Scientific opinion on the regulation of gene-edited plant products within the context of Division 28 of the Food and Drug Regulations (Novel Foods)


1.0 Executive summary

Health Canada has developed its scientific opinion regarding the regulation of gene-edited plant products within the context of the Novel Food Regulations. This opinion helped to inform the development of the Department's new Health Canada Guidance on the Novelty Interpretation of Products of Plant Breeding and Health Canada Guidance on the Pre-Market Assessment of Foods Derived from Retransformants.

The following summarizes the concepts that are addressed in this scientific opinion:

Risk to consumers is based on product characteristics, not the process used to make the product.

For a hazard to present a food safety risk, there must be exposure to the hazard.

Chemical and physical mutagens used in conventional plant breeding and plant development are hazardous. However consumers are not exposed to these mutagens through their food. Similarly, consumers are not exposed to rDNA and gene editing technologies through their food, rather it is the final characteristics of the food to which consumers are exposed. As such, it is those characteristics that matter in relation to the identification, characterization, and management of food safety.

The ways that new characteristics can present food safety risks are limited and specific.

The ways in which new characteristics might impact human health involve the introduction of new or altered toxins or allergens, impacts on nutritional composition of the food, the presence of anti-nutrients or altered metabolism in the food, or a change in food use as described in Health Canada guidance.

Plant crossing and selection can remove undesirable characteristics, supporting food safety.

Plant breeding has advanced significantly over the last two centuries. However, plant crossing and selection is still applied, even for products of rDNA and gene editing technologies. Crossing and selection are used to select against undesirable characteristics, which include characteristics that could result in food safety risks. In relation to gene-edited plants, crossing and selection can be used to remove DNA sequences encoding the gene editing technologies from the plant, as well as to remove unintentional edits.

When undesirable characteristics cannot be eliminated through crossing and selection, plant developers have a history of removing such products from further development altogether, thereby supporting food safety.

2.0 Introduction

Canada's Novel Food Regulations (Division 28, Part B, of the Food and Drug Regulations) were published in 1999. These regulations provide a definition of a "novel food" and requires that all manufacturers and importers submit a notification to Health Canada prior to the sale or advertising for sale of a novel food. Health Canada reviews this notification to determine if the information provided establishes that the food is safe to consume before it can be sold in Canada.

The Novel Food Regulations were formed on the basis that it is the specific characteristics of an organism (i.e., a plant, an animal, or a microorganism) which have the potential to pose a risk to the safety of its derived foods. As such, the regulations were designed to identify foods derived from organisms as novel or not novel based on their characteristics, and not on the specific methods/technologies used to develop those products. This type of regulatory framework is commonly referred to as a product-based system.

Since these regulations were enacted, technological developments have created new tools of genetic modification by which new plant varieties can be developed. These tools have been collectively referred to as gene (or genome) editing technologies.  These technologies are also commonly referred to as plant breeding innovations (PBI). In Europe, Australia and New Zealand, they are also referred to as new breeding techniques (NBTs) or new genomic techniques (NGTs).

In light of these new tools, plant developers highlighted a need for greater clarity, predictability, and transparency surrounding the novel foods regulatory framework. More specifically, plant developers sought to better understand how products developed using gene editing technologies fit within Canada's product-based system for novel foods.

To address this need, Health Canada developed new guidance to clarify which products of plant breeding, including those developed using gene editing technologies, require a pre-market notification as a novel food.

In developing this guidance, Health Canada regulators examined the science of product-based regulation as it pertains to food safety. This included a review of how current plant breeding practices account for characteristics which may impact food safety during the new variety development. This information supports the risk-informed, product-based approach of the developed guidance.

Lastly, regulators considered how gene-edited plant products fit within Health Canada's product-based regulatory approach to all products of plant breeding. This was done through an examination of the nature of gene editing technologies, how they can be applied as tools of plant breeding, and how potential genetic changes resulting from the use of gene editing technologies compared to the genetic changes which may result from other methods of plant breeding.

Based on this review of the available information, it is Health Canada's scientific opinion that gene-edited plant products should be regulated like all other products of plant breeding, by focusing on their final characteristics and not the method of product development.

While it is understood that the application of gene editing technologies spans many different areas such as medicine (e.g., cell and animal models, human disease therapy, etc.) and other areas of biotechnology (e.g., biofuels and gene-edited animals for food use), the scope of this scientific opinion is limited to the area of plant breeding and the plant products that may be developed using these technologies.

3.0 The science of product-based regulation

All products (including foods) carry some level of risk, but whether or not those risks impact public health or the environment depends on the characteristics of the product (NAS, 2016). To ensure the safety of the products it regulates and to maintain public confidence, the Canadian regulatory system is based on a strong science-based approach.

