The TOR/S6K (S6 kinase) branch of the growth modulatory network is negatively regulated by the tumor suppressor Tsc1 (hamartin)/Tsc2 (tuberin) complex. Tuberous sclerosis complex (TSC) is an autosomal-dominant disorder characterized by the formation of hamartomas, benign tumors that arise in various tissues (Pan et al. 2004). In Drosophila, cells devoid of Tsc1/Tsc2 function are increased in size (Gao and Pan 2001; Potter et al. 2001; Tapon et al. 2001). Tsc2 and, more weakly, Tsc1 were found to physically associate with dTOR, thereby inhibiting dTOR kinase activity (Gao et al. 2002). Other studies reported an inhibitory role of PKB on Tsc1-Tsc2 by PKB-mediated phosphorylation of Tsc2 (Inoki et al. 2002; Manning et al. 2002). Recently, the small GTPase Rheb (Ras homolog enriched in brain) has been identified as a new positive growth effector acting downstream of Tsc1-Tsc2 and upstream of TOR. Mechanistically, Tsc2 acts as GTPase-activating protein (GAP) toward Rheb (Garami et al. 2003; Inoki et al. 2003a; Saucedo et al. 2003; Stocker et al. 2003; Tee et al. 2003; Zhang et al. 2003). The molecular mechanism how Rheb relays the signal to TOR is currently unknown. dTOR mutants show a growth deficit that is more pronounced in endoreplicative tissues than in mitotic tissues (Oldham et al. 2000; Zhang et al. 2000). An effector of mTOR is S6K, which upon activation by mTOR phosphorylates ribosomal protein S6. S6K-mediated S6 phosphorylation has been thought to lead to a preferential translation of mRNAs encoding ribosomal proteins and proteins of the translational apparatus although the significance of this S6K function has been questioned recently (Tang et al. 2001; Stolovich et al. 2002; Pende et al. 2004). Inr/TOR signaling activity culminates in the regulation of translation rate by controlling S6K and the translational repressor 4E-BP1 (Gingras et al. 2004). S6K mutant flies are small but in contrast to mutants of the Inr pathway, only cell size is reduced without a change in cell number (Montagne et al. 1999). Therefore, loss of S6K function reduces growth and body size to a lesser extent than loss of other positive components acting further upstream in the cascade.

Growth is modulated by extrinsic factors such as nutrients, temperature, and hypoxia (Palos and Blasko 1979; Frazier et al. 2001; Azevedo et al. 2002). However, their link to the Inr/TOR signaling network are not well defined. Although it is known that starvation results in a reduction in the levels of insulin-like peptides and a reduction in S6K activity in Drosophila (Oldham et al. 2000; Ikeya et al. 2002), little is known about whether temperature or hypoxia regulates the activity of this pathway. It is conceivable that mutations in genes coding for factors mediating the modulation of growth in response to external stimuli have escaped detection because they do not exhibit a very strong phenotype under standard culture conditions. For example, overexpression of the Forkhead transcription factor FOXO3a, the human homolog of Drosophila dFOXO, produces a very subtle small eye phenotype under normal nutrient conditions. Under starvation, however, this phenotype is massively exacerbated, inducing a further eye size reduction and cell death. Furthermore, dFOXO mutant flies are viable and do not show an (over)growth phenotype in an otherwise wild-type background under normal conditions (Junger et al. 2003).

Genes like dFOXO were missed in genetic loss-of-function screens aimed at identifying growth-regulatory genes (1) because they have only mild or missing mutant phenotypes under normal conditions and/or (2) because their function is masked by redundancy.

Genes acting in a redundant manner can be identified in a complementary gain-of-function approach using EP (enhancer/promoter) elements (Rorth 1996). Screening >4000 novel EP lines, we found two insertions in the scylla locus as suppressors of a PKB/PDK1-dependent eye overgrowth phenotype. We identified a second gene in the Drosophila genome with homology to scylla named charybdis. Homologs of these genes also exist in mammals, and they have been implicated in the response of a cell to hypoxia or more generally as stress-induced genes having either pro- or anti-apoptotic functions. We present evidence that scylla and charybdis, like some of their mammalian homologs, are induced by hypoxia and that Scylla and Charybdis act as partially redundant negative regulators of growth by controlling S6K but not PKB activity.

Figure 1. Overexpression of scylla and its paralog charybdis suppresses the hyperactivity of the Inr pathway. (A-E) scylla (C) and/or charybdis (D) overexpression (achieved by EPscy and EPchar, respectively) suppress the PKB/PDK1-dependent big eye phenotype (B (more ...)

We identified two EP insertions (EP22.1, hereafter named EPscy, and EP9.85) in the scylla locus as suppressors of the PKB/PDK1 bulging eye phenotype (Fig. 1C). BLASTP search with the Scylla amino acid sequence revealed another homologous protein encoded in the Drosophila genome termed Charybdis. scylla (scy, CG7590) and charybdis (char, CG7533) are separated by ~232 kb of genomic DNA. Their gene products share a high degree of homology (38% identity, 49% similarity), suggestive of a gene duplication event (Fig. 1F,G). We checked whether charybdis overexpression would behave similarly to scylla in the PKB/PDK1 overexpression assay using EP1035 (hereafter named EPchar). Indeed, the big eye phenotype of the tester system is also suppressed by EPchar (Fig. 1D). UAS transgenes with either the scylla or charybdis cDNA recapitulate the suppression phenotype of the corresponding EP element (data not shown). Coexpression of scylla and charybdis further ameliorates the suppression phenotype to a nearly wild-type situation (Fig. 1E). Notably, scylla or charybdis overexpression on their own using a panel of different eye/wing Gal4 drivers reduced adult organ size. Coexpression of the caspase inhibitors p35 or DIAP1 did not rescue the small eye phenotype induced by expression of either scylla or charybdis in the eye. Moreover, no elevated cell death in eye imaginal disks overexpressing scylla/ charybdis under control of the GMR-Gal4 driver was observed by acridine orange staining (data not shown). This suggests that apoptosis is not the cause for the eye size reduction. Thus, scylla and charybdis overexpression antagonizes the growth-promoting effects of PKB/PDK1 and is sufficient to negatively regulate growth.

scylla and charybdis are both expressed during embryogenesis in dynamic, partially overlapping patterns. In contrast to the broadly expressed scylla mRNA, charybdis transcripts are predominantly restricted to neurons of the CNS and PNS as assessed by mRNA in situ hybridization. During late larval stages scylla mRNA is uniformly expressed without apparent tissue-specific distribution, whereas charybdis mRNA expression could not be detected in third instar imaginal disks (data not shown).

Homologs of Scylla/Charybdis exist throughout the animal kingdom including humans, rat, mouse, Xenopus, and zebrafish. Reminiscent of the situation in Drosophila, mammals possess several paralogous proteins with homology to Scylla/Charybdis. At least two homologs (RTP801/REDD1 and FLJ3691, a hypothetical protein presumably corresponding to REDD2 [Ellisen et al. 2002]) exist in humans, two (RTP801 and SMHS1) in rat, and three (RTP801L, Dig2, and SMHS1) in mice. The overall homology is highest toward the C terminus (Fig. 1G). Charybdis contains a predicted coiled-coil structure (amino acids 168-188).

In mammals, RTP801/REDD1 and other members of the family are induced upon various forms of cellular stress including hypoxia, DNA damage, and dexamethasone treatment (Ellisen et al. 2002; Shoshani et al. 2002; Wang et al. 2003). In fact, RTP801/REDD1 is a direct target of HIF-1, a heterodimeric transcription factor that plays a pivotal role for survival in response to hypoxia (Shoshani et al. 2002). Furthermore, RTP801/REDD1 is controlled by p53 and p63 (Ellisen et al. 2002). Depending on the experimental setup and cell context, RTP801 overexpression either inhibits or increases apoptosis, suggesting a complex regulation and/or dependence on the developmental program of the cell (Shoshani et al. 2002).

Partial scylla deletions were obtained by imprecise excisions of EP9.85, which is integrated in the scylla open reading frame (ORF) and therefore already represents a scylla allele (hereafter named scy^EP9.85) (Fig. 1F). For charybdis, we used a local hop strategy of EPchar to obtain char¹⁸⁰, constituting a new EP insertion (EP1035^*) in the charybdis 5′-untranslated region (UTR) plus the original EPchar (see Materials and Methods). Quantitative real time-PCR showed that in char¹⁸⁰ homozygotes, charybdis mRNA expression is decreased to ~25% of wild-type levels (data not shown). We assume that this charybdis allele is a strong hypomorph, as mRNA levels were only slightly more reduced (to 23% or 15%, respectively) when RNA was extracted from flies heterozygous for char¹⁸⁰ over either one of two independent deficiencies uncovering charybdis and scylla (data not shown).

All scylla mutant combinations and the char¹⁸⁰ homozygotes are viable and fertile without apparent mutant phenotype. scylla and char¹⁸⁰ mutant animals have the same weight as control flies. Measurement of wing size and hair density in the adult wing of scylla mutants revealed no differences in cell size and cell number as compared to control animals (data not shown). We created scylla loss-of-function clones in imaginal disks using FLP/FRT-mediated mitotic recombination to test the effect on growth properties of the mutant tissue. One would expect a growth advantage of cells in clones lacking a bona fide negative growth regulator, as is the case for PTEN. However, larval scylla mutant clones were the same size as their wild-type sister clones. Likewise, clones obtained in adult eyes revealed no increase in cell size of scylla or char¹⁸⁰ mutant ommatidia (data not shown). Thus, loss of Scylla or Charybdis function is dispensable for growth under normal conditions. It is conceivable that Scylla and Charybdis act in a redundant manner. Therefore, it was important to create scylla charybdis double mutants.

We combined char¹⁸⁰ with three of our scylla alleles by meiotic recombination. All double-mutant scylla charybdis combinations produced viable adult flies. Weight analysis of heteroallelic scylla charybdis flies demonstrated that simultaneous loss of Scylla and Charybdis significantly increases body weight. Consistent results were obtained by combining one copy of scy^{EP9.85/31/113} char¹⁸⁰ and the deficiency Df(3L)vin4 uncovering scylla and charybdis. Mutant females were on average 6%-23% and males 9%-17% heavier than control flies (Fig. 2A; data not shown). Conversely, ubiquitous scylla/charybdis overexpression using the Act5CGal4 driver generated flies that are decreased in size and weight (15.1% using EPscy and 14.3% using UAS-char#53) (Fig. 2B, data only shown for males). The majority of scylla/charybdis-overexpressing flies eclosed with a minor delay (0.5-1 d) (data not shown).

Figure 2. Scylla and Charybdis inhibit growth. (A) Simultaneous loss of Scylla and Charybdis leads to an increase in body weight. Control flies (black bars) are represented by either of the two possible double mutants over the TM3 balancer chromosome (first three (more ...)

To assess whether charybdis/scylla overexpression affects cell size, we generated scylla and charybdis gain-of-function flip-out clones in the eye, marked by the absence of the red pigment. A moderate reduction in cell size was observed in cells overexpressing scylla or charybdis (Fig. 2C,D). Ommatidia overexpressing scylla or charybdis exhibit no patterning defects. On the other hand, simultaneous removal of Scylla and Charybdis in clones of photoreceptor cells resulted in slightly enlarged cells (Fig. 2E,F). Consistently, when most of the head capsule and the eyes were made homozygous by means of the eyflp/FRT system in an otherwise heterozygous mutant background, a mild big head phenotype was generated in the double mutant but not in either single mutant (Supplementary Fig. S1).

The gain-of-function clonal analysis in the eye showed that cell size is reduced upon forced scylla/charybdis expression, but it did not address the question whether cell number is affected. To clarify this issue, the wing size and cell number of flies that ubiquitously overexpress scylla were measured. The insect wing is a double-layered epithelial structure, and each cell in the wing secretes a single hair (trichome). Therefore, by counting the number of trichomes per defined area, hair density can be taken as a measure for cell number. Overall wing size was reduced by ~25% in males or ~15% in females overexpressing scylla (Fig. 2G). By extrapolating the number of cells per measured area, we found that cell size is decreased by ~29% in males and ~21% in females (Fig. 2H). We conclude that the size reduction brought about by overexpressing scylla is caused by a reduction in cell size. Cell density is even slightly increased (5.8% in males, 7.6% in females) (data not shown). Our data show that Scylla and Charybdis have a growth-inhibitory role and that they share some functional redundancy.

RTP801/REDD1 was shown to be induced by hypoxia (Shoshani et al. 2002). This prompted us to investigate the effects of hypoxia on the scylla charybdis double mutants. scylla charybdis mutant larvae were raised on normal food at room temperature in a hypoxia chamber containing 9% oxygen during their entire development. In general, control flies as well as scylla charybdis homozygotes were 3-4 d delayed in development under these hypoxic conditions. However, whereas adult homozygous scylla charybdis double mutants could readily be recovered under standard culture conditions in normoxia, homozygous mutants of three independent scylla charybdis allelic combinations (char¹⁸⁰ in combination with scy³¹, scy¹¹³, or scy^EP9.85) were strongly underrepresented compared to normoxic conditions. We also observed the appearance of an increased number of dead pupae in the vials (data not shown). The scylla¹¹³ char¹⁸⁰ double-mutant combination had the strongest effect, and only two escapers (out of 326 scored flies) hatched, whereas under normoxia flies with this genotype were recovered with nearly the expected Mendelian ratio. The eclosed homozygotes raised under hypoxia did not show obvious morphological defects.

Thus, whereas simultaneous loss of Scylla and Charybdis under normoxic conditions results in a slight increase in growth, their absence under reduced oxygen concentrations severely compromises larval development. This indicates that Scylla and Charybdis have a critical function for survival under hypoxic conditions.

Figure 3. scylla and charybdis are induced by hypoxia, and scylla is a target of Drosophila HIF-1. (A-D) Larvae were incubated in a hypoxia chamber for either 17 h under 2% O2 or for 7 h under 2.8% O2 conditions leading to the induction of scylla (B) and charybdis (more ...)

In Drosophila, the bHLH-PAS family comprises Period (Per), Trachealess (Trh), Single-minded (Sim), Spineless (Ss), Dysfusion (Dys), Tango (Tgo), and Similar (Sima). Tgo is the ubiquitously expressed HIF-1β ortholog, which dimerizes with any of the α-subunits. Sima has been shown to fulfill analogous functions to its mammalian homolog HIF-1α (Lavista-Llanos et al. 2002).

To test whether scylla/charybdis transcription is regulated by bHLH-PAS proteins that recognize the same DNA stretch, we overexpressed sim, trh, or sima together with tgo using the Lsp2Gal4-driver that is active specifically in the fat body during the third larval stage. For Sima, we used a form lacking the oxygen-dependent degradation domain (ODD), rendering it refractory to proteolytic destruction under normoxic conditions. Only the coexpression of tgo and sima induced scylla but not charybdis expression as assessed by mRNA in situ hybridization (Fig. 3G,J; data not shown). This does not preclude, however, the possibility that charybdis is a target of Tgo-Sima as we detected its endogenous induction upon hypoxia only in the gut but not in the fat body. Since neither expression of trh with tgo nor sim induced scylla or charybdis expression, the regulation of scylla by the Tgo-Sima heterodimer is specific. Thus, scylla and charybdis, like their mammalian homolog RTP801/REDD1, are induced by hypoxia, and at least scylla appears to be a direct target gene of the HIF-1 homolog Tgo-Sima.

Figure 4. Scylla attenuates growth in response to elevated Inr pathway activity downstream of PKB but upstream of TSC. (A-C) A PKB/PDK1-overexpression eye phenotype is enhanced in a scylla mutant background. Genotypes are as follows: y w (A,A′); y w; GMR-Gal4 (more ...)

Moreover, overexpression of scylla and charybdis not only suppressed the growth phenotype caused by over-activation of the Inr pathway in the eye but to a certain extent also rescued the lethality associated with the ubiquitous increase in Inr pathway activity due to either overexpression of PKB or loss of PTEN. scylla rescued the male-specific lethality caused by ubiquitous expression of PKB and organismal lethality associated with the partial but not complete loss of PTEN function. Similarly, PKB-associated male lethality was also rescued by charybdis overexpression (Supplementary Tables 1, 2). This indicates that scylla and charybdis have the capacity to act as potent negative regulators of insulin signaling downstream of PKB and PDK1.

Several lines of evidence suggest that Scylla and Charybdis act upstream of TSC and Rheb. Tsc1/2 mutant flies can be rescued to adulthood by reducing S6K signaling (Radimerski et al. 2002a), and a mere reduction of one TOR copy in a Tsc1 mutant context results in a rescue to the pupal stage (Gao et al. 2002). We examined whether ubiquitous scylla overexpression could rescue the larval lethality of heteroallelic Tsc1/2 mutant combinations (Tsc1^2G3/Tsc1^Q87X and Tsc2⁵⁶/Tsc2¹⁹²) using the daGal4 or Act5CGal4 drivers in combination with a UAS-scy transgene or EPscy at 18°C, 25°C, and 29°C. Ubiquitous overexpression of scylla/charybdis in a Tsc1/2 mutant background did in no case extend larval development beyond first/second instar, and these larvae died at the same time as Tsc1/2 mutants. Moreover, the big head phenotype of Tsc2¹⁹² (and Tsc2⁵⁶) induced by the eyflp/FRT system was not further enhanced in scy^EP9.85 char¹⁸⁰ Tsc2 triple-mutant heads (Fig. 4F,G). We have shown before (Supplementary Fig. S1) that heads composed almost entirely of scylla charybdis double-mutant cells are enlarged. Conversely, GMRGal4-driven co-overexpression of Tsc1, Tsc2, and scylla or charybdis in the eye did not further reduce the small eye phenotype induced by coexpression of Tsc1 and Tsc2 on their own (data not shown). The absence of an additive growth effect upon loss of Tsc2, scylla, and charybdis or overexpression of Tsc1/2 and scylla or charybdis suggests that they function in the same pathway. These results are consistent with the idea that Scylla and Charybdis act upstream of the TSC complex (see also next section). Our conclusion is further supported by the fact that neither a Rheb-dependent bulging eye phenotype nor organismal lethality could be suppressed by scylla/charybdis coexpression (Fig. 4H; data not shown).

We tested PKB activity of adult female heads overexpressing scylla or charybdis in conjunction with PKB and PDK1 under control of the GMR-Gal4 enhancer. The same experimental setup has previously been used to demonstrate that PDK1 increases PKB activity (Cho et al. 2001). Total fly head protein was extracted and PKB activity was assayed by incorporation of ³²P-labeled phosphate into a synthetic PKB substrate (Crosstide, CT). Although scylla/charybdis overexpression substantially suppressed the PKB/PDK1-induced bulging eye phenotype (Fig. 1C), PKB activity was not reduced in these eyes (Fig. 5A). Moreover, PKB activity was also unaffected in a scylla^-/- background (Fig. 5B).

Figure 5. Scylla reduces S6K activity, but neither Scylla nor Charybdis down-regulates PKB activity. (A) Head extracts of flies overexpressing PKB and PDK1 in conjunction with either scylla or charybdis in the eye were assayed for PKB activity. No reduction could (more ...)

