Sustainable use of CRISPR/Cas in fish aquaculture: the biosafety perspective

Technically great success has been achieved in some fish species with obtaining various GE lines, especially Atlantic salmon and tilapia. These species should be employed as “aquaculture models” to initiate the optimization of aquatic CRISPR/Cas protocols and physiological assessment of potential off-target effects (for food quality and safety). The derivable knowledge from this approach is transferable to other fish species.

The cardinal ERA issues of GE fish are related to release (both intentional and inadvertent) in the environment because such escapees can; a) hybridize with the wild population and can, for SDN-3, lead to dispersion of transgenes (Devlin et al. 2010; Oke et al. 2013; Wringe et al. 2018), and/or b) interfere with existing biodiversity. The environmental risks associated with escapees are (a) changes to the population genetics of closely related species in the receiving environment via mating and alteration to genetic biodiversity (Hayes 2007; Kapuscinski 2007); (b) disturbance in ecological balance via alteration of the food web and destruction of habitat. However, this pertains less to GE fish in inland and contained aquaculture facilities, and more to GE fish in aquaculture systems located in waterways or marine/coastal environments where escapee farmed fish and inadvertent introduction of farmed fish into the environment are possible. Physical and biocontainment barriers are mandatory conditions for ERA of GE fish (Devos et al. 2019; Kapuscinski 2007), nevertheless, these two interventions are not guaranteed. For example, the biocontainment strategy through polyploidy (the induction of 3 or more chromosomes in fish eggs to reduce their fertility), which was employed as part of the transgenesis of the commercialized AAS (Devlin et al. 2010) is leaky because a small percentage remains as diploid fertile fish expressing growth hormone, such that there could be some fertile individuals among escapee salmon (Benfey and Sutterlin 1984). Besides, generating polyploidy in fish is also an animal welfare issue. Similarly, physical containment is not foolproof because escapee farmed Atlantic salmon have been reported at wild salmon spawning grounds (Bergan et al. 1991; Gausen and Moen 1991; Jensen et al. 2013). This has led to criticism on the extent of the physical and biocontainment conditions of the ERA that was conducted for the commercialized AAS (Benessia 2015; Sweet 2019) especially given the difficulty of providing or predicting environmental impacts of release of GE fish. Unfortunately, no information on ERA is publicly available for the commercialized GE tilapia. The challenge to ERA is whether the effects of hybridization and transgene introgression of the released GE fish can affect overall fitness including survival, migration, spawning, reproduction, etc., of the wild population. The sterile GE salmon by the Wargelius group (Wargelius et al. 2016; Güralp et al. 2020), as a proof of prinicple, can prevent hybridization from SDN-1 and SDN-2 fish, and transgene introgression from SDN-3 fish via interbeeding between a released GE fish and wild population, but the impacts of such modification on overall fitness of the fish under natural conditions have not been conducted. Unfortunately, the natural and environmental conditions that influence these factors cannot be studied under controlled laboratory conditions (Leggatt et al. 2017; Sundstrom et al. 2007). However, the concept of combining sterility trait and other traits in the same GE fish increases the prospect of avoiding hybridization with wild relatives and controlling transgene introgression by escapee GE fish from open water commercial fish farms.