While there are a variety of approaches to the regulation of agricultural products internationally (NAS 2016), Canada has taken a product-based approach. Under this approach, it is the final characteristics of the product, rather than the method used to make the product, that determine the level of regulatory oversight (NAS, 1987; NAS, 2016). This is the most scientifically supported regulatory approach as recognized by the United States National Academies of Science (NAS, 2016), and the United States National Research Council (NRC, 2000; NRC, 2002).

The scientific basis underpinning a product-based regulatory system is that there has to be a hazard that people or the environment are exposed to for there to be a risk to consider or manage (NAS, 2016). As the public and the environment are exposed to the final characteristics of a product, not the technology used in its production (NAS, 2016), it is logical to regulate on the basis of those final characteristics. Therefore, even though certain technologies used in plant breeding, like ethyl methanesulfonate (EMS), are extremely toxic (Müller et al., 2009), the hazards these technologies present are not relevant to a product's risk assessmentFootnote 1, unless they are retained within the product itself (e.g., EMS was present in the product). These concepts will be explored in more detail in subsequent sections, with an in-depth consideration of gene editing technologies, and their relationship with modern plant breeding.

An often cited concern is that the science of modern technologies, like recombinant DNA (rDNA) and now gene editing, is relatively new and uncertainties remain regarding the impact these technologies might have on human health and the environment (CBAN, 2020). While this may have been true in the past, and though it is true that the science of any area of study is incomplete (Wagner, W. 1995), there is now over 25 years of regulatory experience evaluating the risks that rDNA products might have on human health (Health Canada, 2021b). To date, there have been no adverse health effects directly attributable to rDNA technology, or from rDNA derived foods (NAS, 2016; EC 2010; Nicolia et al., 2014; Pellegrino et al., 2018). Moreover, rDNA technology has been a practical area of scientific study for almost 50 years (Jackson et al., 1972), and as a result of this extensive experience, many of the original uncertainties are now well characterized, and the unintentional effects of this technology have been established to be no different than those that occur through the use of conventional breeding technologies in regards to product safety (Schnell et al., 2014).

As it relates to gene editing; though these technologies are comparatively new, they have been some of the most intensively studied, and characterized scientific tools of the modern era. This is due to the wide application of these technologies in all domains of biological science, from agricultural biotechnology (Feng et al., 2013) to medicine (Doudna, 2020), spanning all areas of primary research to their direct application in product development and commercialization (Urnov et al., 2010; Joung and Sander, 2013; Adli, 2018; Gao et al., 2020). As will be explained in later sections, the consequence of this intense study of this comparatively new set of technologies is that, like rDNA, robust, scientifically supported insight about both the intentional and the potential unintentional effects of these technologies on the resultant products can be made. In cases where elements of these technologies are present in the final product, existing scientific risk assessment tools used to study and evaluate product characteristics are still effective in characterizing the resulting risk profile of the product.

Based on Health Canada's review of the current evidence, as outlined in the remainder of this document, maintaining a product-based regulatory system remains scientifically sound. Furthermore, as will be demonstrated, it is reasonable to take a product-based approach which accounts for both the product characteristics and the established scientific understanding of how food safety risks can present in new products of agricultural biotechnology (NAS, 2016; NRC 2002, NRC, 2000).

3.1 Risk-based regulatory triggers for pre-market oversight

Risk-based regulatory triggers are what determine if a product should undergo pre-market risk assessment (NAS, 2016), whereas risk assessment (i.e., the pre-market safety assessment itself) is the detailed, data-driven exercise conducted for certain new foods before they are allowed to enter the food supply (NAS, 2016). Though risk-based regulatory triggers and product-based risk assessments are related concepts, they are distinct and separate activities. As this document focuses on the scientific basis for Health Canada's guidance regarding the novelty interpretation of products of plant breeding, the following discussion centers around the risk-based pre-market regulatory trigger. For more detailed information on how Health Canada conducts pre-market assessments of novel foods, the Department has published detailed guidance (Health Canada, 2006). This guidance is based upon the internationally accepted framework established by the Codex Alimentarius Commission (CAC, 2003a; CAC, 2003b) for the assessment of foods derived from biotechnology.

Plants are complex living organisms that are made up of components that have beneficial nutritional value from oils (Dyer et al. 2008), to starch (Jennings, 2019) and protein sources (De Ron et al. 2017), to biologically active phytochemicals that have medicinal properties (Balunas, Kinghorn 2005; Arnason et al., 1981). However, plants can also accumulate toxicants (Spencer and Berman, 2003), allergens (Breiteneder et al., 2004), or anti-nutrients (Burlingame et al., 2009). In spite of these hazards, people have safely consumed countless wild and domesticated plant varieties for nutritional, and, in some cases, medicinal benefit, for thousands of years (Kingsbury, 2009; Singer et al., 2021; Conko et al., 2016; Constable et al., 2007). Based on this information, government regulators must determine which plant products trigger pre-market regulations and require pre-market assessment.

