUU at the potential binding site was converted to GgAgUaU (Fig. 2A). Fluorescent observation showed that the mutant vsx1 A1 3′-UTR-linked GFP was expressed successfully as the wild type A1 3′-UTR-linked GFP (Fig. 2E). Unlike the wild type, however, the expression of the mutant vsx1 A1 3′-UTR-linked GFP could not be inhibited by coinjection with 688 pg agomir-20a (Fig. 2I). This result substantiated that vsx1 A1 mRNA 3′-UTR is a direct target of miR-20a in goldfish.

Of all the 4 examined microRNAs, miR-181c is predicted to form the best helixes with its corresponding seed sequence at the 3′-UTR of vsx1 A1 (Fig. S1). However, experimental examination showed that miR-20a, rather than miR-181c, is the actual binding microRNA at the 3′-UTR of vsx1 A1 and inhibits translation. This could be explained by the site accessibility of microRNA target recognition.³⁰ Putative second structure of vsx1 A1 3′-UTR showed that the binding sequence complement to miR-20a seed sequence is located between wide open regions, while the potential binding sequence complement to miR-181c seed sequence is located within a closed region (Fig. S3A and B). Interestingly, the binding sites of miR-20a and miR-181c target recognition are also located in a wide open region and in a similar closed region, respectively, in the 3′-UTR of zebrafish vsx1 mRNA (Fig. S3C and D). This evolutionary conservation suggests that the 3′-UTR of zebrafish vsx1 mRNA is also the target of miR-20a.

MiR-20a can repress Vsx1 expression in vivo in goldfish and zebrafish

Previous experiments have demonstrated that maternal vsx1 mRNA directs Vsx1 translation during cleavage in zebrafish and the expression pattern of Vsx1 protein can be visualized by zebrafish Vsx1 antibody.²² Because goldfish has vsx1 A1 and A2 loci²⁰ and vsx1 A2 3′-UTR is not the target of miR-20a (Fig. 1B), we investigated whether miR-20a could inhibit endogenous Vsx1 expression in vivo by examining Vsx1 expression in agomir-20a injected zebrafish embryos. Immunohistochenimstry analysis with zebrafish Vsx1 antibody showed that Vsx1 protein was translated in the blastomeres except in a few dorsal blastomeres at dome stage (Fig. 3A–C). When 688 pg of agomir-20a was injected at one cell stage, the Vsx1 protein was undetectable in the embryos during cleavage (Fig. 3D–F). In contrast, injection of antagomir-20a at the same dosage has no detectable effect on Vsx1 expression (Fig. 3G–I). These results demonstrate that miR-20a can effectively inhibit endogenous Vsx1 expression in vivo.

Figure 3.

MiR-20a represses endogenous Vsx1 expression in zebrafish. Localization of Vsx1 protein in uninjected wild-type embryos (A, B, C), agomir-20a injected embryos (D, E, F) and antagomir-20a injected embryos (G, H, I) at dome stage. (B, E, H) Lateral view with animal pole toward the top and dorsal toward the right. (C, F, I) Animal pole views with dorsal toward the right. The injected reagents are indicated at the left of each row. (J) Comparison of endogenous vsx1 mRNA levels among wild type, agomir-20a and angomir-137 injected embryos at different developmental stages. Results are expressed as mean ± SEM, and statistical analyses were done by unpaired t test. *P < 0.001.

To determine how miR-20a represses vsx1 protein expression, we compared vsx1 mRNA profiles among wild type, agomir-20a or control agomir-137 injected zebrafish embryos at different developmental stages. Real-time quantitative PCR analysis revealed that, in comparison with vsx1 mRNA profile in the wild type embryos, the level of vsx1 mRNA was significantly decreased in the agomir-20a injected embryos but was not in the control of agomir-137 injected embryos from 30% epiboly stage onward (Fig. 3J). It is clear that miR-20a can induce the breakdown of vsx1 mRNA by binding to the target site, suggesting that miR-20a represses Vsx1 expression mainly via inducing the breakdown of vsx1 mRNA.

