The genetic program of oocytes can be modified in vivo in the zebrafish ovary

Abstract

Oocytes, the irreplaceable gametes for generating a new organism, are matured in the ovary of living female animals. It is unknown whether any genetic manipulations can be applied to immature oocytes inside the living ovaries. As a proof-of-concept, we here demonstrate genetic amendments of zebrafish immature oocytes within the ovary. Oocyte microinjection in situ (OMIS) stimulates tissue repair responses, but some of the microinjected immature oocytes are matured, ovulated and fertilizable. By OMIS-mediated Cas9 approach, ntla and gata5 loci of oocytes arrested at prophase I of meiosis are successfully edited before fertilization. Through OMIS, high efficiency of biallelic mutations in single or multiple loci using Cas9/gRNAs allows immediate manifestation of mutant phenotypes in F₀ embryos and multiple transgenes can co-express the reporters in F₀ embryos with patterns similar to germline transgenic embryos. Furthermore, maternal knockdown of dnmt1 by antisense morpholino via OMIS results in a dramatic decrease of global DNA methylation level at the dome stage and causes embryonic lethality prior to segmentation period. Therefore, OMIS opens a door to efficiently modify the genome and provides a possibility to repair genetically abnormal oocytes in situ.

oocytes, microinjection, gene editing, transgenesis, morpholino, zebrafish

Introduction

The fertilizable eggs are developed through oogenesis. During mammalian oogenesis, all germ cells in the embryonic ovaries start meiosis to become primary oocytes but get arrested at prophase I of meiosis following chromosome replication (Edson et al., 2009). The first meiosis of mammalian primary oocytes in groups periodically resumes after puberty, which leads to the ovulation of one egg (secondary oocyte) and one polar body. The released egg can proceed into metaphase II of meiosis and become arrested again. After fertilization, the egg starts to complete the second meiosis and releases another polar body. Unlike oogenesis in mammals, germ cells in the zebrafish adult female still proliferate and some of them enter meiosis. Oogenesis process in the zebrafish can be divided into five stages according to their morphology and karyotype in the ovary (Selman et al., 1993; Lubzens et al., 2010) (Supplementary Figure S1A). During stages I and II, oocytes accomplish G1, S, and G2 phases of the first meiosis with the formation of the follicle layers surrounding the oocytes. Stage III oocytes are growing in size and arrested at prophase I, which may last several weeks if the female does not mate to a male fish. When mated to a male, stage III oocytes proceed quickly through stage IV with further increase in size during night, and stimulated by light in the morning, the first polar body is released and stage V oocytes are ovulated with meiotic arrest at metaphase II. Like mammalian eggs, zebrafish ovulated eggs resume the second meiosis with the release of the second polar body following fertilization.

The whole process of oocyte maturation is precisely regulated by pituitary gonadotropins and sex steroids and signal crosstalk between the maturing oocytes and their surrounding somatic cells (Mehlmann, 2005; Nagahama and Yamashita, 2008; Coticchio et al., 2015), which makes it difficult to obtain high-quality mature oocytes for fertilization by culturing immature oocytes in vitro (Farsi et al., 2013; Telfer and Zelinski, 2013; Guzel and Oktem, 2017). Maternal factors that are encoded by maternal-effect genes and deposited in the oocytes play important roles in early development of vertebrate embryos, and their abnormal expression may lead to abortion, birth defects, and even adult diseases (Li et al., 2010; Marlow, 2010; He-Feng Huang, 2014). Conceptually, immature oocytes with genetic defects may be repaired within the ovaries if the genetic program of such oocytes could be artificially modified. However, it is unknown whether the genome of oocytes after chromosome replication during the first meiosis can be genetically manipulated in living females.

In the present, gene functional studies, cell lineage tracing, and human disease modeling usually involve the generation of mutant or transgenic animals by adopting zinc-finger nuclease (Urnov et al., 2005; Remy et al., 2010; McCammon et al., 2011), transcription activator-like effector nucleases (Joung and Sander, 2013), Cas9 nucleases (Hsu et al., 2014; Sternberg and Doudna, 2015; Varshney et al., 2015), and transgenic overexpression (Meng et al., 1999; Gama Sosa et al., 2010) approaches. These manipulations are performed usually by microinjecting reagents into fertilized eggs and occasionally by microinjecting reagents into mature oocytes arrested at metaphase II. Following fertilized egg microinjection (FEM), gene editing or gene integration events may occur within some of embryonic cells at multi-cell stages, resulting in mosaic animals without phenotypes or faithful transgene expression patterns. Therefore, mutant phenotypes or transgene expression patterns are usually observed in F₁ or F₂ generations, and the whole process is time-consuming and labor-intensive. If immature oocytes can tolerate and allow proper functioning of transgenesis or gene editing materials, animals derived from these oocytes (F₀) would straight away display correct mutant phenotypes or transgene expression patterns. Importantly, successful genetic modification in oocytes may allow recovery of defective oocytes.

The zebrafish has become a widely used vertebrate model for studying molecular mechanisms controlling development of embryos. In this study, using the zebrafish model, we aim to test (i) whether immature oocytes in the ovary of living female can tolerate microinjection with no adverse effects on maturation and fertilization; (ii) whether the genome of immature oocytes can be edited before fertilization; and (iii) whether microinjection of gene editing, transgenesis or gene knockdown materials into immature oocytes allows immediate manifestation of phenotypes or efficient expression of a transgene in F₀ embryos derived from microinjected immature oocytes.

Results

Immature oocytes are resistant to microinjection-induced tension

We first asked whether microinjection of immature oocytes inside the living female affects their maturation and fertilization. To this end, we chose adult female fish at 5–12 months old according to their health and fecundity. In the afternoon on Day 1, we anesthetized a female and microsurgically opened its belly to allow visualization of oocytes under a dissection microscope (Supplementary Figure S1B). Then, the cut was sewed immediately, and the operated female was placed into a tank of fish water with low doses of tricaine and antibiotics (penicillin and streptomycin) and allowed to rest for recovery (Figure 1A and B). In the evening next day (Day 2), the operated female was transferred into a mating tank together with a male fish (Figure 1A and Supplementary Figure S2A). In the morning on Day 3, the operated female naturally spawned 50–200 fertilized eggs. This result implies that microsurgical operation of female fish has little effect on mating behavior and spawning. Interestingly, almost all of the operated females survived and could be reused for microsurgical operation again after rest for 3–4 weeks (Supplementary Figure S2A, C, and D). Until now, we have used some females for OMIS as many as four times.

