6

Scientific Questions to Be Addressed

INTRODUCTION

As noted in previous chapters, the development of livestock with heritable genetic modifications (HGMs) offers the potential for enhancing the productivity and sustainability of animal agriculture and for producing novel consumer products. However, methods for producing HGM livestock are still under refinement, and there are gaps in the state of knowledge of the potential effects of application of HGM techniques upon food animals and subsequent effects upon derived foods and the consumer that still need to be resolved. Addressing such knowledge gaps will require additional research that might be undertaken by the U.S. Department of Health and Human Services, the National Institutes of Health, the U.S. Department of Agriculture, other federal government agencies, and non-governmental entities over the next 3-10 years. This chapter identifies knowledge gaps and suggests areas for targeted research and consumer outreach. Following the general timeline and workflow typically used in the development of HGM animals, the chapter discusses issues and concerns in terms of risk management strategies that correspond to Hazard Analysis and Critical Control Points in the development process.

GENOME-EDITING TOOLS

Licensing of CRISPR-Cas9 Technology

CRISPR-Cas9 gene editing technology, the most commonly used clustered regularly interspersed short palindromic repeats (CRISPR)-Cas system, is a patented invention, so any commercial use or application requires a license. Licenses are available from ERS Genomics (Dublin, Ireland) in a wide range of fields,¹ including agriculture and livestock applications. For genome editing of livestock to prove viable, licensing of CRISPR technology must be cost-effective; however, the committee determined that in many cases a CRISPR-Cas9 license may be prohibitively expensive in this context. There are some exceptions when the technology is used for academic or non-profit research. For example, Wageningen University and Research in the Netherlands offers free access to its CRISPR-Cas9 genome-editing technology for non-profit organizations for non-commercial applications in food and agriculture. Similarly, the Broad Institute of MIT and Harvard specifies that no license is necessary to use its

___________________

¹ See https://ersgenomics.com/licensing/.

CRISPR tools, knowledge, methods, and other intellectual property for academic and non-profit research use.² However, Michael Arciero, vice president of intellectual property and corporate development at ERS Genomics, in 2023 said that a Broad Institute license may not be sufficient.³ There has been confusion regarding CRISPR-Cas9 patents in the United States following a high-profile patent interference lawsuit between the Broad Institute and the University of California, Berkeley, the University of Vienna, and Emmanuelle Charpentier, collectively known as CVC. Arciero writes that researchers based in the United States who use CRISPR-Cas9 need a CVC license and may also need a license from the Broad Institute if they work in eukaryotic cells. CVC has over 100 different patents covering different aspects of genome-editing technology, from the single guide (sg) RNA to modifications in trans-activating CRISPR RNA. While ERS Genomics makes the full suite of CVC patents available, additional licenses may be required for the use of CRISPR-Cas9, particularly for use in eukaryotic cells or for certain niche applications. The uncertainty and costs associated with licensing of CRISPR tools can diminish interest in genome editing of livestock, create confusion for both academic and commercial researchers, and impede commercialization. For example, the U.S. Department of Agriculture (USDA) Agricultural Research Service does not currently have a license and needs to partner with industry in order to enter the commercial field. If a U.S. academic institution aims to commercialize an HGM food animal, the costs of procuring CRISPR license rights could pose a major financial barrier and limit the ability to generate HGM animals and conduct research with them to a stage where the animals prove of value to commercial partners. For small start-up companies entering this field, the financial costs could be prohibitive and challenging to enumerate (e.g., spanning not just licensing and research costs but also contractual agreements, legal negotiations, and royalty agreements, among others). For commercial livestock applications, several layers of fees for the use of CRISPR technology are usually negotiated, including an initial fee (typically between $50,000-$100,000, negotiable depending upon the size of the business) and milestone-linked fees, such as achieving regulatory approval (which could be associated with fees of up to $1 million, negotiable depending upon the size of the business) or the first commercial sale. After that, a company may pay royalties on each unit sold, where the unit could be a straw of semen, an embryo, or an animal. The royalty fees are always negotiated and depend upon the business model, expected volumes, and other factors. Royalty fees typically range between 1-3 percent of the price of the trait achieved using licensed CRISPR technology. In the end, the company has to pay a lot of money for the use of CRISPR. All these fees are significantly higher for seed companies and even more so for pharmaceutical companies due to their high profitability. When CRISPR technology first became available, companies sought exclusive rights, which provided a significant competitive advantage. However, the cost of exclusive licenses is significantly higher than that of non-exclusive licenses, and most companies have shifted toward negotiating non-exclusive licenses. The rapid development of different CRISPR-based tools has also made more options available, and their developers are offering more competitive licensing fees. Profluent, for example, recently developed a CRISPR nuclease with the help of artificial intelligence that the company offers at a lower cost than most other CRISPR technologies through its OpenCRISPR™ license.

Further complicating this picture, on September 20, 2024, Emmanuelle Charpentier and Jennifer Doudna, who were jointly awarded the 2020 Nobel Prize in chemistry for developing CRISPR, asked to cancel two of their own seminal patents (Regalado, 2024). This decision could affect who gets to collect the lucrative licensing fees on using the technology. The request to withdraw the pair of European patents came after a decision from a European technical appeals board, which ruled that Charpentier and Doudna’s earliest patent filing did not explain CRISPR well enough for other scientists to use it and therefore does not count as a proper invention. The cancellation of the European patents will affect a broad network of biotechnology companies that have bought and sold rights as they seek to achieve commercial exclusivity to new medical treatments or the right to pursue genetic modification research without doubts about who really owns the technique (Regalado, 2024). These companies include Editas Medicine, allied with the Broad Institute of MIT and Harvard; Caribou Biosciences and Intellia Therapeutics in the United States, both cofounded by Doudna; and Charpentier’s companies, CRISPR Therapeutics and ERS Genomics.

___________________

² See https://www.broadinstitute.org/partnerships/office-strategic-alliances-and-partnering/information-about-licensing-crispr-genome-edi.

