2

Heritable Genetic Modification in Food Animals

HERITABLE GENETIC VARIATION

Technical Background

An understanding of the origin and dynamics of heritable genetic variation within classical animal breeding programs is needed to provide the context for understanding the contributions and concerns posed by animal biotechnology.

The basis for all evolution and, in the context of this report, the development of improved lines of food animals is the existence of genetic variability. Simply stated, if there is no genetic variability upon which to select, genetic progress cannot be achieved. All genetic variability since the origin of life arose from new mutations. Direct genetic manipulation of the genome has added a new source of genetic variation; however, this technical advance does not change the fundamental nature by which plants or animals are improved. Mutations provide the genetic variation that is the basis for all evolution, from the first protocell formed billions of years ago to modern mammals. Evolution is the outcome of increased frequencies of mutations beneficial to the organism. In a natural setting, a beneficial mutation is one that increases an individual’s fitness (Darwin, 1859), that is, the likelihood that the individual possessing that variant will survive and reproduce. This same process of selection has been applied by humans to improve domesticated species for our use. For a breeder of any species, “beneficial” is defined as whatever form is regarded as useful for their purposes.

Alleles of a gene are one of two or more alternative forms of a gene (or DNA sequence) that exist naturally, arise by mutation, and are found at the same place on homologous chromosomes, termed a genetic locus. When a gene is duplicated and translocated to a different location in the genome, the resulting duplicate genes are referred to as paralogs. Currently, the mutations that give rise to alleles at a locus may be classified as: (1) single nucleotide variants (SNVs), or, if more common than 1 percent in frequency, a single nucleotide polymorphism (SNP); (2) small insertions and deletions with an arbitrary size of <50 base pairs (bp) as indels; or (3) larger mutations (>50 bp) as structural variants (SVs) (Mahmoud et al., 2019). Although not a conventional definition, the incorporation of foreign DNA as an SV was included by Hunt et al. (2018). Structural variants include a wide range of events from indels to billions of bp, including whole chromosome arms (Escaramis et al., 2015; Audano et al., 2019; Hunt et al., 2023). The definition of SV has changed from classical usage when SVs could be identified only by using microscopes and chromosome-staining techniques. Consequently, classic SVs were limited to large mutations of

millions of base pairs in length. Newer technologies, such as long-read whole-genome sequencing, have changed this definition to include essentially any alteration changing the length or configuration of the DNA sequence.

Mutations in DNA may directly alter the protein-encoding region of a gene (exon), resulting in an altered protein structure, and may result in a gain or loss of function of the gene product (Monroe et al., 2021; Holm et al., 2024). Mutations in the intergenic non-coding region between exons can result in alternative messenger RNA (mRNA) splice junctions (Boehm and Gehring, 2016) and thereby influence the structure of the protein that is produced, while changes to 5′- and 3′-untranslated regions can change translation initiation, elongation, protein folding, and subsequent protein expression (Jia et al., 2013; Khan et al., 2022). Sequence variants within the intergenic region also may alter gene regulation by changing the affinity of the gene’s promoter region to factors that bind and enhance or inhibit gene transcription or expression, alter epigenetic expression, change chromatin accessibility, or impact enhancer and repressor elements that alter the three-dimensional structure of the DNA molecule that brings distal regulatory elements together (Noordermeer and Duboule, 2013; Acemel et al., 2017; Clark et al., 2020; Orozco et al., 2022). In human populations, the most studied species among mammals, only 5 percent of known disease-associated SNPs have been identified in gene coding sequences, with the remaining 95 percent of disease-associated SNPs located in the intergenic, or non-coding, regions that make up 98 percent of the genome (Orozco et al., 2022). These impacts on phenotype would not have been predicted by genotype alone, demonstrating how little is understood about the complex nature of gene functions and regulation. As a result, phenotypic outcomes stemming from alterations in the genome, sometime as small as single nucleotide changes, can be impossible to predict. For example, scientists are unable to explain how a small duplication in an intergenic region could result in hornless cattle (Allais-Bonnet et al., 2013; Carlson et al., 2016).

Point mutations of DNA arise via three major mechanisms: (1) base misincorporation during replication of non-damaged DNA; (2) accumulation of DNA damage that has not been properly repaired, leading to mutation (Seplyarskiy and Sunyaev, 2021); and (3) movement of transposable elements (Wells and Feschotte, 2020). Mechanisms that can lead to SV include recombination errors, but SVs primarily result from non-allelic homologous recombination and errors in DNA double-strand break repair, such as non-homologous end joining (NHEJ) (Escaramis et al., 2015). Mutations can occur in both somatic and germinal cell lineages, but only those in the germline are heritable and transmitted to the next generation, as the germline generates the eggs or sperm that mediate transmission of genetic information across generations. Such mutations can have a range of effects upon expression of a trait, from very minor, whose effect is not distinguishable from background phenotypic variation caused by environmental effects, to major differences up to and including lethal alterations and complete loss of gene function, producing distinguishable, categorical differences in the resulting phenotype (Zhang et al., 2016; Hua and Springer, 2018; O’Connor, 2021).

Phenotypic traits that are continuous in nature, such as height, weight, and yield of edible product, are usually affected by many different alleles distributed over many genes and as such are termed polygenic traits (Falconer and Mackay, 1996). The kinds of mutations affecting polygenic traits are theorized to be those that act by modifying the rate of gene expression or functionality of the gene products, among other mechanisms (Charneski and Hurst, 2013; Rauscher et al., 2021). The kinds of mutations that change gene efficacy (i.e., gene expression levels and function) include synonymous SNVs in protein-encoding regions resulting from codon usage bias associated with alternative substrate concentrations of synonymous transfer RNAs (tRNAs), which can vary greatly within and between tissues (Agashe et al., 2013; Duechler et al., 2016). Small mutations in other parts of the gene also can change gene efficacy; these parts include promoter regions, enhancers in non-coding intergenic regions, and those in the 3′ and 5′ untranslated regions (UTRs) of genes. The 5′ UTR region has important regulatory functions at the mRNA level (Parker and Xia, 1999), including a key role in regulation of alternative transcription start sites and termination sites that can be the principal driver of transcript isoform diversity across tissues, as demonstrated for the NPY gene in rats (Parker and Xia, 1999; Voronina, 2002), with isoform diversity also evident in livestock species (Ren et al., 2023; Taylor et al., 2016; An et al., 2024). This functional switching could underlie cell-type specific proteomes and functions (Reyes and Huber, 2018). The 3′ UTRs have a different function: they are included in transcribed mRNAs and are involved in processes such as mRNA localization, stability, and translation (Mayr, 2019). In addition, the 3′ UTRs of mRNAs serve as hubs for post-transcriptional control as the targets of microRNAs and RNA-binding proteins (Bae and Miura, 2020). Such regulatory modifications are expected to result in

rather small changes in phenotype, assuming that small expression changes correlate with small phenotypic effects, especially within complex gene networks. However, nature is complex and there will always be exceptions to generalities, especially when binding sites are located in UTRs. For example, in Texel sheep, a single base change in the 3′ UTR of the gene responsible for controlling muscle growth (myostatin) creates a microRNA target site resulting in translational inhibition of myostatin and thus increased muscle mass (Clop et al., 2006). As such, it is easy for evolution to test the effect of such mutations upon fitness through multiple generations of selection and meiotic recombination.

In contrast, any mutation that changes the gene product rather than the rate at which the product is produced may have a major impact upon phenotype. These mutations include SNVs that are non-synonymous (i.e., change the amino acid encoded by a triplet nucleotide sequence), alter a stop codon, change a splice variant, or cause frameshifts, as well as some SVs. When a mutation has a visible or substantial effect on the phenotype, these mutations are generally referred to as monogenic (Hartwell et al., 2021). Examples of monogenic mutations in humans include dwarfism, hair and eye color, cystic fibrosis, Tay-Sachs disease, sickle-cell anemia, and Down syndrome. Animal examples of production interest include double muscling (McPherron and Lee, 1997) and the SLICK coat phenotype (Porto-Neto et al., 2018). In farm animals, only two “genes” have been identified as having causative monogenic mutations in intergenic regions: the polled locus, which results in the lack of horns in cattle and is due to a small duplicated region of DNA, and a SNP in a 12-bp intergenic region that leads to the callipyge phenotype (increased muscle mass) in sheep (Freking et al., 2002; Smit et al., 2003). However, the actual mechanism causing that phenotypic change is unknown (Hennig et al., 2022).

Importantly, selection operates upon individuals; that is, all mutations that affect a given trait are selected together as a common phenotype. Favorable mutations will increase in frequency across the population within which those mutations are beneficial. Those desirable alterations with larger effects on the phenotype will become common more quickly. The power of polygenic selection to bring about major changes in phenotype is well demonstrated. Long-term breeding programs have shown that selection upon such traits as body size or yield can result in changes in phenotype many times greater than the size of the trait originally selected upon (Eisen, 1980; Hill and Bunger, 2004; Moose et al., 2004; Muir et al., 2004; Havenstein et al., 2003a, b, 2007; Rocheford 2009). Polygenic selection also allows other traits that support expression of that phenotype to change at the same time, that is, the integrated traits of the entire organism change in harmony.

Genetic variation for such traits is captured by breeders who simply select individuals that are most meritorious for desired traits for use as breeding stock to produce the next generation. This process usually results in some improvement of the trait in the next generation. The amount of improvement observed is variable and depends upon the direct impact of genes combined with the extent to which random environmental variation impacts expression of the trait. Cumulative random environmental effects can cause environmental merit to become confounded with genetic merit in the selection process, that is, an individual can look meritorious because of chance environmental circumstances rather than having favorable alleles. In the generation after selection, the degree to which environmental effects are confused with genetic merit becomes known (Falconer and Mackay, 1996). If the progeny mean for the selected trait is the same as that of the parents, the trait is perfectly heritable; on the other hand, if the progeny mean of the selected trait is the same as the mean of the prior generation, the trait is not heritable, that is, the superiority of the parents was not transmitted to the offspring and was due to environmental, not genetic effects. The realized heritability of the trait is the degree to which improvement of the trait was realized through actual selection upon that trait.

This classical process of selective breeding was improved in the 1970s by introduction of methods to include pedigree information in the analysis of individual breeding value using Best Linear Unbiased Prediction (Henderson, 1975). This analytic approach helped track which pedigrees (i.e., which relationships) were most highly associated with performance. Application of this method greatly improved the accuracy of selection, especially for lowly heritable traits and traits that could not be measured upon the focal animal, but only upon its relatives, for example, milk production for dairy bulls, disease resistance, and carcass composition.

The next major advancement in commercial breeding programs was augmentation of the pedigree information with genotypic information provided by genotypes of SNVs spaced across the genome, so-called genomic selection (Meuwissen et al., 2001). Genomic selection improved the accuracy of selection even further because it allowed

favorable SNVs to be related to phenotype and thereby to directly place selection pressure upon the underlying genotype. In addition, genomic selection allowed selection of breeding candidates prior to the age when the actual phenotypes would become expressed, thereby reducing the generation interval. For example, in poultry layers, selection for egg production previously required two years per generation because it required one year to attain sexual maturity and then one year to collect egg production data. Now it is possible to select animals to become the next breeders as early as six weeks of age, when genotypes are identified, to produce the next generation.

Modern agriculture remains critically dependent on current and future genetic variation (mutations) to make improvements in agronomically important traits (Mba, 2013). An imminent problem for world agriculture is loss of genetic variation in elite commercial breeding populations due to selective breeding (Muir et al., 2008). That is, since animal breeders propagate only the best of the best, many favorable alleles are lost because they were not in the right combinations in individuals at the time of selection, and due to inbreeding resulting from the limited numbers of breeders selected for reproducing the population. Once lost, those alleles can never be recovered without compromising the performance of the elite population. This problem could be partially overcome by using larger breeding populations or by reducing selection intensity, such that more alleles can be sampled in the next generation. Both strategies are costly, however, and in a competitive commercial environment, animal breeding companies must survive the short term to compete in the long term.

Random mutagenesis has historically, in evolutionary terms, been a solution to this problem when time and cost are not an issue. One solution is to accelerate the rate at which new genetic variability is produced, that is, random mutagenesis (Ma et al., 2021). This approach is especially attractive for traits that do not have existing genetic variability. Mutagenesis has been applied to plant breeding for many decades using chemical agents such as 5-bromouracil and ethyl methanesulfonate, as well as irradiation with UV light, X-rays, gamma-rays, and protons (Tanaka et al., 2010). Several crops have been improved using random mutagenesis, including rice, barley, beans, soybeans, and wheat (Ma et al., 2021; Martínez-Fortún et al., 2022). The joint Food and Agriculture Organization and International Atomic Energy Agency (IAEA, 2022) mutant variety database lists over 3,000 mutant plant varieties officially released or commercially available around the world. The predominant downsides to random mutagenesis are embryonic lethality; the need to screen for undesirable mutations, which are often recessively inherited; and problems associated with incorporating new mutations into elite commercial germplasm. Those issues include identification of the favorable mutations, the number of generations needed to introgress favorable mutations into elite populations, and associated linkage drag of unfavorable alleles. In animals, additional issues include ethical concerns, reproductive limitations, generation interval, and cost. As a result of these challenges, random mutagenesis has not been pursued for the genetic improvement of food animals. However, once a favorable mutation (allele) has been identified, as noted below, gene-editing technology can be used to directly modify existing genes to create desired DNA sequence variants without associated introgression and linkage downsides. An alternative source of useful genetic variability is local and heritage animal breeds, which hold useful alleles that adapt these animals to conditions including less-favorable environmental and climate conditions and to nutrient-poor diets. Their alleles, through crossbreeding, can be introgressed into more widely produced lines.

Assisted reproductive technologies

Ultimately, the goal of genetic modification in a food animal is to deploy a new or improved trait into production populations, and this feat requires germline transmission. Research and advances in technology during the past 100 years have yielded several methods for the generation of food animals with heritable genetic modifications (HGMs), which for mammals requires preimplantation embryo manipulation. As such, future methods for integrating genetic modifications in food animals are likely to involve implementation of strategies involving gene transfer or direct germline gene editing. In mammals, genetic changes can either be introduced into a zygote-stage embryo (a one-cell fertilized oocyte) or into somatic cells in culture followed by transfer of the modified diploid nucleus into an enucleated oocyte (egg) using the process of somatic cell nuclear transfer (SCNT; Table 2-1). In both technologies, the resultant embryos are then transferred into recipient dams to generate a live-born animal. Each approach requires the use of assisted reproductive technologies including generation of preimplantation embryos, either in vivo or in vitro, artificial activation of embryo development, in vitro culture of embryos, and

TABLE 2-1 Methods for Producing Food Animals with Heritable Genetic Modifications

transfer of embryos to establish pregnancies (NRC, 2002). These procedures, however, impact the viability of embryos and can produce unintended epigenetic modifications irrespective of the genetic alterations brought about by gene editing itself (de Waal et al., 2014, 2015).

Generation of Preimplantation Embryos

Success in generating food animals with intended genetic modifications, as well as animals with germline-engineered genomes, relies on the introduction of a transgene or gene-editing components into zygote-stage embryos or the artificial generation of zygote-like embryos by nuclear transfer. The primary means for creating zygotes are: (1) in vivo fertilization followed by recovery from the female reproductive tract, that is, from the oviduct; and (2) in vitro fertilization (IVF) of artificially matured oocytes with either freshly collected or frozen and thawed sperm.

For fertilization in vivo, a combination of estrous synchronization and artificial insemination is used to predict ovulation and time of conception to establish a window of time for flushing of zygotes prior to the first cleavage division. A transgene construct or genome-editing components are then introduced into the embryos in vitro through either microinjection (Park et al., 2017) or electroporation (Miao et al., 2019), both of which are effective at producing HGMs, but which can reduce embryo viability. In this workflow, both the sire and dam are known to the developer and can be used for direct comparison of their genotypes with those of any resulting HGM offspring. The production of embryos in vitro by IVF, introduction of HGM components, in vitro embryo culture, and transfer to recipient females reduces embryo viability and capacity to establish pregnancies. Thus, an efficient HGM workflow for mammals can require 100 or more good-quality oocytes. The litter-bearing reproduction of swine makes in vivo generation of zygote embryos plausible because a single female will ovulate more than 20 oocytes each estrous cycle even in the absence of exogenous hormone stimulation. However, for ruminants, only one to two oocytes are ovulated per estrous cycle. Although exogenous hormone stimulation for superovulation can produce 5-10 oocytes, that number is still insufficient for current HGM workflows. An advantage of in vivo embryo generation is that fertilization rates using artificial insemination are typically high. A disadvantage of this workflow is that the window of development to flush embryos at the beginning of the zygote stage is difficult to control. Moreover, it is often the case that embryos are flushed for treatment with HGM components at the end of the zygote stage or at the two-cell stage, resulting in HGM occurring after the first cleavage division. If HGM occurs in the multicellular embryo stage, mosaicism is a likely outcome.

