Introduction

Genetically modified (GM) animals are an indispensable resource for understanding the mechanisms underlying human diseases and a research base for therapeutic applications. Genome editing with ZFN, TALEN, and CRISPR/Cas9 nuclease has revolutionized the trend of producing GM mice1. Initially, transgenic (Tg) mice were generated by injecting DNA fragments into the pronucleus of fertilized eggs, which expanded the realm of creating genetically modified animals2. This was followed by the production of the specific GM mice, the so-called knock-out mice, by breeding chimeric animals obtained from the blastocysts after precisely genetically modified ES cells were incorporated into their blastocoel, leading to global advancements in genetic modification3. However, these techniques required special equipment and advanced skills and were expensive and time-consuming. Genome editing has broken through the limits of these artisan technologies and reached a level where GM animals could be established with easy operation in just a few months4,5,6.

The nuclease used for genome editing accurately places a double-strand break (DSB) onto the specific nucleotide sequences at the targeted locus. Living organisms (cells) undergo the DSB to connect the sequences through their DNA repair function via the NHEJ pathway or the HDR pathway7. This process incorporates the Indel mutation or knock-in (KI) of the external DNA sequence onto the DSB site. In particular, CRISPR/Cas-based genome editing has been making rapid progress because of its ease and has become a major method for producing GM animals8. CRISPR/Cas9 methods are currently being used to improve Cas9 functions, such as addressing mosaic activity9,10,11 and increasing KI efficiency11,12. Meanwhile, new genome editing technologies have emerged, such as Prime Editing, which uses nCas9 and reverse transcriptase13, and PASTE, which combines Prime Editing and serine integrase14. These methods enable accurate editing without causing DNA double-strand breaks (DSBs). Additionally, progress has been made in the adaptation of other types of Cas proteins, such as Cas12a15,16 and Cas317. The application of these new technologies in the creation of genetically modified animals is highly anticipated.

Most production of GM mice using the CRISPR/Cas system is achieved by transferring CRISPR/Cas components into fertilized eggs18,19,20. Initially, the microinjector-mediated production method was mainly used for Tg mice; however, an electroporation-mediated method was later developed21,22,23,24. This allowed for obtaining edited embryonic genomes with ease of equipment preparation and simplicity of operation. While this method exhibits high KI efficiency for single-stranded oligodeoxynucleotides (ssODN) of up to approximately 200 nt, it is less efficient for the KI of longer DNA fragments spanning 1.5 knt or more25,26. When using microinjection, the KI efficiency for long DNA fragments within the fertilized eggs remains effective at more than approximately 1.5 kb27,28.

Our interest lies in creating GM animals using CRISPR/Cas mediated through the electroporation (EP) of embryos rather than fertilized eggs, specifically two-cell-stage embryos. The two-cell-stage mouse embryo is suitable for cryopreservation29 and its freezing time is widely variable from morning to evening, making them easy to handle. Hence, cryopreservation of two-cell embryos is widely practiced, resulting in many frozen two-cell-stage embryos being preserved at facilities worldwide. There is potential for effectively utilizing these accumulated resources to produce GM animals. Gu et al.30 demonstrated that by using the microinjection of CRISPR/Cas components into two-cell-stage embryos, the KI efficiency of long DNA strands was higher than that in fertilized eggs. These findings suggest that if two-cell-stage embryos can be as easily manipulated as fertilized eggs, they could benefit significantly from embryo genome editing. While several studies have reported genome editing in two-cell stage embryos with a focus on KI efficiency31,32,33,34, the full potential of genome editing in two-cell-stage embryos via EP has yet to be thoroughly explored.

Therefore, we aimed to investigate the generation of CRISPR/Cas-mediated GM mice via EP in two-cell-stage mouse embryos and explore the potential of two-cell-stage embryos as targets for genome editing. We report that EP-mediated genome editing of two-cell mouse embryos is a simple and applicable technique and that two-cell mouse embryos are a useful resource for producing GM mice.

Results

Modified EP method for genome editing in two-cell mouse embryos

Our modified EP system for genome editing in two-cell-stage mouse embryos is shown in Fig. 1. There is a concern that the two blastomeres of a two-cell-stage embryo may fuse together when conventional EP is performed on a two-cell-stage embryo32,34,35. These fused embryos cannot develop to term as they become tetraploid. First, we examined how the EP procedure affected the development in two-cell-stage embryos (Fig. 2). The basic EP system utilized the same conditions as the instrument, with the medium and settings being 20 V for 3 ms (on)/97 ms (off), repeated 5 times in fertilized eggs, which we adopted as the initial condition36. The two-cell embryo groups were prepared with a changed orientation of the electrode (Pre- EP in Fig. 1d and Types A and B in Fig. 2a). Subsequently, they were electroporated using 1–3 sets (20 V for 3 ms (on)/97 ms (off) 5 times per set) (Fig. 2a–c). The EP solvent was a 1:1 mixture of Opti-MEM and 75% PBS media without CRISPR/Cas9 components. Consequently, for type A groups, no fusion of blastomeres in two-cell-stage embryos was observed (Fig. 2b). The development rate into blastocysts remained high but decreased with the number of EP sets (93–73%) (Fig. 2d). In contrast, for the type B groups, the fusion of blastomeres in two-cell-stage embryos was observed (Fig. 2c) and increased with the number of EP sets (Fig. 2f). The rate of blastocysts developed from non-fused in two-cell-stage embryos was high (83–100%), but the rate of blastocysts per total two-cell-stage embryos decreased to 57–60% (Fig. 2e). Thus, these results indicate that blastomere fusion can be avoided by orienting the axis of the contact surface of two blastomeres horizontally to the electrodes and that the EP of two-cell-stage embryos does not affect embryo development (Fig. 2a, Type A).

