One-step generation of triple gene-targeted pigs using CRISPR/Cas9 system

Pig shows multiple superior characteristics in anatomy, physiology, and genome that have made this species to be more suitable models for human diseases, especially for neurodegenerative diseases, because they have similar cerebral convolutions compared with human neocortex. Recently, CRISPR/Cas9 system shows enormous potential for engineering the pig genome. In this study, we expect to generate human Parkinson’s disease pig model using CRISPR/Cas9 system by simultaneously targeting three distinct genomic loci, parkin/DJ-1/PINK1, in Bama miniature pigs. By co-injection of Cas9 mRNA and multiplexing single guide RNAs (sgRNAs) targeting parkin, DJ-1, and PINK1 genes, respectively, into in vivo derived pronuclear embryos, we simultaneously targeted three distinct genomic loci. The gene modified piglets remain healthy and display normal behavior at the age of 10 months. In addition, despite the high number of sgRNAs were employed in the present study, our trio-based whole-genome sequencing analysis suggested that the incidence of off-target events is low. Our results demonstrate that the simplicity, efficiency, and power of the CRISPR/Cas9 system to allow for the modification of multiple genes in pigs and yield results of high medical value.

Parkinson's disease (PD) is the most common neurodegenerative movement disorder in the elderly. While most cases are sporadic, about 5-10% of people are now known to have forms of the disease that occur because of a mutation of one of several specific genes 1 . Recessively inherited loss-of-function mutations in the DJ-1, parkin, and PINK1 genes are linked to familial cases of early-onset PD [2][3][4] . However, neither single nor triple knockout mice lacking DJ-1, parkin, and PINK1 genes can recapitulate the clinical symptoms of idiopathic or inherited PD 5,6 . The lack of a proper model has hindered our ability to develop therapies against human PD. Recently, numerous pig models of neurodegenerative disorders, such as amyotrophic lateral sclerosis (ALS) 7 and Huntington's disease (HD) 8 , have been developed using gene engineering approaches because pigs share many physiological similarities with humans. To understand the pathophysiology of PD, and to develop novel therapies for improved symptomatic management, we were therefore prompted to generate DJ-1, parkin, and PINK1 genes modified pigs to attempt to obtain the relevant disease models of PD.
Recently, through the use of the CRISPR/Cas9 system, biallelic gene knockout pigs were efficiently generated in a single step through a direct cytoplasmic injection of Cas9 mRNA and sgRNA into pig zygotes 9,10 . These findings indicated that the CRISPR/Cas9 system shows enormous potential in establishing large animal models of neurodegenerative diseases by engineering genome in spite of the lack of embryonic stem cell lines for genomic manipulation 11 . Nevertheless, the versatile functionalities of the CRISPR/Cas9 system, such as multiplexed genome editing, remain to be developed and improved in pigs. This requirement is particularly important when simultaneous modification of multiple genes functioning in concert is needed to obtain a desired phenotype. For example, inactivation of all three recessive PD genes, DJ-1, parkin, and PINK1, may have synergistic effects and cause more severe PD pathology than single gene inactivation, which may explain why patients with