³ See https://www.labiotech.eu/expert-advice/five-things-crispr-cas9-license/.

Development of Precise and Efficient Gene-Editing Tools

The design phase for an HGM is the first opportunity and critical control point at which risk can be recognized, managed, and reduced. While revolutionary in their time, the tools of classical transgenesis were very inefficient; further, there was little to no control of the sites where transgene constructs integrated into the host genome or of the integrity and number of transgenes that were introduced (Volobueva et al., 2019; Wallace et al., 2000). Such inefficiency and non-specificity limited the practical utility of classical genetic engineering tools, especially for large livestock species.

The advent of genome editors enabled advances in the efficiency and specificity of methods for developing HGMs. In particular, CRISPR-Cas9 is a class of genome editors comprising a single guide RNA (sgRNA, comprising the CRISPR RNA and tRNA) and the Cas9 protein. Guided by base-pairing between the guide RNA (gRNA) and host DNA adjacent to a protospacer-adjacent motif, Cas9 can create a double-strand break at a desired genomic position (Guo et al., 2023). The use of computational tools, evaluation of target site cleavage efficiency, and use of high-fidelity Cas9 molecules increases on-target efficacy (yielding high sensitivity) and minimizes off-target effects (leading to high specificity) (Liu et al., 2020a,b). Improved genome-editing tools are being developed (Liu, 2024). Improvement of editing tools to increase precision of on-target, intended edits and to reduce the frequency of unintended edits is the first step in a comprehensive multi-step risk management strategy that will ultimately ensure both animal and human health, as well as food safety (Chapter 3). These improvements may include CRISPR-Cas9 editors that use different variants of Cas9 with improved on-target specificity (e.g., SaCas9, eSpCas9, HFCas9, HypaCas9, and Sniper Cas9), as well as other CRISPR systems (e.g., Cas12a, Cas13, base editing, Prime Editing, programmable addition via site-specific targeting elements, CAST, and Cascade). Moving forward, sgRNAs can be engineered to enhance genome-editing fidelity, and the specificity of Cas9 activity can be enhanced by extending or truncating sgRNAs (Chen et al., 2023). As the technology advances, so do the efficiency and specificity of DNA modifications, leading to fewer or (theoretically) no unintended edits. Artificial intelligence tools may also be used to refine the precision of the target identification, improve the design of guide RNA molecules, and thereby reduce the potential for unintended edits (Wessels et al., 2024).

Cas9-mediated double-strand break (DSB) generation is the main source of CRISPR-Cas9 unintended alterations. New versions of gene editors that do not create DSBs usually exhibit greater specificity of genome editing. For instance, base editors such as the adenine base editor can catalyze editing from adenine to guanine, and the cytosine base editor can catalyze editing from cytosine to thymine (Chapter 2). While base and prime editors do not require a double-strand DNA break, these editors can also create new types of unintended alterations such as RNA editing in a gRNA-dependent and also in a random gRNA-independent manner (Jin et al., 2019; Zuo et al., 2019).

With the growing application of CRISPR-Cas systems to livestock, there remains a need to develop new, more precise, and more efficient genome-editing tools (Perisse et al., 2021). Base editing has moved from ex vivo to in vivo applications, with at least nine human clinical trials underway (Liu, 2024). In the case of livestock, base editing has been reported to be effective in generating pigs and sheep with point mutations through direct embryo microinjection (Zhou et al., 2019; Song et al., 2022). However, both the adenine base editor and cytosine base editor strategies are known to induce high rates of unintended mutations compared to other gene-editing approaches (Lee et al., 2018; Fan et al., 2022; Yan et al., 2023). Prime editing, which uses a nickase enzyme to target a genomic sequence and to prime the action of reverse transcriptase to edit a sequence encoded in a guide RNA, can be applied to insert gene-sized DNA sequences into a target site (Chapter 2). This technique has advanced to the point where the first prime editing-based gene therapies for humans are reaching clinical trials (Liu, 2024). Work is ongoing to combine prime editing and site-specific recombinases and transposases to further improve efficiency and site specificity. In light of these advances, improved genome editors merit further development and evaluation in targeted research in food animals over the next 10 years. Such technical advances will be applied in biomedicine over the short to medium term and hold promise for application in mammalian, poultry, and aquaculture systems.

Another way to reduce CRISPR-Cas9 mediated off-target effects is through epigenetic editors, which avoid permanent DNA edits (Huisman et al., 2015). However, epigenetic editors can exert off-target effects at the epigenome level, an issue that needs to be evaluated.

Delivery methods also still need improvement. While adeno-associated virus and lipid nanoparticles are current vectors of interest for in vivo gene therapy (Chen et al., 2023), these have not yet been applied to HGM of

food animals (Chapter 5). Because the titers required for generating any manner of gene edit(s) in the genome of cells that will be used to produce food animals are not known, and due to the fact that unexpected sequence integration events have been documented in the peer-reviewed scientific literature, it will be important to screen the genome of founder animals to demonstrate lack of viral sequence integration events if viral-based delivery methods, including adeno-associated virus, are used for generating food animals with HGMs.

DATA GAPS THAT REQUIRE FURTHER RESEARCH

The sections below detail data gaps regarding genome editing with CRISPR-Cas-based systems that require further research.

Variable Editing, Including Unintended On-Target Effects

Genome editing at the target site is not always precise due to different mechanisms of host DNA repair, including non-homologous end joining and gene conversion. Complex and large-scale unintended on-target effects (i.e., large deletions and rearrangements) have been observed in mouse embryos, human cell lines, and human embryos (Kosicki et al., 2018; Cullot et al., 2019; Rayner et al., 2019; Rezza et al., 2019; Zuccaro et al., 2020; Alanis-Lobato et al., 2021). Integration of donor plasmid and insertion of multiple donor templates at the target site have been observed (Skryabin et al., 2020). Such large-scale on-target mutations would be missed by conventional genome sequencing methods, but, as discussed in Chapter 4, modern DNA screening methods can detect where such structural variants have occurred. However, the actual type of structural variant (inversion, deletion, insertion, rearrangement) needs to be analyzed with care to ensure accuracy, as the respective chromosomes of a pair may have different structural variants in overlapping chromosomal positions. Reconstruction of the exact mutations requires correct resolution of haplotypes.

Process-related effects

Editing outcomes can vary depending on the process used for genome editing (e.g., gRNA design), the mode of Cas nuclease application (e.g., plasmid versus ribonucleoproteins), the type of Cas used, the genomic site selected for editing, and other technical factors. Gene editing is evolving rapidly. As noted in Chapter 2 and above, new editors are being developed and tested, each with its characteristic mode of action and consequent likelihoods of on-target and off-target effects. In addition, bioinformatic tools and predictive in silico approaches for designing gene edits and genome sequencing will improve the detectability of off-target modifications. As information from applications of these technologies in cultured cells, model systems, and food animals grows, it is important to collect and compare the frequencies of intended and unintended edits produced using the respective technologies. Hence, further research is needed to identify how different genome-editing processes affect the likelihood and repeatability of intended and unintended on- and off-target edits not only in cultured cells, but also in whole animals. Ultimately, systems might be developed and demonstrated that minimize concerns associated with off-target editing of food-animal genomes. With the development of larger datasets, application of artificial intelligence-based tools might support improved choice of sites for genome editing and design of guide RNAs with high site specificity.

Lack of standardized detection methods for unintended alterations

As noted in Chapter 5, various off-target detection methods are currently available, yet each has its own strengths and weaknesses (e.g., complex protocols, applicability across different cell types, associated costs, sensitivity, requirement of suitable controls, bias). Implementation of the multiple layers of risk mitigation discussed in Chapter 4 can help in this regard. Further empirical research is needed to benchmark and refine current methods for genome sequencing and bioinformatic detection of unintended alterations. Allocation of resources is needed to develop the genome sequencing and bioinformatic tools to detect and characterize both intended and unintended HGMs.

Best-practice guidelines

There is a need for development of and targeted education and awareness regarding specific techniques that can be used to find unintended editing events. Moreover, these techniques need to be built into existing gene-editing and bioinformatic research and training programs through the establishment of best-practice workflows or platforms to assess the outcomes of HGM applications. Further, should unintended edits be identified at the genomic level, best practices should include a biological assessment of phenotype to determine whether they are of concern. The development of best-practice workflows is required to assess the outcomes of HGM applications using DNA sequencing techniques.

Development of Germline Stem Cells for Genome Editing

The ultimate intent of gene modification in food animals is to develop heritable traits that can be transmitted across generations in a predictable manner. To achieve this, the germ cell lineage must possess a genome that has been modified so that gametes can deliver gene edits or transgenes through sexual reproduction. While genome editing is most often practiced upon one-celled embryos in which the germ cell lineage will arise or upon cultured somatic cells that will be used to derive one-celled embryos via somatic cell nuclear transfer (SCNT), it also could be practiced upon germline stem cells in vitro that have biological capacity to give rise to the sperm- or egg-producing lineages. As noted in Chapter 2, chickens may be subject to gene transfer or genome editing using germline stem cells (Dimitrov et al., 2016; Han and Park, 2018). In addition, direct editing of germline stem cells combined with a surrogate sire breeding strategy could provide an efficient means to introgress desired gene edits into breeding populations that have already been developed for specific traits.

Applying gene-editing strategies to germline stem cells being grown in vitro could allow for screening of both intended and unintended edits prior to use in producing gametes that will transmit the engineered genome to offspring. The generation of gametes from germline stem cells that have been edited in vitro requires surrogate hosts that lack their own germline but have otherwise functional gonads. Sterile surrogate chickens and fish that can produce both sperm and eggs from donor germline stem cells have been generated (Wargelius et al., 2016; Taylor et al., 2017; Ballantyne et al., 2021). In addition, CRISPR-Cas9 gene editing was used to knock out the NANOS2 gene specific to male fertility in mammalian embryos that would be raised to become the surrogate sires via male germline stem cell transplantation (Ciccarelli et al., 2020). These males were born sterile and produced sperm after sperm-producing stem cells from donor animals were transplanted (via a technique known as spermatogonial stem cell transplantation) into their testes. The sperm produced by the surrogates carried only the genetic material of the donor animals. This advancement aids the potential for the use of surrogate sires as a tool for dissemination and regeneration of improved germplasm in all mammalian species. The results demonstrated that allogenic spermatogonial stem cell transplantation in mice, pigs, and goats is immunologically tolerated, leading to extensive engraftment and persistent donor-derived spermatogenesis at a level sufficient to yield fertility. Moreover, inactivation of NANOS2 led to germline ablation in male cattle, which opens the possibility of one day developing surrogate bulls that can be utilized for disseminating desirable and edited genetics in cattle populations globally.

A key to applying direct germline editing and deployment through breeding of surrogate sires is the capacity to grow germline stem cells in vitro in a manner that allows for editing their genomes prior to transplantation. In doing so, such studies embrace the 3Rs (Replacement, Reduction, Refinement) when using animals in research and testing (Russell et al., 1959). At present, it is technically challenging to culture male germline stem cells of all livestock species (Oatley et al., 2016; Giassetti et al., 2019), and further research and development are needed in this space. Direct germline gene editing has been developed as an effective approach to engineer germplasm of poultry and fish, but is an underdeveloped strategy for livestock. Targeting DNA changes in germline stem cells holds strong potential for enabling screening and selection for specific modifications in a cell type that is directly responsible for heritable transmission. Realizing this potential requires the development of methods for propagation of germline stem cells in vitro. When refined, direct germline gene-editing approaches could provide an efficient means to integrate desired gene edits into established breeding populations via combination with surrogate sire technologies. Whether the approaches will overcome animal welfare concerns or present additional

challenges compared with other currently used methods must be addressed through research and development pipelines. Since direct germline gene-editing approaches are underdeveloped compared with other strategies for introducing gene edits such as direct embryo modification and SCNT, investment in research is needed to develop strategies for direct germline gene editing of livestock, including improvement of methods for in vitro propagation of germline stem cells.

