TORC2 has been reported to phosphorylate the HM site of AKT (Sarbassov et al. 2005). To evaluate the impact of TORC2 on AKT phosphorylation, we analyzed HM site phosphorylation (S505) in control and rictor mutant larvae. S505 phosphorylation was robust in wild-type larvae, but was barely detectable in the rictor mutants (Fig. 1D). AKT HM phosphorylation was also absent in the adult flies lacking either Rictor or Sin1 (Fig. 1E,F). HM phosphorylation in the rictor mutant could be restored by ubiquitous expression of a rictor transgene (Fig. 1E). Considering that AKT activity is essential during Drosophila development (Verdu et al. 1999) and that HM phosphorylation is considered to be essential for AKT activity (Alessi et al. 1996), we were surprised to find that S505 phosphorylation was apparently not required to support development.

This prompted us to ask whether HM phosphorylation is required for AKT activity. AKT activity was compared in S2 cells depleted of Rictor or AKT by RNA interference (RNAi). In control cells, insulin treatment led to phosphorylation of numerous AKT substrates, detected by an antibody to the phosphorylated AKT target motif RxRxxS/T (Fig. 1G). A majority of the insulin-induced bands were true AKT substrates, as they were not induced in cells depleted of AKT. Depletion of Rictor strongly reduced phosphorylation of AKT as well as the phosphorylation of the full spectrum of AKT substrates, comparable to cells depleted of AKT (Fig. 1G; Supplementary Fig. S2). This shows that TORC2-mediated HM phosphorylation is needed for full AKT activation, at least under conditions of strong insulin stimulation in vitro. We next examined AKT activity in a rictor mutant larval extract to assess the requirement for HM phosphorylation at a more physiological level of PI3K pathway activation. The signal was much lower than in insulin-treated S2 cells; however, in wild-type extracts we detected one prominent band corresponding to the most abundant band in S2 cells (Fig. 1H). Phosphorylation of this protein was strongly reduced in rictor mutants and was detectable only after the blot was overexposed (Fig. 1H, middle panel). This suggests that enzymatic activity of AKT was considerably reduced, but perhaps not completely eliminated, at physiological levels of PI3K pathway activity in the absence of HM site phosphorylation by TORC2.

The possibility of residual AKT activity in the rictor mutant flies could indicate that they retain low levels of HM phosphorylation, sufficient to maintain the essential functions of AKT. Another possibility is that AKT lacking HM phosphorylation retains biological activity. To distinguish between these alternatives, we generated transgenic flies expressing wild-type and mutant forms of AKT under control of the tubulin promoter (Fig. 2A,B). To test their biological activity, we asked whether the mutant proteins could restore AKT activity in eye imaginal discs containing large regions of AKT mutant tissue. Using the eyeless-FLP recombinase system to produce clones of homozygous mutant cells (Newsome et al. 2000) typically leads to an eye composed of >80% of homozygous mutant tissue and <20% of heterozygous tissue if the mutation does not affect cell growth or survival. Because AKT mutant cells are severely growth impaired, a large portion of the AKT mosaic eyes were composed of heterozygous AKT/+ cells (Fig. 2C). Eye-specific loss of AKT led to severe reduction of eye size. Expression of the wild-type AKT transgene restored growth of the mutant tissue. AKT molecules bearing a mutation in the PH domain, which is essential for PIP3 binding (R45A), or in the activation loop site phosphorylated by PDK1 (T324A), did not show any rescue, indicating that these proteins lack AKT activity. In contrast, the AKT mutant lacking the HM phosphorylation site (S505A) rescued the AKT mutant. This was observed in the increased proportion of the eye consisting of AKT mutant cells (pale color) vs. the more darkly pigmented AKT/+ heterozygous tissue, as well as in terms of restoring eye size (Fig. 2C). Thus, in contrast to what might have been expected, AKT lacking HM phosphorylation retains considerable biological activity when assayed in vivo. Note that the in vivo assay for biological activity of AKT may be more sensitive than in vitro assays for kinase activity.

