SCIENCE AT THE SHINE DOME canberra 5 - 7 may 2004

CELEBRATING 50 YEARS

Symposium: A celebration of Australian science

Friday, 7 May 2004

Dr Chris Helliwell
Senior Research Scientist, CSIRO Plant Industry

Chris HelliwellChris Helliwell is a Senior Research Scientist at CSIRO Plant Industry, Canberra. He received his PhD in plant science from Cambridge University, UK, where he worked on the regulation of expression of the plastocyanin gene, which encodes a component of the photosynthetic electron transfer chain. In 1995 he was awarded a Postdoctoral Fellowship from the Royal Society to work at CSIRO Plant Industry, where he has remained since. Much of his work in Australia has focused on plant hormone biology. He has made a major contribution to understanding the biosynthesis of gibberellin, an important hormone in plant development and reproduction. His research interests also include the role of peptide signalling in plant development, the application of RNAi in plant functional genomics and the role of epigenetic mechanisms in the control of plant development. Chris has published widely in high impact journals and was awarded the 2003 Science Minister's Prize for Life Scientist of the Year.

The biosynthesis of the plant hormone gibberellin

Figure 1
Plants use hormones to control growth and development
(Click on image for a larger version)

You may not know this, but plants also have hormones. This is just a few of them; there is actually a much bigger number than this. You can see that there is a wide variety of chemical structures here. Some of them are very similar to mammalian hormones – this group of compounds called the brassinosteroids is very similar to mammalian steroid hormones – but there is also a large group of molecules present in plants that act as hormones that are unique to plant systems. They include things like ethylene, which is actually a gas, cytokinins and abscisic acid.

These hormones control all sorts of different aspects of plant development. One of the features of the plant is that it can't get up and run away when it is in a bad environment, so it has to be able to adapt to the environment, and these hormones are often involved in these sorts of adaptations. Abscisic acid is particularly important in this. It tends to appear in response to plant stresses, and alters gene expression to help a plant to cope with what is going on.

What I am going to talk about today is this group of hormones called the gibberellins. There are over 100 compounds of this type that have been isolated now from plants and from fungi, but luckily only a few of them actually have biological activity, so we don't need to deal with the whole load of them.

These hormones, as has already been said, control a lot of very important processes in plants, including germination, stem height, flowering and male fertility. This [picture of man on ladder with normal sized and elongated plants] is an experiment that is about as old as this Academy. This guy has done what we call a spray-and-pray experiment on these cabbage plants down here, and it has been spectacularly successful. You can see two of the actions of gibberellin in this slide. One is that the distance between the leaves, the internodal lengths, has been greatly enlarged; that is a growth effect of gibberellins. The other one is a bit harder to see, but at the top there are flowering structures which are not present in these cabbages, so gibberellin, as well as making these plants tall, has made them flower.

Figure 2
Gibberellins in agriculture and horticulture
(Click on image for a larger version)

Gibberellins are actually very important in both agriculture and horticulture, and these are just a few examples where gibberellins have got a major impact. Firstly, many of the cereals that are grown these days are dwarf cereals, and these are the basis of the green revolution.

On the left of the slide are two rice plants, one a tall plant and the other a dwarf plant. These dwarf plants have increased yields. The reason these rice plants are dwarf is that they have reduced levels of endogenous gibberellins. There are also numerous examples of the use of gibberellins in the horticultural industry. In a kind of opposite example to the cabbage that was flowering when gibberellin was applied, if you apply gibberellin to citrus fruits it actually promotes more leaf growth and reduced flowering. This is used commercially to reduce the number of flowers on a plant, and so to get fewer but bigger fruit. Similarly, gibberellins delay the ripening of oranges, and this again has commercial importance. Finally, in the table grape industry, pretty much every table grape you eat has been sprayed with gibberellin to actually get it to that big, juicy state that you want to have.

So I think you can see from this that gibberellins are actually a very important group of hormones in plants. That is the reason why we are interested in working with them and finding out more about how they are made and what they do.

I have split today's talk into two parts. In the first part I am going to talk about the work that we did to understand the biosynthesis of gibberellins, and in particular the isolation of some of the genes involving the gibberellin biosynthetic pathway. Then, in the second part of the talk, I will describe some work that we did to look at whereabouts in the cell gibberellins are made.