These results are consistent with the placement of Scylla downstream of PKB. To test the effect of Scylla on S6K activity, second instar larvae expressing scylla under the control of Act5CGal4 were collected, and larval extracts were assayed for S6K activity. On average, S6K activity was down-regulated by >50% (Fig. 5C). Although there may also be a slight reduction in total S6K protein levels, this effect cannot account for the much stronger reduction in S6K activity. Taken together with the genetic evidence presented above, these results strongly support the argument that Scylla acts between PKB and TSC to regulate S6K activity. Furthermore, in an accompanying paper, Brugarolas et al. (2004) provide direct biochemical evidence that a functional TSC complex is required for RTP801/REDD1 to affect S6 phosphorylation. Altogether, these data indicate that Scylla functions upstream of TSC.

Figure 6. Scylla and Charybdis prolong life span under starvation conditions. (A) Homozygous char180 mutants are more susceptible to starvation than control animals. The average and standard deviations of three independent experiments are shown (n = 114 for char (more ...)

Here we provide evidence that two related proteins, Scylla and Charybdis, are negative modulators of Inr/TOR signaling in response to different external stress situations including hypoxia and starvation. scylla and charybdis single mutants do not show obvious growth phenotypes. scylla charybdis double-mutant flies are also viable and fertile and exhibit a slight increase in body size. Viability of the double mutants is strongly reduced, however, when they are reared under hypoxic conditions. Thus, although Scylla and Charybdis are largely dispensable for normal development, they have a critical role for the endurance of prolonged hypoxia. We show that scylla is transcriptionally induced as a target of Drosophila HIF-1 (the Tango-Similar dimer) and that scylla and charybdis are up-regulated under hypoxic conditions like their mammalian homolog RTP801/REDD1. Furthermore, we show that Scylla negatively regulates Inr/TOR signaling by reducing S6K but not PKB activity.

Dig2, a murine Scylla/Charybdis homolog, is also a stress-responsive protein induced by a variety of treatments including dexamethasone, thapsigargin, tunicamycin, and heat shock (Wang et al. 2003). Together with the finding by Zinke et al. (2002) that scylla/charybdis expression is increased during starvation conditions and our analysis, showing that Scylla and Charybdis act as growth inhibitors, these data support a model wherein Scylla and Charybdis, induced by stresses like hypoxia and starvation, act to dampen growth under certain stress conditions.

The S6K assay demonstrated that Scylla is capable of reducing S6K activity. Thus, Scylla acts upstream of S6K. This result is consistent with our in vivo results. (1) scylla overexpression in the wing reduces wing size by decreasing cell size but not cell number (in fact, cell number is slightly increased), and (2) a S6K scylla mutant combination has the same weight as S6K single mutants. S6K mutants are smaller because of a reduction in cell size but not cell number, making it distinct from other Inr signaling pathway mutants (Montagne et al. 1999). Scylla and Charybdis do not control S6K activity directly but require its upstream regulator TSC. Tsc1/2 mutants cannot be rescued by overexpression of scylla, and the big head phenotype caused by loss of TSC function is not enhanced by the absence of Scylla and Charybdis. Coexpression of scylla or charybdis and Tsc1/2 does not further decrease eye size compared to Tsc1 and Tsc2 co-overexpression on their own. Moreover, a Rheb-dependent big eye phenotype or lethality induced by ubiquitous Rheb expression cannot be suppressed by scylla expression. These results indicate that Scylla regulates S6K activity by acting upstream of Tsc1/2 and Rheb. This function appears to be conserved between mammals and flies because RTP801/REDD1 can reduce S6 phosphorylation only in the presence of TSC (Brugarolas et al. 2004).

Our findings indicate that scylla/charybdis overexpression mainly affects the TSC/TOR/S6K branch of the pathway downstream of PKB. The PKB-FOXO axis appears not to be influenced by Scylla and Charybdis. Eye-specific overexpression of scylla/charybdis in conjunction with FOXO was unable to induce 4EBP (Supplementary Fig. S2). In contrast, simultaneous expression of FOXO and PTEN or a dominant-negative form of PI3K led to a strong induction of the reporter gene (Junger et al. 2003).

