Full activation of PKB/Akt in response to insulin or ionizing radiation is mediated through ATM

The gene mutated in Ataxia Telangiectasia (ATM) has been implicated in several cell functions such as cell cycle control, response to DNA damage and insulin. PKB/Akt has also been implicated in the cellular response to insulin, γ− radiation and cell cycle control. Interestingly, lack of PKB/Akt function in vivo is able to mimic some phenotypic abnormalities associated with Ataxia Telangiectasia (AT). Here we show that ATM is a major determinant of full PKB/Akt activation in response to insulin or γ− radiation. This effect is mediated through the Phosphatidylinositol-3 kinase (PI3k) domain of ATM which affects specifically Akt Serine 473 phosphorylation. This conclusion was inferred from the results obtained in transient transfection assays using exogenous PKB/Akt and ATM in Cos cells. Moreover, the use of ATM inhibitors or siRNA confirmed our observation. Further supporting these results we also observed that biological responses tightly regulated by Akt, such as FKHR activity after insulin treatment or γ radiation response, were altered in cell lines derived from AT patients and Knock Out (KO) mice for ATM in which phosphorylation in Serine 473 was almost abolished. This study proposes new clues in the search of the unknown PDK2 and proposes new explanations for the radiosensitivity or insulin intolerance described more than 30 years ago in AT patients. immunoprecipitation from 293T. 32 P incorporation was detected by film exposure. Aliquots from kinase reaction were used to detect HA-Akt and Flag-ATM.


Introduction
Ataxia telangiectasia (AT) is an autosomal recessive disorder characterized by growth retardation, cerebellar ataxia, cancer susceptibility, oculotaneus telangiectasia, severe immunodeficiency, resistance to insulin and hypersensitivity to gamma radiation among other symptoms (for review see (1). At the cellular level, the role of the gene Mutated in Ataxia Telangiectasia (ATM) includes a wide spectrum of biological processes including, cell cycle control, genome stability, apoptotic machinery, response to genotoxic stress and others (2)(3)(4)(5). The protein encoded by the ATM gene belongs to the super family of Phosphatidylinositol-3 kinase (PI3K) due to the presence of a p110 like domain but no lipid kinase activity has been demonstrated so far (6). This domain is responsible for the phosphorylation of proteins such as p53, ChK1, 4EBP, BRCA1 following a conserved domain -S/T-Q-in which S or T is the phosphoaceptor residue (for a review see (7). In fact, regulation of ATM is mediated by autophosphorylation in residue Ser1981, at the FAT domain, following this conserved motif as has been recently demonstrated (8).
The PKB/Akt, referred now as Akt, protein is a key player of a pathway related to survival, by inhibition of apoptotic signals and promoting cell cycle progression with a clear implication in cancer and other pathologies (9,10). Also, Akt has been shown to be potently activated in response to a wide variety of growth factors such as insulin or IGF-1, and more recently in response to DNA damage (11,12). Phosphorylation in two residues seems to be critical to total Akt activation, a Thr in position 308 mediated were from Sigma. Cells for Akt phosphorylation assays were treated and collected in lysis buffer (25mM HEPES pH 7.5, 0.3 M NaCl, 1.5mM MgCl 2 , 0.2 mM EDTA, 1% Triton X-100, 0.1% SDS, 0.5% Deoxycholic Acid, 20mM B-glycerophosphate) in the presence of protease and phosphatase inhibitors (20µg/ml Aprotinin, 20µg/ml Leupeptin, 1mM PMSF and 0.1mM Na 3 VO 4 , 20mM NaF, 1mM NaPPi) and protein quantity was quantified (Biorad assay, München, Germany). Cells for coimmunoprecipitation assays were collected in HNTG lysis buffer (23). Then, lysates were processed for western blotting (usually 50 µg) or IP using specific antibodies and Protein G (Amershan Pharmacia). Immunocomplexes were extensively washed and resuspended in Sample buffer 5x. Antibody detection was achieved by enhanced chemiluminescence (ECL, Amersham Pharmacia, Uppsala, Sweden)

PCR protocols and generation of GST fusion proteins
Primers and protocol for cloning PI3K domain from Y5Z DNA has been previously described (24). PCR product was cloned onto pEF1/His-C (Invitrogene  . 1c). It has been reported kinase activity in Thr308 mutant Akt after MAPKAP2 in vitro phosphorylation of Ser473 residue, indicating that Ser473 activation can render kinase activity by itself supporting our observations (11).
Interestingly, γ−radiation, a classical ATM stimulus, was also able to induce Akt kinase activity ( fig.1c). However, all the previous experiments were performed by overexpression of ATM or using chemical inhibitors, known to block PI3K also implicated in Akt activation. Therefore, we decided to inhibit specifically ATM using RNA interference technology. Cos cells were transfected with exogenous ATM in the presence of increasing amounts of validated siRNA for ATM (ranging from 0 to 400 nM) and ATM expression was monitorized (upper panel fig. 1d). Therefore, we performed the same experimental approach with endogenous ATM stimulated with Insulin in the presence of HA-tagged Akt (lower panel fig.1d). While a marked decrease was observed in Ser473 phosphorylation, no effect was detected in Thr308. The same experiment was performed in Hela cells with exact results (data not shown).

