3

Potential Hazards to Animals and Consumers

INTRODUCTION

Animal biotechnology offers benefits to agricultural productivity and sustainability (Chapter 2, Tables 2-4 through 2-6), including enhanced disease resistance, improved quality of animal products for consumers, and facilitated dissemination of improved genetics to the animal production industry. However, there also are concerns that it may pose hazards to the health of food animals and to the human consumer of foods derived from animals with heritable genetic modifications (HGMs). This chapter addresses biological mechanisms associated with hazards and harms posed by the development and production of HGM food animals, focusing on animal safety, food safety, and food composition. Food safety is defined as the science and practice of mitigating the risk of biological, chemical, and physical hazards in foods typically consumed by humans in order to prevent the occurrence of foodborne illnesses. Harms that potentially could arise through other risk pathways are also briefly discussed.

As explored in detail in Chapter 4, current knowledge does not support quantitative assessment of risk likelihoods associated with the hazards and harms discussed in this chapter. Only qualitative assessments of risk likelihoods (no, low, moderate, and high) are currently available. Areas needing future research effort are identified throughout this report, and research recommendations are presented in Chapter 6.

HAZARDS FROM HERITABLE GENETIC MODIFICATIONS

It is important to consider any hazards and harms that might arise through HGM animals (discussed in this chapter) and to assess their associated risks (discussed in Chapter 4). While technological advancements are improving the efficiency and fidelity of HGM procedures, unintended consequences are known to occur (Chapter 2). Regarding animal safety, the U.S. Food and Drug Administration (FDA) Center for Veterinary Medicine (FDA-CVM, 2024) has stated its concern regarding the potential for changes in animals’ physiology or behavior that may interfere with their basic functioning, cause suffering, or elevate their susceptibility to disease.

Hazards to the HGM Animal

Harms to the animal may be driven by hazards induced at the molecular level by either gene transfer or genome editing. As discussed below, possible harms to the animal include alterations that reduce fitness, increase disease susceptibility, or reduce animal well-being. These harms to the animal collectively might pose indirect hazards

to humans through consumption of those animal-derived products. Within a risk-analysis context, hazards are those HGMs that might cause harms to the animal, which include: (1) alterations at the intended or unintended genomic site(s) that change the phenotype in an undesirable manner; (2) integration of exogenous DNA such as plasmids, virus, or other genes that result in an undesirable gain of function; (3) conformational changes in the three-dimensional chromatin structure of the DNA that alter expression of genes that change the phenotype in an undesirable manner; (4) epistatic effects of the intended or unintended edits on the expression of other genes resulting in changes in traits of the phenotype; and (5) pleiotropic effects of a transgene or edit upon multiple traits, resulting in a new or undesirable phenotype. These hazards pertain to any mutation, whether natural or human-induced; the distinction is that only the latter is subject to regulatory review. The presence of unanticipated consequences or unintended alterations does not inherently present a safety concern to the HGM animal or to the human consumer of HGM animal products. Any risks would need to be assessed considering all relevant factors and available information. More information on potential harms from these hazards is discussed below and further in Chapters 4 and 5.

Intended and unintended on-target modifications

While the majority of intended genome edits are expected to be small indels of fewer than 20 base pairs, multiple studies have reported various unintended, on-target mutations, including large deletions (of a kilobase or more in size) and complex genomic rearrangements (Kosicki et al., 2018). The FDA Center for Veterinary Medicine (CVM) recently identified unintended alterations when analyzing next-generation DNA sequencing data for multiple products (Moyer, 2023). These mutations have included short indels and duplication events at sites predicted by in silico algorithms with permissive search parameters and via biochemical assays. Unintended alterations in the target area can also include unwanted integration of exogenous DNA.

Off-target modifications

Insertional mutagenesis is a hazard that is especially pertinent to classical gene transfer (microinjection of a DNA construct into a newly formed embryo with an intent of subsequent integration at a random site in the host genome, see Table 2-1), a hazard in which the introduced gene construct is not targeted to a particular genomic site, but rather is integrated at random into the host genome (Palmiter and Brinster, 1986). The insertion of a transgene is a mutagenic event that may affect any gene that happens to be near the site of integration (NRC, 2002). This raises the possibility that the transgene could become integrated within a functional element of the genome, whether into a regulatory element or a protein-encoding sequence, thereby altering or knocking out expression of a critical gene; alternatively, the transgene could become integrated into a coding sequence, rendering the gene product dysfunctional. Insertions with observable phenotypes have been well demonstrated in the mouse model system (Gridley et al., 1987). The frequency of insertional mutations in transgenic mouse lines originally appeared to be about 5-10 percent (Palmiter and Brinster, 1986), a frequency estimated based upon observable phenotypes and prenatal lethality. For example, transgenic mice selected for presence of a transgene were found to carry genetic lesions leading to limb deformity (Woychik et al., 1985) and other skeletal abnormalities (Woychik and Alagramam, 1998). Further, the National Research Council (NRC, 2002) noted that transformed embryos may have inserted DNA sequences in addition to those of an active transgene, that insertional mutagenesis processes frequently led to recessive traits that become apparent only in a homozygote, and that promoters on the transgene may activate expression of other genes in proximity to the transgene. However, the genomic insertion sites of only a small fraction of transgenic lines have been discovered and reported, due in part to limitations in the discovery tools. In subsequent work using targeted locus amplification (de Vree et al., 2014; Hottentot et al., 2017), Goodwin et al. (2019) identified transgene insertion sites of 40 highly used transgenic mouse lines and showed that randomly inserted transgenes disrupted coding sequences of endogenous genes in half of the lines, frequently involving large deletions or structural variations at the insertion site. However, while untargeted gene transfer was important historically, the approach is not likely to be the method of choice for future applications to food animals.

Genome editing using site-specific nucleases (see Chapter 2) provides the ability to introduce specific mutations at targeted genomic locations. While unintended alterations are less frequent with this approach than with

classical transgenesis, they are still an issue warranting screening and characterization in lines developed for practical use in animal production. However, these unintended edits, often termed “off-targets,” can prove difficult to distinguish from de novo mutations, which is relevant in the context of regulatory oversight, as discussed in more detail in Chapter 4. De novo mutation rates vary among species and are more common than was recognized before the era of whole-genome sequencing. A recent review of 150 species (Bergeron et al., 2023) showed mutation rates for various food-animal species, including goats (5.26 × 10^-9), carp (5.62 × 10^-9), chickens (3.63 × 10^-9), and pigs (4.32 × 10^-9). For the genome of pigs (2.8 × 10⁹ nucleotides), the number of new mutations would be expected to be approximately 12 per breeding animal. However, whole-genome sequencing showed an average of 80(±15) de novo mutations in wild-type pigs (Burger et al., 2024). As noted in Chapter 5, “trio sequencing” of parents and offspring can identify new de novo mutations, but beyond screening for the intended edit, no method is available that can discriminate between additional changes that were unintended and caused by the editing mechanism and those that occurred naturally. However, likely off-target locations can be identified by various technologies. Unintended on-target alterations, such as integration of exogenous DNA, also can be identified, as discussed in Chapter 5. For example, in humans, Jonsson et al. (2021) showed that there are, on average, over five different mutations among monozygotic twins that arise during the developmental process. FDA-CVM has observed variable genomic localization of unintended off-target alterations across individual animals (Moyer, 2023). Tsai et al. (2023) attributed large deletions (greater than 90 kilobases) at atypical non-homologous off-target sites to clustered regularly interspersed short palindromic repeats (CRISPR)-Cas-mediated genome editing. DNA strand breakpoints may occur upstream or downstream rather than at the targeted genomic cut site, as has been observed in mice (Shin et al., 2017).