Figure 3
Three enzyme classes are involved in the synthesis of gibberellins
(Click on image for a larger version)

This is a very simplified version of the gibberellin biosynthesis pathway. This was worked out probably 30 or 40 years ago by old-fashioned plant physiologists feeding radio-labelled compounds and looking at what happened to them.

Basically, the pathway starts with this linear molecule, geranylgeranyl diphosphate. The first two reactions are cyclisations of this linear molecule to produce this compound kaurene, which is the first committed step of the pathway. From there, there are a series of oxidation reactions, first on this position here and then over on this 7[-carbon] position here, to give GA12, which is the first gibberellin.

I should say a little bit about the nomenclature here, because this can get quite confusing. The gibberellins are numbered from 1 to 100-and-whatever, and they are just numbered in order of discovery. That adds somewhat to the confusion of these pathways.

Later on I will talk about gibberellin biosynthesis mutants in Arabidopsis, and unfortunately they are also numbered ga1, ga2, 3, 4, 5.

Anyway from this GA called GA12 there is actually a branch in the pathway, and the difference here is that this 13-carbon position can be either hydroxylated or not hydroxylated. This actually differs between different plant species. In Arabidopsis there is very little 13-hydroxylation, but in peas there is actually quite a lot. We don't know whether there is any functional significance to that 13-hydroxylation.

There are then yet more oxidation reactions, and where the pathway ends up is with these two compounds, GA4 and GA1. These are the actual biologically active GAs, so putting those GAs on the plant actually elicits a response. All these other compounds give no response in the plant.

Finally, right on the end here are some inactive metabolites of the active GAs, and these have been hydroxylated at this 2[-carbon] position. This is an irreversible reaction that takes the active GAs out of circulation.

There are three different enzyme classes involved in this pathway. The first group are members of a large protein family called the terpene cyclases. The second group of enzymes are again members of another large multigene family, called the cytochrome P450s. And finally, there is a third group of enzymes called dioxygenases, which catalyse these late steps of the pathway and also these inactivation steps.

Figure 4
The model plant Arabidopsis has been used to understand gibberellin biosynthesis
(Click on image for a larger version)

A lot of our understanding of how this pathway works and of the genes that encode the enzymes in this pathway comes from work with Arabidopsis. The reason we use Arabidopsis is that it is a plant with a very small genome, it has now been completely sequenced, it has got a very short life cycle, and it turns out to be very easy to transform as well. It is also very easy to generate mutants in this plant, in pretty much any characteristic you want. Some of the earliest mutants that were identified were a set of gibberellin-responsive dwarfs. The one I have shown here is known as GA3, and as you can see it is a small plant. Although it has flowered, it has not produced this bolt structure that you see if you apply gibberellin to this plant. Applying gibberellin restores it to its wildtype phenotype.

Figure 5
Dwarf mutants of Arabidopsis define most of the enzymes of the gibberellin biosynthesis pathway
(Click on image for a larger version)

From the work in Arabidopsis and also in other species, most of these enzymes have been isolated. This has been a subject of worldwide interest, so groups in America, Japan and the UK have also been involved in trying to isolate these genes. So the genes for both these two terpene cyclase steps have been isolated. They are now called copalyldiphosphate synthase and kaurene synthase. In this later part of the pathway, all these dioxygenases have been isolated. So initially there were Arabidopsis mutants in these first two steps, the GA 20-oxidase and GA 3-oxidase, which lead to the bioactive gibberellins. It turns out that these enzymes are multifunctional, so the GA 20-oxidase can act either on GA12 or GA53, and similarly the GA 3-oxidase can act on either GA9 or GA20. And these enzymes are also encoded by small multigene families, so there are at least three active 20-oxidases in Arabidopsis and at least two active 3-oxidases in Arabidopsis, and they have different tissue-specific expression patterns.

Other work, not really related to Arabidopsis mutants, has also identified the 2-oxidase, which is this inactivating step at the end of the pathway.

Figure 6
P450-catalyzed steps in the gibberellin biosynthesis pathway
(Click on image for a larger version)

The focus of my research has been these P450-mediated steps from kaurene to kaurenoic acid. These were somewhat harder to work on, because the cytochrome P450s are actually membrane-bound proteins, and this makes them more difficult to isolate and also to express in an active form.

Cytochrome P450s in Arabidopsis, as I have said, are a large multigene family. From the genome sequence we know that there are around 272 cytochrome P450s present. We only know the activity of maybe 10 per cent of those. I think one of the reasons for this is that these enzymes also require cofactors for activity. So the little red blob up here represents heme; that is the same cofactor that is used in hemoglobin. These proteins also require a specific cytochrome P450 reductase for activity.

So, going back to the steps of the pathway, now blown up in more detail: we had an Arabidopsis mutant that was blocked in these first steps between kaurene and kaurenoic acid. We knew this because these plants were accumulating kaurene and could be restored to wildtype by applying kaurenoic acid but not any of these intermediates [kaurenol and kaurenal]. So because we had a mutant we could try and isolate that gene by map-based cloning. This is a process that is very easy in Arabidopsis, and I will very quickly try and explain that.

Figure 7
The Arabidopsis GA3 gene was cloned by mapping
(Click on image for a larger version)

What we had to start off with was our GA3 mutant in one ecotype, called Landsberg, and either side of that mutation we had two other mutant genes which I am just going to call A and B for simplicity. We then cross that to a plant that has got wildtype copies of all three genes and is in a different ecotype [Wildtype Columbia ecotype]. We can distinguish between these two ecotypes because there are small differences in the nucleotide sequence between them: one in every thousand bases turns out to be different and so we can use molecular methods to distinguish between them.

So we cross those two plants, get an F1 population and then self that population. What we are now looking for is events where there has been a recombination between one of these wildtype gene copies A and B and the GA3 mutant gene. This is a slight simplification of the whole scheme.

What we can do in those plants now is to use the fact that we can tell the difference between the red DNA that has come from Columbia and is carrying this wildtype gene and the blue DNA that has come from the GA3 mutant plant, to delineate the region where the gene should be. So this ends up giving us a gene region, and in this case there is a cytochrome P450 gene in the middle of that.

Figure 8
GA3 encodes kaurene oxidase
(Click on image for a larger version)

What we can then do is put that back into a mutant plant, and this produces this plant here, which looks the same as wildtype. So we have complemented that mutation.

As a final step, we wanted to know exactly what the activity of the gene we had isolated was, so we expressed that protein in yeast and fed it with the substrate kaurene, and it produced all these intermediates in the pathway up to kaurenoic acid. And so this enzyme is now referred to as ent-kaurene oxidase.

Figure 9
P450-catalyzed steps in the gibberellin biosynthesis pathway
(Click on image for a larger version)

We had isolated this first gene, but the second set of these P450-mediated reactions was proving a bit more difficult, because there were no Arabidopsis mutants available and we had 272 cytochrome P450s to look at, which is too many to just go and look at them by chance and, hopefully, find the right thing. We had a few clues, though.

In a separate experiment which I don't have time to describe today, we had identified a P450 which is known as CYP88A and is highly expressed in pumpkin seeds. This P450 had also been identified as a mutant gene in two dwarf mutants in maize and barley, but we didn't really know where the block was in the pathway in those mutants. And the final bit of evidence that gave us some hope that we were looking at the right thing is that there are actually two copies of this gene in Arabidopsis, so if one of those genes has mutated the other one can still be active, and so we may not uncover a mutant in that particular gene because of this gene redundancy.

Figure 10
CYP88A catalyzes the three step oxidation of KA to GA12
(Click on image for a larger version)

To test whether this was really a gene in the gibberellin biosynthetic pathway, we expressed this protein in yeast that was also carrying this cytochrome P450 reductase. We were actually quite surprised to find it not only carried out one of these reactions, it actually carried out a three-step reaction from kaurene/kaurenoic acid through to GA12. So it is all oxidations on this 7-carbon position here. So this enzyme is now known as kaurenoic acid oxidase.

Figure 11
Gibberellin biosynthesis is regulated by the amount of active gibberellin
(Click on image for a larger version)

With those two genes isolated, we now know the genes for pretty much all the enzymes of the gibberellin biosynthetic pathway. The only one we are missing is this 13-hydroxylase that catalyses this reaction that seems to be different between different species. We really have no clue about how to isolate that one at the moment.

One of the things that have come out of other people's work is that this pathway is actually under very tight control. The amount of active gibberellin exerts a very tight control on the expression of the genes in the pathway. So the amount of these two gibberellins GA4 and GA1 feeds back on the expression of the GA 20-oxidase genes, and so if you have got increasing amounts of gibberellin it is turning down the pathway. Conversely, the amounts of gibberellin feed forward on the expression of these inactivating enzymes for GA 2-oxidases, and so again when there is high GA this metabolising enzyme is going to be turned on and remove GA from the pathway. This presumably is the way that the plant keeps a very tight control on the amount of gibberellin around, and so when it responds there will be a peak of gibberellin and then it can be restored back to its normal levels.

Figure 12
GA biosynthetic P450s with similar functions have evolved separately in plants and fungi
(Click on image for a larger version)

I mentioned very briefly that gibberellins are also made by fungi. These are actually pathogenic fungi of rice, which make the rice grow tall. Work in Germany has isolated kaurene oxidase and kaurenoic acid oxidase genes from these fungi, and what is quite striking is that these two P450s that come from fungi are quite unrelated to the P450s kaurenoic acid oxidase and kaurene oxidase that catalyse the same reaction steps in plants. It suggests that the gibberellin biosynthetic pathway has probably evolved twice in these different organisms, which I think is quite remarkable given the complexity of the pathway.

I don't have time to show you all the data here, but it appears that, at least in the case of kaurenoic acid oxidase, the reaction mechanism is actually very similar as well.

Figure 13
Where is GA biosynthesis located in the cell?
(Click on image for a larger version)

In the second part of my talk I am going to show you some work that we did to try and understand where gibberellin biosynthesis is located in the cell.

I have already told you that there are these three enzyme classes involved in this biosynthetic pathway. We know that the substrate geranylgeranyl diphosphate is made in the chloroplasts, and P450s are classically located in the endoplasm reticulum, and the dioxygenases are supposed to be soluble. So we thought this was worth investigating further.

The way that we did this was to use a marker called green fluorescent protein [GFP]. This [slide] shows two cells that have been transfected with a DNA vector that will express this protein. It glows green and you can see it under a confocal microscope. GFP on its own tends to accumulate a bit in the nucleus and round here in the cytoplasm.

What we do is to make a fusion protein of whatever protein we want to look at to GFP, and we are hoping that the targeting information in this piece of the protein will take it to the part of the plant cell where it normally goes to. So if we take a chloroplast targeted protein we get this kind of picture, where the chloroplasts have now gone green. This is an endoplasm reticulum targeted protein and you can see these little strands through the cell.

Figure 14
GFP fusions localize gibberellin biosynthesis enzymes in three cell compartments
(Click on image for a larger version)

So we did this experiment using the enzymes in the gibberellin biosynthetic pathway, and not surprisingly the first two steps of the pathway were targeted to chloroplasts – we knew that already. What was more surprising was when we looked at the cytochrome P450s. The kaurenoic acid oxidase is targeted to the endoplasm reticulum, but this first one, kaurene oxidase, is actually targeted to the chloroplast. The late steps are targeted to the cytosol, as we expected.

Figure 15
Kaurene oxidase is located on the outside of chloroplasts
(Click on image for a larger version)

We then went on to look further at where in the chloroplast the kaurene oxidase was going. We can do experiments where we can show which side of the membrane kaurene oxidase is on – which I don't have time to tell you about today – and the results of those experiments were that kaurene oxidase actually sticks onto the outside of the chloroplast. This makes quite a nice model for how gibberellin starts off from geranylgeranyl diphosphate and is made into kaurene, which is quite hydrophobic so presumably it partitions into this membrane, and kaurene oxidase is sitting there ready to take it out into the rest of the pathway.

Just to summarise: most of the genes of the gibberellin biosynthetic pathway have now been isolated, so we are beginning to get a better understanding of how this pathway works. We also know how it is partitioned across the plant cell, and we are beginning to understand how it is regulated.

There are, I guess, some implications for agriculture and horticulture. The dwarf rice plant in that first slide I showed you turns out to be mutant in one of these GA 20-oxidases, which is one of those later steps of the pathway. It may be that this work will have impacts on the citrus and grape industries, where we can perhaps alter the timing or control of when GA is put on, and there may be implications for its use in dwarfing genes in wheat.

Figure 16
High yield dwarf wheat can't detect gibberellin
(Click on image for a larger version)

Very quickly, I will tell you that wheat also has dwarfing genes, which again were very important in the green revolution. These dwarf plants are not actually mutant in gibberellin biosynthesis; they are actually unable to respond to GAs. So it is a different type of mutation.

One of the problems with these types of wheats in Australia is that if you sow them deep, because they don't have any gibberellin there they find it quite hard to germinate. So you get this kind of field here, where the plants are really struggling to emerge.

Figure 17
Gibberellin-responsive plants have better early growth
(Click on image for a larger version)

One prospect is that if we can use gibberellin biosynthesis genes that are knocked out in the vegetative parts of the plant but not during germination, we may be able to get round this type of problem and get wheat fields looking like this, where the plants start growing properly early on.

With that I will finish up and just thank my colleagues at Plant Industry and collaborators at the University of Tasmania and Cambridge.

Questions/discussion

Question: Have you given any thought to blocking the oxidase that actually breaks down the active components, by blocking the GA 2-oxidase?

That has been done. So you can antisense those out, and yes, those plants can accumulate more gibberellin. The problem is that the control is quite tight, so if you increase the amount of GAs you still get this feedback on the 20-oxidases. So the best effects you can have are by doing things like overexpress the 2-oxidase and remove GAs. That is very effective.

Question: You said that some of this stuff is used in agriculture. How is it manufactured?

I said there is a fungus that makes a lot of gibberellin, so I think it is mainly by growing up that fungus and isolating it from the fungus.

Question: You said that this stuff was sprayed on. Will it enter the cell? How does it get into the plant?

I guess we assume it gets into the plant somehow. When we grow Arabidopsis we quite often just grow the plants on plates within the media, but you can also spray it onto the leaves. I guess that you maybe need some sort of surfactant there so it actually gets into the stomata. But it can probably cross cell membranes.

Question: Your Division has been very involved in the GMO debate, a debate not always following entirely logical pathways, particularly from one side of the fence. But if there were any logic, would the people who hate us inserting nasty genes into our foodstuffs react differently if the plant had a gene knocked out? Would a knockout, such as the one that you are describing as potentially useful, silence the GMO critics?

I suspect it probably wouldn't. I guess one of the potentials with this pathway is that it seems to be actually very easy to get mutants that knock out gibberellin biosynthesis to some degree or other. So you may actually be able to do that without making a GMO, and it is probably just a matter of looking at the right plants to get it.

Question: In the fungus that makes gibberellin, does it actually regulate the growth of the fungus or is it just there to affect the host plant?

That is a good question, and I'm not sure I can really answer it very well. Presumably it has evolved to do something to the host plant, although it seems kind of crazy to make the plant grow bigger. I am not sure what advantage that confers on the fungus, but presumably there must have been some strong selection to get that whole pathway operating in a fungus. As far as I am aware, it doesn't have any growth effects on the fungus itself.

Question: Since it is sprayed onto grapes, as you said, if you don't wash your grapes and you eat it, since it is a hormone does it have any effect on humans? And what is it most similar to in human hormones?

As far as I know it has never been shown to have any effects on humans. It is probably most similar to a steroid hormone, but I think it has got fewer rings. The structure is quite different. One of the P450s, kaurenoic acid oxidase, is quite similar to steroid biosynthetic enzymes, but that is probably as far as the similarity goes.

Question: Do any of these gibberellins affect our own cytochrome P450 pathway and therefore the metabolism of many of our drugs?

The relationship between these P450s and the human ones is relatively weak.

Question: If you go back down the evolutionary pathway, where did gibberellins first appear?

I think they are unique to plants, but I am not much of a botanist – I'm not going to give you any Latin names or anything. They presumably evolved separately in fungi.

Question: Do you know if it is in, for example, the Australian bush tomato? I was told when I went on an Aboriginal tour of the Botanic Gardens that Aboriginal women used to use this to prevent having babies. I think it was probably an abortifacient rather than a contraceptive. Certainly the Aborigines knew that you wouldn't have babies if you ate the green Australian bush tomato.

I am sure that gibberellins would be present in all Australian plant species. I would be surprised if that were the active compound in that case.