Leevers: I would like to give a brief summary of what is known about the role of signalling by the insulin/insulin-like growth factor (IGF) pathway during the growth of Drosophila imaginal discs, and then contrast this with what we have heard in the previous papers in this book. There is a pathway in Drosophila that is highly homologous to the insulin/IGF signalling pathway in mammals (Fig. 1 [Leevers]). A number of different labs have studied various genes on this pathway (Edgar 1999, Leevers 1999, Lehner 1999). The pathway is activated by an insulin/IGF receptor homologue, Inr, which autophosphorylates and phosphorylates an insulin receptor substrate protein, Chico (Chen et al 1996, Bohni et al 1999). This results in the activation of a class IA phos-phatidylinositol-3-kinase (PI3K) made up of a catalytic subunit (Dp110) and adaptor subunit (p60) (Leevers et al 1996, Weinkove et al 1999). In addition, at least two downstream components have been identified and studied: they are the serine/threonine kinases, DAkt1 and DS6K, and the Drosophila p70S6 kinase (Verdu et al 1999, Montagne et al 1999). Akt has multiple targets in mammalian systems, though we don't know which are the relevant ones for its role in growth regulation in Drosophila. p70S6 kinase has a specific role and increases translation of a specific family of mRNAs, many of which encode ribosomal proteins.
We have done a number of experiments using an approach similar to that taken by Bruce Edgar to investigate the function of this pathway, by manipulating the activity of the PI3K, Dp110/p60. For example, we have made clones of cells overexpressing active or dominant negative transgenes and looked to see what happens to clonal growth (mass increase) and cell size in the imaginal disc as it is developing. These experiments have shown that if you increase signalling via this pathway, you increase growth rate. They have also shown that increasing signalling via this pathway increases cell size (Weinkove et al 1999). If the activity of this pathway is manipulated in a region of or throughout the imaginal discs, the final size of that imaginal disc is altered as is the size of the adult organ that it gives rise to (Leevers et al 1996).
To summarize, modulating signalling by this pathway alters three parameters: growth rate, cell size and organ size. What is unclear is the nature ofthe relationship
FIG. 1. (Leevers) The insulin/IGF signalling pathway in Drosophila.
between these parameters. Simplistically, these results suggest that when growth rates are altered, cell division is unaffected, so the altered growth rate alters cell size. Likewise, if you alter growth and the developmental programme and patterning machinery are unaffected, then the altered growth rate alters organ size. However, there are several examples of the way in which other molecules modulate growth, which suggest that this explanation of the results is too simplistic.
Firstly, Bruce Edgar has shown that if Myc or Ras activity is increased in the imaginal discs, then like the insulin/IGF pathway, this increases growth rate and cell size (Prober & Edgar 2000, Johnston et al 1999). In contrast to the insulin/IGF pathway, when Ras and Myc activity is increased, organ size is not altered. Somehow the imaginal discs seem to accommodate the changes in growth induced by Ras and Myc. Presumably compensatory apoptosis or effects on cell division may ensure that correct final organ size is achieved. Secondly, Bruce Edgar and Christian Lehner have spoken about thickveins and cyclin D/Cdk4. Overexpressing these molecules promotes growth (i.e. increases clonal growth) but does not affect cell size. Overexpressing these molecules also increases final organ size. Thirdly, if growth rate is altered by slowing global protein synthesis, cell size and organ or organism size are not necessarily affected. This observation is based on studies of a class of mutations in Drosophila called Minutes. Minute genes, when mutated, dominantly slow growth. Not all of them have been cloned, but those that have are ribosomal genes. Heterozygous Minute organisms grow slowly and are developmentally delayed, but in most cases the cells in the imaginal discs of these organisms do reach appropriate sizes. Likewise, Minute organisms usually reach the right final size, so there seems to be a coordinated slowing of growth and development.
"Hunt: How much longer to they take to get there?
Leevers: A two or three day extension of the larval period (when disc growth occurs) is typical. The larval period is normally four days.
Thomas: They are usually hypomorphs.
Nasmyth: Have you or anyone else looked at the epistasis of these effects? PI3 kinase affects all three parameters: growth rate, cell size and organism size. It could be that it independently goes for all three things — this would be the simplest null hypothesis. The prediction of that would be that if you over-produce PI3K in a Minute background or Cdk4 background, you should still get this dramatic effect on cell size and organ size.
"Leevers: If you did that experiment you would probably find that PI3K still increases growth rate. I wouldn't predict that a Minute mutation would kill the effect that PI3K has on growth.
Nasmyth: You would predict, though, that it would still produce a big fly, even though it took longer to get there.
Edgar: I would agree with that.
Bryant: Minutes have small bristles. This would be an affect on cell size.
Leevers: That is true. Most of what we are looking at here in the discs is growth in mitotic cells. Bristle size may mainly reflect post-mitotic growth.