The aim of RA for food safety is to ensure that the process of modification as well as the effects (both intended and unintended) have not resulted in toxicity, allergenicity and/or decreased food quality (i.e. undesirable biochemical composition of the edible tissues) of the modified fish (EFSA 2013). Effects of the CRISPR/Cas modification process, unintended effects as well as intended new traits can affect the fitness and hence the quality of the actual edited fish. Currently, systematic studies have not been conducted on the impact of unintended effects of GM (or CRISPR/Cas) on the quality/safety of edible tissues of GE fish (or any GM or GE modified animal), although unintended changes have been reported to alter disease resistance, foraging behavior, gene expression, reproduction and life-history timing (Abrahams and Sutterlin 1999; Devlin et al. 2015). Extrapolations from specific examples of past experiences with GM and genetic engineering technologies can help deduce potential impacts of CRISPR/Cas on the quality of edited fish for RA purposes. For example, biochemical analyses of the components (carbohydrates, proteins, total fats, vitamins and minerals) of the edible tissues of the AAS show no significant variations compared with the unmodified counterpart, except a slight variation in the concentration of vitamin B6 (Benessia 2015). The mRNA expression levels of the following proteins were reported as increased in a growth-modified transgenic amago salmon (Oncorhynchus masou): haeme oxygenase; leukocyte cell-derived chemotaxin; α-trypsin inhibitors; iron metabolic proteins; and proteins of the reproductive system, while the expressions of lectin, D-6-desaturase, apolipoprotein and pentraxin were reduced (Mori et al. 2007). In a transgenic coho salmon, glutathione levels; glutathione reductase and gamma-glutamiltranspeptidase activities were increased by growth hormone modification (Leggatt et al. 2007). A CRISPR/Cas9-based ablation of elovl2 gene (an essential gene in synthesis of long chain polyunsaturated fatty acids LC-PUFA), resulted in the accumulation of different polyunsaturated fatty acids, and up-regulation of several genes involved in fatty acid metabolism in Atlantic salmon (Datsomor et al. 2019b). The extent to which these can impact the quality of edible tissues of the GE fish is unknown. The AAS was also evaluated for allergenicity via dermal contact (Leggatt 2013) according to present RA standard procedures. However, the impact of the intended increased growth hormone on the edible tissues and its possible health implications was not conducted, neither were the direct and indirect impacts on the entire cellular metabolic network performed (Benessia 2015; Van Eenennaam 2011). It is questionable whether the current standard molecular characterization, toxicity and allergenic studies for characterization of GM food safety is sufficient for products of CRISPR/Cas technology. It has been argued (Abrahams and Sutterlin 1999; Devlin et al. 2015) that GM products should also be evaluated for anti-nutrients or lowered nutrients, this may also be relevant for GE products. Others (Kawall 2020; Agapito-Tenfen et al. 2018) have also argued that in addition to standard molecular characterization, omics evaluation of GE products, especially with regard to unintended changes, will provide relevant additional data for robust RAs. Nonetheless, it is clear that determining whether these unintended changes have (or are associated) with any harmful effects is complex and presents a challenge to the present RA frameworks.

Most GMO regulatory frameworks mainly include safety questions on guidelines for health and ERA. In addition to this, it is important, in the case of GE, to acknowledge the social dimensions of how natural sciences use nature (Palsson et al. 2013). Food has an impact on human lives that is of both cultural and biological importance, and it is therefore not sufficient to consider only measurable risk. Food is also about traditions and ways of life (Myskja and Myhr 2020). Several questions of risk, e.g., future effects of horizontal gene transfer, are not possible to answer in present terms. These become questions about what human changing nature might lead to, and whether this is something that society is willing to accept. Hence, the “[…] blurry line between risk and sustainability demonstrates the significance of including non-safety issues in order to make a decision that is socially acceptable” (Myskja and Myhr 2020).

There is no common practice of evaluating non-safety criteria of GMOs, even though several countries have implemented such measures (Myskja and Myhr 2020). In Norway, GMO regulation frameworks include both environmental/ecological, societal and economic dimensions, through including non-safety criteria—contribution to sustainable development, ethical justifiability and societal utility in the evaluation process of GMOs (Box 1). The criterion of contribution to sustainable development is interesting considering the frequent use of this term, e.g. in global and national aquaculture strategies and reports. However, a framework operationalizing contribution to sustainable development at present has only been developed for GM plants (NBAB 2009, 2011, 2014). An alternative to developing a framework is to make requirements for certifications under international certification schemes; an example for GE fish could be certification under the Aquaculture Stewardship Council (ASC).

The new possibilities of GE, especially CRISPR/Cas, as stated above, call for updated regulatory and RA frameworks and guidelines. Likewise, they demand a new public discussion on the application of the technology in food production amongst other areas of use. Historically, GM has been associated with controversy. The aspects raised in public discourses are connected to animal moral status, the argument of whether genetic engineering is natural or unnatural, the percieved risks and benefits of genetically engineered animals to health and the environment, the purpose of the application, the methods being used and the motivation of the researchers. And finally the species itself—its intrinsic value, species boundaries and animal ethics (Van Eenennam and Young 2018). These aspects may be important for public and market acceptance of GE products.