Fig. 1
figure 1

Overview of the electroporation (EP) system and procedure at two-cell mouse embryos. (a) The EP system consists of a stereo microscope (Olympus SZX16) and an electroporator (BEX CUY21-EDIT II), which were previously used with mouse fertilized eggs36. (b) The EP chamber with platinum electrodes (BEX #LF501PT1-10) was used for this EP system. (c) Microscopic view of a two-cell mouse embryo positioned within the 1-mm gap between the platinum electrodes. (d) Flow chart outlining the experimental procedure for evaluating EP-based CRISPRCas genome editing in two-cell mouse embryos. Pay attention to the contact surface of the two blastomeres of the two-cell-stage embryo, which is positioned perpendicular to the electrodes during the EP stage.

Full size image

1501

Fig. 2
figure 2

Examination of EP to avoid blastomere fusion in a two-cell mouse BDF1×B6 embryo. (a) Two groups, type A and B, were subjected to electroporation (EP). The two-cell embryos were prewashed in EP media and then placed in EP media containing CRISPR reagents, as shown in Fig. 1d, before being manually positioned in type A and B lines for EP. (b) The two-cell embryos electroporated in the type A group did not exhibit any fused embryos. The scale bar indicates 50 μm. (c) In the type B group, two-cell embryos subjected to EP resulted in the observation of several fused embryos (denoted using asterisks). The scale bar also indicates 50 μm. (d) (e) Blastocysts developed from the two-cell embryos electroporated in the type A and B group are shown. The scale bar indicates 50 μm. (f) The table summarizes the rates of fused embryos and blastocyst development in two-cell embryos electroporated one to three times in both the type A and B groups.

Full size image

Next, we examined the blastocyst formation rate in two-cell embryos subjected to varying EP voltage and several repetitions of EP. The experiments were repeated three to five times. The blastocyst formation rate of inbred B6 and BDF1×B6 hybrid strains decreased with an increase in voltage ranging from 15 to 25 V and an increase in EP repetitions (Fig. S1a). The decrease in the blastocyst formation rate of B6 strains was more significant than that in BDF1×B6 hybrid strains with a significant difference at 25 V and 10 repeats of EP condition (B6 42 ± 6% vs. Hybrid 80 ± 8%). Finally, we investigated the relationship between EP voltage and the uptake of mRNA in two-cell embryos (Fig. S1b–c). We prepared EP solutions containing 200 ng/mL EGFP A95 mRNA and placed them between two electrodes on a plate; two-cell embryos of the BDF1×B6 hybrid strain were introduced into the drop, prepared with a changed orientation concerning the electrode (Fig. 2a), and immediately subjected to in vitro EP at 15, 20, and 25 V with five repetitions of 97/3 ms. The experiments were repeated twice, and the average fluorescence intensity of EGFP A95 in resulting four to eight cell embryos cultivated in vitro was measured using a fluorescence microscope. The results show that the uptake of EGFP A95 mRNA by two-cell embryos increased with an increase in EP voltage ranging from 15 to 25 V) (Fig. S1b).

Based on the above results (Fig. 2 and S1), we adopted the following conditions for the experiments below: Type A orientation of two-cell-stage embryos; a 1:1 mixture of media; 1 set of 20 V for 3 ms (on)/97 ms (off) five repeats for the two-cell-stage embryo EP in this study.

Efficiency comparisons and indel mutation analysis in fresh and thawed two-cell embryos and fertilized eggs in genome-edited mice

Frozen-thawed two-cell-stage embryos are potentially useful in genome editing, representing a valuable resource and promising significant benefits. Therefore, to examine the genome editing efficiency of two-cell-stage embryos and to confirm that there was no difference in efficiency between fresh and thawed two-cell-stage embryos, we attempted to create genome-edited mice using our modified EP method with both fresh and frozen-thawed two-cell-stage embryos. We designed and synthesized guide RNAs (gRNAs) targeting base sequences on three genes (Tyr, Adm, and Ramp1, Table S4). Fresh and frozen-thawed two-cell-stage embryos were prepared via in vitro fertilization (IVF) of BDF1×B6 hybrid mice. EP was performed under the following conditions: a 1:1 mixture of media, 20 V with a 3 ms (on)/97 ms (off) duration repeated five times, and the Type A orientation of the two-cell-stage embryo (Figs. 1 and 2a). The concentration of gRNA and Cas9 proteins was 200 ng/µL and 50 ng/ µL, respectively.

The summarized results are presented in Table 1. As anticipated, none of the experiments exhibited fusion of blastomeres in the two-cell-stage embryos. To generate Tyr indel mutated mice, we included fertilized eggs prepared via IVF in BDF1×B6 hybrid mice as a comparative reference. We found that 93% of indel mutated mice were obtained from fresh two-cell-stage embryos, while 81% were obtained from frozen-thawed two-cell-stage embryos; 100% of indel mutated mice were obtained from fertilized eggs. No significant differences were observed among the three groups. Furthermore, the average number of indels per genome in indel-mutated mice obtained from fresh two-cell-stage embryos, frozen-thawed two-cell-stage embryos, and fertilized eggs were 3.6 ± 0.5, 4.2 ± 0.7, and 4.4 ± 0.8, respectively (no significant differences). The coat color of indel mutated mice obtained from the three types of embryos was predominantly white (Fig. S2), indicating null Tyr mutations. We also checked for off-target effects induced by Tyr-gRNA in 42 Tyr indel mice (25 and 17 obtained from fresh and frozen-thawed two-cell embryos, respectively, Table 1). No mutations with off-target probability were observed in the three target regions listed (Table S1). Furthermore, for generating Adm and Ramp1 indel mutated mice, the number and rate of indel mutations obtained from fresh two-cell-stage embryos and frozen-thawed two-cell-stage embryos were not significantly different from those obtained for the generation of Tyr indel mutated mice (Table 1).