Edgar: Some of the Minutes have been reported to give small cell phenotypes in adults. We have looked at a number in discs before the cells differentiate, and we have not seen any effects here.
Hunt: What happens if you just put a low dose of cyclohexamide in their food?
Leevers: I expect you would get a Minute phenotype. That is what happens ifyou put rapamycin, which inhibits p70S6 kinase activation, into their food.
Thomas: We have identified a mutation in the Drosophila S6 kinase (DS6K) in which we obtain a few adults. These flies have smaller cells and the animal is about 50% smaller. Initially we were quite happy with this phenotype because we knew from work that we had done in mammalian systems that this kinase was involved in regulating the expression of ribosomal protein (Rp) messages, which would have a direct effect on translation. We then read Bruce Edgar's paper describing cell cycle mutants and realized that Minute cells are not smaller. Therefore one can distinguish two events. In the DS6K mutants the rate of growth is slowed down quite dramatically. The flies that emerge are quite sick, and only a few survive. They are also quite lethargic. Thus these flies are delayed due to a decreased rate of growth. The second event we need to explain is why the cells of these mutants pass through the cell cycle at a smaller cell size. We thought initially that this had something to with translation rather than the kinase regulating a cell cycle checkpoint for cell size. Indeed, a more simple explanation might be that unlike the Minute, in which only the expression of an Rp message is suppressed, which is then rate-limiting for ribosome biogenesis, in the S6 kinase mutants all Rp mRNAs are being suppressed. The Rp mRNAs are represented by a small gene family, but they can make up 20-30% of the cell's total mRNA. One must also consider that mRNA is in excess over ribosomes. So these Rp mRNAs essentially monopolize the translation machinery. The Rp mRNAs are characterized by this polypyrimidine tract at their transcriptional start site. This tract acts as a repressor, not an enhancer, in the sense that if you change the first C to an A the Rp mRNAs are mobilized into polysomes. What started to emerge for us was the idea that if you suppress the translation of these mRNAs, ribosome biogenesis is decreased, which will account for the slow growth rate. But how does the smaller cell size arise? What we think is happening is that by taking these Rp mRNAs away, you favour the translation of mRNAs which don't have a tract. If these mRNAs encode a cell cycle regulator, which in general have long 5'-untranslated regions and very short half-lives, their expression would be favoured. This would then lead to a cell cycle regulator reaching a critical concentration at a smaller cell size. Interestingly enough, in development, at least in the frog and the fly, ribosome biogenesisis is dramatically up-regulated during oogenesis. In the fly, approximately 50 billion ribosomes are made in a few hours at stage 10 of oogenesis. As soon as embryogenesis ensues, they move off polysomes and are stored in mRNP particles, not degraded, until the completion of embryogenesis. During embryogenesis, ribosome biogenesis is suppressed, i.e. when the cell is passing through multiple rounds of cell division with no growth. The explanation by default was rationalized from an experiment performed by Jim Maller's group. They showed that if you treat Xenopus oocytes with a rapamycin, it induces precocious maturation in the presence of progesterone. They saw that this correlated very nicely with the induction of Mos, and they showed that they could block this effect by microinjecting a rapamycin-resistant allele of the S6 kinase. They then examined mitotic cell cycles, showing that addition of rapamycin caused a more rapid and robust rise of Cdc25. We think then growth is antagonistic to proliferation, and maintains its dominance through monopolizing the translational machinery. Thus ribosome biogenesis is initially acting as an antagonist of proliferation, because Rp mRNAs are simply monopolizing the translational machinery when you stimulate cells to grow. Indeed, in a proliferating cell, about 80% of its energy is being used for ribosome biogenesis. As the cell begins to accumulate ribosomes, the pathway begins to desensitze, Rp mRNAs are diluted, and the cell can begin to translate other messages. I wonder whether this might be an explanation for the fact that in the experiments described by Bruce Edgar, Myc drives expression of cyclin E protein without increasing cyclin E mRNA. What may happen is that by driving Myc you are driving ribosome biogenesis as the pool of ribosomes increases, you begin to translate mRNAs such as cyclin E, which are generally not well recognized by the translational apparatus.
Raff: Among other things, this raises the question of whether cells sense size at all, or sense something else. What is the evidence that an animal cell monitors its size?
Edgar: There is rather little. In the disc cells that we have studied, we initially assumed that they coupled their cell cycle to their growth rate, but on closer examination we found that they have a variable size. They start out large and they get smaller with each division as they go through development.
Raff: Many of these experiments that suggest that cells have a size threshold are done in culture, with saturating amounts of extracellular signals that stimulate cell growth and cell cycle progression. In these conditions, cells will tend to be the same size, but these signals are not saturating in vivo.
