One-Step Homology Mediated CRISPR-Cas Editing in Zygotes for Generating Genome Edited Cattle

Selective breeding and genetic modification have been the cornerstone of animal agriculture. However, the current strategy of breeding animals over multiple generations to introgress novel alleles is not practical in addressing global challenges such as climate change, pandemics, and the predicted need to feed a population of 9 billion by 2050. Consequently, genome editing in zygotes to allow for seamless introgression of novel alleles is required, especially in cattle with long generation intervals. We report for the first time the use of CRISPR-Cas genome editors to introduce novel PRNP allelic variants that have been shown to provide resilience towards human prion pandemics. From one round of embryo injections, we have established six pregnancies and birth of seven edited offspring, with two founders showing >90% targeted homology-directed repair modifications. This study lays out the framework for in vitro optimization, unbiased deep-sequencing to identify editing outcomes, and generation of high frequency homology-directed repair–edited calves.


Introduction
Genetic modification of livestock has a long and storied history, beginning with the domestication of animals. Initially, animals exhibiting traits favorable for domestication and performance were chosen for breeding (mass selection). Then with time, the process became more sophisticated and included pedigree selection, markerassisted selection, and with the sequencing of genomes, genome-wide selection. Transgenics marked the first major milestone that allowed for introgression of novel alleles and traits into livestock animals. 1 This in combination with somatic cell nuclear transfer (SCNT) 2,3 became the main staple for generating genetically modified livestock. The recent discovery and successful validation of genome editors (ZFNs, TALENs, and CRISPR-Cas9) will now allow for genetic modification directly in the zygote bypassing the need for SCNT. [4][5][6][7][8][9][10] Among the genome editors, CRISPR-Cas9 has been widely employed for livestock genome editing. [8][9][10] The system uses a 17-20 nucleotide RNA sequence (''spacer'') as a guide along with a universal sequence (''tracr'') that permits loading into the Cas9 protein (single guide RNA [sgRNA]). [11][12][13] When a precomplexed Cas9 protein and sgRNA ribonucleoprotein (RNP) complex is delivered into the target cell or embryo, the Cas9 introduces double-stranded breaks at target DNA sites. 13 At a lower frequency, the double-stranded breaks can undergo homology-directed repair (HDR) in the presence of homologous repair template, or at a higher frequency by error-prone nonhomologous end-joining (NHEJ) pathways 14 Several recent manuscripts highlighted the feasibility of editing bovine genome in somatic cells, [15][16][17][18][19][20] and direct delivery of editors into zygotes 6,[21][22][23] However, no manuscript to date has reported the generation of live calves with novel variants by CRISPR-Cas-mediated HDR editing of bovine genome directly in zygotes. The main goal of this study was to establish a pipeline for (1) rational design and selection of targeting reagents, (2) high throughput screening approaches to evaluate editing outcomes, (3) optimized techniques for embryo injections to maximize HDR outcomes, and (4) ultimately, generation of genome-edited calves with HDR mediated introgression of novel variants at high efficiency. As a proof of concept, introduction of novel allelic PRNP (prion protein) variants has been chosen for validation of the optimization pipeline and for the generation of cattle resistant to prion diseases.
Misfolded cellular prion proteins (PrP) result in degenerative central nervous system disorders referred to as transmissible spongiform encephalopathies, 24 such as bovine spongiform encephalopathy (BSE) in cattle, scrapie in sheep, and Creutzfeldt-Jakob disease in humans. 24 Even though the ban of ruminant derived feed supplements helped reduce the incidence of BSE cases, the risk of atypical BSE resulting from spontaneous misfolding of endogenous PrP remains a concern. [25][26][27] To address this concern, PRNP null cattle have been generated via SCNT. 28 With the availability of CRISPRs, several groups are revisiting generating PRNP null cattle, although, most attempts have been limited to efforts in vitro. 17,18 Several natural variants of PRNP have been identified that provided protection against inherited prion diseases. 29,30 One particular PrP variant with glycine (G) at position 127 replaced with valine (V), referred to as G127V, was demonstrated to provide dominantnegative protection against the disease (in heterozygosity) in the ''Fore'' population of Papua New Guinea during the Kuru prion epidemic. 31 In an elegant study, mice null for endogenous Prnp and transgenic for human G127V variant conferred resistance against all 18 prion disease isolates. 31 This single amino acid substitution (G127V) was as protective as deletion or knockout of the protein. 31 Even though mouse has been used as a sur-rogate in this and other studies, [31][32][33][34][35][36] replicating these findings in cattle-a natural host to the disease-will be beneficial in both confirming these findings and generating disease-resistant elite cattle breeds. In this study, as a proof of concept, our aim was to engineer G127V allele in cattle using CRISPR-Cas genome editors.