Figure 1

OMIS procedures and efficiency. (A) Illustration of OMIS procedure. Stage III oocytes arrested at prophase I in the microsurgically opened female are microinjected in the afternoon on Day 1. The female mates to a male in the evening on Day 2, during which microinjected oocytes develop to stage IV. Stimulated by light in the morning on Day 3, the microinjected oocytes quickly develop to stage V and are ovulated and fertilized. GVBD, germinal vesicle breakdown. (B) Microsurgical operation of a living female. The arrows in a group of oocytes with the ovary indicate microinjected stage III oocytes with rhodamine B. (C) MOF rates of OMIS-operated oocytes using different buffers. Buffer A: 5.4 mM KCl, 136.8 mM NaCl, 4.2 mM NaHCO₃, 0.44 mM KH₂PO4, 0.25 mM Na₂HPO₄; Buffer B: 10 nM melatonin in A; Buffer C: 50 nM melatonin in A; Buffer D: 100 nM melatonin in A. Buffer E: 50 nM melatonin and 5 mM HEPES in A. nRE, number of rhodamine B-positive embryos; niO, number of microinjected oocytes; nF, number of female fish. Each dot represents the MOF rate for a single female. *P < 0.05. (D) The MOF rates calculated from subsequent OMIS operations. KO, knockout by Cas9; KD, knockdown; transg, transgenesis. Each dot represents the MOF rate for a single female. (E) The OMIS-derived, rhodamine B-positive embryo is observed under a fluorescent microscope. The stage III oocytes were microinjected with 21.6 pg gfp mRNA per oocyte in combination with rhodamine B (left panel), or microinjected with rhodamine B only (right panel) as a no-GFP negative control; and the derived embryos were observed at indicated stages. mpf, minutes post-fertilization; hpf, hours post-fertilization.

Then, we tested whether immature oocytes can tolerate microinjection that would induce the breakage of the cytoplasm membrane and generate tension. We used the rhodamine B as a dye to mark the microinjected oocytes. Rhodamine B was microinjected into 20–60 stage III oocytes arrested at prophase I immediately following microsurgical opening of the belly of fish, and the cut was then sewed. In the morning on Day 3, rhodamine B-positive eggs derived from microinjected oocytes could be distinguished from those derived from uninjected oocytes under fluorescence microscope. Initially, female microinjected with rhodamine B dissolved in ddH₂O rarely spawned rhodamine B-positive eggs. Then, we tested different buffers containing rhodamine B. In order to compare the efficiencies of different microinjection buffers, we calculated the maturation, ovulation, and fertilization (MOF) rate of microinjected oocytes. The average MOF rates ranged from 3.4% to 15.9% among different microinjection solutions (Figure 1C). The optimized buffer consisted of 5.4 mM KCl, 136.8 mM NaCl, 4.2 mM NaHCO₃, 0.44 mM KH₂PO₄, 0.25 mM Na₂HPO₄, and 100 nM melatonin. The inclusion of an appropriate amount of melatonin is essential probably due to its oocyte maturation-stimulatory effect and antioxidative activity (Maitra and Hasan, 2016). The embryos derived from oocytes microinjected with the optimized microinjection buffer developed normally without morphological defects (Supplementary Figure S2B). Thus, microinjection buffer is critical for MOF of microinjected oocytes. We termed our method oocyte microinjection in situ (OMIS).

The optimized buffer was subsequently used for the other genetic manipulations as described below, which gave an average MOF rate of around 20%. The MOF rates varied much among individual females with the highest rate of 74.3% (Figure 1D), suggesting that the quality of the female was important. Usually, each stage III oocyte in ovaries of a female could be microinjected with about 0.2 nl desired reagent. To obtain more OMIS-derived embryos, females of high quality need to be selected 1–2 weeks before OMIS (Supplementary Figure S2A). As described below, the number of embryos offered by established OMIS technology could meet the needs of most usual assays and statistical analysis.

Efficient translation of exogenous mRNA in OMIS-operated oocytes

To investigate whether exogenous mRNA can be translated efficiently in immature oocytes, we co-microinjected gfp mRNA at a dose of 21.6 pg/oocyte with rhodamine B into prophase I arrested oocytes. After spawning, the rhodamine B-positive embryos at the one-cell stage showed strong GFP fluorescence (Figure 1E), which indicates that the translation machinery worked efficiently in the microinjected oocytes. We further validated whether exogenous mRNA microinjected into immature oocytes is stable for a long period ahead of fertilization by microinjection of gfp-nanos-3′UTR mRNA containing the 3′UTR of the nanos1, which allows germplasm-specific expression of GFP. Subsequently, strong GFP fluorescence was observed by confocal microscopy in the primordial germ cells (PGCs) at the bud stage and 33 h post-fertilization (hpf) (Supplementary Figure S3), which suggests that microinjected exogenous mRNA could be stable for a few days in vivo. Expression of exogenous mRNA, such as Cas9 mRNA for gene editing and Tol2 mRNA for transgenesis, is essential for genetic manipulations. Therefore, efficient translation of exogenous mRNA is an essential prerequisite for other applications of OMIS.

OMIS induces high levels of specific matrix metalloproteinases

As described above, the MOF rates of microinjected oocytes varied much among individual females (Figure 1D). We suspected that in vivo microinjection of immature oocytes might introduce stress in response to membrane damage and consequently causes developmental arrest and death of some of microinjected oocytes. To look into possible mechanisms, we compared the transcriptomes of microinjected and uninjected oocytes from the same female fish. Following OMIS with rhodamine B in the afternoon, 10–15 microinjected oocytes (still arrested at prophase I) were isolated from the ovary at 5 h post-OMIS (hpO) or 14 hpO, while uninjected oocytes were isolated at the same time points from the other ovary of the same female. The surrounding somatic cells of these oocytes were separated carefully using tweezers in a specific medium for culturing zebrafish oocytes (Xu et al., 2014). Results showed that microinjected oocytes only contained 47 upregulated and 23 downregulated genes at 5 hpO, and 98 upregulated and 9 downregulated genes at 14 hpO (>2.0-fold change) identified using DESeq2 analysis (P < 0.01 after correction for multiple comparisons) (Figure 2A and B), which suggests that OMIS does not induce dramatic changes in the transcriptional profiles. Gene ontology analysis showed that upregulated genes in microinjected oocytes at 14 hpO were strongly enriched for matrix metalloproteinases (MMPs) (Figure 2D), which play an essential role in tissue remodeling caused by diseases or environmental stresses (Page-McCaw et al., 2007; LeBert et al., 2015). Real-time quantitative PCR analysis confirmed the significant increase of expression levels of mmp9, mmp11b, mmp13a, adamts8a, and timp2b in microinjected oocytes at 14 hpO (Figure 2E and F). Taken together, these data indicate that microinjection mainly induces tissue repairing responses due to physical injury, which might be the main cause of low MOF rates of microinjected oocytes in most cases.

