The Creation of a Multiallele Knockout Genotype in Rabbit Using CRISPR/Cas9 and Its Application in Translational Medicine

Nonrodent animal models have recently become more valuable in preclinical studies. The limitation of nonrodent animal models is that they must demonstrate relatively reliable and predictable responses in addition to representing complex etiologies of a genetically diverse patient population. In our study, we applied CRISPR/Cas9 technology to produce transgenic rabbits. This approach can be useful for creating genetically divergent and homogeneous populations for studies in translational medicine. NADPH oxidase 4 (NOX4) is a promising therapeutic target, as it is linked to several pathologies including stroke, atherosclerosis, and lung and kidney fibrosis. NOX4 knockout (KO) rabbit lines were created in order to study the in vivo effects resulting from a lack of NOX4 protein and loss of gene function. One of the knockout founders was a germline multiallelic knockout male. Its offspring segregated into three distinct NOX4 knockout and a wild-type lines. Mosaicism is a relatively frequent phenomenon in rabbit transgenesis. Our results point to the possible application of mosaicism in preclinical studies. However, careful planning and evaluation of results are necessary. The predicted off-target sites were studied as well, and no signs of off-target events were detected.


Introduction
Since the creation of the first genetic knockout rabbits with the CRISPR/Cas9 method in 2014 [1], the number of publications on CRISPR/Cas9-modified rabbits has increased, touching on a wide array of topics such as improving meat quality [2], altering fur color [3], studying the metabolic features [4,5], modeling human diseases [6], producing recombinant proteins [7], and improving the methodology [8][9][10]. Compared to rodents, rabbits are phylogenetically more closely related to humans, and many physiological processes, such as lipid and carbohydrate metabolism [11,12], disease progression in the case of atherosclerosis [13,14], cardiac diseases [15], and obesity [16,17], are more similar to those in humans. In addition, because of their similar anatomy and bigger size, in physiology and metabolism, as in the case of dog models [49]. Thus, inbred animals (and even nuclear-cloned animals) are the most controlled model systems, and their use can bypass several challenges associated with human studies [50]. On the other hand, biology is characterized by diversity as well as unity [51]. To avoid the questionable validity of an animal model system, genetic diversity should be assured. The dual problem of genetic diversity in animal model systems can be solved using mosaicism in transgenic founders of nonrodent species. In this paper, we show that a transgenic rabbit founder animal can produce multiallelic germline knockout strains that can be maintained as a closely related animal population and with high genetic diversity. Both are necessary for successful translational medical studies.

Ethics Statement
The rabbits used in this study were New Zealand white rabbits (Innovo Ltd., Isaszeg, Hungary). All animal studies were conducted in strict accordance with the recommendations and rules in the Hungarian Code of Practice for the Care and Use of Animals for Scientific Purposes, approved by the Animal Welfare and Research Ethics Committee at Agricultural Biotechnology Institute, National Agricultural Research and Innovation Centre (NARIC-ABC) and registered under permission number ÁK-ENG (PE/EA/58-2-2018) from Pest County's governmental office. All efforts were made to minimize the suffering of animals.

Animals
Laboratory rabbits were maintained in the Animal Care Facility at NARIC-ABC. Animals were housed in individual cages with free access to food and water. Animals were kept under a standard light-dark cycle (06.00-18.00 h) at 19 • C.

Microinjection and Embryo Transfer
The protocol for superovulation of embryo donor does, embryo collection, microinjection of pronuclear-stage embryos, and transfer to recipient females has been described in our previous paper [52].
In summary, sexually mature does received an intramuscular injection of pregnant mare serum gonadotropin (PMSG) (ProSpec, Rehovot Israel), followed by an intravenous injection of human choriogonadotropin (hCG) (Choragon, Ferring GmbH, Kiel, Germany) 72 h later in order to induce superovulation. Donor rabbits were mated after the hCG injection. The oviducts were flushed with PBS-20%FCS (vol/vol) medium for collection of pronuclear-stage embryos. Zygotes were incubated in a Nuaire thermostat (38.5 • C, 5% CO 2 , 80% humidity) in the same embryo culture medium. A mixture of synthetic gRNA and transactivating RNA (Sigma-Merck, Budapest, Hungary) was added to capped Cas9 RNA (150 ng/µL) (Trilink, San Diego, CA, USA) to a final concentration of 5 ng/µL in an embryo microinjection buffer (1 mM Tris, pH 7.5, 0.1 mM EDTA). Cas9 mRNA is a highly purified capped form by CleanCap ® technology preventing 5 degradation and with increased half-life of mRNA. An aliquot of each batch used for microinjection was checked for degradation. Each batch was kept at 37 • C for 4 h previous to microinjection followed by agarose electrophoresis. Without any visible sign of degradation, RNA solution was ready for microinjection and the solution was injected into the cytoplasm of pronuclear-stage embryos. The injected embryos were incubated in embryo culture medium for 2-3 h, followed by the transfer of approximately 30-40 injected embryos into the oviducts of each recipient mother via laparoscopic surgery [52].
Guide RNA was designed using the publicly available E-CRISP program (http://www.e-crisp.org/ E-CRISP/designcrispr.html) according to the second and fourth exon of the NOX4 gene (chromosome 1: 127,338,844-127,541,489, ENSOCUG00000016685.4). Guide RNAs are shown in Figure 1.

DNA Isolation from Blastocysts and Newborn Pups
Injected zygotes were individually collected at the blastocyst stage with a minimal amount of culture media in a 200 μL standard PCR microtube (Merck, Budapest, Hungary) and air-dried. Then, 5 μL of water was added and incubated at 95 °C for 5 min, followed by a 10 min incubation at −70 °C. This process was repeated twice. Embryos were then incubated in a 200 μL final volume of lysis buffer (10 mM Tris pH 8.0, 100 mM NaCl, 10 mM EDTA pH 8.0, 0.5% SDS, 0.4 mg/mL proteinase K) at 37 °C for 2 h. DNA was isolated using the phenol-chloroform method. Equal volumes of phenol and chloroform were added to the mixture, shaken thoroughly, and centrifuged at 13,000 rpm for 10 min. The supernatant was mixed with a double volume of isopropanol, kept at −70 °C overnight, and centrifuged for 30 min at 4 °C. The resulting invisible pellet was then washed twice with 70% ethanol, air-dried, and dissolved in 10 μL water. Genomic DNA from tissue samples was isolated using the conventional phenol-chloroform method [53].

Detection of Mutations in Embryos and Pups by PCR and Sequencing
The PCR primers used for mutation detection and Sanger sequencing are listed in Table 1. PCR reactions were set up with RedTaq Ready mix (Sigma, Budapest, Hungary) with standard PCR conditions (96 °C for 5 min; 33 cycles of 95 °C for 15 s, 58 °C for 15 s, 72 °C for 20 s; and a final step of 72 °C for 5 min followed by holding at 4 °C) in a ProFlex PCR system thermal cycler (Applied Biosystems, Budapest, Hungary). PCR products were gel purified with NucleoSpin Gel and PCR Clean up Columns for gel extraction and a PCR cleanup kit (Macherey Nagel, Dueren, Germany) and cloned into pGEM-T vector (pGEM-T Easy Vector System I, Bioscience, Budapest, Hungary). Then, 4-5 PCR-positive plasmid clones were sequenced. Chromas software (Technelysium Pty Ltd., South Brisbane, Australia) was used for sequence analysis.