Activation of PKB/Akt at Ser473 is mediated through ATM
To evaluate the implication of the PI3K domain of ATM in the Ser473 phosphorylation, we decided to PCR-clone this domain, known to mediate some of the biological properties of ATM such as radiosensitivity (24). The 6-His tagged PI3K domain from ATM was also able to induce Ser473 and Ser15 phosphorylation of Akt and p53 respectively, in a dose dependent fashion inhibited by wortmannin ( fig. 1e).
Expression of the PI3K domain of ATM was evaluated by western blotting anti 6-His tag (data not shown). However p110α, the catalytic subunit of the conventional PI3K, was able to induce phosphorylation in both Akt residues, suggesting a clear difference between these two PI3K families ( fig. 1f).

ATM binds to Akt and mediates Ser473 Akt phosphorylation
In order to gain further insight in to the molecular mechanism of Akt activation mediated by ATM we decided to investigate the possibility of a physical interaction between ATM and Akt. Therefore, 293T cells were transiently transfected with HA-Akt plus Flag-6His-tagged ATM wt or Kd. Extracts were immunoprecipitated against 6-His and blotted against HA. A clear binding between the two molecules was detected, demonstrating that both ATM wt and ATM kd were in vivo associated with HA-Akt ( fig.   2a). In addition, we decided to analyse the phosphorylation status of the Akt bound to ATM, we observed that it was only phosphorylated in Ser473, while no signal was detected in Thr308 ( fig 2a). As a positive control, GFP-p53 was cotransfected with ATM and, we detected binding and p53 phosphorylation at Ser 15 ( fig. 2b). In this scenario we decided to analyse the possibility of a direct phosphorylation in the Ser473 of Akt, a putative phosphoaceptor in the ATM phosphorylation consensus site, although it is not followed by a Q (7). In this context, and as a first approach, we decided to use the whole molecule as a substrate. Thus, HA-tagged Akt wt and a triple mutant in positions 473, 308 and 179 were transfected in Cos cells, immunoprecipitated and used as a substrate for in vitro kinase assay using ATM immunoprecipitated from 293T cells. In this case a 32 P incorporation was detected in the presence of ATM wt form, while we  Defective Akt activation correlates with the AT associated radio sensitivity.
However, using only one stimulus, insulin, could not allow us to propose a general mechanism in which ATM is controlling Akt activity. Therefore we challenged another very well established stimulus for ATM also able to activate Akt, which is γ−radiation ( fig. 1c). Then GM09607 and GM00637 γ-irradiated cells were analysed for Akt activity using an in vitro kinase assay. Lack of Akt activity was detected in AT cells correlating with a sensitive phenotype and reverted by over-expression of ectopic ATM wt ( fig. 5a). Exact results were obtained in the model derived from KO cells ( fig. 5 b).

Discussion
In this report we demonstrate that ATM is implicated in Akt Ser473 Radiosensitivity, the classical hallmark of AT, can be explained by lack of Akt activity as has been observed in Akt1 deficient mice (22), while hyperactivation has been related to resistance (12). Indeed, the increase in the G2/M checkpoint after γ radiation in AT cells (33), is probably due to the low Akt activity, that is a key protein

Figure 2. ATM binds but does not directly phosphorylate Akt in Ser 473
a HA-Akt (4µg) was cotransfected with Flag-6His-ATM wt or kd (4µg)in 293T cells.
Coimmunoprecipitation was done with 6-His antibody and WB was developed with HA antibody (left panel) or with phospho-Akt against Thr308 or Ser473 (right panel).
Positive controls for Akt phosphorylation were performed with insulin (100nM) for 20 minutes.
b Coimmunoprecipitation experiments between p53 Flag-6-His ATM wt or kd following the same approach as in fig. 2a.
c Immunoprecipitated HA-Akt wt (wt) and triple mutant (mut) (2µg) from Cos cells was used as a substrate for in vitro kinase assays. Flag-6-His ATM wt or kd (7µg) were obtained by transfection and immunoprecipitation from 293T. 32 P incorporation was detected by film exposure. Aliquots from kinase reaction were used to detect HA-Akt and Flag-ATM.

Activation of PKB/Akt at Ser473 is mediated through ATM
d GST-Akt wt (wt), GST-Akt Ser473Ala (mut), GST-p53wt (wt) or GST-p53ser15Ala (mut) were used as a substrate for in vitro kinase assays. Flag-6-His ATM wt or kd were obtained as in fig 2c. 32 P incorporation was detected as in fig. 2c. Aliquots from kinase reaction were used to detect GST-Akt, GST-p53 and Flag-ATM.
Results show a representative blot out of three with nearly identical results. c Insulin receptor α subunit levels were measured. Akt was used as a loading control.