Figure 2

RNA profiling of oocytes. (A and B) Scatter plots display correlation between OMIS and uninjected control at 5 hpO (A) and 14 hpO (B). Differentially expressed genes were identified using DESeq2 (P < 0.01 after correction for multiple comparisons) and a 2-fold change between OMIS and control groups. Red and blue dots represent upregulated and downregulated genes in OMIS group compared with control group, respectively. (C) Correlation values of RNA sequencing replicates. There are two replicates for each. (D) GO term analysis of up-regulated genes in OMIS group at 14 hpO. (E and F) Verification of several upregulated matrix metalloproteinase genes by real-time qPCR analysis. *P < 0.05; **P < 0.01; ***P < 0.001.

The genome of oocytes arrested at prophase I is editable

By FEM in zebrafish or MII oocyte microinjection in mice, gRNA and Cas9 mRNA are diluted at multi-cell stages, usually resulting in mosaic animals. It has been reported recently that, through microinjection of CRISPR/Cas9 system into MII oocytes in mice, maternal alleles undergo editing only after the first round of DNA replication after fertilization, while paternal alleles are editable within 3 h after sperm injection (Suzuki et al., 2014). We want to know whether the genome of oocytes in living zebrafish ovaries can be edited using Cas9 approach. After verification of the editing effectiveness of ntla gRNA and Cas9 mRNA by FEM, we microinjected ntla gRNA and Cas9 mRNA together with rhodamine B into stage III oocytes and collected rhodamine-positive eggs about 42 h later (in the morning of Day 3) by gently squeezing the operated female (Figure 3A, upper panel). These eggs were immediately lysed individually to release genomic DNA so that they were actually stage V oocytes arrested at metaphase II. The released genomic DNA from each egg was used to perform single-cell genome analysis by multiple displacement amplification technology (Hou et al., 2015). Since an immature oocyte contains 4 sets of chromosomes (4n) and a mature egg (2n) is attached by the first polar body (2n), it is expected that there might be up to 4 different alleles at the ntla locus in one egg sample. We indeed found that 5 of 11 OMIS-derived eggs had two different alleles at the ntla locus, at least one of which was a mutant form of the target sequence (Figure 3B, D, and E). This result strongly suggests that the gene editing occurs before the release of the first polar body.

Figure 3

Target site sequence analysis of single oocyte genome following OMIS-mediated Cas9 gene editing. (A) Illustration of single oocyte DNA analysis. Top, eggs are squeezed from the OMIS-operated female and immediately lysed individually. Bottom, eggs are squeezed from uninjected female and injected, followed by in vitro culture for 30–60 min. WGA, whole-genome amplification. Microinjection doses: 6.3 pg ntla or 20.5 pg gata5 gRNA plus 43.2 pg Cas9 mRNA for OMIS; 30 pg ntla or 95 pg gata5 gRNA plus 80 pg Cas9 protein for in vitro cultured eggs. (B and C) ntla mutant alleles in six F₀ OMIS eggs and gata5 mutant alleles in one F₀ OMIS egg. The wild-type allele is labeled as w. PCR product from individual eggs was cloned and 12 clones per egg were sequenced. (D) The ratio of eggs carrying mutant alleles. nE, number of eggs. (E and F) Sequences of ntla (E) and gata5 (F) mutant alleles.

To test whether gene editing event occurs in mature oocytes, we microinjected ntla or gata5 gRNA and Cas9 protein into squeezed stage V oocytes (mature oocytes or eggs), cultured in vitro for 30–60 min as described by Xie et al. (2016), and lysed the eggs without fertilization to release the genome for single-cell sequencing (Figure 3A, lower panel). Results showed that none of 12 ntla gRNA- and 20 gata5 gRNA-microinjected eggs carried any mutations at the ntla or gata5 locus (Figure 3D). Xie et al. (2016) also observed that zebrafish mature oocytes microinjected with mc4r or mpv17 gRNA plus Cas9 mRNA did not have detectable mutations based on PCR-based T7E1 assay using DNA mix of several oocytes, which is consistent with our observation. Therefore, it is most likely that the genome of mature oocytes is resistant to editing.

We then tested editing efficiency at gata5 and tyr loci in immature oocytes. Only one of nine OMIS-derived eggs harbored mutations at the gata5 target site (Figure 3C, D, and F) and none of 20 OMIS-derived eggs carried mutations at the tyr target site (data not shown). These results suggest that not every locus in the genome of immature oocytes is editable, which probably depends on the structure of chromatin harboring the target loci.

One-step generation of mutants carrying biallelic mutations via OMIS

Given that OMIS-introduced gRNAs/Cas9 evoke gene editing during oocyte maturation and would allow continuous editing of target genes during embryonic development after fertilization, we would expect biallelic editing of maternal and paternal alleles of a target locus with high efficiency in OMIS-derived embryos. To test this idea, we performed ntla knockout via OMIS in oocytes, which were fertilized with wild-type sperm after maturation (Figure 4A). All of 15 OMIS-derived embryos lacked a tail at 2 days post-fertilization (dpf) (Figure 4B and C), which phenocopied ntla mutants reported previously (Halpern et al., 1993). Sequencing of ntla alleles in these 15 embryos, 12 clones per embryo, revealed a mutation rate of 93.9% (Figure 4D). Especially, a single mutant allele with a 12-bp deletion (mutation type 1, green circle in Figure 4E) was predominant (62.2%), and 4 OMIS-derived embryos carried this mutant allele only (Figure 4E and Supplementary Figure S4). These results indicate that the Cas9 system provided in the OMIS-derived eggs could continuously edit the paternal genome, which results in mutations in both maternal and paternal alleles and manifestation of mutant phenotypes in the OMIS-derived embryos. The underlying mechanism needs to be investigated in the future.

Figure 4

Editing of individual genes by Cas9 approach via OMIS and FEM. (A) Illustration of genomic DNA preparation of individual OMIS-derived mutant embryos after phenotype analysis. (B–E) Editing of the ntla locus. Microinjection doses: 6.3 pg ntla gRNA and 43.2 pg Cas9 mRNA for OMIS; 30 pg ntla gRNA and 200 pg Cas9 mRNA for FEM. (B) Morphology of wild-type and ntla mutant embryos via OMIS and FEM of different categories at 48 hpf. Blue, typical no tail phenotype similar with ntla mutants reported previously. (C) The ratios of OMIS-derived and FEM-derived ntla knockout embryos with different phenotypes. nE, number of embryos. The colors of each bar indicate the embryos with phenotypes in B. (D) The mutagenesis rates of OMIS-derived and FEM-derived ntla knockout embryos with mutant alleles in different categories. The colors of each bar indicate the embryos with phenotypes shown in B. Weighted avg., weighted average with three kinds of embryos via FEM. nC, number of clones. (E) ntla alleles carried by 15 individual OMIS-derived mutant embryos. Twelve DNA clones derived from a single embryo were sequenced. Different alleles have a unique number and color. w, wild-type allele. (F–H) Editing of the tyr locus. Microinjection doses: 15.12 pg tyr gRNA and 43.2 pg Cas9 mRNA for OMIS; 70 pg tyr gRNA and 200 pg Cas9 mRNA for FEM. (F) Morphology of wild-type and tyr mutant embryos via OMIS and FEM of different categories at 48 hpf. (G and H) The ratios of tyr knockout embryos with phenotype (G) and the mutagenesis rates (H) of tyr knockout embryos with mutant alleles in different categories. The colors of each bar indicate the embryos with phenotypes shown in F. Weighted avg., weighted average with different kinds of embryos. nE, number of embryos. nC, number of clones.

