Machine learning model to predict oncologic outcomes for drugs in randomized clinical trials

Predicting oncologic outcome is challenging due to the diversity of cancer histologies and the complex network of underlying biological factors. In this study, we determine whether machine learning (ML) can extract meaningful associations between oncologic outcome and clinical trial, drug-related biomarker and molecular profile information. We analyzed therapeutic clinical trials corresponding to 1102 oncologic outcomes from 104 758 cancer patients with advanced colorectal adenocarcinoma, pancreatic adenocarcinoma, melanoma and nonsmall-cell lung cancer. For each intervention arm, a dataset with the following attributes was curated: line of treatment, the number of cytotoxic chemotherapies, small-molecule inhibitors, or monoclonal antibody agents, drug class, molecular alteration status of the clinical arm's population, cancer type, probability of drug sensitivity (PDS) (integrating the status of genomic, transcriptomic and proteomic biomarkers in the population of interest) and outcome. A total of 467 progression-free survival (PFS) and 369 overall survival (OS) data points were used as training sets to build our ML (random forest) model. Cross-validation sets were used for PFS and OS, obtaining correlation coefficients (r) of 0.82 and 0.70, respectively (outcome vs model's parameters). A total of 156 PFS and 110 OS data points were used as test sets. The Spearman correlation (r_s ) between predicted and actual outcomes was statistically significant (PFS: r_s = 0.879, OS: r_s = 0.878, P < .0001). The better outcome arm was predicted in 81% (PFS: N = 59/73, z = 5.24, P < .0001) and 71% (OS: N = 37/52, z = 2.91, P = .004) of randomized trials. The success of our algorithm to predict clinical outcome may be exploitable as a model to optimize clinical trial design with pharmaceutical agents.

Identify drug repurposing candidates by mining the protein data bank

Predicting off-targets by computational methods is gaining increasing interest in early-stage drug discovery. Here, we present a computational method based on full 3D comparisons of 3D structures. When a similar binding site is detected in the Protein Data Bank (PDB) (or any protein structure database), it is possible that the corresponding ligand also binds to that similar site. On one hand, this target hopping case is probably rare because it requires a high similarity between the binding sites. On the other hand, it could be a strong rational evidence to highlight possible off-target reactions and possibly a potential undesired side effect. This target-based drug repurposing can be extended a significant step further with the capability of searching the full surface of all proteins in the PDB, and therefore not relying on pocket detection. Using this approach, we describe how MED-SuMo reproduces the repurposing of tadalafil from PDE5A to PDE4A and a structure of PDE4A with tadalafil. Searching for local protein similarities generates more hits than for whole binding site similarities and therefore fragment repurposing is more likely to occur than for drug-sized compounds. In this work, we illustrate that by mining the PDB for proteins sharing similarities with the hinge region of protein kinases. The experimentally validated examples, biotin carboxylase and synapsin, are retrieved. Further to fragment repurposing, this approach can be applied to the detection of druggable sites from 3D structures. This is illustrated with detection of the protein kinase hinge motif in the HIV-RT non-nucleosidic allosteric site.

Chemoenzymatic syntheses of prenylated aromatic small molecules using Streptomyces prenyltransferases with relaxed substrate specificities

NphB is a soluble prenyltransferase from Streptomyces sp. strain CL190 that attaches a geranyl group to a 1,3,6,8-tetrahydroxynaphthalene-derived polyketide during the biosynthesis of anti-oxidant naphterpin. Here we report multiple chemoenzymatic syntheses of various prenylated compounds from aromatic substrates including flavonoids using two prenyltransferases NphB and SCO7190, a NphB homolog from Streptomyces coelicolor A3(2), as biocatalysts. NphB catalyzes carbon-carbon-based and carbon-oxygen-based geranylation of a diverse collection of hydroxyl-containing aromatic acceptors. Thus, this simple method using the prenyltransferases can be used to explore novel prenylated aromatic compounds with biological activities. Kinetic studies with NphB reveal that the prenylation reaction follows a sequential ordered mechanism.

A soluble, magnesium-independent prenyltransferase catalyzes reverse and regular C-prenylations and O-prenylations of aromatic substrates

Fnq26 from Streptomyces cinnamonensis DSM 1042 is a new member of the recently identified CloQ/Orf2 class of prenyltransferases. The enzyme was overexpressed in E. coli and purified to apparent homogeneity, resulting in a soluble, monomeric protein of 33.2 kDa. The catalytic activity of Fnq26 is independent of the presence of Mg(2+) or other divalent metal ions. With flaviolin (2,5,7-trihydroxy-1,4-naphthoquinone) as substrate, Fnq26 catalyzes the formation of a carbon-carbon-bond between C-3 (rather than C-1) of geranyl diphosphate and C-3 of flaviolin, i.e. an unusual "reverse" prenylation. With 1,3-dihydroxynaphthalene and 4-hydroxybenzoate as substrates Fnq26 catalyzes O-prenylations.

Structural basis for dual functionality of isoflavonoid O-methyltransferases in the evolution of plant defense responses

In leguminous plants such as pea (Pisum sativum), alfalfa (Medicago sativa), barrel medic (Medicago truncatula), and chickpea (Cicer arietinum), 4'-O-methylation of isoflavonoid natural products occurs early in the biosynthesis of defense chemicals known as phytoalexins. However, among these four species, only pea catalyzes 3-O-methylation that converts the pterocarpanoid isoflavonoid 6a-hydroxymaackiain to pisatin. In pea, pisatin is important for chemical resistance to the pathogenic fungus Nectria hematococca. While barrel medic does not biosynthesize 6a-hydroxymaackiain, when cell suspension cultures are fed 6a-hydroxymaackiain, they accumulate pisatin. In vitro, hydroxyisoflavanone 4'-O-methyltransferase (HI4'OMT) from barrel medic exhibits nearly identical steady state kinetic parameters for the 4'-O-methylation of the isoflavonoid intermediate 2,7,4'-trihydroxyisoflavanone and for the 3-O-methylation of the 6a-hydroxymaackiain isoflavonoid-derived pterocarpanoid intermediate found in pea. Protein x-ray crystal structures of HI4'OMT substrate complexes revealed identically bound conformations for the 2S,3R-stereoisomer of 2,7,4'-trihydroxyisoflavanone and the 6aR,11aR-stereoisomer of 6a-hydroxymaackiain. These results suggest how similar conformations intrinsic to seemingly distinct chemical substrates allowed leguminous plants to use homologous enzymes for two different biosynthetic reactions. The three-dimensional similarity of natural small molecules represents one explanation for how plants may rapidly recruit enzymes for new biosynthetic reactions in response to changing physiological and ecological pressures.

Structural basis for the design of potent and species-specific inhibitors of 3-hydroxy-3-methylglutaryl CoA synthases

3-Hydroxy-3-methylglutaryl CoA synthase (HMGS) catalyzes the first committed step in the mevalonate metabolic pathway for isoprenoid biosynthesis and serves as an alternative target for cholesterol-lowering and antibiotic drugs. We have determined a previously undescribed crystal structure of a eukaryotic HMGS bound covalently to a potent and specific inhibitor F-244 [(E,E)-11-[3-(hydroxymethyl)-4-oxo-2-oxytanyl]-3,5,7-trimethyl-2,4-undecadienenoic acid]. Given the accessibility of synthetic analogs of the F-244 natural product, this inhibited eukaryotic HMGS structure serves as a necessary starting point for structure-based methods that may improve the potency and species-specific selectivity of the next generation of F-244 analogs designed to target particular eukaryotic and prokaryotic HMGS.

Absence of substrate channeling between active sites in the Agrobacterium tumefaciens IspDF and IspE enzymes of the methyl erythritol phosphate pathway

The conversion of 2-C-methyl-d-erythritol 4-phosphate (MEP) to 2-C-methyl-d-erythritol 2,4-cyclodiphosphate (cMEDP) in the MEP entry into the isoprenoid biosynthetic pathway occurs in three consecutive steps catalyzed by the IspD, IspE, and IspF enzymes, respectively. In Agrobacterium tumefaciens the ispD and ispF genes are fused to encode a bifunctional enzyme that catalyzes the first (synthesis of 4-diphosphocytidyl-2-C-methyl d-erythritol) and third (synthesis of 2-C-methyl-d-erythritol 2,4-cyclodiphosphate) steps. Sedimentation velocity experiments indicate that the bifunctional IspDF enzyme and the IspE protein associate in solution, raising the possibility of substrate channeling among the active sites in these two proteins. Kinetic evidence for substrate channeling was sought by measuring the time courses for product formation during incubations of MEP, CTP, and ATP with the IspDF and IspE proteins with and without an excess of the inactive IspE(D152A) mutant in the presence or absence of 30% (v/v) glycerol. The time dependencies indicate that the enzyme-generated intermediates are not transferred from the IspD active site in IspDF to the active site of IspE or from the active site in IspE to the active site of the IspF module of IspDF.

