Dr Doug Gurian-Sherman comments on a recent paper claiming crops can be genetically engineered for drought tolerance using CRISPR

A study [1] has been published by employees of DuPont Pioneer on a drought-tolerant CRISPR maize called ARGOS8. Titled, "ARGOS8 variants generated by CRISPR-Cas9 improve maize grain yield under field drought stress conditions", it states that a field study found that "the ARGOS8 variants increased grain yield by five bushels per acre under flowering stress conditions and had no yield loss under well‐watered conditions".

According the authors, the "results demonstrate the utility of the CRISPR-Cas9 system in generating novel allelic variation for breeding drought-tolerant crops".

So can we expect GM CRISPR'd drought-tolerant maize any time soon? Dr Doug Gurian-Sherman, a research consultant with Minneapolis-based Strategic Expansion and Trainings, LLC, which supports ecologically based sustainable agriculture, advises us not to hold our breath. His comment is below.

Incidentally, it's worth bearing in mind that there are many non-GM drought-tolerant crops, including maize, that are already available to farmers.

Also, farming systems are at least as important as genetics in combating drought. Long-running farming trials by the Rodale Institute in the US found that organic systems yield up to 40% more than conventional chemical methods in times of drought. And in non-drought conditions, organic yields are competitive with conventional after a five-year transition period. These trials included maize and soybeans, the main commodity crops that are genetically engineered in North and South America.
---

Dr Doug Gurian-Sherman comments on the new study:

There are several interesting aspects of this research. First, it should be understood as a research paper, not as a proposal for commercialization. Even though the authors are with DuPont, and their research is likely intended to explore a potential product, there are many such projects that never make it, and that is as expected by the companies themselves. Nonetheless, it is a useful example of what the companies are examining as part of their potential portfolios, and is also useful as a glimpse into the recent status of their work.

The yield improvement from the engineered trait, ARGOS8, was relatively small, only about 4%, and only for drought in the flowering stage, not grain filling. Flowering and grain production are parts of corn crop development that are particularly vulnerable to drought. Traditional breeding has already had considerable success in tackling flowering vulnerability through reducing the interval between the separate male and female parts of corn flowers (called tassels and silk, respectively).

This also raises a question of whether the new engineered trait could complement conventional drought tolerance traits if added to those corn varieties. If it does, without negative tradeoffs (a big question), it might be useful. But if not, its value would be considerably less.

Control corn not disclosed

This raises a related question of the kind of control corn variety used in the experimental field trials to determine the levels of drought tolerance. Was the non-ARGOS8 corn a particularly drought susceptible variety? Or was it typical in terms of drought tolerance, or was it one of the recent non-GM drought-tolerant varieties?

For example, if an average variety, or especially a drought-susceptible variety, was used for comparison with the ARGOS8 plants, that would in effect exaggerate the drought tolerance of the ARGOS8 variety. This is because farmers in drought-prone areas would be expected to grow drought tolerant varieties, not those more susceptible to drought.  

The corn variety used as a control to determine yield characteristics of the ARGOS8 plants was not disclosed in the paper. This represents either poor science writing or possibly a desire to obscure these important data. Until these varieties and their drought characteristics are revealed, the value of the ARGOS8 trait in the real world should be considered essentially undisclosed.  

The levels of experimental drought were not disclosed, except that it varied. And the data are aggregated, so it is hard to know what it means in any detail. Drought conditions were induced experimentally, and this can be very different from natural droughts, which vary considerably in their characteristics. For example, if a natural drought occurred mainly in the earlier stages of corn growth, the drought-tolerance trait may be expected to have little value.

Control corn performed better in one type of drought period

The yield in the control corn variety that was not engineered for drought tolerance was actually higher (by about 3%) when drought stress occurred during grain filling than the yield of the ARGOS8 GM plants, but was not statistically significant. This raises the question of whether these differences were mere experimental artefact, or would be borne out if more experimental (or commercial) replications were performed.

The field trials were also highly controlled in terms of cultivation practices, and seem to have been conducted only during one year in very small plots. So even though they were conducted at several sites in different parts of the country, I would not draw too much significance from their claim of wild-type levels of yield for the GM plants under normal moisture conditions (often a problem with drought tolerance traits). Nor would I draw conclusions regarding their value in real droughts.

A practical problem with GM traits has been the cost to farmers. Companies wish to recuperate high development costs and make a profit (excessive for the big companies because of their concentrated market power). Drought is unpredictable in most areas where corn is grown. Will farmers want to pay for a GM trait that provides only a 4% yield benefit under drought conditions, when those conditions may not even occur, or may only occur occasionally?

For example, it was predicted that the Monsanto GM drought tolerance trait from about five years ago would be welcomed by only about the 15-20% or corn farmers in the US in regions that commonly experience drought stress. And in fact, the available data several years after the commercialization of that trait seemed to show that a very much lower percentage of corn farmers were buying it.

Broader considerations

The ARGOS8 gene controls response to ethylene, an important plant hormone not only for stress tolerance, but also disease resistance and fruit ripening in some species, among other things. Although this gene is only one of several response elements of the ethylene metabolic pathways, it is entirely premature to conclude from the limited field trials disclosed in this paper that there is no negative pleiotropy [the genetic effect of a single gene on multiple phenotypic traits], e.g. unintended effects of the gene that negatively affect yield, under some environments. No responsible plant scientist would accept such limited field data as definitive or even sufficient for commercialization.

And as noted briefly in this paper and many others, natural drought tolerance is a multi-gene trait. The quantitative trait loci (QTL) – the sections of DNA, typically different genes scattered around the genome, that correlate with variation in that trait – that are often associated with drought tolerance are notoriously dependent on particular environmental conditions for their function. They are often also dependent on the particular genetic background of the plant (i.e. the crop variety that the gene is placed into). Thus non-engineered plants with QTL for drought will respond differently to the characteristics of particular droughts and environments. And different combinations of QTL may be more effective under different drought conditions.  

QTL are often seen as not of value, or as a complication, by many industrial breeders and especially big companies, because of their complexity and variability. But they can be a reflection of adaptation to particular environments, which may be  welcomed and viewed as an opportunity under agroecological farming systems that are optimized for local or regional environments.

In other words, industrial farming, as expressed by big GM seed/pesticide companies, is looking for uniformity, which is not specifically adaptive or optimized to local environments. Rather than the genetic diversity of QTLs, they are looking for traits that work well in many environments. And for a complex phenomenon like drought, powerful traits like drought tolerance, without [the problem of] pleiotropy, may be hard to find. Companies do this to lower development and transaction costs. But that is not the best way to develop resilience to multiple and variable local conditions, which not only include drought, but, over time, other stresses as well.  

The push for uniformity by big companies includes overemphasis on a few crops like maize, instead of developing the many crops that have naturally higher levels of drought and heat tolerance. Examples of grains that have these properties include sorghums and millets, and many other types of crops.  

Breeding, e.g. phenotypic selection, for ecological farming systems can optimize resilience to those local conditions, taking advantage of multiple traits and crops. In addition, drought tolerance can be enhanced without the high cost of engineered seeds by agroecological cropping and management systems, e.g. crop and varietal diversity, and developing fertile soil that holds more moisture.

The poor performance of the GM approach to drought tolerance so far can be seen in Monsanto's one commercialized trait, which as of a few years ago was barely grown in the US. But despite that, Monsanto/Bayer is trying to push this trait onto Africa. That trait had about the same performance as the ARGOS8 trait in this paper.

Technical observations

The ARGOS8 paper has some interesting technical aspects.

First, it uses a property of CRISPR-cas9 that is well known among scientists but has been neglected by the media. That is the ability of this gene editing system to insert genes at particular locations in genomes rather the random insertion process of older GM. Some of us have cautioned that this would be an important use of this technology, in addition to the overemphasized single nucleotide mutations (single point mutations) that are often claimed to mimic natural breeding. The use [of CRISPR-cas9] in the ARGOS8 experiments to substitute or add a desired promoter (the part of a gene that drives and controls expression, e.g. when, and how strongly, the gene is "turned on") is a good example of how industry values this characteristic of CRISPR-cas9. It can also be used in this way for transgenic manipulations.

Second, it is curious that a so-called constitutive promoter (which is turned on all the time, all over the plant) was used to drive ARGOS8. The industry and academic scientists have been talking about using drought-specific promoters for years. These have a possibly substantial advantage of not being turned on when drought is not occurring, thus possibly reducing the time when negative unintended effects might occur.

Is this because this paper represents only an intermediate stage of development, or that the known drought promoters do not work as well as was hoped?

In summary, while this paper is interesting, it is very far from showing that the ARGOS8 trait is or could be a commercially useful drought tolerance gene. And it also does not answer the many other downsides or challenges of the GM approach to drought tolerance, or other important crop traits. Although it is an example of only a single drought tolerance gene, it illustrates several more general challenges to using GM to address complex problems in agriculture.  

References

1. Shi J et al (2016). ARGOS8 variants generated by CRISPR-Cas9 improve maize grain yield under field drought stress conditions. Plant Biotechnology Journal 15(2):207–216. https://onlinelibrary.wiley.com/doi/full/10.1111/pbi.12603