For IVF, oocytes are aspirated from ovarian follicles and matured in vitro prior to being fertilized by sperm. The percentage of oocytes that is successfully fertilized and subsequently develop into an embryo varies widely and is impacted by several factors including follicle stage, in vitro maturation conditions, quality of sperm, and in vitro conditions (e.g., media, temperature, and oxygen tension). After fertilization, HGM reagents are introduced via microinjection or electroporation which, as with in vivo-produced embryos, reduces viability. Although laparoscopic guided ovum pickup can be used to source oocytes, the number that can be retrieved is often below the threshold needed to produce enough viable embryos by IVF and treatment with HGM reagents for subsequent transfers and pregnancy establishment. Thus, an in vitro production (IVP) workflow is often utilized which entails sourcing oocytes from ovaries of multiple females that are obtained from abattoirs. Although advantageous because many oocytes will be available for zygote generation and editing, a disadvantage is that although the sire will be known, the dam will not, making direct genotyping comparison between HGM offspring and parents challenging. Another advantage of the IVP workflow is greater control over the timing of embryo development and introduction of HGM reagents at the onset of the zygote stage, which reduces the incidence of mosaicism. Indeed, IVP is the primary strategy for generating cattle, sheep, goat, and swine embryos in an HGM workflow (Park et al., 2017; Miao et al., 2019; Ciccarelli et al., 2020). It should be noted that while slaughterhouse-derived oocytes for IVP are advantageous during the early stages of HGM development (assessing phenotype and testing hypotheses), elite genetics are used when generating animals for breeding stock.

For generation of HGM animals via SCNT, the workflow is initiated with modification of the DNA of somatic cells in vitro. Modifications will vary by individual cell; thus, clonal expansion is necessary, which requires a prolonged period of in vitro growth and cell divisions. Although most genetic errors are repaired correctly, each round of DNA replication required for a cell division can incur errors that go unfixed. These DNA replication errors can be a source of mutations that are independent of HGM machinery processes (Chia et al., 2017). Because the HGM somatic cell genome is diploid, its use in embryo generation requires the enucleation of an oocyte that has been matured in vitro. In addition, the fertilization process is not replicated by the transfer of a diploid genome, and hence artificial activation is required to initiate embryonic development. Further, the epigenetic reprogramming that normally occurs post-fertilization at the zygote embryo stage is often abnormal following nuclear transfer (Bourc’his et al., 2001; Wilmut et al., 2002; Hiendleder et al., 2004; Xue et al., 2002). Although the SCNT technique is advantageous because the genotype will be known prior to embryo generation, the multiple artificial aspects of this approach lead to low efficiency. The efficiency of SCNT, defined as the percentage of transferred embryos that result in healthy offspring, generally ranges from 1 to 5 percent across species (Rodriguez-Osorio et al., 2012; Gouveia et al., 2020), with rates of up to 5 percent reported in pigs (Whitworth and Prather, 2010; Kurome et al., 2013; Li et al., 2013a; Liu et al., 2015; Secher et al., 2017) and up to 10 percent in cattle (Sangalli et al., 2023). Further, developmental abnormalities are often observed in the first-generation offspring (Mordhorst et al., 2019). However, problems encountered due to abnormal epigenetic reprogramming are largely eliminated in the subsequent generations obtained by natural breeding.

In vitro culture of preimplantation embryos

The generation of gene-edited embryos via IVF or SCNT requires in vitro culture to introduce genome-editing components and allow for recovery and/or development prior to transfer to recipient females. For ruminants such as cattle and sheep, embryos subjected to HGM procedures at the zygote stage are cultured for several days to reach the blastocyst stage. For swine, embryos are often cultured to the two-cell or blastocyst stage of development, which can span from a few hours to several days. Regardless of the amount of time, to yield embryos with the greatest chance of establishing a pregnancy, the in vitro conditions must mimic as closely as possible the in vivo environment that supports normal preimplantation development. To achieve these conditions, a plethora of studies have been conducted to define media, temperature, and oxygen tension conditions that support development of swine, sheep, and cattle embryos in vitro; however, conditions remain to be refined for supporting in vitro blastocyst development rates of more than 50 percent (Cordova et al., 2017; Srirattana et al., 2022).

Inheritance of DNA bearing heritable genetic modifications

Regardless of the workflow used, the introduction of HGM components into a preimplantation embryo can modify DNA in early pluripotent (blastomere) cells that give rise to the germline and the soma (the parts of the organism other than the reproductive cells). If the HGM machinery remains active past the one-cell zygote stage, it is possible that up to eight copies of each targeted locus will be subject to differential modification and inheritance at the next cleavage-stage cell division. Thus, animals born directly from HGM embryos (i.e., founders) could have mosaicism of intended and unintended DNA modifications that are soma specific and not transmitted to the next generation via the germline and natural sexual reproduction. As such, detailed molecular characterization of both intended and unintended HGM events in the soma of founder animals is less important than focusing on what is inherited by the first generation of offspring. Because offspring inherit only one copy of each chromosome from a parent, the genome is thereby cleaned of mosaicism following syngamy of haploid genomes provided by sperm and eggs of HGM founders.

Biotechnological Methods for Generating Heritable Genetic Modifications in Food Animals

Gene transfer

A transgenic line is one that has had a sequence of DNA stably introduced into its genome and transmitted to offspring in a Mendelian fashion. The DNA sequence introduced is designed to accomplish a specific phenotypic change in the animal. Thus, the goal of the transgenic approach to genetic improvement is the same as in selective breeding—introducing permanent genetic changes to impact desired traits—but can be accomplished in fewer generations and more reliably (i.e., can achieve changes in current, elite genetics without altering alleles at all other genes) than breeding and selection alone. Therefore, once a gene is associated with a desired trait, that gene can be introduced into the genome of the animal with the goal of adding or modifying the gene product of interest (encoded by that gene) to change that specific characteristic in the animal.

While methodologies for producing animals without natural breeding (i.e., IVF, embryo transfer and culture, or artificial insemination) have been established and techniques for introducing DNA to cells (i.e., transfection, microinjection, electroporation) have existed for decades, the advent of recombinant DNA techniques in the 1970s laid the foundation for the generation of transgenic animals. Recombinant DNA enabled the cutting and joining of DNA molecules in vitro to generate a transgene, that is, a DNA construct designed to drive expression of a gene of interest in an animal. Transgene expression can either be constitutive (in every cell of the animal), tissue specific, or temporal specific based on what promoter elements drive expression of the transgene. The desired promoter and protein-encoding region (complementary DNA or genomic DNA) are combined in a plasmid and propagated in Escherichia coli. The source of DNA for the promoter and protein-encoding elements of a transgene could be from the same species being modified, from a different species, or from a combination of species. Once a transgene is assembled, there are several means by which this piece of DNA can be introduced into the genome of an animal, including pronuclear microinjection, retroviral vectors, or DNA transposons.

The concept of being able to introduce foreign genetic material into the genome of a mammal was first demonstrated by Jaenisch and Mintz in 1974 by injecting the SV40 virus into mouse embryos. Building on this proof of principle, the first transgenic mammals made using recombinant DNA were mice generated in 1981 (Costantini and Lacy, 1981; Gordon and Ruddle, 1981), and the first demonstration of an altered phenotype came a year later with the production of growth hormone (GH) transgenic mice (Palmiter et al., 1982). These transgenic mouse lines were generated by pronuclear microinjection. With this technique, zygotes are collected from donor females or produced in vitro from slaughterhouse-derived ovaries, and the transgene is injected into the pronucleus of each zygote via a glass needle controlled by a micromanipulation setup and visualized using a microscope. Surviving zygotes are then surgically transferred to a reproductively synchronized recipient dam and pregnancies are established. Once born, all offspring are screened for presence of the transgene, and then lines can be established and maintained by natural breeding. With pronuclear microinjection, the transgene integrates into a random region of the genome in only about 1 percent (Rexroad, 1992; Wall, 2001) of the injected zygotes, making the process very

inefficient in most food-animal species (cattle, pigs, sheep, goats, and chickens). These inefficiencies result from several challenges, including the inability to clearly visualize the pronuclei of livestock embryos, poor embryo survival, and low rates of integration into the genome. Nevertheless, this standard method was used to generate the first transgenic food animals (sheep, pigs, and fish) in 1985 (Hammer et al., 1985; Zhu et al., 1985), with examples in goats and cattle following in 1991 (Ebert et al., 1991; Krimpenfort et al., 1991) and chickens in 1994 (Love et al., 1994).