T7 Endonuclease I Assay
A T7 endonuclease I assay was performed to detect mutations in the target sequence [54]. After the extraction of genomic DNA from blastocysts and pups, the genomic region of NOX4 was PCR amplified (primers are shown in Table 1). The PCR product was denatured and reannealed under the These guide RNAs targeting the second and fourth exons of the rabbit NADPH oxidase 4 gene for the CRISPR/Cas9 system. The second exon gRNA with PAM sequence is UGGUGGAGGUAAUGGUACUC-TGG. The fourth exon gRNA with PAM sequence is UAGAGCACCACAAGUUAUA-TGG.

DNA Isolation from Blastocysts and Newborn Pups
Injected zygotes were individually collected at the blastocyst stage with a minimal amount of culture media in a 200 µL standard PCR microtube (Merck, Budapest, Hungary) and air-dried. Then, 5 µL of water was added and incubated at 95 • C for 5 min, followed by a 10 min incubation at −70 • C. This process was repeated twice. Embryos were then incubated in a 200 µL final volume of lysis buffer (10 mM Tris pH 8.0, 100 mM NaCl, 10 mM EDTA pH 8.0, 0.5% SDS, 0.4 mg/mL proteinase K) at 37 • C for 2 h. DNA was isolated using the phenol-chloroform method. Equal volumes of phenol and chloroform were added to the mixture, shaken thoroughly, and centrifuged at 13,000 rpm for 10 min. The supernatant was mixed with a double volume of isopropanol, kept at −70 • C overnight, and centrifuged for 30 min at 4 • C. The resulting invisible pellet was then washed twice with 70% ethanol, air-dried, and dissolved in 10 µL water. Genomic DNA from tissue samples was isolated using the conventional phenol-chloroform method [53].

Detection of Mutations in Embryos and Pups by PCR and Sequencing
The PCR primers used for mutation detection and Sanger sequencing are listed in Table 1. PCR reactions were set up with RedTaq Ready mix (Sigma, Budapest, Hungary) with standard PCR conditions (96 • C for 5 min; 33 cycles of 95 • C for 15 s, 58 • C for 15 s, 72 • C for 20 s; and a final step of 72 • C for 5 min followed by holding at 4 • C) in a ProFlex PCR system thermal cycler (Applied Biosystems, Budapest, Hungary). PCR products were gel purified with NucleoSpin Gel and PCR Clean up Columns for gel extraction and a PCR cleanup kit (Macherey Nagel, Dueren, Germany) and cloned into pGEM-T vector (pGEM-T Easy Vector System I, Bioscience, Budapest, Hungary). Then, 4-5 PCR-positive plasmid clones were sequenced. Chromas software (Technelysium Pty Ltd., South Brisbane, Australia) was used for sequence analysis.

T7 Endonuclease I Assay
A T7 endonuclease I assay was performed to detect mutations in the target sequence [54]. After the extraction of genomic DNA from blastocysts and pups, the genomic region of NOX4 was PCR amplified (primers are shown in Table 1). The PCR product was denatured and reannealed under the following conditions: 95 • C for 5 min, 95-85 • C at −2 • C/s, 85-25 • C at −0.1 • C/s, holding at 4 • C. The reannealed fragments were digested with T7EI (NEB, Ipswich, MA, USA), separated, and visualized on an ethidium bromide-stained 2% agarose gel.

Off-Target Assay
Potential off-target sites of the sgRNAs were predicted using the CRISPR online design tool CRISPR/Cas-OFFinder RGEN (http://www.rgenome.net/cas-offinder/). The eight most promising off-target sites for NOX4 second exon sgRNAs were chosen for PCR and sequencing using REDTaq ® ReadyMix (Sigma Aldrich, Budapest, Hungary) with the corresponding primers ( Table 2). PCR products were analyzed on a 2% agarose gel, then purified and directly Sanger sequenced (Eurofins Genomics, Ebersberg, Germany). Chromas software and online blast tool (https://www.ensembl.org/Oryctolagus_ cuniculus/Tools/Blast) was used for sequence analysis and evaluation of results. Table 2.
The eight most potent off-target sites in the rabbit genome predicted by the CRISPR/Cas-OFFinder RGEN tool. Lowercase letters indicate the sites of mismatches compared to the target sequence.

