Both the U.S. Department of Agriculture (USDA) Animal and Plant Health Inspection Service (APHIS) and the U.S. Food and Drug Administration (FDA) have asked the public for comments on how these agencies might deal with crops and foods derived from new plant varieties produced using new genome editing techniques like CRISPR-Cas9. Public comments are due to both agencies by June 19, 2017 and can be submitted to USDA APHIS here and to FDA here.

I submitted the following comments to these agencies in response to the notices published by APHIS and FDA in the Federal Register last January.

June 16, 2017: Comment to APHIS Re: Proposed revisions to the regulations at 7 CFR part 340

My name is Belinda Martineau, I am trained as a plant molecular geneticist and I was a part of the team of scientists at Calgene, Inc. that carried out safety studies of the Flavr Savr^TM tomato in Calgene’s successful effort to gain U.S. regulatory approval for commercializing that genetically engineered (GE) crop product. I thank USDA APHIS for providing this opportunity for the public to comment on the agency’s proposed revisions to the regulations applying to GE crops and for being transparent throughout this process.

It is long past time to make the piecemeal and loop-hole-filled “coordinated” framework for regulating GE crops and other organisms more science-based and more coordinated among the U.S. regulatory agencies involved. Rather than “concern in the 1980s that if an organism was modified using genetic material taken from a plant pest, or a plant pest was used as a vector or vector agent to carry genetic material into an organism, the resulting GE organism could also be a plant pest,” APHIS has used its authority over plant pests to regulate GE organisms over the last three decades because it appeared to be its only means for doing so. The best option for regulating GE crops, back then and now, would be to pass specific legislation for regulating GE crops, rather than to continue forcing a round technology into a square and scattered regulatory hole—an option that would have to be taken by the legislative branch of our government. APHIS mentioned several putative options related to regulating GE crops that would require new legislation in the Federal Register (Vol. 82, No. 12); a complete overhaul of GE crop regulations via legislation should also be considered.

In the meantime, APHIS now considers it “both appropriate and necessary to begin to evaluate GE plants for noxious weed risk.” And, based on the definition of a noxious weed provided in the 2000 Plant Protection Act (PPA), i.e. ‘‘Any plant or plant product that can directly or indirectly injure or cause damage to crops (including nursery stock or plant products), livestock, poultry, or other interests of agriculture, irrigation, navigation, the natural resources of the United States, the public health, or the environment,’’ this aspect of APHIS authority could be used to regulate most GE crops. (NB: no mention of “invasiveness” is made in the PPA definition of a noxious weed.) For example, APHIS now knows that deregulating glyphosate-resistant GE plants has indirectly caused substantial damage to various crops and other interests of agriculture in tens of U.S. states through the development of glyphosate-resistant superweeds. And based on the scientifically based concept of selection pressure, there is every reason to believe that deregulation and consequent wide cultivation of any other herbicide-resistant GE crops will similarly result in development of superweeds resistant to those herbicides and thus result, indirectly, in similar damage to crops and other interests of agriculture. Thus, based on the 2000 PPA, any herbicide-resistant GE crops should be regulated by APHIS; i.e. the regulatory status of herbicide-resistant GE plants should definitely not be “not regulated” (as they are currently on APHIS’s example list).

Additionally, glyphosate-resistant plants could indirectly injure livestock and/or public health since glyphosate is an animal carcinogen and a probable human carcinogen. Therefore, glyphosate-resistant plants should be categorized as noxious weed risks and should be regulated.

In a truly “coordinated” framework for regulating GE crops, APHIS should take the damaging indirect effects of herbicide-resistant GE plants into account when evaluating GE crops for noxious weed risk as per the 2000 PPA definition, and not simply categorize damage associated with application of an herbicide as a concern for the EPA instead. And the “asynchronous timing of the deregulation of herbicide-resistant plants and the associated herbicide registration” must simply end; if APHIS and EPA can’t coordinate such timing for GE crops and the herbicides or other chemicals designed to be specifically applied to them, the “coordinated” framework is broken.

Based on the 2000 PPA definition it also appears that most, if not all, GE crops could be classified as noxious weed risks in terms of their ability to indirectly cause damage to “other interests of agriculture,” such as the economic interests of U.S. farmers who grow non-GMO crops. When the crops of these farmers become intermixed with GE crops, via cross-pollination or post-harvest, economic losses in sales within and outside of the U.S. have been, and can be again in the future, substantial. USDA has expressed concern over this intermixing problem in the past. And the as yet unconfirmed possibility that a GE corn crop developed for biofuel may have been intermixed with corn used for masa and tamale production last winter (http://www.huffingtonpost.com/entry/gmo-ethanol-corn-contamination-raises-concerns-about_us_58e52857e4b0ee31ab9533dd ) serves as an example of how this intermixing problem could cause economic losses to “other interests of agriculture”…or worse. Therefore APHIS should utilize the PPA of 2000 to regulate GE plants/crops to protect against damage to important “other interests of agriculture” in the U.S.