Although significant, rescue by the HM site mutant AKT was less complete than that obtained with wild-type AKT. This is consistent with the observation that flies lacking TORC2 activity showed reduced, but residual AKT activity (Fig. 1H). In view of the importance of AKT activity in tissue growth (Verdu et al. 1999), we asked whether there might be a more subtle growth defect associated with the reduced AKT activity in flies lacking TORC2 activity. Indeed, rictor mutants raised under controlled conditions on a rich diet displayed a modest (~10%) reduction of body weight (Fig. 3A) and a modest reduction in tissue growth, indicated, for example, by smaller wings (Fig. 3B; Supplementary Fig. S1). Both were fully rescued by ubiquitous expression of a UAS-rictor transgene (Fig. 3A,B). Flies mutant for Sin1 displayed a similar modest growth reduction (Fig. 3C). Together these observations suggest that, at normal physiological levels of Insulin signaling, only minor growth impairment results from loss of AKT HM site phosphorylation. The residual level of AKT activity is thus nearly sufficient to support normal development.

AKT signaling has been shown to regulate tissue growth via two effector routes: the TSC/Rheb/TORC1 pathway (Hay 2005), and by inhibiting the FOXO transcription factor (Junger et al. 2003; Puig et al. 2003). To assess the impact of reduced AKT activity without HM phosphorylation, we analyzed the phosphorylation of two TORC1 targets S6K and 4EBP in control and rictor mutant tissue samples and found no meaningful difference (Fig. 4A,B), consistent with the mild growth defects observed in rictor mutants. To ask whether rictor mutants affect FOXO activity, we used a sensitized assay in which FOXO overexpression in the eye imaginal disc using GMR-GAL4 causes a small rough eye phenotype (Junger et al. 2003). The eye was severely reduced in flies raised at 25°C, but at 18°C the eyes were nearly normal, indicating that the severity of this phenotype is sensitive to the level of excess FOXO activity (Fig. 4C). The phenotype caused by FOXO overexpression was strongly increased in the rictor mutant background at both temperatures.

To directly assess the effect of AKT HM phosphorylation on FOXO regulation in vivo, we overexpressed FOXO in an eye in which the endogenous AKT gene was mutant and AKT activity was instead provided by a transgene expressing the S505A mutant. Removing the possibility of AKT HM phosphorylation strongly enhanced the phenotype caused by FOXO overexpression (Fig. 4C, right panel), producing a phenotype very similar to that in the rictor mutant background. We also analyzed FOXO phosphorylation in S2 cells, in which endogenous AKT was depleted by RNAi and replaced by expression of the S505A AKT transgene. FOXO phosphorylation, visualized by a mobility shift on SDS-PAGE (Puig et al. 2003), was clearly compromised in the absence of AKT HM phosphorylation, compared with similarly treated cells expressing the wild-type AKT transgene (Fig. 4D). Thus, TORC2-mediated regulation of AKT is needed to limit FOXO activity by controlling its phosphorylation level in vivo.

The results presented thus far indicate that loss of HM phosphorylation reduced AKT activity and that this had only a modest effect on tissue growth under normal physiological conditions of Insulin/PI3K activity. To test whether TORC2 would have a greater impact on tissue growth when PI3K signaling is compromised, we made use of the finding that PI3K signaling levels can be modulated by controlling the nutritional status of the animal (Junger et al. 2003; Teleman et al. 2005). We compared growth of flies raised on normal nutrient-rich food vs. food with 10% of the nutritional value. Nutrient-reduced food caused a developmental delay of several days and ~25% reduction in growth of control flies. Surprisingly, instead of an enhanced phenotype, rictor mutants raised on nutrient-reduced food were indistinguishable from the control flies (Fig. 5A). Thus, TORC2 activity is not required for growth when PI3K signaling is low.

