Final Report Summary - COMPUT DRUG DESIGN (Targeting the mutated PI3Ka isoform for the development of novel anti-cancer agents) The two aims of this project are a) to target and deactivate the mutated cancerous PI3Ka protein and b) to gain insights into how the mutated protein becomes over-activated in cancer by studying the mutation effects on PI3Ka structure and dynamics. PI3Ka is implicated in cancer and is involved in several cellular processes such as cell growth, division, and formation of new blood vessels (angiogenesis) that aid cancer cell survival. In cancer cells, PI3Ka is found to be altered in comparison to normal cells. These alterations in the protein, called mutations, are found in 27% of breast cancer patients, 24% of endometrial cancer patients, and 15% of colon cancer patients. The mutations of PI3Ka cause this protein to become overactivated and send signals for continuous replication of the cells, which results to tumor growth. 80% of these mutations are found in two hotspots on PI3Ka: an H1047R substitution close to the C-terminus and a cluster of three charge-reversal mutations (E542K, E545K, Q546K) at the interface between the catalytic and regulatory subunits of the protein. The first aim of this project is to design small molecules that can act as inhibitors of the mutated PI3Ka function. While several PI3Ka inhibitors have been reported in the literature, these molecules are non-selective because they target both the carcinogenic and wild-type (WT) PI3Ka proteins, causing undesirable side-effects. Selective inhibition of the mutated carcinogenic PI3Ka isoform is a promising strategy for the development of novel cancer therapeutics. Through this project, we aim to design and test new potential drugs that will selectively stop the action of the mutated protein PI3Ka found in cancer cells, while leaving the PI3Ka found in normal cells unaffected. Cancer patients, who bear these specific, cancerous PI3Ka mutations in their body as identified by genotyping, will directly benefit from such a treatment. This first aim of the project can be divided in two objectives. Objective 1 is to selectively inhibit the oncogenic E545K mutant of PI3Ka with small-molecules and the Objective 2 is to inhibit the H1047R PI3Ka mutant. To achieve these two objectives, we perform structure-based drug design and collaborate for the synthesis/purchase of potential leads, in vitro and in vivo assaying, and pharmacokinetic evaluation of potent compounds. Following the assays, potent compounds will be optimized with computational methods and re-assayed for enhanced activity.The second aim of this project (Objective 3) is to gain insights into the mechanism of overactication and effects of the mutations on protein structure and dynamics. Molecular Dynamics (MD) simulations have been performed in order to (a) aid our computer-aided drug design studies and (b) assess the effect of the mutants E545K and H1047R on protein structure and dynamics. Moreover, we performed Surface Plasmon Resonance (SPR) experiments in order to investigate the PI3Ka-membrane interactions and to assess the effect of the H1047R PI3Ka mutant on its interaction with the cell membrane. In order to achieve our goal of targeting the mutated PI3Ka protein, we employed a combination of computational and experimental techniques so as to target the H1047R and E545K mutants. Initially, Molecular Dynamics (MD) simulations of a) the WT protein, b) the mutated H1047R PI3Ka protein, c) the mutated E545K PI3Ka were performed. The simulations were analyzed in order to get the representative conformations of both WT and mutant proteins. These representative conformations were used to predict ligand binding sites, where small molecules could potentially bind and inhibit the action of the protein. Thus, binding cavities close to the mutation site were identified on the mutant proteins, which were not found in the WT PI3Ka. These cavities were considered as promising binding sites, which were then used to design mutant selective inhibitors. For the design of mutant selective inhibitors, we used state of the art computer-aided drug design techniques, such as virtual screening, de novo design and lead optimization methods. Approximately 50 compounds were bought and tested first with cell-free in vitro assays utilizing the WT and mutant proteins. Compounds that were found to selectively inhibit the mutant protein in cell-free in vitro assays were then administered to human breast cancer cells that contain the mutated protein as well to normal cancer cells. Our results showed that the chemical compounds that we designed selectively killed mutant human breast cancer cells, but not normal human cells. In view of these exciting results we subsequently aimed to increase the activity (potency) of our initially-designed compounds as well as improve their pharmacological properties (absorption, distribution, metabolism, excretion, and toxicity) by introducing several chemical modifications on the designed candidate drugs. The synthesized compounds were iteratively submitted to cell-free and cell-based assays in order to verify their potency. Finally, two compounds were administered in mouse xenografts to assess in vivo efficacy against the mutated PI3Ka. Xenografts were prepared using cell lines with: a) cancer cells, where PI3Ka is non-mutated and b) cancer cells bearing PI3Ka mutations. Each experiment was performed in 6-8 mice. In the first set, we administered as control plain saline to the xenografts, in the second set we administered the compounds in the xenografts that were grown using cancer cells, where PI3Ka is non-mutated, and lastly in the third set we administered the compounds in the xenografts bearing PI3Ka mutations. Following dosing, the effect on tumor growth was monitored by standard techniques such as palpitation and ultrasound and the compounds’ pharmacological properties was monitored using medicinal chemistry techniques in order to ensure high activity and biodistribution. Our studies showed that our compounds were able to shrink tumors harboring PI3Ka mutations selectively by ~30%, while leaving tumors with WT PI3Ka unaffected. Future plans include further optimization of these lead compounds to achieve better