Reagents
All chemicals were obtained from Sigma-Aldrich Company (St. Louis, MO) unless stated otherwise. All CRISPR-Cas reagents, targeting oligos, and PCR primers used in this study were purchased from IDT DNA Technologies (Coralville, IA) and are shown in Table 1.

Animal experimental assurance
All experiments involving live animals were performed in accordance with the approved guidelines of the University of Maryland and Thomas D. Morris Inc. institutional animal care and use committees. All experimental protocols involving live animals were approved by the institutional animal care and use committees at both Institutes (UMD protocol No. 1418824-1 and TDMI protocol No. 18-005: Genome editing in Cattle).
In vitro maturation, in vitro fertilization, and microinjection Cumulus oocyte complexes (COCs) from abattoir cattle were purchased from ART Inc. (Madison, WI) or DeSoto Biosciences Inc. (Seymour, TN) and shipped to the lab overnight in Maturation medium at 38.5°C. Twenty-two hours (h) after being placed in the maturation medium, in vitro fertilization (IVF) was performed in accordance to an established protocol using frozen semen. 37 Briefly, two straws of thawed semen were mixed with Dulbecco's phosphate buffered saline (DPBS) containing 1 mg/mL BSA to a final volume of 10 mL and centrifuged at 1000 g, 25°C for 4 minutes, followed by two washes of spermatozoa in DPBS. After the final wash, spermatozoa were co-incubated with matured COCs for 6 h in fourwell NUNC dishes (50 COCs/500 lL) in modified Brackett and Oliphant isotonic medium containing 3 mg/mL fatty-acid-free BSA and supplemented with PHE (20 lM D-penicillamine, 10 lM hypotaurine, and 1 lM epinephrine), 38 at a final concentration of 2 · 10 6 spermatozoa/mL at 38.5°C and 5% CO 2 . Six hours after IVF, the presumptive zygotes were vortexed in 0.1% hyaluronidase in HEPES-buffered medium containing 0.01% poly vinyl alcohol (PVA) for 4 minutes to remove the cumulus cells and extraneous spermatozoa. Presumptive zygotes were microinjected with a mixture of commercially sourced Cas9 RNP (Cas9 protein and sgRNA) and single-stranded DNA using a FemtoJet microinjector (Eppendorf, Germany). The microinjected embryos were cultured to blastocyst stage in serum-free medium (BO-IVC; IVF Bioscience, UK) for 8 days at 38.5°C, 5% CO 2 , 5% O 2 , and 100% humidity. Progression to cleavage and blastocyst stage of embryo development rate was recorded on days 2 and 8 post IVF respectively.
Optimization of CRISPR-Cas targeting reagents

Statistical analysis
Statistical analyses were performed with GraphPad Prism, version 8 software. Statistical comparisons of means were made using the unpaired two-tailed Student's t-test; P < 0.05 was considered statistically significant.

Data availability
All sequencing files were deposited on National Center for Biotechnology Information Sequence Read Archive under accession number PRJNA641429.

Results
Optimization of CRISPR-Cas gene targeting reagents An in vitro assay was used to identify optimal sgRNA and targeting single-stranded oligonucleotide (oligos). As shown in Figure 1A, two guide RNAs targeting the sense strand (F-guide) and the antisense strand (Rguide) near the target site were designed and tested.
The goal was to alter residues coding for glycine to valine (GGC to GTA) at the target site as highlighted in the yellow box (Fig. 1A). Successful targeting and sequence conversion will result in the generation of a new AccI restriction enzyme site (Fig. 1A). Commercially sourced sgRNA and Cas9 protein were precomplexed and injected along with a symmetrical sense targeting oligo into IVF zygotes ( Table 1). As shown in Figure 1B and C, the F-guide was not effective in introducing genetic modification at the site as revealed by restriction fragment length polymorphism (RFLP) analysis with AccI ( Fig. 1B and C). Therefore, all subsequent experiments were performed with R-guide. Following initial evidence of an active R-guide, two different 200 bp symmetrical oligos targeting the sense (99 bp TA 99 bp) and antisense (99 bp AT 99 bp) strand on the PRNP locus were similarly tested. In these trials with the R-guide, antisense oligos had better targeting efficiencies compared to the sense oligos (Fig. 1D).
A final variation of the reagents tested was a refinement of the antisense targeting oligo. Three antisense oligos with unique alignments were tested for gene targeting frequency. These include symmetric (identical length target site flanking sequences; 99 bp AT 99bp), asymmetric (longer 5¢ flanking sequence; 96 bp AT 29 bp), and reverse-symmetric antisense oligos (longer 3¢ flanking sequence 33 bp AT 92 bp) ( Table 1). To evaluate editing outcomes, targeted amplicon sequencing of 120 blastocysts (from multiple rounds of injection with the three targeting oligos) was performed using Illumina MiSeq platform (Fig. 1E). A minimum of 3000 reads were obtained per blastocyst (eight days after zygote injection) providing a >10 · coverage across the target site. The reads obtained were aligned to the reference sequence and collated as either HDR, NHEJ, unmodified/wildtype or ''other modifications events.'' Similar to preliminary experiments, all injected embryos showed a range of genetic modifications, from no modifications (unedited) to low frequency HDR with <10% of HDR reads (referred as <10% HDR) modifications, and a few with high frequency HDR reads (>50% HDR). For generating live HDR edited calves, a targeting oligo that yields one-third to one-half of the alleles with HDR modification (>33% or >50% HDR reads) is desirable. Among the three treatment groups, reverse-asymmetric oligo yielded better targeting efficiencies and HDR outcomes, whereas asymmetric oligo performed poorly (Fig. 1F-H). These in vitro optimization experiments identified R-guide, and reverse-asymmetric oligo as ideal candidates for editing this locus.
Embryo injection and embryo transfer to generate edited calves Precomplexed CRISPR RNPs (R-guide) and reverse asymmetric antisense oligo were injected into IVFderived zygotes and transferred into synchronized heifers at blastocyst stage to generate PRNP edited animals. Noninjected embryos served as controls for embryo transfer and for evaluation of embryo culture and embryo transfer technique. A summary of pregnancy outcomes from embryo transfers was shown in Table 2. Six noninjected control blastocysts transferred into 3 recipients (2 528 PARK ET AL. blastocysts/recipient) resulted in one pregnancy (16% efficiency) that went to term and resulted in the birth of a live calf. Another 18 microinjected blastocysts were transferred into nine recipients. Six of the nine recipients went to term and delivered seven live offspring (44% efficiency) and one stillborn calf (two twin pregnancies; Table 2). Of the seven live calves, one calf was a heifer, and the remaining six were bull calves, biasing the ratio toward male offspring.

Genotypic evaluation of edited calves
To genotype the calves, genomic DNA was extracted from ear notch and blood of all offspring. PCR amplification across the target site was performed using a low throughput RFLP analysis ( Fig. 2A) as well as a high throughput Illumina iSeq platform (Fig. 2B). A range of 8,417-25,008 sequence reads from the output FastQ files per each animal were analyzed using CRISPResso 2.0 software (with an exception of one sample with low reads: #752 blood). As shown in Figure 2C, all calves from injected embryos were edited. As expected from in vitro screens, all calves were mosaics and had a varied combination of NHEJ-and HDR-mediated repair outcomes. Among the edited calves, HDR-mediated introduction of G127V variant was identified in five of six calves at frequencies ranging from 9.53% to 94% from ear, and 4% to 93% from blood ( Fig. 2C) samples, with three of the six live calves showing >33% HDR reads (numbers 769, 786, and 789E), and two calves with >90% HDR reads ( 786 and 789E). Two of the six calves (769 and 789) had a greater representation of 1-bp and 4bp out-of-frame missense NHEJ events resulting in PRNP null genotype. We euthanized one of the founders (789E), which developed health problems unrelated to PRNP editing, and collected samples from eight different tissues (in addition to ear and blood sample), isolated DNA and performed genotyping in these different tissues via a similar low throughput RFLP (Fig. 3A) and high throughput iSeq analysis (Fig. 3B) to identify whether the genotypes from ear and blood accurately capture the allele frequency within the entire animal. As shown in Figure 3B, alignment of reads from various tissues were within the range of frequencies identified from ear and blood samples showing similar high HDR frequencies (*94%). To confirm germline transmission of edited alleles, semen from five bull calves at 6 months of age were electro-ejaculated. With the exception of one bull calf, spermatozoa was obtained from the ejaculate (Fig. 4A). iSeq analysis of spermatozoa identified frequencies of HDR edited allele ranging from 23% to 57% among the three edited calves (Fig. 4B).
In summary, from one round of embryo transfers we have generated three founders with high-frequency of germline transmission of G127V alleles (calf number 769, 49.02%; calf 786, 57.15%; and calf 789: 23.06%; sperm), with one of the founders also carrying high frequency of 1-bp deletion PRNP allele (769, 44.18% reads).