By contrast, via FEM with RNPs (gRNA-Cas9 protein), even if the doses of ntla gRNA were ~5 folds higher than those by OMIS and the dose of Cas9 protein was 160 pg/embryo, only 52.3% (46/88) of embryos displayed the no tail phenotype and the others had a smaller tail or a normal tail (Supplementary Figure S6A and B). The mutation rate varied among different classes of the FEM-derived embryos with a weighted average mutation rate of 88.9% and no predominant mutant alleles appeared (Supplementary Figures S6C, D, and S7). Apparently, OMIS with gRNA and Cas9 mRNA still works with a higher mutagenesis efficiency at the ntla locus than FEM with gRNA and Cas9 protein. So, OMIS allows the one-step generation of embryos carrying biallelic mutations, which display the phenotype of corresponding homozygous zygotic mutants.

Next, we compared gene editing efficiency of the tyrosinase (tyr) locus (Figure 4F) (Jao et al., 2013) using OMIS and FEM with tyr gRNA and Cas9 mRNA. The embryos at 48 hpf were categorized into five classes based on morphology: the degree of pigmentation defects in the eyes and trunk (Figure 4F and G). The most severe (blue) class of embryos, which accounted for 80% (12/15) of OMIS-derived embryos, phenocopied ENU-induced tyr/sdy mutants (Kelsh et al., 1996). By contrast, only 11.7% (18/154) of FEM-derived embryos displayed partial loss of pigments (Figure 4G). We sequenced tyr alleles in OMIS-derived embryos and found that the average mutation rate was 95.7% with biallelic mutations in most of embryos in blue class (Figure 4H; Supplementary Figures S8 and S9). Interestingly, unlike ntla/OMIS-derived embryos (Figure 4E), none of tyr/OMIS-derived embryos carries identical mutant alleles (Supplementary Figure S8A), suggesting that gene editing event at the tyr locus occurs at multi-cell stages of development so that different cells of the same embryo harbor different mutations. In contrast, the estimated weighted mutation rate in the FEM-derived embryos was 47.9% (Figure 4H). Even if tyr RNP complex was microinjected during FEM, only 24.4% (40/164) of embryos displayed partial loss of pigments and the rate of mutated alleles was 62.2% (Supplementary Figures S10 and S11), which are still lower than those obtained with OMIS using Cas9 mRNA. Thus, even though the tyr locus of the oocyte cannot be edited prior to fertilization (data not shown), the maternal and paternal alleles can be edited by the Cas9 system provided via OMIS with higher efficiency than FEM with RNPs, resulting in manifestation of mutant phenotypes in the OMIS-derived embryos.

The above results suggest that OMIS allows the one-step generation of embryos carrying biallelic mutations, which display the phenotype of corresponding homozygous zygotic mutants.

One-step generation of compound mutants with multiple loci knockout via OMIS

Double, triple, and even multiple mutants in functionally related genes are ideal for studying complicated biological process, but are hard to be obtained. We expect that OMIS can help achieving this goal. Then, we tested the effectiveness of co-editing ntla, tyr, and gata5 (Chang et al., 2013) through OMIS, and compared it with FEM using Cas9 mRNA. Subsequently, we obtained 15 OMIS- and 217 FEM-derived embryos. These embryos were observed for morphologic defects with a focus on simultaneous manifestation of phenotypes of three genes, e.g. ntg (no tail due to ntla deficiency, no pigments due to tyr deficiency, abnormal heart morphology due to gata5 deficiency) (Reiter et al., 1999), ng, nt, and tg at 2 dpf (Figure 5A). Totally, 46.7% (7/15) of the OMIS-derived embryos exhibited mutant phenotypes for all of the three genes (ntg group) with biallelic mutations in three loci (Figure 5B), which was in sharp contrast to 6.9% (15/217) of the FEM-derived embryos with similar phenotypes (Figure 5C). Therefore, a high percentage of OMIS-derived embryos with similar compound phenotypes would provide repeatable and reliable phenotype assessment.

Figure 5

Simultaneous editing of multiple genes by Cas9 approach via OMIS and FEM. (A–C) Morphology (A), mutant alleles (B), and mutagenesis ratio (C) of embryos with different defects at 48 hpf following co-editing of ntla, tyr, and gata5 loci. For OMIS, each oocyte is microinjected with 11.23 pg ntla gRNA, 17.93 pg tyr gRNA, 20.52 pg gata5 gRNA, and 43.2 pg Cas9 mRNA. For FEM, each fertilized egg was microinjected with 52 pg ntla gRNA, 83 pg tyr gRNA, 95 pg gata5 gRNA, and 200 pg Cas9 mRNA. Compound phenotypes arising from combinatory deficiency: ntg, ntla, tyr, and gata5; ng, ntla and gata5; nt, ntla and tyr; tg, tyr and gata5. nE, number of embryos. The ratio of embryos with defects due to deficiency of a single gene was not calculated, but those embryos are included in the total number. (D) Mutagenesis rates of 19 genes (20 targets) after co-editing. The doses of gRNAs and Cas9 mRNA for OMIS are 5 folds less than those for FEM. Six OMIS-derived embryos were randomly picked up before death (from sphere to 30%-epi stages) and six FEM-derived embryos were randomly picked up at bud stage for sequencing the target regions by deep sequencing. Each dot represents the mutagenesis rate at a target site in a single embryo.

Given that OMIS/Cas9 needs only a very low dose of a specific gRNA for efficient gene editing, we hypothesized that it could work to concurrently edit many more genes. To test this possibility, we randomly chose 19 genes for editing and synthesized 20 gRNAs, 1 or 2 gRNAs for each gene (Supplementary Table S1). All gRNAs after purification were mixed together, but concentrations of each gRNA were variable because of different original concentrations of individual gRNA preparations (Supplementary Tables S2 and S3). For OMIS, a total amount of 149.62 pg 20× gRNAs (each 1.25–14.69 pg), 43.2 pg Cas9 mRNA were microinjected into each oocyte; and for FEM, the doses (per fertilized egg) were about 5 folds of OMIS. The OMIS-derived embryos all deformed by the 30% epiboly stage, which might result from severe disruption of developmental signaling network, while the FEM-derived embryos usually survived throughout segmentation stages. To compare their editing efficiencies, six embryos from each group were randomly selected, and 20 target sequences were amplified from individual embryos for deep sequencing on Illumina HiSeq platform, and the mutation rate at each target site of individual embryos was calculated (Figure 5D; Supplementary Tables S2 and S3). The average mutation rates among the 20 target sites in OMIS-derived embryos ranged from 1.56% (unga target1) to 70.37% (tyr) with a mean of 28.23%, whereas the rates in FEM-derived embryos were from 0.48% (unga target1) to 14.99% (ndr2) with a mean of 4.96%. The average mutation rate at individual sites in OMIS embryos was 5.69 folds higher than that in FEM embryos. These results suggest that OMIS has a prominent advantage over FEM in studying functionally redundant genes.