Results
In the present study, we produced pigs with mutations in multiple targeted genes. By co-injection of Cas9 mRNA and multiplexing sgRNAs into in vivo derived pronuclear embryos, we simultaneously targeted three distinct genomic loci in Bama miniature pigs (Fig. 1A).
The simultaneous use of two adjacent sgRNAs targeting one locus significantly increased the targeting efficiency and improved Cas9-mediated genome targeting 13,14 . Hence, we designed six sgRNAs targeting six different genomic sites encoding pig DJ-1 (sgDJ1-1 and sgDJ1-2), parkin (sgparkin-1 and sgparkin-2) and PINK1 (sgPINK1-1 and sgPINK1-2) (Fig. 1B). The sgRNAs were designed to contain 20-nucleotide customized spacer sequences together with tracrRNA-derived sequences as previously described 15 . A mixture of sgDJ1-1, sgDJ1-2, sgparkin-1, sgparkin-2, sgPINK1-1 and sgPINK1-2 (six sgRNAs) with Cas9 mRNA was injected into one-cell-stage pig embryos. Next, we transferred the injected embryos into surrogate pigs to produce piglets. A total of 34 embryos were delivered to 3 surrogates, and one of surrogates was successfully impregnated and delivered two live-born piglets (Supplementary information, Table S1). We amplified target sites from the genomic DNA of each individual piglet and analyzed the genotype by Sanger sequencing. Surprisingly, both of the founders harbored site-specific indels in the targeting sites. One piglet showed biallelic modification of all three genes, and another showed biallelic modification of the DJ-1 and PINK1 genes and monoallelic mutations of parkin gene (Fig. 1C). These results demonstrated that piglets carrying biallelic mutations in three different genes could be generated in one step through direct zygote injection with high efficiency.
We further examined whether gene modifications caused a depletion of gene expression or related phenotypes. Immunofluorescence ( Fig. 2A) and western blot analysis (Fig. 2B) showed the absence of DJ-1 protein expression in the fibroblasts of these two gene modified piglets, which confirmed that both of the DJ-1 gene alleles were inactivated. Due to the difficulty of obtaining an effective antibody to determine the expression of parkin and PINK1 in pigs, real-time RT-PCR was employed. The relative parkin and PINK1 gene mRNA expression was also dramatically decreased in the fibroblast cells from the two gene modified pig compared with wild-type cells based on real-time RT-PCR ( Fig. 2C), most likely due to nonsense-mediated mRNA decay 16 . Furthermore, because oxidative stress and mitochondrial dysfunction have been linked to the pathogenesis of PD 17 , we also showed that the transcription of some stress defense genes was disrupted in the gene modified pig fibroblast cells (Fig. 2D). However, the two piglets remain healthy with a normal growth rate for Bama miniature pigs and typical symptoms of Parkinson's disease have not been observed in 10-month-old live mutant pigs in this study.
Previous studies have suggested that the CRISPR/Cas9 systems might tolerate sequence mismatches distal from the protospacer adjacent motif (PAM) at the 5′ end of sgRNAs, which would likely induce off-target mutations 18 . Specifically, a high number of sgRNAs were employed in the present study. We thus sought to test the possible off-target effects in genome-modified pigs derived from multiplexing sgRNAs-injected zygote. We performed whole-genome sequencing in the two parents and one of the piglets (Fig. 3A, Supplementary information, Figure S1) to assess the degree of mutagenesis across the entire genome. We largely focused on the identification of small indels introduced by non-homologous end joining (NHEJ)-mediated repair. After the filtering process (methods section), a set of small indels that were most likely to be true positives and potential off-target mutations (rather than the mutations that arose in the parental genomes) were detected by whole-genome sequencing. Consequently, we identified a total of 19 de novo indels in the offspring throughout the genome ( Table 1). The de novo mutation frequency (6.77 × 10 −9 ) is closed to that previously estimated in human (1.5 × 10 −9 ) 19 , which indicates that our CRISPR/Cas9 system does not significantly increase the rate of de novo indel mutations. Further, the low de novo mutation frequency may indirectly reflect the low incidence of off-target events. Additionally, these mutations were all located in the non-coding genomic regions (Table 1), which suggests no obvious effect on the function of the protein. We further asked whether these de novo indels were generated spontaneously or induced by the CRISPR/Cas9 system. To this end, we performed a comprehensive prediction for potential off-target sites by a CasOT program 20 , and we found that no indel mutations located around the predicted/potential off-target regions (Fig. 3B). These results suggested that these mutations were most likely to occur spontaneously. Moreover, the potential off-target sites, which were located within the coding regions, were well covered by sequencing reads (also confirmed by Sanger sequencing, Supplementary information, Figure S2), and no de novo mutations were identified in these off-target sites and flanking regions (Supplementary information, Table S2). In addition, the whole-genome sequencing analysis demonstrated that the present CRISPR/Cas9 targeting did not lead to mosaic mutations (data not shown). We concluded that the CRISPR/Cas9 system may not significantly increase the incidence of off-target cleavages in the pig genome, though a high number of sgRNAs was employed in the present study. Our study is consistent with the recent reports showing off-target mutations are rare in Cas9-modified mice 21 .