Table 1 Efficiency comparisons and indel mutation analysis in fresh and thawed two-cell embryos and fertilized eggs in genome-edited mice.
Full size table

We demonstrated that CRISPR/Cas9 genome-edited mice can be generated from two-cell-stage embryos using our modified EP method. Our results indicate no difference in genome editing efficiency between fresh and frozen-thawed two-cell-stage embryos.

Validation of the utility of EP in two-cell-stage embryos for indel mutagenesis in different genes and mouse strains

To assess the broad applicability of our modified EP method for genome editing in two-cell-stage embryos, we investigated indel mutations in six genes (Adm2, Ddx58, Klf5, Gt(ROSA)26Sor, Trl4, and Trl9) (Tables S2), in addition to the previously studied Tyr, Adm, and Ramp1. The guide RNA sequences targeting the six genes are provided in Table S4. Here, we employed CRISPR/Cas12a to examine the indel mutation capability of the Adm2 gene. The synthesized guide RNA concentration was 200 ng/µL, and both Cas9 and Cas12a proteins were used at 50 ng/µL. We utilized freshly prepared embryos or thawed embryos from BDF1xB6 hybrid mice at the two-cell stage. EP was carried out under identical conditions as described previously (Fig. 1; Table 1), and embryos were allowed to develop to the blastocyst stage in vitro. These blastocysts were used for indel mutation analysis. Notably, fusion of blastomeres in two-cell-stage embryos was not observed, and the blastocyst formation rate after EP ranged from 69 to 90%. The indel mutation rates in the blastocysts were highly efficient, with 100% success for five genes using CRISPR/Cas9 and 1 gene using CRISPR/Cas12a.

Furthermore, we investigated whether the conditions of this EP method affect the efficiency of genome editing and embryonic development rates in two-cell-stage embryos from various mouse strains, including B6, ICR, 129/Sv. Two-cell stage embryos derived from each strain were subjected to EP with Ramp1-gRNA under the same conditions as before (Fig. 1; Table 1), and blastocysts developed from electroporated two-cell-stage embryos in vitro were analyzed for the indel mutation capability. The results are summarized in Table S3. As expected, no fused blastomeres were observed in two-cell-stage embryos from each strain, and the blastocyst formation rate was not different from that of untreated embryos. The Ramp1-Indel mutation rate in blastocysts was 100% in all strains.

Thus, these findings confirm that the conditions of the modified EP method presented in Fig. 1 can be widely applied to genome editing in two-cell-stage embryos.

Efficient ssODN-knock-in editing through modified EP of two-cell stage mouse embryos

We investigated the efficiency of ssODN-KI genome editing using modified EP in two-cell-stage embryos on four genes, Adm, Klf5, Or7a36, and Ramp3 (Fig. 3 and S3). The sequences of gRNAs on the four genes and those of ssODN are summarized in Tables S4 and S5. First, for the KI of the loxP sequence to the Adm gene (Fig. 3a), 108 two-cell stage embryos were electroporated with AM-gRNA-D (200 ng/µl), Cas9 protein (50 ng/µl), and Adm + loxP ssODNs-D (400 ng/µl) (Fig. 3b). No fused blastomeres were observed. These embryos were transferred to pseudopregnant mice, and 38 offspring were obtained. The KI rate of the loxP sequence was evaluated based on the presence of the EcoRI site in the PCR amplification product of the target sequence region (Fig. 3c) and Sanger sequencing (Fig. 3d). Of the 38 offspring, 20 were loxP KI offspring (KI rate 53%), including 15 (KI rate 39%) with loxP/+ and 5 (KI rate 20%) with loxP/loxP. We also investigated off-target effects induced by AM-gRNA-D in the 20 loxPKI mice obtained from fresh and thawed two-cell embryos. No mutations were observed in three regions with off-target probabilities (Table S1).

Fig. 3
figure 3

Efficiency of single-stranded oligodeoxynucleotide (ssODN) knock-in genome editing in two-cell mouse BDF1×B6 embryos. (a) Table summarizes the efficiency of ssODN knock-in in two-cell embryos. (b) Illustration of AM-gRNA-D targeting a location following exon 3 of the murine Adm gene. The sequence recognized by the AM-gRNA-D is highlighted in blue, and the protospacer adjacent motif sequence is highlighted in red. Arrows indicate the positions of the PCR primers (refer to Table S5). All Adm + loxP ssODN sequences are shown in Table S5. (c) Representative full-length image of agarose gel electrophoresis of PCR products amplified from the genomic DNA of pups (nos. 1–14) obtained from the electroporated two-cell embryos. Top panel: results of the restriction fragment length polymorphism assay using the EcoRI enzyme. Bottom panel: PCR products amplified from the target region of the Adm gene. Pups 1, 3, 6, 10, and 13 were heterozygous for the loxP-KI editing. Pups 5 and 8 were homozygous for the loxP-KI editing. M, lambda HindIII + 100-bp ladder markers. (d) Sequences of 5′ and 3′ junction sites in pup 5. Sequence chromatograms show the correct KI of Adm + loxP ssODN-D (as described in Table S5) into the target region of the Adm gene.