The findings from both model animal and food-animal systems illustrate the importance of thorough phenotypic and genotypic characterization of HGM livestock lines during development of production lines. As noted in Chapter 4, multiple layers of risk management are embedded within the process of developing HGM food animals so that substantial, harmful integration events can be culled from HGM animal lines developed for agriculture. Regardless of what mutations are found—on- or off-target, intended or not—a healthy phenotype will prove the most reliable marker of animal well-being.

Unanticipated consequences of the intended HGM

Many genes and proteins have multiple functions, a phenomenon termed pleiotropy; as a result, multiple phenotypic impacts may stem from a single HGM. In the context of gene transfer, pleiotropies have been shown to arise from inappropriately controlled expression of a transgene, often because the promoters fused to the introduced structural gene did not respond to the normal array of homeostatic feedback controls. Growth hormone (GH) has functions in addition to growth promotion (reviewed by Hallerman et al., 2007), including stimulating appetite and intestinal Na⁺-dependent and Na⁺-independent amino acid transport. Exogenous GH administration affects tissue and serum concentrations of insulin-like growth factors IGF-1 and IGF-2. In salmonids, GH injection disrupts circulating levels of insulin and elevates T3 and hepatic 5-deiodinase activity, which converts thyroxine to T3. Through shifts in lipid, protein, and mineral metabolism, elevated levels of GH lead to changes in body conformation and composition. In fishes, GH affects osmoregulatory ability and regulates reproduction. Hence, it is not surprising that a GH transgene might affect phenotypes in addition to growth. Pursel et al. (1989) observed both beneficial and adverse consequences of long-term elevations in plasma levels of bovine GH (bGH) in two lines of transgenic pigs. Two successive generations of pigs expressing the bGH gene showed significant improvements in daily weight gain and feed efficiency and exhibited changes in carcass composition that included a marked reduction in subcutaneous fat. However, long-term elevation of GH was generally detrimental to health; the pigs had a high incidence of gastric ulcers, arthritis, cardiomegaly, dermatitis, and renal disease. The ability to produce pigs exhibiting only the beneficial, growth-promoting effects of GH by a transgenic approach may require better control of transgene expression, a different genetic background, or a modified husbandry regimen. GH genes have also been introduced as transgenes into over a dozen fish species, in many cases leading to pleiotropic effects including morphological abnormalities, physiological perturbations, and decreased performance (Hallerman et al., 2007).

Examples also exist in the context of genome editing. ANTXR1 encodes the anthrax toxin receptor 1, a putative receptor for senecavirus A, which causes vesicular disease in swine. When Chen et al. (2022) used CRISPRCas9 to knock out the gene in pigs, the HGM animals were resistant to infection by the virus, but also exhibited a disease phenotype with a variety of symptoms including partial anodontia, bilateral exophthalmia, cataracts, and reduced lifespan. These animals were not developed further for a production system because of the unexpected phenotype. In another example, CD163 and pAPN are receptors for porcine respiratory and reproductive syndrome virus and transmissible gastroenteritis virus, respectively. Double-knockout swine were resistant to both viruses, but their meat contained elevated levels of heme iron (~6 mg/kg compared to ~4 mg/kg in controls). However, this modest change was not unexpected since knockout of CD163 was known to impair hemoglobin metabolism (Xu et al., 2020).

Animal welfare issues

Another concern about generating HGMs in food animals is the potential for impacting animal welfare. For example, unintended modifications, if harmful to the animal, might ultimately manifest in the phenotype of the HGM animal, for example, as seen in myostatin-edited animals (Guo et al., 2016; Yeh et al., 2017; Matika et al., 2019). Inferring and managing the impact of HGMs on animal welfare is complex because the perception of animal welfare itself is controversial. The use of animals in agricultural systems evokes controversies over how to define animal welfare, economic concerns of producers, and the ethics of using animals in agriculture (Croney et al., 2018a). The debates increase in complexity when concerns about animal welfare are coupled with concerns about sustainability and global food security, and it is unlikely that any proposed solution will simultaneously address all the issues of concern (Croney et al., 2018b). The World Organisation for Animal Health has adopted the following definition:

Based on this definition, genome editing creates harms if it reduces animal health or causes pain, fear, or distress; however, the mental state of an animal can be a nebulous quality to access. A more workable principle for assessing the welfare of HGM animals was suggested by Rollin (1995, p. 169), who asserted that “any animals that are genetically engineered for human use should be no worse off, in terms of suffering, after the new traits are introduced into the genome than the parent stock was prior to the insertion of the new genetic material.” Again, it can be challenging to determine how to define “worse off” if social and mental aspects of animal welfare are included. Therefore, discussion is needed by the animal welfare community regarding what types of HGMs would be considered “beneficial” or at least not rendering the animal “worse off.”

Animal welfare can impact human health through its indirect effects upon qualities of consumable products such as meat, milk, and eggs. Thus, HGMs that impact the health of the animal could also pose risks of harm to human consumers. Should either intended or unintended alterations change food composition in a biologically meaningful manner (as described in Chapters 4 and 5), this may pose concern, depending on the subsequent food processing and inspection steps. For example, while poor animal welfare can increase bacterial load in milk through infections of the mammary tissue (mastitis), pasteurization kills most pathogens. Other animal welfare concerns such as injuries, morbidity, and disease are currently addressed by U.S. Department of Agriculture – Food Safety Inspection Service and Food and Drug Administration standards for harvest (Chapter 4) that require an animal be able to walk and be free of visible abnormalities of the carcass meat, lungs, and organs. Other standards apply to milk (FDA, 2024) and eggs (Table 4-1).

As discussed in Chapter 4, to maintain acceptable levels of animal welfare, most livestock production industries in the United States have developed and implemented certification programs with science-based animal care guidelines. In Chapter 6, it is noted that the degree to which these guidelines are appropriate for HGM food animals warrants directed research. While food-animal regulations apply to HGM animals at the point that they become

food animals, when they are research animals, they are subject to arguably more stringent standards imposed by the Animal Welfare Act, Institutional Care and Use Committees, the Guide for the Care and Use of Agricultural Animals in Research and Teaching (ASAS and PSA, 2020), and the Guide for the Care and Use of Laboratory Animals (NRC, 2011).