Nasrnjth: There is a classic experiment to address whether there is size control. That is to transiently delay the cell cycle so that you produce abnormally large cells. Then you look at the cell division time of subsequent cycles. If these experiments are done with bacteria or yeast, the subsequent cycles are greatly accelerated. These experiments could be done by Bruce Edgar and his imaginal discs. They are technically harder, but they could be done.
Edgar: This would show that there is an equilibrium of some sort at that normal cell size, but does this mean that cells are measuring their size?
"Hunt: Martin Raff, you say that cells are variable in size, but my impression is that if you look at a bit of liver or skin, for example, cells are not that variable.
Raff: I think they are actually surprisingly variable. What would you consider to be the most definitive experiment showing that yeast cells care about their size?
Nasrnjth: Weel was discovered as a size mutant.
Raff: But what is this set point, if not the level of some cell cycle regulator?
"Nasrnjth: In the case of the weel system, not very much is known about how fission yeast is doing it. In budding yeast, it is the level of Cln3.
Raff: So is that size?
Nasrnjth: The level of Cln3 is a pretty good metric for cell size.
Edgar: It is not measuring cell size.
Nasrnjth: Well of course it's not getting a ruler out and measuring it!
Raff: That is my point: you talk about cell size when you mean Cln3.
Eeevers: Perhaps cells measure their growth rate, not their size.
Nasrnjth: Cln3 is a transcriptional activator of the Clnl and Cln2 cyclins, which are somewhat similar to cyclin E. It is a very unstable protein whose rate of synthesis is proportional to cell size. Its concentration is going to be directly and immediately proportional to the rate of synthesis. This stuff goes into the nucleus, which is roughly constant in size.
Raff: But simply by altering the concentrations of extracellular growth factors and/or mitogens, you can change cell size.
Nasmjth: Right, because this regulates things like Cln3.
Raff: So the cell doesn't care about size; it cares about Cln3.
Nasmjth: But why does it care about Cln3? Because it cares about cell size! Now we are getting into a teleological argument. The hypothesis is that the cells do care about cell size, and they have to have a mechanism for measuring it. They do this by measuring the concentration of unstable proteins in the nucleus.
Raff: This has been the dogma, but it is based on little experimental evidence.
Nasmjth: It is not testable: how do you test what is the purpose of something in biology?
Bryant: How is it that the concentration of those proteins you are describing is proportional to size?
Nasmjth: This relates to something I was trying to say earlier on. Generally speaking, probably the bigger the cell, given one copy of that gene, the rate of synthesis of that product will be proportional to cell size. This probably acts at every level of a synthetic and degradative pathway. Everybody grinds up their cells, measures the amount of RNA (which is a measure of ribosomal RNA), and then they do a Northern with GPDH. It is always the same. In yeast these sorts of things have a fairly short half-life, so the level is in a sense a measure of the rate of synthesis. It is always constant relative to the number of ribosomes in a cell. This is what those Northerns tell you, otherwise you wouldn't have nice normalized controls. If you have a small cell or a big cell, this doesn't matter; it always looks the same level. The conclusion is that the rate of synthesis of that mRNA must be proportional to the number of ribosomes, which is a measure of cell size. This is just anecdotal.
Leevers: I think it is more likely that there is a link between protein synthesis, growth rate and the production of a specific cell cycle regulator. This is a great mechanism to ensure a minimum cell size, making sure that cells don't divide until they are growing at a certain rate and producing everything they will need to divide at a reasonable level.
Edgar: I wanted to add a general comment about the PI3K/insulin pathway. All the analysis that has been done in flies so far suggests that it is not a pathway that is used to regulate differential growth in terms of forming and shaping organs, but rather is a nutrition-sensing pathway. Other signalling pathways such as bone morphogenetic protein (BMP), Wnt and extracellular growth factor receptor (EGFR) seem to be the ones stimulating differential growth.
Raff: That is probably also true in mammals in the sense that IGFs tend to stimulate proportional growth. They don't seem to be involved in patterning.
Edgar: The question is, how do those other signals (BMP, Wnt, EGFR) feed into protein synthesis, or whatever regulates growth?
Thomas: In the tissue culture model following insulin stimulation, such signalling elements as protein kinase B, PI3K and MAP kinase are all activated in the absence of amino acids or glucose, but the activation of S6K is blocked. If nutrients are returned to the media, S6K is activated. Indeed, raising nutrient concentrations in the media will activate S6K, without addition of a ligand. This is all being mediated by TOR (target of rapamycin), an effector for nutrient sensors. Raff: What about TOR in mammals?
Thomas: If you challenge animal cells with rapamycin, you block TOR function. If you put in rapamycin resistant alleles you can protect against the antibiotic.
Hunt: There is another side to this coin, which is what happens during starvation. There is a hierarchy of what tissues go first.
Raff: If you starve hydra, they just shrink. Cell proliferation carries on almost normally, but cell death (apoptosis) greatly increases. The animals survive by eating their own apoptotic cells.
Was this article helpful?