In zebrafish embryos, the nodal-related genes ndr1 and ndr2 have overlapping functions and only ndr1;ndr2 double mutants, but not single mutants, lack mesodermal and endodermal tissues in the head and trunk with fused eyes (Feldman et al., 1998). We designed three target sites for ndr1 and four target sites for ndr2 (Supplementary Figure S13A, B, and Tables S1 and S4). All gRNAs were co-injected with Cas9 mRNA for OMIS, but gRNAs-Cas9 protein complexes were injected for FEM. Subsequently, we obtained 23 OMIS- and 151 FEM-derived embryos and observed their morphology at 1 dpf (Supplementary Figure S13C). We found that 21.7% (5/23) of the OMIS-derived embryos exhibited fused eyes with missing of most mesendodermal tissues, which resembled ndr1;ndr2 double mutant phenotype (Supplementary Figure S13C). The most severe defects were found in 2.6% (4/151) of the FEM-derived embryos, which showed closer eyes with missing of anteriormost mesendodermal tissues (Supplementary Figure S13C). But, none of the FEM-derived embryos phenocopied ndr1;ndr2 double mutants. These data demonstrate a value of OMIS-mediated co-editing of multiple genes for briefly looking into function of redundant genes.

One-step generation of germline-like transgenic fish via OMIS

Transgenesis is widely used for investigating functions of genes and labeling tissues and specific cells. We wondered whether OMIS could be used to make germline-like transgenic embryos within a few days. This idea was first tested using the Tol2 transposon-based construct Tg(huc:GFP), which allows GFP expression driven by the huc (also known as alavl3) promoter in neuronal cells (Zhao et al., 2006) (as illustrated in Figure 6A). We obtained 69 OMIS-derived embryos (F₀), 22 (31.9%) of which contained GFP-positive neurons. Most (16/22, 72.7%) of the GFP-positive embryos had GFP expression in most, if not all, of the objective cells (Figure 6B and D). Although 33.9% (62/183) of the FEM-derived embryos expressed GFP, the GFP expressions were apparently mosaic (Figure 6B and D). To test germline transmission, one of the OMIS-derived F₀ embryos was raised to adulthood (male). After mated to a wild-type female, GFP was expressed in 58.7% (88/150) of their F₁ progenies in a pattern very similar to that in most of the OMIS-derived F₀ embryos. By contrast, three adults growing up from GFP-positive FEM embryos gave germline GFP-positive F₁ embryos at an average rate of 4.85% (38/783) (Figure 6C and E). These results suggest that OMIS F₀ embryos have at least one copy of the Tg(huc:GFP) transgene that was integrated into the genome in most (if not all) cells including PGCs at early developmental stages, thus looking like germline transgenic animals.

Figure 6

Simultaneous labeling of individual or multiple tissues with transgenes by OMIS or FEM. (A) Expected expression patterns of the transgenes Tg(huc:GFP or huc:YFP) (in neurons), Tg(fli1a:mCherry) (in vasculature and blood cells), and Tg(crybb1:CFP) (in eye lens). (B–E) Transgenesis efficiency using Tg(huc:GFP) construct by OMIS or FEM. For OMIS, each oocyte was microinjected with 1.08 pg Tg(huc:GFP) plasmid DNA and 21.6 pg TPase mRNA; and for FEM, each fertilized egg was microinjected with 5 pg Tg(huc:GFP) DNA and 100 pg TPase mRNA. The embryos are observed at 24 hpf by confocal microscopy. The boxed areas are enlarged for better view. (B) Representative F₀ OMIS-derived (upper panel) and FEM-derived (lower panel) embryos with different numbers of GFP-positive neurons. (C) Germline transmission of the transgene Tg(huc:GFP). The OMIS-derived F₀ males are mated to wild-type females and their progenies (F₁) show two transgene expression patterns, strong and weak. Transgenic F₁ embryos with different patterns are also found from crosses between three FEM-derived F₀ males and wild-type females. (D and E) The rate of embryos in each category of F₀ (D) and F₁ embryos (E). (F–K) Transgenesis efficiency using Tg(huc:GFP), Tg(fli1a:mCherry), and Tg (crybb1:CFP) constructs by OMIS or FEM. (F and G) Expression patterns of the reporters in a representative OMIS (F) or FEM (G) F₀ embryo. For OMIS, each oocyte was microinjected with 1.08 pg of each construct DNA and 21.6 pg TPase mRNA; and for FEM, each fertilized egg was microinjected with 5 pg of each construct DNA and 100 pg of TPase mRNA. The boxed areas are enlarged for better view. (H) The ratios of F₀ embryos with expression of transgenes. 3×, all three transgenes are expressed; 2×, two of transgenes are expressed; 1×, only one of transgenes is expressed; Neg, none of transgenes are expressed, blue. nE, number of embryos. (I) Expression patterns of transgenes in a representative F₁ embryo derived from an OMIS founder. The boxed areas are enlarged for better view. (J and K) Germline transmission rates in F₁ progenies of OMIS-derived founders (J) and FEM-derived founders (K). The labels are the same as in H.

Simultaneous germline labeling of different types of cells in animals with different reporters is very useful for lineage tracking, which is extremely difficult to be attained by conventional transgenesis via FEM. We performed co-microinjection of three transgene constructs, Tg(huc:YFP), Tg(fli1a:mCherry) for mCherry expression in the vascular system and some blood cells (Lawson and Weinstein, 2002), and Tg(crybb1:CFP) for CFP expression in the eye lens (Emelyanov and Parinov, 2008) (Figure 6A) by OMIS and FEM. Among 39 OMIS-derived embryos, 25.6% (10/39) of embryos expressed all three reporters, and 40% (4/10) of them showed germline-like expression patterns of all three reporters (Figure 6F and H). For FEM, we observed 20.77% (27/130) embryos expressing all three reporters, but none of them showed germline-like expression patterns (Figure 6G and H). When adults growing up from OMIS embryos with germline-like expression patterns of three reporters mated with wild-type fish, about half of their F₁ progenies (n = 134) showed strong and specific expression patterns for each transgene. Interestingly, 5.97% (8/134) of these F₁ embryos co-expressed all of three transgenes (Figure 6I and J), while <5% of F₁ embryos from FEM-derived adults (n = 1049) showed expression of individual transgene and none of them co-expressed all of three transgenes (Figure 6K). These results imply that OMIS is able to generate stable lines co-expressing multiple transgenes in just one generation (~3 months), which would not be achieved by FEM.