Structural basis for the promiscuous biosynthetic prenylation of aromatic natural products

The anti-oxidant naphterpin is a natural product containing a polyketide-based aromatic core with an attached 10-carbon geranyl group derived from isoprenoid (terpene) metabolism. Hybrid natural products such as naphterpin that contain 5-carbon (dimethylallyl), 10-carbon (geranyl) or 15-carbon (farnesyl) isoprenoid chains possess biological activities distinct from their non-prenylated aromatic precursors. These hybrid natural products represent new anti-microbial, anti-oxidant, anti-inflammatory, anti-viral and anti-cancer compounds. A small number of aromatic prenyltransferases (PTases) responsible for prenyl group attachment have only recently been isolated and characterized. Here we report the gene identification, biochemical characterization and high-resolution X-ray crystal structures of an architecturally novel aromatic PTase, Orf2 from Streptomyces sp. strain CL190, with substrates and substrate analogues bound. In vivo, Orf2 attaches a geranyl group to a 1,3,6,8-tetrahydroxynaphthalene-derived polyketide during naphterpin biosynthesis. In vitro, Orf2 catalyses carbon-carbon-based and carbon-oxygen-based prenylation of a diverse collection of hydroxyl-containing aromatic acceptors of synthetic, microbial and plant origin. These crystal structures, coupled with in vitro assays, provide a basis for understanding and potentially manipulating the regio-specific prenylation of aromatic small molecules using this structurally unique family of aromatic PTases.

Kinetic analysis of Escherichia coli 2-C-methyl-D-erythritol-4-phosphate cytidyltransferase, wild type and mutants, reveals roles of active site amino acids

Escherichia coli 2-C-methyl-D-erythritol-4-phosphate cytidyltransferase (YgbP or IspD) catalyzes the conversion of 2-C-methyl-D-erythritol 4-phosphate (MEP) and cytidine triphosphate (CTP) to 4-diphosphocytidyl-2-C-methylerythritol (CDPME). Pulse chase experiments established that the reaction involves an ordered sequential mechanism with mandatory initial binding of CTP. On the basis of analysis of the previously reported crystal structures of apo-YgbP as well as YgbP complexed with both CTP.Mg(2+) and CDPME.Mg(2+) [Richard, S. B., Bowman, M. E., Kwiatkowski, W., Kang, I., Chow, C., Lillo, A. M., Cane, D. E., and Noel, J. P. (2001) Nat. Struct. Biol. 8, 641-648], a group of active site residues were selected for site-directed mutagenesis and steady-state kinetic analysis. Both Lys27 and Lys213 were shown to be essential to catalytic activity, consistent with their proposed role in stabilization of a pentacoordinate phosphate transition state resulting from in-line attack of the MEP phosphate on the alpha-phosphate of CTP. In addition, Thr140, Arg109, Asp106, and Thr165 were all shown to play critical roles in the binding and proper orientation of the MEP substrate.

The Oligomeric states of Haloarcula marismortui malate dehydrogenase are modulated by solvent components as shown by crystallographic and biochemical studies