Full size image

Furthermore, we investigated the KI efficiency of Flagx3 for the Klf5 and Or7a36 genes and loxP sequences for the Ramp3 genes, respectively (Fig. S3). To this end, two-cell-stage embryos were electroporated with Klf5 + Flagx3 ssODNs (400 ng/µL) and Ramp3-gRNA (200 ng/µL), Cas9 protein (50 ng/µL), and Ramp3 + loxP ssODNs (400 ng/µL) (Fig. S3 and Table S5). No fused blastomeres were observed. The treated two-cell-stage embryos were cultivated into blastocysts. The ssODN KI rate was evaluated through the RFLP assay using ClaI (Klf5) or EcoRI (Ramp3) enzyme in the PCR amplification product of the target sequence region using crude DNA solution from the blastocyst as a template. For Or7a36, the increase in size of the PCR products was used. Klf5 ssODN-treated two-cell-stage embryos showed a blastocyst formation rate of 80% and a ssODN KI rate of 63%. The Klf5-gRNA used was confirmed to have a genome editing efficiency of approximately 44% in offspring derived from fertilized eggs36. Or7a36 ssODN-treated two-cell-stage embryos showed a blastocyst formation rate of 75% and a ssODN KI rate of 39%. Ramp3 ssODN-treated two-cell-stage embryos showed a blastocyst formation rate of 90% and a ssODN KI rate of 4%. The Ramp3-gRNA used demonstrated a low KI rate of 5% in fertilized eggs, along with a blastocyst formation rate of 85%. Thus, the low KI rate of Ramp3-gRNA was not altered by the EP of two-cell-stage embryos.

Based on the above results, we demonstrated that CRISPR/Cas9-mediated ssODN KI genome editing of two-cell-stage embryos via EP is feasible and achieves efficiency equivalent to that of fertilized eggs.

Efficiency of KI editing of dsDNA and ssDNA fragments via modified EP in mouse two-cell-stage embryos

We investigated the efficiency of DNA fragment-KI genome editing using modified EP in two-cell-stage embryos on 2 genes, Ramp2, and Gt(ROSA)26Sor (Fig. 4, S4, and Table S6). To this end, two-cell embryos were electroporated with 1.23 kb dsDNA or 1.23 knt ssDNA, along with Ramp2-gRNA and Cas9 protein (Fig. 4a). As a control, fertilized eggs were treated similarly. The development rates of blastocysts derived from electroporated two-cell-stage embryos and fertilized eggs with 1.23 kb dsDNA and 1.23 knt ssDNA were 78%, 80%, 67%, and 73%, respectively (Fig. 4b). The KI rates of these blastocysts were 19%, 32%, 14%, and 27%, respectively (no significant differences). Subsequently, similar experiments were conducted using 1.80 kb dsDNA and 1.80 knt ssDNA. Consequently, the KI rates were 0%, 0%, 0%, and 0%, respectively, while the development rates of these blastocysts were 64%, 62%, 51%, and 53%, respectively.

Fig. 4
figure 4

Efficiency of knock-in genome editing using various sizes of dsDNA and ssDNA fragments through modified EP of two-cell-stage mouse BDF1×B6 embryos. (a) Left illustration of Ramp2-gRNA targeting the intron of the murine Ramp2 gene. The sequence recognized by the Ramp2-gRNA is highlighted in blue, and the protospacer adjacent motif sequence is highlighted in red. Arrows indicate the positions of the PCR primers (Table S5). Both 1.23 kb dsDNA and 1.23 knt ssDNA, as well as both 1.80 kb dsDNA and 1.80 knt ssDNA, were utilized. They consist of both 5′ and 3′ Ramp2 homology arms, and EGFP cDNA (Table S6). Right illustration of R26-gRNA targeting the intron of the murine Gt(ROSA)26Sor gene . Both 0.66 kb dsDNA and 0.66 knt ssDNA, as well as both 1.36 kb dsDNA and 1.36 knt ssDNA, were utilized (Table S6). (b) (c) Table summarizes the efficiency of various sizes of DNA fragments about knock-in genome editing in the Ramp2 and Gt(ROSA)26Sor locus in fertilized eggs and two-cell embryos. They were electroporated with 100 ng/µL of dsDNA or 50 ng/µL of ssDNA along with 200 ng/µL of guide RNA and 50 ng/µL of Cas9 protein. Each experiment was conducted at least three times.

Full size image

Next, two-cell embryos and fertilized eggs were electroporated with 0.66 kb dsDNA or 0.66 knt ssDNA, along with R26 -gRNA and Cas9 protein (Fig. 4c). The results showed that the KI rates were 17%, 17%, 14%, and 10%, respectively (no significant differences) with the development rates of these blastocysts being 76%, 73%, 78%, and 80%, respectively (Fig. 4c). Alternatively, the results obtained from similar experiments were conducted using 1.36 kb dsDNA and 1.36 knt ssDNA, the KI rates being 0%, 20%, 0%, and 11%, respectively, with the development rates of these blastocysts at 80%, 87%, 68%, and 73%, respectively.

Based on these results, we observed that ssDNA generally leads to a higher KI editing rate compared to dsDNA. When the size of the KI DNA exceeded 1.80 kb for dsDNA or 1.80 kb for ssDNA, the efficiency of KI editing decreased. Additionally, using two-cell embryos tended to result in a higher KI editing rate compared to using fertilized eggs.

