Targeted mutagenesis in chicken using CRISPR/Cas9 system

The CRISPR/Cas9 system is a simple and powerful tool for genome editing in various organisms including livestock animals. However, the system has not been applied to poultry because of the difficulty in accessing their zygotes. Here we report the implementation of CRISPR/Cas9-mediated gene targeting in chickens. Two egg white genes, ovalbumin and ovomucoid, were efficiently (>90%) mutagenized in cultured chicken primordial germ cells (PGCs) by transfection of circular plasmids encoding Cas9, a single guide RNA, and a gene encoding drug resistance, followed by transient antibiotic selection. We transplanted CRISPR-induced mutant-ovomucoid PGCs into recipient chicken embryos and established three germline chimeric roosters (G0). All of the roosters had donor-derived mutant-ovomucoid spermatozoa, and the two with a high transmission rate of donor-derived gametes produced heterozygous mutant ovomucoid chickens as about half of their donor-derived offspring in the next generation (G1). Furthermore, we generated ovomucoid homozygous mutant offspring (G2) by crossing the G1 mutant chickens. Taken together, these results demonstrate that the CRISPR/Cas9 system is a simple and effective gene-targeting method in chickens.

OVA and OVM mutation by CRISPR/Cas9 in chicken PGCs. We next examined whether OVA and OVM in chicken PGCs could be disrupted by transfection with the two selected pX330 plasmids. Cultured PGCs derived from Barred Plymouth Rock (BPR) male embryos were transiently transfected with pX330-Neo-OVATg3 or pX330-Neo-OVMTg2, selected with or without 0.5 mg/ml neomycin at 2-4 days post-transfection, and then cultured for up to 2 weeks without neomycin. A subset of the PGCs was collected, and OVA and OVM sequences around target sites were PCR amplified and analyzed following TA cloning. We found OVA and OVM mutations with frequencies of 34% and 13%, respectively, in the TA clones derived from neomycin-treated PGCs in contrast to respective mutation frequencies of 6.7% and 0% without neomycin selection ( Table 1). Given that neomycin treatment resulted in more-efficient mutation of the two genes, transient antibiotic selection was efficacious for CRISPR/Cas9-mediated gene editing in PGCs. To potentially increase the proportion of mutant alleles in PGCs with the CRISPR/Cas9 method, we tested two additional antibiotic selection techniques, puromycin and zeocin, after insertion of their corresponding resistant genes into the pX330 plasmid. As shown in Table 1, transient Co-transfection with pCAG-EGxxFP-OVM and pX330 plasmids containing OVM sgRNA sequences OVMTg1-4 or pX330 without sgRNA sequences. Right two images in (a,b) show EGFP fluorescence generated by co-transfection with pCAG-EGxxFP-Centrin1 (Cetn1) and pX330/Cetn1 plasmids as an experimental control and by transfection with pCAG-EGFP as a positive control. Scale bar = 100 μm. (c,d) Percentage of EGFP-positive cells relative to all cells from experiments carried out as in (a,b), which reflects the efficiency of DSB-mediated homology-dependent repair. Data are shown as the mean ± standard deviation (SD) from four regions of a single culture. antibiotic selection at 2-4 days post-transfection with puromycin (1 μg/ml) and zeocin (50 μg/ml) resulted in mutational frequencies of 100% and 92% in OVA and 91% and 92% in OVM, respectively. These high frequencies of mutations with puromycin and zeocin as compared with neomycin can be attributed to the differences in antibiotic sensitivities of untransfected PGCs. Cultured chicken PGCs were rather more sensitive to puromycin and zeocin than to neomycin (Fig. S1). Thus PGCs in which the pX330 plasmid was not successfully transfected might survive after the transient treatment of neomycin as compared with exposure to puromycin and zeocin. In the sequenced clones, 9 mutational classes with 1-to 10-bp deletions in OVA and 12 classes with 1-to 21-bp deletions in OVM were observed (Fig. 3). Notably, although various deletion mutations were detected, no insertion mutations were found in the analyzed sequences. The majority of these deletions induced frameshift mutations and consequent changes in OVA and OVM proteins. These results show that endogenous genes can be efficiently targeted by CRISPR/Cas9 mutagenesis by using proper antibiotic selection in chicken PGCs. Consistent with this idea, expression of a PGC-specific endogenous protein, chicken vasa homolog (CVH) 21 , was frequently reduced or absent after transfection of a pX330-Puro plasmid encoding CVH-specific sgRNAs followed by transient selection with puromycin ( Fig. S2).