It has been demonstrated that maternal Vsx1 plays pivotal roles in restricting the expression of axial mesoderm gene within dorsal midline and regulating convergent extension movements during early embryogenesis in zebrafish, maternal vsx1 knockdown could result in dorsalized phenotype with disorganized dorsal midline structures and dramatically reduced brain.^21,22 To confirm that miR-20a targets vsx1 mRNA in vivo, we analyzed the phenotype of agomir-20a injected embryos. 86% of agomir-20a injected zebrafish embryos (N = 153) exhibited aberrant convergent extension movements during gastrulation (Fig. 4F and G) and dorsalized phenotype with disorganized dorsal midline structures and dramatically reduced brain at 24 hpf (Fig. 4H–J), resembling the abnormal phenotype of maternal vsx1 knockdown embryos. ^21,22 The control embryos injected with the same dose of agomir-137 developed as normal as the wild type embryos during gastrulation and at 24 hpf (Fig. 4K–O). We further examined the expressions of flh and ntl, the direct downstream target genes of Vsx1, at 40% epiboly stage and bud stage, respectively, in agomir-20a injected embryos by whole mount in situ hybridization (WISH). In comparison with the wild type embryos, the expression of flh in 67% of the agomir-20a injected embryos (N = 21) was ventrally expanded at 40% epiboly stage (Fig. S4A–F), and the expression of ntl in 85% of the agomir-20a injected embryos (N = 26) was expanded in width but shortened in length at bud stage (Fig. S4G–K). These expression patterns of flh and ntl were similar to their abnormal expression patterns in maternal vsx1 knockdown embryos.^21,22 Taken together, these results substantiated that agomir-20a can effectively target vsx1 mRNA and repress Vsx1 expression in vivo.

Figure 4.

Aberrant development of miR-20a overexpression embryos in zebrafish and goldfish. The embryos shown in (A, B, F, G, K, L, P, Q, U, V) are at late gastrula stage. Zebrafish embryos shown in (C–E, H–J and M–O) are at 26 hpf stage. Goldfish embryos shown in (R–T and W–Y) are at 48 hpf stage. (B, D, G, I, L, N, Q, S, V, X) lateral views with anterior to the top and dorsal to the right. (E, J, O, T and Y) are the magnified images of (D, I, N, S and X), respectively, at the head region.

In goldfish, maternal vsx1 mRNA is transcribed from vsx1 A2 whose translation could not be inhibited by miR-20a, vsx1 A1 mRNA, the target of miR-20a, is initially transcribed at segmentation stage.²⁰ Therefore, injection of agomir-20a at one cell stage in goldfish should be unable to elicit aberrant convergent extension movements during gastrulation but could disturb later development. Our results showed that it was indeed the case. Agomir-20a injection at one cell stage in goldfish did not elicited obvious abnormality of the convergent extension movements during gastrulation (Fig. 4U and V), but resulted in impaired head, eye and trunk formation in 87% of the injected embryos (N = 206) at later developmental stages (Fig. 4W-Y). These results demonstrate that miR-20a can target endogenious vsx1 mRNA by binding to the 3′-UTR and repress Vsx1 expression in both zebrafish and goldfish.

Spatiotemporal expression pattern of miR-20a during embryogenesis

MiR-20a has been identified in many vertebrates including in zebrafish and the sequence is conservative in all the examined embryos (http://www.mirbase.org/index.shtml). To investigate whether miRNA-20a has a role in regulating vsx1 translation during embryogenesis, we first examined miRNA-20a expression in goldfish embryos by deep sequencing. Pre-miRNA-20a, matured miR-20a-5p and miR-20a-3p were detected in the 72 hpf embryos. Pre-miRNA-20a contains 152 nucleotides and folds into a hairpin structure (Fig. 5A). High-throughput sequencing data showed that miR-20a-5p and miR-20a-3p covered 91% (2779 read count) and 9% (283 read count) proportion, respectively, of total matured miR-20a (Fig. 5B). The matured goldfish miR-20a-5p contains 23 nucleotides with the same sequences of identified miR-20a in zebrafish (Accession number: MI0001907 in miRBase).

Figure 5.

