Study shows GM gene-edited rice is not “substantially equivalent” to the parental non-edited rice. Report: Claire Robinson and Prof Michael Antoniou

Following a recent podcast interview () we were asked whether there is any solid scientific research looking at how gene expression or molecular composition in genetically modified (GM) plants differs from conventionally bred plants. As this is an interesting and important question, we thought it would be worth exploring the evidence for how GM plants differ from conventional ones.

First-generation GMOs

There is good evidence of unintended changes between older-style, first-generation (“transgenic” – involving random insertion of foreign gene(s)) genetically modified organisms (GMOs) and their non-GMO counterparts. For example, a review of “omics" molecular profiling studies by Benevenuto et al (2022) concluded: “Several altered metabolic pathways have been found in the comparative omics studies assessing unintended effects in GM crops.”

This review includes the work of Prof Michael Antoniou and colleagues on glyphosate-tolerant NK603 GM maize vs its non-GM isogenic (with the same genetic background, but without genetic modification) parent. This study found that there were major protein and biochemical (metabolite) differences between the GM and non-GM maize and that these differences were due to the GM transformation process, leading the authors to conclude, “NK603 and its isogenic control are not substantially equivalent.”

This conclusion was unpopular with the agricultural biotechnology industry and allied scientists such as Alison Van Eenennaam; certain other scientists; the European Food Safety Authority, which years before had issued an opinion that NK603 was as safe as conventional maize; and the UK’s Science Media Centre (more about them here). They all mobilised to attempt to smear the paper, prompting the authors of the original research to publish a response to their critics in the same journal.

Why did research that only measured various compounds in a GM maize compared with its non-GM close relative excite such fury? The problem was that GM crops are approved by regulators worldwide based on variations on the assumption that they are “substantially equivalent” to the non-GM parent, apart from the intended genetic tweak. But Prof Antoniou’s research gave the lie to that assumption – and exposed the shortcomings in regulators’ approach to measuring equivalence. To establish substantial equivalence on the compositional level, regulators generally only require a gross comparative analysis between the GM crop/food and its non-GM counterpart, e.g. amounts of protein, fat, and carbohydrate present, much like what you might read on a packet of cornflakes.

However, this is not sufficiently detailed to pick up differences in the types of proteins or metabolites present in the GM crop. That’s the case even though these factors could make the difference between it being safe or unsafe to eat and safe or unsafe for the environment. Prof Antoniou’s research did pick up those differences, though the published study didn’t claim that there were any health implications because that question isn’t addressed by this type of study. But far from welcoming this new scientific knowledge, the response from some sectors of the scientific establishment was “How dare you look?”

Benevenuto et al’s review also includes many other research studies demonstrating that when in-depth molecular compositional profiling (comparative “omics”) analysis is used, GM crops and foods are invariably found to be not substantially equivalent to their non-GM counterpart. For example, using protein profiling (proteomics), Zolla et al showed that GM maize MON810 produced an allergenic protein, zein, that was not present in the non-GM parent.

“New GMOs”

Regarding “new GMOs” made with gene editing tools, there is very little research looking at differences between the GM gene-edited plant and its non-GM parent. Based on the fact that many of the processes of gene editing are shared by older-style transgenic genetic engineering, we could expect that there would be some overlap in the unexpected effects. But few are looking. And the UK government has gone so far as to bar government scientists from ever finding out – through direct analysis – that a “new GMO”, so-called “precision bred”, plant is not equivalent to a conventionally bred one.

However, one study has already shown remarkable differences between gene-edited rice plants and their non-GM, non-gene-edited parent. Xiao-Jing L et al (2023) made comparisons between:

• non-gene-edited parent vs base-edited (a form of gene editing) rice, and
• non-gene-edited vs CRISPR/Cas gene-edited rice.

No backcrossing with non-GM plants was done to try to remove GM process-induced mutations (DNA damage). This is a common technique used to “clean up” genomes from unintended effects of gene editing. But according to an analysis by Dr Yves Bertheau, former director of research at the French National Institute for Agricultural Research, “cleanup” backcrossing is not generally done thoroughly. It is limited by developers’ habits, equipment, expertise, and the available material. Also, there are no standardised binding guidelines for examining unintended changes and their subsequent elimination. As a result, unintended genetic damage can remain in the final marketed gene-edited product, with potentially dangerous consequences. Therefore the absence of backcrossing in Xiao-Jing L et al (2023) is not a reason to discount the findings as irrelevant to “real-world” gene editing.

In the study, all plants were grown in the same location, but the authors don’t say if they were grown at the same time. Growing the plants in the same location and at same time minimises compositional variation that could arise from environmental influences, so that any differences seen will likely be the result of the gene editing and base editing processes. At least half of that requirement was met.

The authors performed untargeted transcriptomics (total gene expression profiling) and proteomics (protein expression profiling) on the plants. But they analysed the leaves, not the grain, the part of the plant that is eaten by humans. Therefore no human health conclusions are possible.

As the authors wanted to undertake transcriptomics analysis, they were forced to analyse growing leaves to look for gene expression patterns. Mature dried edible rice grains cannot be used for transcriptomics analysis as they are in a quiescent state, so gene expression is absent. Nevertheless, the authors could have performed proteomics and metabolomics on the grains and this could have yielded insightful information regarding effects on nutritional quality and possible toxicity – but they didn’t.

This study is titled, “No obvious unintended effects was [sic] found in gene editing rice through transcriptional and proteomic analysis”. That’s strange, because they did find major unintended differences in the gene-edited rice compared with the non-edited parent.

Transcriptomics revealed 520 and 566 differentially expressed genes in the Cas9/Nip (CRISPR/Cas gene-edited) and ABE/Nip [base-edited] comparisons, respectively. In other words, the study showed these processes produced large changes in gene expression patterns, albeit in leaves. The authors concluded that the changes were mainly due to “environmental adaptation” – an odd claim, since the plants were grown in the same location to minimise environmental differences. And although the authors don’t say so, the plants were probably grown at same time, so their conclusion that environmental adaptation was the cause of the observed gene expression changes is not valid. Evidently, the gene editing and base editing processes produced the changes.

We would point out that these are process-induced unintentional changes, which will not be evaluated for safety in regions of the world where the GMO industry succeeds in deregulating gene-edited crops. The industry is pushing for product-based regulation, which only looks at the intended end product and ignores the unintended effects of the genetic manipulation processes used to make it.

In the study, biochemical pathway analysis based on the transcriptomics (gene expression pattern) changes predicted alterations in the metabolism of various substances, including terpenoids and polyketones (crucial for ecological adaptation). Also predicted were changes in plant–pathogen interactions and plant signal transduction – the process by which plants perceive environmental cues, such as light, water, and pathogens, and internal hormone signals, converting them into specific responses. All these predicted biochemical function changes could have profound implications for the way the plants would perform in farmers’ fields, as well as posing potential dangers to health and the environment. These possibilities would need to be tested in further experiments.

Proteomics revealed 298 and 54 differentially expressed proteins in the Cas9/Nip (CRISPR/Cas) and ABE/Nip (base editing type gene editing) comparisons, respectively. This shows that the gene expression changes led to protein profile changes. A biochemical pathway analysis linked with protein profile alterations showed changes in biosynthesis of secondary metabolite and metabolic pathways.

The authors concluded that no new proteins were produced, so there are no health consequences – but as they didn’t look at the edible part of the plant, this conclusion is unsound. Also, proteomics does not pick up ALL proteins in the sample analysed, so it is possible they could have missed new proteins being produced by both the on-target and off-target gene editing process-induced mutations around the genome. And crucially, the authors did not conduct metabolomics (small molecule biochemical profiling), which could spot toxins that are not proteins and which could have shown whether the predictions of the transcriptomics were accurate.

To conclude, analysing the leaves showed CRISPR/Cas gene editing and base-editing gene editing caused major changes in gene expression and protein profiles.

This begs the question: If these changes occurred in the leaves, what happened in the edible grain? Would there have been changes in proteins or metabolites, including the appearance of novel toxins or allergens?

This study is proof minimally that the processes of CRISPR gene editing and base-editing gene editing are mutagenic to a degree that results in large changes in gene expression and consequent protein profile. Unfortunately, the authors did not conduct whole genome sequencing to look at unintended DNA damage from these processes and correlate this to the observed changes in gene/protein expression patterns.

Study confirms GMWatch predictions

It has been known for many years that gene editing can cause large-scale DNA damage at both the intended gene editing site as well as “off-target”, elsewhere in the genome. However, what is ignored by both gene editors and regulators is the large-scale genome-wide DNA damage that occurs through the obligatory associated processes of plant tissue culture and plant cell transformation components of the gene editing procedure. This genome-wide unintended DNA damage can number in the hundreds or thousands of mutation sites. Such large-scale damage will inevitably be present, regardless of how “precisely” the gene edit is targeted.

We have predicted in numerous GMWatch articles that this genome-wide DNA damage, resulting from the gene editing process as a whole, could result in large-scale changes in gene expression patterns, leading to altered protein and metabolic function. The study by Xiao-Jing L et al (2023) confirms that our predictions were correct. The concerns that the study raises, although these are not addressed in this publication, are that altered patterns of gene function and biochemistry in the edible parts of the plant could lead to the production of novel toxins and allergens.

The study also has implications for the UK government’s Genetic Technology (Precision Breeding) Regulations. The regulations focus exclusively on the intended gene-edited trait, ignoring the large-scale genome-wide unintended changes of the type shown in the study. The study’s findings undermine the regulations’ assumption that a gene-edited plant that is self-declared by the developer to be “precision bred” is equivalent to a plant arising from “traditional processes” such as conventional breeding.

Multiple problems with gene editing

Finally, Xiao-Jing L et al (2023) reveals unintended large-scale alterations in gene expression patterns, which in all likelihood are underpinned by genome-wide DNA damage resulting from the gene editing procedure as a whole. However, this is just one consideration. Other problems that have been found to arise from gene editing include the unintended insertion of transgenes, such as the antibiotic resistance genes that were unexpectedly found in gene-edited cattle. In another example, this time from the plant kingdom, researchers found fragments of foreign bacterial DNA in the genome of a plant that they had gene edited using the same method employed by the developer of a gene-edited mushroom that was claimed to be free from foreign DNA.

The effects of such foreign DNA on consumers’ health and the environment are not yet known. The same applies to the gene expression and protein profile changes in the gene-edited rice, found in Xiao-Jing L et al (2023).

Conclusion

Shockingly, Xiao-Jing L et al (2023) was the only study we could find that used cutting-edge molecular profiling (omics) to obtain in-depth functional data on the consequences of the gene editing and base editing procedures. The authors compared non-gene-edited parent and gene-edited plants produced via CRISPR/Cas and base editing, and found that these processes are mutagenic, resulting in changes in gene expression and protein profiles.

The types of crude compositional data historically required by regulators for GMO approvals – measuring total protein, carbohydrate, and fat content in the GM plant and the non-GM parent – are not detailed enough to show GM process-induced changes in gene expression, protein profile, and metabolites that could affect health and the environment. For instance, it’s not how much protein that is important in risk assessment, but what specific types of proteins are present. In other words, the devil is in the detail. And that applies to GM gene editing and base editing as much as to older-style transgenic GM techniques.

We hope the findings of this study will lead others to undertake similar molecular profiling “omics” analyses in gene-edited plants to more comprehensively determine the outcomes of gene editing processes and the safety of the products.

The study:

Liu Xiao-Jing et al (2023). No obvious unintended effects was found in gene editing rice through transcriptional and proteomic analysis. GM Crops & Food, 14:1, 1-16, DOI:10.1080/21645698.2023.2229927. https://doi.org/10.1080/21645698.2023.2229927

Image: Shutterstock (licensed purchase)