Animal welfare issues pertaining to cloning

In some gene editing or gene transfer experiments, HGM protocols are applied to cultured somatic cells, followed by screening to identify cells that have been modified successfully. Somatic cell nuclear transfer (SCNT) would then be practiced to generate whole animals from modified cultured cells; that is, the nucleus of a cultured somatic cell would be transferred into an enucleated metaphase-II oocyte for generation of a new individual, which other than the new HGM would be identical to the somatic cell donor (Chapter 2, Table 2-1). While SCNT provides a practical way of generating HGM animals with valued traits, it also poses concerns regarding the potential for shortened chromosomal telomeres, low early survival rate for cloned individuals, and large offspring syndrome (NRC, 2002). These issues have been lessened by the development of improved cloning protocols and are restricted to individuals who are themselves cloned. Further, since these traits are not considered heritable, these concerns would not be at issue for an HGM animal line that is brought forward for regulatory review for possible commercialization. Cloning of animals has been actively used in many countries including the United States (Selokar et al., 2022), and FDA has reported regulatory findings concluding that the milk and meat from cloned cattle, pigs, and goats are safe for human consumption (FDA-CVM, 2008a, b).

HAZARDS FOR CONSUMERS

A second set of concerns relates to potential hazards posed to the consumers of foods derived from HGM animals. The FDA-CVM (2024) has stated that factors that might be considered regarding food safety include the potential for elevated levels of hormones or proteins or production of novel substances that could be harmful to human health if consumed, and whether the nutritional composition of the food is altered. To set the context for a discussion of the potential for such considerations to arise in the context of HGM animal-derived products, it is reasonable to ask the question: “What hazards exist in the natural foods we eat?” The U.S. Centers for Disease Control and Prevention (CDC, 2024a) estimates that about 48 million people get sick, 128,000 are hospitalized, and 3,000 die from foodborne diseases each year in the United States. A meta-analysis of foods that were implicated in carcinogenesis (Schoenfeld and Iaonnidis, 2013) showed that only one food (olives) was never associated and two foods (bacon and salt) always were associated with that harm endpoint. Thus, in the context of discussion of food safety, almost no food is absolutely safe.

Bioactive compounds or novel substances

In some applications, the aim of gene transfer or site-directed nuclease 3-type genome editing (i.e., editing that adds a new DNA sequence to the genome of the host) is to introduce modifications that result in expression of proteins or polypeptides that will be biologically active in the HGM animal. Examples include multiple cases in which GH transgenes were incorporated into food animals with the intent of increasing growth rate (see Chapter 2). The National Research Council (NRC, 2002) noted the possibility that novel, introduced molecules could retain unwanted bioactivity after consumption of an HGM animal product, presenting a food safety concern. Bioactive compounds might exert effects in the digestive system prior to absorption, and indeed, that is the intent in some applications. For example, recombinant bile salt-stimulated lipase expressed in the milk of transgenic cattle was intended for oral administration in humans for treatment of patients suffering from pancreatitis (Wang et al., 2017). Lysozyme secreted in the milk of transgenic goats retarded spoilage of the milk and was used to promote the recovery of piglets experiencing diarrhea (Cooper et al., 2013).

The fate of the bioactive protein or polypeptide molecule during and after digestion is a key issue. Cooking will break down most complex proteins and polypeptides. During digestion, proteins and larger polypeptides are

mostly broken down into small peptide fragments and free amino acids in the digestive tract. Di- and tripeptides that are absorbed into digestive epithelial cells are broken down enzymatically into amino acids. Few intact small peptides are absorbed into the bloodstream during digestion. Degradation of the intact protein would destroy its bioactivity.

Experience with protein drug delivery suggests that there is no protein drug delivery unless the molecule is encapsulated in nanostructures designed to be absorbed and then released. However, there are reports in the literature demonstrating that intact bioactive proteins, especially allergens, can be absorbed whole by the gut. Detectable amounts (5-33 μg/L) of immunoactive bovine β-lactoglobulin were found in 18 human milk samples from 38 mothers (Jakobsson et al., 1985); once some of the mothers adopted a cow milk-free diet, the contents fell to non-detectable amounts in two mothers and to 6 μg/L in the third, and all three infants became free from colic. In another study, detectable amounts (5-800 μg/L) of bovine β-lactoglobulin were found in 93 of 232 milk samples from 25 human mothers (Axelsson et al., 1986); six mothers with allergic symptoms such as asthma, hay fever, or eczema all had detectable amounts of β-lactoglobulin in their milk. The two mothers with detectable β-lactoglobulin in all milk samples had the highest serum values, and their infants suffered from gastrointestinal symptoms, weight decline, and exanthema. β-lactoglobulin (>0.1 μg/L) and ovalbumin were also detected in breast milk in 15 of 24 healthy lactating Japanese women who consumed at least 200 mL of cow milk per day for 1 week before sampling, adding further evidence that β-lactoglobulin concentrations in breast milk are related to consumption of cow milk. Consumption of raw or cooked eggs also can lead to dose-dependent concentrations of ovalbumin in human breast milk (Palmer et al., 2005; Metcalfe et al., 2016). The allergenicity of oral protein exposure may be different from other sorts of bioactivity, as absorption is not required to activate mucosal cell receptors. However, proteome profiling has shown non-human peptides in human milk (Zhu et al., 2019), although bioactivity was not assessed in that study. More generally, bioactive peptides have been shown to be transported across the intestinal epithelium into the bloodstream, with subsequent absorption in various tissues (Caira et al., 2022); however, for most of these molecules, no concentration data for body fluids or tissues are available, and this information would be needed to assess their bioavailability and possible bioactivity.

Nawaz et al. (2019) cited evidence supporting the presence of fragments of food-derived DNA, including DNA from HGM crops, in the blood and tissues of human consumers; however, there was no evidence for HGM crop-derived DNA functioning or being expressed following transfer to gut bacteria or consumer somatic cells. There was limited evidence suggesting that plant food-derived microRNAs can survive digestion, enter the body, and affect gene expression patterns. Food safety concerns may be posed for individuals whose digestive system has been compromised by disease, injury, or advanced age when ingesting foods containing bioactive proteins or peptides (Samadi et al., 2018). In addition, allergenicity or intolerance might remain a problem for sensitive individuals, as noted below.

The first wave of gene transfer experiments involving food animals included many studies in which GH transgenes were introduced, and the potential for bioactivity of those transgenes in the consumer was considered by the National Research Council (NRC, 2002). FDA also evaluated the food safety of GH proteins when recombinant bovine GH (or somatotropin) was administered to increase milk yield in dairy cattle (Juskevich and Guyer, 1990). Among other evidence, FDA cited data showing that non-primate GH proteins or fragments of the GH molecule are not biologically active in humans, nor are insulin-like growth factors secreted by the host animal in response to GH administration. In addition, the FDA-CVM Veterinary Medicine Advisory Committee (FDA-CVM-VMAC) assessed the bioactivity of the Chinook salmon GH gene expressed by the AquAdvantage Atlantic salmon (FDA-CVM-VMAC, (FDA-CVM-VMAC, 2010), among other food safety issues. In that assessment, GH and other hormones associated with the somatotropic axis (insulin-like growth factor 1 [IGF-1], estradiol, testosterone, 17-ketotestosterone, T3, and T4) were identified as potential hazards for the consumption of AquAdvantage salmon in food, and the IGF-1 levels of the mature diploid transgenic salmon were found to exceed those of the comparator salmon by more than 10 percent. However, the FDA-CVM-VMAC concluded that even if there were increases in the amounts of these normally occurring substances, they would not likely affect any biologically meaningful interactions with human GH receptors due to interspecies differences in the ability of these substances to bind to homologous receptors in mammals or to cause physiological changes via such binding. Even if the expression of IGF-1 were present at the highest levels measured, and even if consumers of salmon ate nothing but AquAdvantage

salmon containing this likely upper-bound level of IGF-1, the margin of exposure to this component of the food products would be well within levels of exposure from other dietary sources of salmon, posing no additional risk.

The likelihood that a bioactive product poses a hazard to the consumer will vary among HGMs, foods or food products, and consumers, posing a low to moderate food safety concern. For a susceptible individual, for example, one with a compromised digestive system, such a hazard could pose severe consequences (Samadi et al., 2018; NRC, 2002). Such hazards would be expressly considered in case-by-case risk assessment (see Chapter 4).

Food composition

As noted above, regulatory authorities in the United States (FDA-CVM, 2024) and internationally (FAO and WHO, 2008) have expressed concerns about the composition of food derived from HGM animals. Case studies of changes in composition of foods derived from HGM animals, while relatively few in number, can help to inform assessments and elucidate the associated scientific issues. Carlson et al. (2016) used transcription activator-like effector nuclease-mediated genome editing to produce five polled Holstein calves, and Young et al. (2020) crossed a genome-edited bull that was homozygous for the dominant P_C Celtic POLLED allele with horned cows (pp) to obtain six heterozygous (Pcp) polled calves. The offspring did not differ in their growth, health, development, or breeding potential from control animals (Trott et al., 2022). Trott et al. (2022) also compared the nutritional composition of milk and meat from these offspring with contemporary conventional comparators and compiled compositional data on selected raw beef cuts (shoulder and rump) from global databases to document the range of naturally occurring nutritional variation in these muscle tissues. All nutrient values in the meat from genome-edited and control offspring fell within the range for beef nutrient profile data available in public databases. Milk from only one gene-edited female offspring was tested, however, and the number of comparator controls was also limited. Milk composition varied within controls, between genome-edited and control offspring, and over time. All values except sulfur percentage were within the ranges reported in the peer-reviewed literature. Both the offspring of the genome-edited bull and contemporary controls produced some milk samples that were outside of FDA’s legal standard for the identity of milk when analyzed on an individual quarter basis. However, in commercial practice, milk in final, packaged form for human consumption is mixed in bulk from a large number of animals, and the percentage of milkfat may be adjusted to meet required nutrient specifications. Chapter 4 considers other case studies of food composition involving the AquaBounty Atlantic salmon (Box 4-1), Revivicor GalSafe pig (Box 4-2), and Acceligen SLICK cattle (Box 4-3).

Attempts to demonstrate that the composition of products derived from HGM animals does not fall outside the range of “normal” composition is difficult because available nutritional profiles do not reflect the entire range of nutrient profiles possible for a given food item. Food nutrient data are currently curated and available to anyone as a “comprehensive source of food composition data with multiple distinct data types” at the U.S. Department of Agriculture (USDA) Agricultural Research Service’s FoodData Central website (USDA-ARS, 2024). Launched in 2019, the site was developed as a compendium of several databases (Fukagawa et al., 2022): Foundation Foods, Experimental Foods (USDA-ARS, 2024), Standard Reference Legacy (Haytowitz et al., 2019), the Food and Nutrient Database for Dietary Studies 2019-2020 (USDA-ARS, 2022), and the USDA Global Branded Food Products Database (Pehrsson et al., 2024).

Foundation Foods reflects nutrient values for many single-ingredient foods along with the underlying metadata reflecting details about sampling, analytical approach, and numbers of samples used for data generation. Experimental Foods reflects nutrient data from USDA-sponsored research projects collected under specific research conditions (and thus are applicable only under those circumstances). Standard Reference Legacy is a set of nutrient lists released in 2018, but is no longer updated. The Food and Nutrient Database for Dietary Studies 2019-2020 contains data derived from the Foundation Foods and Standard Reference Legacy datasets for foods and beverages described in What We Eat in America, the dietary intake component of the National Health and Nutrition Examination Survey (CDC-NCHS, 2022). The USDA Global Branded Food Products Database includes data deposited as a result of public-private partnerships for food industry branded and private-label products.

While these data are valuable resources, review of the Foundation Foods metadata reveals important knowledge gaps regarding the range of variation in food composition that exists across a single food item, particularly

for meats and other foods derived from animals. For evaluation of the impact of an HGM on nutrient composition of meat products, a defined range of “normal” nutrient composition is generally lacking or may be based on a very small number of unrelated samples. For example, a search of nutrient composition of “redfish” in the Food DataCentral databases (USDA-ARS, 2024) returns only nutrient data collected from perch in the Food and Nutrient Database for Dietary Studies database, reflecting the lack of currently available data that is specific to redfish. In fact, these same perch data are returned as the nutrient profiles in the database for any query of orange roughy, crappie, redfish, walleye, or rockfish species.

Comparative data for beef products are also limited. A search for “ribeye” returns information on beef ribeye from the Foundation Foods database and provides analytical data for energy, protein, lipid, carbohydrates, and mineral content based on data from eight beefsteak samples and lipid composition from seven samples, all purchased at one study location, although sourced from different suppliers in six cities across the United States. Further, most U.S. beef products are commonly sold under at least three quality grades reflecting the percentage of intramuscular fat or marbling—USDA Prime (≥10 percent intramuscular fat), USDA Choice (4.0-9.9 percent), and USDA Select (3.0-3.9 percent) (Savell et al., 1986; Wilson et al., 1999), but the research literature has not generally defined the range of “normal” lipid content for beef, prepared as typically consumed.

To address a need for modern nutrient compositional analyses for beef across different conditions, a Nutrient Database Improvement study was conducted (Roseland et al., 2018). This work reflected a “market basket” approach that resembled nationwide average products and began with 164 representative beef carcasses collected at seven packing plants and distributed to three universities for processing. The project provided nutrient profiles for 12 retail beef cuts across different cooking and fabrication methods. Cholesterol, fatty acid, vitamin, and mineral composition were evaluated. Results of this study showed that different cuts and preparation methods resulted in differing profiles, even within a given carcass. In describing the nutritive value of an 85-gram serving of one cut of beef in the context of recommended daily values, the authors noted that a serving would provide 29-52 percent of the recommended daily value for zinc, 20-23 percent of niacin, 19-28 percent of vitamin B6, 17-48 percent of vitamin B12, 9-16 percent of iron, and 8-32 percent of total fat based on a 2,000-calorie diet. Results of this study illustrated that absolute nutrient content for any given food item is affected by a complex array of factors, with potential impacts far greater than differences that may result from an HGM, including an intentional gene edit or an unintended effect of an edit.

While studies such as those by Roseland et al. (2018) contributed valuable and needed data, USDA scientists recognized that their Foundation Foods and Experimental Foods databases needed to move away from the market basket concept to enable curation of important metadata demonstrating the variability of nutrients in foods due to genetic variation and environmental, management, and processing conditions (Fukagawa et al., 2022). This approach to data collection and curation will be useful in supporting compositional evaluation of foods derived from HGM animals.

The existing composition data for animal-derived foods is insufficient for purposes of assessing the impacts of HGMs on food composition, an issue further discussed in Chapters 5 and 6.

Human food allergies and intolerances

Allergenicity and food intolerance are leading food safety issues worldwide. Allergenicity results from an antigen inducing an abnormal immune response (an overreaction) and is different from a normal immune response in that it does not result in a protective or prophylactic effect, but rather causes physiological function disorder or tissue damage. The immunoglobulin E (IgE) component of the immune system is implicated in the expression of allergies. Exposure to triggering antigens in foods can, in severe cases, lead to anaphylactic shock, which, if not treated with epinephrine to suppress immune response, can be life-threatening.

Food allergies are common, affecting about 3-4 percent of adults and 6-8 percent of children in the United States. Most reactions (about 90 percent) are caused by the nine most common food allergens (milk, eggs, fish, shellfish, tree nuts, peanuts, wheat, soybeans, and sesame), four of which are animal-derived allergens and all of which are required to be listed on food labels in the United States. However, almost all foods have proteins that can cause IgE-mediated allergic reactions. Food allergy is a complicated potential hazard involving the complex

and variable human immune system, influenced by other genetic and environmental factors. Sensitization is the development of IgE antibodies upon first exposure to a food protein, and the allergic reaction (elicitation) follows with subsequent exposure. The symptoms of a food allergy are varied and, as noted above, can prove life-threating in some people if not quickly treated. Food allergen exposure by skin or inhalation is thought to be a major component of sensitization, in addition to oral consumption of the allergen.

Mammalian meats are not commonly allergenic because humans tend not to be allergic to proteins that are similar to those present in the human body. The animal food allergens usually originate from organs or tissues that are not present in the human body (such as chicken eggs), are secreted (such as cow milk), or are from non-mammalian animals (fish and shellfish) that are evolutionarily different from humans and thus have more proteins that are distinct from those of humans.

An emerging mammalian meat (beef, pork, lamb meat, or organs) allergic disease, alpha-galactose syndrome (alpha-gal) (FDA, 2020), has a novel pathophysiology involving IgE antibodies to a sugar moiety, galactosealpha-1,3-galatose. Humans and some non-human primates do not have the alpha-gal sugar on their cells. The initial sensitization is caused by a bite from one of several tick species (de la Fuente et al., 2019); the tick’s saliva transfers this sugar molecule to the human, where it is recognized as an allergen. The alpha-gal allergic reaction following the next exposure to red meat occurs about 3-6 hours after ingestion, in contrast to the typical food allergy symptoms that occur soon (minutes to 2 hours) after exposure to the allergen. The identification of a Brazil nut allergen in a transgenic soybean used for human food is an example of an unanticipated allergen causing allergic disease. To improve the nutritional quality of soybeans, methionine-rich 2S albumin from the Brazil nut was introduced into transgenic soybeans. This caused a classical IgE-mediated reaction in humans with Brazil nut allergy (Nordlee et al., 1996).

The regulatory experience with HGM crops and with conventionally bred food animals contribute the scientific justification for allergen risk assessment for foods derived from HGM food animals. Chapter 5 reviews the methods for allergenicity testing and identifies knowledge gaps needing further research, which are summarized as future research needs in Chapter 6.

An allergen risk-assessment workflow process for HGM animals intended for human consumption is needed. Elements to consider should include the source of the transgene, bioinformatic screening for known allergenic proteins, possibly a pepsinolysis protein digestion assay (see Chapter 5), and human serum reactivity testing (see Codex Alimentarius, FAO and WHO, 2008). The possibility that frameshifts induced by non-homologous end joining can result in substantial runs of new amino acids can be assessed computationally for predicted allergenicity. Resistance to digestion by the enzyme pepsin is characteristic of many allergenic proteins and has been used in laboratory testing to determine stability of the protein during human digestion (Astwood et al., 1996). However, recent evidence does not support this as a reliable method to determine risk and the European Food Safety Authority Panel on Genetically Modified Organisms (EFSA GMO Panel, 2021) recommended that this assay should be validated or discontinued, a point discussed in Chapter 6. All available evidence from allergenicity testing should be considered for the assessment of risk for allergenicity from HGM products with the potential to enter the human food supply (Metcalfe et al., 1996).

Proteins other than those specifically encoded by a transgene may change in abundance when an HGM animal is produced. This becomes especially important when regulatory proteins are introduced, up-regulated, or down-regulated. If the HGM animal is itself allergenic (e.g., shellfish), there is a possibility that changes in the amounts of allergenic proteins may occur. Proteomics (see Chapter 5) represents a useful approach to describe relative amounts of specific proteins in two similar samples. Depending on the workflow, known allergens can be specifically quantified or any protein that changes significantly can be identified via proteomics. Moreover, untargeted proteomics can be used for relative quantitation, providing additional information regarding the transgene product (e.g., sequence verification, modification, truncation), as well as information that can be used for toxicology and nutrition assessments. In conducting analyses of allergenicity, the key question is whether the HGM animal product falls outside the normal expected range in terms of the presence of allergenic proteins (Johnson, 2024).

A second possible hazard related to food safety is food intolerance. While often confused in popular discussion, food allergy and food intolerance are the outcome of distinct mechanisms. Food allergies occur when the body has an immune response to certain foods and can prove life-threatening. Non-allergic hypersensitivity reactions,

or digestive intolerances, in sensitive individuals may result from food components such as gluten; fermentable oligosaccharides, disaccharides, monosaccharides and polyols; histamine or other vasoactive amines; sulfites; and food additives (Tuck et al., 2019; Gargano et al., 2021). Food intolerances occur when the digestive system cannot break down certain foods—leading to gas and bloating, diarrhea, heartburn, nausea, or stomach pain—but are not life-threatening. For example, lactose intolerance is the result of lactase deficiency, which prevents the body from properly breaking down lactose, a milk sugar. Food intolerance affects an estimated 20 percent of the U.S. population (Tuck et al., 2019). Determining the cause of food intolerance or sensitivity is not a well-developed science. Other than the hydrogen breath test to diagnose lactose intolerance, there are no proven skin or blood tests for these issues (Cleveland Clinic, 2022). In addition to FDA-required labeling of the nine established allergens, the USDA Food Safety Inspection Service also recommends voluntary statements on food labels for the presence of allergens and additional components that inform consumers with sensitivities or intolerances, for example, “contains milk, whey (from milk), soy” (Post et al., 2007). In this case, consumer safety is protected by ingredient lists and existing food composition data.