Efficient knockdown of the maternal gene dnmt1 via OMIS

Maternal genes may express and store their protein products in oocytes during oogenesis, which makes it difficult to efficiently interfere with their functions through FEM. We tested whether OMIS could be used to effectively knockdown the maternal gene dnmt1, which encodes a DNA (cytosine-5)-methyltransferase and is required for the maintenance of cytosine-5 methylation pattern during early development in mammals (Reik et al., 2001), as an example. When the morpholino-modified antisense oligonucleotide dnmt1-MO was injected into one-cell stage embryos, Dnmt1 protein level was not obviously reduced in embryos at the 4-cell and 256-cell stages (Figure 7A), and consequently, the 5mC level showed little changes at the 256-cell and dome stages (Figure 7B), which indicates that maternal Dnmt1 protein could not be removed by knockdown after fertilization. As reported before (Rai et al., 2006), the FEM-derived dnmt1 morphants only develop visible phenotype at 3 dpf. In contrast, dnmt1 knockdown through OMIS with 1.08 pg dnmt1-MO per oocyte resulted in almost complete absence of Dnmt1 protein in embryos even at the 4-cell stage (Figure 7A), and loss of 5mC at the 256-cell and dome stages (Figure 7B). The OMIS-derived dnmt1 morphants showed developmental delay at the dome stage, failed to undergo epiboly, and gradually died at 10 hpf (Figure 7C). These results imply that the translation of maternal mRNA can be blocked by antisense oligonucleotides and existed protein may be degraded during oocyte maturation.

Figure 7

Knockdown effects of dnmt1 following OMIS or FEM. (A and B) Dnmt1 protein (A) and 5mC (B) levels are visualized at indicated stages by confocal microscopy following dnmt1 knockdown via OMIS and FEM. The doses of dnmt1-MO and the control MO (cMO) were 1.08 ng per oocytes or 5 ng per embryo. (C) Phenotypes of OMIS-derived dnmt1 morphants (upper panel) and OMIS-derived embryos microinjected with cMO (lower panel) at different stages.

Discussion

In this study, we have established the OMIS technology to manipulate immature oocytes within the ovaries of the living adult females, which proves the in situ manipulability of immature oocytes for the first time. Importantly, we discover that the genome of primary oocytes can be edited by CRISPR/Cas9 approach although these oocytes are not undergoing chromosome replication after OMIS. Compared with microinjection in fertilized eggs, OMIS is more efficient, even if at much lower doses of reagents, for gene knockdown, gene editing, and transgenesis. OMIS is a valuable tool in several circumstances. First, maternal mutants due to deficiency of a maternal gene could not be rescued by its ectopic expression after fertilization. Second, a maternal gene could not be knocked down due to the presence of its protein in oocytes. Third, many functionally redundant genes need to be simultaneously edited biallelically. Fourth, many tissues or cell types need to be simultaneously labeled by multiple transgenic reporters.

The MOF rates of OMIS-derived embryos are variable among individual adult females, ranging from zero to 74.3%. We found that microinjecting of immature oocytes stimulates tissue repairing responses, which may result in maturation failure and apoptosis of some oocytes. The variation of MOF rates among females means that different females respond to the microinjection stimulation differentially, which may be related to their genetic backgrounds. The females with a high MOF rate could be used to establish lines for better performance.

We also tried OMIS for immature oocytes at stages I and II, which are much smaller than stage III oocytes and contain much less amount of maternal material. When the optimized microinjection solution for OMIS of stage III oocytes was used for OMIS of stage I or stage II oocytes, we failed to obtain any OMIS-derived embryos. This may be explained by the promoting effect of melatonin on oocyte maturation. Then, we decided to exclude melatonin from the microinjection solution for OMIS of early oocytes. One week after OMIS of stage I and stage II oocytes, the females were mated to males for the first time and then mated once a week. We obtained 30 OMIS-derived embryos out of 1257 embryos from 70 females totally. The rate (2.39%) was much lower than that for OMIS of stage III oocytes. We guess that the current microinjection solution is not appropriate for stage I/II oocytes. Therefore, future effort should be made to optimize OMIS conditions for stage I/II oocytes.

In avian species, early embryonic development occurs in the oviducts and a laid ‘egg’ carries embryos consisting of some 50000 cells. This feature makes it extremely difficult to generate transgenic or mutant lines in birds. OMIS may be of the choice to overcome that obstacle.

Human immature oocytes can be cultured and fertilized in vitro (Cha and Chian, 1998), which provide an opportunity to repair defective oocytes in vitro. However, in vitro maturation of human oocytes usually gives rise to lower rates of maturation, fertilization, clinical pregnancy, and lower live birth (Farsi et al., 2013), suggesting that it is difficult to simulate in vivo oocyte maturation environments. The establishment of OMIS in mammals would provide an approach to repair defective oocytes in situ. Unlike immature oocytes within the fish ovaries that are well separated from each other, immature oocytes in mammalian ovaries are surrounded by granulosa and thecal cells, which would make in situ microinjection of oocytes more difficult. Theoretically, development of OMIS technology is possible in mammals.

Materials and methods

Zebrafish strains

The wild-type India and AB strains were used throughout the experiments. Unless otherwise specified, the adult fish were fed with live adult brine shrimp in the morning and evening; and manually ground dry trout pellets at noon. For females after OMIS, they were fed with live adult brine shrimp thrice a day. All embryos were raised in Holtfreter’s solution at 28.5°C and staged as described previously (Kimmel et al., 1995). Oocytes were staged as previously described (Selman et al., 1993). Ethical approval was obtained from the Animal Care and Use Committee of Tsinghua University. All microsurgeries were performed under anesthesia, and all efforts were made to minimize the distresses of fish.

Fertilized egg microinjection

Cas9 mRNA or Cas9 protein (NEB, M0646), gRNAs, and morpholinos were injected into the yolk of one-cell stage embryos, while plasmids DNAs and Tol2 transposon mRNAs were co-injected into the cytoplasm. All these reagents were diluted in RNase-free water. Injection doses were given in the text or corresponding figure legends.

Microinjection solution for OMIS

The basic microinjection solution for OMIS consisted of 5.4 mM KCl, 136.8 mM NaCl, 4.2 mM NaHCO₃, 0.44 mM KH₂PO₄, and 0.25 mM Na₂HPO4 in RNase-free water. The optimized solution contained 100 nM melatonin (Abcam, ab141052). The solution should be fresh. DNA, RNA, or other reagents were added to the solution at a desired concentration. In order to identify embryos derived from the microinjected oocytes, rhodamine B (Sigma, R8881) was added to the microinjection solution.