Off-Targets
Off- Target

Testing of Guides and Production of Founder Knockout Animals
The mutation efficiency of guide RNAs was tested in cultured blastocysts. Both guides were effective in inducing mutations in the rabbit NOX4 gene. However, the guide RNA targeting the second exon of NOX4 had slightly higher mutation efficiency. Table 3 shows the ratio of total injected zygotes that developed to blastocyst stage to mutant blastocysts for the two guides designed for the second and fourth exons (18% and 16%, respectively). Table 3. Results of microinjection of Cas9 and targeting gRNAs into rabbit zygotes. The second and fourth exons were targeted, and both guides were successful in inducing targeting events with almost equal efficiency. In addition, it is beneficial to introduce an early stop codon in the second exon in order to avoid the presence of a functional truncated peptide. Hence, we chose the RNA guide targeting the second exon of the NOX4 gene to produce live animals.
Based on the results of embryo testing, a total of five knockout founder animals were produced, two males and three females, as summarized in Table 4. All mutations were heterozygous 7 bp and 8 bp deletions directly upstream of the PAM site with a couple of base differences: Animals #2, #3 and #4 acquired identical mutations (Figure 2), while animal #5 carried three different mutated alleles. We established heterozygous and homozygous genome-edited rabbit sublines by consecutive brother-sister mating from founder animal #5. In addition, it is beneficial to introduce an early stop codon in the second exon in order to avoid the presence of a functional truncated peptide. Hence, we chose the RNA guide targeting the second exon of the NOX4 gene to produce live animals.
Based on the results of embryo testing, a total of five knockout founder animals were produced, two males and three females, as summarized in Table 4. All mutations were heterozygous 7 bp and 8 bp deletions directly upstream of the PAM site with a couple of base differences: Animals #2, #3 and #4 acquired identical mutations (Figure 2), while animal #5 carried three different mutated alleles. We established heterozygous and homozygous genome-edited rabbit sublines by consecutive brother-sister mating from founder animal #5.  Animal #5 passed on four different alleles to the F1 generation (Figures 3 and 4). The most frequent allele was an 8 bp deletion (∆8). The other allele contained a 128 bp deletion (∆128), including an intronic region. The third allele type was the most irregular, with deletions and insertions adding up to a 104 bp deletion in total (∆104) upstream of the PAM sequence. A wild-type allele also Animal #5 passed on four different alleles to the F1 generation (Figures 3 and 4). The most frequent allele was an 8 bp deletion (∆8). The other allele contained a 128 bp deletion (∆128), including an intronic region. The third allele type was the most irregular, with deletions and insertions adding up to a 104 bp deletion in total (∆104) upstream of the PAM sequence. A wild-type allele also appeared in the F1 generation, demonstrating the presence of a fourth allele. We analyzed the expression level of NOX4 in different tissues of a selected homozygous individual from line ∆8 by RT-qPCR. Expression pattern matched to an expected -/-phenotype (data not shown).
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 15 appeared in the F1 generation, demonstrating the presence of a fourth allele. We analyzed the expression level of NOX4 in different tissues of a selected homozygous individual from line ∆8 by RT-qPCR. Expression pattern matched to an expected -/-phenotype (data not shown).

Off-Target Analysis
The selected potential off-target sites were amplified by PCR using genomic DNA from both the wild-type animal and founder rabbit #5 and evaluated by DNA sequencing. Eight candidate offtarget sites were analyzed. In all these cases, no difference was observed (see Figure 5), so no offtarget event was detected at any of the genomic locations. The second exon is underlined with a black line. The ∆8 mutation is a small deletion of eight base pairs, indicated in yellow. In ∆104, a TAAT insertion is present, highlighted in blue, and a couple of single base pair changes appeared in the sequence of ∆104 and ∆8, which, in the case of the ∆104 mutated allele, was probably the consequence of imperfect NHEJ repair. However, in the ∆8 mutated allele, it might be a small nucleotide polymorphism not related to NHEJ.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 15 appeared in the F1 generation, demonstrating the presence of a fourth allele. We analyzed the expression level of NOX4 in different tissues of a selected homozygous individual from line ∆8 by RT-qPCR. Expression pattern matched to an expected -/-phenotype (data not shown).