In accordance with risk-based regulation, products that trigger pre-market review should have new characteristics that pose a food safety hazard to which people will be exposed (NAS, 2016). On this basis, some new products do not require pre-market safety assessment because they do not exhibit hazardous characteristics in relation to what is already available as food. Furthermore, even if a new product demonstrated new hazards, there is not always exposure to those hazards, and as such there is minimal risk to public health.

A good example of characteristics that do not pose a new hazard, is the regulatory status of Next Generation Waxy Corn in Canada (Health Canada, 2021a). Next Generation Waxy Corn is a gene-edited corn variety whose genome is edited to remove a gene involved in amylose biosynthesis with no other genomic alterations (Gao et al., 2020). The result of this genome edit is a corn variety which generates 100% amylopectin in corn kernels. This characteristic is substantially equivalent to waxy corn varieties that are already part of food (Huang et al., 2020), and as a result, this new characteristic does not justify pre-market risk assessment in accordance with a risk-based approach to regulation.

A good example to illustrate the latter principle (i.e., lack of exposure to a hazard) is the consumption of stone fruits, like apricots and peaches. Stone fruits have cyanogenic glycosides in their pits that serve as a defense against herbivory (Geller et al., 2006; Gleadow et al., 2014; Vetter, 2000). For consumers, cyanogenic glycosides are food safety hazards because they hydrolyze to release hydrogen cyanide, which can cause cyanide poisoning in high enough amounts (Geller et al., 2006; Akil et al., 2013). However, people generally do not eat the pits of stone fruit. Therefore, while stone fruits contain cyanide-producing molecules that are hazardous to consumers, the location of those molecules in the fruit means that people are not expected to be exposed to them. Therefore, as long as consumers do not eat the pit, there is no food safety risk from the cyanogenic glucosides in stone fruit.

The absence of exposure to a hazard eliminating the food safety risk can be extended to explain why it is a product's final characteristics, and not the technology used in its development, that informs the level of appropriate regulatory oversight. As was mentioned before, certain technologies that have been used in plant development for a century, like ethyl methanesulfonate (EMS) (Müller et al., 2009), and physical mutagens (e.g., electromagnetic radiation) (Ahloowalia et al., 2004), can pose genotoxic hazards. However, the public is not exposed to EMS or the physical mutagens used in plant development when they consume those mutagenized plant varieties. By consequence, there is no risk of genotoxicity to consumers from these mutagenized plant lines. This principle is further supported by primary literature which has reported the global release of over 1650 plant varieties developed using physical and chemical mutagens (Maluszynski et al., 2000; Ahloowalia et al., 2004), with no known adverse public health effects that can be attributed to these conventional plant breeding technologies (NAS, 2016).

The ways in which new characteristics may impact food safety are expressed by the Codex Alimentarius Commission (CAC, 2003a), in the primary literature (Delaney et al., 2018), and are already reflected in how pre-market safety assessments of novel foods are conducted in Canada (Health Canada, 2006). These resources indicate that regulators should take into account both intended and unintended effects of a new product characteristic, identify whether those effects may create new or alter existing hazards, and determine whether those new or altered hazards are relevant to human health (CAC, 2003a; CAC, 2003b). The ways in which new characteristics might impact human health involve the introduction of new or altered toxins or allergens, impacts on nutritional composition of the food, or the presence of anti-nutrients or other factors that might influence food nutrition (CAC, 2003a). As a result, it is scientifically defensible to incorporate these criteria in determining whether a product has a new characteristic that requires pre-market safety assessment or not, and not just in structuring the risk assessment of these products (NAS, 2016).

4.0 Plant breeding

Building upon the discussion of the scientific basis of the product-based regulatory approach, it is important to explore the history and basic principles involving plant breeding to understand how products of plant breeding are generated, and the ways in which plant breeding practices can manage risks, as well as the limits of these practices to manage risks.

Plant breeding is the science of continually improving desirable characteristics and reducing or eliminating undesirable characteristics in plants. What is considered desirable or undesirable for a plant is a function of the producer's and consumer's needs for the plant. Generally speaking, characteristics that improve the reliability and fitness of a plant for a given area of cultivation, or characteristics that improve and standardize the harvest and processing of the plant, are desirable characteristics for the producer (Kingsbury, 2009). For consumers, characteristics that improve appearance, flavour and reduce the cost of the plant at market are also seen as desirable (Kingsbury, 2009). For both the producer and consumer, yield is a desirable characteristic, especially in circumstances where demand for the plant might outstrip supply (Kingsbury, 2009).

