Zebrafish (D. rerio, AB strain) were maintained at 28.5°C under standard conditions. Embryos were raised and maintained at 28.5°C and staged according to standard morphological criteria (Kimmel et al., 1995). All handling of fishes was carried out in accordance with the guidelines on the care and use of animals for scientific purposes set up by the Institutional Animal Care and Use Committee (IACUC) of the Shanghai Ocean University (SHOU), Shanghai, China. This research was approved by the IACUC of SHOU.

Embryos were collected at timed developmental stages. Total RNA was extracted from whole embryos using TRIzol Reagent according to the manufacturer’s protocol (Invitrogen). For quantification of miRNAs, miRNA-specific stem-loop RT primers were designed with the software primer 5.0. The isolated RNA was reverse transcribed into cDNA by miRNA-specific stem-loop RT primers and PrimeScript® RT reagent Kit (Takara). qRT-PCR was performed using the miRNA-specific stem-loop RT primers and SYBR Green Master Mix following the manufacturer’s protocol (Takara). For quantification of pri-miR-202 and protein coding transcripts, total RNA from the embryos were reverse transcribed using random primers supplied in the PrimeScript® RT kit following the same protocol as above (Takara). All samples were performed in triplicates, and expression level of target genes was calculated with the 2^-△△CT. U6 was used as the internal control. The primers used are listed in Supplementary Table S5.

Antagomirs to miR-202-3p, miR-202-5p and the scrambled antagomir (as negative control) were designed and synthesized by GenePharma (China). The sequences are provided in Supplementary Table S5. Fertilized eggs from wild type zebrafish at the one-cell stage were injected with 1 nl of each antagomir (8 µM) by using a microinjector (Eppendorf). The injected embryos were maintained at 28.5°C for development.

Deletion of the miR-202 locus from the zebrafish genome was carried out using the CRISPR-Cas9 system. CRISPR-Cas9 target sites were designed using an online tool ZiFiT Targeter software (http://zifit.partners.org/ZiFiT). Two gRNAs were chosen to delete the miR-202 locus; the primers are listed in Supplementary Table S5. Capped Cas9 mRNA was synthesized in vitro by mMESSAGE mMACHINE T7 ULTRA kit (Ambion), and purified using RNeasy Mini Kit (Qiagen). gRNAs were synthesized using MAXIscript T7 kit (Ambion) following the manufacturer’s protocol and purified. Approximately 400 pg mRNA encoding Cas9 and 100 pg gRNA were injected into each embryo. The embryos were raised and maintained at 28.5°C.

To screen F0 miR-202 mutant zebrafish, genomic DNA was isolated from embryos produced by crossing microinjected F0 zebrafish with wild type partners. The target region was amplified by PCR using the specific primer pairs that were designed to distinguish wild type and mutated alleles (Supplementary Figure S1, Supplementary Table S5). The F0 parents who produced the miR-202 mutant embryos were identified. Mutation status of their miR-202 locus was further verified through PCR amplification and sequencing.

The embryonic fatality of the homozygous miR-202 mutant rendered unavailable sexually mature miR-202^−/− individuals for reproduction. Therefore, investigation of miR-202^−/− phenotypes and underlying mechanisms relied on precise genotyping of embryos produced from heterozygous miR-202 parents. Genomic DNA was isolated from a single embryo using the alkaline lysis method: a timed embryo produced by heterozygous parents was submersed in 20 µl of 50 mM NaOH and heated to 95 °C for 10min. The tube was then vortexed and heated again, and 2 µl of Tris-HCl (1 M, pH = 8.0) was added to neutralize the solution. The tube was centrifuged and the supernatant was collect for PCR amplification with the proper primer (Supplementary Figure S1A). In the cases when embryos were taken prior to 10 hpf, the nested PCR was used for genotyping (Supplementary Figure S1B). The primers are listed in Supplementary Table S5.

Agomirs for miR-202-3p, miR-202-5p and pre-miR-202 were chemically synthesized (Sangon Biotech) based on their native sequences (Supplementary Table S5). Series of dilutions of each agomirs or mixture or precusor (10 μM, 20 and 30 µM) were microinjected into the one-cell fertilized eggs obtained from the miR-202 heterozygous parents with 1 nL using a microinjector (Eppendorf). Developmental status of the injected embryos were observed real-time under a stereomicroscope (Zeiss). Embryos that survived to 12 hpf were picked out, counted, and genotyped.