Identification and expression of miR-20a in goldfish. (A) The sequence of pre-miR-20a detected by next generation sequencing in 72 hpf goldfish embryos. cau-miR-20a-5p is represented in red and cau-miR-20a-3p in light blue. (B) Proportion of cau-miR-20a-5p and cau-miR-20a-3p by high-throughput sequencing. (C) Expression profile of cau-miR-20a at different developmental stages. Error bars indicate standard error of the mean. (D–I) Expression pattern of cau-miR-20a in goldfish embryos at late-gastrula (D, E), 8-somite (F, G) and 20-somite (H, I) stages. (E) Lateral view with animal pole toward the top. (G, I) Anterior-Ventral view with forebrain toward the bottom. (J) In situ hybridization of cau-miR-20a on the eye frozen section of 72 hpf goldfish embryo. GCL: ganglion cell layer; INL: inner nuclear layer; ONL: outer nuclear layer.

Next, we examined the spatiotemporal expression pattern of miR-20a-5p at different developmental stages in goldfish and zebrafish. Real-time quantitative RT-PCR (qRT-PCR) analysis showed that miR-20a-5p was undetectable at blastula and early gastrula stages. Its expression level was rapidly increased from late gastrula stage to 8-somite stage but gradually decreased to a relatively low level at late embryonic stages (Fig. 5C). WISH detected weak but ubiquitous expression of miR-20a-5p at late gastrula stage (Fig. 5D and E), the expression level is significantly increased in the developing eye, brain and anterior spinal cord from 8-somit stage onward (Fig. 5F–I). We also examined miR-20a-5p expression in zebrafish by WISH and observed a similar spatial expression pattern (Fig. S5).

Using Locked nucleic acid modified DNA oligonucleotide probes, which have been shown to increase the sensitivity for the detection of miRNA,^39-41 we visualized the expression pattern of miR-20a-5p in the retina at 72 hpf, at this time the neural retina of the goldfish eye has developed into a layered array of different neuronal types, by in situ hybridization on the frozen sections of the eye. MiR-20a-5p was detected in ganglion cells layer (GCL) and outer nuclear layer (ONL) but was undetectable in inner nuclear layer (INL) (Fig. 5J). This spatial expression pattern of miR-20a-5p in the retina correlated well with that the activity of vsx1 is restricted within the bipolar neurons.

vsx1 A1 3′-UTR mediated protein expression is inhibited when and where miR-20a is expressed

Previous experiments have showed that vsx1 A1 3′-UTR can mediate a dynamic tissue-specific translation pattern of GFP sensor, mimicking the dynamic translation pattern of endogenous vsx1 mRNA in both time and space, during embryogenesis in goldfish.^20,29 To establish that the expression of endogenous miR-20a is associated with the inhibition of vsx1 3′-UTR mediated protein expression during normal development, we examined whether functional inactivation of endogenous miR-20a could induce ubiquitous expression of vsx1 A1 3′-UTR-linked GFP at different developmental stages. At early gastrula stage, ubiquitously expression of GFP was detected in both the control and 1.0 ng antagomir-20a injected embryos (Fig. 6A and D). After 72 hpf, however, the expression of GFP was restricted within the eye of wild type control embryos (Fig. 6B), but was maintained ubiquitously in the miR-20a decoyed embryos (Fig. 6E). In the eye cryosections of embryos at 72 hpf, GFP was specifically expressed in inner nuclear layer (INL) cells in the control embryo ([20] and Fig. 6C), but was expressed in outer nuclear layer (ONL), INL and ganglion cell layer (GCL) cells in the miR-20a decoyed embryos (Fig. 6F). In comparison with the spatiotemporal expression pattern of miR-20a, it is clear that vsx1 A1 3′-UTR mediated protein expression is inhibited when and where miR-20a is expressed during embryogenesis. This result suggests that miR-20a is crucial for establishing a correct spatiotemporal expression pattern of Vsx1 protein.

Figure 6.

Comparison of vsx1 A1 3′-UTR mediated GFP translation patterns in wild type and miR-20a decoyed embryos during embryogenesis. (A, B) Patterns of vsx1 A1 3′-UTR mediated GFP translation in wild type embryos at early-gastrula (A) and 72 hpf (B) stages. (D, E) Patterns of vsx1 A1 3′-UTR mediated GFP translation in antagomir-20a injected embryos at early-gastrula (D) and 72 hpf (E) stages. (C, F) GFP localization detected by immunofluorescent staining in the eye cryosections of wild type (C) and antagomir-20a injected (F) embryos at 72 hpf stage. Dashed lines border the boundaries between different cell layers. L: lense; GCL: ganglion cell layer; INL: inner nuclear layer; ONL: outer nuclear layer.