The use of microinjection to deliver lentiviral vectors carrying a transgene to the perivitelline space of zygotes increased the efficiency of transgenic animal production, with 10-30 percent of the zygotes injected resulting in a transgenic founder (Hofmann et al., 2003) and 60-90 percent expressing the transgene. As with pronuclear microinjection, the virus harboring the transgene integrates into a random location within the genome. While more efficient, this approach has several drawbacks, including restrictions on transgene size, a high frequency of mosaics, and challenges associated with insertion sites being close to genes. The lentiviral approach has been used in pigs (Hofmann et al., 2003), cattle (Hofmann et al., 2004; Tessanne et al., 2012), and sheep (Cornetta et al., 2013), among other species, for a variety of applications. As with standard microinjection, the random nature of viral insertion makes it impossible to target genetic changes to specific sites in the genome. Hence, both pronuclear microinjection and the use of retroviral vectors are effective techniques only for gene addition.

Since the first report of a transgenic animal via microinjection, there has been non-zygote-based research on methods that enable the site-specific integration of DNA into the genome. For these methodologies, embryonic stem cells (ESCs) or somatic cells are modified in vitro; after the modified cells are screened for site-specific insertion, SCNT is used to create an animal with the desired change. Targeted genetic modification was first demonstrated in mice rather than in food-animal species due to the feasibility of culturing and maintaining mouse ESCs in vitro. ESCs are undifferentiated (i.e., their fate has not yet become restricted to a certain cell lineage), pluripotent (capable of developing into many cell types, germ or somatic) cells derived from early embryos (the inner cell mass of blastocysts) that can be maintained in culture. A specific change in DNA sequence can be directed to a specific site in the genome by introducing a targeting vector composed of the DNA sequence bearing the modification of interest flanked by homology arms consisting of regions of DNA complementary to the genomic sequence flanking the site of the desired HGM. The targeting vector can be introduced via standard transfection or electroporation procedures whereby the DNA enters the cell and can be a substrate for homology-directed repair, inserting the DNA at the correct location in the genome. This type of site-specific integration is still very inefficient (1 in 10⁴-10⁶ cells) but the HGM can be screened for with cells in culture. Once a modified ESC line is isolated, the modified cells can be injected into a blastocyst-stage embryo where they incorporate into the inner cell mass and contribute to the formation of the individual, resulting in a chimeric founder animal. The resulting chimeric founders are then bred, and while germline competence is often low, if the modified cells contribute to the germline, a transgenic line can be established. While ESCs from food-animal species have proven difficult to isolate, cells with some properties indicative of pluripotency from cattle have recently been reported (Bogliotti et al., 2018). However, no germline competence, which is a hallmark of true pluripotency, has been demonstrated to date in cells from food-animal species.

SCNT-based techniques have also been applied to target genetic modification in livestock species. Using this approach, a somatic cell is grown in culture, genetically modified, and subsequently used to create an animal by fusing the modified cell with an enucleated egg (Wilmut et al., 1997). The reconstructed embryos are then transferred to surrogate dams to develop into offspring with the genetic content of the cell used for cloning. While this technique enabled site-specific modification in food animals and all offspring were transgenic, it was very inefficient, with only 0.4-1.7 percent of transferred embryos resulting in a live birth depending on the cell type used (Campbell et al., 1996). Until the advent of gene-editing technologies, this was the main method for generating targeted modifications of specific genes in cattle and pigs (Hodges and Stice, 2003; Lai and Prather, 2003).

To date, classical transgenesis has been used in food-animal species for specialized non-agricultural purposes (e.g., medical models, xenotransplantation, and animal bioreactors for pharmaceutical production), as well as for the improvement of animals and their food products (reviewed in Garas et al., 2015). For example, transgenic approaches have been applied to develop animals that are more resistant to diseases, such as cattle that are more resistant to mastitis (Wall et al., 2005) and bovine spongiform encephalopathy (BSE; Richt et al., 2007) and chick-

ens that are more resistant to avian influenza (Lyall et al., 2011). Transgenic food animals have also been generated to modify meat and milk composition, including pigs rich in omega-3 fatty acids (Lai et al., 2006), dairy cows (van Berkel et al., 2002) and goats (Maga et al., 2003, 2006) expressing antimicrobial constituents of human milk in the mammary gland, and transgenic cows lacking a major milk allergen (Jabed et al., 2012). Transgenic pigs better able to utilize phosphorous sources in their diet by salivary expression of phytate secrete 75 percent less phosphorus in their manure, lessening the environmental impact of production (Golovan et al., 2001). Following work based on the GH-transgenic mice, applications toward improved productivity focused on growth; examples include the development of transgenic Atlantic salmon that are more efficient at converting feed into body mass, allowing them to reach market weight twice as fast (Du et al., 1992), as well as multiple examples in pigs (Hammer et al., 1985) including pigs that express GH constitutively, those that express insulin-like growth factor 1 (IGF-1) in muscle (Monaco et al., 2005), and those that express increased levels of α-lactalbumin to increase milk production and thereby enhance the growth rate of piglets (Bleck et al., 1998).

In most of these examples, the transgene inserted had the desired impact upon the phenotype of the animals. However, unlike the GH-transgenic mice and salmon, GH-transgenic pigs suffered from a variety of ailments including ulcers and impaired reproduction (Pursel et al., 1990). This led researchers to shift their focus to improving growth by restricting the expression of the growth promoting gene, IGF1, to muscle, which resulted in leaner body mass and improved feed efficiency (Pursel et al., 1999). Another instance in which a goal was not met was in the development of transgenic goats that expressed spider silk in their milk (Karatzas et al., 1999); in that case, the silk fibers interfered with mammary function when expressed at a high level, causing a welfare issue. It should be noted that in both cases, the use of genetic engineering technology in itself was not the cause of compromised welfare, but rather the predictions regarding the intended applications were not proven true.

The classical transgenesis approach can result in DNA sequences other than the transgene becoming introduced into the genome. With microinjection, the transgene construct is assembled in a circular plasmid vector, which is linearized prior to injection, often containing some plasmid DNA at the ends of the transgene. In addition, the transgene can integrate into the host genome as concatemers of multiple transgene copies. With retroviral vectors, a proviral DNA sequence is inserted alongside the transgene. With ESC or SCNT techniques, antibiotic-resistance cassettes are present in the transgene. The purpose of this is to provide a means to select cells that have taken up the transgene by culturing the cells in media containing the antibiotic encoded by the cassette; cells with the transgene are able to grow in this environment, while those that have not taken up the transgene will not survive. Several strategies can be employed to remove antibiotic-resistance genes, but the presence of plasmid, viral, or antibiotic-resistance-encoding DNA can pose additional hazards (see Chapter 3). With newer genome-editing technologies, the addition of such “accessory” DNA sequences is no longer necessary to make the desired genetic modification. Thus, newer genome-editing technologies that are based on clustered regularly interspersed short palindromic repeats (CRISPR) pose fewer potential risks to both the HGM animal and consumers of HGM animal-derived food products.

Genome editing

Making precise, site-specific changes in the animal genome has long been a goal for many applications, including foundational, biomedical, and applied research. The ability to introduce and select for a specific change to a specific gene was first reported in the late 1980s (Thomas and Capecchi, 1986). Subsequent generation of the first gene-targeted knockout mouse (Thomas and Capecchi, 1987) culminated in the 2007 Nobel Prize in Physiology or Medicine being awarded to Mario Capecchi, Martin Evans, and Oliver Smithies for their discoveries enabling gene targeting in mice via the use of ESCs (Nobel Assembly, 2007). This groundbreaking work laid the foundation for the current protocols for gene targeting in use today in the field of gene editing. Indeed, Jennifer Doudna and Emmanuelle Charpentier received the Nobel Prize in Chemistry in 2020 for the development of a method for gene editing (Royal Swedish Academy of Sciences, 2020), highlighting the importance and potential of these technologies.