Genomic DNA and total RNA were concurrently isolated from single embryo collected from mating of miR-202^+/- parents at 3.5 hpf. After genotyping of the embryos with DNA, the total RNAs from same-genotype embryos, namely miR-202 homozygous, or heterozygous, or wild type embryos were pooled to gain a sufficient amount of total RNA for each genotype for sequencing. RNA-seq was performed by NovoGenes (Tianjin, China). RNA-seq reads were trimmed using Trimmomatic (Bolger et al., 2014) (Ver. 0.33 AVGQUAL:20 TRAILING:20 MINLEN:50). The clean Illumina paired-end reads of each sample were mapped to the annotated zebrafish genome (GRCz10) using HISAT2 aligner (Kim et al., 2015) (Ver. 2.0.4). Cufflinks was used to count the reads for each gene and transformed to FPKM. Differentially expressed genes (DEGs) between the genotypes were determined using the edgeR (Robinson et al., 2010) package developed in R. Compare homo_3.5 h with hete_3.5 h and wt_3.5 h, respectively, for log2 fold change >1 or < −1 and p_value <0.05 was defined as differentially expressed genes (DEGs). DEGs related with the maternally inherited mRNAs were identified by adopting the following criteria: 1) if FPKM (wt_0 h≥homo_3.5 h≥max (hete_3.5h, wt_3.5 h), the gene was taken to indicate insufficient degradation (ID); 2) if FPKM (homo_3.5 h > max (hete_3.5h, wt_3.5h, wt_0 h), the gene was associated with over-expression (OE); and 3) if FPKM (min (hete_3.5h, wt_3.5 h)≥homo_3.5 h≥wt_0 h), the gene was regarded as insufficient expression (IE). Almost no genes were over degraded in miR-202^−/− embryos and were thus not considered for GO and KEGG enrichment.

The wild type and abnormally developing embryos of miR-202^+/- pairs were collected at 4hpf, and the embryos were removed from the egg shell. Three groups of normal WT (A_4hpf) and three groups of abnormal embryo (D_4hpf) samples (each having about 50 embryos) were used to extract protein for proteomic analysis. Proteomic analysis was performed using LC-MS/MS on a QExactive mass spectrometer with an Easy-nLC system (Thermo Fisher Scientifc). The LC-MS/MS data were analyzed using Proteome Discovery (Version 2.2, Thermo Fisher Scientific) with the zebrafish Uniprot database (uniprot-danio + rerio_170221.fasta). To quantify protein, the abundance value was normalized with the median value of the whole protein set and only unique peptides were used. Differential protein screening was performed at criteria of 1.2 and 0.833 fold change (FC).

A digoxigenin (DIG) labelled RNA probe of miR-202-3p (accession number: MIMAT0001864) and a scrambled RNA probe (NC) were synthesized by Exiqon (Denmark). Whole mount in situ hybridization (WISH) was performed as previously described (Thisse et al., 2008). Shell-removed embryos were fixed in 4% PFA (paraformaldehyde) at 4°C overnight, dehydrated in methanol and rehydrated by a series of methanol/PBST gradients, and then treated with proteinase K and re-fixed in 4% PFA. Embryos were pre-hybridized with hybridization mixture (HM) at 58°C for 2–4 h and hybridized with DIG-labelled miR-202-3p anti-sense probe or scrambled probe (NC) at 58°C overnight. After hybridization, embryos were washed in a series of saline sodium citrate (SSC) gradients. Subsequently, the embryos were blocked in MAB buffer with 1% blocking solution (Roche) for 3 h at room temperature and incubated in alkaline phosphatase conjugated anti-DIG antibody (1:5,000 diluted in blocking solution, Roche) at 4°C overnight. The embryos were washed four times in PBST for 15 min, and the signal was developed using NBT/BCIP Staining solution. The images were documented with a stereomicroscope (Zeiss) equipped with a digital camera.