MiR-20a is essential for normal embryogenesis by preventing ectopic expression of Vsx1 protein

To determine whether miR-20a is essential for normal development, we decoyed miR-20a in goldfish and zebrafish by injecting 1.0 ng and 688 pg antagomir-20a, respectively, into the embryos at one cell stage. No abnormality was observed before segmentation stage in antagomir-20a injected zebrafish and goldfish embryos, consistent with that miR-20a is not expressed at late gastrula stage. Defects of embryogenesis were observable from 6–8 somite stage onward and severely impaired head, eye and trunk were observed in 75% and 53% antagomir-20a injected goldfish (N = 257) and zebrafish (N = 136) embryos, respectively, at later developmental stages (Fig. 7 D–F and J–L). This result indicates that miR-20a is essential for normal embryogenesis in zebrafish and goldfish.

Figure 7.

Functional inactivation of miR-20a elicits aberrant development in both zebrafish and goldfish. Zebrafish and goldfish embryos shown in images are at 26 hpf stage and 48 hpf, respectively. (B, E, H, K) Lateral view with anterior to the top and dorsal to the right. (C, F, I, L) are the magnified images of (B, E, H, K), respectively, at the head region with anterior to the left and dorsal to the top.

Previous investigation observed that zebrafish vsx1 transcripts exist initially in the ventrolateral region of zebrafish embryo at blastula and gastrula stages,²¹ subsequently in the presumptive neurons of hindbrain and spinal cord during segmentation stages and finally in the retina.^{18,23,24,26,27} To determine whether inhibition of Vsx1 expression by miR-20a is essential for normal embryogenesis, we examined whether releasing the inhibition by decoying miR-20a could elicit excessive generation of vsx1 marked neurons. Whole mount in situ hybridization showed that, at 18 hpf, wild type embryos generated 9 distinct vsx1 marked neurons in the central tier of spinal cord and the neurons were arranged in regular space corresponding to the somites (Fig. 8B and E). 87% of antagomiR-20a injected embryo (N = 15) generated additional vsx1 marked neurons (from 13 to 16) and the neurons were arranged irregularly (Fig. 8A). In association with aberrant neuron formation in the spinal cord, no distinguishable somite was formed in miR-20a decoyed embryos (Fig. 8A). To confirm that miR-20a can suppress the generation of spinal cord neurons, we directly examined the generation of vsx1 marked neurons in the central tier of spinal cord in miR-20a overexpression embryos. In 70% of agomir-20a injected embryos (N = 20), the generation of vsx1 mRNA marked neurons in the central tier of the spinal cord was significantly suppressed (Fig. 8C and F). These results demonstrate that miR-20a is essential for preventing ectopic Vsx1 expression and establishing different cell identity during neurogenesis.

Figure 8.

Comparison of vsx1 marked neurons generation in miR-20a decoyed and overexpression zebrafish embryos. All the wild type, antagomir-20a and agomir-20a injected zebrafish embryos are at 18 hpf. (D, E, F) are the magnified image of the black solid box region in (A, B, C), respectively. Arrows indicate vsx1 marked neurons. The injected reagents are indicated at the top of each column.

Discussion

Transcriptional factor Vsx1 plays different roles at different developmental stages. ^18,21-27 It is initially transcribed in a wide embryonic region but gradually restricted within distinct cell types at subsequent developmental stages. Therefore, post-transcriptional regulation might be required for the transition from one developmental stage to the next and the establishment of different cell identities. In this study, we demonstrated that miR-20a can repress Vsx1 expression by binding to the 3′-UTR of vsx1 mRNA in goldfish and zebrafish. Decoying miR-20a resulted in severely impaired head, eye and trunk formation in association with excessive generation of vsx1 marked neurons in the central tier of spinal cord and defect of somites. These results demonstrate that miR-20a is essential for normal embryogenesis through restricting Vsx1 expression in goldfish and zebrafish, and that post-transcriptional regulation is an essential mechanism for Vsx1 playing different roles in diverse developmental processes.