Generation of OVM heterozygous mutant chickens. Cultured chicken PGCs differentiate into functional gametes after transplantation into recipient chicken embryos 7 . To determine whether the PGCs carrying CRISPR/Cas9-induced mutations have germline competency and the capacity to generate offspring, we produced germline chimeric chickens and analyzed their offspring (Fig. 4a). Because OVA mutant chickens had already been generated 13 , we focused on the generation of novel OVM mutants. PGCs from the same pool of cells that were transfected with pX330-Puro-OVMTg2, transiently selected with puromycin, and sequenced were used as donor cells, and thus the vast majority (~90%) were expected to contain mutated OVM. The donor PGCs, which were derived from the BPR strain and thus carry the recessive gene (i/i) for black feather color, were expanded and transplanted into the blood of recipient chick embryos from the White Leghorn (WL), which carries the gene for white feather color (I/I). Prior to transplantation, recipient WL embryos were exposed to 5-6 Gy gamma radiation to ablate endogenous PGCs and increase the contribution from donor PGCs 8 . Three presumptive male germline chimeras (G0) were raised to sexual maturity, and the OVM genomic DNA in their semen was PCR amplified, TA cloned, and sequence analyzed. More than 90% of TA clones from two roosters (#372 and #376) exhibited OVM mutations with 1-to 12-bp deletions around the OVMTg2 target site ( Table 2 and Fig. 4b). These two roosters were crossed with wild-type BPR hens (i/i), and the chicks that hatched were analyzed for feather color. Analysis of the progeny from the roosters demonstrated that the frequency of donor-derived (i/i) progeny as assayed by black feather color was 79% (#372) and 67% (#376). Sequence analysis revealed that 58% (#372) and 48% (#376), respectively, of the donor-derived progeny had an OVM mutation with a 1-to 31-bp deletion ( Table 2 and Fig. 4c-d). These results indicate that the donor PGCs mutagenized by the CRISPR/Cas9 system could differentiate into functional gametes and produce genetically modified progeny. From among the monoallelic OVM mutant chicks, males and females with frameshift mutations were selected as heterozygous OVM knockout (OVM +/− ) chicks. All the OVM +/− chicks were healthy and showed no obvious abnormalities.

Off-target and plasmid integration analysis.
In previous studies, frequent off-target mutations have been observed in CRISPR-mediated mutagenesis in mammalian cell lines 22,23 . To evaluate the off-target effects of CRISPR in chickens, we analyzed potential off-target sequences in the genomes from these mutant chickens.
Three candidates were chosen, as they carried sequences that matched the OVMTg2 seed sequence: two covered a 12-nucleotide stretch followed by the 5′ -NGG-3′ protospacer adjacent motif (PAM) sequence; the other, a span of 17 nucleotides, was followed by the 5′ -NAG-3′ PAM sequence (Table S2). These sequences were PCR amplified and analyzed by sequencing. For a total of 43 sequences from three potential off-target sites present in 19 mutant chickens, we did not detect any mutations (Table S2). These off-target mutations were not observed in genomic DNA from transfected PGCs that had not been transplanted into recipient embryos (Table S2). These results suggest that off-target mutations occur infrequently in the OVM mutant chickens and PGCs generated by the CRISPR/Cas9 system. To analyze the possibility of plasmid integration in knockout chicks, we searched for the puromycin resistance gene in the genome of OVM +/− chicks by PCR analysis (Fig. S3). As the designed primer set could detect 10 fg of plasmid DNA, this assay should be able to detect as few as 0.05 copies of the puromycin resistance gene when 10 ng of genomic DNA is analyzed. No obvious transgene product was detected in eight OVM +/− chicks, suggesting that random plasmid integration in the G1 mutated chicken genome occurs at a low frequency. We also examined puromycin sensitivity of PGCs that were transfected but not subsequently transplanted (Fig. S3). The PGCs were sensitive to puromycin, suggesting that random integration of the plasmid rarely occurred in the PGCs.
Generation of OVM homozygous mutant chickens. Finally, we examined whether homozygous OVM knockout chickens can be generated by crossing OVM +/− males and females. A sexually mature G1 male (#529) and G1 female (#475), which have an OVM mutation with a 19-and 4-bp deletion, respectively, were crossed, and the genomes of G1 parents and G2 progeny were analyzed by fragment analysis 24 (Fig. 5a,b). As shown in Fig. 5b, the frequency of OVM −/− progeny was 24% (8/33), corresponding to Mendelian inheritance. Both male and female OVM −/− chicks hatched and were healthy (Fig. 5c). This result indicated that the G1 mutations introduced with the CRISPR/Cas9 system do not noticeably affect chromosomal distribution during meiosis or gamete differentiation.

Discussion
To our knowledge, this study represents the first example of the generation of CRISPR/Cas9-mediated mutant chickens. A single plasmid transfection and transient antibiotic selection resulted in efficient targeted mutation in PGCs (> 90%), and, therefore, chimerism of spermatozoa in PGC-transplanted G0 chimera chickens could easily be evaluated. The percentage of heterozygous OVM mutant chicks among the donor-derived offspring (G1) was around 50% from two of the three G0 founders (Table 2), and the majority of mutants resulted in frameshift mutations (Fig. 4e). In addition, OVM −/− mutant offspring (G2) were obtained by crossing a OVM +/− male and female G1 chicken, indicating that the CRISPR/Cas9 system is quite useful for generating gene-specific knockout in chickens. To date, conventional homologous recombination-based approaches and the TALEN method have been used to generate knockout chicken 12,13 . Because CRISPR/Cas9-mediated knockout requires simple plasmid  construction compared with the other methods 14,15 , our results validate this as a convenient and easy-to-use method for gene targeting in chickens.
In contrast to several other organisms, zygote manipulation in chickens has rarely been successful 4 . Therefore, instead of zygotes, germline-competent cultured PGCs have been used to generate genetically modified chickens. Several transgenic and knockout chickens were generated by gene alteration of PGCs followed by transplantation of the PGCs into recipients, rearing of the recipients to sexual maturity, and identification of the genetically modified offspring 7-13 . Here we showed induction of PGC mutations in two genes with >90% efficiency using the CRISPR/Cas9 system in combination with antibiotic selection (Table 1). Park et al. have investigated PGC mutation using TALEN with fluorescence-activated cell sorting selection and established mutant offspring. They observed PGC mutation in OVA with 33.3% efficiency 13 . The differences in the mutational frequencies between the two studies reflect the differential mutational efficiencies of TALEN and CRISPR/Cas9. Consistent with this idea, comparative analyses of TALEN-and CRISPR/Cas9-mediated genetic mutation have shown a superior mutational frequency induced by CRISPR/Cas9 in vitro and in vivo 25,26 .
A potential disadvantage of the CRISPR/Cas system is the possibility of off-target effects. In the current study, we observed no obvious effects of the OVMTg2 sgRNA/Cas9 combination on potential off-target sites in the mutant chickens (Table S2). Several studies have reported frequent off-target mutagenic effects of CRISPR/Cas9 in transformed cell lines 22,23 ; however, it has also been reported that off-target mutations are very rare and not a serious problem for CRISPR/Cas9-mediated genome editing in human stem cells 27,28 . Although additional targeted and/or genome-wide analyses may be required for precise evaluation of off-target effects associated with mutagenesis by CRISPR/Cas9 in chickens, our results suggest that specific gene targeting occurs with the CRISPR/Cas9 system in chicken PGCs.
Both OVA and OVM are major allergenic proteins. Of the egg white proteins, OVA is the most abundant allergenic protein, whereas OVM is the dominant allergen 29 . Hens with homozygous disruptions in these allergen genes may produce low-allergenicity eggs, which could be tolerated by a person who has egg white allergies. In particular, allergenicity of OVM has been hard to eliminate from egg whites because of the strong stability of OVM during heating and proteolytic digestion 30,31 ; therefore, genetic ablation of OVM may provide a new and effective method to reduce egg white-based allergic reactions. Heterozygous mutants in OVA have been generated by TALEN 13 , and in the current study we established heterozygous and homozygous OVM mutants by CRISPR/Cas9 (Fig. 4). It will be of immediate interest to test whether eggs from the homozygous knockout hens show reduced allergenicity. Though time intensive, it is possible to envisage crossing OVA and OVM mutant chickens to produce lines carrying mutations in multiple allergy-related genes. Alternatively, given the efficient mutagenesis in chicken PGCs by CRISPR/Cas9, hens with the disruption of multiple allergenic genes may be generated more rapidly via their simultaneous disruption in PGCs.