Some ERA-relevant studies on transgenic fish have been conducted but these relied mainly on laboratory/confined field (Devlin 2007) and modelling (Ahrens and Devlin 2011; Li 2014) studies and not on field studies. In the laboratory/confined field studies, the approach is the application of semi-natural conditions and use of surrogate models in nature (Devlin et al. 2006). EcoPath and Ecosim modelling have been applied in predicting effects of releases of growth hormone transgenic salmon (Ahrens and Devlin 2011; Li 2014). For example, computer-based modelling simulations found that presence of transgene can potentially shift genetic backgrounds and phenotypes of both GM and non-GM fish away from the naturally selected optima (Ahrens and Devlin 2011). However, it is not practicable to obtain data that accurately depict pathway to harm of GM fish when released into natural conditions (Devlin et al. 2015; Sundstrom et al. 2007). This has raised uncertainties as to the extent data generated from these studies can be used in ERA, such as uncertainties related to extrapolating results from confined tests to natural ecosystems, pleiotropic effects and phenotypic trade-offs between traits and genotype-by-environment interactions (Benessia 2015; Devlin 2007).

There is at present no systematic data available on the pysiopathology of GE fish. Even for the most studied Salmonidae subfamily, different methodologies and research objectives have been used. For example, in the widely studied growth hormone-expressing transgenic coho salmon, amago salmon and Atlantic salmon, few studies on biochemical alterations caused by deregulation of growth hormone expression are available (Leggatt et al. 2007; Mori et al. 2007). The most available comprehensive study is the biochemical characterization on AAS by the producer—Aquabounty (AquaBounty Technologies Inc. 2010; Benessia 2015), but no independent systematic studies on AAS have been reported in the published literature. Studies on safety of GE fish for human consumption should not be limited to the direct effect of the transgene (VMAC 2010) for SDN-3 fish, nor allergens (Van Eenennaam 2011), but also extended to the direct and indirect effects on the entire metabolic network (for SDN-1, 2 and 3). In this regard, the application of new and robust molecular biology analytical techniques such as omics (in particular proteomics and metabolomics) can be useful (Agapito-Tenfen et al. 2018; Eckerstorfer et al. 2019a; Kawall 2020). For example, a research group (Datsomor et al. 2019b) recently used lipidomic and transcriptomic analyses to characterize the impact of a CRISPR/Cas9-elovl2 knockout event on lipid biosynthetic pathway.

Many countries have agreed on Agenda 2030 and committed to the United Nations’ 17 global sustainable development goals (SDGs) for a better future (UN 2015). The overarching goal by this commitment is to make possible dignified human life while at the same time permanently and on a sustained basis protect natural conditions of life. The three key parts of sustainability is included within the SDGs: economy, society and the environment. These could be used as guidance for what measures within sustainable development should be assessed. The scope of SDGs is wide and includes 169 specific targets (UN 2015), which are important and internationally accepted goals. Adopting these widely accepted goals could also ease defending the inclusion of non-safety issues internationally. In addition to the UN SDGs, both protection of biodiversity, ecosystems and development of sustainable food production systems are part of ´A European Green Deal´ (European Commission Grean Deal 2019). For example, if use of the AAS and the GE tilapia (as well as future GE fish products) is to be evaluated for contribution to sustainability, it would be advantageous if this evaluation aligns with the action plans of the Green Deal.

Sustainable development is globally defined as development which does not reduce the possibilities of future generations while simultaneously ensuring the possibilities of present generations (Brundtland 1987). Within such a definition, using GE for ensuring a stable and efficient aquaculture production may be accepted, if the technology meets the demands of safety and risk issues, and if it contributes to social and economc sustainability, e.g., upholding transparency in the food chain, creating work opportunities and supporting local communities. Further, the welfare of animals must be morally acceptable for it to be sustainable (Blix and Myhr 2021; Broom 2010). In the Norwegian Animal Welfare Act, all animals are stated to have intrinsic value independent of their use-value to humans (Norwegian Animal Welfare Act 2009). If intrinsic value of animals is also to be taken into consideration, then a sustainable use of GE cannot compromise the moral status of the animal (Blix and Myhr 2021). Defining sustainable development as something that has to be morally acceptable also widens the extent to which perspectives of stakeholders and the public are included in evaluation of GE foods—one of the intentions of non-safety criteria (Zetterberg and Edvardsson Björnberg 2017).