Releasing the inhibition of Vsx1 expression by decoying miR-20a resulted in excessive generation and irregular arrangement of vsx1 marked neurons in the central tier of spinal cord in association with impaired somite formation. These observations suggest that there might be some remaining vsx1 mRNAs transcribed at previous stage in the cells around the neurons in the central tier of spinal cord during neurogenesis. The remaining vsx1 mRNA might promote ectopic neuron generation and disturb head, eye and trunk development if the activity is not inhibited by miR-20a.

It has been shown that miR-20a can target integrin ITGβ8 and is a suppressor of oral squamous cell carcinoma migration.⁴² Decoying miR-20a did not elicit defects of embryogenesis before 6–8 somite stage. The abnormal phenotype resulted from miR-20a overexpression resembled the phenotype of vsx1 knockdown embryos. These observations suggest that inhibition of ITGβ8 by miR-20a is unlikely to have detectable effect on the migration of embryonic cells and morphogenesis. However, it is interesting to investigate whether the inhibition of ITGB8 by miR-20a has a role in organogenesis.

In zebrafish, vsx1 is specifically transcribed in neurons in the central tier of the spinal cord along the entire rostral-caudal axis during neurogenesis.^23,24 Our result showed that miR-20a is ubiquitously expressed during these stages. But it remains unclear whether miR-20a is expressed in the neurons in the central tier of spinal cord. However, overexpression of miR-20a suppressed the generation of vsx1 marked neurons in the central tier of spinal cord, suggesting that miR-20a could not inhibit Vsx1 expression in these neurons. This might be due to the absence of miR-20a or the high level of transcription of vsx1 in these neurons.

During goldfish and zebrafish retinogenesis, vsx1 is initially expressed in undifferentiated, presumptive neural retina, then upregulated selectively in presumptive bipolar cells, and plays roles in regulating the proliferation and differentiation of retinal progenitors as well as the functional maintenance of bipolar cells.^18,23,25,26 In Xenopus levis, it has been demonstrated that miRNAs are used to control the timing of the generation of bipolar neurons by mediating the translational inhibition of vsx1 mRNA in the retinal progenitors at early developmental stages.¹⁵ In this study, we show that miR-20a regulates vsx1 3′-UTR-mediated translation in a tissue-specific manner. MiR-20a is expressed in most of the retinal neuron layers but not in the inner nuclear layer (Fig. 5J). Decoying miR-20a resulted in vsx1 A1 3′-UTR-linked gfp reporter sensor expressed in all the layers (Fig. 6F), not restricted within the inner nuclear layer as observed in the wild type control (Fig. 6C). This result suggests that miR-20a plays a role in sorting the retinal neurons into functional layers during development by controlling the spatial pattern of Vsx1 protein expression.

An interesting question is how the spatiotemporal expression pattern of miR-20a, which correlates negatively with the expression pattern of its target gene vsx1, is regulated during embryogenesis. Previous study has observed that differential expression of miR-163 in related species is resulted from cis- and/or trans-regulatory changes.⁴³ It is possible that the spatiotemporal expression pattern of miR-20a is mainly regulated at transcriptional level and controlled by a set of transcriptional factors. Further studies on the identification of these transcriptional factors and their interaction with the cis-elements at the promoter region of miR-20a gene will help in understanding why the transcription of miR-20a is negatively correlated with vsx1.

Previous study demonstrated that Vsx1 is a substrate of the ubiquitin/proteinsome pathway, suggesting that ubiquitin-dependent proteolysis may regulate Vsx1-dependent cellular differentiation during retinogenesis in zebrafish.⁴⁴ Polyubiquitination of Vsx1 is also likely an essential post-transcriptional regulatory mechanism in dynamically restricting Vsx1 activity within the neurons in the central tier of spinal cord during neurogenesis in fish.

