Shrinkage control: regulation of insulin-mediated growth by FOXO transcription factors

The insulin signaling pathway regulates organismal growth in response to nutrient conditions by controlling a range of metabolic and biosynthetic processes. Recent studies in Drosophila have shown how transcriptional responses to reduced insulin and nutrient levels can act to inhibit growth.

In the transition to multicellularity during evolution individual cells gave up autonomous control over whether to grow and divide, live or die. These processes are regulated instead by a variety of intercellular signals and the network of signal-transduction pathways they activate. Thus, proliferation of a population of cells can be regulated in concert in response to triggers that reflect the needs of the whole organism, such as patterning cues, developmental stage, and environmental conditions. Over the past several years, studies in mammalian cell culture and model organisms such as Drosophila have identified as a dedicated regulator of cell growth and proliferation in response to nutrition the signaling pathway from insulin at the cell surface to phosphatidylinositol (PI) 3-kinase and the protein kinase Akt (also called protein kinase B, PKB) inside the cell [1]. Mutations in this pathway result in profound changes in cell, organ and organism size, and its activation is a critical step in a number of types of cancer. Intensive efforts have therefore been directed towards gaining a molecular understanding of the mechanisms by which insulin signaling promotes growth. Three recent studies [2][3][4], including a paper by Jünger et al. in this issue of Journal of Biology [2], have now addressed the role played by gene expression in mediating insulin-controlled growth in Drosophila.

Signaling responses to insulin
The proximal steps downstream of insulin binding are well understood [5] (Figure 1). In response to ligand binding, the insulin receptor phosphorylates insulin receptor substrate (IRS) proteins (encoded by the chico gene in Drosophila), which act as docking sites for the class I PI 3-kinase. Activated PI 3-kinase increases the levels of the second messenger phosphatidylinositol 3,4,5-triphosphate (PIP 3 ) at the cell membrane; the accumulation of PIP 3 is opposed by the phosphatase activity of a negative regulator of insulin signaling, the tumor suppressor PTEN. An important downstream effector of PIP 3 is the serine threonine protein kinase Akt/PKB. In response to PI 3-kinase activation, interaction between PIP 3 and the pleckstrin homology domain of Akt causes recruitment of Akt to the cell membrane, where it is further activated by one or more additional kinases. Akt appears to be the major critical target of PIP 3 signaling in Drosophila, as mutations in Akt that block its ability to bind PIP 3 can restore viability to animals with high levels of PIP 3 caused by mutations in PTEN [6].
Two signaling branches downstream of Akt have been identified ( Figure 1). One branch of this pathway leads to activation of the target of rapamycin (TOR) and p70 S6 kinases, which promote cell growth through a number of effects including stimulation of ribosome biogenesis [7]. The direct target of Akt in this case appears to be the product of the tuberous sclerosis complex 2 gene [8], TSC2, which was recently found to function as a negative regulator of the small GTPase Rheb, an upstream activator of TOR [9]. Akt phosphorylates and inactivates TSC2, thereby allowing increased activity of Rheb, TOR, and S6 kinase.
A second pathway downstream of Akt was initially identified through genetic studies in Caenorhabditis elegans. Insulin signaling mediates responses to nutrient levels in C. elegans by regulating the formation of a developmentally arrested juvenile form known as the dauer, which can survive starvation conditions for an extended period [10]. Loss-of-function mutations in insulin signaling components mimic starvation, leading to inappropriate dauer formation. A number of years ago, Daf16 was identified as a negative regulator of this insulin-dependent response in worms [11]. Mutations in daf16 can completely suppress the dauer induction caused by reduced insulin signaling. Daf16 was found to encode a transcriptional regulator of the Forkheadbox type O (FOXO) class of Forkhead-related factors, thus indicating that control of gene expression is a major output of insulin signaling in worms. Subsequent studies in cultured mammalian cells extended these results, showing that FOXO factors are negatively regulated by the insulin/PI 3kinase/Akt pathway. In response to increased insulin levels, activated Akt phosphorylates FOXO on multiple sites, resulting in its nuclear exclusion [12]. Upon reduced insulin signaling, FOXO becomes dephosphorylated and accumulates in the nucleus, where it acts to regulate the transcription of a number of target genes.

Growth control by FOXO factors
Could FOXO-regulated transcription play a role in growth regulation by the insulin/PI 3-kinase pathway? Several lines of evidence point to such a role. First, overexpression of any of the three mammalian FOXO homologs, FOXO1, FOXO3a or FOXO4, leads to growth arrest in a variety of cell types [12]. Increased levels of insulin can suppress the growth arrest caused by overexpression of wild-type FOXO, but not of FOXO mutants lacking Akt phosphorylation sites. Second, FOXO factors regulate expression of a number of regulators of cell proliferation including p27 kip1 , cyclin D, and the Retinoblastoma-related protein p107. Induction of p27 kip1 , an inhibitor of cyclin-dependent kinases, appears to be a critical step in cell-cycle arrest by FOXO. The transcription of p27 kip1 is directly induced by FOXO factors in response to low insulin levels, and cells lacking the kip1 gene are highly resistant to growth inhibition by expression of FOXO or inactivation of PI 3-kinase [13]. In addition, transcription of cyclin D is negatively regulated by FOXO, and forced expression of cyclin D can partially bypass FOXO-induced arrest [14]. Finally, a number of chromosomal translocations involving FOXO members are associated with neoplasias. For example, a t(1;13)(p36q14) translocation found in rhabdomyosarcomas results in fusion of a portion of FOXO1 with the PAX7 gene [15].
A potential limitation to the conclusions from these studies is that most were performed in cultured, transformed cells using non-physiological levels of transgene expression. Thus, the relevance of FOXO factors and their potential targets in growth mediated by insulin and PI 3-kinase in vivo The dFOXO protein mediates a transcriptional response to insulin signaling. Under conditions of abundant nutrients, dFOXO is retained in an inactive state in the cytoplasm due to phosphorylation by Akt. When insulin levels fall, dFOXO is dephosphorylated and translocated into the nucleus, where it stimulates transcription of 4E-BP and presumably other negative regulators of growth. In addition, active dFOXO increases expression of the insulin receptor gene [4], which may result in increased insulin sensitivity under low insulin conditions. remains unclear. Indeed, genetic studies have suggested that downregulation of TSC2 and subsequent activation of the TOR/S6 kinase pathway may be the central function of insulin signaling in regulating cell growth [16].
As now described by Puig et al. [4], Jünger et al. [2] and Kramer et al. [3], addressing this question in Drosophila allows analysis of both overexpressed and endogenous FOXO in a variety of in vivo conditions. The fly genome encodes a single FOXO ortholog, dFOXO, whose sequence includes three Akt phosphorylation consensus sites similar to those found in mammalian FOXOs and nematode Daf16. As in these proteins, phosphorylation of dFOXO is stimulated by Akt activation in response to insulin, and this results in turn in its cytoplasmic localization and transcriptional inactivation [4]. Each of the three studies [2][3][4] demonstrates that overexpression of dFOXO or mammalian FOXO proteins in developing Drosophila tissues results in a significant reduction in growth. Importantly, more severe phenotypes are obtained by expression of FOXO proteins lacking their Akt phosphorylation sites, or by coexpression of wild type dFOXO with an inhibitory version of PI 3kinase. The degree of growth suppression by dFOXO also increases in response to nutrient deprivation [2], which has been shown to reduce the levels of insulin-like protein expression. Together these results provide in vivo support for the idea that FOXO proteins are negative regulators of growth in response to conditions of low insulin signaling.
Although these experiments were conducted in vivo, the results suffer the usual caveats of studies based on overexpression. Indeed, it was found that the growth inhibition caused by dFOXO expression is due in part to induction of necrotic cell death [2], a phenotype not observed upon complete loss of insulin/PI 3-kinase signaling. This suggests that the overexpression phenotypes may not reflect normal FOXO function. To directly test the physiological requirement for dFOXO in regulating growth, Jünger et al. [2] generated loss-of-function mutations in the dFOXO gene. The predicted phenotype of disrupting a negative growth regulator is unrestrained growth, as observed in PTEN and TSC mutants. Surprisingly, this was not the case in the dFOXO mutants: flies lacking dFOXO were found to grow to a normal size [2]. Thus, despite its ability to potently inhibit growth when overexpressed, dFOXO is apparently not required for growth suppression under normal developmental conditions. In contrast, a genetic requirement for dFOXO was observed when insulin-signaling levels were experimentally lowered. Loss of FOXO significantly suppressed the reduced growth phenotype of mutations in the insulin receptor, chico, PI 3-kinase and Akt genes [2]. Thus, under normal conditions, insulin/PI 3-kinase signaling appears to be sufficient to maintain dFOXO in a phosphorylated state, rendering it inactive, cytoplasmic, and therefore largely irrelevant. When insulin signaling is reduced, however, dFOXO is required to provide full growth inhibition.
Like most models however, the current one has difficulty incorporating a few experimental observations. Although most parts of the fly grew normally in the dFOXO mutant, the wings were found to be reduced in size, an unexpected result for a growth-suppressor mutation. In addition, dFOXO mutants suppressed the overgrowth phenotype caused by mutations in PTEN, a negative regulator of insulin signaling. These results suggest that in some situations dFOXO may play a positive role in regulating growth. Recent studies have found that transient downregulation of Akt signaling and activation of FOXO3a is required for mitotic progression in NIH 3T3 cells [17]. This finding may partly explain why dFOXO mutants do not have an overgrowth phenotypethey fail to go through sufficient mitoses -and may also account for previous observations that constitutive expression of PI 3-kinase in the Drosophila wing can increase the rate of cell growth but not cell division [18].
Insulin signaling regulates growth by controlling both cell size and cell number, and mutations in different components of this pathway in Drosophila have been shown to cause distinct effects on these parameters. For example, the small flies resulting from mutations in the chico/IRS1 gene are comprised of both smaller and fewer cells [19], whereas loss of dS6K function causes a reduction in cell size without affecting cell number [20]. Where does dFOXO fit into this scheme? In general, most of the results in the recent studies [2][3][4] suggest that dFOXO exerts its effects largely through changes in cell number: dFOXO mutants were found to suppress the reduction in cell number but not cell size caused by chico mutations [2]. Furthermore, Puig et al. [4] found that the small eyes and wings resulting from dFOXO overexpression were comprised of fewer cells of normal size [4]. Thus, changes in cell size and cell number are genetically separable outcomes of insulin signaling, and dFOXO represents the first identified insulin signaling component that regulates primarily cell number.
These distinctions become somewhat blurred, however, when one considers the actual cellular processes that control the final number and size of cells in an organism, namely cell growth, cell division, and cell death. In the case of dFOXO overexpression, for example, the reduction in cell number but not cell size implies that rates of cell growth and division are decreased in a balanced fashion, thus maintaining normal cell size (Figure 2). In chico mutants, on the other hand, this balance must be slightly disrupted, with the rate of cell growth being reduced to a greater extent than that of cell division, resulting in both fewer and smaller cells. Thus, seemingly qualitative differences amongst insulin-signaling components in their effects on final cell size and number may reflect rather modest or even trivial differences during development, such as the developmental stage at which a gene product becomes limiting. Indeed, in contrast to the conclusions of Puig et al. [4], Kramer et al. [3] found that overexpression of dFOXO caused reductions in both cell size and number; this discrepancy is likely to be due in part to differences in timing of overexpression, with Kramer et al. expressing dFOXO later in development, in primarily post-mitotic cells, thereby preventing a balanced reduction of growth and division. Thus, classifications of insulin signaling components on the basis of their effect on cell number and cell size probably represent somewhat artificial distinctions that do not reflect critical differences in their cellular functions.
What are the transcriptional targets that contribute to growth regulation by insulin signaling? The results of genome-wide expression analyses suggest that the number of FOXO-regulated genes is likely to be rather large. Puig et al. [4] identified 277 genes that were upregulated in cultured Drosophila cells expressing constitutively active dFOXO. Jünger et al. [2] took a complementary approach, identifying genes whose expression decreased in response to insulin. In addition, the expression profiles of Drosophila larvae subjected to nutrient deprivation in vivo have been assayed [21]. One target gene identified in each of these studies is d4E-BP, a negative regulator of translation that acts by binding and inhibiting the translation-initiation factor eIF4E. The 4E-BPs are well-established targets of phosphorylation by the TOR-dependent pathway, which disrupts the association between 4E-BP and eIF4E; the current results therefore indicate that both the expression and activity of d4E-BP are negatively regulated by insulin signaling (Figure 1). Interestingly, loss-of-function mutations in d4E-BP appear to have no effect on growth in an otherwise wild-type background, but they were found to suppress the reduction in growth caused by reduced insulin signaling, in a manner remarkably similar to that of dFOXO mutants [2]. In addition, Puig et al. [4] also identified the insulin receptor gene as being transcriptionally activated by dFOXO, suggesting a negative feedback loop that may serve to buffer the effects of alterations in insulin levels.
Together, these new studies in Drosophila significantly broaden our understanding of the multiple layers of insulin-mediated growth regulation. Control of gene expression by FOXO factors in response to insulin allows integration of transcriptional activities with other growthrelated processes regulated by insulin, such as protein synthesis, carbohydrate metabolism and survival. A challenge for the future is to explore how these processes interact, and to determine what role transcription plays in their regulation. For example, by coordinating the expression of genes that induce growth arrest with genes required to survive quiescence, FOXO factors may provide a comprehensive response to conditions of low insulin or nutrient levels [22]. In addition, it will be important to understand how differences in cell type and developmental context can influence the transcriptional and physiological response to FOXO activity, regulating cell growth and proliferation in some cases and differentiation in others. Identification of the physiologically relevant target genes in these processes should provide further insights into the important process of insulin signaling.