In 2002, the Eurobarometer on Biotechnology showed that “[…] the majority of Europeans [did] not support GM foods” (Gaskell 2003). This attitude did not change according to the findings of the Eurobarometer of 2010 (European Commission/TNS 2010). The concerns for this group of foods were safety for health, and the issues of whether it was necessary to apply GM in place of conventional breeding. In 2019, the concern seemed to have dropped slightly—only 27% of the respondents expressed concern for GM ingredients in foods or drinks (European Commission/EFSA 2019). Predictions are for GE products to be more accepted, especially for products of SDN-1, which do not involve cis or transgenesis. The main objectives for using GE are more efficient and sustainable food production, as well as economic profitability. In spring 2020, two comprehensive surveys on public perception of gene technology in food production were conducted and published in Norway by the Norwegian Biotechnology Advisory Board (NBAB 2020), and by SIFO (Consumption Research Norway) (Bugge 2020). NBAB (N = 2016) used the term ´gene editing` and showed that the Norwegian consumers´ attitudes depend heavily on what the technology is used for. With regards to use on animals, the consumers are mainly concerned with using the technology for improving health in production animals and reducing the environmental impact of protein production industries. Increasing growth or changing visual traits like color of flesh is regarded less important amongst the consumers (NBAB 2020). In the SIFO report (N = 1066), the term GMO was used in their survey, and 47% of the respondents expressed that GMOs collided with their view of what ethical food production looks like. With regards to the possible negative effects of GMOs, the respondents were most concerned with nature and ecosystems, followed by concern for their own health (Bugge 2020).

Similar surveys have also been conducted in China. Recently Cui and Shoemaker (2018) published a nationwide study (N = 2063) on public perception of GM food where a majority of the respondents answered that they are either neutral (46.7%) or opposed (41.4%) to GM food. The study also approached linking acceptance to self-percieved knowledge about GM foods, and states that there seems to be statistical evidence for a positive connection between more knowledge and acceptance. The study also emphasised a lack of trust in the public towards authorities, but also in biologists opinion, and further calls for more effort “[…] to gain confidence, trust and support from public domain” (Cui and Shoemaker 2018). However, the issues that the respondents are most concerned with are “how to identify GM food” and they also want more information about “general scientific knowledge on GM food safety” (Cui and Shoemaker 2018). This underpins the need for dissemination of knowledge connected to GM and GE foods, and that the public seeks information which will not necessarily lead to rejecting GM foods, but rather for more informed actions.

The results from these surveys indicate that the discourse on GE foods is not unambiguous. It has been emphasized that some important issues on which public acceptance is based are—naturalness, trust, one's own health, the environment and the origin of the food products that we eat (Myskja and Myhr 2020). There is a need for upholding this discussion about the origin and type of foods we want to accept and why these are acceptable (De Graeff et al. 2019). This has also recently been brought up by the European Group on Ethics in Science and New Technologies in the report `Ethics of Genome Editing`. Regarding use of GE on animals, key identified questions are how the new technology affects the balance between the three R´s (refine, replace, reduce) in research, and in general, how it affects animal welfare as well as what the implications for biodiversity are (EGE 2021).

Using disease resistance in GE salmon and the SDN-1 approach as examples, data on unintended off- and on-target effects as well as data on detection, traceability and surveillance can be obtained using the scheme described in Fig. 1. Gratacap et al. (2019b) have shown that combining in-vitro and in-vivo large-scale genome wide screening approach can be used to identify disease resistance alleles in aquaculture species. We further propose that studies integrating new genetic tools in omics can be used to identify recurring and consistent features of CRISPR/Cas knockout mutations in salmon and/or salmonid-derived cells using e.g. disease-related genes of Atlantic salmon, which can be relevant for detection and surveillance of GE fish. Non-random repair outcome of NHEJ of DSBs generated by CRISPR/Cas9 across cell-lines, experimental replicates and reagent delivery have been reported (van Overbeek et al 2016; Shou et al 2018). Repair outcomes were unique to each target and determined by protospaceer adjacent motif (PAM) sequence rather than genomic sequence (van Overbeek et al 2016; Shou et al 2018; Chakrabarti et al 2019), and are reproducible and predictable (van Overbeek et al 2016; Shou et al 2018; Chakrabarti et al 2019). Thus, integrated multiomics (sequencing, bioinformatics, transcriptomics, proteomics and metabolomics) can be used to profile the resulting repair-products of CRISPR/Cas9 DSB of specific genes (e.g. disease resistance genes) with the aim of identifying unique and recurring characteristics. These features can serve as unique genetic signatures or molecular markers for detection, tracing and surveillance of GE fish developed by CRISPR/Cas SDN-1 approach, providing important data for RA of food safety. This approach can also be used to identify off-target mutations and their impact on the GE fish.

The SIFO and NBAB reports have divergent interests and methods for asking questions and have thus interpreted the answers somewhat differently. This has been emphasized and criticized (Antonsen et al 2020a, b; Carson et al. 2021). The divergent results, and the criticism of the questions asked indicate that more work is needed in this area, especially for mapping consumer perception and acceptability. We would like to direct a similar reaction to the Chinese survey on public perception by Cui and Shoemaker (2018). Analysing large surveys based on a pre-determined belief that GE organisms are safe to eat and could be made safe for the environment will not give a clear understanding of the public perception. It is important not to write off negative public perception or lack of acceptability as a lack of knowledge amongst the public. The NBAB has attempted such a comparison by asking “control” questions in the survey, in order to test the respondents` level of knowledge. Some of the criticisms in (Antonsen et al 2020a, b) focused on this and how it not only degrades public perception, but also that the control questions used were poorly formulated and misleading. Acceptance of GE organisms in food production including aquaculture is not only a question of risk and physical effects, it is also about human relation to animals and nature, instrumentalization of animals, dignity and characteristics of species, and the […] increasingly imbalanced power distribution between humans and animals” (De Graeff et al. 2019). This should be further emphasized in future surveys where stakeholders´ and public acceptance of GE are studied; see Fig. 1, (6), RRI. As shown in the NBAB study, using technology for improving health in production animals and reducing the environmental impact may be considered more acceptable than other purposes.

Future surveys aiming at mapping public perception should take all this into account. A survey might not, as seen in recent studies (NBAB 2020; Bugge 2020; Cui and Shoemaker 2018) give clear answers to whether the public supports or is opposed to the use of GE animals in food production. Acceptance has several dimensions and aspects which do not necessarily lead to a clear answer for opposing or supporting the technology (Van Eenennam and Young 2018). We encourage future surveys to be aware of this and instead of asking leading questions, focus the surveys so that they are open with the aim to identify what the most pressing concerns are.

Contribution to sustainable development is one of the non-safety criteria in Norwegian Gene Technology Act, but the regulation is not yet translated into guidelines on how to evaluate GE (and GM) fishes. This could be done by looking at global goals and strategies such as UN SDGs and EU Green Deal. Further, a framework for evaluating GE animals should be based on relevant regulatory frameworks and policy documents for animal and fish welfare (Blix and Myhr 2021). In Norway the Animal Welfare Act equalizes terrestrial and aquatic animals, designating them with rights to be protected from harm, stress and strain, in addition to other intrinsic values (Norwegian Animal Welfare Act 2009). Internationally, fish has not been assigned the same status. However, the Animal World Health Organization has developed international standards for fish welfare contained in the Aquatic Code (The Aquatic Code 2019). A framework for evaluating GE fish should use the Aquatic Code, which focuses on how to avoid disease, as a guiding minimum for how welfare is to be understood with regards to fish. In addition, it should also be considered whether the intrinsic value of fish should be taken into consideration. This could be a valuable guiding principle (Trøite and Myskja 2018) for determining whether GE diminishes the integrity of animal; alternatively what kind of GE is acceptable? (Blix and Myhr 2021).