Discussion
The major goal of this manuscript was to establish a gene targeting reagent optimization pipeline to generate high frequency HDR modifications in embryos, so that the findings can be readily translated to generating live edited animals. As evidenced in the manuscript, two major competing forces need to be balanced for generating genome edited cattle for eventual commercial applicationsthe first being high efficiencies of genome editing and the other being pregnancy outcomes. The later which is  dependent on the quality of the embryos, is also influenced by choice of editing reagents. In a nutshell, optimizing genome editing reagents is a necessary first step for improving editing outcomes and for generating high quality embryos for successful pregnancies. As we have observed and similarly noted by others in the field, not all CRISPR sgRNAs and targeting reagents are equal in their targeting efficiency. For example, the F-guide with a CRISPR cut-site farthest (5 bp) from the targeted site resulted in no discernible gene targeting compared to the R-guide with a cut-site that was 2 bp from the target site and embedded within the spacer sequence; similarly, targeting oligo cis-to cut-site (antisense oligo) resulted in higher efficiencies than the sense oligo. In a further refinement of this screening process, we adopted a high throughput Illumina miSeq platform to perform targeted amplification at 10 · coverage across the cut-site. This high throughput screen presents a quantitative rather than a qualitative readout and provides an unbiased assessment of various genome editing outcomes. This is FIG. 4. Genotyping of spermatozoa from a live calf with high HDR frequencies. (A) Genomic DNA from spermatozoa was isolated for genotyping. The PCR amplicons were purified, digested with AccI, and resolved on agarose gel to identify successful gene targeting outcomes. Genomic DNA from euthanized calf with high HDR edits (789 E) and wild type cells were used as controls. We could not recover semen from one of the electroejaculated bull calf (827) and was not incorporated in the iSeq run. the first manuscript, to our knowledge, that systematically investigated the use of CRISPR-Cas system for HDR-mediated gene targeting directly in embryos for generating live gene targeted calves. Following the ex vivo optimization of reagents that resulted in optimal targeting efficiencies, we have proceeded with embryo injections and performed nonsurgical embryo transfers. All six resultant offspring were gene edited confirming high efficiency of editing following CRISPR RNP injections, consistent with our optimization experiments. Likewise, as reported by several groups, mosaicism in the injected embryos was a consistent outcome. 40 Mosaicism is especially a major concern for genome editing in cattle with long pregnancy and generation intervals. Recent publications have attempted to overcome this bottleneck, including modifying the CRISPR reagents and timing of the injections. 41,42 This will remain to be tested in livestock and will be a focus of our future investigations. Another interesting observation was inconsistency in the representation of edited alleles between soma (tissues) and germline (sperm). Although, we did not identify greater variation among the frequencies of edited alleles among various somatic tissues analyzed (Fig. 3), the same was not the case for germline (Fig. 4). This could be because, the primordial germ cells-the precursors for sperm and eggs-emerge from a small cluster of cells in the primitive streak stage gastrulating embryo and likely represent edits within a smaller cohort of cells. Regardless, from a round of embryo transfers we have generated five edited animals; two of the founders have high HDR frequencies (calves 769 and 786) and will be used for mating and propagation of the HDR edits. Likewise, two calves have high NHEJ efficiencies with either a one nucleotide (769) or a four-nucleotide deletion (789) that result in out-offrame deletions, will be similarly utilized for generating PRNP null calves. These will be part of the future efforts and are beyond the scope of current publication. In conclusion, we have established a pipeline for reagent optimization, high throughput screening, and embryo transfer to generate gene targeted live cattle via zygotic injections.