OMIS procedure

One to two weeks before OMIS, adult females at 5–12 months old were mated to males. Only those females that produced large amount of embryos with normal development were selected for OMIS operation. At the first day of OMIS, a female was anesthetized 2–4 h after its lunch in 550 μg/ml tricaine (Sigma, A5040) in a petri dish. After being wiped to remove excess water with paper tissue, the fish was placed on a damp sponge immersed in a buffer of 5.4 mM KCl, 136.8 mM NaCl, 4.2 mM NaHCO₃, 0.44 mM KH₂PO₄, 0.25 mM Na₂HPO₄, and 0.5% (w/v) BSA (Solarbio, A8020). A cut of 4–6 mm long was made on one side of belly with surgical scissors (WPI, 504026) and forceps (WPI, 500341). Approximately 0.2 nl of reagents diluted in microinjection solution was microinjected into each oocyte. Usually, 30–60 oocytes at stage III in the nearside ovary were individually microinjected, using a Narishige MN-151 3-Axis joystick mechanical micromanipulator connected to a foot pedal controlled MPPI-2 pressure injector (ASI) under a stereo-microscope (Motic, SMZ-161) at 30× magnification. Ideally, microinjection needs to be completed within 60 sec. Injection solution should contain a dye or fluorescent protein or mRNA for fluorescent protein expression, which allows the identification of injected oocytes. After injection, the wound of the female was sewed with a surgical sewing needle (WPI, 501948) and suture carefully and quickly. All of these operations were made quickly and gently to minimize the pain of fish. Once the microsurgical operation was completed, the female was put into fish water supplemented with 32 μg/ml tricaine, 10 units/ml penicillin, and 10 μg/ml streptomycin (HyClone, SV30010), followed by blowing tenderly with a pipette till it woke up. The fish was then transferred to fish water containing gradually reduced concentrations of tricaine for three times. In the evening of the second day, the female was paired with a male of good quality, but they were separated by a barrier. In the morning of the third day, once the barrier was removed, the pair started to chase and lay fertilized eggs. The OMIS-derived fertilized eggs or embryos are selected by observing dye or reporter. Because a few embryos can be obtained from one OMIS-operated female, the number of OMIS-derived embryos available for subsequent analysis will be lower than that of embryos microinjected after fertilization. The female may be reused for OMIS after rest for 3–4 weeks. The injection doses were indicated in the text or corresponding figure legends.

Oocyte RNA-sequence library preparation and sequencing

Oocytes were isolated from injected females at 5 or 14 h after OMIS and somatic cells surrounding oocytes were detached carefully. Total RNA of oocytes were purified using RNeasy Mini kit (QIAGEN, 74104) as recommended protocols. Total RNA (5 μg) was DNase I (Fermentas, EN0521) treated at 37°C for 1 h. Poly-A tailed mRNA was collected using DynabeadsTM mRNA Purification Kit (Invitrogen 61006). Extracted RNA was fragmented with RNA Fragmentation Buffer (NEB, E6186A) at 95°C for 5 min. Reaction was stopped and RNA was purified by AMPure beads. The strand-specific RNA libraries were prepared as described previously (Parkhomchuk et al., 2009). Libraries were treated with UDG (AMPErase, N808-0096) at 37°C for 1 h followed by DNA amplification using Phusion HF DNA polymerase (NEB, M0530L). The amplified DNA was size-selected using AMPure Beads for 200–500 bp DNA fragments. All libraries were sequenced by Illumina HiSeq X Ten platform according to manufacturer’s instruction.

RNA-seq data processing

RNA-seq data were firstly processed using Trim Galore! with default parameters to trim the adapter-containing and low quality reads. The filtered data were then mapped to the zebrafish reference genome (danRer7) by STAR (version: STAR_2.5.3a_modified). The gene expression level was normalized to fragments per kilobase of transcript per million mapped (FPKM) values using Cufflinks (version 2.2.1) (Trapnell et al., 2012). And differentially expressed genes were identified using DESeq2 analysis (P < 0.01 after correction for multiple comparisons) and a 2-fold change between OMIS and control groups.

RT-PCR and real-time quantitative PCR

Total RNA were purified as described previously from injected and uninjected oocytes at 5 hpO and 14 hpO. RT-PCR was performed using GoScript^TM reverse transcription system (Promega, A2790) as recommended protocol. The primer sequences for Real Time qPCR were as follows: mmp9, 5′-GGCTCTGGACCAGCCATTCAAAC-3′ (forward) and 5′-CCAATCACAGCCGTATCTCTG-3′ (reverse); mmp11b, 5′-CAACAGTATGGGTACTGACCTC-3′ (forward) and 5′-CACCCAGTAATTCTGGCCTTG-3′ (reverse); mmp13a, 5′-CATCAGCTGACCTACAGGATTGAG-3′ (forward) and 5′-GTGTTTGGGCCATAAAGTGACTGG-3′ (reverse); timp2b, 5′-GGCAATGACGCTTATGGCTA-3′ (forward) and 5′-GGGACACTGGACAATCTTGC-3′ (reverse); adamts8a, 5′-CATGAACTCGGCCATGTACTC-3′ (forward) and 5′-CATTGACGGGCACCTTCTGCTC-3′ (reverse); β-actin, 5′-GCCTTCCTTCCTGGGTATGG-3′ (forward) and 5′-CCAAGATGGAGCCACCGAT-3′ (reverse).

Genomic analysis of single OMIS-derived egg

After OMIS, females in the morning on Day 3 were squeezed gently to collect mature oocytes (eggs) that are distinguished by rhodamine B. Each oocyte was dechorionated manually in the buffer containing 5.4 mM KCl, 136.8 mM NaCl, 4.2 mM NaHCO₃, 0.44 mM KH₂PO₄, 0.25 mM Na₂HPO₄, and 0.5% (w/v) BSA to avoid egg activation and transferred into 4 μl PBS on ice. We used REPLI-g Single Cell Kit (QIAGEN, 150345) to accomplish the cell lysis and whole-genome amplification steps. Buffer D2 (3 μl) was added into the tube with a single oocyte, and incubated for 10 min at 65°C. After that, 3 μl stop solution was added into the tube to stop the reaction. Next, we added 40 μl mixture of phi29 DNA polymerase and its buffer into the 10 μl cell lysis system we had in last step and incubated the tube at 30°C for 4 h and inactivated polymerase at 65°C for 3 min. Then, the amplified genomic DNA was used for amplifying the genomic region surrounding ntla, gata5, or tyr gRNA target sites by PCR and the products were cloned into EZ-T^TM cloning vector (Genstar, T168-101). Through sequencing of PCR products, mutant oocytes were identified and 12 clones per oocyte were picked up for sequencing by Sanger sequencing (GENEWIZ).