There are currently two classes of genome-editing tools: engineered nucleases (proteins custom-designed to bind to and cut DNA) and CRISPR-Cas (elements of a bacterial defense system) (Table 2-2). Each class of tools

TABLE 2-2 Common Methods for Making Site-Specific Modifications in the Genomes of Food-Animal Species

accomplishes the same objective by enabling site-specific cuts in the genome but does so in different fashions. An engineered nuclease is a hybrid protein consisting of a DNA-binding domain coupled to a DNA-cleavage domain. To function, a pair of proteins is made, each containing a DNA-binding domain and a DNA cleavage domain. The DNA-binding domains are designed to match the DNA sequence on either side of the target region. When the pair finds and binds to their matching region in the genome, the two cleavage domains come together and are activated, cutting the DNA between them, leaving a double-strand break at a desired spot in the genome. The two types of engineered nucleases, zinc finger nucleases (ZFNs) and transcription activator-like effector nucleases (TALENs), differ in the type of DNA-binding domain that they possess.

The DNA-binding domain of ZFNs is a zinc finger, a three-dimensional structural motif with a finger-like shape characteristic of classes of proteins in nature that recognize and bind to DNA in a sequence-specific manner (Cathomen and Joung, 2008). More specifically, these zinc finger motifs can recognize a triplet of base pairs in a certain order. For ZFNs, “custom” zinc finger proteins are designed that string together 3 to 6 of these zinc finger recognition units to enable binding to a specific sequence of DNA ranging from 9 to 18 base pairs, respectively. Attached to this DNA-binding domain is a DNA cleavage domain comprising the type II restriction endonuclease FokI, an enzyme that will cleave DNA with no sequence specificity. Thus, the DNA-binding domain guides the DNA cleavage enzyme to the desired site in the genome to produce a cut. To function, ZFNs have to be used in pairs, each binding to opposite DNA strands on either side of the desired cut site oriented so that the FokI domains meet in the middle. When both ZFNs are bound to the DNA, FokI is activated and will cut the DNA, producing a double-strand break (Kim et al., 1996). Building on this base knowledge and earlier work that demonstrated the ability to introduce an intentional double-strand break in genomic DNA using a meganuclease (Rouet et al., 1994), the first use of ZFNs for gene-editing purposes in animals was reported by Bibikova et al. (2002) and helped lay the foundation for the field of gene editing as we know it today.

The DNA-binding domain of TALENs is based on a virulence factor from the pathogenic bacterium Xan-thomonas. This virulence factor is a member of the transcription activator-like (TAL) effector family that, once in a cell, can mimic transcription factors that bind to the host DNA and disrupt normal function of the host cell. The DNA-binding domain of the TAL protein is present in a series of repeats, each with a conserved 33- to 34-amino acid sequence with the 12th and 13th amino acids being variable and termed the repeat variable diresidue (RVD). The amino acids present at the RVD determine the DNA base to which TAL will bind. Like ZFNs, custom TAL-like proteins can be designed that string together the RVD for any series of base pairs and are coupled to the FokI DNA cleavage domain (Christian et al., 2010). Generally, 18 RVD are strung together on either side of the cut site, with the only constraint being that each target is preceeded by a thymine (T) base. TALENs are also used in pairs, with FokI being activated to cut when dimerized (i.e., when both TALENs bind).

The clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated protein (Cas) system accomplishes the same outcome as ZFNs and TALENs (making a cut at a precise site in the genome) but does so through an RNA-guided nuclease, an enzyme that cleaves DNA, whereby a small, chimeric non-coding guide RNA (gRNA) directs a Cas nuclease to the target site where the cut is made. The CRIPSR-Cas9 system

was derived from elements of a bacterial defense system. Fundamental research showed that bacteria contain arrays in their genome representing sequence-matching parts of invading phage genomes separated by repeating (palindromic) DNA sequences (Mojica et al., 2005). These sequences, termed spacers, are complementary to a protospacer in the viral genome that is flanked by a short (usually 3-bp) sequence termed the protospacer adjacent motif (PAM). These spacers comprise a mechanism by which bacteria could make a memory of DNA from invading phages to then be able to target them for destruction if the bacterium becomes infected again. The array of spacers gets transcribed into individual short RNA sequences termed CRISPR RNA (crRNA) with the aid of trans-activating CRISPR RNA (tracrRNA) that, once formed, associate with the Cas nuclease, one of the enzymes produced by the CRIPSR system. The complex is then guided by the sequence of the crRNA to find its complementary phage DNA and target it for cleavage by Cas, thus limiting the ability of the phage to propagate itself inside the bacterial host (Barrangou et al., 2007).

Once the basics of this system became known, further research was conducted to adapt the system to work outside of a bacterial cell and to target DNA other than phage sequences. Pivotal to the ability to use this system for gene editing was the foundational work by Charpentier, Doudna, and others that defined the structure of an RNA molecule to guide Cas9 to cleave DNA in vitro and demonstrated that any sequence could be targeted if it is preceded by a PAM site (Jinek et al., 2012). This foundational work led the way to gene editing in higher-order cells, first demonstrated in human and mouse cells in 2013 (Cong et al., 2013; Mali et al., 2013) and subsequently in a variety of other organisms ranging from yeast (DiCarlo et al., 2013) to flies (Gratz et al., 2013), worms (Friedland et al., 2013), fishes (Hwang et al., 2013), and plants (Li et al., 2013a).

This CRISPR-Cas9 system requires two components: a single guide RNA (sgRNA) molecule (tracrRNA-crRNA hybrid) and the Cas9 enzyme. A gRNA can be designed to target any region of DNA in the genome if it precedes a PAM site. In the case of the most-used CRISPR-Cas9 system from Streptococcus pyogenes, the PAM site is NGG, where N represents any base. Thus, by introducing a gene-specific gRNA and Cas9 to any cell, both strands of the target DNA are cleaved, resulting in a double-strand break in the targeted sequence within the genome. Other versions of this system include Cas9 nickases (where only one strand of DNA is cleaved), base editors (modified Cas9 that deaminates bases to change their coding; Komor et al., 2016) and prime editing (modified Cas9 that can introduce short, desired changes with no double-strand break required (Anzalone et al., 2019). Multiple Cas enzymes (Cas12, Cas13) derived from other bacteria also have been identified with specificities for DNA or RNA. CRISPR-based systems such as programmable addition via site-specific targeting elements (Yarnall et al., 2023) are being developed that improve insertion efficiency and reduce off-target events (Table 2-3). However, with the exception of base and prime editing, most tools are being developed for human gene therapy and have not yet been used in food-animal species.

TABLE 2-3 CRISPR-Cas and Derivative Gene-Editing Methods

NOTES: Cas9n = Cas9 nickase; gRNA = guide RNA; HDR = homology-directed repair; NHEJ = non-homologous end joining; pegRNA = prime-editing guide RNA.

The common feature among all ZFN, TALEN, and CRISPR-Cas9 gene-editing tools is the creation of a double-strand DNA break at a desired location in the genome. Once this break is made, endogenous DNA repair pathways initiate repair of the damaged DNA. The two primary methods of DNA repair in eukaryotic cells are non-homologous end joining (NHEJ) and homology-directed repair (HDR). The NHEJ pathway imprecisely joins the broken DNA ends in a template-independent manner. This error-prone repair pathway frequently results in insertions or deletions (indels) in the repaired DNA sequence. The indels are heterogeneous in size, but frequently cause frameshift mutations that disrupt the protein-encoding sequence, often resulting in a premature stop codon and/or a non-functional gene product. As such, NHEJ-mediated repair is often used experimentally to inactivate or “knock out” a gene of interest.

In contrast, the HDR pathway utilizes a homologous DNA sequence as a template to precisely repair the damaged DNA. Whereas NHEJ is active throughout the cell cycle, HDR is functional only during the replicative phases (S to G2) of the cell cycle. HDR typically uses a homologous template from a sister chromatid to restore the lost genetic information in a highly specific and precise manner. Experimentally, this pathway can be exploited by providing an exogenous DNA repair template containing the desired DNA change (the edit) along with regions of homology that are necessary for error-free repair. HDR-mediated repair is used experimentally to make precise DNA edits or to insert or “knock in” a new DNA sequence.

Following the generation of a double-stranded DNA break, both repair pathways are active in the cell when a DNA repair template is present. However, the HDR pathway is less efficient than the NHEJ pathway. Thus, when a precise DNA edit is desired, scientists must screen populations of cells to identify those containing the desired repair. The efficiency of this process has direct implications for the clinical or agricultural applications of this technology.