4.1 From plant selection to plant tissue culture

Human populations have been manipulating plant characteristics for over 10,000 years (Purugganan et al., 2009).  Plant domestication started with intuitive and later deliberate selection where plants with desirable characteristics were identified from wild varieties or partially domesticated cultivars (Kingsbury, 2009; Purugganan et al., 2009). The selected plant with a desired characteristic was often a naturally occurring mutant within the existing plant variety or a hybrid resulting from an unintentional cross between related species (Kingsbury, 2009). It was not until the 1800s when sophisticated, plant cross-breeding programs began, and these cross-breeding programs were made more predictable by the rediscovery of Mendel's Laws of Heredity in the early 1900s (Kingsbury, 2009).  At this point in plant breeding history, there was significant interest in accelerating the rates at which desirable plant characteristics could be created and identified (Kingsbury, 2009). However, this was limited by natural forms and rates of genetic mutation within sexually compatible plant populations (Kingsbury, 2009). With the discovery of chemical- and later radiation-based mutagenesis, the rate at which useful characteristics could be generated was greatly accelerated above the natural rates of mutation (Singer et al., 2021; Graham et al., 2020). Furthermore, with improved understanding of plant sexual reproduction, and later tissue culture techniques, the genetic range from which desirable characteristics could be sourced was also expanded (Kingsbury, 2009).

4.2 Recombinant DNA and gene editing

For much of the twentieth century, following the advent of chemical- and radiation-based mutagenesis, plant breeding advanced through improved understanding of plant genetics, plant tissue culture techniques, and the development of tools to accelerate plant cross-breeding programs (Kingsbury, 2009; Moose and Mumm, 2008; Evenson and Gollin, 2003). The development of recombinant DNA (rDNA) technology, and more recently gene editing technologies, significantly changed the sources of new characteristics that plant developers can work with. With these technologies plant developers can source desirable characteristics from beyond sexually compatible plants (Kingsbury, 2009; Slater et al., 2008; Klee et al., 1987; Sanford, 1990).

These technologies have the potential for introducing new useful characteristics to plants that can't be introduced using conventional plant breeding technologies, however, there remain some limitations. For example, it's often stated that rDNA, and now gene editing, will let plant developers generate new plants and market them faster than conventional plant breeding technologies (Niler, 2018). Though there are examples of this occurring (Gao et al. 2020), these examples are more the exception rather than the rule for rDNA and gene editing. The reason for this is that rDNA and gene editing technologies depend on plant transformation and plant tissue culture techniques and not all plants, or plant tissues are receptive to these tools (Slater et al., 2008; Altpeter et al., 2016; Jones, 2009; Birch, 1997).

While desirable characteristics can now be sourced from beyond sexually compatible plants, these characteristics can only be introduced into specific plant varieties. More often than not, these receptive varieties are not what is sold to producers to grow for food because these varieties do not perform well enough in agricultural settings. As a result, following plant transformation or gene editing, the plant developer will cross the transformed or gene-edited plant with commercial plant varieties over multiple generations to move the new characteristic into a commercial plant variety. Though this process takes a significant amount of time, and undercuts the full potential of these technologies in quickly delivering new products to market, there is a benefit of this approach from a food safety perspective. The standard practices of plant breeding are used to remove undesirable characteristics that could affect food safety, and which could include unintended changes that result from rDNA or gene editing (Shimelis and Laing, 2012; Hickety et al., 2019; Allard, 1999).

Through this evolution of plant breeding technologies, plant developers have overcome the limitations of natural rates of mutation, greatly accelerating these rates to produce new desirable characteristics. They have also overcome the limitations imposed by plant reproductive barriers, to be able to access desirable characteristics beyond sexually compatible plants. However, these achievements have not been without certain limitations imposed by plant tissue culture and the receptivity of plant varieties to modern technologies. A positive outcome of these limitations is that standard practices of plant breeding continue to be used in most plant development programs to produce products safe for food use, including those plants developed using rDNA and gene editing (NAS, 2016; Prakash, 2001; Delaney et al., 2018; Conko et al., 2016).

5.0 Gene editing technologies and plant breeding

Gene editing technologies refer to biotechnology tools that can be used to generate specific modifications to the genome of living organisms by adding, removing, or altering genetic sequences at precise locations. Gene editing can accomplish the same objectives as conventional breeding technologies (Lassoued et al., 2019; Mao et al., 2019; Parry et al., 2009), and rDNA technologies (Fichtner et al., 2014), but with higher precision, and with no more unintentional genetic changes than those other technologies (Chen et al., 2019; Graham et al., 2020).

There are a number of tools described as a 'gene editing technology'. At present, it is mainly used to describe CRISPR-Cas, but it also refers to Oligonucleotide Directed Mutagenesis (ODM), Transcription Activator-like Effector Nucleases (TALENs), Zinc-Finger Nucleases (ZFNs) and meganucleases, as well as variations of these technologies. In addition to the references provided below, numerous other extensive reviews and perspectives have been published on these technologies (Doudna and Charpentier, 2014; Komor et al., 2017; Hua et al., 2019). These technologies continue to evolve with improvements to precision, efficiency and to identify experimental conditions to allow for gene editing of new plant species (Young et al., 2019; references within Graham et al., 2020).