This suggested that TORC2 activity might be needed in vivo to permit high-level PI3K/AKT signaling. To test this, we compared the effects of genetically elevated PI3K activity in control and rictor mutant flies. Overexpressing the catalytic subunit of PI3K caused tissue overgrowth (as described previously; Leevers et al. 1996) and increased phosphorylation of AKT on S505 (Fig. 5B,C). In rictor mutants, S505 phosphorylation was undetectable, and intriguingly, overgrowth was efficiently suppressed despite elevated PI3K activity (Fig. 5B,C). Removal of PTEN provides an independent means to elevate PIP3 levels and increase AKT activity. PTEN mutant clones cause dramatic hyperplasia in the developing eye by increasing both cell number and cell size (Fig. 5D; Goberdhan et al. 1999; Gao et al. 2000). Removing rictor in this background efficiently suppressed the tissue overgrowth caused by loss of PTEN, leading to a nearly normal-sized eye (Fig. 5D). The observed suppression was likely to be due to abolished AKT HM phosphorylation, as PI3K-induced hyperplasia was also efficiently suppressed in the eye tissue in which endogenous AKT was replaced by the S505A mutant (Supplementary Fig. S3). We next assessed the ability of rictor to suppress overgrowth of the eye caused by clones mutant for TSC1, which activates TORC1 signaling downstream from AKT, and causes massive tissue overgrowth (Gao and Pan 2001; Potter et al. 2001; Tapon et al. 2001). Loss of TSC1 caused robust overgrowth of the eye, but this was not suppressed by simultaneously removing rictor (Fig. 5E). Thus, the growth-limiting effects of rictor are specific to activation of the pathway at or above the level of AKT.

The PI3K/AKT signaling pathway is conserved between Drosophila and mammalian species. The lack of genetic redundancy among pathway components makes Drosophila a useful system in which to dissect the roles of the individual pathway members in vivo. Earlier analyses of other pathway members have shown that the Insulin receptor, PI3K, PDK1, and AKT are each essential for viability, and that mutant tissue displays severe undergrowth (Supplementary Fig. S4; Verdu et al. 1999; Weinkove et al. 1999; Brogiolo et al. 2001; Rintelen et al. 2001). Mutants of Drosophila insulin receptor substrate homolog, chico, are semiviable but severely growth impaired (Böhni et al. 1999). Although individual AKT mutants are viable in mouse, the essential nature of AKT is likely to be masked by genetic redundancy among the three AKT genes (Chen et al. 2001; Cho et al. 2001a, b; Peng et al. 2003). Previous studies in cultured cells have suggested that TORC2 is an important regulator of AKT phosphorylation and activity, and that this phosphorylation event is required for AKT kinase activity (Alessi et al. 1996; Sarbassov et al. 2005). It was recently shown that loss of TORC2 activity in rictor mutant mice leads to loss of AKT HM phosphorylation and to embryonic lethality (Shiota et al. 2006; Yang et al. 2006), suggesting that HM phosphorylation is essential for AKT activity in the mouse. In contrast, our findings show that TORC2-mediated phosphorylation on the HM site is not essential for AKT activity in vivo. Indeed, although AKT activity was reduced, considerable residual activity was found in flies lacking TORC2 activity. Flies expressing a mutant form of AKT lacking the HM phosphorylation site also showed considerable AKT activity in vivo. Our findings indicate that the maximal level of AKT activity is limited in the absence of HM phosphorylation. Under normal physiological conditions in Drosophila, this reduced level of AKT activity is almost sufficient to support normal growth. But without HM phosphorylation, AKT cannot transduce the higher-than-normal levels of PI3K pathway activity that result from mutation of the tumor suppressor PTEN or increased insulin stimulation. When considered in this context, the lethality of rictor mutant mice could reflect a higher threshold in the requirement for AKT activity in some biological process in mouse than in fly, but the possibility of essential TORC2 targets other than AKT cannot be excluded.