Consistent with an interplay between the Inr and TOR/S6K pathways, Inr lethality is suppressed by heterozygosity of Tsc1 (Gao and Pan 2001). Furthermore, overexpressed PKB phosphorylates and inactivates Tsc2 and thereby activates S6K (Inoki et al. 2002; Manning et al. 2002). Our finding that scylla overexpression is sufficient to rescue the lethality associated with PKB overexpression indicates that the lethality caused by PKB overexpression is due to the hyperactivation of the TOR/S6K pathway. Thus, oncogenic activation of PI3K/PKB signaling seems to be mainly mediated by TOR/S6K signaling. This may explain the beneficial effect of Rapamycin treatment (or its derivatives CCl-779 and RAD001) on PTEN-deficient tumors or cells overexpressing PKB (Neshat et al. 2001; Podsypanina et al. 2001; Shi et al. 2002; Majumder et al. 2004).

We have shown that directed expression of Tgo-Sima in the fat body induces scylla expression. That this regulation is physiologically relevant can be inferred from three findings. First, scylla is also induced under hypoxic conditions. Second, directed expression of other bHLH-PAS proteins like Tgo-Trh or Sim alone did not induce scylla expression. Third, survival of flies lacking scylla and charybdis function is severely compromised under hypoxic conditions. Figure 7 summarizes our current model of Scylla and Charybdis function. scylla and charybdis are induced in response to external stress stimuli (e.g., hypoxia and starvation) to inhibit growth downstream of PKB but upstream of Tsc1/2. Scylla suppresses growth by reducing S6K activity. This could be achieved by relieving the inhibitory effect of PKB on Tsc2. Alternatively, Scylla/Charybdis could be negatively regulated targets of PKB. This is unlikely, however, since Scylla and Charybdis lack PKB consensus phosphorylation sites. AMPK, activated by drops in energy levels, may also contribute to the induction process of scylla and charybdis for growth inhibition, presumably under prolonged stress exposure. However, it is also possible that AMPK is controlled by Scylla and/or Charybdis. AMPK decreases protein synthesis by inhibition of S6K in a Rapamycin-sensitive manner, suggesting that mTOR is involved in mediating AMPK signaling (Kimura et al. 2003). AMPK also phosphorylates Tsc2, an event important for the cellular energy response pathway (Inoki et al. 2003b).

Figure 7. Model of scylla/charybdis regulation and their effects. scylla and charybdis are stress-induced, negative growth modulators that inhibit growth under adverse environmental conditions. They are likely to feed into Inr/TOR/S6K signaling downstream of PKB (more ...)

The shortened scylla construct UAS-scy^Δ1-12 was PCR-generated by using pJR3 as the DNA source with an ATG translational start site introduced by PCR mutagenesis. EP9.85, inserted 36 bp after the predicted translational start, shows the same effects as EPscy, which lies further upstream of the scylla transcription start site. We consider it likely that the last 3 nt of EP9.85 (i.e., ATG) serve as starter methionine to which amino acid 13 of the original Scylla protein is connected in frame.

To generate the UAS-char transgene, we ordered EST LP03309 (Research Genetics) corresponding to charybdis and checked the sequence for mutations. charybdis cDNA was excised from pOT2 as an EcoRI/XhoI fragment and inserted into the pUAST vector.

Transgenic flies were generated by P element-mediated germline transformation. Constructs were injected into yw embryos. At least two independent lines could be established for each construct.

char¹⁸⁰ was identified in a local hop screen, mobilizing the original w⁺-marked EPchar insertion (causing a faint orange eye color). After the mobilization, we scored animals having a darker w⁺ eye color than EPchar and allowed recombination to occur by crossing single females to y w males. Only lines in which all the offspring consistently displayed the darker eye color (i.e., where presumably no recombination has occurred) were further analyzed by PCR using primers that would amplify the charybdis genomic locus (but not when a new EP had inserted). Out of 331 single females showing a darker eye color, 112 lines were established and analyzed by single-fly PCR. Only char¹⁸⁰ was identified (contains the original EPchar element plus a new inverted EP insertion [indicated as EP1035^* in Fig. 1F] ~450 bp upstream of the char ORF). Quantitative RT-PCR of homozygous char¹⁸⁰ flies revealed a 75% reduction in mRNA levels (Q RT-PCR data available on request).