Of the three G0 germline chimeras from this study, two (#372 and #376) were estimated to have a relatively high contribution from the mutated donor sperm, but one (#387) had a low contribution in this regard ( Table 2). A difference in the proportion of genetically modified sperm in G0 individuals has frequently been observed in PGC-transplanted germline chimeras 7,8 and is presumed to result from differences in donor PGC contribution in the recipient testis. The variability of germline transmission among chimeras may be attributable to variation in the number of injected PGCs, bleeding and/or vascular occlusion after PGC transplantation, and differential expansion of PGCs within the gonad 7 . Because this variability is hard to circumvent, estimation of donor contribution in G0 semen and the subsequent avoidance of chickens with low donor contribution as founders are important for the efficient establishment of genetically modified chickens in the G1 population 8 . Although donor (BPR) and recipient (WL) sperm are hard to distinguish by DNA analysis in general, we could easily estimate the relative donor contribution in G0 semen based on OVM sequence analysis (Table 2) because of the high frequency of OVM mutations in transplanted PGCs ( Table 1). The evaluation of donor contribution in G0 semen without the need for test crosses can reduce the number of experimental animals needed, and thus chicken mutagenesis mediated by the CRISPR/Cas9 system in this manner may have an advantage with regard to the ethical treatment of animals.
To generate homozygous mutant chickens, the current methods of chicken genome editing (TALEN and CRISPR/Cas9) require the interbreeding of the G1 generation of heterozygous mutant male and female chickens. The interbreeding is not difficult but requires a time-consuming process that takes ~6 months for chickens of the G1 generation to reach sexual maturity. To shorten the time needed for generating homozygous mutant chickens, it is conceivable that the G0 male and female chimeras carrying mutated gametes could be interbred. In addition to males, if G0 female germline chimeras were generated, homozygous mutant chickens could be obtained in G1, one generation earlier than is achieved with the current method. Because these technologies have thus far been unsuccessful in female-derived PGCs 7,8,32 , better methods for female-derived PGCs should be developed for the rapid generation of homozygous mutant chickens. Culture and mutation methods for female-derived PGCs would also enable the introduction of mutations on the female-specific W chromosome.
We have demonstrated a simple and efficient method to generate knockout chickens using the CRISPR/Cas9 system in PGCs. Because the chicken embryo has been an important model for studying developmental biology 33 , a convenient knockout technology will be a powerful tool not only for the improvement of chicken products but also for developmental studies including chicken embryonic pattern formation, tissue morphogenesis, and organogenesis. In many types of cells and organisms, CRISPR/Cas9 has been used to generate not only knockout but also knock-in genotypes. In addition, CRISPR/Cas technology has been expanded to include various genome-engineering applications such as large-scale genomic deletions, transcriptional modification, and epigenetic regulation 34,35 . We envisage that many of these advances may be applicable in PGCs and chicken by modification of the method developed in our study. Expansion of the various genome-engineering technologies with the CRISPR/Cas system may be expected to lead to fundamental advances such as a greater understanding of PGC development and chicken embryonic morphogenesis, genetically based methods to improve chicken health and productivity, and production of knock-in chickens for agricultural and biomedical applications. Plasmid construction. The plasmids expressing hCas9 and sgRNA were generated by inserting complementary oligonucleotides corresponding to each target sequence (Fig. 1, Fig. S2, and Table S1) into the BbsI site of pX330 (Addgene, Cambridge, MA) 14 . For antibiotic selection, neomycin-, puromycin-, and zeocin-resistant genes were inserted into the NotI site of pX330. Two reporter plasmids for the SSA assay, pCAG-EGxxFP-OVA and pCAG-EGxxFP-OVM, were generated by ligating PCR-amplified genomic fragments of OVA (481 bp) and OVM (457 bp), respectively, containing the sgRNA target sites into the EcoRI site of pCAG-EGxxFP 20 . Primers for PCR amplification are shown in Table S1, and PCR was carried out with PrimeSTAR HS DNA polymerase (TaKaRa, Otsu, Japan), as follows: 98 °C for 2 min and 30 cycles of 98 °C for 10 s, 55 °C for 5 s, and 72 °C for 30 s. pCAG-EGxxFP vector and pX330/Cetn1, which was used as an experimental control for the SSA assay (see below), were kindly provided by Dr. Ikawa (Osaka University).

Methods
Scientific RepoRts | 6:23980 | DOI: 10.1038/srep23980 SSA recombination assay. The ability of pX330 plasmids (pX330-Neo-OVATg1-4 and pX330-Neo-OVMTg1-4) to disrupt the target gene was evaluated by the SSA assay as described 20 . In brief, HEK293T cells (4 × 10 5 ) were seeded into individual wells of poly-l-lysine-coated 6-well plates. After 24 h, co-transfection of pX330-sgRNA sequence-carrying plasmids and pCAG-EGxxFP target-site plasmids (0.5 μg each) was carried out with Lipofectamine 2000 (Life Technologies, Carlsbad, CA). After an additional 48 h, cells that were positive and those that were negative for EGFP fluorescence were scored in four regions (0.25 mm 2 each), and the percentage of positive cells relative to all cells was determined.