POST-EDITING CHALLENGES AND PRE-COMMERCIAL DEVELOPMENT

Identification of Unintended Modifications in Silico and In Vitro

The next critical control point in the development of HGM animals is to investigate unintended alterations that may have occurred in the editing process and are inherited in the next generation. Unintended changes are of concern because of the potential hazard that they may present to normal gene function, fluxes through key biochemical pathways, host homeostasis, animal health and welfare, and ultimately food safety. The health of HGM animals would be screened before an HGM animal line would be submitted to regulatory review with the intent to enter commercial production systems (Chapter 4), and in this context, it is worth noting that the progenitor HGM animals are often not the animals used for production. Yet, the detection and assessment of the effects of unintended modifications is of interest for developing the scientific basis for ensuring animal health. Hence, addressing the issue of unintended changes could increase public trust and social acceptance of products derived from transgenic and genome-edited livestock (deGraeff et al., 2019; Fan et al., 2022). In silico tools (open online software) can predict a potential CRISPR-Cas9 off-target site because the prediction algorithms are based on sgRNA sequences. These in silico tools, however, are insufficient when considering complex intranuclear microenvironments involving epigenetic and chromatin organization states. Other cell-free detection methods reconstitute nuclease reaction on DNA or chromatin extracted from cells to identify genomic cleavages, for example, digested genome sequencing using cell-free chromatin DNA offers higher accuracy for suggesting off-target sites for further investigation (Guo et al., 2023). Additional research is necessary to advance in silico tools that simulate in vivo or in vitro conditions to facilitate rapid and accurate prediction of unintended edits.

Reference Genomes and Pangenomes

Detection of unintended edits remains challenging due to natural genetic variation, including structural variation (SV). Existing mutations, as distinct from those induced by the editing process, are initially detected as deviations from a reference genome (Chapter 5). While most food animals have extensive information available about naturally occurring variants, it is not a simple matter to integrate and update variant information with new genome releases. For example, the National Center for Biotechnology Information resources that collect single nucleotide polymorphism and SV data stopped providing updates for non-human organisms in 2017 (NCBI, 2017), and there have been multiple updates to food-animal genomes and the development of breed- and line-specific genomes in the years since. An alternative approach is to use pangenomes so that SVs are recognized and incorporated into the graphical representation of the genome, allowing for more informed choices for gene editing experiments (Chapter 5). Pangenomes have been completed for a few species, including chickens (Rice et al., 2023) and pigs (Li et al., 2017; Tian et al., 2020), and are in progress for ruminants. However, there is an urgent need to develop a pangenome for all food-animal species to support accurate detection of indels and SVs. Funding needs be allocated to develop gapless genomes and pangenomes for all food-animal species. This research will enhance the ability to identify structural variants and make more informed choices regarding genome editing, while also facilitating the evaluation of unintended alterations.

These pangenome efforts should also provide high-resolution chromatin structure information (e.g., chromosome conformation capture sequencing or Hi-C sequencing) to enable mapping of higher-order chromosome folding and elucidate interactions between DNA fragments that are close in three-dimensional space but far apart in the linear genome. The outcome of Hi-C sequencing will vary based upon the cell type analyzed, and this approach would require analysis of multiple tissue types to be comprehensive, but data from such studies would help to

determine any possible effects from alterations of food-animal genomes. Furthermore, while several aquaculture genomes have been completed (e.g., channel catfish, Atlantic salmon), a dedicated effort to provide high-quality reference genomes and pangenomes for crucial aquaculture species (including fishes, crustaceans, and shellfishes) is needed.

Assessing Consequences of Mutation Following Identification of Unintended Genomic Modifications

As discussed in Chapter 5, whole-genome sequencing can be used to identify changes between individuals and between generations, but it remains difficult to differentiate natural variations from genetic alterations. However, assessment of potential harm to an HGM animal requires the verification of proper function of the transgene or edit. Any impacts of unintended effects caused by expression of the HGM are difficult to determine solely from -omics data but will ultimately manifest in the phenotype, and would need to be addressed if biologically relevant. While multi-omics approaches have been used to great benefit to understand biological mechanisms related to gene expression and relationships between genes, the application of these approaches to the outcomes of gene editing or transgene insertions should be hypothesis-driven rather than conducted through an open-ended search. Investigators, regulators, and interested parties should be cognizant of the difference between a statistically significant change and a biologically relevant effect. Assessments of any HGMs must be based upon phenotypic changes in the animal, and these assessments should be focused on the pre-commercial generation, rather than the initial HGM progenitor.

ASSESSMENT AND ANIMAL WELFARE

HGMs may impact animal welfare in positive or negative ways. Improvement of disease resistance would reduce animal suffering and mortality. However, unintended alterations could manifest as negative effects upon the phenotype of the HGM, for example, as seen in some myostatin-edited animals (Guo et al., 2016; Yeh et al., 2017; Matika et al., 2019). While observable phenotypic outcomes would lead to culling of affected animals during the development of an HGM line as a step in the risk management process (Chapter 4), the impact of HGMs on more subtle endpoints may be more difficult to assess.

The impact of HGMs on animal welfare is particularly difficult to access because the perception of animal welfare is itself controversial. The use of animals in agricultural systems involves controversies over: (1) how to define animal welfare; (2) economic concerns of producers; and (3) ethical debate over the use of animals in agriculture (Chapter 3). When concerns about animal welfare are coupled with concerns about sustainability and global food security, the problem of welfare in animal agriculture becomes an intractable problem because it is unlikely that any solution will simultaneously address all issues of concern. Further, the World Organisation for Animal Health (WOAH, 2016) has adopted a definition of animal welfare that includes the physical and mental state of an animal in relation to the conditions in which it lives and dies. As noted in Chapter 3, the physical health of an animal can be assessed objectively. In contrast, the mental state of an animal is a nebulous quality to assess (see Chapter 3). This problem is exacerbated in non-mammalian species (e.g., poultry and fishes), as what constitutes a healthy mental state in non-mammals is less well studied. More research and discussions with stakeholders must be conducted to determine clear, reproducible, and quantitative assessments for food-animal welfare, particularly with respect to what constitutes a healthy mental state in animals. This effort should encompass all vertebrate food animals, including poultry and fishes.

Discussion of animal welfare to define best practices for assessing and maintaining animal welfare during the development and production of HGM animal lines is recommended. In particular, discussion should consider whether behavioral change might be assessed as a possible marker for changed animal welfare. Discussion is needed by the animal welfare community regarding what types of HGMs would be considered “beneficial.” Shriver (2009, 2020) poses an ethical argument that HGMs to knock out pain receptors in domestic livestock would reduce the capacity to suffer and would lead to better consequences than maintaining the status quo, leading to a world in which there is much less unnecessary suffering. However, given that pain evolved as a warning mechanism to

avoid further injury, its reduction may not be an acceptable goal since this could potentially lead to even greater injuries and reduced health and well-being. Further discussion and clarification are needed among the animal welfare community and other stakeholders regarding what types of HGMs would be considered “beneficial” or at least not rendering the animal “worse off.”

In the context of HGM food animals, there are no set guidelines for quantifying changes in animal welfare related to the application of HGM methods. As noted in Chapter 5, there are existing procedures in place for assessing farmed animal welfare which could be applied to the production of HGM animals. Animal welfare laws and programs should apply to the welfare of all food animals, irrespective of whether they are derived from conventional breeding programs or as a result of transgenesis or genome editing.

PHENOTYPIC ASSESSMENT OF HGM EFFECTS

While multi-omics approaches offer promising methods to identify changes in the genome, transcriptome, proteome, and the contribution of metabolites to biological effects in early generations of the HGM line, these approaches should be used specifically and cautiously. The application of multi-omics approaches needs to be appropriate to the transgene or gene edit being evaluated and the anticipated effects of that HGM, and should be hypothesis-driven with a clear understanding of what a biologically relevant outcome would look like (as opposed to a statistically significant change, as discussed Chapter 4 and above). This approach requires that the correct comparator is used so that results can be compared to the corresponding food products found in commerce.

During the research and development phase of creating food animals with HGMs, experimental rigor mandates that genotypic alterations driving intended phenotypic changes are fully elucidated in prototype founders. Once the trait-driving genetic alterations are determined to be heritable through the germline of founders, focus then shifts to assessing the durability of the phenotype in subsequent generations and production-relevant settings, and continued assessment of genotype becomes less important. Reproductive performance is a useful indicator of overall production and the health of the animals. Hence, assessment of HGM food animals should focus on durability of phenotypes within the pre-commercial generation, rather than on the HGM founder that will be multiple generations removed from animals that ultimately are utilized for food production. Phenotypic durability assessment of the HGM animal should focus upon its health, development, and reproductive performance. The food safety assessment should focus upon the similarity of foods from HGM animals in comparison with corresponding foods derived from conventionally bred animals, and whether there is a material difference from existing foods that are already considered safe to consume.

IMPACTS OF HERITABLE GENETIC MODIFICATION ON FOOD COMPOSITION

A leading issue for the consumer and a key focus of regulatory agency interest (FAO and WHO, 2008; FDA-CVM, 2024a) is the potential impact of HGMs on the composition of derived animal foods (Chapters 3 and 4). To assess food safety, it is important to know the range of variation of key nutritional components of foods derived from an HGM animal to provide an appropriate context for assessing the significance of any differences between HGM animals and conventional comparators. A key threshold question, then, is what is the “normal” composition of comparator animal-derived foods? In particular, what is the range of variation in the composition of a food derived from a specific breed or across multiple breeds? The existing database on composition of animal food products is generally inadequate for purposes of assessing the impacts of HGM on food composition.

Nesbitt et al. (2024) evaluated both quality traits and composition of meat derived from homozygous CD163 gene-edited and homozygous-unedited (null) pigs with heritable resistance to porcine respiratory and reproductive syndrome virus (PRRSV). No significant differences were identified between homozygous-edited (CD163ΔE7/ΔE7) and homozygous-null (CD163+/+) animals across all characteristics recorded. The values for all measured variables fell within the normal range, as described in reference publications cited by the authors. Thus, there was no evidence to suggest that the CD163 edit had any adverse impact on meat quality or composition, or on postmortem physical carcass defects. This study appropriately compared the composition of meat and meat products from the CD163-edited pigs with those of unedited pigs of the same breeding line in the company’s overall

breeding population. In some contexts, such as HGM cattle produced from abattoir-sourced eggs, such a direct comparator would not be available. Further, it would be useful to have data not only on the HGM line and its direct comparator, but also across the breed and species to gain a broader understanding of the effect of the HGM upon food composition. The availability of species-wide data would, for example, have strengthened interpretation of the significance of changes in food composition for products of the AquAdvantage salmon, polled cattle, and PRRSV-resistant pigs (Chapter 4).

Concomitant with improved knowledge about food-animal product composition is the need to expand available datasets of nutrient profiles from foods derived from animals generally, including HGM animals, and to ensure that these data are publicly accessible and interoperable (Wilkinson et al., 2016). Given the need for more extensive and intensive data on the composition of animal-derived foods, the committee identified a need for more original research and much greater sharing and archiving of data collected by the academic, public, and private sectors. The restructuring of USDA’s composition and nutrient databases, particularly Foundation Foods and Experimental Foods, was conducted with an emphasis on capturing metadata that characterizes sources of food variability (Fukagawa et al., 2022), providing an excellent platform for expansion. Sustained support for expansion and for public access to these resources will facilitate informed risk assessment of foods derived from HGM animals, as well as help to elucidate the effects of genotype variations on animal phenotypes. Significant investment is recommended in (1) research that enables the generation of nutrient profiles of animal-derived foods that reflect the range of nutrient variation that exists due to natural genetic variation across breeds and lines of animals, environmental effects, animal production and management practices, and processing and cooking methods; (2) sustained support for the expansion of USDA’s FoodData Central⁴ to provide resources for the submission, curation, and maintenance of nutrient profile data; and (3) development and awareness of community standards for deposition of research data.

Scientific and technical advancements raise the possibility of developing an updated, more defensible approach for assessing the safety of foods derived from HGM animals (Chapter 4). Brune et al. (2021) proposed that compositional assessment of HGM organisms should follow a stepwise approach to determine what, if any, compositional data generation is necessary. If HGMs are designed to affect or regulate biochemical pathways and cascades, hypothesis-driven compositional studies looking at the affected pathways or cascades might well be warranted (Herman et al., 2009). The list of compositional analytes that pose a potential human food safety hazard would be based upon a hypothesis generated by the nature of the trait being introduced (Waters et al., 2021). Experiments then can be designed to measure the appropriate key nutrients to test that specific hypothesis.

Statistically different alterations do not necessarily equate to biologically relevant changes in physiology, animal health, or welfare (Chapter 4). Multi-omics approaches may be useful for inferring whether an observed difference in proteins or metabolites is biologically relevant. Hypothesis-driven compositional analysis for assessing the safety of HGM animal-derived foods is recommended.

ALLERGENICITY AND FOOD INTOLERANCE ASPECTS OF HGM ANIMAL FOOD COMPOSITION

As noted in Chapter 3, food allergies are common, affecting about 3-4 percent of adults and 6-8 percent of children in the United States. The Food Allergen Labeling and Consumer Protection Act of 2004 identified eight principal food allergens (milk, eggs, fish, shellfish, tree nuts, peanuts, wheat, soybeans), and in 2021, sesame was added to the allergen list. Responding to the concerns of the American public, the U.S. Food and Drug Administration (FDA, 2024) requires manufacturers of products containing these food items to label their products as containing allergens and identify the specific allergens they contain.

Beyond these nine principal food items, most foods have proteins that potentially can cause IgE-mediated allergic reactions (Chapter 3). Food allergy is a complex response to a potential hazard (allergenic proteins in foods) involving the complex and variable human immune system. Individuals are all immunologically unique regarding response to allergens, and this response is affected by both genetic and environmental factors. Sensitization is the development of immunoglobulin E (IgE) antibodies upon first exposure to a food protein, and allergic reaction (elicitation) follows upon subsequent exposure. Food allergen exposure by skin or inhalation is thought to be a

___________________

⁴ https://fdc.nal.usda.gov

major component of sensitization, whereas early ingestion of these allergenic proteins induces tolerance in many cases (Chapter 3). Allergenicity risk assessment therefore focuses upon the risk of elicitation.

Toxicity and allergenicity

Both hazard identification and exposure are required to estimate any risks posed by food products from HGM animals. Traditional food animals that are commercially produced do not have the capability to synthesize toxic compounds like plants do, although they may accumulate or sequester toxins (Chapter 3). Molecular and microbiological assays already exist and are deployed to detect toxins in food products. Likewise, lessons learned from transgenic and genome-edited food crops make it unlikely that toxigenic or allergenic genes would be purposefully transferred into food animals (Chapter 4). The committee proposes a system for determining the relative risk for increased levels of toxicity or allergenicity based upon (1) whether the gene edit could be achieved by conventional breeding or would require a transgenic approach and (2) whether the altered gene’s product is expressed in tissues that are consumed (Chapters 3 and 5).

If a novel product is expressed in HGM animals, then protein characterization and hazard identification encompassing toxicity and allergenicity would be performed on the resulting food products as a part of the core protein characterization studies for food safety assessment. Waters et al. (2021) proposed that additional studies should be required only when a food safety hazard is identified in the first set of core molecular and protein characterization studies (Brune et al., 2021) and that problem formulation should be employed to design any hypothesis-driven supplementary studies to be performed to characterize hazard and exposure only when a hazard is identified. The effect of the transgene or genome edit should be considered; for novel products expressed in HGM animals, toxicity and allergenicity assays would be performed on the resulting novel food products as required.

Methods for determining allergenicity

Mammalian meats are not commonly allergenic because humans tend not to be allergic to proteins similar to those in the human body. The animal food allergens are usually (1) structures that are not in humans, for example, in eggs; (2) secreted proteins, for example, milk; or (3) products from animals such as fish and shellfish that are evolutionarily distant from mammals and hence express proteins not found in humans. As of the writing of this report, no proven allergic reaction has been reported as a result of any food animal being genetically modified. That being said, it is important to know how similar a protein needs to be to an allergen in order for it to prove allergenic. The Food and Agriculture Organization and World Health Organization (FAO and WHO, 2001) and the EFSA GMO Panel (2010) stated that a primary sequence with 35 percent identity in the amino acid sequence of the expressed protein across a sliding 80-mer residue window is sufficient to elicit an allergenic response. However, there has been little testing because of the difficulty in obtaining true allergen/non-allergen sequence, along with a dearth of research in clinical food allergy. Abdelmoteleb et al. (2021) indicated that this testing approach may be too conservative.

Several lessons learned from experiences with genetic modifications of crops can help improve methods for characterizing the potential allergenicity of products from HGM animals. The most frequently used allergen sequence database is AllergenOnline (Goodman et al., 2016). This database was developed and is maintained by the Food Allergy Research and Resource Program (FARRP) in the Department of Food Science and Technology at the University of Nebraska-Lincoln. AllergenOnline provides access to a peer-reviewed allergen list and sequence-searchable database that is intended for the identification of proteins that may present a risk of allergenic cross-reactivity. The website was designed to help in assessing the safety of proteins that may be introduced into foods through genetic engineering or through food processing methods. The objective is to identify proteins that may require additional tests, such as serum IgE binding, basophil histamine release, or in vivo challenge to evaluate potential cross-reactivity. Focusing primarily on reactivity of proteins with IgE from clinically relevant human sera is suggestive evidence, but not proof of allergenicity. In addition, some proteins in this and other databases may not be allergens, and the degree of allergenicity is not considered.

Many food allergens are susceptible to pepsinolysis, while many non-allergenic compounds are resistant to pepsinolysis (EFSA GMO Panel, 2020). The current regulatory requirements for allergen risk assessment in

genetically modified organisms should be updated. The 35 percent identity/80-mer sliding-window approach should be validated, or changed to a more discriminatory, validated approach (EFSA GMO Panel, 2020). There is currently no food safety regulatory oversight consideration of self-tolerance where proteins like those in the human body are regarded as unlikely to be allergenic. Additionally, enhancements to allergen databases are needed, especially to include allergenic potency and possible abundance information to help facilitate risk assessment. The pepsinolysis test is poorly predictive of allergenicity and therefore needs to be either validated or removed. Novel workflows, including proteomics, should be developed, validated, and standardized to ensure utility and consistency of application.

The adoption of matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (MALDI-ToF-MS) for allergen analysis may emulate the detection capability (sensitivity) of established allergen detection techniques while reducing technical support and having multiplexing potential equivalent to competing techniques such as enzyme-linked immunosorbent assay and liquid chromatography-mass spectrometry (Birse et al., 2023). Moreover, affinity-based MALDI appears capable of higher throughput, at a lower cost per sample, than any existing technique, thereby enabling repeated sub-sampling as a way to reduce representative sampling issues.

Allergenicity detection methods should be updated and standardized to ensure utility and consistency in application before being considered for possible inclusion in evaluation by regulatory authorities.

HAZARD OF DISEASE TRANSMISSION FROM HGM ANIMALS TO OTHER ANIMALS AND HUMANS

Both classical gene transfer and genome editing have been applied to develop disease-resistant or disease-resilient lines of food animals (Chapter 2). Some experiments have given rise to lines with promising levels of disease resistance, for example, PRRSV-resistant pigs (Whitworth et al., 2016; Burkard et al., 2017), which are currently being incorporated into commercial production systems (Burger et al., 2024). While disease-resistant and disease-resilient animals provide improved productivity and resilience of food production systems, as noted in Chapter 3, they also pose the hazard that disease-resistant food-animal populations might provide selective pressure for pathogens to evolve with altered host range or virulence. Further research should be directed toward identifying and characterizing key determinants of disease and transmission likelihood to improve the identification of field isolates that may pose a threat to human and animal health.

However, there are multiple approaches that together can be used to reduce pathogen evolution, as discussed in Chapter 5. In genetically modified plants, stacking multiple transgenes conferring herbivore or pathogen resistance has been shown to make counter-evolution of herbivore or pathogen less likely (Zhao et al., 2003; Thompson et al., 2012; Shehryar et al., 2020), but not zero (Luo et al., 2021). Similarly, for food-animal systems, stacking to target multiple genes or pathways simultaneously (rather than targeting a single pathway or process for resisting infection) would reduce the risk that a pathogen would evolve to circumvent the barrier to infection or transmissibility. HGM animals that are disease resistant can be tested against new field isolates and for continued reduced viral loads or reduced transmission, as appropriate. Research is recommended to identify new targets for HGMs to enhance the resistance of food animals to disease. This is considered necessary to know. Deployment of disease-resistant HGM animals should be accompanied by rigorous testing and ongoing surveillance of commercial lines to ensure continued efficacy. This is considered necessary to ensure continued food-animal health and production. Gene stacking should be utilized to develop HGM disease-resistant and disease-resilient animals, as appropriate (e.g., where there is enough information about pathogen-resistance pathways).

Surveillance within the HGM disease-resistant population might include post-slaughter testing for disease or seroconversion, and this should be combined with improved disease surveillance in food animals along with integrated wildlife monitoring. As noted in Chapter 3, an ongoing global outbreak of avian influenza has devastated bird and marine mammal populations. Although the ongoing outbreak is being monitored, the level of effort being devoted has been criticized as inadequate (Gao, 2018; Wallace-Wells, 2024). The Centers for Disease Control and Prevention’s One Health Program must be adequately resourced to fully perform its mandate and to provide surveillance for potential breakthrough pathogens from HGM food animals. Improved, integrated wildlife monitoring is recommended to support the One Health effort.

With current knowledge, the risk of a pathogen evolving the ability to evade the immune response of livestock or humans is difficult to impossible to predict. Directed research into the determinants of heightened pathogenicity and transmissibility in livestock is warranted. Further research needs to be directed toward understanding the risk of pathogen evolution within HGM and conventionally bred food animals and spillover into other animal and human populations.

INTERNATIONAL HARMONIZATION OF ASSESSMENT METHODS AND POLICY

Developers of HGM food animals need to secure regulatory approval before they can market their lines for commercial production. Approval of a product for marketing within the United States or any country suggests that the product will enter not only domestic production, but also international trade. International trade in animal products is important to producers, consumers, and the animal production sector.

International trade in the products of animal biotechnology would be promoted by adoption of similar food safety assessment methods and reporting requirements across countries and supranational groupings. Some countries have proven unwilling to review food safety data generated in other countries; for example, regulators in Panama asserted that data for food safety assessment of the AquAdvantage salmon should be generated anew in Panama, a requirement not justified on a scientific level. International trade in the products of animal biotechnology would benefit from alignment or harmonization of regulatory polices among countries and supranational groupings (Hallerman et al., 2024). Strict regulation of animal biotechnology is effectively a barrier to trade in such products, for example, with the European Union.

Scientists, developers, and regulators have met in several virtual and in-person workshops and international forums aimed at promoting understanding of HGM technologies and science-based, risk-proportionate regulatory processes. The U.S. Department of Agriculture–Foreign Agricultural Service has provided important funding and logistical support for these workshops, as well as for regional workshops aimed at building regulatory capacity for oversight of animal biotechnology. The USDA Foreign Agricultural Service and/or other agencies need to continue to support these international harmonization activities over the next 3-10 years.

COMMUNICATION, PUBLIC OUTREACH, AND EDUCATION

Communication and public outreach

Agricultural production of HGM animals poses both benefits and risks. Ultimately, however, the level of production of HGM animals will be determined by the marketplace. Hence, public acceptance of food products from HGM animals is at issue. Therefore, continuous engagement, education, and transparency are key.

The study charge to this committee focused on assessing animal biotechnology and the potential biological risks associated with developing food animals with HGMs, a focus reflected in the fact that all members of the committee are trained in the biological sciences or allied areas. The committee was not tasked with assessing best practices for public engagement but recognizes the importance of this issue. There is a general consensus that in many cases, risk management cannot simply focus on biological mechanisms, given that two biologically equivalent risks are generally not experienced as equal by the public (Fischhoff et al., 1978; NRC, 1996). If risk analyses are to inform policies regarded as legitimate in the eyes of the public, the public must be engaged in the process (NRC 1983, 1996; NASEM, 2016a, b). To effectively engage the public, efforts must incorporate lessons from decades of scientific evaluations of past approaches (Dankwa-Mullan et al., 2021). For example, research has demonstrated that the once highly regarded “deficit model,” in which experts simply educate the public, is often not useful (Dudo and Besley, 2016). This has been demonstrated in cases of controversial emerging technologies such as genetic engineering of crops (NASEM, 2016a). Recent research on public perceptions of gene editing of food animals indicates that biological risks are often not among the top concerns (Yunes et al., 2021; Winther et al., 2023; Borgdorf and Meijboom, 2024; Kuo et al., 2024). New insights from social science research could enable more effective engagement with the public on issues related to food animals with HGMs. A future study group should be established by the National Academies of Sciences, Engineering, and Medicine to address the need for

public engagement activities related to HGM technologies, especially as applied to food animals. This study group should develop a set of best practices for engagement based on scientific assessments of past efforts at public engagement regarding emerging technologies. Such a committee should include both natural and social scientists.

RECOMMENDATIONS

Based on the committee’s deliberations and assessment, the following recommendations are provided:

1. Targeted research should be conducted to further develop and evaluate improved genome editors.
2. Resources should be allocated to investigate how various genome-editing techniques influence the frequency, consistency, and predictability of both intended and unintended alterations.
3. Best-practice workflows should be developed to assess the outcomes of HGM applications.
4. Additional research should be conducted to advance the use of in silico tools that simulate in vivo or in vitro conditions to facilitate rapid and accurate prediction of unintended edits.
5. When viral-based delivery methods such as adeno-associated virus are used for HGM of food animals, HGM candidate founder animals should be extensively screened to demonstrate that no viral vector integration has occurred and ensure that lines to be commercialized do not have unidentified or uncharacterized viral integration events.
6. Research investments should be allocated toward improving the in vitro culture of germline stem cells from food-animal species.
7. Assessment of HGM food animals should focus on durability of phenotypes within the pre-commercial generation, rather than on the HGM founder that will be multiple generations removed from animals that are ultimately processed as a food product. Phenotypic durability assessment of the HGM animal should focus upon its health, development, and reproductive performance.
8. Food safety assessment for HGM animal-derived products should focus upon substantial equivalence, that is, on whether the phenotype compares to that of existing food products that are already considered safe to consume.
9. Funding should be allocated to develop gapless genomes and pangenomes for food-animal species.
10. Support research and facilitate discussions with stakeholders, aimed at determining clear, reproducible, and quantitative assessments of food-animal welfare, particularly with respect to what constitutes a healthy mental state, and to define best practices for assessing and maintaining animal welfare during the development and production of HGM animal lines.
11. Facilitate further discussion and clarification among the animal welfare community and other stakeholders regarding what types of HGMs would be considered “beneficial” or at least not rendering the animal “worse off.”
12. Development and use of HGM animals should be conducted in accordance with existing animal welfare laws and programs, which are applicable to the welfare of all food animals, irrespective of whether they are derived from conventional breeding or as a result of HGM procedures.
13. Support significant investment in (1) research that enables the generation of nutrient profiles of animal-derived foods that reflect the range of nutrient variation that exists due to natural genetic variation across breeds and lines of animals, environmental effects, animal production and management practices, and processing and cooking methods; (2) the sustained expansion of USDA’s FoodData Central repository to ensure resources for the submission, curation, and maintenance of nutrient profile data; and (3) development and awareness of community standards for deposition of research data.
14. Compositional analyses required for regulatory assessment of whether HGM animal-derived foods are safe for consumption should be hypothesis-driven.
15. Given that the phenotypic effect of the genome edit is the key consideration for risk assessment, toxicity and allergenicity assays should be performed on the novel food products from HGM animals as required.

16. Allergenicity detection methods should be validated and standardized to ensure utility and consistency in application before being considered for possible inclusion in evaluation by regulatory authorities.
17. Further research should be directed toward identifying and characterizing key determinants of disease acquisition and transmission likelihood to improve the identification of field isolates of pathogens that may pose a threat to human and animal health.
18. Support research to identify new targets for HGMs to enhance the resistance of food animals to disease. Deployment of disease-resistant HGM animals should be accompanied by rigorous testing and ongoing surveillance of commercial lines to ensure continued efficacy. The utilization of gene stacking to develop HGM disease-resistant and disease-resilient animals is recommended, as appropriate (e.g., where there is enough information about pathogen-resistance pathways).
19. Further research should be directed toward understanding the risk of pathogen evolution within HGM and conventionally bred food animals and spillover into other animal and human populations.
20. The USDA Foreign Agricultural Service and/or other agencies should continue to support international regulatory harmonization activities over the coming 3-10 years.
21. A study group should be established by the National Academies of Sciences, Engineering, and Medicine to address the need for public engagement activities related to HGM technologies, especially as applied to food animals. This study group should develop a set of best practices for engagement based on scientific assessments of past efforts at public engagement regarding emerging technologies. Such a committee should include both natural and social scientists.

REFERENCES