Direct generation of floxed adm mice via sequential EP of fertilized egg and then in two-cell-stage embryos

Genome editing of fertilized eggs through EP has been well-studied21,22,23. If EP-based genome editing in two-cell embryos can be easily implemented, it will pave the way for sequential genome editing using fertilized eggs and two-cell embryos, increasing opportunities for genome editing32. To assess this challenge (Fig. 5a), we designed and synthesized 2 gRNAs, Adm-gRNA-A and Adm-gRNA-D targeting the Adm gene (Table S4), and 2 loxP sequences, Adm-loxP ssODN-A and Adm-loxP ssODN-D for KI on the Adm gene (Table S5). The EP was performed under the same conditions described, except that the voltage was reduced from 20 to 15. During the EP of the two-cell-stage embryos, Type A orientation was performed (Figs. 1 and 2a). The results are summarized in Fig. 5b. First, 60 fertilized eggs were subjected to EP with AM-gRNA-D (200 ng/µL), Cas9 protein (50 ng/µL), and Adm + loxP ssODNs-D (400 ng/µL). Second, the following day, the 60 two-cell stage embryos derived from the electroporated fertilized eggs were subjected to a second round of electroporation with AM-gRNA-A (200 ng/µL), Cas9 protein (50 ng/µL), and Adm + loxP ssODNs-A (400 ng/µL). The resulting embryos were then transferred into pseudopregnant ICR females. We examined the fetuses at 15.5 days post-coitum because knockout for Adm results in fetus lethality37. Consequently, we obtained 13 fetuses from a total of 60 two-cell stage embryos and observed no lethal fetuses. To evaluate the KI rate of Adm loxP ssODN-A and D sequences, we performed an RFLP assay using EcoRI enzyme on the PCR product of the target sequence region (Fig. S5) amplified by genomic DNA isolated from the limbs of the 13 fetuses. A 46% KI rate of loxP ssODN-D on the Adm gene was found in the fertilized eggs. The KI rate of loxP ssODN-A on the gene was 54% in the two-cell-stage embryos. Based on the analysis of In vitro Cre assay (Fig. 5c and d), the floxed KI rate of the Adm gene was determined to be 15% (2 of 13 individuals). These results demonstrate that the modified EP of genome editing at the two-cell stage contributes to the overall success of genome editing in both fertilized eggs and two-cell-stage embryos.

Fig. 5
figure 5

Sequential electroporation (EP) of fertilized egg and two-cell-stage BDF1×B6 embryos for directly generating floxed Adm mice. (a) Schematic illustration of the experimental procedure based on the sequential EP method32. (b) Table summarizes the result of sequential EP of fertilized egg stage and then two-cell-stage embryos. (c) In vitro Cre and PCR analysis for identifying a founder mouse with floxed Adm allele. (d) Agarose gel electrophoresis of PCR products amplified from the in vitro Cre-treated genomic DNA of the fetus (no. 1, 3, 9, 11, 12, 13, as shown in Fig. S5). Fetus no. 1, 3, 9, and 11 are positive for both loxP sites on the Adm locus. Fetus no. 12 and 13 have only loxP-A and -D on the Adm locus, respectively, and were used for PCR control. This result indicates that fetuses 3 and 9 are founder mice with the floxed Adm allele. M, Lambda HindIII + 100-bp ladder markers. (e) Sequence chromatograms of PCR products from fetuses 3 and 9, as detected in (d).

Full size image

Discussion

In this study, we improved the EP method to prevent blastomere fusion in both fresh and thawed two-cell stage-embryos, thereby enabling efficient genome editing. Additionally, we demonstrated for the first time that Cas12a is capable of efficient genome editing in two-cell-stage embryos. A key characteristic of genome editing in two-cell-stage embryos, as revealed by our study findings, is that indel mutation rates and ssODN KI efficiencies in two-cell-stage embryos were equivalent to those in fertilized eggs. Furthermore, both two-cell-stage embryos and fertilized eggs showed comparable levels of mosaicism. Notably, two-cell-stage embryos tended to exhibit higher KI efficiencies for longer DNA sequences than fertilized eggs. When the DNA sequence to be knocked in is within 1.3 kb, EP in two-cell-stage embryos may offer a more efficient approach for generating KI mice than fertilized eggs.

A concern when using two-cell-stage embryos for EP is the phenomenon of blastomere fusion32,34,35. Horii et al.38 addressed the issue of approximately 20% blastomere fusion in two-cell stage embryos caused by EP. They successfully reduced the blastomere fusion rate to approximately 10% or less without altering KI efficiency by treating two-cell stage embryos with calcium (Ca)-free conditions or actin polymerization inhibitors prior to EP. Here, we demonstrated that this fusion phenomenon of blastomeres can be easily avoided by incorporating a modification into the EP procedure, where the two cleavage planes of the two-cell-stage embryos are aligned horizontally with the electrodes. In this study, except for two instances where it seemed there was a maladjustment of the cleavage planes of the two-cell-stage embryos, no examples of blastomere fusion were observed. Adjusting the cleavage planes of the two-cell-stage embryos to be horizontal with the electrodes is relatively straightforward once accustomed. Currently, when performing EP, we are able to line up around 60 two-cell stage embryos in the gap between the platinum electrodes. This method can be readily adopted by researchers and technicians conducting embryo genome editing via EP using fertilized eggs.

It was crucial to determine whether frozen-thawed two-cell-stage embryos could be used for embryo genome editing via EP. We demonstrated that frozen-thawed two-cell-stage embryos exhibit indel mutation abilities comparable to those of fresh two-cell-stage embryos. The indel mutation rates targeting the Tyr, Adm, Ramp1 genes were equivalent between frozen-thawed and fresh two-cell-stage embryos. Furthermore, we observed high rates of indel mutations in fresh two-cell-stage embryos in 6 genes: 5 genes (Klf5, Gt(ROSA)26Sor, Ddx58, Tlr4, and Tlr9) targeted by Cas9, and one gene (Adm2) targeted by Cas12a. In addition, the conditions for EP of this two-cell-stage embryo were effective with no difference in the BDF1xB6, B6, ICR, and 129/Sv strains. Thus, there is no difference between frozen and thawed two-cell-stage embryos and fresh two-cell-stage embryos, and they can be used to produce genetically modified animals.

Comparing the embryo genome editing abilities between two-cell-stage embryos and fertilized eggs is a critical aspect. We demonstrated that the indel mutation rate targeting the Tyr gene in two-cell-stage embryos is equivalent to that in fertilized eggs. Additionally, the coat color of mice with indel mutations in the Tyr gene differed between embryos derived from the black BDF1xB6 mouse strain, with 100% exhibiting white fur from two-cell-stage embryos, compared to 93% from fertilized eggs of the same strain, indicating that two-cell-stage embryos possess indel mutation abilities comparable to those of fertilized eggs.

When utilizing two-cell-stage embryos for embryo genome editing, a concern was the possibility of higher mosaic rates compared to fertilized eggs. However, unexpectedly, the results obtained in this study showed no difference in mosaic rates between two-cell-stage embryos and fertilized eggs. One possible reason for this might be the cell cycle of the two-cell-stage embryos. In our study, we used two-cell-stage embryos corresponding to the developmental stage at 30–33 h post-fertilization, during which time the cell cycle is in the G2 phase. The G2 phase of two-cell-stage embryos has a longer duration of 10–11 h39,40. CRISPR/Cas components are introduced into the embryo during the G2 phase, inducing Indel or KI mutations, and subsequently degraded. Consequently, there may be no difference in the mosaic rate between two-cell-stage embryos and fertilized eggs. This outcome suggests that two-cell-stage embryos represent a comparable resource for embryo genome editing as fertilized eggs. In some cases, gRNAs with nucleotide-resistant modifications are used to prolong their functionality without degradation within the fertilized egg, but the use of these modified gRNAs may increase mosaic rates in two-cell-stage embryo genome editing.

In this study, we investigated the DNA KI efficiency in two-cell-stage embryos using ssODN (< 191 nt), dsDNA (0.66–1.80 kb), and ssDNA (0.66–1.80 knt).

Regarding ssODN KI, the rate of loxP sequence KI to the Adm gene in offspring mice was 53%, with 25% of them being homozygous for loxP/loxP. Furthermore, the KI of the Flagx3 sequence to the Klf5 and Or7a36 genes in blastocysts was achieved at rates of 63% and 39%, respectively. Previously, we reported a KI rate of 44% for the Flagx3 sequence to the Klf5 gene in fertilized eggs36. These results indicate that two-cell-stage embryos can knock in short DNA fragments at rates comparable to fertilized eggs. However, the KI rate of loxP sequences into the Ramp3 gene remained low in two-cell-stage embryos when using Ramp3-gRNA, which is known for its low editing efficiency in fertilized eggs. To achieve high KI efficiency, appropriate gRNA and ssODN settings are crucial. The KI rate of the EGFP sequence to the Ramp2 gene tended to be higher with ssDNA (1.23 knt) compared to that with dsDNA (1.23 kb). Similar trends were observed in fertilized eggs. This was also observed in the KI rate of the U6Tyr sequence to the Gt(ROSA)26Sor gene. However, when the KI DNA sequence length was between 1.80 (kb or knt), KI events were not observed in fertilized eggs or two-cell-stage embryos. This result aligns closely with the DNA size reported for KI through EP in fertilized eggs25,26, and the efficiency appears to be comparable in two-cell-stage embryos. Notably, our analyses demonstrated that KI rates tend to be higher in two-cell-stage embryos than in fertilized eggs, supporting the observations of Gu et al.30 In instances where the DNA sequence to be knocked in is less than 1.3 kb, genome editing at the two-cell stage may be more advantageous than fertilized eggs, suggesting that this straightforward EP method could be particularly beneficial. Conversely, Gu et al. successfully achieved the KI of a 7.5 kb DNA fragment into two-cell-stage embryos through microinjection30. This indicates that the length of the KI DNA sequence in our two-cell-stage embryos may be limited by the efficiency of delivery via EP into both the zona pellucida and the embryo itself. Additionally, Bagheri et al.31 demonstrated the successful KI of approximately 7-kb-long DNA by injecting all CRISPR components along with a DNA template into the sub-zona pellucida space of two-cell-stage embryos, followed by EP; however, their KI rates ranged from 2.2 to 3%. We also attempted to introduce a 5.5 kb plasmid into the sub-zona pellucida space of two-cell-stage embryos by creating an opening in the zona pellucida using a piezoelectric micromanipulator, followed by EP under the current conditions. However, successful plasmid integration was not achieved. The differences observed may stem from variations in EP conditions, indicating that further investigation is warranted to improve KI efficiency. Recently, Davis et al.34 introduced a method for delivering DNA templates into two-cell-stage embryos using adeno-associated virus (AAV), similar to the approach employed by Chen et al.41 in fertilized eggs, wherein CRISPR/Cas9 components were delivered through EP. This approach yielded high KI rates for long DNA fragments. However, as the use of AAVs may not be readily adopted in many laboratories, various modifications are required in the future.

