Main

Those students in fly laboratories that view Drosophila more as a model than as an organism are most familiar with its striking capacity to regulate size. When a new generation of flies is neglected in the old vial instead of being transferred to a fresh one, further progeny suffers from nutrient limitation and consequently develops into tiny flies up to five times smaller than those reared with optimal care. Of course, even without neglecting flies, we all know that environmental factors such as nutrition or functional load can regulate the size of organs and organisms within the limits set by the genome, but the molecular basis of size control in animals has rarely been addressed1. Recently, however, Galloni and Edgar2 described the initial results of a systematic screen for mutations that result in an arrest of larval growth in Drosophila. The first identified genes function in the context of translation of messenger RNA into protein. In a complementary study, Böhni and colleagues3 show that chico, which encodes a Drosophila homologue of mammalian insulin-receptor substrates 1–4, is required for growth of flies to normal size. The growth-regulatory role of insulin-like growth factors is not, therefore, restricted to mammals, and translational control is presumably a primary target of this conserved signalling pathway.

At this point, it is worth noting that the size of organs and organisms is regulated by both cell number and cell size. Cell number is itself determined by the balance between cell proliferation and cell death. Cell size is also regulated independently.

Drosophila larvae hatch from the egg after one day of embryonic development and pupate four days later, having increased 200-fold in size. A screen for mutations that result in arrest of this larval growth is an obvious way to try to identify growth-regulatory genes. But this strategy suffers from the problem that too many mutations can harm cells or larvae, and hence result in slow growth before the suffering larvae eventually die. To identify genes more specifically involved in growth regulation, Galloni and Edgar2 took larval-growth-arrest mutations through a series of follow-up investigations, the rationale behind which was based on their previous analysis of the effects of nutrient deprivation on larval growth4.

In this study4, newly hatched wild-type larvae survived for about eight days when provided with sucrose as an energy source. Under these conditions, however, larvae fail to grow, because growth requires amino acids in the diet. Interestingly, a few characteristic brain neuroblasts continue to incorporate bromodeoxyuridine (BrdU) — a marker of DNA replication — despite starvation. Ectopic expression of the transcription factor E2F/DP also promotes the S phase of the cell cycle (DNA replication) in many tissues of the starved larvae. So the lack of dietary amino acids does not result in a general inability to synthesize the cellular proteins required for cell-cycle progression. Dietary amino acids might instead be required for the production of signals that induce cell growth. This interpretation is supported by the observation that BrdU incorporation in the central nervous system isolated from starved larvae requires, in addition to dietary amino acids, an unknown signal that can be provided by co-culture with the larval fat body, the insect equivalent of the vertebrate liver.

On the basis of the results of these starvation studies, a screen for mutations that cause larval growth arrest should ideally reveal those that do not interfere with survival for at least eight days — such extended survival would mean an absence of general metabolic abnormalities. Ectopic E2F/DP expression should induce vigorous BrdU incorporation in these mutants. Galloni and Edgar identified this class of mutations2.

They then used clonal analysis to determine whether or not these mutations behave in a cell-autonomous manner. Clones of cells homozygous for mutations in genes involved in the production of a diffusible growth-inducing signal are expected to behave differently to clones homozygous for mutations in genes required for the interpretation of such signals, as cell proliferation in the former clones, but not in the latter, should be rescued by signals generated in the non-homozygous background. Both types of mutation were found2.

Finally, as they screened mutations generated by ‘P element’ transposon insertions, Galloni and Edgar were able to succeed rapidly in the molecular analysis of selected genes. They found seven allelic P-element insertions within the gene encoding eIF4A, a protein that initiates mRNA translation into protein. The bonsai gene was found to encode a protein with a stretch of similarity with prokaryotic and yeast mitochondrial ribosomal protein S15. In the case of plume and poney, the P-element insertions might affect the genes encoding the ribosomal protein L30 and the aspartyl-transfer-RNA synthase, respectively.

Do these results convey more than just the trivial fact that protein synthesis is required for growth? The answer should not be given without keeping in mind that all of the mutants identified are perfectly capable of synthesizing DNA after ectopic expression of a transcription factor (E2F/DP), arguing against the idea that the growth arrest is the consequence of a general inability to express the genome. Moreover, the proliferation of bonsai mutant cell clones in wing and eye imaginal discs is normal, consistent with the theory that bonsai is involved in the production of a signal required for cell growth, not proliferation, in larvae. Analysis of other genes identified in this screen may reveal the proteins of which translation is so important.

Studies of mammalian cells are filling in the molecular details of the long-recognized importance of translational control in the context of growth regulation (Fig. 1). eIF4F, the protein complex that recruits the small ribosomal subunit to mRNA ready for translation, and the ribosomal protein S6 appear to be particularly important1,5. The eIF4F complex is composed of eIF4A, G and E. eIF4E binds the cap structure at the 5′ end of the mRNA. eIF4F activity is inhibited by eIF4E-binding proteins (4E-BPs) which inhibit binding of eIF4E to G. Translational repression by 4E-BPs is opposed by phosphorylation of these proteins, which is induced by a signalling pathway that also regulates S6. This pathway involves phosphatidylinositol-3-OH kinase (PI(3)K), the phosphoinositide phosphatase PTEN, the serine/threonine kinase Akt/protein kinase B (PKB), the kinases PDK1 and FRAP/mTOR, and p70S6kinase. The pathway is part of the complicated signalling network that controls cell metabolism, proliferation and apoptosis in mammalian cells. It is activated by receptor tyrosine kinases for platelet-derived growth factor, insulin and insulin-like growth factor (IGF) upon ligand binding. In the case of insulin and IGFs, receptor activation results in phosphorylation of insulin-receptor-substrate (IRS) proteins and so generates binding sites on the IRS proteins for PI(3)K.

Figure 1: Size control by insulin/insulin-like-growth-factor (IGF) signalling and translational regulation.
figure 1

This pathway has been implicated in the regulation of cell, organ and organism size. Activation of the receptor tyrosine kinases for insulin/IGF by ligands results in autophosphorylation (circled ‘P’s) of tyrosine (Y) residues. Insulin-receptor-substrate (IRS) proteins are then recruited. (The Drosophila gene chico encodes a protein related to the mammalian IRS 1–4 proteins.) Receptor kinase activity phosphorylates tyrosine residues in IRS proteins, resulting in the recruitment of phosphatidylinositol-3-OH kinase (PI(3)K) through an adaptor subunit. The Drosophila insulin receptor has an extra carboxy-terminal domain (red) that can also be recognized by the PI(3)K adaptor. Activation of PI(3)K generates phosphatidylinositol-3-phosphate (PtdIns(3)P) which in turn activates protein kinase B (PKB). Further downstream events include translational regulation, presumably through activation of p70S6kinase and the eIF4F complex (subunits of which are eIF4A, G and E; this complex is inhibited by the eIF4E-binding protein, 4E-BP). The proteins in lilac boxes are involved in size regulation, according to genetic analyses in Drosophila.

Full size image

Böhni and colleagues3 have analysed the function of the gene encoding a Drosophila IRS homologue. This gene, named chico, is required for normal growth. Mutants grow less extensively during larval stages and develop into flies that are more than 50% smaller than control flies. Except for being smaller, the mutant flies do not have morphological pattern defects. These tiny flies lay small eggs and are composed of fewer and smaller cells, just like the flies that develop in adverse food conditions. The mutant cells do not appear to have a higher propensity to undergo apoptosis but they progress slowly through the cell cycle.

Genetic mosaics show that chico is autonomously required for the regulation of the size of cells and organs (at least in the analysed case of eye and head capsule). chico interacts genetically with the Drosophila insulin receptor and with PI(3)K, which have been implicated previously in size regulation at, respectively, the organism and the cellular levels6,7. Phenotypes produced by mutation of Drosophila PKB and p70S6kinase (H. Stocker and E. Hafen, personal communication; J. Montagne, M.J. Stewart and G. Thomas, personal communication) are similar to the chico mutant phenotype. It is tempting to speculate that the insulin-receptor pathway normally regulates growth in response to nutrient conditions through translational control. Poor nutrition might cause a drop in insulin-like peptide production, perhaps in the fat body, resulting in lower levels of these peptides in the haemolymph. This may result in lower activation of the insulin-receptor pathway and so less opposition to translational repression by the eIF4E-binding proteins.

Böhni et al.’s results3 further emphasize the functional and structural conservation of the insulin signalling pathway in metazoan evolution. In Caenorhabditis elegans, this pathway is required for the physiological response to nutrient limitation during development; such a limitation results in a developmental arrest and differentiation of dauer larvae with a characteristically altered metabolism and increased longevity8. In mammals, insulin and IGF are also required for normal growth during development and for metabolic regulation. With regard to metabolic control, it is interesting that chico mutant flies have an almost twofold higher lipid content than wild-type flies3. However, the role of the Drosophila pathway appears to be more specific than might have been anticipated from the complex signalling network described in mammalian cell culture studies: chico does not appear to be involved in apoptosis or pattern formation.

Given the evolutionary conservation of the insulin signalling pathway, Böhni et al. speculate that it might have been a target for the modulation of body and organ size during evolution. But we need to postulate further mechanisms to explain the vast evolutionary variations in body size and proportions within species. These variations are often not accompanied by changes in cell size, as occurs in response to variations in chico function. Moreover, the reduction of wings to halteres that has evolved in dipteran flies is controlled ontogenetically by the homeotic gene Ubx. If Ubx were to achieve relative growth inhibition in the haltere imaginal disc by inhibiting the insulin signalling pathway, we would expect disproportionate growth of wing and haltere discs in chico mutants. The proportional size reduction of all body aspects in chico mutants argues for an involvement of the insulin signalling pathway specifically in the global integration of growth control at the organism level.

Elimination of chico function is not the only way to generate small cells in flies (but it is unusual in that it also affects organism/organ size). Clones of haploid cells that are smaller than diploid cells have no effect on overall wing size and pattern9. Clones of smaller cells generated by accelerating progression through the cell cycle, for instance by overexpression of E2F/DP, also have no effect on overall wing size and pattern10. Judging from the organ-autonomous size reduction observed when Böhni et al. generated homozygous chico mutant eye imaginal discs in a heterozygous background, large clones of chico mutant cells in the wing might be expected to affect wing size. Studies of wing discs will be important to evaluate whether pattern regulation really fails in the case of chico mutant cell clones, in contrast with the efficient regulation observed in clones of haploid or rapidly cycling cells (or slowly growing minute mutant cells).

The growth of organisms and organs is regulated by growth factors, mitogens and survival factors. If we are to understand the regulation of growth during animal development, we need careful analysis of the rates of cell growth, cell-cycle progression and apoptosis in an organism accessible to genetic manipulation. The work of Galloni and Edgar2 and of Böhni et al.3 exemplifies the tremendous value of Drosophila when exploring this difficult issue.