Discussion
The genetic manipulation of pigs is a crucial approach for improving livestock production and the study of human disease. However, the generation of gene modified pigs has been impeded by the slow, tedious, and expensive procedure of genetic modification due to the lack of embryonic stem cell lines. The most exciting finding in this study is the demonstration of robust multiplex, biallelic gene targeting in the pig without induction of off-target mutattions. A unique capability of the CRISPR/Cas9 system is multiplex genome engineering by delivering a single Cas9 enzyme and two or more single guide RNAs (sgRNAs) targeted to distinct genomic sites. Through the co-injection of Cas9 mRNA and the multiplexing of sgRNAs into in vivo derived pronuclear embryos, we simultaneously targeted three distinct genomic loci in Bama miniature pigs in a highly efficient manner. Several previous studies have reported that simultaneous use of dual sgRNAs to target an individual gene significantly improved the CRISPR/Cas9-mediated genome targeting efficiency 13,14,22 . In line with this, we found that dual gRNAs could indeed simultaneously target three distinct genomic loci in Bama miniature pigs in a highly efficient manner. This could be because different sgRNAs have substantially variable DNA cleavage efficiency, and dual gRNAs could guarantee the CRISPR system has maximize occurrence of genetic modification at the target site. Actually, in our present study, the allelic mutations at the parkin target site from piglet 1# were induced by different sgRNAs. Recently, dual sgRNAs were used to induce a deletion of the chromosome sequence  between the two sgRNA-targeted loci, which is a very useful technique for creating null alleles and targeting noncoding regions in the genome such as enhancers and silencers that control expression of disease-relevant genes 22 . However, Wang et al., reported that combination of some two sgRNAs may exhibit an antagonistic effect in targeting. Such antagonistic effect reminded us that when multiple gRNAs were used in combination, the side effect should be taken into consideration 23 .
Although CRISPR/Cas9 systems can serve as the basis of a simple and highly efficient method for performing genome editing in various organisms, they do not have perfect specificity and can induce high-frequency off-target mutagenesis at sites in the genome other than the desired on-target site 18 . The major concern about the present study is the potential off-target effect caused by high number of sgRNAs. Unlike previously reports, in which limited numbers of potential target DNA sequences with point or combined mismatches in silico predicted by comparison with the authentic targeting sites were tested, we performed whole-genome sequencing in the parents and one of the piglets (1#) to assess the degree of mutagenesis across the entire genome. Surprisingly, sequencing assay showed that our CRISPR/Cas9 system did not introduce significant off-target cleavages in the pig genome, though a high number of sgRNAs were employed in the present study. In additional, we also did not find any mosaic mutations in the piglets by whole-genome sequencing. It suggested that the choice of sgRNAs that represent unique sequences in the genome and the use of optimization of the injection method would suffice to avoid off-target mutations and reduce the mosaic rate and thus cut the cost and time required for the production of mutant animals.
We expect that the newly developed CRISPR/Cas9 technology will promote the generation of human diseases pig models and pigs with tailored economic traits, and yield results of high medical and agricultural value.

Material and Methods
Chemical and Reagents. Unless 1  137135750  GAGTCAAGCCTATGAGTATA  A  AAT  AAAGTAGTGACCAAATATTA   1  192684122  AGACAAAGGTGGTCTTTTGA  T  TTGAAGACACATCA  TGAAGACAGATCAAGGTAGC   10  12607948  GAATCTAGATTTACACTCAA  A  ACAGTAC  CAGAATAGATTTATGATAGC   11  T  A  C  C  A  T  A  T  A  T  T  C  C  A  A  A  A  G  A  A  G  T  A  A  A  T  A  T  T  T  A  A  A  A  T  C  A  A  C  G  T  A  T  T  G  0  5  2  1  4  3  0  6   13  7678576  ATCCAAATTATAGAAATCAT  AC  A  CAACTATACATCAAAAAAAC   13  7678587  AGAAATCATACAACTATACA  A  A  T  C  A  A  T  C  G  A  A  C  A  A  A  A  A  A  A  C  C  T  T   13  103844162  ACATGTATCTTTTCGAATTA  G  GT  TTTTTTTTTTCCCCAGAATA   13  160142098  TTTTTATGACTGTAGTTTTG  TAGTAG  T  AGTAGAGTCTGAAGTCAGGG   14  107749627  GACTTTGTTGAACACCTGAA  ACTAACACTGTTTC  A  CTAACACTGTTTCCTAAGTC 15 8  18066061  CTTTGTTGAAACCACTATTT  ATTTTCTTTTC  A  TTTTCTTTTCTTTTCTTTTC   8  T  T  A  C  T  A  C  T  T  G  T  A  C  T  T  G  A  T  T  T  T  A  A  C  C  A  C  A  T  A  T  A  T  T  A  T  A  A  G  G  A  T  C  7  5  0  6  9  2  9  6   9  38150005  GGATATCAATAAATAGAGTT  AG  A  GGAAAAAAATACCCAGGAAG   9  90939919  CTAGTTATCACCA  T  A  T  T  T  G  A  T  G  T  G  A  T  T  A  T  T  T  T  T  A  T  T  A  T  A  T  A  T  A   9 92437048 TATGTACATATACACACACA CAT C ATATATATATTTTCTTTCTA Table 1. Unique de novo indels detected by whole-genome sequencing. No indels lie in the coding sequences. Indels that lie in the introns are highlighted in Bold. Other indels lie in the intergenic region. Red highlights indicate potential PAMs (NGG) within 20 bases upstream of the indel.