In situ hybridization on paraffin sections was performed as described previously (Jørgensen et al., 2010). Briefly, ovary tissue or embryos were dissected in 1 × PBS and fixed in 4% PFA overnight at 4°C. Fixed tissues were embedded in paraffin and sectioned using a paraffin slicer microtome (Leica) at 10-µm thickness and transferred to special coating glass slides (Leica). Slides were hybridized overnight with 1 μg/ml digoxigenin-labeled probe at 65°C in HM solution. After washing in SSC buffer, slides were incubated with alkaline phosphatase-coupled anti-digoxigenin antibodies overnight at room temperature. Slides were then dehydrated through ethanol series and xylene (Sigma-Aldrich) then mounted using Entellan (Electron Microscopy Sciences). Images were acquired using a confocal microscope (Zeiss).

F1 heterozygous zebrafish adults were crossed in the appropriate breeding tanks. EGFP mRNA was transcribed in vitro as above. The fertilized eggs (one-cell stage) were collected immediately and injected with EGFP mRNA (100 pg) and then cultured at 28.5°C for 3 hours. Embryos were screened by fluorescence analysis with a stereomicroscope (Zeiss). Embryos with bright and weak green fluorescence intensities were selected for genotyping verification.

F1 heterozygous zebrafish adults were crossed in the appropriate breeding tanks. Fertilized eggs were collected and stored at 28.5°C to allow the embryos to develop for 3 hours. Embryos were then washed with E3 medium (Cold spring Harbor Protocols 2011, pdb. rec66449, doi: 10.1101/pdb.rec066449 (2011)) and then immediately incubated with a general Oxidative Stress Indicator (CM-H2DCFDA) (Invitrogen) at a final concentration of 3 µM. Embryos were incubated in the dark for 15 min at 28.5°C. At the end of the incubation, the ROS-detection solution was immediately removed and embryos were washed three times with E3. Fluorescence intensity of the embryos was analyzed by stereomicroscope (Zeiss). Embryos with bright and weak green fluorescence were selected for genotyping verification.

SYTOX nuclear green stain is impermeable to living cells, but stains nuclei in a syncitium (or otherwise following membrane degradation) (Goonesinghe et al., 2012). To visualise migration of YSL nuclei relative to the blastoderm margin during epiboly, embryos from heterozygous parents were injected with 1 nl of 0.5 mM Sytox Green fluorescent nucleic acid dye (Invitrogen, United States) into the yolk cell at 3 hpf and then visualised at 4 hpf and 6 hpf under a fluorescence stereomicroscope (Zeiss). Embryos were kept in E3 medium for genotyping verification. Images were captured and processed using a Zeiss AxioCam MR and AxioVision 4.5 software.

F2 embryos from heterozygous parents were used for immunofluorescence staining. Embryos were fixed overnight in 4% PFA at 4°C, and then were peeled off the egg shells. Embryos were permeabilized in 0.5% Triton-X-100 for 30 min at room temperature. After 1 h blocking in 1% BSA/PBS at room temperature, embryos were incubated overnight at 4 °C with primary antibody anti-ZO-1 at 1:200 (Thermal Fisher Scientific). After three washes, embryos were incubated with secondary antibody at 1:3,000 (Thermal Fisher Scientific) for 2 h at room temperature. TSA-F green fluorescent dye staining (1:100) was used to amplify signals by incubation at room temperature for 30 min in the dark, and then washed with PBS for at least 1 h. DAPI (500 ng/ml) was added to counterstain the nuclei followed by washing with PBS for three times. Photographs were taken using confocal microscopy (Zeiss). Following photography, the embryos were then genotyped individually.

F2 embryos from heterozygous miR-202 parents were used for TUNEL staining. Embryos were fixed in 4% PFA at 4°C overnight, then removed egg shell from embryos. TUNEL staining was performed using a commercially available kit (Thermo Fisher Scientific) by following the manufacturer’s instructions. The embryos were stained with FITC-dUTP Labeling Mix and DAPI (500 ng/ml), then were analyzed under a laser confocal microscope (Zeiss). After being photographed, each embryo was genotyped.

To validate the function of the miR-202-3p target genes in the embryonic lethal phenotype of miR-202-3p deficiency, shRNAs against the Seven genes (nfkbiaa, perp, mgll, atp1b1a,nfil3-5, pleca, nfe2l2b) which were predicted to be miR-202-3p targets and upregulated in the miR-202^−/− embryos were designed through the BLOCK-iT™ RNAi Designer (Thermo Fisher Scientific), chemically synthesized (Sangon Biotech) and cloned to pLKO.1 plasmid (Addgene). shRNA plasmid was microinjected into WT embryos together with the miR-202-3p antagomir, in a final concentration of 200 ng/μL and 8 nM, respectively. Then injected embryos were raised and maintained at 28.5 °C and observed for developmental status; rescue rate was calculated for every 2 h. The shRNAs are listed in Supplementary Table S5.

We manipulated the mRNA contents of miR-202-3p target genes, nfkbiaa, perp and mgll in developing embryos for validating the function of these genes in embryonic development. To down-regulate a gene, a single type or a mix of the plasmid constructs containing specific shRNA was microinjected into the wild type embryos, in a final concentration of 200 ng/μl, and 100 ng/μl for each one in the mixture. To increase the mRNA content of a specific gene in developing embryos, mRNA was microinjected into wild type embryos with final concentration of 400 ng/μl for a single gene or 200 ng/μl for each one in the mixture. Each embryo was injected in 1 nl volume. Embryos after injection were cultured at 28.5°C and observed for developmental status every hour. The shRNA sequences and primers for cDNA amplification are listed in Supplementary Table S5.

The developmental stages of embryos were examined using a stereomicroscope (Zeiss) by observing the morphological appearance. An embryo was considered to be dead if lysed cells were visible under the microscope. An embryo was considered to be abnormal if development was slower than in the wild type, and developmental termination was registered for an embryo if no morphological progression was observed within a period of 1 hour. Rescue rate is calculated through a two step procedure: 1) counting all live embryos at 12 hpf and genotyping each embryo; 2) calculating the ratio of the living miR-202^−/− embryos to the total number of embryos examined. The ratio is regarded as the rescue rate of a reagent because through large scale phenotype and genotype analyses, we had established that no miR-202^−/− embryos would survive beyond 12 hpf without rescue.

Dual luciferase assay was carried out to validate the authenticity of the predicted target genes of miR-202-3p. Native and mutated 3′UTRs of the candidate genes were amplified from zebrafish embryonic cDNA and cloned into the pmirGLO Dual-Luciferase miRNA Target Expression Vector (Promega) and sequenced. HEK293T cells were plated in a 96-well plate and incubated at 37°C for 24 h miR-202-3p agomir (or scrambled agomir) and pmirGLO-3′UTR (or mutated 3′ UTR) construct were co-transfected into the HEK293T by using Attractence Transfection Reagent (QIAGEN). The transfected Cells were continuously incubated at 37 °C for 24 h, and luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega) following the manufacturer’s protocol in a luminometer. Data were first normalized to residual luminescence then to an agomir negative control. The 3′UTRs or mutation sites and primers are shown in Supplementary Figure S3 and Supplementary Table S5.

Statistical analysis was conducted using the Student’s t-test (two-tailed). All values are shown as mean ± s.d. p < 0.05 were considered statistically significant. One asterisk, two asterisks and three asterisks indicate < 0.05, p < 0.01 and p < 0.001, respectively.

To test whether miR-202-3p and miR-202-5p play a role in early embryonic development, we microinjected the antagomirs of miR-202-3p and miR-202-5p respectively into fertilized embryos. Injected embryos developed normally at the initial stages, similar with the control group (injected with a scrambled antagomir) and the wild type embryos. However, starting from 4 hpf, embryos injected with the miR-202-3p antagomir demonstrated developmental stoppage and the blastomere disassociated from 6 hpf to 12hpf (Figure 1A). Once the threshold amount of miR-202-3p antagomir (8 µM) was reached, developmental failure occurred in over 90% of the embryos during the blastula stage. In sharp contrast, no developmental abnormality was observed in the embryos injected with the miR-202-5p antagomir or the control antagomir at the same amounts (Figure 1B). The time course statistics for blastomere cytolysis in injected embryos are shown in Figure 1C, which is based on more than two thousand injected embryos for each antagomir. qRT-PCR was performed to evaluate the efficiency of miRNA knockdown with their respective antagomirs; both miR-202-3p and miR-202-5p antagomirs functioned effectively (Figure 1D). These results suggested that miR-202-3p might be essential for embryonic development at the mid-blastula stage.

To ensure that the blastula lethality phenotype that occurred in the miR-202-3p antagomir injected embryos is an authentic function of this miRNA, we targeted the miR-202 locus for deletion using the CRISPR-Cas9 technology in zebrafish. Two founders (a male and a female) of zebrafish with deletions of 833 bp or 837 bp fragment were generated (Figure 2A). Heterozygous F1 lines were generated by mating the two founders to wild type female or male fish. The miR-202^+/- fishes developed normally like the wild type fish. However, the genotype of F2 embryos were inherited at the expected Mendelian ratio at 4 hpf, while no zygous miR-202 deletion zebrafish can survive to adulthood, indicating that miR-202 is crucial for zebrafish development (Figure 2B). In situ hybridization of ovary sections prepared from the wild type and miR-202^+/- fishes showed that in both fishes, miR-202-3p was expressed in the developed oocytes with similar intensities but absent from the underdeveloped oocytes (Figure 2C), further suggesting maternal carryover of miR-202-3p at the initial developmental stages. By cross-mating between the miR-202^+/- F1 fishes, developmentally timed F2 embryos were collected at 4 hpf. Whole mount in situ hybridization on these embryos demonstrated differential staining intensities for miR-202-3p, corresponding to the wild type, miR-202^+/- and miR-202^−/− genotypes, respectively (Figure 2D). qRT-PCR analyses on genotyped embryos clearly indicated reduced miR-202-3p expression in the miR-202^+/- embryos and only residual levels of miR-202-3p in the miR-202^−/− embryos at 4 hpf, which was carried over from the oocyte. However, it could not detect the miR-202-3p in the miR-202^−/− embryos at 6 hpf (Figure 2E).

The miR-202^+/- F1 females were paired with F1 miR-202^+/- males to examine the viability of embryos. While the majority of F2 embryos developed normally, about 1/4 stopped development and become visibly abnormal starting from 4 hpf, followed by cytolysis within next few hours (Figure 3A). This fatal phenotype highly resembled that of the embryos in which miR-202-3p is knocked down in the timing of occurrence and severity of the phenotype. In the more than 2000 F2 embryos genotyped, more than 90% of miR-202^−/− embryos demonstrated developmental termination and cytolysis within 4–6 hpf, and no miR-202^−/− embryos survived beyond 12 hpf (Figure 3B). No developmental abnormality occurred in the miR-202^+/− embryos at this stage.

To verify whether the fatal miR-202^−/− phenotype was a bona fide outcome from miR-202 deletion, we performed rescue attempts by injecting agomir of miR-202-3p, miR-202-5p, pre-miR-202, a combination of miR-202-3p and miR-202-5p, and a scrambled miR-202-3p (i.e., NC) to the F2 embryos. By extensive genotyping of viable embryos at 4, 6, 8, 10, and 12 hpf for each rescue agent, we found 80 ± 12%, 75 ± 12% and 77 ± 16% of miR-202^−/− embryos demonstrated delayed development not the cytolysis at the 12 hpf time point for miR-202-3p, pre-miR-202, the combination of miR-202-3p and miR-202-5p, respectively, while none of the miR-202^−/− embryos were rescued by miR-202-5p and the scrambled miRNA (Figures 3C,D). Thus the fatal phenotype of miR-202^−/− at the blastula stage could be rescued if miR-202-3p is present in the rescue agents. Rescuing efficiencies of miR-202-3p and pre-miR-202 both peaked at 20 μM, with under- and over-doses resulting in two to three folds lower rescue rates (Figure 3E). Interestingly, both overdosed and inadequate amounts of miR-202-3p resulted in severely delayed embryo development and embryos underwent cytolysis at the blastula stage. These results indicate that miR-202-3p is required for embryonic viability and its expression be precisely regulated during MBT.