Off-Target Analysis
The selected potential off-target sites were amplified by PCR using genomic DNA from both the wild-type animal and founder rabbit #5 and evaluated by DNA sequencing. Eight candidate offtarget sites were analyzed. In all these cases, no difference was observed (see Figure 5), so no offtarget event was detected at any of the genomic locations.

Off-Target Analysis
The selected potential off-target sites were amplified by PCR using genomic DNA from both the wild-type animal and founder rabbit #5 and evaluated by DNA sequencing. Eight candidate off-target sites were analyzed. In all these cases, no difference was observed (see Figure 5), so no off-target event was detected at any of the genomic locations.

Heritability of Deletions
Founder animal #5 was used as a sperm donor in four different artificial inseminations. Female partner animals originated from two different available sub-breeds of New Zealand White rabbits to represent different genotypes. Later, the same females were used in two more consecutive matings to represent animal populations with close relationships (both parents were the same). Animal #5, the founder male, had a total of four litters with an average litter size of 3-10 pups/litter. The three mutants and the wild-type allele appeared with varying frequencies among the pups in the F1 generation. The most frequent allele was ∆8 with 50%, followed by ∆128 and ∆104 with 36% and 10%, respectively. The wild-type allele appeared in only one female (4%). There was an alteration from Mendelian inheritance according to the sex of the animals. Two-thirds of the population was female, so male offspring was underrepresented by 14% compared to the total number of animals. This difference might be due to the small sample size (28 pups in total). The female/male ratio showed the same pattern among all three mutated genotypes (70:30, 64:36, and 67:33 in ∆128, ∆8, and ∆104, respectively). The data are shown in Table 5. In the F2 generation, when a heterozygous male and female were bred, homozygous NOX4 KO mutants of both sexes were born. We initialized the analyses of phenotype, such as litter size and preweaning lethality, and have not found any difference compared to wild-type litters. We have proof of homozygous animal survival and of the fertility of both sexes. More data with a higher sample size are needed to confirm or disprove the differences in sex ratio in NOX4 knockout rabbits. The establishment of homozygous knockout population is in progress.