All crosses were performed at 25°C if not stated otherwise.

scylla-overexpressing wings were mounted in Euparal Mounting solution (TAAB Laboratories). Wing area was determined using Photoshop and NIH Image 1.61 to count pixels. Relative cell number and cell size was determined by extrapolating the counted hairs in a defined area (0.1 mm)² on the dorsal side of the wing proximal to the posterior cross-vein and posterior to vein five.

Adult flies used for scanning electron microscopy were stored in 70% acetone before they were critical-point dried and coated with gold to be examined under the scanning microscope.

Adult fly heads were cut in half using a razor blade and shortly stored in Ringers on ice. Eyes were then fixed as described in Basler et al. (1991).

Heat-shock-induced Gal4 overexpression clones (y w hsflp; GMR > w⁺ > Gal4, where > corresponds to FRT sites) were induced 24-48 h after egg laying (AEL) by a 1-h heat shock at 37°C. The heat shock induces FLP/FRT-mediated mitotic recombination leading to the excision of the w⁺ marker. scylla charybdis double-mutant clones were induced 24-48 h AEL by a 30-min heat shock at 38°C.

mRNA in situ hybridizations were done essentially as described (Lehmann and Tautz 1994; O'Neill and Bier 1994). In vitro transcription was done using the DIG RNA labeling kit (Roche). DNA fragments corresponding to parts of the scylla and charybdis full-length mRNAs were PCR-amplified and sub-cloned into pCRII-TOPO vector. For the in vitro transcription, SP6 and T7 polymerases were used to produce sense and anti-sense RNAs.

For the S6K assay, second instar larvae were collected by floatation in 30% glycerol. Following a brief wash in water, batches of 20-30 larvae were transferred to 1.5-mL Eppendorf tubes and flash-frozen in liquid nitrogen. The S6K assay was done essentially as described (Oldham et al. 2000). Either 25 or 40 μg of total protein was used for the S6K assay with H2B (Histone type VII-S, Sigma: H4255) as a substrate. Protein concentrations were determined using the RC DC Protein Assay (Bio-Rad).

³²P incorporation into Crosstide or H2B was quantified and background-corrected using a PhosphorImager and image-quant software (Molecular Dynamics).

Western blotting: The following antibodies and dilutions were used: S6K D-20 antibody (Montagne et al. 1999) at 1:200 dilution, PKB antibody (Andjelkovic et al. 1995) at 1:1000 and mouse anti-α-tubulin (Sigma) at 1:2000. HRP-conjugated secondary antibodies (Dako A/S, Glostrup) were diluted 1:2000. Since the same PKB antibody used in immunoprecipitation was also used in Western blotting, HRP-conjugated protein-A (ICN Biomedicals) at 1:5000 was used instead of secondary antibody. Signals were detected using ECL Western blotting detection reagents (Amersham Biosciences).

We are greatly indebted to Thomas Radimerski for his support and the introduction into the PKB- and S6K activity assays and critical reading of the manuscript, George Thomas for the dS6K antibodies and diverse reagents, and Brian Hemmings for the dPKB antibody and the PKB substrate Crosstide. We appreciate the support of Jorge Soliz and Max Gassmann for letting us use the hypoxia chamber. We thank Kathrin Doepfner for the help in generating the char¹⁸⁰ mutant; Christof Hugentobler for excellent technical support; Pablo Wappner, Benny Shilo, and Stephen T. Crews for fly stocks; Matthias Knirr and Christoph M. Schuster for fly stocks and sharing results prior to publication; Christian Frei and Hugo Stocker for critical reading of the manuscript and insightful comments; and the entire Hafen lab for discussions.