![[Go to Home page]](http://www.science.org.au/images/smallseg.gif){kind=small} |
Australian Academy of Science |
| About the Academy |
Home > Events > Australian Frontiers of Science > 2003
Gene silencing
in plants and other eukaryotes
I would like to tell you about gene silencing in plants and other eukaryotes. You can see from my title that, in my world, plants come first! I want to give you two take-home messages from this talk. Firstly, that the gene silencing technique that I am going to tell you about, often called 'RNA interference' in animals, is very powerful. And secondly, that the pathway by which it operates was not even contemplated 15 years ago, but actually plays a key role in multicellular life.
Ten or so years ago, we were working on virus resistance in plants, and unwittingly encountering this pathway. We were following the lead of Roger Beachy, who had said that if you take a virus-derived gene (the coat protein gene) and express its protein from a transgene in a plant, you can get virus protection. So we, like lots of other people, took a virus gene (in our case from potato virus Y) and put it into plants (in our case potato). What we got were some plants that were completely immune to potato virus Y (figure 1). However, we found was that these virus-resistant plants contained almost no protein or messenger RNA from the transgene. This was very confusing, but very soon, two unsung American researchers (Bill Dougherty and John Lindbo) made a pioneering discovery. They showed that it was not the protein, but the RNA from the transgene, that was conferring the resistance. Somehow, the RNA triggers a mechanism in the plant that destroys not only the transgene RNA but also the RNA genome of the invading virus (which shares sequence homology with the transgene). Much of this sequence-specific degradation model that they proposed has been validated in the intervening years. Ironically, these researchers left Science soon after their pioneering work to become a salesman and a monk! Our contribution to understanding the mechanism was to demonstre that the feature of the transgene RNA, that triggered the RNA degradation mechanism, was that it was double stranded. Lindbo and Docherty had suggested that it was excessive levels of the transgene RNA.
Let me show you how it works: Imagine that the image in figure 2 is a plant cell infected by a virus. When a virus enters, it injects its single-stranded genomic RNA into the cell which then replicates. So this is the plus strand, which is copied to form a new minus strand, which is then copied to form a new plus strand, and so on. Double-stranded RNA is unusual in cells.
Plants have the capacity to recognise double-stranded RNA (figure 3) and to trigger a mechanism that destroys that double-stranded RNA. The plant uses the sequences of the double-stranded RNA as a guide to recognise the single-stranded RNA of the same sequence and destroy it, thus protecting the plant against the virus. I guess what you are probably saying to yourselves now is, 'Well, if this is always the case, then all plants should be resistant to all viruses.' Obviously, this is not the case or there would be no plant viruses – I am describing a mechanism which in some cases successfully defends a plant against a virus but in other cases it can be overcome by a virus.
We know the nitty-gritty detail of the mechanism from some beautiful work in Drosophila and nematodes. Again, if you imagine that that is the cell, then there is this enzyme here which is called Dicer and it completely ignores any single-stranded RNA that is there. But if you introduce double-stranded RNA into the cell, it says, 'Yes, I recognise you' and chops (dices) it up (figure 4).
We will just look at one of these Dicers now. It chops up the dsRNA, and the important point is that it chops it up into 21-base pieces. The Dicer then transfers one strand of 21nt dsRNA fragment to a complex (called RISC) which is a nuclease. Now the RISC has a guide-RNA (or a siRNA) that it uses to scrutinize single-stranded RNAs. When the complex finds an RNA with which the siRNA has perfect complementarity, the RISC specifically cleaves it [and zapping sound effects] (figure 5).
Of course, when you have put in long double-stranded RNA the cell generates lots of siRNA-loaded RISCs that can collectively recognise contiguous sequences and chop them up [with more zapping noises] (figure 6). What I would ask you to remember is that these guides are 21 nucleotides and these bits that are cut off are also 21 nucleotides. So how had we got these original plants that were resistant in figure 1? What we had done in these very initial plants is, by chance, we had introduced the transgene in such a way that it was producing double-stranded RNA corresponding to the virus sequence. Consequently the RISC complex was already charged up with guide RNAs corresponding to the viral sequence. So when the virus injects its RNA into the cell, the plant is prepared, and simply munches it up.
We thought, 'This is great. Now we know how we can kill off any RNA that we like in a cell. All that we have to do is to trick the cell into believing that the messenger RNA, of the gene that we want to destroy, is a virus.' And rather than put in two strands of RNA, we came up with the idea that we could make a transgene which would produce the self-complementary RNA – which looks like dsRNA. To do this, we placed a promoter in front of this coding region, which has some virus sequence at the front end in a sense orientation, and at the back end this sequence is flipped around and in the antisense orientation. So, when the RNA is transcribed from the transgene, it produces an RNA – which we call a hairpin RNA. It folds back on itself and hybridises to produce duplexed RNA (figure 7). Then, if we introduce this transgene into a plant, it will produce what looks like a replicating virus. Which single stranded RNA is specifically degraded depends on the particular inversely-oriented sequence.
Figure 8 shows an example of one of our first attempts to silence gene activity using this hairpin RNA approach. Here is a plant expressing the GUS gene – a 'blue' gene. If we then transform this plant with a transgene that produces a hairpin RNA corresponding to the GUS sequence, it silences the blue gene so the plant is now white. If you look at this gel here [at bottom], it shows that RNA extracts from the white plants (W) contain 21-base RNAs that have been derived from chopping up the hairpin RNA and the target GUS mRNA, whereas these small RNAs are not present in extracts from the blue plants (B).
Once we had worked out this technology, we have been applying it to a range of different genes (figure 9). I won't go into fine detail about the genes we have silenced, except to say that they control a wide range of plant characteristics and that we have applied this technology in a number of different plant species. We have found that we can silence genes with really high efficiency, so that virtually every transgenic plant that we make shows good silencing of the target gene.
Figure 10 shows some examples of the phenotypes we have generated. Here, we have knocked out a flowering time gene. This is a wild-type plant, and these silenced plants are flowering early. This phenotype relates to silencing a plant gene involved in the perception of red light: these plants don't 'see' the red light, so they show elongated growth, as if they are in the dark. Here we have knocked out a seed coat pigment gene. Here, we have knocked out the gene for UV protection, so the plant photobleaches. Here, we have knocked out a gene which responds to ethylene. Here we have knocked out a gene that makes the anthers, the male part of the plant, so this plant only has female reproductive organs. And here, we see again the silencing of the GUS gene.
I am sure that you are all aware of the fibre crop cotton (figure 11). Cotton oil is also an important commercial product. The trouble with conventional cotton oil is that it is an 'unhealthy' oil. It contains a very high level of linoleic acid and a low level of oleic acid in its oil profile.
So we have made a hairpin gene construct that targets the delta 12-desaturase gene, which is responsible for the conversion of oleic acid to linoleic acid. You can see that this has reduced the linoleic acid in the profile to 4 per cent and boosted the oleic acid up to 78 per cent (figure 12). So here is an example of an important commercial application of this technology.
Another application that we are very interested in is in the area of genomics. The genome of Arabidopsis, like that of humans, mice, Drosophila, yeasts and other animals, has been sequenced. Within Arabidopsis there are 25,000 genes, but we really don't know what many of them are doing (figure 13). So we are using our hairpin RNA strategy to knock out all the genes in the Arabidopsis genome, one by one.
This is being done in a major European network to with which we are collaborating (figure 14). The idea is to knock out every gene, individually, and look at the resulting phenotype to see if that phenotype will give us a clue as to what that particular gene was doing. This has been done in nematodes quite extensively.
Here are a couple of papers showing how using double stranded RNA have been used to target the genes of chromosomes 1 and 3 (figure 15). This has led to the identification of the function of a whole host of different genes. Hopefully, I have shown you that we can use this strategy to knock out basically whatever gene we like. We do this in plants, with these large hairpin constructs. In medical applications, instead of using large hairpin constructs, one uses these pre-digested 21 mers, introduced as small bits of RNA, so that interferon-type responses are not triggered. These medical applications appear to be working very well. I would like, in the remainder of my presentation, not to talk about further applications of the technology, but to talk about the pathway itself. I think this is a really exciting, new, breaking field that shows that the pathway is of key importance to multicellular eukaryotic life.
Biologists will know of the central dogma (figure 16) that DNA goes to RNA goes to protein, and as Levon was saying, what happens is this: if this is your gene, you have a promoter region that when you have transcription factors binding to it can activate the transcription of this coding region to make RNA which is then translated to produce the protein. Alternatively, you can have other transcription factors that will bind to this area and stop the expression. It is generally thought that transcription factors are regulating when and where and how much of a particular protein is made. And also, according to this dogma, RNA is regarded simply as a messenger that acts between DNA and RNA. But what I would like to show you is that these small 21nt RNAs, that I have been talking about, play a central role in regulating gene expression. Particularly in regulating gene expression in such a way that the organism can undergo its essential developmental changes. Go back to the experiment involving the blue and white plants shown in figure 8. I showed you before that in these plants, which contain the GUS and the anti-GUS hairpin constructs, one can detect these small 21-mer RNAs. The probe has actually detected them, showing that they are related to the GUS sequence. When people were isolating these small RNAs and sequencing them, and finding that they were derived from the GUS mRNA or hairpin RNA,' they also found other small RNAs that were not derived from the hairpin or target RNAs. In fact, these small RNAs are also present in non-hairpin plants. There are a whole variety of 21-mer RNAs that are being produced during the normal growth cycle of a plant. What has really become apparent, over the last year and a half is that, in the same way that we were introducing hairpin RNA transgenes to target a particular plant gene for our own human desires, the plant is doing a similar thing undirected by us. And, in fact, this happens in all the eukaryotes.
From their DNA, often from the regions that we have previously called junk or non-coding sequences, plants and animals are actually producing their own type of RNA hairpins (figure 17 shows what happens). In plants, these RNAs with a hairpin structure are transcribed, they get chopped up by the Dicer protein, and the small RNAs are incorporated into the RISC complex. The RISC complexes then cleave specific endogenous gene mRNAs, preventing their translation.
In animals (figure 18), hairpin RNAs are processed in much the same way to make small RNAs but these then, somehow, inhibit the translation, rather than cleave, the target mRNAs. By either method the final outcome is the specific silencing of endogenous genes. The small RNAs directing this process are now termed: microRNAs and are largely targeted against transcription factor mRNAs.
These microRNAs (figure 19) are made at a specific times during development, and they either activate the destruction of or prevent the translation of their target mRNAs, thus stopping particular transcription factors from being made at a certain time or in a certain tissue. This represents a whole new paradigm shift: transcription factors are dictating developmental switches, but they are in turn regulated by a higher control – microRNAs. To give you further insight into what this is all about: microRNAs were first identified in worms[nematodes] (there are about 120 of them) and they are made from pre-microRNAs (hairpin RNAs). The microRNAs were first identified for their role in regulating the moulting process. In nematodes containing a mutation, which prevents the production of some specific microRNAs, the target genes are not inhibited and the nematode larvae get stuck in their juvenile form. In Drosophila, it has very recently been shown that microRNAs regulate fat metabolism, programmed cell death, stress responses and developmental changes. In plants the same sorts of observations have been made. Flowering time, leaf shape, floral identity, hormone signalling, all these developmental changes, are being regulated by these small RNAs. Even in yeast, it has been shown that the small RNA silencing pathway is required for centromere structure. If the pathway is not functioning (in a strain with a Dicer mutation), the centromere breaks down and the chromosomes are unable to attach to the spindle properly and partition correctly into the daughter cells of a dividing cell.
We seem to have stumbled over this pathway which is essential to nucleate and multicellular life. If we look at where we can find the pathway, it is in animals, in plants and in fungi, but seemingly not in bacteria (figure 20). We have not really located the point in evolution that the pathway took shape, but it is tempting to suggest that the microRNA pathway evolved to facilitate multicellular life. These small RNAs have the capacity to move from cell to cell, so it is tempting to suggest that they communicate between cells to direct cell differentiation – which is the prerequisite to the formation of complex multicellular organisms. I would like to finish by acknowledging my excellent colleagues for their work.
Session 3 discussion |