To date, gene editing has been successfully used to modify several types of plants, including but not limited to: sugarcane, citrus fruits, grapes, wheat, tobacco, potato, flax, cotton, and maize (references within Zaman et al., 2019 and Songstad et al., 2017), and it is expected that plant developers will use gene editing to improve many other plants for food use, among other uses.

5.1 Oligonucleotide directed mutagenesis

Oligonucleotide Directed Mutagenesis (ODM) uses specifically designed oligonucleotides that have two important features; their sequences nearly match a target DNA sequence while harboring a nucleotide mismatch. During ODM, the oligonucleotide anneals to the target DNA site, triggering mismatch repair, resulting in the specified DNA change that corresponds to the mismatched nucleotide (Songstad et al., 2017). This technique, and variations on it, have been used to gene edit plants including tobacco and maize (Zhu et al., 1999; Beetham et al., 1999).

5.2 Meganucleases, TALENs and ZFNs

Meganucleases are naturally occurring enzymes that can cleave DNA through recognition of specific target sequences that are 12-40 nucleotides in length (references within Songstad et al., 2017; Takeuchi et al., 2015). While the large recognition sequence limits target site selection options, this characteristic also results in highly specific meganuclease targeting, which lowers the probability of off-target edit sites being present in a given genome. Meganucleases have been used to gene edit plants including cotton and maize (D'Halluin et al., 2013; Gao et al., 2010), though none have been developed for commercial purposes.

TALENs and ZFNs technologies use programmable, sequence-specific DNA-binding proteins to target a genomic location, which is then cleaved by the DNA nuclease domain of the TALEN or ZFN to induce a double-stranded break. Following this, the cell repairs this DNA break by one of two possible mechanisms: template-independent non-homologous end joining (NHEJ) or template-dependent homology-directed repair (HDR). TALEN technology has been used to generate a variety of gene-edited plants including potato (Clasen et al., 2016), soybean (Haun et al., 2014), tobacco (Li et al., 2016), and sugarcane (Jung et al., 2016). Additionally, ZFN technology has been used to produce gene-edited plants such as tobacco (Townsend et al., 2009), Arabidopsis thaliana (Zhang et al., 2010), and maize (Shukla et al., 2009).

5.3 CRISPR-Cas

The CRISPR-Cas gene editing system has quickly become a popular gene editing technology among scientists and product developers (Hsu et al., 2014; Schaeffer and Nakata, 2015; Zaman et al., 2019; Knott and Doudna, 2018). Its popularity comes from the system's simplicity in terms of design and technical execution, cost to acquire and deploy, and the potential to precisely modify the genome of any living organism (Knott and Doudna, 2018).

The CRISPR-Cas system is a naturally-occurring, acquired viral pathogen defence mechanism in archaea and bacteria (Bhaya et al. 2011; Hill et al. 2018). The naturally-occurring system is composed of three components: the CRISPR associated (Cas) endonuclease, the CRISPR RNA (crRNA), and the transactivating CRISPR RNA (trRNA) (Bhaya et al., 2011; Jinek et al., 2012). The engineering of this viral defense system into a programmable gene editing tool involved reducing it to a two-component system comprising the Cas endonuclease, which cleaves DNA, and a guide RNA (gRNA), which can be designed to target nearly any DNA sequence (Jinek et al., 2012).

In delivering the endonuclease and gRNA to living tissues, plant developers have a variety of approaches at their disposal. They can deliver the functional endonuclease/gRNA complex, or they can deliver plasmids encoding the endonuclease and gRNA sequences to living plant cell cultures (Mao et al., 2019). The end result of these approaches being a transient presence of gene editing tools in living tissues.  Alternatively, plant developers can stably transform living plant cells with endonuclease- and gRNA-encoding sequences using particle bombardment or Agrobacterium mediated transformation (Mao et al., 2019). The end result of these approaches being a plant cell line that possesses genomic sequences encoding and expressing the gene-editing tools.

In terms of editing DNA, the gRNA binds to its complementary genomic target and forms a complex with the Cas endonuclease at the target DNA site. Formation of this complex leads to double-stranded DNA cleavage at an upstream location called a protospacer adjacent motif (PAM). The PAM is an essential, but constraining DNA sequence motif near the target DNA sequence (Jinek et al., 2012; Schaeffer and Nakata, 2015; Kleinstiver et al., 2015; Walton et al., 2019). Following DNA cleavage, the cell will then repair the site using NHEJ or HDR. NHEJ repairs the cut site without a template, and typically generates small insertion/deletions of a few nucleotides. By inserting or deleting a few nucleotides at the target site, the resulting DNA sequence can alter gene function or expression. Alternatively, the cell can use HDR to repair the cut site by copying the sequence from a homologous donor template, which is introduced along with the gene editing tools, and which introduces precise changes to the genome at the targeted site.

