PI3K Charges Ahead

See allHide authors and affiliations

Science  13 Jul 2007:
Vol. 317, Issue 5835, pp. 206-207
DOI: 10.1126/science.1146073

Jennifer Y. Lee, Jeffrey A. Engelman, Lewis C. Cantley*

The phosphoinositide 3-kinase (PI3K) signaling pathway is crucial to the viability of many cancers (1-3). PI3K inhibitors block cancer cell growth and survival, and some genetic alterations, such as loss of the PTEN tumor suppressor, increase PI3K activity in cancer cells. After the seminal discovery that PIK3CA—the gene encoding the PI3K catalytic subunit (p110α)—is mutated in cancers (4), these mutant p110α proteins were found to have constitutive PI3K activity and the capacity to transform normal cells into cancer cells (5). Mutations in PIK3CA occur throughout p110α, but two hotspot regions—the helical and kinase domains—comprise more than 80% of the mutations. However, the molecular mechanisms by which these mutations increase PI3K activity have remained a mystery. On page 239 of this issue, Miled et al. (6) elucidate how one of the hotspot mutations increases PI3K activity. The findings reveal a new mechanism for activating PI3K and suggest new possibilities for therapeutic agents that target this enzyme.

PI3K (specifically, class IA PI3K) is a heterodimer consisting of a p85 regulatory subunit and a p110 catalytic subunit (1, 7). Normally, PI3K, which resides in the cytoplasm, is activated upon binding to either a receptor tyrosine kinase at the cell surface (for example, the platelet-derived growth factor receptor) or to adaptor molecules (such as insulin receptor substrate-1, which is phosphorylated by activated insulin receptor). The p85 regulatory subunit mediates these binding events through two Src homology 2 (SH2) domains that interact with phosphorylated tyrosine residues on the receptor or adaptor protein. As a result, PI3K localizes to the plasma membrane, where it phosphorylates the membrane lipid phosphatidylinositol 4,5-bisphosphate (PIP2) to produce phosphatidylinositol 3,4,5-trisphosphate (PIP3). This leads to activation of downstream signaling pathways that control cell growth and survival.

The p85-p110α interaction is remarkably strong, and p110α is stable only when it is bound to p85 (8). This heterodimer is formed by the binding of the p85 inter-SH2 (iSH2) domain to the p110α adaptor-binding domain (ABD) (see the figure). However, the structure of this interaction had remained unknown until Miled and colleagues crystallized the p110α ABD bound to the p85 iSH2 domain. They found that the ABD forms a ubiquitin-like domain that contacts the antiparallel coiled-coil region of iSH2. However, previous data implied additional interactions between p85 and p110α—the p85 N-terminal SH2 domain (nSH2) inhibits purified p110α in vitro (9). This suggested that binding of the p85 nSH2 domain to a receptor or adaptor releases p110α from this inhibitory interaction with p85, without disrupting the iSH2-ABD interaction (see the figure).

In light of these structural findings, the authors examined two classes of p110α oncogenic mutations: the prevalent helical domain mutations and the less common ABD mutations. Several mutations in the helical domain result in an amino acid of opposite charge, such as a glutamic acid (negative charge) to lysine (positive charge) change at position 545 (E545K). Through biochemical studies, the authors found that, unlike wild-type p110α, the E545K mutant is not inhibited by p85 nSH2. Thus, this region of the p110α helical domain likely binds to p85 nSH2, an interaction that normally maintains wild-type PI3K in a low-activity state. The authors hypothesized that important charge-charge interactions occur between the p110α helical domain and p85 nSH2 domain, and they mutated all nSH2 basic residues to acidic ones. They identified two important residues in p85 (Lys379 and Arg340) that are required to inhibit p110α. When they reversed the charge on these basic residues by mutating to glutamates, the mutant p85 nSH2 domains effectively inhibited p110α E545K but not wild-type p110α. Thus, the authors propose that p110α E545K is oncogenic because it is not inhibited by the p85 nSH2 domain (see the figure).

A cancer-linked mutation in PI3K.

The major stabilizing interaction in PI3K involves the p85 iSH2 and p110α ABD domains. Inhibition of PI3K is mediated by a charge-charge interaction (shown as plus and minus signs) between the p85 nSH2 domain and the helical (H) domain of p110α. PI3K localizes to the membrane where interaction with an activated receptor relieves the inhibition. In PI3K with an oncogenic p110α charge-reversal mutation in the helical domain (E545K), the inhibitory interactions are abrogated, resulting in constitutive PI3K activation.


It is less clear how the ABD mutations activate PI3K. Although the ABD binds to p85, ABD mutations are located on the exposed surface oriented away from the iSH2 domain, suggesting that ABD mutations do not directly interfere with p85-p110α interaction. Instead, they may distort orientation of the ABD with respect to the catalytic core, affecting the intrinsic enzymatic activity of p110α or its interactions with other proteins.

How does the E545K mutant promote cell growth and survival in the absence of growth factors? Although the mutation abrogates intermolecular inhibition, it does not explain its membrane localization. The p85 nSH2 domain, unencumbered by interaction with the p110α helical domain, might remain more tightly associated with receptors and adaptors, protecting critical tyrosine residues from dephosphorylation and thereby prolonging PIP3 production. It is also possible that the mutant PI3K localizes to the membrane through membrane-bound Ras protein (p110α has a Ras-binding domain) or random encounters with membrane lipid substrate. The research by Miled et al. will hopefully spur efforts to crystallize the holoenzyme with both the wild-type and mutant p110α proteins.

PI3K inhibitors are now being evaluated in clinical trials. Analogous to the success of drugs that block kinases—trastuzumab in HER2-amplified breast cancers, imatinib in Philadelphia-chromosome chronic myelogenous leukemia, and gefitinib in EGFR-mutant lung cancers—cancers with genetic activation of PI3K signaling could be susceptible to PI3K inhibitors. Thus, cancers with PIK3CA mutations (or PTEN loss) will be carefully investigated for sensitivity to PI3K inhibitors. The p110α E545K mutant may be susceptible to compounds that bind to its unique helical domain surface. Such a specific therapy would be expected not to inhibit wild-type PI3K, thus reducing unwanted side effects. Moreover, different PIK3CA mutations may be functionally distinct and their effects on cellular responses to inhibitors could also be variable. Thus, we are reminded not to necessarily group all cancers with PIK3CA mutations together when analyzing cancers and their response to targeted therapies.


Navigate This Article