The three-dimensional crystal structure of the (R207S, R292S) mutant of malate dehydrogenase from Haloarcula marismortui was solved at 1.95A resolution in order to determine the role of salt bridges and solvent ions in halophilic adaptation and quaternary structure stability. The mutations, located at the dimer-dimer interface, disrupt two inter-dimeric salt bridge clusters that are essential for wild-type tetramer stabilisation. Previous experiments in solution, performed on the double mutant, had shown a tetrameric structure in 4M NaCl, which dissociated into active dimers in 2M NaCl. In order to establish if the active dimeric form is a product of the mutation, or if it also exists in the wild-type protein, complementary studies were performed on the wild-type enzyme by analytical centrifugation and small angle neutron scattering experiments. They showed the existence of active dimers in NaF, KF, Na(2)SO(4), even in the absence of NADH, and in the presence of NADH at concentrations of NaCl below 0.3M. The crystal structure shows a tetramer that, in the absence of the salt bridge clusters, appears to be stabilized by a network of ordered water molecules and by Cl(-) binding at the dimer-dimer interface. The double mutant and wild-type dimer folds are essentially identical (the r.m.s. deviation between equivalent C(alpha) positions is 0.39A). Chloride ions are also observed at the monomer-monomer interfaces of the mutant, contributing to the stability of each dimer against low salt dissociation. Our results support the hypothesis that extensive binding of water and salt is an important feature of adaptation to a halophilic environment.

Structure and mechanism of 2-C-methyl-D-erythritol 2,4-cyclodiphosphate synthase. An enzyme in the mevalonate-independent isoprenoid biosynthetic pathway

The enzyme 2-C-methyl-D-erythritol 2,4-cyclodiphosphate (MECDP) synthase catalyzes the conversion of 4-diphosphocytidyl-2-C-methyl-D-erythritol 2-phosphate (CDP-ME2P) to MECDP, a highly unusual cyclodiphosphate-containing intermediate on the mevalonate-independent pathway to isopentenyl diphosphate and dimethylallyl diphosphate. We now report two x-ray crystal structures of MECDP synthase refined to 2.8-A resolution. The first structure contains a bound Mn(2+) cation, and the second structure contains CMP, MECDP, and Mn(2+). The protein adopts a homotrimeric quaternary structure built around a central hydrophobic cavity and three externally facing active sites. Each of these active sites is located between two adjacent monomers. A tetrahedrally arranged transition metal binding site, potentially occupied by Mn(2+), sits at the base of the active site cleft. A phosphate oxygen of MECDP and the side chains of Asp(8), His(10), and His(42) occupy the metal ion coordination sphere. These structures reveal for the first time the structural determinants underlying substrate, product, and Mn(2+) recognition and the likely catalytic mechanism accompanying the biosynthesis of the cyclodiphosphate-containing isoprenoid precursor, MECDP.

Structure of 4-diphosphocytidyl-2-C- methylerythritol synthetase involved in mevalonate- independent isoprenoid biosynthesis

The YgbP protein of Escherichia coli encodes the enzyme 4-diphosphocytidyl-2-C-methylerythritol (CDP-ME) synthetase, a member of the cytidyltransferase family of enzymes. CDP-ME is an intermediate in the mevalonate-independent pathway for isoprenoid biosynthesis in a number of prokaryotic organisms, algae, the plant plastids and the malaria parasite. Because vertebrates synthesize isoprenoid precursors using a mevalonate pathway, CDP-ME synthetase and other enzymes of the mevalonate-independent pathway for isoprenoid production represent attractive targets for the structure-based design of selective antibacterial, herbicidal and antimalarial drugs. The high-resolution structures of E. coli CDP-ME synthetase in the apo form and complexed with both CTP-Mg2+ and CDP-ME-Mg2+ reveal the stereochemical principles underlying both substrate and product recognition as well as catalysis in CDP-ME synthetase. Moreover, these complexes represent the first experimental structures for any cytidyltransferase with both substrates and products bound.

Insights into the molecular relationships between malate and lactate dehydrogenases: structural and biochemical properties of monomeric and dimeric intermediates of a mutant of tetrameric L-[LDH-like] malate dehydrogenase from the halophilic archaeon Haloarcula marismortui

L-Malate (MalDH) and L-lactate (LDH) dehydrogenases belong to the same family of NAD-dependent enzymes. LDHs are tetramers, whereas MalDHs can be either dimeric or tetrameric. To gain insight into molecular relationships between LDHs and MalDHs, we studied folding intermediates of a mutant of the LDH-like MalDH (a protein with LDH-like structure and MalDH enzymatic activity) from the halophilic archaeon Haloarcula marismortui (Hm MalDH). Crystallographic analysis of Hm MalDH had shown a tetramer made up of two dimers interacting mainly via complex salt bridge clusters. In the R207S/R292S Hm MalDH mutant, these salt bridges are disrupted. Its structural parameters, determined by neutron scattering and analytical centrifugation under different conditions, showed the protein to be a tetramer in 4 M NaCl. At lower salt concentrations, stable oligomeric intermediates could be trapped at a given pH, temperature, or NaCl solvent concentration. The spectroscopic properties and enzymatic behavior of monomeric, dimeric, and tetrameric species were thus characterized. The properties of the dimeric intermediate were compared to those of dimeric intermediates of LDH and dimeric MalDHs. A detailed analysis of the putative dimer-dimer contact regions in these enzymes provided an explanation of why some can form tetramers and others cannot. The study presented here makes Hm MalDH the best characterized example so far of an LDH-like MalDH.

Halophilic adaptation: novel solvent protein interactions observed in the 2.9 and 2.6 A resolution structures of the wild type and a mutant of malate dehydrogenase from Haloarcula marismortui

Previous biophysical studies of tetrameric malate dehydrogenase from the halophilic archaeon Haloarcula marismortui (Hm MalDH) have revealed the importance of protein-solvent interactions for its adaptation to molar salt conditions that strongly affect protein solubility, stability, and activity, in general. The structures of the E267R stability mutant of apo (-NADH) Hm MalDH determined to 2.6 A resolution and of apo (-NADH) wild type Hm MalDH determined to 2.9 A resolution, presented here, highlight a variety of novel protein-solvent features involved in halophilic adaptation. The tetramer appears to be stabilized by ordered water molecule networks and intersubunit complex salt bridges "locked" in by bound solvent chloride and sodium ions. The E267R mutation points into a central ordered water cavity, disrupting protein-solvent interactions. The analysis of the crystal structures showed that halophilic adaptation is not aimed uniquely at "protecting" the enzyme from the extreme salt conditions, as may have been expected, but, on the contrary, consists of mechanisms that harness the high ionic concentration in the environment.

Lymphedematous skin and subcutis: in vivo high resolution magnetic resonance imaging evaluation

Physico-chemical and morphologic parameters of skin layers and subcutaneous tissue in lymphedematous limb were studied in vivo using magnetic resonance imaging. High resolution images were obtained with a depth resolution of about 70 microm, using a specific surface gradient coil specially designed for skin imaging and connected to a standard whole-body imager at 1.5 T. Twenty-one patients with unilateral lower extremity lymphedema (11 primary and 10 secondary) were examined. Skin thickness, relaxation times, and relative proton density were calculated in lymphedematous limbs and in contralateral extremities. In diseased limbs, the average skin thickness (2.17 mm) was significantly larger (p = 1.5 x 10(-4)) than that of contralateral limb (1.14 mm). Major cutaneous alterations due to lymphedema took place in dermis. In lymphedematous dermis, the significant increase of relaxation time values could be due to a shift in the equilibrium of water inside this tissue in relation to the interactions between macromolecules and water molecules. In lymphedematous epidermis our results showed an increase in the number of free water protons. Information about water and fat distribution in lymphedema was also obtained using chemical shift weighted images. Our results demonstrated a water retention diffusely spread over the entire dermis, and an important fluid retention located in the interlobular spacing and beside the superficial fascia. Inside the subcutis, the mean thickness of the superficial fat lobules was increased more than that of the deep fat lobules. From all the various measurements we could not distinguish primary from secondary lymphedema.

In vivo high-resolution MR imaging of the skin in a whole-body system at 1.5 T

High-resolution magnetic resonance (MR) images of the skin were acquired with a whole-body MR system at 1.5 T by adding a specific imaging module: A saddle-shaped surface gradient coil was connected in place of one of the gradient coils of the system, and a surface radio-frequency coil with a 1.5-cm radius was placed at the center of the gradient coil. The images, acquired in 3 minutes 25 seconds, represent a field of view of 18 x 50 mm2, which corresponds to a pixel size of 70 x 390 micro2; the section thickness was 1.2 mm. With this spatial resolution, the different layers of the skin are clearly delineated: Epidermis appears as a high-signal-intensity layer, while dermis appears hypointense due to its very short T2. Pilosebaceous units appear as inclusions of epidermis inside dermis. The high quality of the images obtained enables in vivo MR characterization of skin.