To elucidate mechanisms of miR-202-3p function in early development, we conducted transcriptome comparisons among the WT, miR-202^+/−, and miR-202^−/− embryos collected at 3.5 hpf. RNA and DNA from single embryos were isolated concurrently and genotyping was carried out using the DNA. RNA from embryos of same genotype was pooled for RNA-seq. As expected, gene expression profiles from the miR-202^−/− embryos showed greater divergence from the heterozygous and wild type embryos while the latter two are more similar (Figure 4A). The expression levels of the DEGs were then compared with wild type embryos at the one-cell/fertilized egg stage to determine how the DEGs were related with the maternally inherited mRNAs. This comparison allowed us to divide the DEGs into three subgroups: insufficient degradation (ID) of maternal RNA, overexpression (OE) of zygotic genes, and insufficient expression (IE) of zygotic genes in the miR-202^−/− embryos (Figure 4B, Supplementary Table S1). There are 54 genes including 7 ribosomal proteins that were insufficiently degraded, 158 genes, such as nfkbiaa, perp, mgll involved in cell proliferation, apoptosis and cell-cell adhesion are over-expressed, and 43 genes insufficiently expressed (Figure 4C). We further analyzed the three subgroups of DEGs for KEGG enrichment; 15 pathways including ribosome, oxidative phosphorylation, apoptosis, cell junction, inflammatory and metabolic related pathways were identified (Figure 4D, Supplementary Table S2). A larger portion of DEGs (63%) belong to the over-expressed subgroup, suggesting a suppressive role of miR-202 during zygotic genome activation. Due to failure in the initiation of epibolic movement in miR-202^−/− embryos, we are curious whether loss of miR-202 would affect the formation the Yolk Syncytial Layer (YSL), the structure critical for initiating epibolic cell movement and reorganization. We found transcription of the YSL marker genes (Xu et al., 2012) was normal in miR-202^−/− embryos (Figure 4E). Staining of the F2 embryos using SYTOX also showed proper formation of YSL in all embryos including the miR-202 mutants at 4 hpf. However, with development progressed, YSL in the wild type embryos was shown to have a complete epibolic movement of cells towards the vegetal pole, but epibolic movement failed in the miR-202^−/− embryos, which clearly pinpointed the timing of developmental termination was prior to epiboly in the mutant (Figure 4F).

To evaluate the proteomic consequences of miR-202 deletion, we analyzed the proteome of miR-202^−/− embryos and compared it to the wild type at 4 hpf. We observed a total of 115 differentially expressed proteins, of which 100 were down-regulated and 15 were up-regulated in miR-202^−/− embryos (Supplementary Table S3). A most peculiar feature of the miR-202 null embryonic proteome is the reduction of 34 ribosomal proteins (Figure 4G, Supplementary Table S3).

The identification of multiple dysregulated KEGG pathways from the transcriptome comparisons prompted us to evaluate the cellular consequences that the loss of miR-202 might produce. Due to the striking presence of dysregulated ribosomal proteins in the transcripts and proteome, we first verified whether protein synthesis was impaired. Exogenous introduction of polyadenylated EGFP mRNAs to the embryos produced by miR-202^+/- parents showed almost complete abolishment of protein synthesis in the miR-202 null embryos (Figure 5A). The level of reactive oxygen species (ROS) was also measured to evaluate whether oxidative stress is an outcome of the dysregulated pathways. We microinjected CM-H2DCFDA, an indicator of cellular ROS level into the embryos from the miR-202^+/- parents. About 4-fold higher ROS intensities were detected in the miR-202 null embryos (Figure 5B). In addition, cell-cell adhesion was also affected as demonstrated by the significantly reduced presence of zonula occludens-1 (ZO-1), the marker for tight junction (Figure 5C).

Overexpression of the apoptotic pathway in the miR-202^−/− embryos hinted that apoptosis could be the ultimate fate of the miR-202^−/− embryos. We performed TUNEL staining on sections of the F2 embryos followed with genotyping of the embryos. Apoptotic signals were widely detected from the blastomere cells in the miR-202^−/− embryos as compared to no detected signals in the miR-202^+/- and wild type embryos (Figure 5D). More than 90% of the cells in the blastomere at 3.5 hpf were visibly undergoing apoptosis in the miR-202^−/− embryos. Taken together, results strongly indicated that miR-202, through its precisely controlled product miR-202-3p, is essential to maintain cellular homeostasis during MBT.