There is no record of any HGM animals that can pose novel food allergen or intolerance endpoints, although there are some examples from crops. Transgenic soybeans containing methionine-rich 2S albumin from Brazil nut elicited a classical IgE-mediated reaction in humans with Brazil nut allergy (Nordlee et al., 1996). Another example is StarLink corn created by Aventis. StarLink corn included a genetic modification incorporating the protein Cry9C from the soil bacterium Bacillus thuringiensis (Bt), which was intended to kill caterpillars. In 1998, the U.S. Environmental Protection Agency gave limited approval for StarLink corn for use as animal feed. The Bt corn was not approved for human consumption because the Cry9C gene has two significant characteristics of known allergens (not being broken down by gastric juices or by heat). However, the corn was later found in the human food supply. That finding proved a major test for U.S. regulatory agencies and had many outcomes, including FDA recalls of over 300 food products that contained StarLink corn, including taco shells, French fries, and beer. The value of the U.S. corn crop decreased by an estimated 7 percent for at least a year, and exports to major trading partners such as Japan and South Korea declined. The incident led to lawsuits and congressional hearings. Although allergenicity tests did not substantiate the occurrence of allergenic response to the transgene product, a U.S. District Court judge in 2003 approved a $110 million nationwide settlement. Aventis voluntarily withdrew its license for the variety, which is no longer grown (Segarra and Rawson, 2001). These crop examples highlight the importance of applying allergenicity testing and regulatory processes to correctly identify and respond to potential harms posed by the genetic modification of products in the human food supply. The issue of whether our tools are sufficient to detect and predict allergenicity and thereby to anticipate and predict risk is considered in more detail in Chapter 5.

As noted above, levels of allergenicity- or intolerance-related compounds in foods derived from an HGM animal would be expected to be within the normal range of its equivalent product from a non-modified animal. Within this context, there are several different situations in which HGM of food animals may affect derived food products:

1. Alteration to a gene expressed in a tissue that is not typically consumed (e.g., creation of polled cattle). In this case, the altered gene would not reasonably be expected to result in altered food products (e.g., milk, meat) from these animals. This type of modification should be deemed to pose no risk and evaluated to confirm that no unanticipated genetic alterations occur in the animals.
2. Alteration to a gene expressed in a tissue that is consumed, but not known to be associated with food-related allergies or hypersensitivities (e.g., knockout of myostatin in a meat-producing animal). In this case, since myostatin occurs naturally in muscle tissue prior to its processing to meat, the product intended for human consumption should be analyzed to determine whether its composition is within the normal range of the equivalent non-modified food product.
3. Alteration to a gene expressed in a tissue that is consumed and is known to be associated with allergic response or food hypersensitivity (e.g., GalSafe pigs, which do not produce alpha-gal, a sugar molecule that can trigger a rare allergy; FDA, 2020). In this case, testing to verify any changes to allergenicity would be required. As part of its assessment of intentional gene alterations as a “drug,” FDA assesses effectiveness; in this case, the assessment of effectiveness of reduced allergenicity would need to include assessment of allergen content.

4. Alteration to a gene expressed in a tissue that is typically consumed but not typically expressed in that species (e.g., expression of alligator cathelicidin in fish). In this case, the product of the exogenous gene may need to be tested for allergenicity and be evaluated to determine whether additional labeling of food items might be required.

In cases where a known allergen is likely to be produced in an HGM food animal, the products intended for human consumption should be assessed for food allergenicity in the same manner as they are already evaluated by FDA in non-HGM animal products (requiring data to be provided on evidence of IgE-mediated food allergy, prevalence of an IgE-mediated food allergy in the U.S. population, severity of IgE-mediated food allergic reactions, and allergenic potency). Similarly, altered expression of genes known to contribute to food sensitivities (e.g., egg proteins, whey, vasoactive amines) should be evaluated to determine whether additional labeling of food items might be required.

Toxicity

Many plants protect themselves from herbivory by expressing toxic compounds (Mithöfer and Boland, 2012; Fürstenberg-Hägg et al., 2013; Chaudhary et al., 2018). In contrast, animals do not generally protect themselves from predation by expressing toxic compounds, with some notable exceptions. Some insects protect themselves from predation by sequestering toxic compounds from plants that they consume (Beran and Petschenka, 2022); for example, monarch butterflies sequester cardenolide toxins from milkweed (Petschenka and Agrawal, 2015). Such processes are rare, but not unknown, in vertebrates consumed as food. For example, shellfish sequester toxins from red tide algae upon which they feed (Bricelj and Shumway, 1998) and pufferfishes sequester tetrodotoxin and related compounds originating from bacteria and bioaccumulated up the food chain (Veeruraj et al., 2016). Some birds also sequester toxins (Ligabue-Braun and Carlini, 2015); examples in North America include the ruffed grouse, which acquires andromedotoxin from mountain laurel.

Toxins are well studied compounds, and genes for toxins are unlikely to be purposefully transferred into a food animal. Further, proteins are generally broken down during the processes of cooking and digestion into small polypeptides and amino acids; hence, direct toxicity of the proteins would be unusual (NRC, 2002). Any transferred proteins that would remain intact or otherwise pose a food safety concern would be evaluated in the pre-market food safety evaluation and regulatory review processes. The likelihood of toxins entering the human food supply is low.

OTHER POTENTIAL HAZARDS

Beyond hazards to the HGM animal, the production system, and the consumer of derived products, other potential hazards might reasonably be anticipated to occur.

Hazards related to infectious disease and disease resistance

Advances in modern animal husbandry have increased production efficiencies. However, outbreaks of animal infectious diseases represent one of the most significant risks associated with intensive farming. Zoonotic and foreign animal diseases pose a significant threat to human health, animal welfare, the economy, and global food security. To mitigate the risk of disease outbreaks, on-farm biosecurity measures, vaccination programs, and stamping-out policies have been implemented.

The use of genome editing to establish disease resistance in an animal represents a significant breakthrough in molecular technology. The overwhelming majority of current disease-resistant animals are produced to combat viral pathogens. Viral diseases are ideal targets for genome editing due to their well-defined genetic interactions with their hosts. These interactions include viral dependence on attachment receptors to enter host cells, dependence on host machinery for virus replication and release, and the significant economic benefit that resistance to the

virus would impart. Because viruses have an absolute dependence on their host for progeny succession, genome editing is well suited to producing animals for disease resistance.

Perhaps the most easily conceptualized approach is receptor knockout. As viruses have evolved through time to attach to specific molecular receptors found in a host, removal or editing of that receptor has been demonstrated to render the animal resistant to infection. Numerous examples in farmed animals have already been demonstrated, as discussed in Chapter 2. Another approach is to provide wide-spectrum resilience to bacteria through the expression of antimicrobial peptides; one example is the development of transgenic channel catfish that express cecropin (Dunham et al., 2002).

In the absence of a known virus receptor, alternative strategies can be employed to engineer animals with enhanced resilience to disease. For example, the receptor for avian influenza virus (AIV) has not yet been fully characterized but appears to be dependent on carbohydrate moieties. Separate approaches have been described in chickens that enhance resilience against this zoonotic disease. In the first instance, expression of RNA interference RNA decoy molecules targeting the viral genome resulted in chickens that did not transmit AIV, but unfortunately still succumbed to infection (Lyall et al., 2011). In a different approach, the knockout of host ANP32A, a protein involved with viral polymerase activity, resulted in animals that were resistant to a low-dose low-pathogenic AIV challenge, but complete resistance was not achieved at increased doses (Idoko-Akoh et al., 2023). The significance of AIV that makes it a particularly attractive target for developing disease-resistant animals is outlined in Box 3-1.