With so much interest in the CRISPR-Cas gene editing system, there have been efforts to identify natural Cas endonuclease homologs, or generate engineered variants that recognize different PAM sequences, thereby lessening constrains on target site selection for CRISPR-Cas gene editing (Kleinstiver et al., 2015; Walton et al., 2019; Knott and Doudna, 2018; Murugan et al., 2017). Beyond expanding recognized PAM characteristics, natural or engineered varieties of Cas endonucleases with modified endonuclease functions have been identified, further expanding the versatility of the tool (Knott and Doudna, 2018; Murugan et al., 2017).

The identification and development of Cas variants have expanded the CRISPR-Cas toolbox, and the utility of this system has been illustrated by modified CRISPR technologies such as base-editing and prime editing, which do not use DNA double-strand breaks or a repair template for targeted genome modification. Base editing methods use a modified CRISPR/Cas9 tethered to a deaminase enzyme. An A or C deamination event at the target site generates a base mismatch, which is repaired and results in a substitution (C>T/G>A or T>C/A>G) at the target base (Rees and Liu, 2018). Unlike base editing, prime editing does not require precisely-positioned PAM sites (Anzalone et al., 2019) for targeted modification. The prime editing system uses an impaired version of Cas9 fused with a reverse transcriptase that is programmed by a prime editing guide RNA (pegRNA), which specifies both the target site and the desired edit. Prime editing can be used for single base substitutions, as well as multiple base substitutions, transversions (purineàpyrimidine or pyrimidineàpurine), insertions, and deletions (Anzalone et al., 2019).

6.0 How can gene editing technologies be applied to plant breeding?

Plant breeding depends on genetic variation for identification and selection of useful characteristics in plants. In conventional breeding programs, this variation has evolved and continues to evolve naturally over time, and/or is introduced to the species by means of wide crosses (McCouch, 2004), tissue culture (Jain, 2001), or the application of chemical and/or physical mutagens (Oladosu et al., 2016). Plant developers also employ rDNA technology, where genetic sequences are inserted into a plant genome using biotechnology to expand upon the natural genetic diversity in the plant species.

Gene editing has already been used in research settings to improve agronomic characteristics including: increased quality and yield (references within Zaman et al., 2019), increased resistance to bacterial pathogens (Peng et al., 2017), increased resistance to fungal disease (Wang et al., 2014), increased resistance to viral pathogens (Ali et al., 2015), and introduced tolerance to herbicides (Butler et al., 2016). In terms of modified composition, gene editing has been used to adjust the carbohydrate composition of corn kernels (Gao et al., 2020). Plant developers have indicated that gene editing will be used to introduce new characteristics and change existing characteristics in commercial plant varieties to improve agronomic performance (e.g., plant height, size, yield, environmental stress tolerance, disease and pest resistance) and/or to modify composition and nutritional quality and/or other qualities linked to consumer preference (Zhang et al., 2018; Cui et al., 2020; Wang et al., 2019; Ogbonnaya et al., 2017; P R, Singh et al., 2016 and references within Tshikunde et al., 2019). As a potential application for consumer preference, gene editing has been used to reduce the browning of mushrooms (Waltz, 2016).

The ways plant developers will use gene editing technologies to accomplish these objectives could involve the introduction of small insertions or deletions at precise genomic sites, precise deletions of larger segments of genomic DNA, as well as insertions of whole genes and their regulatory elements (Chen et al., 2019; Zhang et al., 2018). Furthermore, plant developers have indicated that gene editing can help identify useful characteristics in regions of plant genomes that developers currently have difficulty manipulating using conventional breeding technologies. As a result, gene editing has the potential to identify and improve new characteristics within the plant species (Van Eck, 2020; Rönspies et al., 2021).

6.1 Creating genetic variation in plants using gene editing

Gene editing technologies are the most recent tools that have been developed to expand the genetic variation that plant developers need to meet the demands of producers and consumers and assist with global food security and sustainability objectives. How gene editing differs in relation to conventional and rDNA-based plant breeding genetic variation is that the gene-edited variation can be introduced at precise, predetermined locations within the plant genome (Chen et al., 2019). The precision of gene editing can simplify the food safety evaluation of a new, gene-edited plant and therefore offers the possibility for new products to be developed and commercialized in a timely and efficient manner.

7.0 Gene editing technology – Key considerations for regulators

The following section will discuss how gene editing technologies are delivered to living cells, the unintended effects in plant genomes that can occur using gene editing, how the components of gene editing technologies can, in some cases, become part of the final plant product, and other tools that plant developers currently use to ensure safety of their products.

7.1 Delivering the gene editing technologies to living cells

Like all other forms of plant biotechnology, gene editing technologies need to be delivered to living cells to work. Where gene editing technologies differ from conventional and rDNA plant development tools is in the way that DNA sequences encoding the components of gene editing technologies can, in some cases, become part of the product, and therefore the technology itself can contribute a new characteristic to the plant (Wolt et al., 2016; Zhao and Wolt, 2017).