Materials and Methods

Animals and ethics statement

Male and female red-cap goldfish (Carassius auratus) were bred separately in aerated water at breeding season. Zebrafish (Danio rerio) were maintained at 28.5°C in a circulation system of 14 h light and 10 h dark. Goldfish fertilized eggs were obtained by squeezing sperms and eggs simultaneously into a culture dish. Zebrafish fertilized eggs were collected after fertilization. Both goldfish and zebrafish fertilized eggs were dechorionated with 0.25% trypsin in 1 × PBS before microinjection. Goldfish and zebrafish embryos were maintained at their optimal development temperature of 20–22°C and 28.5°C, respectively, and staged by morphology as well as by time as described previously.⁴⁵ The development of goldfish embryos is much slower than that of zebrafish embryos. This study was approved by the Ethics Committee of Laboratory Animal Center of Zhejiang University (Zju201306-1-11-060).

MiRNA target predictions

The vsx1 cDNA sequences were acquired from GenBank (http://www.ncbi.nih.gov) and accession numbers were as follows: NM131333 for zebrafish (Danio rerio), FJ447487 for goldfish (Carassius auratus) vsx1A1 as well as GU471459 for vsx1A2. Three 3′-UTRs were aligned by online program Clustalw2 (http://www.ebi.ac.uk/clastalw). Predictions of miRNA potential binding sites on the 3′-UTR of goldfish vsx1A1 and A2 were performed by combinatorial utilization of 3 different online miRNA target prediction algorithms: PITA (http://132.77.150.113/pubs/mir07/mir07_prediction.html#),³⁰ TargetScanFish Release 6.2³³ (http://www.targetscan.org/fish_62/) and BiBiServ server.³⁴ Seed sequences with one G : U wobble or a single mismatch within the 8 mer seed sequence were considered, and only miRNA-mRNA duplexes with a free energy below –5 kcal/mol were selected. The structure of miRNA-mRNA was imitated by the RNAhybrid (http://bibiserv.techfak.uni-bielefeld.de/rnahybrid/). Online program RNA Movies (http://bibiserv.techfak.uni-bielefeld.de/rnamovies/) was used to visualize the possible secondary structure of 3′-UTR.

Green fluorescent protein reporter plasmids construction

GFP coding sequence was PCR-amplified and inserted into a pCS2 vector, named pCS2-GFP. vsx1 A1 3′-UTR-linked or vsx1 A2 3′-UTR-linked GFP reporter sensors, named pCS2-GFP-vsx1A13′-UTR and pCS2-GFP-vsx1A2 3′-UTR, were generated by subcloning the full-length 3′-UTR of the goldfish vsx1A1 or A2 into the 3′ end of pCS2-GFP.^20,29 A mutant pCS2-GFP-vsx1A13′-UTR plasmid, named pCS-GFP-vsx1A13′-UTR-Mut, was generated by inducing the mutant sequence at the target site of miR-20a with the QuickChange II Site-Directed Mutagenesis Kit (Stratagene) and the primer pair of 5′-AAAAGTCTCCAGAAAGGATTCTAAAGgAgTaTACCTACTGTGATGAT GAATGTTTTG-3′ (forward) and 5′-CAAAACATTCATCATCACAGTAGGT AtAcTcCTTTAGAATCCTTTCTGGAGACTTTT-3′ (reverse, lower cases indicate the mutant sites). All the constructs were confirmed by sequencing and extracted by Endo-Free Plasmid Mini Kit (Omega Biotech Corporation) for microinjection.

Agomirs and antagomirs synthesis

Oligonucleotides agomirs and antagomirs identical to and against the matured miR-20a, respectively, were designed and synthesized by GenePharma. Samples were chemically modified with methoxy, cholesterol and 4 phosphorothioate backbone at 3′ end as well as 2 phosphorothioate backbone at 5′ end. Stock solutions (20 μM/L) were stored at −80°C before use.

Microinjection

All the samples were injected into the blastodisc at 1 to 2-cell stage. For co-injection, the desired samples were mixed thoroughly prior to injection. The proper injection doses of agomir-20a and antagomir-20a were determined by injecting different dose of the reagents and examining the effects. Injection of 688 pg of agomir-20a per embryo in both goldfish and zebrafish elicited high ratio of abnormal embryos with specific phenotype. Injection of 688 pg and 1.0 ng antagomir-20a per embryo in zebrafish and goldfish, respectively, elicited high ratio of abnormal embryos with specific phenotype (due to that goldfish egg is much bigger than zebrafish egg). These doses were used for our experiment.