In the absence of a vaccine for high-pathogenicity avian influenza virus in poultry (which has not been approved for use in the United States), genome-editing approaches offer a new strategy for potentially controlling infection and limiting transmission to other birds and mammals (Looi et al., 2018; Idoko-Akoh et al., 2023; Kapczynski, 2024b). However, this possibility also raises the question of whether genome-edited animals with disease resistance or resilience could create a novel host pool within which AIV could evolve to become more transmissible or pathogenic to birds or mammals.

Influenza viruses are known for their rapid evolution, primarily driven by two mechanisms: antigenic drift and antigenic shift. Antigenic drift involves the gradual accumulation of small genetic changes. In contrast, antigenic shift occurs when two or more influenza viruses co-infect a host, leading to the reassortment of genetic material and the creation of new, potentially pandemic-causing strains. The consequences of this type of reassortment include the three major human pandemics in the 20th century (H1N1, H2N2, and H3N2), for which all of the implicated viruses contained viral segments of avian origin (Sharif, 2024). It is crucial to understand how animals that are resistant or resilient to influenza virus would impact viral evolution. By carefully monitoring these animals, valuable insights into the potential effects of genetic interventions on viral evolution and the emergence of novel viral variants can be gained. With current knowledge, it is difficult to assess the likelihood that avian influenza-resistant chickens would cause enhanced virus expansion into human populations.

When a virus has zoonotic potential, utilizing a single target or mechanism to generate disease resistance poses the risk that a pathogen may evolve to escape from control, with the potential that the mutant may be more pathogenic to the animal or to humans. For zoonotic viruses, disease resilience in the HGM animal is not a viable approach, since it may encourage generation of a reservoir with close interaction with the human population through agricultural workers. The degree to which a genetic modification protects the host from disease will affect the level of risk posed by subsequent evolution of the pathogen. That is, if HGM-induced resistance to pathogen infection or replication in the HGM animal is complete, then there is very little chance that the pathogen will adapt as there will be no de novo pathogens produced in the system. However, if resistance is incomplete, then pathogens have the potential to evolve, with the associated potential that virulence or tropism may also change. Existing host range is also a significant factor. Pathogens with a broad host range are already able to adapt to varied cellular environments, and thus may be better able to adapt to HGM animals than pathogens that infect a single species or group of related species. There is also a plausible pathway to harm if disease resistance in the HGM animal is achieved by changing an endogenous protein such that it becomes more similar to a homolog from another species (including humans) which could drive pathogen evolution to include altered host tropism. Assessment of the risks posed by disease-resistant HGM livestock will require directed research into the evolution of the pathogen in both animal and human populations (see Chapter 6). A holistic strategy for managing the hazards posed by avian influenza virus may require both vaccination of animals and humans and development of AIV-resistant poultry (Looi et al., 2018).

In summary, while development of disease-resistant or disease-resilient animals offers the potential to reduce losses to disease, it may also allow the pathogen to evolve against the barrier created. This may potentially create a risk of harm to animals and humans if that evolution increases host tropism or allows the virus to evade antiviral treatments.

Transfer of antibiotic resistance

Unintended promotion of antibiotic resistance is a public health concern (Manyi-Loh et al., 2018) that has been associated with the production of HGM crops (Goldstein et al., 2005) and animals (Kleter and Kuiper, 2002). Antibiotic-resistance genes are often included in expression vectors (often plasmids) utilized in classical transgenesis. Expression of antibiotic resistance is useful for selecting for plasmid-bearing E. coli cells when a large quantity of expression vectors is prepared in anticipation of a gene transfer experiment. Antibiotic resistance also is useful for selecting for transformed cultured cells in anticipation of an SCNT-based production of an HGM animal. There have been reports of template plasmid integration at the target site in genome-editing experiments, for example, with zinc finger nucleases in cultured cells and with CRISPR-Cas9 procedures carried out upon the nematode Caenorhabditis elegans (Dickinson et al., 2013), zebrafish (Gutierrez-Triana et al., 2018), and mice (Skryabin et al., 2020). This raises the concern that heightened frequency of expression of resistance to the antibiotic would select for heightened frequency of resistance to that antibiotic in pathogenic microbes that could be shed from the HGM animal into the environment, such that the antibiotic would become ineffective for treatment of human or animal disease. Well-designed experiments would excise the transgene construct (the promoter, structural gene, and poly-A tail) from the plasmid vector before use or would use another sort of selectable marker, while poor purification or omission of that step could lead to retention of the antibiotic-resistance gene and its introduction into the genome of an HGM animal.

Vector integration errors may be underreported or overlooked (Norris et al., 2020). Utilization of appropriate molecular screening methods would enable detection of plasmid integration and integration of multiple template copies (see Chapter 5). Researchers are cognizant of the issue of antibiotic resistance in the design and execution of current experiments producing new lines of HGM animals. Selectable markers other than antibiotic resistance are now preferred, and a directed search can be used to detect antibiotic-resistance genes in whole-genome sequences of early-generation HGM animals. Moreover, scientists now have access to approaches that do not require use of plasmids with selectable markers to edit genome sequences. As noted in Chapter 4, such individuals would be removed from further development of HGM lines and any hazard to animal or human health would thereby be addressed.

Mobilization of viruses from animals used for xenotransplantation

As noted in Chapter 2, transgenic and genome-edited animals, some lines with multiple transgenes and edits, have been produced for purposes of xenotransplantation of organs, tissues, or cells to human patients who are unable to secure transplants from human sources. Despite the fact that humans have larger genetic differences from pigs than from non-human primates, pigs have numerous advantages as xenotransplant donor animals (Denner and Tönjes, 2012). Xenotransplantations of porcine pancreatic islets, hearts, and kidneys to human patients have been reported (Rabin, 2023, 2024a, b), and the development of genetically modified pigs designed to elicit decreased host immune responses, along with advances in immunosuppressive strategies, have resulted in prolonged xenograft survivals (Fishman, 2019) of greater than 2 weeks, long enough that at least one patient was able to return home after xenotransplantation (Rabin, 2024a). Xenotransplantation promises huge benefits by creating a large supply of organs, tissues, and cells where shortages exist, with a great potential benefit to human patients. While pigs intended for xenotransplantation would be produced under strict bioconfinement, and screening for infectious viruses would be practiced, the intimate contact between porcine cells and the human body poses the hazard that xenotransplantation may be associated with transmission of potentially zoonotic porcine viruses (Denner and Tönjes, 2012; Denner, 2022). Viruses including porcine endogenous retroviruses (PERVs), herpesviruses, hepatitis E virus, and porcine circoviruses are of concern (Mao et al., 2023). Of special interest are the roughly 60 PERVs that are integrated in the genome of all pigs, which can infect human cells. Because they are integrated into the porcine genome, these viruses cannot be eliminated by bioconfinement practices. PERVs have not, however, been shown to cause disease or even viremia in pigs. In the first clinical trials treating diabetic patients with pig islet cells, no porcine viruses were transmitted. However, porcine cytomegalovirus was transmitted in the first pig heart transplantation to a human patient and possibly contributed to the death of the patient (Denner, 2022; Rabin, 2022). In that case, while the virus was detected in the recipient, it remains unclear whether porcine cytomegalovirus can infect primate cells, including human cells (Burwitz et al., 2016). PERVs were not transmitted in clinical, preclinical, or infection experiments.

