Living with Lethal PIP3 Levels: Viability of Flies Lacking PTEN Restored by a PH Domain Mutation in Akt/PKB

See allHide authors and affiliations

Science  15 Mar 2002:
Vol. 295, Issue 5562, pp. 2088-2091
DOI: 10.1126/science.1068094


The phosphoinositide phosphatase PTEN is mutated in many human cancers. Although the role of PTEN has been studied extensively, the relative contributions of its numerous potential downstream effectors to deregulated growth and tumorigenesis remain uncertain. We provide genetic evidence in Drosophila melanogaster for the paramount importance of the protein kinase Akt [also called protein kinase B (PKB)] in mediating the effects of increased phosphatidylinositol 3,4,5-trisphosphate (PIP3) concentrations that are caused by the loss of PTEN function. A mutation in the pleckstrin homology (PH) domain of Akt that reduces its affinity for PIP3 sufficed to rescue the lethality of flies devoid of PTEN activity. Thus, Akt appears to be the only critical target activated by increased PIP3 concentrations in Drosophila.

Mutations in the tumor suppressor gene PTEN (the phosphatase and tensin homolog on chromosome 10) are frequent in glioblastomas, endometrial carcinoma, melanomas, and prostate cancer (1). Furthermore, two dominant hamartoma syndromes, Cowden disease and Bannayan-Zonana syndrome, are linked to germ line mutations in PTEN (1). The PTEN protein carries a phosphatase domain resembling those of dual-specificity protein phosphatases (2–4). Although it can dephosphorylate protein substrates such as focal adhesion kinase (5) and the adapter protein Shc (6), PTEN's predominant enzymatic activity appears to be the dephosphorylation of phosphoinositides at the D3 position. Because PTEN uses the second messenger PIP3 as a substrate, PTEN antagonizes the function of phosphatidylinositol-3 kinase (PI3K) (7, 8). Immortalized mouse embryonic fibroblasts or embryonic stem cells lacking PTEN function show an approximately twofold increase in PIP3 concentrations (9, 10). PIP3 interacts with a wide variety of PH domain-containing proteins, including the serine-threonine kinases Akt (also called PKB) and phosphoinositide-dependent kinase 1 (PDK1), Btk family tyrosine kinases, guanine nucleotide exchange factors for the Rho and Arf families of small guanosine triphosphatases, and phospholipase Cγ (11, 12). The plethora of proteins that are potentially regulated by PIP3 provides widespread signaling potential for this lipid second messenger.

Genetic analyses in model organisms have implicated PTEN as a negative regulator of insulin receptor signaling. In the nematodeCaenorhabditis elegans, PTEN antagonizes the activity of the PI3K AGE-1 in the regulation of metabolism, development, and life span (13–16). In the fruit flyDrosophila melanogaster, PTEN counteracts signaling downstream of the insulin receptor to control cellular growth (17–19). There are, however, additional phenotypes associated with mutations in PTEN that cannot easily be reconciled with an exclusive function of PTEN in insulin receptor signaling [for example, the burst vulva phenotype in C. elegans (13) and defects in the actin cytoskeleton inDrosophila (17)]. To better understand the consequences of loss of PTEN function, it would be useful to know which important downstream effectors react to increased PIP3 concentrations and whether PTEN has other physiological substrates in addition to PIP3.

The protein kinase Akt is an important component of insulin receptor signaling (20). Akt is recruited to the plasma membrane by virtue of the interaction of its NH2–terminally located PH domain with PIP3. At the membrane, subsequent phosphorylation events by PDK1 and an unidentified kinase lead to the full activation of Akt (21–23). InPTEN-deficient mouse embryonic fibroblasts and embryonic stem cells, Akt is phosphorylated and activated (9, 10). The phenotypes associated withAkt mutations in both C. elegans andDrosophila are consistent with its role in signal transduction downstream of the insulin receptor (24–27).

We monitored three properties of Drosophila Akt (dAkt) separately: kinase activity, abundance of the protein, and membrane localization. We relied entirely on mutations in the endogenous gene encoding dAkt to avoid potential side effects caused by overexpression of mutant proteins. dAkt1 encodes a catalytically inactive protein, dAktF327I (25). The viabledAkt4226 allele contains a P-element insertion upstream of the dAkt gene and therefore results in the reduced expression of wild-type dAkt protein (19,28). Finally, we characterized the viable hypomorphic mutation dAkt3 (29) that selectively impairs the membrane recruitment of dAkt in response to increased concentrations of PIP3. Sequencing of genomic DNA extracted fromdAkt3 homozygous flies revealed a single nucleotide exchange resulting in the substitution of a serine residue for a nonconserved glycine at the end of the sixth β sheet of the PH domain. To address the mechanism by which this Gly99 → Ser99 (G99S) mutation in the PH domain affects dAkt, we compared the amount of dAkt protein and activity in wild-type anddAkt3 mutant larvae. Whereas no apparent difference in expression of the protein was observed (Fig. 1A, inset), dAkt activity from the mutant larvae represented only 30% of that in wild-type larval extracts (Fig. 1A). We also expressed epitope-tagged forms of wild-type dAkt, catalytically inactive dAktF327I, and PH domain mutant dAktG99S in insulin-responsive human embryonic kidney (HEK) 293 cells. All three proteins were expressed in similar amounts (30). dAktG99S activity from insulin-stimulated cells was reduced by about 90% as compared to that of the wild-type kinase (Fig. 1B). All forms of dAkt proteins were detected in the cytosol of unstimulated cells (Fig. 2, A, D, and G). Stimulation of the cells with insulin for 5 min resulted in association of the wild-type and the catalytically inactive enzymes with the plasma membrane, but failed to recruit the dAktG99S mutant protein (Fig. 2, B, E, and H). In contrast, treatment of HEK 293 cells with the protein-tyrosine phosphatase inhibitor pervanadate, a potent activator of Akt (31), led to membrane recruitment of all dAkt proteins (Fig. 2, C, F, and J). Consistently, pervanadate treatment stimulated dAktG99S activity to 80% of the wild-type level. However, pervanadate-induced activation of the mutant protein occurred more slowly than did that of the wild-type kinase (30). Taken together, these data indicate that the G99S substitution reduces the association of dAkt with the plasma membrane, probably by affecting the affinity of its PH domain for PIP3. Thus, dAkt3 enabled us to study the consequences of impaired recruitment of dAkt to the plasma membrane.