efficacies against mutated PI3Ka. In addition to discovering candidate drugs, we also employed computational and experimental techniques to gain insights into the mechanism of overactivation of the mutant H1047R and E545K proteins. Our results are in excellent agreement with experimental data and allow us to provide atomic-detail insights into the mechanism of mutant PI3Ka overactivation by monitoring structural and dynamical elements of the WT and mutant proteins.Investigation of the H1047R mutation overactivation:The crystal structure of the WT PI3Ka was published by Huang et al. in 2007 and of the mutant H1047R PI3Ka by Mandelker et al. in 2009. In the structure of the WT PI3Ka complex, His1047 is close to the C-terminal end of the activation loop. The His1047 is within H-bonding distance of the backbone carbonyl of Leu-956, a residue within the activation loop. The mutant arginine side chain of 1047 points 90° away from the position of its WT histidine counterpart and breaks the bond with Leu-956. Moreover, biochemical experiments suggested that 1047R points toward the cell membrane and its positively-charged side chain interacts with negatively-charged phospholipids. This interaction might facilitate the activation of the enzyme by enhancing the accessibility of the substrate to the binding site of the enzyme. In this study, we have modeled PI3Ka in order to gain insights into the overactivation mechanism of the commonly-expressed H1047R mutant through (MD) simulations (10 independent production runs x 100 ns) and Functional Mode Analysis (FMA) and have used SPR experiments to validate our results. Our results lead to a model of the overactivation mechanism of the commonly-expressed PIK3CA mutant H1047R based on structural and dynamical differences with its WT counterpart. We propose a self-inhibitory mechanism, whereby the C-terminus shields the DRH motif from performing futile ATP hydrolysis coupled with an enhanced ability of the mutant H1047R protein to bind to anionic lipids. Moreover, Functional Mode Analysis reveals that the H1047R mutation alters the twisting motion of the N-lobe of the kinase domain with respect to the C-lobe and shifts the position of the conserved P-loop residues in the vicinity of the active site. These findings demonstrate the dynamical and structural differences of the two proteins in atomic detail and propose a mechanism of overactivation for the mutant protein. Moreover, we performed MD simulations in aqueous solution for the wild type (WT) human, WT murine, and H1047R human mutant PI3Ka proteins bound to a known PI3Ka inhibitor, PIK-108, which is bound to the catalytic site as well as a novel non-ATP site. We verify the existence of the second binding site in the vicinity of the hotspot H1047R PI3Ka mutant through binding site identification, MD simulations, and principal component analysis (PCA). PIK-108 remains stable in both sites in all three variants throughout the course of the simulation. We demonstrate that the pose and interactions of PIK-108 in the catalytic site are similar in the case of murine WT and human mutant forms, while they are significantly different in the case of human WT. PIK-108 binding in the non-ATP pocket differs among the three variants, providing insights into the design of mutant-specific inhibitors. Finally, we assess the binding site implication in PI3Ka allostery in terms of its communication with the active site. Investigation of the E545K mutation overactivation:The major stabilizing interaction in PI3Ka involves the regulatory (p85a) and catalytic (p110a) domains. While a crystal structure of the E545K mutant does not exist, the interface between the regulator and catalytic subunits of PI3Ka has been resolved in the H1047R crystal structure. In this structure, the hotspot residue E545 (negative residue, catalytic domain) is bonded to K379 (positive residue, regulatory domain) through a salt bridge. Inhibition of PI3Ka activity is mediated by this positive-negative charge interaction that keeps the wild-type PI3Ka in a low activity state. Indeed, the E545K charge reversal mutation in the helical domain of the catalytic p110? is located exactly at the position where phosphopetides (pY)-peptides bind the regulatory subunit. Therefore, it has been suggested that E545K mutant increases lipid kinase activity by disrupting the p85a-p110a interface, mimicking the pY-activated state of the protein. Miled et al. (2007) have recently proposed that in this oncogenic charge-reversal mutation in the helical domain (E545K), the inhibitory interactions of the regulatory domain are abrogated, resulting in constitutive PI3Ka activation. Understanding the mechanism of mutational (E545K) activation will give insight into the early stages of the catalytic cycle of PI3K? and contribute toward designing novel mutant specific therapeutics. Therefore, we have performed MD simulations of the WT and E545K mutant proteins (8 independent simulations of 200 – 800 ns each) for the identification of early events and intermediate structures during the activation pathway of PI3Ka kinases. First we constructed the model of the E545K performing a single substitution of residue Glu545 to Lys. Using MD simulations we aim at characterizing the early stages of mutational activation that could involve local rearrangement (ns) and not major conformational changes that would probably require much longer time scales (µs). Indeed, in our simulations the overall structure of the dimer protein does not change significantly upon the E545K mutation. However, a closer look reveals that the intrinsic inter and intra domain plasticity of the system does change significantly upon introduction of the mutation. According to our results the disruption of the interface of the catalytic subunit (C2 domain) with the regulatory subunit (iSH2 domain), which results in a loss of communication between the two subunit as well as the increase in the inter-domain plasticity of the kinase domain (local deformation) could be the initial and early events involved in the mutational (E545K) over-activation of PI3Ka. Future plans include further analysis of the essential dynamics of the systems (PCA) and the inter-domain hydrogen bond network in order to support this finding.