Cross-species PERV or other viral infection would require a series of events (NRC, 2002), that is, expression of infectious virus, localized infection of host cells, spreading infection in the host, persistent viremia, and expression of disease. To pose a risk to society, such an infection would require transmission to close contacts and epidemic transmission. Each step is associated with low, but yet unquantified likelihood. To prevent transmission of porcine viruses to human patients, Denner (2022) developed highly sensitive PCR-based, immunological, and other methods for detection of xenotransplantation-relevant viruses for screening of donor pigs and xenotransplant recipients. Diagnostic and control measures for porcine viruses include methods for detection of genomic integration sites, vaccines, RNA interference, antiviral pigs, farm biosecurity, and drugs (Mao et al., 2023). Research needs pertaining to assessing risk posed by xenotransplantation are discussed in Chapter 6.

Horizontal gene transfer

Humans and animals are constantly exposed to foreign DNA from a broad range of food and feed sources, whether from wild-type, domesticated, or HGM organisms. Until a few years ago, it was assumed that ingested DNA is completely degraded in the digestive tract of humans and animals. However, Rizzi et al. (2012) showed

that a small percentage of ingested DNA could survive digestion, although the DNA was limited to small fragments of approximately 200 base pairs by the end of the digestive process. Moreover, there was no finding of horizontal gene transfer (HGT) in the animal or even among bacteria in the gut. The risk for uptake by the animal was the same for control and HGM-derived foods. With the global commercialization of HGM food and feed products, however, there has been a renewed interest in the fate and effects of HGM-derived extracellular DNA in the body of the consumer (Van Eenennaam and Young, 2014). This has led to increased interest in understanding the processes involved in degradability, stability, mutagenic potential, and expressibility of extracellular food-derived DNA (Nawaz et al., 2019). The key concern is HGT, the stable transfer of genetic material from one organism to another without reproduction or human intervention (Keese, 2008). HGT occurs by passage of donor genetic material across cellular boundaries by conjugation, transformation, transduction, or other diverse mechanisms of DNA and RNA uptake, followed by heritable incorporation into the genome of the recipient organism. While the genome of almost every organism reveals the signature of numerous ancient HGT events (Boto, 2014; Soucy et al., 2015), most commonly involving the transmission of genes from viruses or mobile genetic elements, the presence of these sequences is evidence of ancient cellular infections and not the legacy of consumption of a food product.

HGT first became an issue of public concern in the 1970s when it was recognized that antibiotic-resistance genes naturally spread among pathogenic bacteria, and concerns were renewed more recently in the context of the commercial production of genetically modified crops (Dröge et al., 1998). The frequency of HGT from plants to other eukaryotes or prokaryotes is extremely low, however, lower than background rates (Keese, 2008). Nawaz et al. (2019) reviewed the fate of DNA derived from HGM plant foods in the human body. Despite intense research effort, no study reported integration of transgenes from HGM crops into the genomes of either animal or human consumers. On the basis of its understanding of the process required for HGT from plants to animals and data on genetically engineered (GE) organisms, the National Academies of Sciences, Engineering, and Medicine (NASEM, 2016) concluded that HGT from GE crops or non-GE crops to humans is highly unlikely and does not pose a health risk.

Mosley et al. (2020) evaluated the potential for transfer of a transgene from transgenic to non-transgenic pigs during parturition, mating, gestation, or lactation; their results showed no HGT during these processes. No empirical studies have identified any incidence of HGT of DNA from conventional or HGM animal-derived foods to consumers. The possibility of HGT has been suggested for transgenic animals that were transformed using transposon-based vectors (NRC, 2002). Transposons and related mobile genetic elements have been observed at thousands of copies per genome in a wide range of eukaryotes, including mammals, birds, and humans (Boto, 2014; Soucy et al., 2015). The concern is the possibility of HGT occurring via transposition among diverse hosts, leading to insertion into functional genes and thereby causing unexpected genetic damage (NRC, 2002). There is also concern that incomplete digestion of genetically modified food products in the gastrointestinal tract could result in the horizontal transfer of genes to the microflora and somatic cells of the intestine (Deepa, 2015; Ghimire et al., 2023). Focusing on the multicellular host, the greater concern would be transfer to germ cells; however, these are less accessible than somatic cells (Keese, 2008). The possibility of intergenerational transmission of any recombinant DNA would be insignificant because the compound probability of DNA reaching germ cells and becoming incorporated into the genome with a suitable promoter is even lower (Keese, 2008). The National Research Council (NRC, 2002) noted that the hazard of HGT can be addressed by not including transposase in transgene constructs so that once the construct is inserted into the host genome, the element is immobilized. HGT would not be at issue for genome-editing-mediated HGM. To date, no study has conclusively demonstrated the transfer of recombinant DNA from recombinant DNA plants to naturally occurring bacterial or host cells in the gastrointestinal tract of various mammals (Rizzi et al., 2012). Based on current understanding of the process required for HGT from food items to animals and empirical data examining the possibility of HGT from HGM organisms, horizontal gene transfer from consumed food products to the germline lacks plausibility and scientific credibility and does not pose a health risk.

KEY FINDINGS

While HGM of livestock poses benefits of increased productivity and food system resiliency, there are also concerns regarding hazards to the health of livestock and to the human consumer of foods derived from HGM animals.

1. The presence of unanticipated molecular consequences or unintended alterations does not inherently present an animal safety concern; any risks would be assessed considering all relevant factors and available information. Products that might pose food safety concern would be evaluated in pre-market food safety evaluation and regulatory review. The likelihood that a bioactive product poses a hazard to the consumer will vary among HGMs, foods, and consumers, posing a low to moderate level of food safety concern.
2. The regulatory experiences with HGM crops and conventionally bred food animals contribute the scientific justification for allergen risk assessment for products derived from HGM food animals. In conducting analyses of allergenicity, the key question is whether the HGM animal product falls outside the normal expected range in terms of the presence of allergenic proteins. Although there are some examples from HGM crops, no HGM animals have shown to be posing risk of a novel food allergen or intolerance endpoints. The crop examples highlight how allergenicity testing and regulatory processes correctly identified and responded to potential harms posed by the genetic modification of products in the human food supply.
3. The likelihood of toxins being expressed by HGM animals is low. Researchers are cognizant of the issue of antibiotic resistance in the design and execution of current experiments producing new lines of HGM animals, and other selectable markers are now used.
4. With current knowledge, the possibility of heightened risk of disease spillover from disease-resistant HGM animals into human populations presents a difficult set of risks to assess. Assessment of such risk will require research into the evolution of pathogens in livestock populations. Managing the risks posed by livestock pathogens includes biosecurity, vaccination, and use of resistant lines. Based on current understanding of the process required for horizontal gene transfer from food items to animals and empirical data examining the possibility of HGT from HGM organisms, horizontal gene transfer from consumed food products to the germline lacks plausibility and scientific credibility and does not pose a health risk.

REFERENCES