PGC culture and transfection. PGCs were isolated from blood of BPR embryos at stage 14-16 (Hamburger-Hamilton, HH) and incubated in forced-air incubators (P-008B Biotype; Showa Furanki, Saitama, Japan) at 38.5 °C and 60-80% relative humidity. Cells were cultured on Mitomycin C-treated buffalo rat liver (BRL) feeder cells in BRL-conditioned medium with supplements as described 36 . PGC transfection for genome editing was carried out using Lipofectamine 2000 according to the manufacturer's protocol with these minor modifications: (i) 1.6 μg of pX330-sgRNA sequence-carrying plasmid, and (ii) Lipofectamine 2000 reagents were diluted in 100 μl of OPTI-MEM (Thermo Fisher Scientific, Waltham, MA) and then were incubated with 0.5-1 × 10 5 PGCs for 5 min, followed by a 2-h incubation at 37 °C with 400 μl of antibiotic-free BRL-conditioned medium. Then, PGCs were transferred to 6-well plates with BRL feeder cells and were cultured at 37 °C, 5% CO 2 . PGCs were treated with or without 500 μg/ml neomycin (Nacalai, Kyoto, Japan), 1 μg/ml puromycin (InvivoGen, San Diego, CA), or 50 μg/ml zeocin (InvivoGen) from day 2 to day 4 after transfection to enrich for pX330-transfected cells. The remaining PGCs were then transferred to antibiotic-free medium and allowed to proliferate for transplantation and analysis.
Genomic DNA sequence analysis. Genomic DNA was extracted from PGCs, chicken semen, or shafts of chick feathers using the DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA). Genomic fragments containing sgRNA target sites were PCR amplified using MightyAmp DNA polymerase Ver.2 (TaKaRa) with specific primer sets (Table S1), TA cloned into pGEM-T Easy vector (Promega, Madison, WI), and sequenced with the ABI 310 Genetic Analyzer (Applied Biosystems, Foster City, CA). PCR conditions were as follows: 98 °C for 2 min and 40 cycles of 98 °C for 10 s, 60 °C for 10 s, and 72 °C for 30 s. For screening of OVM mutant chickens (G1), PCR products were purified using QIAquick PCR purification kit (Qiagen) and sequenced with specific primers (Table S1).
Gamma irradiation, PGC microinjection, and embryo culture. WL eggs were irradiated at 1 Gy/min with 137 Cs with a Gammacell 40 irradiator (Atomic Energy of Canada Ltd., Chalk River, ON, Canada). Irradiated embryos were allowed to develop to stage 13-15 HH, and then 1,000-2,000 PGCs, the majority of which contained OVM mutations based on an evaluation of mutation frequency (see Fig. 4a), were injected into the bloodstream of embryos as described 36 . The male embryos, as detected by PCR 32 , were allowed to develop in forced-air incubators with tilting at a 60° angle two times each hour at 38.5 °C and 60-80% relative humidity until hatch-out.
Progeny test. Male putative chimeric chickens were raised to sexual maturity. The proportion of mutated sperm was estimated by sequence analysis of the OVM target site region in TA-cloned PCR amplicons from semen genomic DNA. Two roosters estimated to have >90% mutated cells in their semen were crossed with wild-type female BPR (i/i) chickens by artificial insemination. Offspring that were derived from donor (BPR; i/i) and recipient (WL; I/I) PGCs were identified by their black (i/i) and white (I/i) feather color, respectively. The sequences of OVM target site regions in donor-derived offspring were analyzed as described above. For G2 production, a sexually mature OVM +/− male and female were crossed.
Off-target analysis. Potential off-targets of OVMTg2 sgRNA were predicted by a BLAST search of the 13 nucleotides located 5′ to the PAM motif consisting of 5′ -NGG-3′ 37 or of the 19 nucleotides 5′ to a possible PAM motif, 5′ -NAG-3′ 38 . The search was performed against the chicken genome sequence (Gallus_gallus-4.0 assembly; GenBank Assembly ID GCA_000002315.2) in the Ensembl database (http://www.ensembl.org/index.html), and three potential off-target sites were identified with the highest homology score in each criterion (Table S2). Genomic DNA from 19 mutant chickens was analyzed for off-target effects by PCR amplification, followed by sequence analysis (Table S1). Fragment analysis. Genomic DNA from G1 parents and G2 offspring was PCR amplified with 6-FAM 5′ -labeled reverse primer and unlabeled forward primer against OVM (Table S1). PCR conditions were as follows: