Starvation for dietary AA leads to profound changes in metabolism of Drosophila larvae. These changes include decreases in ribosome and protein synthesis [18, 24, 25], alterations in the storage and metabolism of fat and carbohydrate [6, 25, 26], and induction of autophagy [27]. These alterations are essential for arresting larval growth and development, and maintaining homeostasis under poor nutrient conditions[24, 25, 28]. Our microarray data suggest that many of these changes are mediated through nutrient-dependent modulation of gene expression. For example, mRNA levels of both protein synthesis and ribosome biogenesis genes are reduced upon AA starvation. Similarly genes involved in mitochondrial biogenesis and function, such as mitochondrial ribosomal proteins and a mitochondrial transcription factor, TFAM, are also suppressed by AA starvation. In contrast, several genes involved in proteolysis and autophagy are rapidly induced by removal of dietary AA.
A previous report also described both protein and complete nutrient starvation responses in whole larvae [14]. Although this study used different protocols to analyze gene expression, similar, but fewer, genes were identified compared to our study. More recently, Telemann et al (2008) [16] examined transcriptional responses to complete starvation in either fat or muscle. They identified many genes whose regulation was specific to either tissue, but they also reported changes in similar classes of genes especially those involved in ribosome synthesis, mRNA translation and mitochondrial biogenesis. Our analyses were done using whole larvae therefore we are probably identifying strong transcriptional changes that are occurring in most larval cells. Together, these previous reports and our data identify the major changes in metabolic gene expression mediated by dietary AA in Drosophila larvae.
The bulk of the larval organs (such as the gut, fat body and muscle) are composed of polyploid cells which all undergo nutrition-dependent endoreplication [5, 29]. AA starvation leads to a shutdown of cell cycle progression in these tissues [5, 29]. However, our microarray analyses on starved larvae did not reveal significant changes in mRNA levels of cell cycle regulators. This finding suggests that nutrition-dependent control of cell cycle progression may occur through post-transcriptional changes in cell cycle genes. This type of regulation has been described in yeast in which nutrient-dependent alterations in ribosome synthesis and mRNA translation can control translation of cell cycle genes [30].
The insulin/PI3K and TOR signaling pathways are the major mediators of nutrition-dependent cell and organismal growth in Drosophila larvae. Our microarray analyses allowed us to compare the expression profiles of AA starvation with modulation of PI3K and TOR signaling. These comparisons suggest that only about half of genes whose expression is altered by AA starvation also respond to modulation of PI3K and/or TOR signaling. The strongest overlap was between TOR and AA starvation-regulated genes. Virtually all of these overlapping genes were involved in the control of some aspect of cellular metabolism, particularly the control of carbohydrate and lipid metabolism. However, we did see some notable differences between starvation and TOR mutants. In particular, TOR mutants showed little change in the expression of ribosome biogenesis and protein synthesis genes. TOR mutants have reduced nucleolar size and lower levels of rRNA synthesis [18, 20], two effects that are indicative of reduced ribosome synthesis and that are phencopied by protein starvation [18]. Also, inhibition of TOR in cultured Drosophila cells does inhibit expression of ribosome biogenesis genes [31]. It is possible these differences may reflect tissue specific effects (i.e. cell culture versus larvae). However, it is likely that many of the effects of TOR on ribosome synthesis may be mediated through post-transcriptional control of downstream effectors e.g. TIF-IA [18] and 4E-BP. Indeed many of the nutrition-dependent functions of TOR probably reflect transcription-independent effects of TOR on processes such as mRNA translation [32], autophagy [28] and endocytosis [33].
We saw a much weaker overlap between genes affected by AA starvation versus PI3K overexpression. For example, given the role of insulin/PI3K signaling in nutrition-dependent growth, we predicted that genes suppressed by starvation might be induced by PI3K overexpression. However, this prediction was correct for only a small subset (12%) of AA-starvation suppressed genes. In particular we found that genes involved in ribosome biogenesis and protein synthesis, two processes strongly influenced by protein starvation, were generally unaffected by PI3K. Thus, not all AA starvation-regulated genes respond to activation of the insulin/PI3K pathway. The transcription factor FOXO is a conserved downstream target of the PI3K pathway. High PI3K activity leads to nuclear export and inhibition of FOXO [34]. It has been suggested that many transcriptional responses to PI3K activation are mediated through repression of FOXO activity. Indeed, a recent paper implicated FOXO in mediating many transcriptional effects of AA starvation [16]. This suggestion would seem at odds with our comparison of AA starvation and PI3K-responsive genes. These differences between our findings with PI3K overexpression and previous examination of FOXO mutants may be due to different experimental approaches. For example, we examined gene expression following overexpression of PI3K in both fed and starved larvae, whereas Telemann et al. examined starvation expression profiles in either fat or muscle of FOXO mutants. In addition, it is likely that many of the effects of PI3K overexpression on both transcription and growth can occur independently of FOXO. For example, FOXO mutants have no growth phenotypes in nutrient rich food [35], while overexpression of PI3K can induce marked cell growth under identical conditions [6], suggesting that the effects of PI3K involve more than FOXO inhibition alone.