The various outcomes that can occur from genome modification by a site-directed nuclease (SDN) fall into three categories. SDN1 modifications include the small indels that occur following the NHEJ repair mechanism. At the molecular level, SDN1 mutations are of the same type as those that occur naturally as the result of exposure to chemicals or radiation. SDN2 modifications are small, precise genetic modifications achieved by the HDR pathway using an exogenously provided DNA repair template. SDN2 modifications can include specific base substitutions or small indels that modify the target gene. SDN3 modifications are those that result in the incorporation of new genetic material, such as transgene insertion. The type of modification that is generated can impact the regulatory requirements for gene-edited crops and food animals (Wray-Cahen et al., 2024). For example, SDN1 and SDN2 modifications can be made to genes without introducing any foreign DNA, potentially avoiding regulatory concerns regarding transgenesis.

Breeding and Evaluation of HGM lines

The application of gene transfer or gene-editing techniques, when successful, results in the production of individuals that may become founders of HGM animal lines. The founders are evaluated for incorporation of the intended modification, ascertaining whether the intended modification was indeed achieved and whether these prospective founders exhibit healthy phenotypes and express the targeted trait. Founders that meet these criteria will then be bred to produce individuals that bear the intended modification in both homologous chromosomes of a pair. These offspring will be further evaluated for expected Mendelian transmission of the HGM, healthy phenotype, and expression of the desired trait. Only with success to this point might a developer present the line for regulatory review and for multiplication for production of a possible commercial breeding stock. Thus, line development is a mutigenerational undertaking. As noted in Chapters 3 and 4, these generations provide the opportunity for risk assessment and management.

GROUP-SPECIFIC METHODS, APPLICATIONS, AND ISSUES

Mammals, poultry, and fishes present different reproductive strategies and biologies, leading to different choices of experimental protocols for gene transfer or genome editing (Figure 2-1). This section discusses methods, agricultural applications, and key issues for the development of HGM in each animal group.

Mammals

The initial achievement of creating transgenic food animals resulted from the translation of methods developed with mice for the microinjection of recombinant DNA into pronuclear-stage embryos. The advent of SCNT methodology allowed introduction of genetic modifications into somatic cells in vitro followed by generation of whole animals. Over the past decade, CRISPR-based genome editing has replaced TALEN and ZFN platforms to enable the precision engineering of food-animal genomes through either direct application to zygote-stage embryos or cells to be subject to SCNT to yield whole animals (Figure 2-2).

Agricultural applications of transgenesis in mammals used for milk, meat, and fiber production (cattle, pigs, sheep, and goats) include modifying production-related traits, disease resistance, and food composition. One of the main ways of improving the efficiency of livestock production systems is accelerating the time to maturity (i.e., growth rate) or yield of product (meat, milk) per animal. As such, early use of transgenes focused on increasing growth rate via growth hormone expression in sheep (Rexroad et al., 1989; Ward and Brown, 1998) and pigs (Pursel et al., 1989, 1990). Pigs expressing phytase in their salivary glands were better able to utilize dietary phosphorus and exhibited improved feed efficiency in addition to secreting less phosphate in their manure (Golovan et al., 2001). Other growth-related applications in pigs promoted the growth of nursing pigs by expressing an antimicrobial enzyme in the milk to protect piglets from infection (Tong et al., 2011) or by increasing the amount of milk that a sow produces (Wheeler et al., 2001), as milk production accounts for 44 percent of the pre-weaning growth of pigs. Transgenic animals also have been produced with the goal of producing more meat (i.e., muscle) by either enhancing muscle development (Pursel et al., 1992) or preventing myostatin, a protein regulating muscle growth, from functioning to result in “double-muscled” cattle (Tessanne et al., 2012) and goats (Zhou et al., 2013). Researchers have also applied transgenesis to increase wool growth in sheep (Damak et al., 1996; Bawden et al., 1998).

Transgenesis has also been used to enhance disease resistance in food-producing mammals. For instance, transgenic cattle (Richt et al., 2007) and goats (Golding et al., 2006) have been generated that lack the ability to make prion proteins. Misfolded prion proteins result in prion diseases such as BSE (“mad cow disease”), in cattle and PrP(CJD) in Creutzfeldt-Jakob disease in humans, and thus animals lacking prions are thought to be less susceptible to BSE. Mastitis is an udder infection commonly caused by the bacterium Staphylococcus aureus, and transgenic dairy cows expressing lysostaphin, a bacterial hydrolase specific to S. aureus, in their mammary gland are resistant to S. aureus mastitis (Wall et al., 2005). Several applications in pigs rendered them more resistant to viral infection caused by influenza (Muller et al., 1992) and foot-and-mouth disease virus (Hu et al., 2015).

Researchers have also sought to modify the composition of meat and milk to benefit human health. Transgenic pigs (Saeki et al., 2004; Lai et al., 2006), cows (Wu et al., 2012), and sheep (Zhang et al., 2013) expressing genes to modify fatty acid composition have been produced, resulting in meat with increased levels of heart-healthy omega-3 fatty acids. A desaturase gene has been expressed in the milk of goats to decrease the levels of saturated fat (Reh et al., 2004). Increasing the protein content of milk has potential benefits for consumers as well as processors of milk (i.e., cheesemakers). Work in dairy cows showed it was possible to increase the protein content of milk by overexpressing two endogenous milk proteins, β- and κ-casein (Brophy et al., 2003). Of the milk proteins, β-lactoglobulin is the major milk allergen in bovine milk, and transgenic cows lacking this protein have been generated (Jabed et al., 2012) as a potential source of milk that will not elicit human allergy reactions. Lastly, several groups have expressed antimicrobial constituents of human milk, lysozyme and lactoferrin, in the milk of cows and goats in attempts to improve both udder and human health. Milk from lactoferrin transgenic cows (van Berkel et al., 2002) and goats (Zhang et al., 2008) and lysozyme transgenic cows (Yang et al., 2011) and goats

(Maga et al., 2006) exhibited antimicrobial activity and thus the potential to reduce the incidence and severity of mastitis and influence the gut microbiota composition of consumers of the milk. Indeed, work with lysozyme transgenic goats found that consumption of the milk by young pigs as a human-relevant animal model shifted the intestinal microbiota (Maga et al., 2012) and could help treat (Cooper et al., 2013) and prevent E. coli-induced diarrhea (Garas et al., 2017, 2018), demonstrating a potential human health benefit.

Applications of gene editing in livestock species used for food have focused on a wide array of traits relating to disease resistance, improved meat or fiber, food safety or quality, animal welfare, environmental adaptation, and control over reproduction (Table 2-4). Most of these applications have focused on making animals more resistant to common diseases. Porcine reproductive and respiratory syndrome virus (PRRSV) is a costly and common disease in pigs worldwide that is not well-controlled with vaccines. Pioneering work done by two groups demonstrated it was possible to delete (Burkard et al., 2017), mutate (Whitworth et al., 2016), or replace (Wells et al., 2017) a domain of the protein implicated in virus binding (exon 7 of CD163) via gene editing without disturbing the function of the protein, thus rendering pigs resistant to infection with PRRSV. Other groups have since replicated these findings for PRRSV via complete (Wang et al., 2019; Hung et al., 2022) or partial (Guo et al., 2019) removal, mutation (Yang et al., 2018; Tanihara et al., 2021) or swapping (Chen et al., 2019) of exon 7 of CD163 and other regions of the gene (Salgado et al., 2024). Other diseases of interest to the swine production industry have been addressed by either editing viral receptors to make pigs more resistant to transmissible gastroenteritis virus (TGEV) (Whitworth et al., 2019; Xu et al., 2020) and porcine epidemic diarrhea virus (Tu et al., 2019) or by targeting viral replication to make pigs less susceptible to classical swine fever virus (Xie et al., 2018; Qi et al., 2022).

