Unveiling molecular insights into the mechanism of activation of oncogenic phosphoinositide 3-kinase mutants

An article by Liu et al. published in the Proceedings of National Academy of Science USA describes cryo-electron microcopy (cryo-EM) structures of three oncogenic gain-of-function mutants of p110α, the catalytic subunit of phosphoinositide 3-kinases (PI3Ks), occupied by the catalytic inhibitor BYL-719 and in complex with the regulatory subunit p85α (1). In an earlier manuscript published in 2021 (2), the same team described the cryo-EM structures of wild-type (WT) p110α-p85α heterodimers occupied by the catalytic inhibitor BYL-719 and of WT p110α-p85α heterodimers in complex with a phospho-tyrosine (P-Tyr) peptide. The P-Tyr peptide binds to p85α and thereby stimulates the kinase activity of p110α by an allosteric mechanism. Comparison of the three oncogenic p110α mutant structures to the earlier reported structures of WT p110α and P-Tyr-activated WT p110α provides important mechanistic insights into normal allosteric activation of WT p110α as well as aberrant activation induced by oncogenic mutations in p110α that function as critical drivers of a variety of cancers.

PI3Ks are enzymes that catalyze the phosphorylation on the 3’ position of hydroxyl groups of phosphatidylinositol molecules located in the cell membrane. The formation of phosphorylated phosphatidylinositol molecules (phosphoinositides) is induced by a variety of extracellular and intracellular signals and plays a critical role in controlling cell proliferation, metabolism, cell migration, immune response, and other fundamental cellular processes (3–5). Moreover, the exact balance of the PI3K activity is essential for normal development and tissue homeostasis; aberrant inactivation or activation via genetic gain-of-function mutations results in severe developmental disorders or cancers, respectively (5–7).

The structural analyses described in the manuscript by Liu et al. (1) provide insights into the mechanism of aberrant oncogenic activation of members of class-I PI3Ks. Class-I PI3Ks were shown to be activated by a variety of extracellular signals including receptor tyrosine kinases (RTKs), immune receptors, cytokine receptors, and G-protein coupled receptors, among other extracellular cues (8–11). Class-I PI3Ks are heterodimeric enzymes composed of different isoforms of p85, a regulatory subunit that forms a complex with a catalytic kinase subunit p110, the latter being responsible for catalyzing the phosphorylation of the headgroup of phosphatidylinositol-4,5-bisphosphate (PtdIns(4, 5)P₂) (8, 9). The p85 regulatory subunit contains two src homology 2 (SH2) domains that bind to specific P-Tyr-containing sequences in activated RTKs (e.g., EGF receptor) or tyrosine-phosphorylated docking proteins (e.g., IRS1 and IRS2). By binding to these specific sequences, PI3K translocates to the cell membrane in close proximity to its substrate PtdIns(4, 5)P₂ (10). The generated PtdIns(3, 4, 5)P₃ molecules at the cell membrane in turn serve as specific binding sites for pleckstrin homology domains of a variety of signaling molecules, including the protein kinase AKT, the docking proteins GAB1 and GAB2, and other intracellular signaling molecules that mediate their cellular responses at the cell membrane. These mechanisms represent important steps in the intracellular signaling pathways activated by RTKs that directly recruit and activate PI3K, e.g., PDGF receptor, or by insulin receptor that indirectly recruits and activates PI3K via the closely associated docking proteins IRS1 and IRS2 (10).

The structural analyses described in the manuscript by Liu et al. (1) provide insight into the mechanism of aberrant oncogenic activation of members of class-I PI3Ks.

To expand the current understanding of the mechanism of oncogenic activation of PI3Kα at the molecular level, the authors employed cryo-EM to determine distinct conformational states and the dynamic nature of oncogenic PI3Kα mutants and compared their findings to their findings for WT PI3Kα. The cryo-EM map of the oncogenic PI3Kα H1047R mutant shows three structural differences which can explain the enhanced kinase activity of this mutant compared to WT PI3Kα. First, while residue H1047 in the WT enzyme stabilizes the activation loop of the kinase domain via a hydrogen bond interaction, mutation R1047H disrupts this interaction, releasing the activation loop and likely thereby simulating kinase activity. Second, in the cryo-EM map, the C-terminal tail of p110α in the PI3Kα H1047R mutant appears unstructured, indicating a high degree of flexibility which may facilitate interactions between the C-terminal tail and the cell membrane. Third, the authors used 3D variability analysis, as implemented in the cryo-EM data processing software cryoSPARC, to analyze the dynamics of the PI3Kα H1047R mutant and compare it to the dynamics of WT PI3Kα. These analyses show that the iSH2 and ABD domains of the PI3Kα H1047R mutant exhibit significantly increased mobility, weakening the inhibitory contacts between the regulatory p85α and catalytic p110α subunits. Taken together, these structural findings provide a potential explanation for how the H1047R mutation enhances kinase activity of PI3Kα.

The authors further determined the cryo-EM maps of the oncogenic mutants E542K and E545K of PI3Kα. In WT PI3Kα, negatively charged glutamate residues at positions 542 and 545 of the catalytic p110α subunit form salt bridges with positively charged residues of the regulatory p85α subunit. Mutations E542K and E545K disturb these salt bridges, fully disengaging p85α from p110α except for a contact between their iSH2 and ABD domains. Disengagement of p85α relieves an inhibitory constraint upon p110α, thus enhancing kinase activity. It is noteworthy that p110α is fully defined in the cryo-EM maps of PI3Kα E542K and E545K, whereas p85α is not visible in the maps due to a nonrigidly defined contact between the two subunits. Relative to p110α, p85α is highly flexible and scattered over countless conformations, essentially resulting in “averaging out” of the p85α subunit in the cryo-EM maps.

PI3Ks are enzymes that have evolved to act in proximity to their substrate at the cell membrane. Translocation of PI3K to the cell membrane is mediated via the SH2 domains of the regulatory p85 subunit through binding to specific binding sites composed of P-Tyr-containing sequences in the cytoplasmic domain of RTKs. Furthermore, p85 can also translocate to the cell membrane by binding to a family of membrane-attached docking proteins that are phosphorylated in response to activation of immune receptors, RTKs, cytokines, and other extracellular cues. These docking proteins are phosphorylated on multiple P-Tyr sites, which in turn triggers recruitment of several SH2-containing signaling molecules. The reduced dimensionality at the cell membrane enables direct association of p110 with the cell membrane to facilitate efficient enzymatic catalysis. Moreover, as p85 contains also an SH3 domain, it is likely that additional proteins with proline-containing SH3-specific binding sites may likewise form complexes with PI3K.

Comprehensive elucidation of the mechanisms underlying how the enzymatic activity of PI3K family members are regulated normally or by oncogenic gain-of-function mutations should consider the natural environment at the cell membrane and the multiple interactions that take place between PI3Ks with membrane receptors and with other signaling molecules. The information from the cryo-EM structures presented in the manuscript by Liu et al. underscores the importance of conformational dynamics of heterodimeric PI3K molecules and the importance of different conformational states of oncogenic PI3K mutants similarly as previously described (12, 13). Unfortunately, due to technical reasons, it is currently challenging to express stable heterodimeric PI3K molecules in amounts sufficient for structural studies unless the cultured cells used for protein expression are treated with the PI3K inhibitor BYL-719, or with other inhibitors. It is not clear whether the structural integrity of the PI3K complexes may be adversely affected by gradual dissociation of bound BYL-719 molecules, or how binding of BYL-719 may affect the conformational states and the dynamic properties of the PI3K complexes that play a role in PI3K activation. Clearly, the cryo-EM structures of the three oncogenic PI3K mutants described in the manuscript by Liu et al. provide an important step toward elucidating the mechanism of action of PI3Ks at a molecular level. The manuscript also provides structurally guided insights and a useful road map for the development of new selective therapeutics for treating cancers driven by oncogenic PI3K mutants (Fig. 1).

Fig. 1.

Cryo-EM structure of WT PI3Kα (PDB ID 7MYN). Left: Surface representation highlighting the catalytic p110α and regulatory p85α subunits. Right: Ribbon diagram highlighting the individual domains of p110α and p85α, the binding site for inhibitor BYL-719, and the location of three oncogenic mutations in p110α.