The advantage of enabling genome editing via EP in two-cell-stage embryos is the extended period available for embryo genome editing. This implies that various embryo genome editing techniques, such as Indel mutations, ssODN KI (< 150 nt), and DNA KI up to 1–1.5 kb (knt), can be performed multiple times from fertilized eggs to two-cell-stage embryos. Indeed, Horii et al.32,38 developed the “sequential EP method,” where they sequentially introduced loxP sequences into fertilized eggs and two-cell-stage embryos to demonstrate the production of floxed mice. Bernas et al.33 also successfully applied this method to generate floxed mice. Therefore, adopting an EP method for two-cell-stage embryos that do not cause blastomere fusion, we attempted to create floxed mice of the Adm gene using the sequential EP method. As expected, we efficiently obtained floxed mice for the Adm gene without blastomere fusion in two-cell-stage embryos. This example suggests that addressing the extended embryonic development period from fertilized egg to the two-cell stage for genome editing could be valuable, paving the way for the establishment of new methods for developing various embryonic genome editing technologies.

In conclusion, this study emphasizes the potential of two-cell-stage embryos as a valuable resource for genome editing and enhanced animal model production. Moreover, realizing genome editing in two-cell-stage embryos via EP extends the window for diverse editing techniques from fertilized eggs, offering a promising avenue for future research in embryo genome editing techniques.

Methods

Ethics statement for animal experimentation

All animal experiments were conducted in accordance with the ethical guidelines of Shinshu University. All experiments were approved by the Shinshu University Ethics Committee for Animal Experiments (Permit No. 024022). This study was conducted in accordance with the Reporting of In Vivo Experiments on Animals (ARRIVE) guidelines. Euthanasia was carried out using a combination of 0.3 mg/kg medetomidine (Nippon Zenyaku Kogyo Co. Ltd., Koriyama, Japan), 4.0 mg/kg midazolam (Astellas Pharma Inc., Tokyo, Japan), and 5.0 mg/kg butorphanol (Meiji Seika Pharma Co. Ltd., Tokyo Japan), followed by cervical dislocation.

Embryo culture and EP solutions

For embryo manipulation, KSOM and 25 mM HEPES-containing KSOM were prepared following the protocol established by Sakurai and Shindo19. For RNA and DNA solutions, 75% phosphate-buffered saline solution without Ca²⁺/Mg²⁺ (PBS, Thermo) was used. To prepare 75% PBS, PBS and double-distilled water were mixed at a ratio of 3:1. The EP medium consisted of a 1:1 mixture of Opti-MEM (Thermo Fisher Scientific K.K., Tokyo, Japan) and 75% PBS.

Preparation of EGFPA95 mRNA, gRNA, Cas9 RNP, ssODN

The preparation of EGFPA95 mRNA was described by Sakurai et al.36. All gRNAs used in this study, except for Adm2 (Table S4) were prepared through the T7 RNA polymerase-mediated synthesis described by Sakurai et al.42. Adm2 gRNAs (Table S4) were synthesized by Integrated DNA Technologies, Inc. (IDT; Coralville, Iowa, USA). Cas9 and Cas12a proteins were purchased from IDT.　Cas RNPs were prepared by incubating both gRNA(s) and Cas protein in a 1:1 mixture of Opti-MEM and a 75% PBS for 20–30 min at 23–25 ℃ before EP. Adm + loxP ssODN-A (140 nt), Adm + loxP ssODN-D (181 nt), Klf5 + Flagx3 ssODNs (169 nt), Or7a36 + Flagx3 (191 nt) and Ramp3 + Flagx3 ssODNs (169 nt) were synthesized by IDT as dried materials. The ssODN sequences are shown in Table S5.

Preparation of long ssDNA, and dsDNA

The double-strand (ds) DNA and single-strand (ss) DNA sequences of Ramp2-EGFP KI and Gt(ROSA)26Sor -U6 tyr KI are shown in Table S6. For the 1.8 kb dsDNA of Ramp2-EGFP KI, the 5’ arm (510 bp) and 3’ arm (530 bp) regions of Ramp2 were first amplified using primers R2-5 S/R2-5 A and R2-3 S/R2-3 A with C57BL/6J genomic DNA as the template, and Tsk Gflex DNA polymerase (Takara Bio Inc., Shiga, Japan). EGFP cDNA was amplified using primers EGFP-S/A with pR2HR (Sakurai et al., 2014) as the template. The construct pRamp2KI-EGFP was created by joining the three PCR fragments with pBluescript II (EcoRV site) using Gibson assembly (NEB Japan Inc., Tokyo, Japan), and the sequence fidelity was confirmed. Next, the R2-5 S primer was phosphorylated using T4 polynucleotide kinase (Takara). The dsDNA of Ramp2-EGFP KI as PCR products was then amplified with phosphorylated R2-5 S and R2-3 A using pRamp2-EGFP KI as the template and Tsk Gflex DNA polymerase. This PCR product was used for the KI experiment as dsDNA. Finally, 1.8 knt of ssDNA were prepared from the dsDNA using the Guide-it™ Long ssDNA Production kit (Takara Bio Inc., Shiga, Japan). The ssDNA was purified via ethanol precipitation, dissolved in 75% PBS, quantified using absorption spectroscopy and agarose gel electrophoresis, and stored at − 20 °C until use.

Additionally, 1.23 kb of dsDNA and 1.23 knt of ssDNA of Ramp2-EGFP KI were prepared through PCR using pRamp2-EGFP KI as the template and the ssDNA Production kit with the dsDNA, respectively. Similarly, long dsDNA (0.66 and 1.36 kb) and ssDNA (0.66 and 1.36 knt) of Gt(ROSA)26Sor -U6 tyr KI were prepared. The construct p Gt(ROSA)26Sor -U6 tyr-gRNA KI was created using Gibson assembly with 400 bp of the Gt(ROSA)26Sor 5’ arm, 500 bp of the 3’ arm regions, and 460 bp of U6 tyr-gRNA. The ssDNAs were prepared via PCR using p Gt(ROSA)26Sor -U6 tyr-gRNA KI as the template, followed by the ssDNA Production kit for the dsDNA.

