Last week a legal decision in New Zealand overturned a previous opinion that using gene editing to create variants would not create genetically-modified organisms (GMOs). At the centre of this is how the law defines a GMO, but let’s leave that for part two and take a stroll through the past.
Since antiquity humans have sought to improve agricultural species.
Charles Darwin’s famous theory was for evolution by descent with modification. Every species is a genetically-modified version of its ancestral species. The variations are introduced by natural processes and the selection by natural selection or other population genetics effects. In this sense every species is a genetically-modified organism – us, too.
We’re also modified by mutations creating genetic variants not seen in either of our parents’ genomes.
Our genomes are a mixture of our parents’ genomes, but in addition to the mixing of our parents’ genes we also have about 50-100 genetic changes that are not present in either of our parents’ genomes. These de novo mutations are also the stuff of evolution. They introduce brand-new changes. It’s part of what makes individuals, individual. (Only perhaps one or two of these 50-100 changes are likely to affect us in an easily identified way.)
Look at enough fields, forests, herds, buzzing insects—whatever—you’ll find a few individuals that look different. Their genetic modifications have made them visibly different. Other changes you might not see from the outside, but they’re there, altering the chemistry of the organism (animal, plant, fungus, bacteria, virus).
One way you can develop new crops or animals is to grow large numbers of them and select those few that are different, whose differences look potentially useful. They might have plumper fruit or bigger roots, for example. You might need to follow this with in-breeding to establish the trait robustly.
That’s a (very) slow process, particularly for plants or animals that take several years to mature, and a pot-luck one that involves growing a very large number of the animals or plants to hope to uncover potentially useful variants.
You could try speed things up by increasing the rate variants occur. One approach would be irradiate ‘seed’ tissue or use chemical mutagens to create random changes in their DNA. They have been widely used – the joint FAO/IAEA Programme Mutant Variety Database lists 3218 mutant plant varieties. These aren’t all new either, the entries go back for over 80 years.
(It’s worth thinking about that number when you think of food safety – about 75% of those are crop species, they’ve been out a long time but we don’t hear of many, if any, major issues about food safety.)
Potentially you can get more than your bargained for. This sort of mutagenesis is random, hitting anywhere in the genome. It does not have to be limited to hitting just one place in the genome. If a breeder selects for a trait that is based on mutagenesis at one point in the genome, they might also be getting changes at other places in the genome too. You don’t really know what you’ve got.
Despite that we still have no idea of the exact genetic changes introduced, the screening of the traits of the plants by breeders seems to be enough to maintain safety levels. That unintended experiment has been running for more than 60 years and while philosophically it’s a lousy way of testing it, it seems to be fine. Generate variants in a rough-and-ready manner; screening by the traits seems to be enough.
While faster than traditional selective breeding alone, this is still a process of randomly generating changes and selecting from them.
Another approach would be to alter particular genes selectively. This requires knowing enough about the genes to have a particular target in mind. As an example I’ve written about an attempt to increase calcium levels in carrots by making a shorter version of the CAX1 gene that would make it more persistent at storing calcium. You have to know what CAX1 does first. You’re not treating a plant (or it’s genome) as a black box that you randomly change hoping for an useful variant to pop up. You have to know how CAX1 works in storing calcium and that by shortening it, it might stuff more calcium into the cells. (Note this example hasn’t introduced a new gene, but altered an existing one.)
Earlier, we saw how variants can occur and how we can be different from our parents through new mutations. Mutations can occur several ways, but one common theme is that the repair of them can sometimes not work leaving a ‘broken’ gene.
Animals (and plants) rely on DNA damage repair to keep their DNA in proper shape. We are constantly bombarded with radiation from the sun and elsewhere. Small doses of chemicals make their way into a few of our many, many cells and muck up those few cells. Some of this damages the DNA in these cells. Our cells have DNA repair enzymes that detect chemical oddities in our DNA, cut out the offending bit and patch in a fix. Occasionally the fixing process itself doesn’t quite work and that particular place in the out genome stays damaged.
These repair processes are pretty important. Without them we’d have all sorts of trouble. One example is a city in Brazil with a high incidence of xeroderma pigmentosum (XP). People with XP have a damaged DNA repair gene – their cells are unable to properly repair DNA damage caused by ultraviolet light. In severe cases they have to avoid all sunlight. Their skins literally peel off and they suffer from skin cancers. The majority of people with XP don’t live past the age of 20. XP shows how important DNA repair is to us and reminds us that we’re all under ‘attack’ but evolution has seen a working solution adopted for most of us.
Repair processes aren’t infallible; sometimes they don’t work. Mutations that can’t be repaired stick around. If the changes occur in the germ cells that form the next generation, that change is passed on.
There are different DNA repair systems. One repair process is to rejoin DNA that has been cut or broken, so one way to trigger DNA repair is to cut the DNA. This cutting of the DNA mimics what happens in natural mutations.
There are natural proteins that cut DNA, they’re called nucleases. They have been used by molecular biologists for a long time to cut and join pieces of DNA.
Each position in DNA has one of four bases linked to a chain that tethers them together as strand of bases. These enzymes cut the phosphodiester chain, often loosely called the ‘backbone’. If you imagine DNA as a spiral staircase, the bases would be the steps and the backbone would be the spiral girders supporting the staircase.
Natural nucleases recognise and bind to short sequences of DNA – a few DNA bases with a particular sequence. For example, the EcoRI nuclease recognises the DNA sequence GAATTC, cutting ‘backbone’ between the ‘G’ and the adjacent ‘A’.
Very short DNA sequences like this are found many, many times in a genome. Our genome, for example is about 3 billion base pairs long. There are only four DNA bases, so there are many thousands of EcoRI sites in our genome.
Nucleases have two parts. One part recognises the DNA sequence it binds to (the DNA-binding region). The other part is an enzyme that does the chemistry of cutting the phosphodiester backbone.
Some proteins can be made to bind very long DNA sequences that occur only once within a genome.
By hooking together one of these proteins and a DNA-cutting domain you can make an enzyme that will cut only at once place in a genome.
That’s the idea behind ZFN-1 and TALENs.
In the case of ZFNs—zinc finger nucleases—a chain of zinc finger DNA-binding modules is made that recognises a long DNA sequence.
For TALENs—that’s transcription activator-like effector nucleases—modified versions of the DNA-binding regions of TALEs are joined to a nuclease.
You could use these enzymes to cut and add new genetic material, or you could just cut the DNA and let the plant’s DNA repair system kick in – every once in a while it’ll fail and the position will stay broken.
This can be used to knock out particular genes.
ZFN-1 and TALENs are used in pairs, so that the DNA is cut in both strands. Double strand breaks trigger the plant cell’s DNA repair system. The DNA repair system is occasionally faulty; when that happens the gene that was cut is knocked out.
The court case that sparked this article examined an opinion that use of ZFN-1 or TALEN gene editing techniques did not create GMOs in response to an appeal to the opinion. Scion’s question for an opinion from EPA is a general one, rather than based detailed plans for a particular project, but one possible project they named was developing pines that are less prone to wilding seedlings. Let’s use this as an example.
If you know a gene in pine is important for, say, fertility or for cones to develop, you could knock out that gene leaving the rest of the plant’s genes exactly as they were. It’d be the same as if natural radiation knocked the gene out – you’d end up with a variant individual that is unable to produce offspring.
No DNA from another species is being added, nor DNA from the same species added for that matter. What you are trying to do is create a variant with one gene disabled.
I’ve tried to keep the above neutral, to sketch out some background without taking a stance. In part two, I’ll offer my opinions on the ruling and some thoughts about the Act. (I may dive into the court ruling and the Act itself either in part two or as a third part, but I’m sure only die-hards would read that…!)
Update: As a follow-up, readers might like the guest article by Peter Langridge, the CEO of the Australian Centre for Plant Functional Genomics, GM techniques: from the field to the laboratory (and back again).
A broader GM term might be genetic manipulation; this encompasses selective breeding, inbreeding, etc. There’s a nice article on genetic manipulation just out on i09, Genetic Manipulation: The First 50,000 Years. It’s well worth reading – give it a go.
I could write a separate article explaining how the ZFN-1 and TALEN methods work in more detail and perhaps what has been done with them if there is interest. (One use of genetic editing is to try correct genetic diseases, for example.) There is also newer technique, the CRISPR/Cas system (pronounced ‘crisper’), that some may be interested in. Feel free to let me know if you’d be interested.
1. Some expert commentary on the court ruling is available at the New Zealand Science Media Centre website. Being someone involved in science communication, of the responses, I like Assoc. Prof. Peter Deardon’s take; you can read it over there.
2. The ‘other genetic process’ is a grab-bag to include things like genetic drift, founder effects and so on. If you don’t know what these are, that’s OK and don’t worry too much. It’s just a reminder that there’s more that just classical selection.
3. You might think I’m playing word games, but it reveals that if we take the term genetically-modified organism literally—life with altered genetics compared to its ‘parental’ form—we’d have to include all life. Obviously this is not what is it that the law thinks ‘genetically modified organism’ means.
4. ‘Mutation’ is an annoying word for people in science communication. We probably mostly have science fiction to blame for that people see it meaning something negative. Variants is better, I think, if a little bland. (You’ll see in my article on Bengkala’s deaf, I referred to the deafness as a genetic variant for that reason.)
5. I’m using generalised values, different estimates arise from different studies. It’s worth remembering that the majority of these mutations will have no effect on the organism. There is plenty of debate about what the precise mutation rate is and what the implications are. I don’t want to get distracted by that here, hence the generalised values. You can read one on-line discussion at the Sandwalk blog; read the comments to get a small taste of the flavour of debate on this. The age of the father plays a role in the rate of de novo mutations. For those who would like a technical take on the subject, one fairly recent review is Ku et al., Human Genetics, 2012, 6:27, A new era in the discovery of de novo mutations underlying human genetic disease (PDF file).
6. I’m being informal here, it’s not just seeds but you get the idea.
7. Breeders could also activate ‘jumping genes’ (transposons). When these land within another gene they knock out the gene they landed within. While this is used in research experiments, I’ve no idea if commercial species have been developed this way.
8. Even there, there are degrees of specificity. The latest gene editing techniques are very specific (but not necessarily perfect). Some other techniques are (intentionally) random within a small portion of the genome.
9. The first restriction nuclease, HindII, was discovered in 1968 (published in 1970 – I guess science was slower then! – Smith et al J. Mol. Biol. (51): 379–391 (1970). “A Restriction Enzyme from Hemophilus influenzae”.) For the keen: Both of HindII and EcoRI proteins cut both sides of the DNA. HindII cuts at the same place leaving blunt ends; EcoRI cuts leaving a dangling stretch of a few bases with no matching second strand, called a sticky end because the single-stranded portion can bind to a matching single-stranded DNA of the complementary sequence.
10. TALEs were originally found in bacteria (Xanthomonas) that release them to help them infect plants by the TALEs entering the plant and activating (turning on) genes in the plant that make it easier for the bacteria to infect the plant. (Plants have evolved resistance genes competing against the action of TALE proteins.)
Other articles in Code for life: