Onur Dagliyan

Optimization Based Tumor Classification from Microarray Gene Expression Data

Background An important use of data obtained from microarray measurements is the classification of tumor types with respect to genes that are either up or down regulated in specific cancer types. A number of algorithms have been proposed to obtain such classifications. These algorithms usually require parameter optimization to obtain accurate results depending on the type of data. Additionally, it is highly critical to find an optimal set of markers among those up or down regulated genes that can be clinically utilized to build assays for the diagnosis or to follow progression of specific cancer types. In this paper, we employ a mixed integer programming based classification algorithm named hyper-box enclosure method (HBE) for the classification of some cancer types with a minimal set of predictor genes. This optimization based method which is a user friendly and efficient classifier may allow the clinicians to diagnose and follow progression of certain cancer types. Methodology/Principal Findings We apply HBE algorithm to some well known data sets such as leukemia, prostate cancer, diffuse large B-cell lymphoma (DLBCL), small round blue cell tumors (SRBCT) to find some predictor genes that can be utilized for diagnosis and prognosis in a robust manner with a high accuracy. Our approach does not require any modification or parameter optimization for each data set. Additionally, information gain attribute evaluator, relief attribute evaluator and correlation-based feature selection methods are employed for the gene selection. The results are compared with those from other studies and biological roles of selected genes in corresponding cancer type are described. Conclusions/Significance The performance of our algorithm overall was better than the other algorithms reported in the literature and classifiers found in WEKA data-mining package. Since it does not require a parameter optimization and it performs consistently very high prediction rate on different type of data sets, HBE method is an effective and consistent tool for cancer type prediction with a small number of gene markers.

Structural and dynamic determinants of protein-peptide recognition.

Protein-peptide interactions play important roles in many cellular processes, including signal transduction, trafficking, and immune recognition. Protein conformational changes upon binding, an ill-defined peptide binding surface, and the large number of peptide degrees of freedom make the prediction of protein-peptide interactions particularly challenging. To address these challenges, we perform rapid molecular dynamics simulations in order to examine the energetic and dynamic aspects of protein-peptide binding. We find that, in most cases, we recapitulate the native binding sites and native-like poses of protein-peptide complexes. Inclusion of electrostatic interactions in simulations significantly improves the prediction accuracy. Our results also highlight the importance of protein conformational flexibility, especially side-chain movement, which allows the peptide to optimize its conformation. Our findings not only demonstrate the importance of sufficient sampling of the protein and peptide conformations, but also reveal the possible effects of electrostatics and conformational flexibility on peptide recognition.

Rational design of a ligand-controlled protein conformational switch

Design of a regulatable multistate protein is a challenge for protein engineering. Here we design a protein with a unique topology, called uniRapR, whose conformation is controlled by the binding of a small molecule. We confirm switching and control ability of uniRapR in silico, in vitro, and in vivo. As a proof of concept, uniRapR is used as an artificial regulatory domain to control activity of kinases. By activating Src kinase using uniRapR in single cells and whole organism, we observe two unique phenotypes consistent with its role in metastasis. Activation of Src kinase leads to rapid induction of protrusion with polarized spreading in HeLa cells, and morphological changes with loss of cell–cell contacts in the epidermal tissue of zebrafish. The rational creation of uniRapR exemplifies the strength of computational protein design, and offers a powerful means for targeted activation of many pathways to study signaling in living organisms.

Engineering extrinsic disorder to control protein activity in living cells

Engineering control of cellular proteins The ability to switch proteins between active and inactive conformations can give insight into their function. Dagliyan et al. present a method to insert domains that control protein activity. They computationally identified protein loops that are coupled to the active site. Sensory domains inserted into these loops could modulate protein activity when their conformation was changed by light or ligand binding. The authors engineered domains into three different classes of proteins involved in cell signaling and found that switching the proteins between active and inactive states could control the shape and movement of living cells. Science, this issue p. 1441 Sensory domains inserted into identified allosteric sites modulate structural disorder to control proteins in cells. Optogenetic and chemogenetic control of proteins has revealed otherwise inaccessible facets of signaling dynamics. Here, we use light- or ligand-sensitive domains to modulate the structural disorder of diverse proteins, thereby generating robust allosteric switches. Sensory domains were inserted into nonconserved, surface-exposed loops that were tight and identified computationally as allosterically coupled to active sites. Allosteric switches introduced into motility signaling proteins (kinases, guanosine triphosphatases, and guanine exchange factors) controlled conversion between conformations closely resembling natural active and inactive states, as well as modulated the morphodynamics of living cells. Our results illustrate a broadly applicable approach to design physiological protein switches.

Knowledge-Based Design of a Biosensor to Quantify Localized ERK Activation in Living Cells

Investigation of protein activation in living cells is fundamental to understanding how proteins are influenced by the full complement of upstream regulators they experience. Here, we describe the generation ofxa0a biosensor based on the DARPin binding scaffoldxa0suited for intracellular applications. Combining library selection and knowledge-based design, we created an ERK activity biosensor by derivatizing a DARPin specific for phosphorylated ERK with a solvatochromatic merocyanine dye, whose fluorescence increases upon pERK binding. The biosensor specifically responded to pERK2, recognized by its conformation, but not to ERK2 or other closely related mitogen-activated kinases tested. Activated endogenous ERK was visualized in mouse embryo fibroblasts, revealing greater activation in the nucleus, perinuclear regions, and especially the nucleoli. The DARPin-based biosensor will serve as a useful tool for studying biological functions of ERK inxa0vitro and inxa0vivo.

Engineered kinase activation reveals unique morphodynamic phenotypes and associated trafficking for Src family isoforms

Significance Src family kinases (SFKs), critical in many aspects of homeostasis and disease, occur as multiple isoforms. It has been difficult to dissect the unique function of each isoform because their structures are so similar. Here we specifically activated each SFK isoform through insertion of an engineered domain. The domain caused the kinases to be catalytically inactive until they were reactivated by the small molecule rapamycin. Computational methods for quantifying dynamic changes in cell shape revealed that activation of each isoform produced dramatically different cell behaviors. Quantitative analysis showed that these behaviors correlated with specific patterns of subcellular trafficking, and depended on isoform acylation. The Src kinase family comprises nine homologous members whose distinct expression patterns and cellular distributions indicate that they have unique roles. These roles have not been determined because genetic manipulation has not produced clearly distinct phenotypes, and the kinases’ homology complicates generation of specific inhibitors. Through insertion of a modified FK506 binding protein (insertable FKBP12, iFKBP) into the protein kinase isoforms Fyn, Src, Lyn, and Yes, we engineered kinase analogs that can be activated within minutes in living cells (RapR analogs). Combining our RapR analogs with computational tools for quantifying and characterizing cellular dynamics, we demonstrate that Src family isoforms produce very different phenotypes, encompassing cell spreading, polarized motility, and production of long, thin cell extensions. Activation of Src and Fyn led to patterns of kinase translocation that correlated with morphological changes in temporally distinct stages. Phenotypes were dependent on N-terminal acylation, not on Src homology 3 (SH3) and Src homology 2 (SH2) domains, and correlated with movement between a perinuclear compartment, adhesions, and the plasma membrane.

Classification of cytochrome P450 inhibitors with respect to binding free energy and pIC50 using common molecular descriptors.

Virtual screening of chemical libraries following experimental assays of drug candidates is a common procedure in structure based drug discovery. However, the relationship between binding free energies and biological activities (pIC50) of drug candidates is still an unsolved issue that limits the efficiency and speed of drug development processes. In this study, the relationship between them is investigated based on a common molecular descriptor set for human cytochrome P450 enzymes (CYPs). CYPs play an important role in drug-drug interactions, drug metabolism, and toxicity. Therefore, in silico prediction of CYP inhibition by drug candidates is one of the major considerations in drug discovery. The combination of partial least-squares regression (PLSR) and a variety of classification algorithms were employed by considering this relationship as a classification problem. Our results indicate that PLSR with classification is a powerful tool to predict more than one output such as binding free energy and pIC50 simultaneously. PLSR with mixed-integer linear programming based hyperboxes predicts the binding free energy and pIC50 with a mean accuracy of 87.18% (min: 81.67% max: 97.05%) and 88.09% (min: 79.83% max: 92.90%), respectively, for the cytochrome p450 superfamily using the common 6 molecular descriptors with a 10-fold cross-validation.

Structure Based Discovery of Small Molecules to Regulate the Activity of Human Insulin Degrading Enzyme

Background Insulin-degrading enzyme (IDE) is an allosteric Zn+2 metalloprotease involved in the degradation of many peptides including amyloid-β, and insulin that play key roles in Alzheimers disease (AD) and type 2 diabetes mellitus (T2DM), respectively. Therefore, the use of therapeutic agents that regulate the activity of IDE would be a viable approach towards generating pharmaceutical treatments for these diseases. Crystal structure of IDE revealed that N-terminal has an exosite which is ∼30 Å away from the catalytic region and serves as a regulation site by orientation of the substrates of IDE to the catalytic site. It is possible to find small molecules that bind to the exosite of IDE and enhance its proteolytic activity towards different substrates. Methodology/Principal Findings In this study, we applied structure based drug design method combined with experimental methods to discover four novel molecules that enhance the activity of human IDE. The novel compounds, designated as D3, D4, D6, and D10 enhanced IDE mediated proteolysis of substrate V, insulin and amyloid-β, while enhanced degradation profiles were obtained towards substrate V and insulin in the presence of D10 only. Conclusion/Significance This paper describes the first examples of a computer-aided discovery of IDE regulators, showing that in vitro and in vivo activation of this important enzyme with small molecules is possible.

Identification of Novel Integrin Binding Partners for Calcium and Integrin Binding Protein 1 (CIB1): Structural and Thermodynamic Basis of CIB1 Promiscuity

The short cytoplasmic tails of the α- and β-chains of integrin adhesion receptors regulate integrin activation and cell signaling. Significantly less is known about proteins that bind to α-integrin cytoplasmic tails (CTs) as opposed to β-CTs to regulate integrins. Calcium and integrin binding protein 1 (CIB1) was previously identified as an αIIb binding partner that inhibits agonist-induced activation of the platelet-specific integrin, αIIbβ3. A sequence alignment of all α-integrin CTs revealed that key residues in the CIB1 binding site of αIIb are well-conserved, and was used to delineate a consensus binding site (I/L-x-x-x-L/M-W/Y-K-x-G-F-F). Because the CIB1 binding site of αIIb is conserved in all α-integrins and CIB1 expression is ubiquitous, we asked if CIB1 could interact with other α-integrin CTs. We predicted that multiple α-integrin CTs were capable of binding to the same hydrophobic binding pocket on CIB1 with docking models generated by all-atom replica exchange discrete molecular dynamics. After demonstrating novel in vivo interactions between CIB1 and other whole integrin complexes with co-immunoprecipitations, we validated the modeled predictions with solid-phase competitive binding assays, which showed that other α-integrin CTs compete with the αIIb CT for binding to CIB1 in vitro. Isothermal titration calorimetry measurements indicated that this binding is driven by hydrophobic interactions and depends on residues in the CIB1 consensus binding site. These new mechanistic details of CIB1-integrin binding imply that CIB1 could bind to all integrin complexes and act as a broad regulator of integrin function.

FRET binding antenna reports spatiotemporal dynamics of GDI–Cdc42 GTPase interactions

Guanine-nucleotide dissociation inhibitors (GDI) are negative regulators of Rho family GTPases that sequester the GTPases away from the membrane. Here we ask how GDI-Cdc42 interaction regulates localized Cdc42 activation for cell motility. The sensitivity of cells to overexpression of Rho family pathway components led us to a new biosensor design (GDI.Cdc42 FLARE), in which Cdc42 was modified with a FRET ‘binding antenna’ that selectively reported Cdc42 binding to endogenous GDI. Similar antennae could also report GDI-Rac1 and GDI-RhoA interaction. Through computational multiplexing and simultaneous imaging, we determined the spatiotemporal dynamics of GDI-Cdc42 interaction and Cdc42 activation during cell protrusion and retraction. This revealed a remarkably tight coordination of GTPase release and activation on a time scale of 10 seconds, suggesting that GDI-Cdc42 interactions are a critical component in the spatiotemporal regulation of Cdc42 activity, and not merely a mechanism for global sequestration of an inactivated pool of signaling molecules.