Genomic analysis of microinjected single unfertilized egg

Unfertilized eggs were collected by squeezing wild-type females, and transferred into the same buffer as described before to avoid egg activation. These eggs were injected with gRNAs and Cas9 protein and put at 28.5°C for 30–60 min. Through the same whole-genome amplification method mentioned above, amplified genomic DNA from individual eggs were directly used for amplifying the target site by PCR. The remaining analysis was the same as OMIS-derived eggs.

Clone sequencing and PCR products for deep sequencing

To identify the effectiveness of CRISPR/Cas9 by OMIS or by FEM, after phenotype observation and classification, genomic DNA from single injected embryo was extracted by lysis in 30 μl of 50 mM NaOH at 95°C for 30 min and then added 3 μl 1 M Tris-HCl (pH = 8.0) to the lysate. The genomic region surrounding a target site was amplified by PCR with primers listed in Supplementary Table S3 and cloned into EZ-T^TM cloning vector. Then 8–12 clones per embryo were isolated and sequenced by Sanger sequencing (GENEWIZ). For sequencing target sites with 20× gRNAs, six OMIS- or FEM-derived embryos were picked up randomly, and their genomic DNAs were extracted by the same way. The genomic regions surrounding 20 target regions were individually amplified by PCR with specific primers. Then, PCR products from each embryo were pooled as one sample for 2 × 300 bp paired-end sequencing on Illumina HiSeq platform (GENEWIZ). Paired-end sequencing data of PCR amplicons were first trimmed with Trim Galore! (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore) and aligned to the zebrafish reference genome (Zv9) by bwa mem, and only unique mapped reads were kept for subsequent analyses. INDELs were identified with CrispRVariants as previously described (Burger et al., 2016; Lindsay et al., 2016).

Constructs for transgenesis

The 3.1-kb promoter/enhancer fragment of zebrafish huC (also known as elavl3) gene (Zhao et al., 2006) was inserted into a Tol2-based construct containing GFP or YFP coding sequence to make the constructs huc:GFP or huc:YFP. The pT2Fliep-mCherry-GM130 plasmid was kindly provided by Dr Holger Gerhardt. fli1a and crybb1 promoters were amplified by PCR from this construct and subcloned into the Tol2-based vector containing mCherry or CFP coding sequence to generate fli1a:mCherry or crybb1:CFP construct, respectively.

RNA synthesis and morpholinos

The construct containing humanized Cas9 cDNA with double nuclear localization signals (Chang et al., 2013) was kindly provided by Dr Jingwei Xiong. For making gfp mRNA, the construct pCS2-GFP was linearized by NotI (NEB, R0189L) digestion. The construct pCS2-GFP-nanos-3′UTR (Koprunner et al., 2001) was kindly provided by Dr Eraz Raz, and also linearized by NotI digestion. The full-length TPase coding sequence from Medaka fish was cloned into pCS2 vector and linearized by NotI digestion. mRNAs were synthesized using mMESSAGE mMACHINE T7/Sp6 kit (Ambion, AM1344, AM1340) and purified using RNeasy Mini kit. The gRNAs were designed with ZiFiT Targeter Version 4.2 tool (http://zifit.partners.org/ZiFiT/). Before gRNA synthesis, we sequenced target sites in different zebrafish strains to be used to guarantee the efficiency of gRNAs. For making gRNAs, double-stranded DNA for a specific gRNA target site was PCR-amplified from gRNA scaffold in pMD 19-T vector with the primers capped by T7 promoter, which was used as the template of gRNA synthesis. All gRNA target sequences are shown in Supplementary Table S3. After gel extraction of the template, gRNA was generated using MEGAscript T7 kit (Ambion, AM1334). Following in vitro transcription, the gRNA was purified using mirVana miRNA isolation kit (Ambion, AM1561) to a final volume of 30–40 μl. The sequences of dnmt1-MO (Rai et al., 2006; Anderson et al., 2009) and standard control MO (cMO) were 5′-ACAATGAGGTCTTGGTAGGCATTTC-3′ and 5′-CCTCTTACCTCAGTTACAATTTATA-3′, respectively. MOs were dissolved in RNase-free water, and heated to 65°C for 10 min before microinjection.

Immunofluorescence and imaging

Whole-mount immunofluorescence with anti-5mC (Abcam, ab10805) and anti-DNMT1 (Santa Cruz, sc-20701) antibodies were performed essentially as before (Wu et al., 2014). Briefly, embryos at desired stages were fixed by 4% polyformaldehyde for at least 1 day at 4°C, dechorionated manually and dehydrated with methanol at −20°C for 1 h. Then, the embryos were rehydrated with 0.4% PBST (0.4% Triton-X 100 in PBS) for several times, 5 min each, and treated with 2 M HCl for 1 h and then with 100 mM Tris-HCl (pH8.5) for 15 min at room temperature. Next, the embryos were incubated in the block solution for 1 h at room temperature and then in antibody solution overnight at 4°C. On the second day, the embryos were washed with 0.4% PBST for several times and incubated in fluorescence-conjugated secondary antibodies (488 nm-conjugated anti-mouse, 647 nm-conjugated anti-rabbit; Jackson ImmunoResearch) for 2–3 h at room temperature, followed by 0.4% PBST washing several times and DAPI (Invitrogen, D1306) staining. Immunostained embryos were mounted, and imaged using Zeiss 710META laser scanning confocal microscope.

Live embryos were anesthetized at desired stages with 0.02% tricaine and mounted in 1% low melting agarose (Amresco, 0815) for observation under Olympus MVX10 microscope or Zeiss 710META laser scanning confocal microscope. Phenotype pictures were taken using Nikon SMZ1500 microscope.

Statistical analysis

An average from multiple samples was expressed as mean ± SD (standard deviation). Significance of difference between groups was analyzed by Student’s t-test (two-tailed). Significant levels were indicated in the corresponding context.

Data availability

The RNA-seq data of oocytes are available from NCBI GEO under accession code GSE111677. All the other data supporting the findings of this study are available within the article and its supplementary information files and/or from the corresponding author upon request.

Acknowledgements

We thank Dr Wei Xie (Tsinghua University) for sequencing and analyzing total RNA and mutant alleles, Dr Qinghua Tao (Tsinghua University) for anti-DNMT1 antibody. We are grateful to Dr Juhui Qiu, Miss Yaping Meng, Dr Bo Gong, and Dr Jing Chen for providing gRNAs and constructs. We also thank the other members of Meng lab for discussion and staff at the Cell Facility in Tsinghua Center of Biomedical Analysis for assistance with imaging.

Funding

This work was supported by the National Natural Science Foundation of China (31330052 and 31590832).

Conflict of interest

none declared.

Author contributions

X.W. and W.S. performed the experiments and analyzed the data. B.Z. performed RNA sequencing and analyzed deep sequencing data. X.W. and A.M. wrote the manuscript. A.M. conceived the research.

Author notes

Xiaotong Wu and Weimin Shen contributed equally to this work.