I agree with APHIS’s definitions of genetic engineering and GE organism, but not with its exclusion “from its definition of GE organism, certain organisms that are created using techniques that fall within the scope of genetic engineering….” It is inappropriate and not scientifically sound to minimize concern over the off-target effects which occur as a result of the use of the new genome-editing types of genetic engineering techniques, as well as those that can occur when using the older Agrobacterium-based and biolistics techniques. APHIS states that chemical and radiation-based mutagenesis can create “thousands of mutations in a single organism, and most of the plant breeders’ subsequent efforts involve eliminating unwanted mutations by repeated crosses and selection….” But an off-target activity rate of ≥50% has been reported in the scientific literature for the CRISPR/Cas9 genome editing technique, for example (http://www.sciencedirect.com/science/article/pii/S216225311630049X ), and thousands of off-target changes were recently reported in a CRISPR–Cas9-edited mouse (https://www.nature.com/nmeth/journal/v14/n6/full/nmeth.4293.html ), indicating that “single base pair substitutions, as well as deletions of differing sizes” are not as “precisely administered” using genetic engineering as implied by APHIS. And the fact that these off-target changes may comprise “small genetic differences” which may be “not phenotypically observable” does not mean that they are not of concern; most mutations are deleterious, and crops can produce toxic compounds that are not phenotypically observable. As long as there are such high rates of off-target effects associated with genetic engineering, all GE organisms should be evaluated by APHIS for plant pest and noxious weed risk, no exceptions.

The language in the proposed preliminary/provisional Regulatory Status Evaluation Requirements posted on the USDA website (https://www.aphis.usda.gov/aphis/ourfocus/biotechnology/biotech-rule-revision/2016-340-rule/reg-status-eval-data-reqs ), appears to be appropriate for evaluations of the regulatory status of all new crops produced using any GE technology—including genome editing and synthetic biology. But based on the history of GE crops, which includes indirect damage to U.S. crops via superweeds and economic losses due to contamination of non-GMO crops with GE crops, and the fact that off-target effects can occur during the genetic engineering process, it is appropriate for APHIS to require all new GE products to undergo its evaluation process, no exceptions; all developers of GE crops should be required to request a regulatory status evaluation for each new GE product that is developed, on a case-by-case basis prior to each product’s commercialization, and to fulfill the (putative) Regulatory Status Evaluation Requirements described on the USDA website.

I agree with the requirement in step 2. of the proposed preliminary/provisional Regulatory Status Evaluation Requirements, requiring that developers provide “The genotype of the modified organism, including a detailed description of the differences in genotype between the organism subject to the inquiry and the recipient organism.” However, the more specific requirements listed under step 2. fail to explicitly mention a requirement to characterize modified organisms for off-target effects; just as USDA APHIS is proposing that developers submit alignments of modified sequences with the unmodified sequences for intentionally inserted or edited sequences, a truly “detailed description of the differences in genotype between the organism subject to the inquiry and the recipient organism” would require alignments of the entire genomes of the two in order to identify possible off-target effects. The most recent National Academy of Sciences report on GE organisms recommended that techniques used in the various “omics” fields of biology should be employed in characterizing GE organisms; APHIS should require the use of whole-genome sequencing (WGS) to identify all off-target mutations in GE organisms (as was used in the genome-edited mouse study mentioned above: https://www.nature.com/nmeth/journal/v14/n6/full/nmeth.4293.html ).

June 16, 2017: Comments to FDA Re: Docket No. FDA-2016-N-4389, Genome Editing in New Plant Varieties Used for Foods; Request for Comments

My name is Belinda Martineau, I am trained as a plant molecular geneticist and I was a part of the team of scientists at Calgene, Inc. that carried out safety studies of the Flavr Savr^TM tomato in Calgene’s successful effort to gain U.S. regulatory approval for commercializing that genetically engineered (GE) crop product, which included FDA’s approval of the kanamycin-resistance-conferring protein in those tomatoes as a food additive. I thank FDA for providing this opportunity for the public to help inform the agency’s “thinking about foods derived from new plant varieties produced using genome editing techniques.”

Comments on Question 1.

The greater the ability of a developer of a new plant variety to specifically target an intended change to a specific, hopefully “safe” locus in a crop genome; and the fewer off-target, unintended disruptions (such as insertional mutations, additions of minimally characterized vector backbone sequences, and changes in the relative positions of protein-coding genes, microRNA genes and other elements of genome architecture like enhancers and silencers) that are made elsewhere in the crop genome; and the more collective knowledge that is available about the technique and that is known to and utilized by the developer; and the more fail-safe systems the technique has, the safer the resulting new plant variety will be. A comparison of genome editing and other means for developing new plant varieties in terms of these safety-related characteristics is shown in the table (below).

Unfortunately, scientific knowledge of and experience with plant varieties developed with in vitro recombinant DNA technologies and commercialized over the last 23 years have not provided much in the way of helpful information to be applied to safety assessment of new plant varieties produced using genome editing. If FDA had required developers to conduct long-term animal feeding studies of these new plant varieties, for example, we would now have a better idea of how often unintended, off-target effects might have negative implications for food safety. In fact, even when a long-term animal feeding study did suggest that Monsanto’s NK603 GE corn variety might have negative unintended effects (even if the study’s conclusions were deemed “inconclusive” after thorough examination of all the raw data associated with that study by the editor-in-chief of the peer-reviewed journal it was originally published in: https://www.elsevier.com/about/press-releases/research-and-journals/elsevier-announces-article-retraction-from-journal-food-and-chemical-toxicology ), FDA did not require that developer to provide additional data and/or information to support the safety of their product; I think FDA should have done so. And the flaws associated with the older non-targeted genetic engineering techniques remain; I know of a company that did not know until recently that backbone vector sequences can be inadvertently inserted into new plant varieties transformed using the Agrobacterium-based method. If FDA had required developers to check to ensure that they had not inadvertently inserted their genes of interest into an endogenous plant gene or vector backbone sequences into the recipient plant genome, and if FDA had required developers to conduct long-term animal feeding studies, we would now know a lot more about how to deal with off-target effects.

Comments on Question 2 and 3.

The fact that developers can select a specific gene in a crop to “edit” as opposed to the relatively random insertion of genes–often (27-63% of the time) into another gene thereby mutating that gene–that occurs when using the older genetic engineering methods is a major advantage of the new genome editing technologies. (Although these technologies will also be used to insert foreign genes into plant genomes, and selection of a “safe” site for intended insertion will be an issue in those cases.) However, an off-target activity rate of ≥50% has been reported in the scientific literature for the CRISPR/Cas9 genome editing technique, for example (http://www.sciencedirect.com/science/article/pii/S216225311630049X ), and thousands of off-target changes were recently reported in a CRISPR–Cas9-edited mouse (https://www.nature.com/nmeth/journal/v14/n6/full/nmeth.4293.html ). A professor of plant biology at UC Davis stated at a public lecture recently that of the many, many peer-reviewed scientific papers he had read in which the CRISPR method had been used in the reported study, every single one of them also reported off-target effects. These high rates of off-target effects, and the mutations they may result in, make genome-edited plant varieties more likely than traditional-bred plants to present food safety risks. Most mutations are deleterious. Until we know more about such off-target effects and how they might be reduced/eliminated, all products/foods created using these new technologies should be regulated in the U.S., no exceptions. The high rates of off-target events, combined with the newness and consequent lack of experience in using these new technologies to produce foods for human consumption, provide support for FDA taking this cautious and prudent stance with regard to genome-edited foods at this time.

Comments on Question 4.

The step FDA should take to ensure that small firms—as well as large firms and all aspiring developers of genome-edited food or feed organisms—engage with FDA about the safety of such GE foods or feeds is to make engagement with the agency mandatory prior to commercialization of such products. FDA should also establish a mandatory safety evaluation process for these products. The most recent National Academy of Sciences report on GE organisms recommended that techniques used in the various “omics” fields of biology should be employed in characterizing GE organisms; therefore, FDA should, for example, require the use of whole-genome sequencing (WGS) to identify all off-target changes/mutations in all GE organisms (as was used in the genome-edited mouse study mentioned above: https://www.nature.com/nmeth/journal/v14/n6/full/nmeth.4293.html ). Advances in molecular technology should be used not just to create new GE foods or feeds, but to ensure that they are safe for long-term consumption as well.

Summary of differences and similarities of various means for developing new plant varieties:

* Changes in the position of a gene within a genome, e.g. abnormally juxtaposing a gene with heterochromatin (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3721279/ ), can silence or otherwise alter normal gene expression patterns.

** Approximately 20% of the time, this method also leads to insertion of vector “backbone” sequences (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC160498/pdf/061032.pdf ) which, depending on the vector used, may be associated with additional safety risks.