It is important to note that although this is one way to apply gene editing, it is not necessary to include these DNA sequences in the plant genome to successfully gene edit the plant. Given the choice, DNA-free gene editing (Tsanova et al., 2021) appears to be the preferred approach by plant developers. This was expressed during an October 16, 2020 Expert Panel meeting (Health Canada, 2021c) between government regulators and plant breeding experts from both academia and industry.  At this meeting it was explained that should a plant developer need to integrate gene editing sequences in to the plant genome, developers are unlikely to advance this particular product to commercialization without having subsequently removed these sequences.

Should a plant developer advance a gene-edited plant to commercialization where the DNA sequences encoding the gene-editing technology are retained in the plant genome, then this product would be identified as containing foreign DNA, and therefore foods derived from this plant product would be considered novel (i.e., require a pre-market safety assessment).

7.2 Unintended effects that result from off-target edits

Off-target edits are genetic changes that result from the gene editing technologies working at genomic sites other than the intended edit site (Wolt et al. 2016). This can occur because gene editing technologies target genomic sequences by protein-based (Bedell et al. 2012; Bibikova et al. 2003) or guide-RNA sequence-based DNA recognition (Doench et al. 2016). As a result, genomic sequences that share sequence similarity to the intended edit site may also be edited by gene editing technologies.

It is also important to note that unintended genetic changes can occur in close proximity to the intended targeted sequence. These are sometimes referred to as unintended "on-target" changes. Though the locations of these changes could be argued to be 'on-target', the ways in which developers manage these sorts of unintentional changes are the same as off-target edits, and therefore the principles described in relation to off-targets apply to these unintentional on-target edits.

As product developers will be primarily targeting genetic sequences that will alter biological processes with the intention of re-creating existing, or generating new or altered characteristics in plants, there is a possibility that off-target edit sites may inadvertently impact secondary biological processes, thereby introducing unintended characteristics and altering the risk profile of the plant-derived foods (Wolt et al., 2016; Zhao and Wolt, 2017). The first principle supporting this contention is that similar sequences, and certainly sequence homology between genetic sequences, can be predictive of biological structure and function (Gotz et al., 2008). Indeed, sequence similarity and homology are already used to evaluate for possible allergenic or toxic sequences in the pre-market safety assessment of novel foods (Health Canada, 2006) as sequence similarity and homology are useful and predictive when evaluating product safety.

As a matter of context, conventional breeding technologies, as well as rDNA technology, can also introduce unintended genetic changes which may result in unintended characteristics, and thus alter the risk profile of the plant-derived food. Health Canada and the Canadian Food Inspection Agency (CFIA) have previously determined that these sources of unintended genetic changes do not pose higher risks than other plant development technologies and therefore do not require a higher level of scrutiny (Schnell et al., 2014).

It is important to note that because off-target edits can be predicted, they can also be mitigated with good gene editing tool design, selection, and experimental conditions. Furthermore, the presence of off-target edits can be directly investigated using targeted DNA sequencing (Wolt et al., 2016; Liang et al., 2017); something that cannot be reasonably accomplished for random mutations. Additionally, scientific studies evaluating off-target edits suggest that, though they are possible, they rarely occur even at predicted off-target sites in plants (Young et al., 2019, Hahn and Nekrasov, 2019 and reviewed in Fichtner et al., 2014). And should off-target edits be present in a gene-edited plant, plant developers are able to remove them from most plants using breeding and selection, and/or backcrossing (Kaiser et al., 2020).

As a result of these considerations, any risks posed by the occurrence of an off-target edit can be well managed through proper characterization and existing plant breeding practices.

7.3 Gene editing technologies – one of many tools plant developers use together to create new plants

Gene editing technologies are unlikely to be used in isolation when developing a new plant for reasons that were discussed previously (Section 4.2). Indeed, with the exception of some vegetatively propagated plants (e.g., sugarcane, potato, apples, etc.), gene-edited plants will likely undergo further breeding by crossing and selection to move the gene-edited characteristic from the original, gene-edited plant variety to one or more commercial production varieties that will be grown for food. As part of this step, plant developers use conventional breeding technologies to create large populations of plants from which the majority are eliminated and only a few plants are selected and retained for further development (Allard, 1999). Through this process of crossing and selection, plant developers eliminate undesirable characteristics from the population while advancing desirable characteristics (Kaiser et al., 2020; Glenn et al., 2017). Much of this field trial work occurs over the course of five or more years, and in diverse growing environments to evaluate the agronomic performance and plant quality (Kaiser et al., 2020).