Mice, fertilized eggs, and two-cell-stage embryos

This study used B6D2F1×B6 hybrid (sCAT without Cas9 transgene19) and 129+ Ter/SvJcl (129, CLEA Japan, Tokyo, https://www.clea-japan.com/en/products/inbred/item_a0460). The hybrid and 129 mice were maintained in-house. C57BL/6J (B6, aged 3–10 weeks) were purchased from THE JACKSON LABORATORY JAPAN, INC (Yokohama, Japan). ICR mice (aged 8–10 weeks) were obtained from ICR (CLEA Japan (Tokyo, Japan), SLC (Shizuoka, Japan).

Fertilized eggs were obtained via a standard IVF protocol described by Sakurai and Shindo19 and were used for experiments approximately 9–10 h after insemination (time of insemination = 0 h). Two-cell-stage embryos were obtained by culturing the fertilized eggs in KSOM media19 in vitro and were used for experiments 30–33 h after insemination. Additionally, we used some two-cell-stage embryos that were frozen by vitrification and thawed, following Nakao et al.,29. The thawed two-cell-stage embryos were used in the experiments at the equivalent of 30–33 h after insemination.

EP, in vitro embryo culture, and embryo transfer

EP was performed with a CUY21EDIT electroporator II (BEX Co., Tokyo, Japan) and platinum electrodes set in a plastic plate (#LF501PT1-10; 1 mm gap, 10 mm length, 3 mm width, and 0.5 mm height; BEX Co.) under a stereoscopic microscope (Fig. 1a, b, c). We used the method of Hashimoto and Takemoto21 with some modifications. Briefly, we used 25 mM HEPES-containing KSOM19 and a 1:1 mixture of Opti-MEM (Thermo) and 75% PBS for the EP media containing CRISPR/Cas components (gRNPs; gRNPs and (ssODN or ssDNAs or dsDNAs)). The overview of the EP system and procedure in this study are shown in Fig. 1. Before EP, the embryos were washed with the 1:1 mixture media and then washed with the 1:1 mixture containing CRISPR/Cas components. A 5-µL drop containing several sets of CRISPR/Cas components in the 1:1 mixture was placed between the electrodes and kept under observation with a dissecting microscope, Next, 5–60 embryos (washed) were placed into the drop. In the case of the two-cell-stage embryo, the contact surface of the two blastomeres was oriented perpendicular to the electrode. EP was carried out at 15–20 V, alternating between on and off for 3 ms and 97 ms, respectively, one to five times as one set of EP. In some experiments, two to three sets of EPs were performed. After EP, the embryos were immediately collected from the electrode chamber and washed with 25 mM HEPES-containing KSOM. Electroporated fertilized eggs and some electroporated two-cell-stage embryos were cultured in KSOM medium up to the blastocyst stage at 37 °C under conditions of 95% humidity and 5% CO2 or were transferred into the oviducts of pseudopregnant ICR females to develop to full term.

Analyses of genome editing

We performed genome editing on a total of nine loci (Adm, Adm2, Ddx58, Klf5, Ramp1, Ramp2, Ramp3, Gt(ROSA)26Sor, Tlr4, Tlr9, and Tyr) for indel editing and six loci (Adm, Klf5, Or7a36, Ramp2, Ramp3, and Gt(ROSA)26Sor) for KI editing. For blastocyst analyses, a crude DNA solution was prepared from each single blastocyst42. For fetus analyses, genomic DNA was isolated from the fetus’s hand. For pup analyses, genomic DNA was isolated from a part of the pup’s ear. PCR primers for amplifying regions spanning the mutated and KI sequences are shown in Table S5. PCR was performed using Tsk Gflex DNA polymerase (Takara) with the following conditions: 98 °C for 10 s and 68 °C for 60 s, for a total of 35 cycles (45 cycles for blastocyst samples). To analyze indels, the PCR products of each sample were sequenced using a BigDye Terminator Cycle Sequencing Kit ver3.1 and an ABI Genetic Analyzer 3130 (Applied Biosystems, Life Technologies Japan, Ltd., Tokyo, Japan). The data were analyzed using Genetyx-Mac ver.13.0.3 (Software Development Co. Ltd., Tokyo, Japan), ClustalW (http://www.genome.jp/tools-bin/clustalw), and Inference of CRISPR Edits (ICE) software (Synthego Corporation, Silicon Valley, CA, USA). For KI analyses, the presence of ssODN KI at the Adm locus was determined by EcoRI sites in the PCR product generated by amplifying the region targeted by Adm-gRNA. Similarly, the presence of ssODN KI at the Klf5 and Ramp3 loci was determined by the presence of ClaI sites in the PCR products. The ssODN KI at the Or7a36 locus was determined by the size of PCR products. Successful KI of long dsDNA and ssDNA of the EGFP fragments into the Ramp2 locus, and the U6 Tyr-gRNA sequence into the Gt(ROSA)26Sor locus, were determined by the size of PCR products generated by amplifying with primers located outside the homologous arms. The PCR products were sequenced as described above. The in vitro Cre assay was conducted following Sentmanat et al.43.

Statistical analysis

Data are presented as means ± SE. Percentage data were arcsine-transformed before statistical analysis. The number of indels per genome and the KI efficiency of both dsDNA and ssDNA were analyzed using Tukey-Kramer’s test or Welch’s t-test. Statistical significance was determined at a threshold of P < 0.05.