RNA extraction and deep sequencing

Total RNA was extracted from embryos of different stages utilizing mirVana™ miRNA Isolation Kit (Ambion) according to the manufacturer's instructions and treated by the TURBO DNA-free™ Kit (Ambion) to digest contaminated genomic DNA. RNA isolated from goldfish at 72 hours post fertilization (hpf) was sent for Next Generation Sequencing (Sangon Biotech) in a HiSeq2000 apparatus.

Real-time quantitative RT-PCR

SYBR® primeScript^TM RT-PCR kit (TaKaRa) was used for real-time quantitative RT-PCR of vsx1 with the following primers 5′-AGCCAGCAGGAATGCACAA-3′ and 5′-GAATCGTCCGCTCCATTAG-3′. β-actin was employed as the internal standard, using the primers 5′-TCCCTTGCTCCTTCCACCA-3' and 5′-GGAAGGGCCAGACTCATCGTA-3′. The TaqMan miRNA Assay (Applied Biosystems) was used to determine expression level of mature miR-20a (Assay ID: 000580). cDNA was synthesized using stem-loop RT primers and the TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems). Quantitative real-time PCR reactions was performed in a 384-wall plate at 95°C for initial denaturation for 10 min, followed by 40 cycles of 95°C for 15s and 60°C for 1 min using the LightCycler® 480 system (Roche). The relative expression of miRNA was normalized against U6 snRNA (Assay ID: 001973), calculated by comparative threshold (Ct) method and was plotted as a histogram with GraphPad Prism5 program software (Roche). All PCR was carried out in triplicate.

Whole mount in situ hybridization

Regular WISH was performed as described previously⁴⁶ with minor modifications. 5′-DIG-labeled RNA probe was transcribed with T7/T3 RNA polymerase and digoxigenin mix (Roche) in vitro from the linearized pBluescript II SK plasmid containing the vsx1, flh or ntl fragment. WISH for miRNA was proceed by adapting the protocol of Sweetman.⁴⁷ Double digoxigenin-labeled, locked nucleic acid (LNA) modified oligonucleotide probe for miR-20a (Lot: 38208–15), U6 (Lot: 99002–01) and scrambled miRNA (Lot: 99004–01) control were purchased from Exqion and hybridized at 48°C, 53°C and 56°C, respectively. U6 small nuclear RNA probe and scrambled miRNA probe were considered as the positive control and the negative control, repectively.

Tissue section preparation and in situ hybridization

Embryos were fixed in 4% paraformaldehyde at 4°C overnight, then transferred to 20% sucrose in 1×PBS overnight, subsequently embedded in OCT media and cut as 14 μm sections on a cryostat. Frozen sections were subjected to in situ hybridization or immunofluorescent staining.

ISH for miR-20a in eyes of 72 hpf goldfish on frosen sections was performed as Kloosterman's protocol⁴¹ with some adaptations. Reagents and apparatus used were DEPC treated. Hybridizations were conducted overnight at 48°C and 4°C. LNA miR-20a probe labeled with double digoxigenin was obtained from Exqion. All the slides were mounted with a drop of mounting medium. The in situ signal was visualized by microscopy and recorded.

Immunofluorescent staining

Sample cryosections were fixed in ice-cold actone for 10 min, washed twice with ice-cold 1×PBS and blocked in 10% goat serum/PBS for 1 h. Sections were incubated overnight with GFP antibody (Lot: cw0258, Cwbiotech) diluted at 1:500, followed by incubation with FITC conjugated secondary antibody (Lot: cw0113, Cwbiotech) diluted at 1:100. Images were captured on a Nikon camera after DAPI staining and slide mounting.

Immunohistochemistry

Immunohistochemistry was manipulated by standard procedures²¹ with polyclonal antibody against ZF-VSX1 (1:500, Abmart) and goat anti-IgG secondary antibody (1:1000, Lot: 05181013, Cwbiotech). DAB was gained from SIGMA (Lot: D4293).

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by funds of NSFC30971654 and State Key Basic Research Project of China (2010CB126301). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Supplemental Material

Supplemental data for this article can be accessed on the publisher's website.