Using miRNAMap2 (Hsu et al., 2008) and TargetScanFish (Ulitsky et al., 2012), we identified 24 possible target genes of miR-202-3p from the DEGs (Supplementary Table S4). We then selected seven upregulated genes (nfkbiaa, mgll, nfil3-5, atp1b1a, pleca, nfe2l2b and perp) for further analysis based on their potential involvement in cell proliferation and apoptosis, cell adhesion, and metabolism (Table 1). Since all these genes are up-regulated when miR-202-3p is deleted or knocked down, we first tested whether minimizing of the elevation of these transcripts could alleviate the fatal phenotype when a miR-202-3p antagomir was present. We thus co-injected miR-202-3p antagomir with a designed shRNA targeting one of the seven genes (Figure 6A). We verified that the gene-specific shRNAs functioned properly to suppress the up-regulation of the intended genes before and after 4 hpf (Supplementary Figure S2). We counted the number of normally developing embryos for each of the shRNAs at 6 hpf–the timepoint when the miR-202-3p antagomir would have led to cytolysis if no shRNA had been injected. We found that down-regulated expression of three genes: nfkbiaa, perp and mgll was able to rescue about 80, 50 and 45% of the miR-202-3p knockdown embryos from developmental failures at MBT, respectively (Figure 6B).

To verify that these three genes are direct targets of miR-202-3p, native and mutated forms of the 3′UTRs of these genes were cloned to a luciferase report vector (Supplementary Figure S3). Both native and mutated vector were co-transfected with miR-202-3p agomir and scrambled agomir into HEK293T cells to measure luciferase activities. Results validated that nfkbiaa, perp and mgll are direct targets of miR-202-3p (Figure 6C).

We further investigated the inter-relationship among the trio of miR-202-3p target genes. We first examined the expression patterns of the trio in wild type embryos and in miR-202^−/− embryos by qRT-PCR. Data showed that all three genes were transcribed around 4 hpf during ZGA (Figures 7A–C) and were overexpressed when miR-202 is deleted (Figure 7D), coinciding the timing of the fatal phenotype of the miR-202^−/− mutant. To explore the relative roles of the trio in generating the phenotype, we cloned the transcripts of these genes, in vitro transcribed them, and microinjected them into fertilized eggs of wild type fish. Examining at 12 hpf, approximately 80%, 50%, 45% and 20% of injected embryos exhibited developmental delay or embryonic mortality when mixture of the three target gene mRNAs, nfkbiaa mRNA, perp mRNA and mgll mRNA were introduced, respectively (Figure 7E). These results further validated involvement of all three genes in producing the phenotype, but nfkbiaa and perp played bigger roles in the process than mgll. Interestingly, similar ratios of developmental delay and embryonic mortality were observed when the trio were supressed by their respective shRNAs or a mixture of the three shRNAs in wild type embryos (Figure 7F). The similar fatal phenotypes produced by over- and down-regulation of the trio indicated transcription of the trio, especially those of nfkbiaa and perp, need to be tightly controlled for proper embryogenesis to proceed. These findings suggested the need of tight control of miR-202-3p level during MBT.

As the similar phenotypes resulted from over- or down-regulation of the individual genes hinted potential existence of inter-regulation among the trio. The notion was proved true as we found that when nfkbiaa was over-expressed or down-regulated, transcription of perp responded in the same direction during the developmental period from 2-6 hpf (Figure 8A). Similarly, over-expression or down-regulation of perp elicited the same responses from nfkbiaa (Figure 8B). Therefore, nfkbiaa and perp are inter-regulated. An inter-regulation relationship between genes was also true between nfkbiaa and mgll, but in a slightly different manner, in which over- and under-expression of nfkbiaa resulted in down-regulation of mgll (Supplementary Figure S4). Our results showed that nfkbiaa, perp and mgll form an interconnected regulatory network and the two major factors, nfkbiaa and perp are positively regulating on each other.

Taking all results together, we concluded that miR-202-3p suppressed over-expression of nfkbiaa, perp and mgll during MZT. These three target genes formed interconnected regulatory networks with each other. Reciprocally, nfkbiaa and perp regulate expression of the miR-202 locus, these inter-regulation loops acted concertedly to maintain miR-202-3p at a relatively constant level (Figure 9). This tightly regulated miR-202-3p-mediated network is essential to maintain embryonic viability during MBT.