Figure 1

Reduced kinase activity caused by an amino acid substitution in the PH domain of dAkt. (A) Effect of the G99S substitution in the PH domain on dAkt kinase activity from larval extracts (42). Activity from wild-type larvae was considered to be 100%. Inset, dAkt protein was detected in 40 μg of larval extracts using the same antiserum. (B) Reduced insulin-induced activation of the G99S mutant dAkt. The dAkt constructs were expressed in HEK 293 cells (43). Transfected cells were starved for 24 hours before stimulation with insulin for the indicated time periods, and dAkt kinase activity was determined (44). The activity of wild-type dAkt from unstimulated cells was considered to be relative activity = 1.

Figure 2

Reduced membrane localization of the G99S mutation of dAkt. HEK 293 cells plated on coverslips were transfected with epitope-tagged wild-type (A toC), G99S (D to F), and F327I (G to J) dAkt and deprived of serum for 16 hours before stimulation with insulin (B, E, H) or pervanadate (C, F, J) for 5 min. Fixed and permeabilized cells were incubated first with the monoclonal antibody 12CA5 to the HA epitope and then with fluorescein isothiocyanate–conjugated secondary antibody. An analysis by confocal microscopy revealed the subcellular localization of the dAkt variants.

We combined the dAkt alleles with null mutations indPTEN (32). Animals lacking dPTEN function die during larval stages. A reduction in dAkt expression using the viable dAkt4226 allele did not rescue the lethality associated with dPTEN. Similarly, animals doubly mutant for dPTEN and dAkt1 did not survive. Thus, either dAkt activation is not the sole reason for the lethality caused by loss of PTEN, or dAkt function is not dispensable in the absence of dPTEN. The latter hypothesis is strongly supported by results obtained with the dAkt allele that selectively impairs the membrane recruitment of dAkt. Flies devoid of functional dPTEN were rescued to viability by any dAkt allelic combination that included dAkt3 (Fig. 3A) (33). The rescued flies did not display morphological defects that would be expected in light of the phenotypes ascribed to clones of dPTEN mutant cells (17). Tangential sections through compound eyes revealed essentially normal ommatidial and rhabdomeric structures, and the wings of the rescued flies showed no abnormalities in the venation, such as missing crossveins (34). We determined the PIP3/PIP2 ratio by metabolic labeling of phospholipids from larvae (35). PIP3 concentrations were increased in the dPTEN dAkt doubly mutant larvae as compared to those of wild-type larvae (Fig. 3B), excluding the possibility that PIP3 concentrations remain within physiological limits by a feedback regulation mechanism involving dAkt. This suggests that the potential activation of a number of PH domain–containing proteins other than dAkt does not interfere with viability.

Figure 3

Restored viability of flies lacking dPTEN function by the PH domain mutation in dAkt. (A) Morphology and weight of dPTEN mutant flies rescued bydAkt3 /dAkt1 . The left panel shows female flies, the middle panel shows male flies, and the right panel shows the weight of adult male flies. (B) PIP3 concentrations in flies devoid of dPTEN function rescued by the dAkt3 mutation.

Our results indicate that the activation of dAkt is the only crucial outcome of the loss of dPTEN function. Activation of dAkt should therefore mimic the dPTEN loss-of-function phenotype. We expressed a constitutively activated membrane-anchored dAkt during eye development (36). The resulting eyes were increased in size due to enlarged ommatidia (Fig. 4B), a phenotype similar to that seen in clones of dPTEN mutant cells (17–19). This overgrowth phenotype is independent of upstream signals, because it was still evident in achico or Dp110/PI3K mutant background (Fig. 4D) (34).

Figure 4

Growth in the developingDrosophila eye promoted by activated dAkt. (A toD) Scanning electron micrographs of compound eyes of female flies; the anterior is to the left. (A) Wild type. (B) Overexpressing a membrane-tethered version of dAkt (GMR-Gal4 UAS-myr-dAkt). (C) chico mutant. (D) GMR-Gal4 UAS-myr-dAktin a chico mutant background (45).

We conclude that flies devoid of the tumor suppressor dPTEN can live with abnormally high concentrations of PIP3 if only the affinity of dAkt for PIP3 is decreased. Thus, the PH domain–mediated translocation of dAkt to the membrane and its subsequent activation is the only lethal event triggered by increased PIP3 concentrations. Because the PH domain of Akt interacts with the substrate of PTEN's lipid phosphatase activity, we also conclude that PTEN does not exert any essential function other than the dephosphorylation of PIP3.

  • * Present address: Department of Vascular and Metabolic Diseases, F. Hoffmann-La Roche AG, CH-4070 Basel, Switzerland.

  • To whom correspondence should be addressed. E-mail: hafen{at}


View Abstract

Navigate This Article