Our study indicates that cells employ multiple and possibly mutually exclusive mechanisms to activate AKT (). The reasons why RTKs would employ two distinct modes of AKT activation are not entirely clear. However, a fraction of AKT appears to utilize this alternative mode of activation in normal and prominently in cancerous cells. Our studies showed that even in the presence of PI3K inhibitor, ligand bound HER2/ErbB-2 or EGFR activated Ack1 which in turn Tyr-phosphorylated and activated AKT. AKT is frequently activated in pancreatic cancer which has been shown to be highly correlated to HER-2/neu
. Moreover, many of the pancreatic cell lines and tumors expressing activated AKT had retained wild-type PTEN 
. We noticed that PanIN, pancreatic adenocarcinoma and breast tumors of MMTV-neu
mice exhibit significantly higher levels of pTyr284-Ack1 and pTyr176-AKT (unpublished data). Taken collectively, our data may explain AKT activation in those tumors that display amplification/activation of RTKs but have normal PI3K/PTEN levels. We propose that other tumors that possess somatic autoactivating mutations or amplification in non-receptor tyrosine kinases could use similar mechanisms for AKT activation 
Tyr176-phosphorylation leads to AKT activation, a model.
Are there conditions when Tyr176 modification is not needed for AKT activation? Some of the conditions when Tyr176 phosphorylation of AKT is not required for AKT activation could be; 1) Presence of constitutively active PIK3CA
mutations, observed in colorectal, glioblastomas, gastric breast and lung cancers 
. 2) Loss of tumor suppressor PTEN resulting in increased levels of cellular PIP3, occur commonly in prostate cancer, endometrial cancer, and glioblastoma, among others 
. 3) A rare somatic activating mutation, E17K in the pH domain which facilitates AKT recruitment to the membrane in PIP3-independent manner 
We have used the term AKT ‘translocation’ to indicate emergence of (cytosolic) AKT in the plasma membrane in response to growth factors. Our data () demonstrate that AKT in the plasma membrane is phosphorylated at Tyr 176 and mutation of this site in AKT abrogates appearance of AKT in the plasma membrane (). Based on the evidence, our model () suggests that as Ack1 signaling pathway is initiated at the plasma membrane by RTKs. Ack1 associates with growth factor-bound RTKs (via Mig6 homology domain in Ack1 carboxy terminal proline rich region) and is activated 
. Ack1 is constitutively bound to AKT (); Activated Ack1 directly phosphorylates AKT at Tyr176, thus facilitating accumulation of Tyr176-phosphorylated AKT at the plasma membrane. Tyr176-phosphorylated AKT preferentially binds PA, a plasma membrane phospholipid as opposed to unphosphorylated AKT (refer to Fig. S8
for details). PH domain in AKT is a lipid binding domain and thus might be involved in the membrane binding of Tyr176-phosphorylated AKT. Collectively, our data suggests that Ack1 mediated AKT Tyr176-phosphorylation is driving this translocation process. Thus, although AKT Tyr176-phosphorylation and its migration to the plasma membrane is PIP3 independent, the recruitment of Tyr176 AKT in the plasma membrane may require a functional PH domain.
In contrast to AKT, pTyr176-AKT specifically binds the plasma membrane anionic phospholipid, PA (Fig. S8
). Tyr176-phosphorylation could induce conformational changes in the AKT PH domain to enable binding to PA. The PH domain of Son of sevenless (SOS) and PX domains of p47phox
have previously been shown to possess a phosphoinositide-binding pocket and a second anion binding pocket which enables them to interact with PA facilitating plasma membrane recruitment 
. We speculate that AKT too might possess a masked anion binding pocket, and Tyr-phosphorylation induced conformational changes could unmask this pocket allowing it to bind PA.
In endogenous systems Ack1 associates with AKT2 albeit weakly as compared to AKT1 (). AKT isoforms are differentially distributed among different cellular compartments 
with majority of AKT1 in the cytosol, and AKT2 in the mitochondria. Additionally AKT2 protein appears to be not as abundant as AKT1 in MCF-7 and MEFs (Fig. S1A
). Thus, weak interaction with AKT2 could be a combined outcome of differential cellular distribution and lower protein levels. However, our unpublised data demonstrates significant tyrosine phosphorylation of AKT2 upon coexpression of Ack1 and AKT2 in HEK293T cells, suggesting that both AKT1 and 2 are Ack1 substrates.
This study demonstrates that Tyr176-phosphorylation is sufficient for AKT membrane localization followed by PDK1/PDK2 mediated activation, defining the upstream Ack1 kinase activity as ‘PDK3’. We do not rule out the possibility that other tyrosine kinases may be able to target AKT for Tyr176-phosphorylation. Ack1 knockout mice are not currently available. However, when they are developed, they would allow us to investigate whether AKT can be phosphorylated at Tyr176 by other receptor or non-receptor tyrosine kinases in response to growth factors. Multiple non-receptor tyrosine kinases were earlier shown to increase AKT activity 
, however, precise mechanism of AKT activation by any of the Tyr-modifications is not clear, nor is their role in initiation or progression of cancer. To our knowledge, this report provides the first demonstration for a role of Tyr-phosphorylated AKT in its compartmentalization, which allowed us to delineate its critical role in AKT kinase activation, its potential to initiate neoplasia in mouse prostates and promote disease progression in human breast cancers. Large numbers of tumors are reliant upon AKT activation for survival and growth making it an attractive target for molecular therapeutics 
. The assay that was used during development of AKT inhibitors was primarily based on AKT Ser473-phosphorylation. Our data indicates that a new class of AKT inhibitors can be identified based on AKT Tyr176-phosphorylation. These novel inhibitors that block AKT membrane localization and activation could have major implications in cancer, diabetes and obesity research.