One key finding from our work was the role of Myc in nutrition-mediated gene expression. We found that the gene expression profiles of both Myc mutants and Myc-overexpression exhibit a strong overlap with AA starvation-responsive genes. Many of the overlapping genes were involved in ribosome synthesis, mRNA translation and mitochondrial function. Moreover, we showed that Myc overexpression could bypass a requirement for AA and reverse repression of many of genes induced by starvation. These findings suggest that Myc acts as a downstream mediator of AA-dependent regulation of gene transcription and expression of metabolic genes in Drosophila larvae. Most of these Myc- and AA starvation-sensitive genes contain either the consensus Myc binding site or alternate E-box motifs that may also bind Myc, within their promoters. A recent paper reported a similar enrichment of Myc binding sites within the promoters of genes inhibited by rapamycin treatment in cultured Drosophila S2 cells, particularly ribosome synthesis genes [16]. This latter study reported that Myc association with the promoters of these genes was inhibited by treatment with the TOR inhibitor, rapamycin [16]. Our in vivo ‘by-pass’ experiments extend these cell culture findings to show that Myc functions downstream of nutrition to control metabolic gene expression in developing larvae. Two recent reports indicated that Myc mRNA levels were FOXO-regulated in a tissue specific manner [16, 36], hence providing one potential mechanism by which Myc might be regulated downstream of AA availability. Regulation of Myc protein levels by either translation control or ubiquitin mediated-degradation may also be a downstream effect of AA.
An important finding from our work was the marked differences in the expression profiles of PI3K-overexpression versus Myc-overexpression. For example, we found that Myc induced expression of a large number ribosome biogenesis and protein synthesis genes (Fig 4) whereas PI3K-overexpression did not. Consistent with this, we previously reported that Myc overexpression had strong effects on rRNA synthesis, whereas PI3K overexpression had little or no effect [21]. Thus, while both Myc and PI3K can induce quantitatively similar increases in growth, comparisons between their gene expression profiles suggest qualitatively different effects of Myc versus PI3K on transcription and metabolism. Indeed, Myc and PI3K produce morphologically different effects on cell growth in the larval fat body [37]. Importantly, these differences between PI3K and Myc suggest that the effects of insulin/PI3K on transcription are probably in large part independent of Myc function. In contrast, a recent paper reported that in cultured Drosophila S2 cells, insulin treatment (and presumably increased PI3K activity) promoted Myc binding to the promoters of ribosome biogenesis genes and increased expression of these genes [16]. These results contrast with our data and may reflect differences between responses to increased PI3K activity in in vitro cultured S2 cells versus developing larvae. Nevertheless, our data do point to differences in the way PI3K and Myc regulate gene expression and cell growth in vivo. These qualitative differences between how PI3K and Myc function may have implications for situations in which growth is deregulated, such as cancer. Thus, increases in PI3K pathway activity and Myc may each induce distinct effects on metabolism, and consequently cooperate to drive aberrant cell growth and induce transformation [38]. In fact, increased Myc levels and overactivation of effectors of the PI3K pathway are observed in many cancers and may function synergistically to promote tumor progression [39].
Nutrient dependent changes in gene transcription have been well studied in budding yeast [1, 2]. In this unicellular organism alterations in extracellular levels of nutrients and amino acids can elicit widespread changes in transcription of large classes of metabolic genes, particularly those involved in ribosome biogenesis and protein synthesis [2, 4]. In many cases the cis-regulatory promoter elements and corresponding DNA-binding factors that mediate these responses have been defined [2]. However, these factors often have no obvious homologs in metazoans. In Drosophila, our work and that of others suggests Myc and TORC2-CREB function as nutrition/PI3K/TOR-regulated transcription factors [16, 19]. However, no additional factors have been described. In this paper we surveyed the promoters of genes whose expression is AA starvation-sensitive for enriched DNA motifs. One motif in particular (Motif1) emerged from this analysis. Motif 1 was previously identified as a core promoter element in a large number of genes in Drosophila, although the exact binding factor(s) are not known [23]. Our data suggest that Motif1 is required for transcription of AA starvation-responsive genes. Moreover, Motif-dependent transcription is inhibited by AA starvation. These findings raise the possibility that motif 1 is a binding site for AA starvation-regulated transcription factor(s). It is likely that this putative factor(s) is an activator rather than a repressor since mutation of motif1 in the transgenic reporters led to reduced transcription as opposed to de-repression (and increased transcription). Given the location of motif 1 within the core promoters of many Drosophila genes and its proximity to the transcription start site [23], the factor(s) that bind Motif 1 may also either be, or associate with, core promoter factors. This view is consistent with emerging data suggesting that regulation of core promoter factors can control differential gene expression. For example, in mammalian cells, induction of myogenesis involves differential expression and recruitment of core transcription factors [40]. Also, in yeast the expression of stress- versus growth-regulatory genes in response to environmental cues depends on differential recruitment of core promoter factors [1]. We found Motif1 in the promoters of many classes of AA starvation-responsive metabolic genes, particularly those involved in protein synthesis and ribosome biogenesis. Moreover, a previous genome-wide binding analysis of dMyc showed that the presence of Motif1 significantly correlates with the presence of consensus Myc-binding sites [41]. These findings together with our data suggest that a putative motif-1-binding factor and Myc may function to control metabolic gene transcription in response to nutrient availability in Drosophila. Whether transcription through both promoter elements is cooperative or additive is unclear. However, since many AA starvation-responsive genes (e.g. cluster A and B genes, ribosome biogenesis genes) contain only one or other of the two motifs, motif1 and the E-box can probably function independently of each other.