Discussion
The results of our study show the successful generation of several NADPH oxidase 4 knockout rabbit lines. The NOX4 gene plays an important role in biological processes, and reactive oxygen species have a modulatory effect on the physiological functions of diverse mammal organs. NOX4 knockout mouse models have recently been demonstrated to be useful models to study the regulation of insulin secretion [55]. Other research has shown the role of NOX4 in liver fibrosis in connection with gut microbiota [56]. In addition, the protective effect of NOX4 knockdown has also been demonstrated in sepsis-induced lung injury in mice [57]. The broad spectrum of NOX4 function creates the need for better animal models to make the best mammalian system for the purposes of translational medicine.
The efficiency of CRISPR/Cas9 to induce site specific mutations in rabbits is a complex issue. Chromatin structure of the targeted genomic region affects the availability of the site to CRIPSR/Cas9 complexes. An open chromatin structure predicts greater efficiency. The concentration of gRNA is also a crucial element. The encountered 16-18% efficiency of mutated blastocysts is within the normal range. However, the efficiency of targeting varies greatly between 10-100% according to scientific data [1].
Five founder animals were generated in total, of which three acquired identical 8 bp deletions. This phenomenon seems striking, but robust analysis of CRISPR-mediated genome editing in cell culture systems has shown that DNA editing varies considerably between sites, with some targets shown to have a highly preferred indel while others display a wide range of infrequent indels [58]. Presumably, the pattern is dependent on both DNA sequence and chromatin structure.
Interestingly, one of the NOX4 KO founder males turned out to be a genetic mosaic animal with germline transmission of three distinct knockout alleles besides the wild-type allele to its offspring. The ∆8 subline of founder animal #5 carried a short 8 bp deletion directly upstream of the PAM site, which is a common type of modification induced by the CRISPR/Cas9 method according to our own observations and the literature [59]. The other two modified alleles carried longer deletions, a 128 bp deletion in the case of the ∆128 subline, and a more exciting deletion-insertion-rearrangement type of modification in the case of the ∆104 rabbit subline. Complex targeting events in genome-edited animals have previously been detected [60]. Both modifications-∆104 and ∆128-make the detection of the knockout allele a simple one-step procedure, with allele-specific PCR sparing us the burden of T7 assays in our future experiments.
We found a deviation of the male/female ratio from the expected Mendelian pattern of inheritance: 14% fewer males were born in the F1 litters of the #5 founder male. These data must be closely examined in our future studies and while evaluating F2-generation data.
Mosaic events during genome editing are generally unwanted events. There are known strategies to reduce the genetic mosaicism produced by CRISPR/Cas9 system [61]. Mosaic events can be found during rabbit genome editing using CRISPR/Cas9 methods after allele screening [2,8,[62][63][64][65][66]. Our opinion is that mosaic events can be useful in animal models of human diseases. To profit from the fact that mosaic animals represent different edited alleles, we developed a new breeding strategy ( Figure 6).
The mosaic animal is mated consecutively with its counterpart, resulting in litters whose newborns are closely related but represent different transgenic lines with different targeting events. We suppose that analyzing these lines would result in predictable and equal phenotypic differences, ensuring the observations are the outcome of genome editing but not off-target events. On the other hand, mosaic founders are mated with different genetic background partners, which will result in a population of newborns with a diverse genetic heritage. In this case, small differences during phenotyping are tolerated and can presumably be explained. Most of human diseases are multifactorial and multigenic. Even single inheritance disorders can show diverse appearance depending on genetic background. To create the best animal model of the analyzed disorder, it is necessary to keep genetic divergence within the model population. On the other hand, genetic diversity can cause uninterpretable deviations sometimes. This is one important reason why most studies use inbred mice as model animals. Inbred strains usually react in a predictable way. Our breeding strategy involved combining the advantages of inbred and diverse genetic populations. Although this strategy uses more animals and the phenotypic examination is more complex, the resulted biological answer closely resembles the progression of human diseases. Using our new mating system, genetically divergent and homogeneous genome-edited populations can be produced in parallel, which is beneficial in translational medicine studies. The mosaic animal is mated consecutively with its counterpart, resulting in litters whose newborns are closely related but represent different transgenic lines with different targeting events. We suppose that analyzing these lines would result in predictable and equal phenotypic differences, ensuring the observations are the outcome of genome editing but not off-target events. On the other hand, mosaic founders are mated with different genetic background partners, which will result in a population of newborns with a diverse genetic heritage. In this case, small differences during phenotyping are tolerated and can presumably be explained. Most of human diseases are multifactorial and multigenic. Even single inheritance disorders can show diverse appearance depending on genetic background. To create the best animal model of the analyzed disorder, it is necessary to keep genetic divergence within the model population. On the other hand, genetic diversity can cause uninterpretable deviations sometimes. This is one important reason why most studies use inbred mice as model animals. Inbred strains usually react in a predictable way. Our breeding strategy involved combining the advantages of inbred and diverse genetic populations. Although this strategy uses more animals and the phenotypic examination is more complex, the resulted biological answer closely resembles the progression of human diseases. Using our new mating system, genetically divergent and homogeneous genome-edited populations can be produced in parallel, which is beneficial in translational medicine studies.