Bovine viral diarrhea virus impacts the health and well-being of cattle worldwide by causing severe respiratory and intestinal illness and is highly contagious, transmissible even from a dam to her unborn calf. Gene editing was used to generate animals with reduced susceptibility to this major viral pathogen by modifying a gene to prevent virus binding (Workman et al., 2023). Gene editing has also been used to knock in a gene that encodes an antibacterial enzyme with specificity to a major cause of mastitis, a common disease affecting the health and welfare of dairy cattle (Liu et al., 2013). In addition, cattle with the ability to resist BSE (Park et al., 2020), tuberculosis (Wu et al., 2015), pneumonia caused by the bacterial pathogen Mannheimia (Pasteurella) haemolytica (Shanthalingam et al., 2016), and correction of a genetic disorder (isoleucyl-tRNA synthetase syndrome) in Japanese black cattle that impairs protein synthesis and leads to reduced prenatal growth and survival (Ikeda et al., 2017) have been reported. All of these diseases impact animal welfare, and there are several other applications of gene editing focused on improving animal welfare and well-being. These include dairy cattle that do not grow horns, thereby eliminating the painful disbudding practice (Carlson et al., 2016), as well as male pigs that do not require surgical castration that commonly is performed at a young age to prevent development of boar taint, the unpleasant smell and taste of meat from mature male pigs, as well as aggression (Flórez et al., 2023). Another application relating to the ability of cattle to adapt to their environment is coat color dilution (Laible et al., 2021; Rodriguez-Villamil et al., 2021; Cuellar et al., 2024) to alleviate heat stress and thereby improve the thermotolerance of dairy animals living in warming climates.

Gene editing has been used to modify traits to improve the yield or quality of food-animal products and breeding practices. These include multiple examples of producing cattle (Proudfoot et al., 2015), pigs (Qian et al., 2015), sheep (Proudfoot et al., 2015), and goats (Ni et al., 2014) that recapitulate a naturally occurring double-muscle phenotype, thereby producing more meat per animal to improve production efficiency. Sheep (Hu et al., 2017) and rabbits (Xu et al., 2020) with longer wool have been produced, thereby increasing yield. With respect to food quality, pigs with lower content of saturated fat (Li, M., et al., 2018) and dairy cows lacking a major human allergen (Zhou et al., 2017) have been generated, changing food product composition to improve human health. Finally, work has been done to ablate the germline in sheep (McLean et al., 2021), cattle, goats, and pigs (Park et al., 2017; Ciccarelli et al., 2020; Mueller et al., 2023) to enable novel surrogate breeding strategies to conserve or disseminate desirable genetics using germline complementation.

TABLE 2-4 Applications of Gene Transfer and Gene Editing in Mammals Used for Food

NOTES: KI = knock-in; KO = knockout; nt = nucleotide; SCNT = somatic cell nuclear transfer; ZFN = zinc finger

Poultry

Most of the meat-based protein produced globally is derived from poultry production. Most of that protein is from broiler-type chickens that have been selectively and intensively bred over the past 70 years by commercial poultry genetics companies. The resulting broiler chicken of today is a highly efficient, nutritious animal with a feed conversion rate of 1.6 that reaches market weight of 1.8 kg in 42 days. Additionally, eggs from laying chickens are a nutritious source of protein and fat, and the modern layer hen will produce around 300 eggs a year. In 2022, 70 billion broiler chickens were slaughtered, and 123.6 million tons of chicken meat and 87.0 million tons of hen eggs were produced globally (FAO, 2024), making poultry the most important farmed animal for protein production. Chickens are farmed intensively in the commercial farm setting and extensively in small-holder farms in almost all countries and climate conditions. Successful poultry farming in both settings depends upon the use of prophylactic vaccinations against the most common chicken pathogens, including Marek’s disease virus (MDV), Newcastle disease virus, infectious bursal disease virus, and avian infectious bronchitis. Avian influenza virus (AIV) is a constant global threat to poultry flocks in all countries and is a cause for concern for its recurring jumps into other food-animal species, wild animal populations, and humans (see Chapter 3). However, the use of vaccines as a control mechanism against AIV varies across countries.

Owing, to their unique reproductive biology, avian species are not readily amenable to genetic modification using the reproductive techniques that have been developed for mammalian species (Figure 2-1). Viral vectors have proven useful to introduce transgenes randomly into the chicken genome (Han and Park, 2018, Lee et al., 2020a). The founder animals are usually highly chimeric, but offspring bred from these founders have single, defined genomic integration events that are stably inherited by successive generations. As noted above, transgenic tools have been applied to poultry species, and several examples of chicken and quail have been produced with novel or improved production traits. The recombinant protein lysosomal alpha lipase is produced in chicken egg albumin and is clinically approved for treatment of Wolman syndrome in the United States and the European Union (European Medicines Agency, 2015; Shirley, 2015). A transgenic chicken demonstrating a lower propensity to transmit AIV has been developed (Lyall et al., 2011). Genome-editing technology in poultry will be useful to validate existing genetic variants that are present in commercial poultry flocks and create novel genetic changes that will be beneficial to poultry production through increased meat and egg production, lower environmental impacts, and reduced animal losses to poultry diseases.

Due to the complex structure of the avian egg, variations of the techniques developed for HGM in mammalian species are used to deliver modifications to poultry. Adenovirus-encoding CRISPR guide RNAs and Cas9 protein are delivered directly into the cleavage-stage embryo of the laid egg. Small random indels are created at the target locus that will be present in the mosaic founder animal and will be transmitted individually in each edited germ cell. The offspring will be a ratio of heterozygote edited and non-edited animals containing individual indels. Further, genome-editing tools can also be injected directly to an early-stage embryo of the laid egg via the embryonic vascular system to target migratory primordial germ cells, or, less commonly, semen delivery to generate indels in founders and their offspring (Tyack et al., 2013; Cooper et al., 2019). Both delivery methods are applicable to all bird species but have low efficiencies in generating genome-edited animals. Chickens are the only food-animal species in which techniques have been demonstrated to culture the primordial germ cells in vitro and thereby direct modification of the early embryonic germ cells (Han and Park, 2018). When returned to a surrogate host, the modified germ cells form functional gametes that transmit the HGM to the offspring.

Genome-editing techniques have been used to generate chickens with modified traits (Table 2-5). Some involve simple indels, while others entail the directed modification of a few amino acids or the targeted insertion of a transgene. As shown with other livestock animals, the disruption of the myostatin gene in quail, chicken, and ducks caused an increase in muscle mass and a reduction in overall body fat (Kim et al., 2020; Lee et al., 2020b, 2022). Sterile chickens that were modified using genome-editing tools were generated using both TALENs and CRISPR-Cas9 (Taylor et al., 2017; Ballantyne et al., 2021) with the goal that they could serve as surrogate hosts for the germ cells of other poultry breeds. Alteration of the allergenic properties of eggs would benefit the 0.1 percent of the adult human population with egg allergies, and chickens containing altered allergenicity by removal of allergenic egg-white proteins have been produced (Park et al., 2014; Oishi et al., 2016; Ezaki et al.,

TABLE 2-5 Applications of Gene Transfer and Genome Editing in Poultry

NOTES: KI = knock-in; KO = knockout; nt = nucleotide; SCNT = somatic cell nuclear transfer; ZFN = zinc finger

2023). Perhaps the biggest potential benefit of genome editing in poultry is the development of egg-laying hens that would produce only female chicks. Currently, the yearly slaughter of billions of male chicks is performed in hatcheries because male chickens do not lay eggs and are superfluous to the egg-layer industry (Doran et al., 2017). Researchers in several countries are using genome-editing technology to develop the capability to detect male chicks in the unhatched egg in order to facilitate their removal from the incubator prior to hatch. One side benefit of this technology is that only the males are genetically modified, and thus the hatched female chicks are not genome edited (Tizard et al., 2019).

As mentioned above, control of infectious diseases in poultry typically relies on vaccines, as well as biosecurity practices and culling of infected flocks. However, vaccines put immune pressure on pathogens to escape this containment and require booster vaccinations to provide adequate protection over the life of longer-lived birds. Further, vaccination of poultry at a commercial scale is logistically challenging. This has led to efforts to develop HGM poultry with disease resistance. RNA interference molecules targeting viral genomes have been reported to be successful against both MDV and AIV (Chen et al., 2009; Lyall et al., 2011). More recently, expression of Cas9 protein and sgRNAs targeting MDV integrated in the chicken genome was able to inhibit virus replication (Challagulla et al., 2021). Using CRISPR-Cas9 with the receptor KO strategy, Koslová et al. (2020) demonstrated in vitro and in vivo resistance to avian leukosis virus (subgroup J) infection through single amino acid deletion of the NHE1 receptor. Concerningly, however, follow-up reports determined that rapid viral mutations ablated host resistance, highlighting the need for multiple gene edits and surveillance for viral evolution in gene-edited animals (Matoušková et al., 2024). Finally, a chicken containing two amino acid edits to the ANP32A gene demonstrated altered susceptibility to a low-dose challenge with low-pathogenic AIV (Idoko-Akoh et al., 2023). Theoretically, chickens resistant to known and emerging infectious diseases, particularly viruses, can be developed using HGM technology.

Fishes

The contribution of aquaculture to the world supply of fish grew from 4 percent in 1970 to over 50 percent in 2020 (FAO, 2020). Increasing effort has been invested in genetic improvement of aquaculture stocks, involving both classical selective breeding and biotechnology-based methods. Fishes provide straightforward systems for biotechnological manipulation. Protocols for artificial induction of spawning exist for cultured species. Fishes have high fecundity, the eggs are relatively large, fertilization is external and easily conducted in vitro, and embryonic and larval development occur outside the mother. Egg incubation and larval rearing methods are established for cultured species. Crustaceans and mollusks are generally less amenable to biotechnological manipulation, however, as crustaceans have internal fertilization and mollusks have very small eggs. Yet, progress is also being made in applying biotechnology to these groups.

A large research effort has been devoted to the production and evaluation of transgenic fishes expressing a variety of traits (Wang et al., 2021). Accelerated growth has been pursued by transfer of growth hormone genes, with notable results in Atlantic salmon (Du et al., 1992), mud loach (Nam et al., 2001), and other species (reviewed by Hallerman et al., 2007; Devlin et al., 2020). Enhanced nonspecific disease resistance has been approached by introduction of the cecropin gene via gene transfer (Dunham et al., 2002) and the alligator cathelicidin gene by genome editing (Wang et al., 2023) into channel catfish, and the lactoferrin into grass carp cells by genome editing (Mao et al., 2004). Reproductive confinement of aquaculture stocks was approached by transferring an antisense gonadotropin releasing hormone gene into rainbow trout (Uzbekova et al., 2000), transgenic knockdown of primordial germ cells in channel catfish (Li, H., et al., 2018), and other approaches in common carp and other species (Thresher et al., 2009). A phytase transgene was introduced into medaka to promote better utilization of the major phosphorous storage compound in soybean meal (Hostetler et al., 2005). The one transgenic fish that is commercially available for consumption is the AquAdvantage GH-transgenic Atlantic salmon that was approved by the U.S. Food and Drug Administration (FDA) in 2015 (FDA-CVM, 2015). For those salmon, fertilized eggs are produced in Canada and the fish are grown and matured to market size in a recirculating aquaculture facility in Indiana. The AquAdvantage GH-transgenic Atlantic salmon is the only transgenic animal whose food products are in routine production globally.

Over the past decade, gene-editing techniques have come to replace classical gene transfer for the biotechnological manipulation of aquaculture species (Blix et al., 2021; Hallerman, 2021; Moran et al., 2024; see Table 2-6).

TABLE 2-6 Applications of Gene Transfer and Genome Editing in Aquaculture Species

NOTES: KI = knock-in; KO = knockout; nt = nucleotide; SCNT = somatic cell nuclear transfer; ZFN = zinc finger nuclease.

A leading application is knockout of the myostatin (GDF8) gene, whose product limits muscle growth, resulting in heightened filet production in channel catfish (Khalil et al., 2017), olive flounder (Kim et al., 2019), and yellow catfish (Zhang et al., 2020). Myostatin-knockout red sea bream and leptin receptor-knockout tiger puffer (Kim et al., 2019, Kishimoto et al., 2019, Ohama et al., 2020) produced by the Regional Fish Institute in Japan have been commercialized and represent the first gene-edited animals marketed for food use globally (Regional Fish Institute, 2023). Induction of sterility was approached by knockout of the luteinizing hormone gene in channel catfish (Qin et al., 2016) and the dead end gene (dnd) in Atlantic salmon (Wargelius et al., 2016); restoration of sterility in the Atlantic salmon also has been demonstrated (Güralp et al., 2020). Knockdown of dead end led to sterility in sterlet, posing the possibility of using this common sturgeon as a surrogate species for producing imperiled sturgeon species (Baloch et al., 2019). Resistance of cultured grass carp cells to reovirus infection was observed following knockout of the junction adhesion molecule-A gene (Ma et al., 2018). Coloration is an important trait for marketed fishes, and reduction of melanin by disruption of the tyrosinase gene has been achieved in goldfish (Liu et al., 2019) and large-scale loach (Xu et al., 2019).

Perspective across animal production sectors

In light of the developments discussed in the three preceding sections, the potential contribution of animal biotechnology to genetic improvement of livestock is substantial, and the goals and capabilities of these technologies are both distinct from and complementary to those of conventional selective breeding. Both classical transgenesis and genome editing offer the prospect of producing animals with unique, desirable phenotypes that improve production sustainability, efficiency, and profitability. In addition, gene editing also can expedite the introgression of a desired gene variant into a recipient line without the need for applying the traditional approach of crossbreeding and repeated backcrossing. The traditional approach reduces the performance of an otherwise high-performance line because introgressing a desired variant from a resource line requires generations of backcrossing to recover the largely intact genetic background of the high-performance line, requiring several generations of time and investment.

Genetic modification has the potential to facilitate improvement of desirable production, health, and welfare traits of animals used for food production.

CURRENT STATE OF ANIMAL BIOTECHNOLOGY

While many applications aimed at increasing the productivity and sustainability of animal agriculture are in development, only a few have been approved in the United States or other countries (Table 2-7). Some HGM animal lines (e.g., Acceligen’s SLICK cattle and Washington State University’s NANOS2 Surrogate Sire breeding program) are still in the research-and-development stage but have navigated the U.S. federal approval process. Only three HGM animals are in routine production for use as food. Among animals produced through classical transgenesis, the AquAdvantage growth hormone-transgenic Atlantic salmon is in production (AquaBounty, 2024), although only in facilities operated by the developer. Among animals produced using genome editing, only myostatin-knockout red sea bream and leptin receptor-knockout tiger puffer and olive flounder are in production in Japan in facilities operated by the Regional Fish Institute (2024). However, none of these products are under widespread production by conventional farmers.

In addition to applications in agriculture (Murray and Maga, 2016), transgenic and gene-edited agricultural animals are also developed to provide model systems for biomedical research (Rogers, 2016; Whitelaw et al., 2016; Zhang et al., 2022). Biopharmaceutical compounds produced by transgenic “biopharm” animals have received FDA approval for use as drugs. FDA granted approval to Alexion to produce sebelipase alfa (an enzyme for replacement therapy for the treatment of lysosomal acid lipase deficiency or Wolman disease) in the egg white of transgenic chickens (Mullard, 2016). FDA also approved ATryn, a recombinant antithrombin indicated for prevention of peri-operative and peri-partum thromboembolic events in hereditary antithrombin-deficient patients produced

TABLE 2-7 Regulatory Determinations for HGM Animals for Food Use by Country

SOURCE: Wray-Cahen et al., 2024.

by GTC Biotherapeutics, Inc. using transgenic goats (U.S. Food and Drug Administration, 2018a). Ruconest, a recombinant C1 esterase inhibitor indicated for treatment of acute attacks in adult and adolescent patients with hereditary angioedema is produced by Salix Pharmaceuticals, Inc. in the milk of transgenic rabbits (U.S. Food and Drug Administration, 2018b).

Several research groups and companies are developing pigs as the source of cells, tissues, and organs for xenotransplantation to human patients (Eyestone et al., 2020). FDA approved initiation of the first clinical xenotransplantation trial in 2019 using porcine skin grafts from GGTA1 knockout pigs (Holzer et al., 2019). Recent transplants of porcine kidneys into human patients showed promising results (Rabin, 2024a, b). Further, xenotransplantation of several other organs from gene-edited pigs, including heart (Rabin, 2022) and liver (Mallapaty, 2024), have been reported in human patients.

KEY FINDINGS

Conclusion 2-1: The potential contribution of animal biotechnology to genetic improvement of livestock is substantial, and both distinct from and complementary to the goals of conventional selective breeding. Both classical transgenesis and genome editing offer the prospect of producing animals with unique desirable phenotypes that improve production sustainability, efficiency, and profitability. In addition, gene editing can expedite the introgression of a desired gene variant into a founder line without the need for applying the traditional approach of crossbreeding and repeated backcrossing. Genetic modification has the potential to facilitate desirable production, health, and welfare traits of animals used for food production. Gene transfer, and especially genome editing, have considerable potential to improve the sustainability of animal production by improving efficiency and welfare of the animal, as well as providing the basis for value-added products for the consumer.

REFERENCES