by Pierre Baduel
figures by Kaitlyn Choi

Summary: The legality of genetically modified organisms (GMOs) remains a controversial issue around the world, especially in Europe. Recent discoveries in the field of epigenetics, or heritable information not encoded in the DNA, have revealed promising alternatives to genetic engineering. As reported in Proceedings of the National Academy of Sciences (PNAS) this year, a team of British researchers managed to completely shut down a gene in a plant without making any modifications to its DNA sequence. Several hurdles remain before this technique can be applied in the field, but crop companies already see tremendous potential for this new development.

RNA interference and its epigenetic implications

When DNA was discovered, it was thought to be the sole repository of the heritable information that defines an organism. In the 1990s, however, with the successful sequencing of the human genome, it became clear that very little (1-2%) of the genome actually consisted of genes. Instead, researchers discovered vast numbers of “selfish elements,” DNA sequences that are capable of copying themselves across the genome to enhance their own inheritance. Selfish element activity can be detrimental to the organism as a whole; these parasitic components occasionally land in the middle of genes, which interferes with the normal workings of the cell. In response to such a threat, hosts have evolved epigenetic mechanisms to silence selfish elements and keep them inactive.

Figure 1. (A) A genetic modification (mutation or insertion of a new gene) will modify the DNA sequence of an organism (B) An epigenetic modification will not modify the DNA sequence but will modify the way it is used by the organism. For example, RNA interference will deposit silencing marks (here in red) on a gene, thus preventing its use.

Like non-selfish genes, selfish elements need proteins to accomplish the copying. Therefore, like any other gene, they are first transcribed into RNA molecules, which serve as messengers that are then “translated” into proteins. However, if these RNA messengers are constantly degraded before being translated into a protein, then the selfish element is effectively silenced. David Baulcombe and colleagues discovered this mechanism in 1999, and opened the door of the now-blooming field of RNA interference (RNAi). They found out that if they synthesized a small RNA molecule that perfectly bound a target messenger RNA and injected it into a plant cell, the cellular machinery would degrade the targeted RNA. [1]. However, if scientists continue to inject these small RNA molecules, this mechanism wastes a lot of energy, as the cell constantly produces mRNA just to degrade it immediately afterwards. Therefore, with continued exposure to this small RNA, the cell compensates by adding long-term silencing marks to the DNA itself, which prevents the production of any messenger RNA. This long-term silencing is called RNA-directed DNA methylation (RdDM). This series of marks constitutes a second “layer” of information on the DNA that was coined epigenetics: it was information, but not coded in DNA like traditional genes. Was this information heritable?

RNAi was soon generalized to the other kingdoms of life, and RdDM was found to occur not only on selfish elements, but also on genes artificially inserted in the genome by humans or pathogens, called transgenes. This was actually how RdDM was discovered in 1994. The transgenes were recognized as foreign by the plant, targeted by RdDM and silenced. But until now it was unknown whether this pathway could also silence a non-transgenic, non-selfish gene, also called a naïve gene. In the aforementioned PNAS study, David Baulcombe and Donna Bond proved it could, [2] and also showed that the silencing was heritable.

A new study establishes a heritable RNAi strategy to silence naïve genes

Using viruses to inject the small RNAs, Baulcombe and Bond targeted a gene that actively inhibits flowering. If the RNA-mediated silencing worked, the inhibitor would be inactivated and the plants would flower early. The plants infected with the viruses did not flower early, and many of their progeny were also late-flowering. But a small percentage of their offspring did flower early. The second generation was even more affected. As the offspring of intermediate-flowering plants flowered early, it was as if the silencing was progressively increasing between generations. When the plant antiviral defense mechanisms were inactivated in a separate experiment, the researchers turned this small percentage of offspring that flowered early into almost 100%! These early-flowering plants were all free of the virus because only the parents were infected. The inactivation of the inhibitor was found in the pollen of the parents, but not at all in their leaves. This confirmed that although silencing began to occur in the parents, it was too late in development to be seen in the leaves, and it was instead transmitted to the offspring by the pollen after at least two generations. This is the first proof of the feasibility of heritable, epigenetic silencing of naïve genes.

Figure 2. (A) In the parents, the flowering inhibitor is active and the plants flower late. (B) After infection with the virus, the small RNAs target the inhibitor gene and silence it in the pollen of the parents. (C) The offspring inherit the silenced gene from the pollen and flower early.

Modifying crops traits but not their genome: issues and advantages

Taking this first proof of concept from the laboratory to application in the field will require a number of questions to be answered. Over how many generations does the silencing last? Can we increase the efficiency without getting rid of the plants’ antiviral mechanisms? And foremost, how can we avoid spraying genetically modified viruses in our fields? If another carrier of the small RNAs could be used, this would be a viable strategy to entirely bypass genetic engineering. Indeed, as Baulcombe and his collaborators did, we can now change a trait of a plant without modifying its DNA. The trait Baulcombe’s team targeted, reduced flowering times, is of agronomic interest and could improve a crop’s adaptation to local environments. Other interesting traits could also be studied; for example, resistance to drought, heat-shock, and flooding. If silencing a gene is found to lead to valuable properties for a crop, then this strategy may produce seeds with improved traits, but no genetic modifications. This means that improved traits are no longer obtained by adding foreign DNA into a plant, but rather simply by inactivating what is already there. In this way, the question of heritability is of importance. Will the seeds bear improved traits for one or two generations and turn back to their natural state without repeated treatment? Or will they conserve their traits indefinitely like classic GMOs? The number of generations the trait is transmitted is thus very important, as it will impact the economic models used by companies and farmers.

RNAi herbicides: the El Dorado of weed control?

Figure 3. (A) Palmer amaranth in the wild (B) Resistance is obtained by overexpression of the enzyme targeted by glyphosate. RNAi against this gene would restore sensitivity to the herbicide.

This strategy cannot be applied to genes that need to be activated to be useful or functions that are not already found in the crop of interest (RNAi can only silence, not add or enhance). As the small RNAs do not introduce any foreign DNA, they can only act on genes already present in the plant genome. For example, introducing new resistance to pesticides or herbicides could never be attained through this method. However, what is attainable is the silencing of existing resistances. And as we know, weed resistance to the most common herbicides does not take long to appear when these products are used at industrial scales globally (see this article). Palmer amaranth, for instance, is a well-known super weed that became resistant to glyphosate, the active ingredient of Monsanto’s Roundup. The resistant weeds simply produce higher levels of the enzyme targeted by the herbicide, allowing them to overcome its effects. Using RNAi to shut down such a gene would therefore restore the vulnerability of the weed to glyphosate. As such, companies like Monsanto leaped on the technology.

Doug Sammons, a senior fellow studying glyphosate resistance at Monsanto, stated in a Canadian agricultural magazine, “There’s been such a shortage of new areas (for weed control) discovered in the last 20 years….This is really a new opportunity.” [3] Monsanto’s new BioDirect technology, which uses RNAi, has already been announced [4] and discussed even though it is still in Research & Development. Even though the targeting of small RNAs can be fine-tuned to a specific gene of a specific weed, off-target risks are a cause of concern for regulatory organizations like the European Food Safety Agency [5] or the Environmental Protection Agency. What if they survive digestion and are transmitted to humans? What if the small RNAs find targets in other genes and other plant species? In the case of glyphosate, the gene targeted belongs to a family of closely related genes that could easily be confused as targets by the small RNAs. This off-targeting would have to be carefully controlled as it could lead to deleterious effects on the crop itself. Furthermore, if the effects are transmitted to progeny and could still be seen in later generations, the question of heritability will here again be of utmost importance.

As shown by Baulcombe and Bond, RNAi strategies have tremendous potential for crop companies. However, taking this new, and potentially game changing, technology from the laboratory to RNAi herbicides or epigenetically modified crops in the field, will require more proof to ease important societal concerns.

Pierre Baduel is a 3rd year graduate student in Organismic and Evolutionary Biology studying plant genetics. As a member of the corps des Mines, one of the French corps d’État of inter-ministerial engineers, he is also very interested in the interaction between science, public policy and innovation.

This article is part of the August 2015 Special Edition, Genetically Modified Organisms and Our Food.


1. J. Hamilton and D. Baulcombe, “A Species of Small Antisense RNA in Posttranscriptional Gene Silencing in Plants,” Science, pp. 950-952, 1999.
2. D. Bond and D. Baulcombe, “Epigenetic transitions leading to heritable, RNA-mediated de novo silencing in Arabidopsis thaliana,” PNAS, vol. 112, no. 3, pp. 917-922, 2015.
3. R. Arnason, “RNAi may hold key to glyphosate resistance,” The Western Producer, 10 February 2014.
4. BioDirect. Monsanto.
5. E. F. S. Authority, “Risk assessment considerations for RNAi-based GM plants,” in Brussels, Belgium, International Scientific Workshop, 2014.

2 thoughts on “Epigenetics in Plant Breeding: Hard Science, Soft Tool

Leave a Reply

Your email address will not be published. Required fields are marked *