To help with these activities, plant developers generate characteristic-linked genetic markers to follow the new characteristic as it is moved from the original gene-edited plant into the commercial plant varieties that will be grown for food (Kaiser et al. 2020). Plant developers have also established genome-wide genetic markers for their commercial plant varieties. With these genome-wide markers, the number of backcrossing required to reconstitute the commercial plant variety, with the new gene-edited characteristic, can be evaluated. Furthermore, with these genome-wide markers, developers can select for factors that contribute to the agronomic, consumer preference, and nutritional characteristics of the plant. Lastly, for plants known to naturally produce toxins, these genome-wide markers can help developers manage the levels of plant-produced toxins, thereby improving food safety (Kaiser et al., 2020).

In the limited circumstances where and undesirable characteristic cannot be removed due to close genetic linkages with the desirable characteristic's genetic locus, or other confounding factors, plant developers have a history of removing such products from further development (Prakash, 2001; Conko et al., 2016).

8.0 Pre-market and post-market product safety

The ways in which product characteristics can influence food safety have been well reported and discussed in the primary literature (CAC, 2003a; CAC, 2003b; Delaney et al., 2018). As a result, it is possible to identify clear risk-based criteria as to whether a product should require pre-market regulatory oversight based on these characteristics.  Products of plant breeding that possess genetic modifications that; (1) do not alter an endogenous proteinFootnote 2 in a way that introduces or increases similarity with a known allergen or toxin relevant to human health; (2) do not increase levels of a known endogenous allergen, a known endogenous toxin, or a known endogenous anti-nutrient beyond the documented ranges observed for these analytes in the plant species; (3) do not have an impact on key nutritional composition and/or metabolism; and (4) do not intentionally change the food use of the plant have no new plausible food safety hazard, and therefore do not justify pre-market assessment. Products of plant breeding which do not meet these criteria may present a food safety hazard and thus justify pre-market regulatory oversight on a plausible risk basis. In these cases, Health Canada would conduct pre-market assessments according to the detailed guidance that the Department has published (Health Canada, 2006). This guidance is based upon the internationally accepted framework established by the Codex Alimentarius Commission (CAC, 2003a) for the assessment of foods derived from biotechnology.

However, regulatory requirements that support food safety extend beyond pre-market regulatory requirements. There are post-market requirements established in federal regulation for all food products that provide further controls to ensure food safety. For example, the Food and Drugs Act (Food and Drugs Act, R.S.C., 1985, c. F-27, Part I, 4(1)) requires that all food must lack poisonous or harmful substances, be fit for human consumption, must not be adulterated and be stored appropriately. Foods that do not satisfy these conditions cannot be sold, and must be removed from the market. Furthermore, in relation to foods derived from plant sources, for the 53 crop kinds subject to registration in Canada, the Seeds Regulations (Seed Regulations, C.R.C., C. 1400, Part III, Variety Registration 74. (a)) effectively establishes that if a variety may be detrimental to human or animal health and safety or the environment, the Registrar is required to cancel that registration for cause and order the removal of seed of that variety from the marketplace.

As a result, with the overlapping pre- and post-market regulatory oversight controls in place as part of federal regulations, combined with industry-led plant variety performance monitoring, Canada retains a robust food safety system to ensure that Canadians have access to safe and nutritious food.

9.0 Conclusions

The identification, characterization, and management of public health risks from food functions best when the focus of regulatory programs is on the final characteristics of products rather than the technologies used to make those products. The reason for this is that for a hazard to present a public health risk, the public must be exposed to the hazard. As is the case of chemical and physical mutagens used in plant development, so is the case of rDNA and gene editing technologies in regards to food safety. In none of these cases are people typically exposed to these technologies through food, rather it is the final characteristics of the food to which consumers are exposed.

Furthermore, while plant breeding has advanced significantly over the last two centuries, the basic plant breeding technologies of crossing and selection are still applied, even for products of rDNA and gene editing technologies. This matters in regards to managing food safety risks because conventional plant breeding technologies, like crossing and selection, are used to select against undesirable characteristics that could result in food safety risks. In rare instances where the undesirable characteristics cannot be eliminated due to genetic linkage or other confounding reasons, plant developers have a history of removing such products from further development altogether, thereby supporting food safety.

As such, it is the scientific opinion of Health Canada that gene editing technologies do not present any unique or specifically identifiable food safety concerns as compared to other technologies of plant development. Therefore, gene-edited plant products should be regulated like all other products of plant breeding within the Novel Food Regulations (i.e., by the characteristics they exhibit and how these characteristics impact food safety).

10.0 References

Footnote 1

Within the context of novel foods, Health Canada uses the term "pre-market safety assessment" interchangeably with the term "risk assessment".

Return to footnote 1 referrer

Footnote 2

Note: endogenous proteins which demonstrate similarity with a known allergen or toxin prior to genetic modification are excluded from this category unless the degree of similarity is increased.

Return to footnote 2 referrer

Page details

Date modified: