key: cord-0041992-i0ecxgus authors: nan title: Abstracts of publications related to QASR date: 2006-09-19 journal: nan DOI: 10.1002/qsar.19900090309 sha: 27175294e587d99e3fcd462857f0168e9d58474b doc_id: 41992 cord_uid: i0ecxgus nan tive Mechanisms p.2-25. Edited by Magee, P.S., Henry, D.R., Block, J.H., American Chemical Society, Washington, 1989. Results: An overview is given on the approaches for the discovery and design concepts of bioactive molecules: a) Natural products derived from plant extracts and their chemically modified derivatives (cardiac glycosides, atropine, cocaine, penicillins, cephalosporins, tetracyclines and actinomycins, pyrethrins and cyclosporin; b) Biochemically active molecules and their synthetic derivatives: acetylcholine, histamine, cortisonelhydrocortisone, indole-3-acetic acid (phenoxyacetic acid herbicides); c) Principles of selective toxicity is discussed exemplified by trimethoprimlmethotrexate, tetracyclines, acylovir, azidothymidine, antifungal agents; d) Metabolism of xenobiotics; e) Exploitation of secondary effects (serendipity); f) Receptor mapping; g) Quantitative structure-activity relationship studies; h) Empirical screening (shotgun approach). Results: Past and present of QSAR is overviewed: a) Historical roots; b) The role of QSAR models in rational drug design, together with a simplified diagram of the steps involved in drug development, including the place of QSAR investigations; c) Classification of QSAR models: structure-cryptic (property-activity) models, structure-implicit (quantum chemical) models, structure-explicit (structure-activity) and structure-graphics (computer graphics) models; d) A non-empirical QSAR model based on quantities introduced for identification of chemical structures, using Szymansk's and Randic's identification (ID) numbers, including applications for alkyltriazines. BioEssays, 1989, 11(5) , 136-141. Results: A review is given on recent observations about receptor structure and the dynamic nature of drug receptors and the significance of receptor dynamics for drug design: a) Receptors are classified according to structure and function (i) ion channels (nicotinic acetylcholine, GABA, Glycine); (ii) G protein linked [adrenergic ( c x ,~) , muscarinic acetylcholine, angiotensin, substance K, rhodopsin]; (iii) tyrosine kinase (insulin, IGF, EGF, PDGF); (iv) guanylate cyclase (atrial natriuretic peptide, speractin); b) Protein conformational changes can be best studied on allosteric proteins whose crystal structure is available (e.g. hemoglobin, aspartate transcarbamylase, tryptophan repressor) (no high resolution of a receptor structure is known); c) Receptor conformational changes can be studied by several indirect approaches (i) spectral properties of covalent or reversibly bound fluorescent reporter groups; (ii) the sensitivity of the receptor to various enzymes; (iii) the sedimentation of chromatographic properties of the receptor; the affinity of binding of radioligands; (iv) the functional state of the receptor; d) There are many unanswered questions: e.g. (i) are there relatively few conformational states for receptors with fluctuations around them or many stable conformational states; (ii) How can static structural information be used in drug design when multiple receptor conformations exist. Title: Designing molecules and crystals by computer. (Review) Author: Koide, A. IBM Japan Limited, Tokyo Scientific Center, Tokyo Research Laboratory 5-19 Sanban-cho, Chiyoda-ku, Tokyo 102, Japan. Source: IBM Systems Journal 1989, 28(4), 613 -627. Results: An overview is given on three computer aided design (CAD) systems developed by IBM Tokyo Scientific Center: a) Molecular Design Support System providing a strategic combination of simulation programs for industrial research and development optimizing computational time involved and the depth of the resulting information; b) MolWorld on IBM Personal Systems intended to create an intelligent visual environment for rapidly building energetically stable 3D molecular geometries for further simulation study; c) Molecular Orbital Graphics System designed to run on IBM mainframe computers offering highly interactive visualization environment for molecular electronic structures; d) The systems allow interactive data communication among the simulation programs for their strategically combined use; e) The structure and functions of MolWorld is illustrated on modeling the alanine molecule: (i) data model of molecular structures; (ii) chemical formula input; (iii) generation of 3D molecular structure; (iv) formulation of bonding model; (v) interactive molecular orbital graphics; (vi) methods of visualizing electronic structures; (vii) use of molecular orbital graphics for chemical reactions. Title: Interfacing statistics, quantum chemistry, and molecular modeling. (Review) Author: Magee, P.S. BIOSAR Research Project Vallejo CA 94591, USA. Source: ACS Symposium Series 1989, No.413 . In: Probing Bioactive Mechanisms p.37-56. Edited by Magee, P.S., Henry, D.R., Block, J.H., American Chemical Society, Washington, 1989. Results: A review is given on the application and overlap of quantum chemical, classical modeling and statistical approaches for the Quant. Struct.-Act. Relat. 9, 234 -293 (1990) Abstr. 225-228 235 understanding of binding events at the molecular level. A new com-a) QSAR of CNS drugs has been systematically discussed according plementary method called statistical docking experiment is also to the following classes: (i) general (nonspecific) CNS depressants: presented: general anesthetics; hypnotics and sedatives; (ii) general Insights obtained using energy-minimized structures; Activation in the bound state, types and energies of interactions at the receptor site and in crystal; Four successful examples (significant regression equations) are given for the modeling of binding events using physico-chemical descriptors and correlation analysis: (i) binding of a diverse set of pyridines to silica gel during thin-layer chromatography; (ii) binding of meta-substituted N-methyl-arylcarbamates to bovine erythrocyte AChE; (iii) binding of meta-substituted N-methyl-arylcarbamates to AChE obtained from susceptible and resistant green rice leafhoppers; (iv) activity of phenols inhibiting oxidative phosphorylation of ADP to ATP in yeast; A new statistical method for mapping of binding sites has been developed based on the hypermolecule approach, identifying key positions of binding and nature of the energy exchange between the hypermolecule atoms and the receptor site; Two examples are given on the successful application of statistical modeling (statistical docking experiment) based on the hypermolecule approach: (i) inhibition of housefly head AChE by metasubstituted N-methyl-arylcarbamates (n = 36, r = 0.841, s = 0.390, F = 25.82); (ii) inhibition of housefly head AChE by orthosubstituted N-methyl-arylcarbamates (n = 46, r = 0.829, s = 0.485, F = 14.24). (nonspecific) CNS stimulants; (iii) selective modifiers of CNS functions: anticonvulsants, antiparkinsonism drugs, analgetics and psychopharmacological agents; (iv) miscellaneous: drugs interacting with central a-adrenoreceptors, drugs interacting with histamine receptors, cholinergic and anticholinergic drugs; b) The review indicates that the fundamental property of the molecules which mostly influence the activity of CNS drugs is hydrophobicity (they have to pass the cell membrane and the bloodbrain barrier); c) Electronic parameters, indicative of dipole-dipole or charge-dipole interactions, charge-transfer phenomena, hydrogen-bond formation, are another important factor governing the activity of most CNS agents; d) Topographical, lipophylic and electronic structures of CNS pharmacophores are reviewed; e) 191 QSAR equations, 24 tables and 3 figures from 294 references are shown and discussed. The relevant template for each atom in the molecule is mapped into a bit array and the appropriate atomic position is marked; Volume comparisons (e.g. common volume or excluded volume) are made by bit-wise Boolean operations; The algorithm for the visualization of the molecular surface comprising the calculated van der Walls volume is given; Comparisons of CPU times required for the calculation of the van der Waals molecular volumes of various compounds using the methods of Stouch and Jurs, Pearlman, Gavazotti and the new method showed that similar or better results can be achieved using the new algorithm with VAX-class computers on molecules containing up to several hundred atoms. Abtstr. 229-232 Quant. Struct.-Act. Relat. 9, 234-293 (1990) One of the important goal of protein engineering is the design of isosteric analogues of proteins; Major software packages are available for molecular modeling are among others developed by (i) BioDesign, Inc., Pasadena, California; (ii) Biosym Technologies, San Diego, California; (iii) Tripos, St. Louis, Missouri; (iv) Polygen, Waltham, Massachusetts; (v) Chemical Design Ltd. Oxford; The molecular modelling packages use three basic parameters: (i) descriptive energy field; (ii)algorithm for performing molecular mechanics calculations; (iii) algorithm for performing molecular dynamics calculations; Modelling study of the binding events occurring between the envelop protein (gp120) of the AIDS (HIV) virus and its cellular receptor (CD4) protein supported the hypothesis that this domain was directly involved in binding the gp120 envelop protein leading to the design of conformationally restricted synthetic peptides binding to CD4. 19901229 Title: Finding Washington, 1989. Results: A new technique called "homology graphing" has been developed for the analysis of sequence-function relationships in proteins which can be used for sequence based drug design and the search for lead structures: a) As target protein is inhibited by the ligands of other proteins having sequence similarity, computer programs have been developed for the search of the similarity of proteins; b) Proteins are organized into hierarchical groups of families and superfamilies based on their global sequence similarities; c) Global sequence similarities were used to find inhibitors of acetolactate synthase (ALS) and resulted in a quinone derivative as a lead structure of new ALS inhibitors; d) Local sequence similarities of bacterial and mammal glutathione synthase (GSH) were used to find inhibitors of GSH; e) It was shown that the sequence segment of GSH was similar to dihydrofolate reductase (DHFR) is part of the ATP-binding site; f) Biological bases of local similarity between sequences of different proteins were indicated: molecular evolution of proteins and functionally important local regions; g) Homology graph, as a measure of sequence similarity was defined; h) Sequence-chemical structure relationship based on homology graph and the procedure to find lead structures was illustrated by an example resulting in a list of 33 potential inhibitors selected by the procedure based on the sequence segment from residue 150 to 210 of the sequence of tobacco ALS. Source: ACS Symposium Series 1989, No.413. In: Probing Bioactive Mechanisms p.198-214. Edited by Magee, P.S.. Henry, D.R., Block, J.H., American Chemical Society, Washington. 1989. Results: A review is given on the molecular design of the following major types of antifungal compound in relation to biochemistry, molecular modeling and target site fit: a) Squalene epoxidase inhibitors (allilamines and thiocarbanilates) blocking conversion of squalene 2,3-oxidosqualene; b) Inhibitors of sterol C-14 demethylation by cytochrome P-450 (piperazines pyridines, pyrimidines, imidazoles and triazoles); c) Inhibitors of sterol A' -t A' isornerization andlor sterol reductase inhibitors (morpholines); d) Benzimidazoles specifically interfering with the formation of microtubules and the activity phenylcarbamates on benzimidazole resistant strains; e) Carboxamides specifically blocking the membrane bound succinate ubiquinone oxidoreductase activity in the mitochondria1 electron transport chain in Basidiomycetes; f) Melanin biosynthesis inhibitors selectively interfering with the polyketide pathway to melanin in Pyricularia oryzae by blocking NADPH dependent reductase reactions of the pathway (fthalide, PCBA, chlobentiazone, tricyclazole, pyroquilon, PP389). Title: Quantitative modeling of soil sorption for xenobiotic chemicals. (Review) Author: Sabljic, A. Theoretical Chemistry Group, Department of Physical Chemistry, Institute Rudjer Boskovic HPOB 1016, YU-41001 Zagreb, Croatia, Yugoslavia. Source: Environ. Health Perspect. 1989, 83(2), 179 -190. Results: The environmental fate of organic pollutants depends strongly on their distribution between different environmental compartments. A review is given on modeling the soil sorption behavior of xenobiotic chemicals: a) Distribution of xenobiotic chemicals in the environment and principles of its statistical modeling; b) Quantitative structure-activity relationship (QSAR) models relating chemical, biological or environmental activity of the pollutants to their structural descriptors or physico-chemical properties such as logP values and water solubilities; c) Analysis of the QSAR existing models showed (i) low precision of water solubility and logP data; (ii) violations of some basic statistical laws; d) Molecular connectivity model has proved to be the most successful structural parameter modeling soil sorption; e) Highly significant linear regression equations are cited between k , values and the first order molecular connectivity index ( ' x ) of a wide range of organic pollutants such as polycyclic aromatic hydrocarbons (PAHs) and pesticides (organic phosphates, triazines, acetanilides, uracils, carbamates, etc.) with r values ranging from 0.976 to 0.986 and s values ranging from 0.202 to 0.300; f) The molecular connectivity model was extended by the addition of a single semiempirical variable (polarity correction factor) resulting in a highly significant linear regression equations between the calculated and measured KO, values of the total set of compounds (n = 215, r = 0.969, s = 0.279, F = 3291); g) Molecular surface areas and the polarity of the compounds were found to be responsible for the majority of the variance in the soil sorption data of a set of structurally diverse compounds. Title: Strategies for the use of computational SAR methods in assessing genotoxicity. (Review) Results: A review is given on the overall strategy and computational SAR methods for the evaluation of the potential health effects of chemicals. The main features of this strategy are discussed as follows: a) Generalized SAR model outlining the strategy of developing information for the structure-activity assessment of the potential biological effects of a chemical or a class of chemicals; b) Models for predicting health effects taking into account a multitude of possible mechanisms: c) Theoretical models for the mechanism of the key steps of differential activity at the molecular level; d) SAR strategies using linear-free energy methods such as the Hansch approach; e) Correlative SAR methods using multivariate techniques for descriptor generation and an empirical analysis of data sets with large number of variables (SIMCA, ADAPT, TOPKAT, CASE, etc.); f) Data base considerations describing three major peer-reviewed genetic toxicology data bases (i) National Toxicology Program (NTP) containing short term in vitro and in vivo genetic tests; (ii) data base developed by the EPA Gene-Tox Program containing 73 different short term bioassays for more than 4000 compounds, used in conjunction with ADAPT, CASE and TOPKAT; (iii) Genetic Activity Profile (GAP) in form of bar graphs displaying information on various tests using a given chemical. Title: Quantitative structure-activity relationships. Principles, and Authors: Benigni,, R.; Andreoli, C.; Giuliani, A. applications to mutagenicity and carcinogenicity. (Review) Laboratory of Toxicology and Ecotoxicology, Istituto Superiore di Sanita Rome, Italy. Source: Mutat. Res. 1989, 221(3), 197 -216. Results: Methods developed for the investigation for the relationships between structure and toxic effects of compounds are summarized: a) The extra-thermodynamic approach: the Hansch paradigm, physical chemical properties that influence biological activity and their parametrization, originality of the Hansch approach, receptors and pharmacophores: the natural content of the Hansch approach, predictive value of QSARs, a statistifa1 tool: multiple linear regression analysis, the problem of correlations among molecular descriptors, other mathematical utilizations of extrathermodynamic parameters; b) The substructural approach: when topological (substructural) descriptors are needed, how to use topological decriptors; c) QSAR in mutagenicity and carcinogenicity: general problems, specific versions of the substructural approach used for mutagenicity and carcinogenicity, applications to mutagenicity and carcinogenicity. Title: Linking structure and data. (Review) Author: Bawden, D. Source: Chem. Britain 1989, 25(Nov) , I107 -1108. Address not given. Results: The integration of information from different sources, particularly linking structural with non-structural information is an important consideration in chemical information technology. A review is given on integrated systems: a) SOCRATES chemicallbiological data system for chemical structure and substructure searching combined with the retrieval of biological and physicochemical data, compound availability, testing history, etc.; b) Psidom suite of PC based structure handling routines combining chemical structure with the retrieval of text and data; c) Cambridge Crystal Structure Databank on X-ray data of organic compounds integrating information on chemical structure, crystal conformation, numerical information on structure determination, bibliographic reference and keywording; d) computer aided organic synthesis for structure and substructure search, reaction retrieval, synthetic analysis and planning, stereochemical analysis, product prediction and thermal hazard analysis. Title: Determination of three-dimensional structures of proteins and nucleic acids in solution by nuclear magnetic resonance spectroscopy. Source: Critical Rev. Biochem. Mol. Biol. 1989, 24(5) , 479 -564. Results: A comprehensive review is given on the use of NMR spectroscopy for the determination of 3D structures of proteins and nucleic acids in solution discussing the following subjects: a) Theoretical basis of two-dimensional (2D) NMR and the Nuclear Overhauser Effect (NOE) measurements for the determination of 3D structures is given; b) Sequential resonance assignment for identifying spin systems of protein NMR spectra and nucleic acid spectra, selective isotope labeling for extension to larger systems and the use of site specific mutagenesis; c) Measurement and calculation of structural restraints of the molecules (i) interproton distances; (ii) torsion angle restrains; (iii) 4 backbone torsion angle restraints; (iv) side chain torsion angle restraints; (v) stereospecific assignments; (vi) dihedral angle restraints in nucleic acids; d) Determination of secondary structure in proteins; e) Determination of tertiary structure in proteins using (i) metric matrix distance geometry (ii) minimization in torsion angle space; (iii) restrained molecular dynamics; (iv) dynamical simulated annealing; (v) folding an extended strand by dynamical simulated annealing; (vi) hybrid metric matrix distance geometry-dynamical simulated annealing method; (vii) dynamical simulated annealing starting from a random array of atoms; f) Evaluation of the quality of structures generated from NMR data illustrated by studies for the structure determination of proteins and oligonucleotides using various algorithms and computer programs; g) Comparisons of solution and X-ray structures of (i) globular proteins; (ii) related proteins; (iii) nonglobular proteins and polypeptides; h) Evaluation of attainable precision of the determination of solution structures of proteins for which no X-ray structures exist (i) BDS-I (small 43-residue protein from the sea Anemone sulcata; (ii) Hirudin (small 65-residue protein from leech which is a potent natural inhibitor of coagulation); i) Structure determination by NMR is the starting point for the investigation of the dynamics of conformational changes upon ligand Abtstr. 236-239 Quant. Struct.-Act. Relat. 9, 234 -293 (1990) binding, unfolding kinetics, conformational equilibria between different conformational states, fast and slow internal dynamics and other phenomena. Title: ALADDIN. An integrated tool for computer-assisted molecular design and pharmacophore recognition from geometric, steric, and substructure searching of three-dimensional molecular structures. ( ALADDIN has the ability to (i) objectively describe receptor map hypothesis; (ii) scan a database to retrieve untested compounds which is predicted to be active by a receptor map hypothesis; (iii) quantitatively compare receptor map hypotheses for the same biological activity; (iv) design compounds that probe the bioactive conformation of a flexible ligand; (v) design new compounds that a receptor map hypothesis predicts to be active; (vi) design compounds based on structures from protein X-ray crystallography; A search made by ALADDIN in a database for molecules that should have D2 dopaminergic activity recognized unexpected D2 dopamine agonist activity of existing molecules; A comparison of two superposition rules for D2 agonists was performed by ALADDIN resulted in a clear discrimination between active and inactive compounds; A compound set was designed that match each of the three lowenergy conformations of dopamine resulting in novel active analogues of known compounds; Mimics of some sidc ~1 1 . 1 1 1 1 . 111 p.piide beta turns were designed, in order to demonstrate that ALADDIN can find small molecules that match a portion of a peptide chain and/or backbone; Results: Lately a number of chemical information systems based on three-dimensional (3-D) molecular structures have been developed and used in many laboratories: a) CONCORD uses empirical rules and simplified energy minimization to rapidly generate approximate but usually highly accurate 3-D molecular structures from chemical notation or molecular connection table input; b) Chemical Abstracts Service (CAS) has added 3-D coordinates for some 4 million organic substances to the CAS registry file; c) Cambridge Structural Database System contains X-ray and neutron diffraction crystal structures for tens of thousands of compounds; d) MACCS3D developed by Molecular Design Ltd., contains the standard MACCS-I structures to which additional 3-D data, such as Cartesian coordinates, partial atomic charges and molecular mechanics energy are added; MACCS3D allows exact match, geometric, submodel and substructure searching of 3-D models with geometric constrains specified to certain degree of tolerance; two 3-D databases are also available from Molecular Design that can be searched using MACCS3D [Drug Data Report (10,000 models) and Fine Chemicals Directory (90,000 models)]; e) ALADDIN (Daylight Chemical Information Systems) is also searches databases of 3-D structures to find compounds that meet biological, substructural and geometric criteria such as ranges of distances, angles defined by three points (dihedral angles) and plane angles that the geometric object must match. Aladdin is one of a number of menus working within the framework provided by daylight's chemical information system. Title: Improved access to supercomputers boosts chemical applica-Author: Borman, S. C&EN 1155 Sixteenth St., N.W., Washington DC 20036, USA. Source: C&EN 1989, 67(29) , 29-37. Results: Supercomputers have been much more accessible by scientists and engineers in the past few years in part as a result of the establishment of National Science Foundation (NSF) supercomputer centers. The most powerful class of supercomputers have program execution rates of 100 million to 1 billion floating-point operations per second, memory storage capacities of some ten million to 100 miltion computer words and a standard digital word size of 64 bits, the equivalent of about 15 decimal digits. The following examples are given for the use of supercomputer resources for chemical calculations and modeling: a) Modeling of key chromophores in the photosynthetic reaction center of Rhodopseudomonas viridis showing the heme group, the iron atom and the chlorophyll which absorbs light and causes rapid transfer of electron to pheophtin and then to the quinone; modeling includes a significant part of the protein having about 2000 atoms out of a total of some 12,000; Quant. Struct.-Act. Relat. 9, 234-293 (1990) Abstr. 240-242 239 b) Modeling of transition state of reaction between chloride and methyl chloride including electron clouds and water molecules surrounding the reaction site; c) Analysis of nucleic acid and protein sequences to evaluate the secondary structure of these biopolymers; d) Construction of a graphical image of hexafluoropropylene oxide dimer, a model for DuPont Krytox high performance lubricant; e) Calculation of the heats of formation of diaminobenzene isomers indicated that the target para isomer was 3 kcal/mol less stable then the meta isomer byproduct therefore the development for its large scale catalytic synthesis was not undertaken (saving was estimated to be $1 to $2 million). , b) Comparison of the newly defined Eo, parameter with the Taft-Kutter-Hansch E, (TKH E,) parameter showed characteristic steric effects of ortho-alkoxy and n-bonded planar type substituents (e.g. NO,, Ph); c) In various correlation analyses using retrospective data Eo, satisfactorily represented the steric effects of ortho-substituents on reactivity and biological activity of various organic compounds; d) Semi-empirical AM1 calculations using a hydrocarbon model to study the steric effects of a number of ortho-substituents resulted in the calculation of the ES value (difference in the heat of formation between ortho-substituted toluene and t-butylbenzene) which linearly correlated with the Eo, and the TKH E, parameters; e) Effects of di-ortho substitution on lipophilicity could be mostly expressed by the summed effect of the 2-and 6-position substituents; t) Highly significant regression equations were calculated for the pK, values of di-ortho-substituted benzoic acids using various substituent parameters; g) Quantitative analysis of the effect of ortho-substitution is difficult because it is a result of overlapping steric and electronic effects. Title: Calculation of partition coefficient of N-bridgehead com- ( I I ) is more lipophilic than propanolol-4-sulphate (IV)]. Fig. 1 shows the relationship between lipophilicity and pH for the compounds (circle represents (I), triangle (II), rhomboid (m) and square gv 234 -293 (1990) Abstr. 245-246 241 f (Rekker's constant, characterizing hydrophobicity). Results: A good agreement was found between the observed and calculated logP values of I1 (3.98 and 3.65, respectively) and for III. The hydrophobicity of I was found to be significantly lower than that of I1 (2.78 and 4.72, respectively) . The large deviation was attributed to the surface reduction as a result of condensed ring formation in I. Since interesting pharmacological activities have been reported for several derivatives of this type of compounds, the hydrophobicity of the unsubstituted 1 lH-indolo[3,2-c]quinoline has been calculated to be 2.22: (1) [interaction energy between a molecule and the binding site model was assumed to be the sum of its atomic contributions according to the expres-E , . , , ,~~(~) was the interaction energy parameter between the site region rand the atom-type of atom a and AG(b) was the total interaction energy for the binding mode b (binding mode was regarded as feasible when the molecule was in its energetically most favorable conformation)]. sion AG(b) = Erelion rEatomi a in r Er,typc(a). where Results: For development of the binding site model, first a simple geometry was proposed and AGm-5 AGm,calc I AGm+) was calculated for the whole set of compounds. If the calculated binding energy of any of the compounds was outside of the above boundary, the proposed site geometry was rejected and a more complex one was considered. This procedure had been repeated until all molecules in the set could be fitted within the experimental data range. As a result a 3D, five-region Voronoi binding site model has been developed for the PAHs containing a trigonal pyramid (rl) in the center and portions r2rs having infinite volumes and indicated by boundary planes. Region r, represented access to the solvent and regions r3rs were blocked for binding ( Fig. 1) : pyrene is shown in its optimal binding mode with its atom barely touching the boundary surfaces and edges: Calculations showed that benzene and other monoaromatic ring compounds should be very weak competitors for the B[a]P site. The model correctly predicted the binding energy of nine competitors outside of the training set. '% (Wiener index calculated as the sum of all unique shortest distances between atoms in the hydrogen suppressed graph of the compound); (Wiener index calculated as the sum of all geometric distances between atoms in the hydrogen suppressed molecule of the compound). Results: The traditional 2D Wiener number is defined as the sum of the lengths of all possible routes in the molecular graph. Here the length is proposed to be calculated as the real three-dimensional length between atoms: this is the 3D Wiener number. This number has many of the advantageous features of the related and very much studied 2D Wiener number. Additionally, it is highly discriminative and its use in quantitative structure-property relation studies (QSPR) appears to be encouraging, according to the preliminary calculations. Of these the most convincing is the set of statistical parameters for the linear correlation between the experimental and calculated enthalpy functions of 3DW the lower alkanes not shown here. Three different models have been tried and in all cases the 3D Wiener number seemed to be superior to the 2D one as it is reflected in (Eqs.1-6). a) GABA receptors in human mouse, rat and bovine brain tissues, membrane preparations and cellular uptake systems; b) GABA receptors in cat and rat spinal cord preparations; c) Cultured astrocytes. As in the equations nearly all indicator variables had negative regression coefficients it was concluded that instead of searching for better analogs, the research should be directed toward degradable pro-GABA or pro-muscimol derivatives that are efficiently taken up into the central nervous system (CNS). 0.585(+2.22) IRNG + 4.16 (3) Title: Synthesis and QSAR of 1-aryl-4-(~-2-quinolyI/l-isoqui-noly1ethyl)piperazines and some related compounds as hypotensive agents. Authors (1) Based on Eq. 1, an optimal logP is predicted (logPo = 4.23). The highest activity was produced by the 1-(3-methylphenyl)-4-(~-2-qui-Data determined: Chemical descriptors: Abtstr. 251-252 Quant. Struct.-Act. Relat. 9, 234 -293 (1990) nolylethyl) piperazine, its logP value being near to the optimal value (4.52). L.og(BPH) values calculated by Eq. 1 agree well with the observed ones. Source: Toxicology 1989, 58(2), 197 -210. Compounds: 3,5-Dimethoxyphenol, 4-chlorophenol, 2.6-dichlorophenol, 4-methyl-2-nitropheno1, 2,4dichlorophenol, 2,4,6-trichlorophenol, 2,3,4,5-tetrachlorophenol, 2,4,6-triiodophenol, pentachlorophenol. Biological material: Chinese hamster ovary (CHO) cells. Data taken from the literature: Ezoc; ECsoc; ECZOA; EC~OA [concentration (mmol1L) of the compound leading to a 20 or 50 % inhibition of the cell growth or adenosine uptake, respectively]. Data determined: EGO; ECSO [concentration (mmol1L) of the compound leading to a 20 or 50 % inhibition of the Na+/K+-ATPase activity, respectively]. Chemical descriptors: logP (logarithm of the partition coefficient in I-octanollwater); u (Hammett's constant, characterizing the electron-withdrawing power of the substituent); E, (Taft's constant, characterizing steric effects of the substituent); x (molecular connectivity index, calculated by Koch's method). Results: Highly significant linear relationships were calculated between log (ECzo) and logP (r = -0.963). the relationship between log(EC,,) and u being less good (r = -0.767). Combining the two parameters the relationship has improved (Eq. I): (1) (logarithm of the partition coefficient in I-octanollwater); (Hansch-Fujita's substituent constant characterizing hydrophobicity); (Hammett's constant, characterizing the electron-withdrawing power of the substituent); (STERIMOL steric parameter, characterizing the steric effect of the meta substituents); (RPLC derived hydrophobic substituent constant, defined by Chen and Horv6th, and extrapolated to 0 X methanol); (indicator variable 1 for the present 0 for the absence of hydrogen bonding substituents). Results: Logk' values were determined for the benzenesulfonamides and correlated with chemical descriptors. A highly significant linear relationship between logk' and logP was calculated (Eq. 1): ( pK, (negative logarithm of the acidic dissociation constant); logP (logarithm of the partition coefficient in I-octanol/water). Results: Relationships between Ki values and the chemical descriptors were investigated for CPZ and its listed metabolites. Relationship between log(l/Ki) and logP was calculated (Eq. 1) no numerical intercept (c) is given: ( In spite of the complexity of the full mechanism of inhibition involving at least six transition states and five distinct intermediates, a significant linear regression equation was calculated for Ki (Eq. 3): Since the crystal structure of the acyl-enzyme complex, the acylation and deacylation rate were available, it was concluded that the inhibition begins with the histidine 57 catalyzed attack of serine 195 0, at the benzoxazinone C4, while the carbonyl oxygen occupies the oxyanion hole formed by glycine 194 and serine 195. Title: Antifolate and antibacterial activities of 5-substituted Authors: Harris, N.V.; Smith, C.; Bowden, K. Rhone Results: It was shown earlier that binding of diaminoquinazolines to DHFR correlated with the torsional angle of the 4-amino group of the quinazoline nucleus. It was postulated that the interaction between the adjacent 5-substituent and the 4-amino group was very important in determining DHFR binding of the compounds possibly, because of the influence on the hydrogen-bond formed between the 4-amino group and a residue at the active site. The existence of such interaction in 5-substituted 2,4-diaminoquinazolines were shown by measuring a , , and 6~~1 values. The UI and UOR electronic parameters correlated well with chemical shifts of the 2-NH, groups (Eq. 1) but showed poor correlation for the 4-NH, group (Eq. 2), respectively: (1) The equations suggest that the through-ring resonance interactions between the 5-substituent and the adjacent 4-amino group are disrupted by some other effects which might have significance for binding. a) An extensive set of compounds based on the nalidixic acid structure of type I. where R', R3, R6 and R7 are various substituents; X6 and X* = C, N (for nalidixic acid: X6 = C, X8 = N, R' = Et, R' = COOH. R6 = H, R7 = Me); b) Subset of (I) (set A) containing fifty two 6,7-disubstituted 1 -alky l-1,4-dihydro-4-oxoquinoline-3-carboxylic acids; Compounds: Abtstr. 260-261 Quant. Struct.-Act. Relat. 9, 234-293 (1990) c) Subset of (I) (set B) containing one hundred and sixty two xylic acids; d) Subset of (I) (set C) containing eighty five 1,4-dihydr0-4-oxo-1,8-naphthyridine-3-carboxylic acids with substituted azetidinyl, pyrrolidinyl and piperidinyl rings at position 7, fluorine at position 6 and ethyl, vinyl or 2-fluoroethyl substituent at position 1. Biological material: Ps. aeruginosa V-1, E. coli NIHJ JC-2, S. The study showed that the most active compounds have fluorine in position 6, R7 can be a wide variety of nitrogen containing substituent and the best predictor for R7 is its lipophilicity. Compounds: 17 Phytoalexins: pisatin, 3,6a-dihydroxy-8,9-(methylenedioxy)pterocarpan, 6a, 1 la-dehydropisatin, 3-hydroxy-8,9-(methylenedioxy)-6a, 1 1 a-dehydropterocarpan, (*)-3-hydroxy-9-methoxypterocarpan, (+)-3-hydroxy-9-zmethoxypterocarpan, (-)-3-zhydroxy-9-methoxypterocarpan, vestitol, sativan, formonenetin, coumestrol, 4'-O-methylcoumestro1, phaseoilin, phaseollinisoflavan, 2'-methoxyphaseollin-isoflavan, glyceollin, 6a-11 a-dehydroglyceollin, tuberosin, 6a, 1 ladehydrotuberosin. (capacity factor determined RP-HPLC). calculated for logP of six reference compounds using their k' values (Eq. 1): (1) n = 6 r = 0.993 s not given F not given The lipophilicity of the phytoalexins were within the range of log P = 1.5 -4.2. It was found that the antifungal activity of similar compounds positively correlated with antifungal activity but no equation could be calculated for the whole set of compounds. It was suggested, however, that compounds with logP values higher than 3.5 were retained in the membranes, therefore phytoalexins with slightly lower lipophilicity, as well as greater fungitoxicity and systemic activity should be searched. Certain structural features seemed to correlate with antifungal activity such as the presence of phenolic OH and benzylic hydrogen. It was suggested that the ability of the ortho OH group to form fairly stable intramolecular hydrogen bond may contribute to the greater stability of the Shiff base hnctional group and the higher biological activity of the substances (various subsets required different equations). Results showed that compounds with increasing lipophilicity and electron donating substituents at the 3-and 5-positions have high inhibitory activity. I-[(3'-allyl-2'-hydroxybenzilidene)amino]-3-hydroxyguanidine was found to be the most active compound. The use of parameter focusing of the substituent hydrophobic constant and electronic constants was suggested for the selection of further substituents to design effective compounds. Biological material: a) Rabbits; b) Rats; c) Guinea pig. Data taken from the literature: analogue Results: PRP, ECJOH, ECSOB, ECSOT values were measured and presented for the C,, PAF analogue and compared with that of other analogues. C,, PAF analogue was less potent than the c 1 6 or CIS PAF analogues and equivalent to C,, PAF analogue, showing that the activity decreased with lipophilicity. A highly significant parabolic relationship was calculated between log(RPS) and Cf (Eq. 1): The maximum activity was calculated Cf = 6.78, this corresponds to the Cl6 PAF. (energy minimizatipn of the compounds were calculated using the free valence geometry energy minimization method); (Molecular shape analysis according to Hopfinger was used to quantitatively compare the shape similarity of analogs in their minimum energy conformer states (within 8 kcal/mol of their global minimum energy ( Fig. 1 shows the superposition of the reference conformations of the phenylalanine and tryptophane analogues). Quant. Struct.-Act. Relat. 9, 234-293 (1990) Chemical descriptors: logP (logarithm of the partition coefficient in 1 -octanol/water); (Hansch-Fujita's substituent constant characterizing hydrophobicity of a substituent on the aromatic ring and the hydrophobicity of the aromatic ring itself, respectively) ; [common overlap steric volumes (A3) between pairs of superimposed molecules in a common low energy conformation]; [dipole moment (Debeyes) of the whole molecule and of the aromatic ring, respectively, calculated using the CNDOI2 method] ; Quantum chemical indices (partial atomic charges calculated by the CNDOI2 method); 0 1 -0 4 [torsion angles (deg) (Fig. 1 ) rotated during the conformational analysis of the compounds]. Results: Significant parabolic regression equations were calculated for the antigelling activity of the phenylalanine and tryptophan analogues (Eq. 1 and Eq. 2, respectively): The different QSAR for the phenylalanine and tryptophan analogues indicated that they interact with hemoglobin in different ways or at different sites. For the phenylalanine analogues the hydrophobicity of the side chain, the aromatic dipole moment and the steric overlap volume explained about 50 %, 20 % and 10 % of the variance in antigelling activity, respectively. For the tryptophan analogues the square of the dipole moment or the steric overlap volume explained 70% or 60% of the variance in RA, respectively, being the two descriptors highly correlated. The results show that the tryptophan analogs have a relatively tight fit with the receptor site. Title: S-Aryl (tetramethyl) isothiouronium salts as possible antimicrobial agents, IV. In both Eq. 3 and Eq. 4, log(l/C) depended primarily on electronic factors (Eu' ) and only secondarily on hydrophobicity (CTobsd). A threshold logP value for the active isothiuronium salts was indicated, as the compounds with logP values between -0.70 and -1.58 were found to be totally inactive with the exception of the nitro-derivatives. Title: Comparative QSAR study of the chitin synthesis inhibitory activity of benzoyl-ureas versus benzoyl-biurets. Source: Tagungsbericht 1989, No.274 \ R* ponents explaining 69.61 %, 19.02 % and 9.30 % of the variance. Fig. 1 shows the minimum energy conformation of a highly active representative of the urea analogs (dimilin) with 5.6 A distance between the 1 and 15 carbon atoms. Fig. 2 shows the low energy conformation of the corresponding biuret analog with the two benzene rings in appro:imately the same plane and with the same C1-C18 distance (5.6 A) allowing to fit a hypothetical benzoylurea phamacophore. The similarity of the regression equations and the modelling study supported the hypothesis that the benzoylbiurets act by the same mechanism as the benzoylureas. Biological material: 8 Insect species: Aedes aegypti, Musca domestica, Chilo suppressalis, Hylemya platura, Oncopeltus suppressalis, Oncopeltus fasciatus, Pieris brassicae, Leptinotarsa decemlineata. [concentration of the benzoylurea derivative (various dimensions) required to kill 50% of insect larvae (A. aegypti, M. domestica, C. suppressalis, H. platura, 0. suppressalis, 0. fasciatus, P. brassicae or L. decemlineata]. Data determined: LCSO [concentration of the biuret analogue (ppm) required to kill 50% of insect larvae (A. aegypti or M. domestica]; Molecular modeling (models of the compounds were built using MOLIDEA); Conformational analysis (minimum energy conformations of the compounds were calculated using molecular mechanics method). Chemical descriptors: The thesis is devoted to the quantitative analysis of the uncoupling activity of substituted phenols using chemical descriptors in order to obtain further information on the mode of action of phenol uncouplers: The study of the partition coefficient of substituted phenols in liposomelwater system [P(L/W)] showed that (i) P(L/W) depended primarily on the logP value; (ii) influence of steric and electronic parameters depended on the type of the lipid involved; QSAR analysis of uncoupling phenols in rat-liver mitochondria identified the relevant physicochemical parameters required for phenols being protonophore in inner mitochondrial membrane and quantitatively separated the potency as the protonophore in the inner mitochondrial membrane and the incorporation factor (IogP); Protonophoric potency of substituted phenols was linearly related to uncoupling activity when certain critical physicochemical parameters of the experiment were taken into account; Linear relationship was calculated between uncoupling activities of substituted phenols and related uncouplers in the mitochondria from the flight muscles of house flies and in spinach chloroplasts; The results indicated a shuttle type mechanism for the uncoupling action of substituted phenols. Title: Uncoupling properties of a chlorophenol series on Acer cell 234 -293 (1990) Compounds: 22 Chlorinated phenols substituted with 2-C1, 3-C1, 2,4,5-Cl, 2,4,6-CI, pentachlorophenol, 4-CI-2-Me. 4-C1-3-Me, 4-C1-2,3-Me, 4-C1-3,5-Me, 4-CI-2-ally1, 4-C1-2-Pr-5-Me, 4Z1, 2,3-C1, 2,443, 2,5-C1, 2,6-Cl, 3,4-C1, 3,5-CI, 2,3,6-C1, 2-Cl-6-NO2, 2,4-C1-6-NO,, 2-CI-4,6-NO,. Biological material: Acer pseudoplatanus L. cell suspensions. Data determined: Dso [concentration of the compound (pmollL) required for 50 % uncoupling effect registered by measuring the oxygen consumption rate by polarography]; [minimal concentration of the compound (pnol/L) required for giving a full uncoupling effect]. Chemical descriptors: logP (logarithm of the partition coefficient in 1-octanol/water); MR (molar refractivity); ED (Steric parameter representing the perimeter of coplanary molecules projected onto the aromatic plane); A (Angular parameter expressing the hindrance in the neighborhood of the hydroxyl group in positions 2 and 6, respectively) ; UI, 02 (Hammett's constants, characterizing the electron-withdrawing power of the para-substituent and the ortho-or 4-nitro substituents, respectively). Results: Highly significant linear regression equations were calculated for the uncoupling effects of chlorophenols in Acer cell suspensions: The equations for the uncoupling effects in the whole cells and those calculated previously for isolated mitochondria or chloroplasts possess similar structures. 0.202(*2.049) a, -4.250 (2) Title: Effects of 3' substituents on diphenyl ether compounds. Results: SAR suggested that the space for the N' and NZ substituents in the PSI1 binding site is relatively large. The variation of the number of the carbon atoms of R2 on the photosynthetic inhibitory activity is shown in Fig. 1 (Hansch-Fujita's substituent constant characterizing hydrophobicity); Chemical descriptors: 0.32(*0.29) IOR + 0.41(&0.43) HB + 5.62 The biological activity of three out of the 30 (DPE-16, 19 and 28) substituted diphenyl esters were measured and listed. IGR values were measured for the three compounds and compared with that of A-23 and Methoprene. It was found that the position of acetamido group in the phenol moiety when it is in the ortho position Abtstr. 272-273 234 -293 (1990) increases the lipophilicity of the compound with a logP value of 2.54. If the same group is in mr para position, the logP values are 2.16 and 1.99, respectively and they are comparatively ineffective. When both the ortho positions are substituted with tertiary butyl groups (DPE-28) the logP value is relatively higher (3.30) which increases the lipophilicity of the compound and explains the pronounced IDR activity at relatively low concentrations. Abstr. Results: A highly significant linear regression equation was calculated for the descriptors of R' (R' = i-PrO was eliminated as an outlier) (Eq. 1): The compound with R' = EtO, R2 = Me and Z = 0 was found to be an effective, broad spectrum insecticide. The replacement of the quaternary carbon with a silicon atom cansimplify the synthesis of test compounds and thus can be advantageously utilized for the preparation of large compound sets for QSAR studies. The data suggest that the initial electron loss from the given compounds is the preeminent factor effecting the reaction rate. A single mechanism is suggested over the entire range of reactivities, where a transition state with a considerable positive charge is involved. Title: Connection models of structure and activity: II. Estimation of electronoacceptor and electronodonor functions of active centers in the molecules of physiologically active materials. Research Institute of Physiology Active Materials Chernogolovka, Moskow district, USSR. Engl. summary). Authors Chemical descriptors: logP (logarithm of hydrophobicity). Results: Calculations for electronoacceptor and electronodonor entharpic and free energy factors on the base of functional groups were made according to the principle of independence of active centers: Data determined: Linear correlation was found between the calculated and measured characteristics: The accuracy of the fitting was the same as the measurement error of AH,,, and AGm.The entropy might be calculated from enthalpy, Gibbs energy and temperature: The good linear correlations between the measured and calculated data show that the functional group approaches might be used for these compound types. The substituent effects for the a-acceptorlr-donor substituents (F, C1, Br, I) were found to be very much larger for the C6FsR relative to the nitrobenzenes. These results indicate that the extra electron enters a o*-orbital, which is localized on the C-R atoms. for the structure-solubility relationship of aliphatic alcohols. The study indicated that solubility of aliphatic alcohols depends primarily on molecular connectivity ('x), the number of carbon atoms in the alkyl chain (n'), the number of hydrogens on the a-carbon atom (normal, iso, secondary, ternary) and the degree of branching (VG): (1) n not given r not given s not given F not given Eq. 1 was found to be a highly significant predictor of S (Eq. 2): The result support Kier's, furthermore Kier and Hall's earlier models on the structural dependence of water solubility of alcohols. -lOg(S) = 113 'X + (113)* SG -2.5075 Title: Linear free energy relationships for peroxy radical-phenol reactions. Influence of the para-substituent, the orthodi-tert-butyl groups and the peroxy radical. k (reaction rate constant (M -'s -I ) of the reaction between the reaction of cumyl-, 1 -phenylethyl-and t-butyl-peroxy radicals and ortho-para-substituted phenol inhibitors). Data taken from the literature: Chemical descriptors: u+ R. UI, UR (Charton's electronic substituent constant and its decomposition to inductive and resonance components, respectively for the characterization of the para substituent); (indicator variable 1 for the presence or absence of the t-Bu groups in 2,6-position of the phenols). Results: Highly significant linear regression equations were calculated by stepwise regression analysis for logK in spite of the diverse data set originating from different laboratories using different peroxy radicals (Eq. 1, Eq. 2): Quant. Struct.-Act. Relat. 9, 234 -293 (1990) (2) n = 32 r = 0.848 s = 0.432 F = 37.2 I C~" was not selected by stepwise regression indicating that the orthodi-t-Bu substitution had no significant effect on the rate of hydrogen abstraction from phenols by the radicals. The form of the equations for different subsets of the phenols and radicals indicated that the reaction mechanism was the same for the different peroxy radicals. Title: A fractal study of aliphatic compounds. A quantitative structure-property correlation through topological indices and bulk parameters. The following descriptors are considered as 'bulk parameters': VW (van der Waals volume, calculated from the van der Waals radii of the atoms); MW (molecular weight); SD (steric density of the functional group). Results: Highly significant equations are presented for calculating VW, SD and Mw r values ranging from 0.92 to 1.00, other statistics and the number of investigations are not given. Q and 0 values calculated by these equations were introduced to the equation given above and the physicochemical properties were calculated. The observed and calculated IogV,, D and p values are presented and compared for the alkanes, alcohols, acids and nitriles. The observed and calculated physicochemical parameters agreed well. Fractal nature of the alkyl chain length was discussed and a relationship was presented between the fractal-dimensioned alkyl chain length and a generalized topological index. Title: Application of micellar liquid chromatography to modeling of organic compounds by quantitative structure-activity relationships. Chemical descriptors: logP (logarithm of the partition coefficient in 1-octanol/water). Results: In a series of experiment with the listed compounds micellar liquid chromatography has been applied to model hydrophobicity of organic compounds in a biological system. The measured logk' values of the substituted benzenes were found to be superior predictors of logP. Fig. 1 shows the plot of logP versus logk' of the substituted benzenes. Highly significant correlation was calculated for the logk' values of phenols (open squares) (n = 6, r = 0.985), for the rest of the compounds (full squares) (n = 16, r = 0.990) and for the entire set (n = 22, r = 0.922). Further experiments using various surfactant types in the mobil phase suggested that logk' values generated on a lamellar phase may be better predictors of hydrophilicity than logP obtained from binary solvent systems. Title: Isoxazolinyldioxepins. 2. The partitioning characteristics and the complexing ability of some oxazolinyldioxepin diastereoisomers. Authors Quant. Struct.-Act. Relat. 9, 234 -293 (1990) Source: J. Chem. Soc. Perkin Trans. I1 1989 . No. 11, 1935 -1937 Compounds: 10 Oxazolinyldioxepin derivatives of type I and 11, where X = H, F, CI, CF3 or CH3. Data determined: logk' [logarithm of the capacity factor, measured by reversed-phase liquid chromatography (RPLC)]; MEP (molecular electrostatic computed by Geesner-Prettre and Pullman's VSSPOT procedure). Chemical descriptor: logP (logarithm of the partition coefficient in 1-octanollwater). Results: The logk' and logP values were measured for the two type of diastereomers and a highly significant linear relationship between logk' and logP was presented (r = 0.995): The MEPs of I and 11's F-derivatives were determined and presented, "a" for type I, "b" for type II: The complex forming ability of the diastereoisomers with mono-cations was investigated and explained in terms of the structures and electronic properties of the compounds. Results: Linear relationships are presented plotting y versus n for the 9 hydrophobic sorbents (Fig. 1 ) and the slopes of these straight lines are suggested for experimental determination of . q, values. ~0 values determined by the suggested method are listed. While no linear relationships were found between Kd and n, y depend linearly on n for the test compounds [alkanols (I). alkane diols (2) Results: Three linear models were fittedwith independent variabIes of log(P), MR and O x . The best fitting parameters (independent of composition) were obtained from the following models (no statistical characteristics is presented): (1) (2) The two types of correlations (with structural and with moving phase parameters) together might be used for the optimization of chromatographic separation of complex mixtures of sulphur-containing substances. (zero order molecular bonding type connectivity index); Ig(K) = a0 + a1 P' + a2 logP + a3 P' logP Ig(K) = a0 + a1 tg(cM) + a, logP + a3 tg(cM) logP -3-1 19901285 The Kd values derived by the suggested method were compared by Kd values calculated by Martin's rule and a good agreement was found. Title: Mathematical description of the chromatographic behaviour of isosorbide esters separated by thin layer chromatography. Compounds: 9 isosorbide esters: isosorbide (l), 1-5-monoacetate, 1-2-monoacetate, 1-5-mononitrate, 1-2-mononitrate, l-diacetate, 1-5-nitro-2-acetate, 1-2-nitro-5-acetate, l-dinitrate. Rn, R~I [retention factors obtained by thin-layer chromatography in benzene/ethylacetate/isopropanol/ (7:3: 1.5) and in dichloromethane/diisopropylether/isopropanol (20:4:2: 1) eluent systems, respectively]. Data determined: Chemical descriptors: (information index, based on the distribution of the elements in the topological distance matrix); (the geometrical analogue); (Randic connectivity index); (maximum geometric distance in the molecule); Compounds: 46 highly diverse chemicals. Grouped according to the following properties: contains (ester or amide or anhydride) or (heterocyclic N) or (0 bound to C) or (unbranched alkyl group with greater than 4 carbons). Data determined: AERUD Chemical descriptors: (aerobic ultimate degradation in receiving waters). 2 v x 4x, nCI M, (molecular weight). Results: The paper has aimed at developing a model for predicting AERUD. The data sets were collected from 22 biodegradation experts. The experts estimated the biodegradationtime that might be required for AERUD on the time scales of days, weeks, months and longer. 46 (valence second order molecular connectivity index); (fourth order path/cluster connectivity index); (number of covalently bound chlorine atoms); highly diverse chemicals but typical in wastewater treatment systems were examined. Zero to six order molecular and cluster connectivity indexes were calculated using computer programs wrinen in FOR-TRAN for IBM PC/XT. The best fitted linear regression model is: [first order rate constant: transport or transformation parameter (mol/Pa. h)]. Results: The QWASI fugacity model describes the fate of a (contaminating) chemical, such as organo-chlorine compounds, pesticides or metals. The lake model consists of water, bottom and suspended sediments, and air. The model includes the following processes: advective flow, volatilization, sediment deposition, resuspension and burial, sediment-water diffusion, wet and dry atmospheric deposition, and degrading reactions. The steady state solution of the model is illustrated by application to PCBs in Lake Ontario using the equilibrium criterion of fugacity as the variable controlling environmental fate of the chemical. The applications are based upon inaccurate data. Use of fugacity is inappropriate for involatile chemicals, such as metals, or ionic species, because fugacities are calculated from a basis of vapor phase concentrations. For these materials the use of the equilibrium concentration activity is more appropriate since activities are calculated from a water phase base. Thus, a new equilibrium criterion, termed the "aquivalent" concentration (or equivalent aqueous concentration) is suggested as being preferable. This concentration has the advantage of being applicable in all phases, such as water, air and sediments. The formalism developed in the QWASI approach can also be applied, making possible a ready comparison of the relative rates (and thus, the importance) of diverse environmental fate processes. All these are illustrated by applying the model on a steady state basis to Quant. Struct.-Act. Relat. 9, 234-293 (1990) Abstr. 288-289 261 the PCB example and to the fate of lead in Lake Ontario. The estimated and observed concentrations of PCBs and lead in Lake Ontario agree well: the largest difference in the case of PCBs in rain amounts to a factor of three. In other phases, and especially in the case of lead, the difference is usually less than 30 per cent. Although in order to judge the biological effects of a contaminant it is of fundamental importance to know its transport and transformations, and the present model has been proven to useful to describe this; direct biological implications are not deduced at the present stage. The similar slopes of the equations show that these compounds exert their cytotoxicity primarily by alkylation. While the majority of the tested compounds showed no hypoxia-selective cytotoxicity (ratio awa 1 .O), the 4-N02 and 3-NO2 substituted compounds were more toxic to UV4 cells under hypoxic conditions (ratio = 3.2 for the compound with R = 4-N02), indicating cellular reduction of the nitro-group. The measured hypoxic selectivity of the 3-No2 and 4-N0, substituted compounds was a fraction of the calculated ratio (measured 220 fold and calculated 3500 fold by Eq. 2 between the 4-N02 and 4-NH, substituted compounds). The main reason for the difference between the calculated and measured hypoxic selectivity is suggested to be the low reduction potential of the 4-N02 and 3-NO, groups (E = -500 mV and E = -470 mV, respectively). 19901288 Title: Quantitative structure-activity relationships for the cytotoxici- (Hammett's constant, characterizing the electron-withdrawing power of the substituent); (Hammett's polar electronic constant characterizing the electron withdrawing power of the substituent for anilines). Results: Significant linear regression equations were calculated for the halflife (T1/2), growth inhibition (150) and clonogenicity data (CTlo) using Hammett constants (Eq. 1, Eq. 2, Eq. 3): (1) n = 11 r = 0.96 s = 0.24 F not given 234 -293 (1990) type, test animals, the mean level of toxicity and the form of the equation. E.g. analysis of the toxicity of phenols showed a transition between simple dependence from logP to exclusive dependence to reactivity factors indicating two separate classes of phenol toxicity (Eq. 1 for mouse i.p. toxicity, and Eq. 2 for rat oral toxicity): Results: An additivity model, pLC50 = Cni A Ti 4-To, where ni is the number of ith substituents in a benzene derivative, AT; is the toxicity contribution of the ilh substituent and To is the toxicity of the parent compound (benzene), was used for predicting toxicity of LO Similar correlation was found between mutagenicity and u (Fig. 2) indicating that both biochemical and chemical processes involve a pH dependent nucleophilic ring opening (including the protonation of the aziridin nitrogen as rate controlling step) and in.,uenced by electronic and steric factors (equation not given). resonance effect). (electron density on N1 in the HOMO calculated by the MNDO). Results: Highly significant linear relationships between log( 1 IC) and logp, &HOMO (Eq. 1); IogP, qHOMO (Eq. 2) are presented indicating that the more hydrophobic and more electron-rich triazines are more active according to the Ames test: Substructures [a total of 32355 fragments were generated from the 189 compounds using the program CASE (Computer-Automated Structure Evaluation) system]. Results: A comparative classification of the compounds were performed using CASE for identifying molecular fragments associated with cancerogenic activity (biophores) as well as deactivating fragments (biophobes). CASE identified 2 1 biophores and 2 biophobes from the 32355 fragments of the 189 compounds with a less than 12.5% probability of being associated with carcinogenicity as a chance. The sensitivity and specificity of the analysis was unexpectedly high: 1.00 and 0.86, respectively. The predictive power of CASE Biological material: Chemical descriptors: was tested using the identified biophores and biophobes on a group of chemicals not present in the data base. The ability of CASE to correctly predict carcinogens and presumed non-carcinogens was found to be very good. It was suggested that non-genotoxic carcinogens may act by a broader mechanism rather than being chemical specific. Compounds: 9 compounds of type I where R = H, CH3, C,H5, CzH7, C3H7, CzH9, C4H9, C3H5, C4H7, SC4H7; 3 compounds of t Y P I1 where R = H, C~HEOH, SC4H80H. Data determined: p t pi" (a priori probability of appearance of the i-th active compound); (a priori probability of appearance of the i-th nonactive compound). (the first order molecular connectivity index); (the second order molecular connectivity index); (information-theoretic index on graph distances calculated by the Wiener index according to Gutmann and Platt); Chemical descriptors: (rank of smell, where the rank is defined to equal with one for the most active compound). Results: The authors' previously proposed structure-activity relationship approach was applied for structure-odor relationship. 13 different compounds of groups I and I1 were examined using the topological indices W, R, I, x as independent variables and v as the dependent variable. The best correlation was obtained between R and v ( Fig. I) Results: LogP and IogP, values were determined for the nitroimidazole derivatives. Significant linear equations were calculated, the best one related for logP and logP,r (Eq. 1): (1) logP = 1.14 lOgP,I + 0.37 n = 9 r = 0.92 s not given F not given Chemical descriptors: logP Descriptors (logarithm of the partition coefficient in I-octanoll water); (12 indicator variables taking the value of 1 for the presence of cr/p-hydroxy , 6A-fluoro, 6A-methy1, 9Afluoro, 17-hydroxy, I6A-fluoro. 16,17-acetonide, 21-deoxy, 21-acetate, 21-propionate, 2 I-butyrate or 2 1-isobutyrate, respectively). Results: A data set of 43 steroids were compiled after removing those ones containing unique substituents. The set was divided into two categories of approximately equal membership by defining a threshold logP value of 1.45. A descriptor set was created and the non-significant ones were eliminated using the weight-sign change feature selection technique. Linear leaning machine was applied to calculate the weight vectors and complete convergence was achieved in the training procedure. The predictive ability of the linear pattern classifier thus obtained was tested using the leave one out procedure. The predictive ability was found to be 8 I .4 %. The predictive ability of the approach was found to be good and improvement was expected with larger data set. Steroids, however, containing new substituents would have to be subjected to a repeated pattern-recognition calculation. lengthlbreadth descriptors (2 descriptors)]. Results: For modeling the shape of the compounds, SIMCA was used: the approach was to generate disjoint principal models of clustered points in a multidimensional space. The number of clusters for each structure was determined by using hierarchical cluster analysis. Fig. 1 shows the orthogonal views of a schematic representation of the SIM-CA models for the atom clusters in senecionine: Each compound in turn was used as a reference structure. Every other structure was superimposed on the reference using the ends of the corresponding binding moment vector plus the ring nitrogen atom. Canonical correlation analysis was used for calculating the correlation between the five biological activity data and shape descriptors of 21 structures. The best correlation was observed for Jurs' shadow descriptors. The MSA and SIMCA descriptors were comparable. The model was able to express both the amount and direction of shape the differences, and also for encoding relevant information for correlation with the biological activity. Compounds: 6 N-substituted 3-methyl-4-nitropyrazole-5-carboxamides (11), 6 N-substituted 4-amino-3-methylpyrazole-5-carboxamides (III), 14 N-substituted 3-methyl-4-diazopyrazole-5-carboxamides and N-piperidiny 1-N-( 1,3-dimethyl-4-nitrosopyrazol-5-yl)-urea (VII) . Title: Structure-activity correlations for psychotomimetics. 1. Phenylalkylamines: electronic, volume, and hydrophobicity parameters. Abtstr. 300 Quant. Struct.-Act. Relat. 9, 234-293 (1990) Data determined: EDso Conformational analysis g [dose of the compound (mg/kg) which causes 50% of the rats which were trained on 1 rngfkg reference compound to respond as they would to the training drug]; (geometries of the compounds were calculated using MMF2 from starting geometries determined by the program EUCLID). Discriminant analysis resulted in a function containing six variables which misclassified only one compound in the training set. When the data was repeatedly split randomty into a training and a test sb, the misclassification rate was 9% (15 out of 161 classifications). Fig. 2 shows the plot of the two canonical varieties from discriminant analysis visualizing the separation of hallucinogenic and nonhallucinogenic derivatives (meaning of symbols are the same as in Fig. 1 ). Multiple regression analysis (MRA) was found to be the most useful for identifying relevant and discarding redundant variables. Highly significant parabolic regression equations were calculated for the human activity data (A) (n = 50, r ranging from 0.9004 to 0.9563, F not given) and for animal data (EDSO) (n = 16, r = 0.8679 and r = 0.9825, F not given). Eight descriptors were found to be highly significant. Among these the importance of directional hydrophobicity and volume effects indicated that steric and hydrophobic interactions participate in the interaction with the receptor.MRA indicated a strong interaction between the meta-and para-substituents and the presence of the formation of charge transfer complex by accepting charge. Data did not support the hypothesis that the human activity data and animal ( 1 ) Data taken from the literature: Sweet(n) (sweet taste of the compound, where n represent the number of times a sample has to be diluted to match the taste of 3% sucrose solution). [class fit distances of a compound to sweet and nonsweet class (dimension not given) calculated by principal component analysis]. Chemical descriptors: MR (molar refractivity); BI, L (STERIMOL steric parameters, characterizing the steric effect of the substituent); r (Hansch-Fujita's substituent constant characterizing hydrophobicity); urn, up (Hammett's constants, characterizing the electron-withdrawing power of the substituent in meta-and para-position, respectively). Results: No statistically significant regression equation was obtained by the Hansch-Fujita approach using the chemical descriptors listed. D', D2 Quant. Stact.-Act. Relat. 9, 234 -293 (1990) Abstr. 301-302 267 Principal component analysis of the data set extracted 2 principal components, explaining 64 % of the variance of the sweet compounds. The sweet compounds clustered in a relatively confined region of the 15D space whereas the tasteless and bitter compounds were scattered around the sweet compounds. A Coomans plot, however., indicated, when plotting D' versus D2, that sweet and nonsweet compounds could be well separated along the D' axis ( Fig. 1, Title: Conformation of cyclopeptides. Factor analysis. A convenient tool for simplifying conformational studies of condensed poly-ring systems. Prolyl-type cyclopeptides. Authors conformations of the six-membered DOP-ring family may be reproduced by means of a superposition of the canonical twist (T), boat (B) and chair (C) forms. Physically, the coefficients have the meaning of relative contributions (amplitudes) of the T, B and C forms into the total conformation of the ring. Here factor analysis (FA) and principal component analysis was used in conformational studies of 30 various X-ray conformers of DOP/PYR. A correspondence was found between factors identified and RPT, when the rings are considered separately. This fact allows a physical interpretation of the FA results: two or three puckering variables were found for the DOP and PYR rings expressing the absolute amplitudes of the basic pucker modes. Subse-quent FA treatment of the condensed system revealed five conformational variables necessary and sufficicnt to describe the twolring puckering completely. Each of the basic pucker modes defines a unique pattern of conformational variation of the whole two-ring system. The results demonstrate that FA is a powerful technique in analysing condensed poly-ring systems, not amenable to the RPT treatment. Title: Preprocessing, variable selection, and classification rules in the application of SIMCA pattern recognition to mass spectral data. Authors: Dunn m*, W.J.; Emery, S.L.; Glen, G.W; Scott, D.R. College of Pharmacy, The University of Illinois at Chicago 833 South Wood, Chicago IL 60612, USA. Source: Environ. Sci. Technol. 1989, 23(12) , 1499-1505. Compounds: A diverse set of 121 compounds observed in ambient air classified as (1) nonhalogenated benzenes; (2) chlorine containing compounds; (3) bromo-and bromochloro compounds; (4) aliphatic hydrocarbons; (5) miscellaneous oxygen-containing hyhocarbon-like compounds (aliphatic alcohols, aldehydes and ketones). pattern recognition was applied to autocorrelation-transformed mass spectra of the compounds using providing chemical class assignment for an unknown]; m/z;, mlz,, m/zl (first three principal components scores of SIMCA). Results: SIMCA pattern recognition method was applied on a training set of 78 toxic compounds targeted for routine monitoring in ambient air. The analysis resulted in very good classification and identification of the compounds (87 % and 84 %, respectively). However, the training procedure proved to be inadequate as a number hydrocarbons from field samples (GCIMS analysis) were incorrectly classified as chlorocarbons. A new approaches for the preprocessing (scaling the MS data by taking the square root of the intensitiesfollowed by autocorrelation transform), variable selection (only the 16 most intense ions in the MS spectrum were taken), and for the classification rules of SIMCA has been introduced to improve results on real data. Fig, 1 and As a result of the revised rules the classification performance has been greatly improved for field data (97 -94 %). Title: A QSAR model for the estimation of carcinogenicity. 234-293 (1990) It was suggested that the mechanism of mutagenicity of the CIMeB[a]Ps measured in the Ames test is probably more complex than the simple reactivity of carbocation intermediates. [dipole interaction potential (dimension not given)]; [molecular electrostatic potential (kcallmol) in a plane]; [molecular electrostatic field map (kcall mol), mapping the 1E(R)J values of a molecule surface in a plane, predicting the directions and energies of the interactions with small polar molecules at distances greater than the van der Waals sphere]; (construction of surfaces corresponding to a given value of potential); 3D MEP and MEF maps (3D maps weregenerated by superimposing the equipotential curves corresponding to a value of 20 kcallmol in the case of MEP. and 1.5 kcallmol in the case of MEF, computed in several planes perpendicular to the mean plane of the analogues in low energy conformations, stacking over each other in 1 A distance). Results: The three vasopressin analogues differ significantly in their biological activities. Both MEP and MEF maps of the of the biologically active (Mpa')-AVP and (Cpp')-AVP are similar, but they are different from that of the inactive (Ths')-AVP. Fig. 1, Fig. 2 Abtstr. 306-307 Quant. Struct.-Act. Relat. 9, 234-293 (1990) A new method for calculating the points of the equipotential curves was also presented. Crystal structure (crystal coordinates of the molecules were determined by X-ray diffraction methods). Data taken from the literature: [electrostatic molecular potential (eV) were calculated using AM-1 type semiempirical MO calculations]; Conformational analysis [minimum energy conformations were calculated using X-ray structures as input geometries followed by AM 1-method (Fletcher-Powell algorithm)]. Chemical descriptors: UI, u2 [rotational angles of the N-ally1 group (deg)]. Results: Four similar energy minima were located by AM-I calculations for both NAMH+ and NLPH+. The energy minima for the protonated NAM + and NLPH + were the most populated ones with conformational enantiomers relative to the involved N-allyl-piperidine moiety (37 % and 44 %, respectively). It was shown that the isopotential curve localization of EMP contour maps were very similar for the corresponding conformations of both NLPH * and NAMH + indicating that both molecules should interact a the same anionic sites of the opioid receptor, ( p morphine receptor). Fig. 2 and Fig. 3 shows the EMP contour maps of NAMH' and NLPH + , respectively, in their preferred conformations: Compounds: Esfenvalerate (SS and SR isomers) of type I, 3-phenoxybenzyl 2-(4-ethoxyphenyl)-3,3,3-trifluoropropyl ether (R Quant. Struct.-Act. Relat. 9, 234-293 (1990) Abstr. 308 271 isomer) (11). a-cyano-3-phenoxybenzyl 2-(4-chlorophenyl)-2-methylpropionate (S isomer) (III) and deltamethrin (IV). " CN Data determined: Conformational analysis (minimum energy conformations of the compounds in vacuum were calculated using AM1 molecular orbital method and Broyden-Fletcher-Goldfarb-Shanno method integrated into MOPAC); (root mean square, indicating the goodness of fit between two conformers in 3D); (logarithm of the partition coefficient in I-octanol/water estimated using CLOGP program); [heat of formation of the most stable conformer (kcallmol)]. RMS logP E Chemical descriptors: 0 1 -0 6 Results: It was assumed that the 3D positions of the benzene rings of the pyrethroids are decisive for good insecticidal activity. The lower energy conformers of (I) (SS and SR isomers), (11) (R isomer), (111) (S isomer) and deltamethrin (IV) were compared by superimposition. Inspite of their opposite configuration, esfenvalerate (I) (SS isomer) and the new type pyrethroid I1 (R isomer) were reasonably superimposed, indicating that the positions of the benzene rings in space are important and the bonds between them are not directly determinant (Fig. 1) Crystal structure (X-ray crystal coordinates of penicillopepsin was obtained from the Protein Data Bank). Data determined: Electrostatic potential [electrostatic potential of the protein atoms (kcallmol) is calculated using the partial charges in the AMBER united atom force field); Docking (The DOCK program was used to find molecules that have a good geometric fit to the receptor). Results: A second generation computer-assisted drug design method has been developed utilizing a rapid and automatic algorithm of locating sterically reasonable orientations of small molecules in a receptor site of known 3D structure. It includes also a scoring scheme ranking the orientations by how well the compounds fit the receptor site. In the first step a large database (Cambridge Crystallographic Database) is searched for small molecules with shapes complementary to the receptor structure. The second step is a docking procedure investigating the electrostatic and hydrogen bonding properties of the receptor displayed by the MIDAS graphics package. The steps of the design procedure is given. The algorithm includes a simple scoring function approximating a soft van der Waals potential summing up the interaction between the receptor and ligand atoms. Directional hydrogen bonding is localized using electrostatic potential of the receptor at contact points with the substrate. The shape search of (I) was described in detail. A new method has been developed for the construction of a hypothetical active site (HASL), and the estimation of the binding of potential inhibitors to this site. The molecules were quantitatively compared to one another through the use of their HASL representations. After repeated fitting one molecule lattice to another, they were merged to form a composite lattice reflecting spatial and atomic requirements of all the molecules simultaneously. The total pKi value of an inhibitor was divided to additive values among its lattice points presumed to account for the binding of every part of the molecule. Using an iterative method, a self consistent mathematical model was produced distributing the partial pKi values of the training set in a predicting manner in the lattice. The HASL model could be used quantitatively and predictively model enzyme-inhibitor interaction. A lattice resolution of 2 -3 A was found to be optimal. A learning set of 37 E. coli DHFR inhibitors were chosen to test the predictive power of the HASL model at various resolutions. Binding predictions (pKi values) were calculated for the entire inhibitor set at each resolution and plotted separately for the learning and test set members at 2.8 A resolution (Fig. I) : . Ala-). Data determined: Molecular models (3D structure of the molecules have been constructed and displayed using the program GEOM communicating with Cambridge X-ray data bank, Brookhaven protein data bank, Sandoz X-ray data bank SYBYL and DISMAN; Quant. Struct.-Act. Relat. 9, 234-293 (1990) Abstr. 311-312 273 Distance geometry [nuclear Overhauser enhancements (NOE) and spin-spin coupling constants were measured by 2D NMR methods, semiempirically calibrated as proton-proton distance (A) and dihedral angle (deg) constrains and used in distance geometry calculations (DISMAN) andlor in restrained molecular dynamics calculations to determine 3D structure of molecules in solution]; Crystal structure (atomic coordinates of the compounds were determined by X-ray crystallography); RMS [root mean square deviation (A) of the corresponding atoms of two superimposed molecular structures]. Chemical descriptors: Results: Distance geometry calculations were carried out using GEOM and DISMAN, to identify all conformations of the compounds in solution which were consistent with experimental data obtained by NOE measurements. The application of GEOM was demonstrated by modelling cycbsporin A with and without a limited set of H-bond constrains and with a full NMR data set. In case of cyclosporin A, 100 randomly generated linear analogues of the cyclic structure were formed from the monomers. Geometric cyclization was achieved using DISMAN, resulting in many different but stereochemically correct conformations of cyclosporin A. Superposition of the backbones Of the 10 best cyclic conformers showed RMS deviations between 1.8 A and 3.1 A. Fig. 1 shows the superposition of a DISMAN generated ring conformation (thick line) with its X-ray structure of cyclosporin A (thin line) with H-bond constraints (RMS = 1.25 A): Fig.1 37 Distance and 4 dihedrL-angle constraints have been extracted from NOE and vicinal coupling data and used to generate the conformation and the cyclization conditions of the hexapeptide (Fig. 2) (position of residual distance violations and their direction is shown by arrows): Although the described method is not exhaustive, it explores a much greater variety of initial structures than had been previously possible. Title: A new MODEL parameter set for @-lactams. Authors: Durkin, K.A.; Sherrod, M.J.; Liotta*, D. Department of Chemistry, Emory University Atlanta GL 30322, USA. Source: J. Org. Chem. 1989, 54(25) , 5839-5841. Compounds: 22 @Lactam antibiotics of diverse structure. Data taken from the literature: Crystal structures (crystal coordinates of the p-lactams were determined using X-ray diffraction method). Results: Superposition of the X-ray structures and the calculated geometries of 8-lactams using the original parameter set in the MM2 force field in MODEL gave satisfactory rms values. A lack of planarity of the 8-lactam ring and significant differences in the calculated bond lengths and anglesaround the &lactam nitrogen were detected, however. = 0, s, so, SO,]. In order to improve fit, a new atom type with new parameters has been developed for the P-lactam nitrogen (wild atom type 6 0 coded with symbol 2 2 in MODEL). The new parameters were evaluated by comparison of the calculated and X-ray geometries of the 22 0-lactams. Using the new parameter set, the X-ray data were satisfactorily reproduced except for the sulfone 8-lactams. It was indicated that the AMPAC data were not suitable for the sulfones as the hypervalent sulfur compounds are not well described in the AM1 Hamiltonian. An additional parameters was, however, derived giving good structural data unrelated to the AMPAC information. It is not known which the new parameter sets is the best for the sulfone /3-lactams. Title: A molecular modelling study of the interaction of Compounds Noradrenalin. Biological material: a) cDNA of the hamster lung 0,-adrenergic receptor and &-adrenergic receptor kinase; b) Bacterio-ovine-and bovine-rhodopsin and rhodopsin kinase. Protein primary sequence (amino acid sequence of the hamster lung P,-adrenergic receptor has been deduced by cloning the gene and the cDNA of the hamster lung &adrenergic receptor);, (The COSMIC molecular modeling program was used for modeling a-helices in a hydrophobic environment using p and w torsion angles of -59" and -44", respectively, according to Blundell et al.); (the two highest lying occupied and the two lowest lying unoccupied orbitals, respectively, calculated using INDO molecular orbital calculation); Crystal structure (crystal coordinates of noradrenaline has been determined by X-ray diffractometry); Conformation analysis (minimum energy conformation of the &-adrenergic receptor model has been calculated using molecular mechanics method). Results: Strong experimental evidences suggested that rhodopsin and 8,-adrenergic receptor had similar secondary structure. Thus, it was assumed, that similarly to bacterio-ovine-and bovine-rhodopsins, D2-Adr ener gic receptor -. . B2-adrenergic receptor possesses a structure consisting of seven ahelices traversing the cell membrane. Fig. 1 shows the postulated arrangements of the a-helices of rhodopsin and the &-receptor. Using the experimental data, a model of the &-adrenergic receptor has been generated for the study of its interaction with noradrenaline. A possible binding site was created. Successful docking indicated that HOMO and LUMO orbitals contributed to the binding in a chargetransfer interaction between Trp-109 and noradrenaline. A hydrogen bond was detected between the threonine residue of the model receptor and the noradrenaline side chain hydroxyl explaining why chirality was found to be important for the activity of adrenergic substances. Title: Three-dimensional steric molecular modeling of the [binding affinity (nM) of the compounds to the 5-HT3 receptor]. Data determined: Molecular modeling (3D molecular models of each 19 compound were made using CAMSEQlM Molecular Modeling System); [distance (A) from the center of the aromatic ring to the ring-embedded nitrogen, when the nitrogen is placed in the same plane as the aromatic ring]. Results: In order to derive rules for the 5-HT3 pharmacophore, a molecular graphics-based analysis was made using six core structures. The structures were aligned so as to overlay the aromatic rings and to place the ring embedded nitrogen atom in the same plane as the aromatic ring. Nine steric rules were derived from the analysis common to all 19 potent 5-HT3 agents. Fig. 1 shows the 3D representation of the six overlaid 5-HT3 core structures using CAMSEQIM: The 5-HT3 inactivity of atropine could be explained because its steric properties differed from those the active ICS 205-930 only by a single atom and failed to meet two of the nine hypothetical criteria. UV-visible spectra [spectrophotometric studies of mixtures of the dyes and nicotine in 10% (vlv) aqueous ethanol mixture at 29"CI. Results: Cyanine dyes demonstrate a multitude of biological activities which may be due to the interference of the adsorbed dye molecule on active sites of the living cell. It was shown by UV-and visible spectrophotometry that the hydroxy styryl cyanine dyes and nicotine formed 1 : 1 charge-transfer complexes. The absorption band of the complex formed between nicotine and dye was detected at wavelengths longer than those of the individual pure substances having identical concentrations to those in mixture. Fig. 1 shows that the two partially positive centres of the dye (2-(2-hydroxystyryl)-pyridinium-I-ethyliodide) were located at a similar distance than the two nitrogen atoms of pyridine or pyrrolidinyl moieties of nicotine allowing the suggested 1: I parallel stucking interaction between the two molecule: Molecular modeling (200 Conformations were calculated using a distance geometry algorithm and energy minimized by a modified MM2 force field in MOLEDIT). Results: All conformers within 5 kcallmol of the lowest energy conformer were superposed on the X-ray structure of MK-329. Crystal structure (crystal coordinates of the proteins were determined using X-ray diffraction method). Data taken from the literature: was less than 1 A); (probability that a given tetrapeptide sequence is superimposable on the ribonuclease A structure); [probability that the ith residue (amino acid) will occur in the jth conformational state of the tetrapeptide which is superimposable to ribonuclease A]. Results: It was suggested that the five tetrapeptides were essential components of larger peptides and might be responsible for their biological activity (binding to the CD4 receptor). Earlier it was hypothesized that the critical tetrapeptide located in a segment of ribonuclease A, would assume low energy conformations (residues 22 -25, a @-bend, having a segment homologous to the sequence of peptide T). Low energy conformers of the tetrapeptides could be superimposed to the native structure of segment 22-25 of ribonuclease A. Fig. shows the superimposition of peptide T (full square): Many low energy conformers could be calculated for the tetrapeptides but for the polio sequence. The P, value for most tetrapeptides were 5 -10 times higher that the value of the less active polio sequence. The results supported the hypothesis that the active peptide T adopts the native ribonuclease @-bend. Title: Potential cardiotonics. 4. Synthesis, cardiovascular activity, molecule-and crystal structure of 5-phenyl-and 5-(pyrid-4-yl)- Data determined: [dose of the compound (mollkg) required for 30 4% increase of the heart beat frequency of guinea pig or dog heart]; [dose of the compound (mollkg) required for 10 % decrease of systolic or diastolic blood pressure of dog]; Crystal structure (atomic coordinates of the compounds were determined by X-ray diffraction); Molecular modeling (molecule models were built using MOLPAC); MEP [molecular electrostatic potential (MEP) (dimension not given) was calculated using CNDOIZ]. Results: Milrinon and its oxygen containing bioisoster possess highly similar crystal structure and MEP isopotential maps ( Fig. 1 and Fig. 2) Both compounds show strong positive inotropic and vasodilatoric activity. It was suggested that the negative potential region around the thiocarbonyl group such as the carbonyl group in milrinon imitates the negative potential field around the phosphate group of CAMP. Title: Molecular mechanics calculations of cyclosporin A analogues. Effect of chirality and degree of substitution on the side chain conformations of (2S,3R,4R,6E)-3-hydroxy-4-methyl-2-(meth~lamino)-6octenoic acid and related derivatives. [solution conformation of CsA in CDCh has been elucidated via molecular dynamics simulation incorporating 58 distance constrains obtained from IR spectroscopy and nuclear Overhauser effect (NOE) data]; (conformational analysis was performed using the SEARCH subroutine within SYBYL); Energy minimization (low energy conformers were calculated using molecular mechanics withinMacroModel Ver. 1.5 applying an all-atom version of the AMBER force field). Results A total of 12 conformations of CsA have been identified within 4 kcalfmol of the minimum energy conformer. Population analysis showed that one conformer dominates in solution. Fig. 1 shows the superposition of the peptide backbone of the crystal and solution structures of CsA (crystal structure is drawn with thick line and the solution structure with thin line). It was shown that the Boltzmann distribution between active and inactive conformers correlated with the order of the immunosuppressive activity. A common bioactive conformer serving as a standard for further design has been proposed for CsA and its analogs. Abtstr. 320 Quant. Struct.-Act. Relat. 9, 234 -293 (1990) 4 Data determined: Molecular modeling (models of (I), (11) and (111) were built using SYBYL based on X-ray coordinates of the compounds); Conformational analysis (minimum energy conformations of the compounds were calculated using the SEARCH option of SYBYL and MNDO method; [interaction energy of the molecules (kcall mol) with a hypothetical receptor probe (negatively charged oxygen atom) calculated by GRID]. Results: The specific receptor area of the sodium channel was modeled with a negatively charged oxygen probe (carboxyl group), interacting with the positively charged (protonated) ligand. Fig. 1 shows areas for energetically favorable interaction (areas I, 11 OH 0 Ho2C *. . Quant. Struct.-Act. Relat. 9, 234 -293 (1990) Abstr. 321-322 279 Biological material: a) Aspergillus ochraceus; b) Carbopeptidase A. Data determined kobr [first order rate coefficient (IO-6/sec) of the hydrochloric acid hydrolysis of ochratoxin A and B]; X-ray crystallography (Coordinates of the crystal structure of ochratoxin A and B was obtained using X-ray diffraction); (models of ochratoxin A and B was built using ALCHEMY); ["C NMR chemical shifts (ppm) of the amide and ester carbonyls of the ochratoxins]. Chemical descriptors: PKa (negative logarithm of the acidic dissociation constant). Results: A reversal of the hydrolysis rate between ochratoxin A and B was observed comparing the hydrolysis rates obtained in vitro (carbopeptidase A) and in vivo (hydrochloric acid). The difference in hydrolysis rates cannot be due to conformation since the two toxins have the same conformation in both crystal and in solution. Fig. 1 shows the fit of ochratoxin A and B based on superimposing the phenolic carbon atoms. It is suggested that the relative large steric bulk of the chloro atom hinders the fit between ochratoxin A and the receptor site of carbopeptidase A. Thus, probably the slower metabolism is the reason, why ochratoxin A is more toxic than ochratoxin B. Title: Inhibitors of cholesterol biosynthesis. 1. Trans-6-(2-Pyrrol- Charge distribution studies showed that compactin had two distinct regions of relatively large partial charges corresponding to the pyrrol ring and the isobutyric acid side chain. Experiments for more closely mimicking the polar regions associated with the high activity of compactin indicated that potency of the new compounds was relatively insensitive to the polarity of the R' group. It was also suggested that an electron deficient pyrrole ring was required for high potency. Title: Synthesis and biological activity of new HMG-CoA reductase inhibitors. 1. Lactones of pyridine-and pyrimidine-substituted 3.5dihydroxy-6-heptenoicf-heptanoic) acids. Chemical descriptors: Results: An attempt was made to correlate electrophysiological activity with the effect of the position of the aryl group on the conformation of the side chain using molecular modeling. The study suggested that the compounds with class 111 activity prefer a gauche (A in Fig. 1 ) and compounds in which class I activity prefer trans relationship of the nitrogens (B in Fig. 1) : The study indicated that the point of attachment of the aryl moiety had an effect on the side chain conformation which appeared to be a controlling factor of the electrophysiological profile of these compounds. Title: A molecular mechanics analysis of molecular recognition by cyclodextrin mimics of a-chymotrypsin. Authors ( 1 ) Quant. Struct.-Act. Relat. 9. 234 -293 ( 1990) Biological material: Chymotrypsin. Data taken from the literature: Crystal structure (crystal coordinates of the macrocycles determined using X-ray diffraction analysis). Data determined: Molecular modeling Structure superposition (models of B-CD and in chains by Nmethylformamide and N-dimethyl-formamide substituted (capped) B-CD were built using the AMBER program and the coordinates for building the N-methylformamide substituent were calculated using MNDO in the MOPAC program); (energy minimization of the molecules were calculated in vacuo using molecular mechanics program with the AMBER force field); (energy minimized structures of B-CD and capped B-CD were separately fit to the Xray structure of the B-CD complex); [molecular electrostatic potential (kcallmol) of B-CD and capped B-CD were approximated by the Coulombic interaction between a positive point charge and the static charge distribution of the molecule, modeled by the potential derived atomic point charges at the nuclei and visualized as 2D MEP map]. Results: B-CD and capped B-CD were analyzed as biomimetic models of the active site of chymotrypsin. Capped B-CD was shown to be the more effective biomimetic catalyst. Capping also altered certain structural features of molecular recognition. The orientation of the secondary hydroxyls were altereddue to twisting of some of the glucose units. Secondary hydroxyl oxygen mimics the Ser-195 of chymotrypsin in initiating the acyl transfer event through nucleophilic attack on the substrate. Fig. 1 shows the energy minimized structures of B-CD (A) and capped B-CD (B) (fragment number is given in parenthesis). The MEP maps of B-CD and capped B-CD showed that the qualitative features of the electrostatic recognition were practically the same in the two mimics. BiologicaI material: Four monocotyledonous (Johnson grass, yellow foxtail, barnyard grass, yellow millet) and four dicotyledonous weed species (velvetleaf, morning glory, prickly sida, sicklepod). Data determined: [pre-emergence and postemergence herbicidal activities of the compounds were measured and rated using a scale ranging from 0 (no activity) to 9 (complete kill]; [measure of the compound's ability (dimension not given) to translocate upwards in plants through xylem vessels); [soil sorption coefficient calculated by the formula K, = C,/C,, where C, is the concentration of the compound (pg compoundlg soil) and C. is the concentration of the compound (pg compoundlml) in water solution in equilibrium with the soil]; (models of the compounds were built using MACCS and PRXBLD programs); TSCF Kd Molecular modeling Quant. Struct.-Act. Relat. 9, 234 -293 (1990) Abstr. 327-328 283 Conformational analysis (minimum energy conformations of the compounds were calculated using MM2 molecular mechanics method); (molecules were visualized using program MOGLI on an Evans and Sutherland Picture System 33); [total energies, orbital eigenvalues, atomic charges and dipole moments of simple model analogs of type I were calculated using PRDDO (Partial Retention of Diatomic Overlap) level of approximation]. Electronic structure Chemical descriptors: logP (logarithm of the partition coefficient in 1-octanollwater). Results: Conformational analyses and high level quantum mechanical calculations of the conformational preferences showed that the compounds with R = 4-C1 and 5-CI substituents adopt a coplanar structure stabilized by intramolecular hydrogen bond, whereas the 3-C analogue does not (Fig. 1 ): Higher logP values (0.6 -1.0 logarithmic unit difference), higher Kd and TSCF values of the 4-CI and 5-CI substituted compounds relative to the 3-CI analog were interpreted as the result of the intramolecular hydrogen bond and were consistent with the observation that the 4-CI and 5-CI analogs were active as post-emergence but not pre-emergence herbicides while the 3-CI derivative was active in both modes. Title: Application of molecular modeling techniques to pheromones of the marine brown algae Cutleria multifida and Ectocarpus siliculosus (Phaeophyceae). Metalloproteins as chemoreceptors? (geometrical models of the compounds were constructed using information from the Cambridge Structural Data Base (CSD) and calculated using molecular mechanics methods in SYBYL); (minimum energy conformations of the compounds were calculated using molecular mechanics method within SYBYL). Chemical descriptors: KFCQ [partition coefficient in FC72/water (FC72 = fluorocarbon Results: As both ectocarpene (I) and multifidene (11) trigger mutual cross reactions between male gametes of Ectocarpus siliculosus and Cutleria multifida males it was supposed that a common mode of binding should exist for the two structurally different pheromones. The active analogue approach was applied to model the pheromone receptor by superposing the minimum energy conformations of active structural analogues (HI, IV, V, VI) on ectocarpene and multifidene. The common active conformation of (I) and (11) was extracted by systematic superimposition of the analogues. To explain the function of the double bonds in the pheromones, the presence of a receptor bound metal cation was assumed. Simultaneous optimization,of both structures without and with a receptor bound metal cation resulted in virtually the same conformations. Fig. 1 shows the mapping of multifidene onto ectocarpene in their biologically relevant conformations. solvent)]. Title: Critical differences in the binding of aryl phosphate and carbamate inhibitors of acetylcholinesterases. Conformational analysis [minimum energy conformations of (Asn-Ala-Asn-Pro)9 was calculated using CHARMM (Chemistry at Harvard Macromolecular Mechanics), AMBER (Assisted Model Building with Energy Refinement) and ECEPP (Empirical Conformational Energy Program for Peptides) potential energy functions]; [root mean square deviation (A) of the position of the corresponding atoms of two superimposed molecular sructures], Results: 24 Low energy conformations of (Asn-Ala-Asn-Pro)9 has been determined using CHARMM, ECEPP and AMBER in order to determine their final conformations and relative energies. The final conformations were compared calculating the rms values of their C" atoms and matching the parameters of the energy minimized (Asn-Ala-Asn-Pro), peptide to that of the ideal helix or coiled coil. The similarity of the final conformations obtained by using any two different potentials starting from the same conformation varied from the satisfactory to highly unacceptable. The extent of difference between any pairs of the final conformations generated by two different potential energy functions were not significantly different. The lowestenergy conformation calculated by each of the energy potentials for any starting conformation was a left handed helix and pair-wise superposition of the C" atoms in the final conformations showed small rms values (1 .O -1.3 A) . It was suggested that the native conformation of (Asn-Ala-Asn-Pro), in the CS protein may be a left-handed helix, since all three potential energy functions generated such conformation. 234 -293 (1990) Source: Proteins 1989 Proteins , 6(2), 193 -209, 1989 . Biological material: Crambin. Data determined Phi-psi probability plot (probabilities of the occurrences of phi-psi dihedral angle pairs for each amino acid were determined and plotted using the data of approximately 100 proteins from the Brookhaven Protein Data Bank); (optimization technique for the reproduction the folding process converging to the native minimum energy structure by dynamically sampling many different conformations of the simplified protein backbone). Chemical descriptors: Phi-psi values [dihedral angles (deg) defined by the bonds on either side of the a-carbon atom of the amino acid residue in a protein]. Results: A simplified model has been developed for the representation of protein structures. Protein folding was simulated assuming a freely rotating rigid chain where the effect of each side chain approximated by a single atom. Phi-psi probabilities were used to determine the potentials representing the attraction or repulsion between the different amino acid residues. Many characteristics of native proteins have been successfully reproduced by the model: (i) the optimization was started from protein models with random conformations and led to protein models with secondary structural features (a-helices and 0strands) similar by nature and site to that of the native protein; (ii) the formation of secondary structure was found to be sequence specific influenced by long-range interactions; (iii) the association of certain pairs of cysteine residues were preferred compared to other cysteine pairs depending on folding; (iv) the empirical potentials obtained from phi-ps probabilities led to the formation of a hydrophobic core of the model peptide. x [dihedral angle (deg) ]. Results: Four kinds of Monte Carlo simulations of about 20,000 steps of the conformations of crambin were carried out by using the second derivative matrix of energy functions (starting from native and unfolded conformations both in two kinds of systems, in vacuo and in solution). Fig. 1 shows the native (a) and theunfolded (b) conformation of crambin. Starting from native conformation, the differences between the mean properties of the simulated crambin conformations obtained from in vacuo and solution calculations were not very large. The fluctuations around the mean conformation during simulation were smaller in solution than in vacuo, however. Simulation starting from the unfolded conformation resulted in a more intensive fluctuation of the structure in solution than in vacuo indicating the importance of the hydration energy term in the model. The conformations generated in the simulations starting from the native conformation deviate slightly from the Xray conformation (rms = 0.70 A and I . 10 A for in vacuo and solution simulations, respectively). The results indicate that the simulations of the protein with hydration energyare more realistic that the simulations without hydration energy. Fig.1 Results: Earlier studies overestimated the catalytic rate decrease of the hypothetical D102A point mutant of thrombin (20 orders of magnitude decrease calculated instead of the 4 order of magnitude measured). The source of error was due to an overestimation of V and neglecting the effects of the surrounding water molecules and induced dipoles. To compensate for these errors, a scale factor of 0.12 was introduced into the calculations. As aresult of the rescaling, one magnitude increase of tat for the D121 mutant and two magnitudes decrease of k,, of the K41 mutant of ribonuclease A was predicted. It was shown that the effect of the mutations on the catalytic rate depended almost entirely on steric factors. It was suggested that in mutants of serine proteases where the buried Asp is replaced by Ala or Asp, the kcat value will decrease between 4 -6 orders of magnitude. Title: High-resolution structure of an HIV zinc fingerlike domain via a new NMR-based distance geometry approach. Authors: Summers*, M.F.; South, T.L.; Kim [root mean square deviation (A) of the corresponding atoms of two superimposed molecular sructures]. Results: The atomic resolution structure of an HIV zinc fingerlike domain has been generated by a new NMR-based DG method using 2D NOESY backcalculation. The quality of the structures thus obtained were evaluated on the basis of the consistence with the experimental data (comparison of measured and back-calculated NMR spectra) rather than comparing it tostructural informations from other sources (e.g. X-ray data). The method provided a quantitative measure of consistence between experimental and calculated data which allowed for the use of tighter interproton distance constraints. The folding of the C( l)-F(2)-N(3)-C(4)-G(s)-K(6) residues were found to be virtually identical with the folding of the related residues in the X-ray structure of the iron domain of rubredoxin (RMS values 0.46 and 0.35 A). The backbone folding of the peptide was found to be. significantly different from that of the "classical" DNA-binding Zn-finger. Fig. 1 shows the wire frame model of all the back%ne atoms and certain side chain atoms of the peptide (DG struciure) (dashed lines indicate hydrogen atoms): active site of the protease dimer in an extended conformation with extensive van der Waals and hydrogen bonding and was more than 80 % excluded from contact with the surrounding water (Fig. I , where the inhibitor is shown in thicker lines and the hydrogen bonds in dashed lines): Data determined: AGO,, AGO, [standard free energy (callmol) of transfer of a molecule from an apolar phase to an aqueous phase, observed or calculated by Eq. 1: AGO, = C Aui Ai, where Aoi is the atomic solvation parameter of atomic group i, Ai is the accessible surface area of atom i, respectively]. Results: Atomic Solvation Parameters (ASPS) characterizing the free energy change per unit area for transfer of a chemical group from the protein interior to aqueous surroundings were determined. AGO, and AGO, were determined and compared, and a highly significant linear relationship is presented (Fig. 1) . One letter symbols indicate amino acid side chains: Fig .1 The binding of the inhibitor induced substantial movement in the en-'2 zyme around the residues 77 to 82 in both subunits at places exceeding i - The structure of glutamine synthetase is discussed. It was established that hydrophobic interactions are important for the intersubunit interactions, and the hydrophobic interactions between the two rings of subunits are stronger than between the subunits within a ring. The Cterminal helix contribute strongly to the inter-ring hydrophobic interaction. ASPS are suggested to estimate the contribution of the hydrophobic energy to protein folding and subunit assembly and the binding of small molecules to proteins. Title: Determination of the complete three-dimensional structure of the trypsin inhibitor from squash seeds in aqueous solution by nuclear magnetic resonance and a combination of distance geometry and dynamical simulated annealing. Authors: Holak*, T.A.; Gondol, D.; Otlewski, J.; Wilusz, T. Max-Planck-hstitut fiir Biocbemie D-8033 Martinsried bei Miinchen, Federal Republic of Germany. Title: Interpretation of protein folding and binding with Atomic Crystal structure (atomic coordinates of CMTI-I was determined by Solvation Parameters. X-ray diffraction method). Results: In order to obtain information of the 3D structure of the free CMTI-I in solution, a total of 34 inhibitor structures were calculated by a combination of distance-geometry and dynamical simulated annealing methods, resulting in well defined 3D positions for the backbone and side-chain atoms. Fig. 1 shows the superposition of the backbone (N, C", C, 0) atoms of the structures best fitted to residues 2 to 29 (binding loop): The average RMS difference between the individual structures and the minimized mean stfucture was 0.35(*0.08) A for the backbone atoms and 0.89(+0.17) A for all heavy atoms. Title: Electron transport in sulfate reducing bacteria. Molecular modeling and NMR studies of the rubredoxin-tetraheme-cytochrome-c3 complex. Biological material: a) Sulfate reducing bacterium (Desulfovibrio vulgaris); b) Rubredoxin (iron-sulfur protein); c) Tetraheme cytochrome c3 from D. vulgaris; e) Flavodoxin. detected in the segments from the residues 16 to 18 and 25 -25. Fig. 1 shows the best superposition (residues 2 to 29) of the NMR and crystal structure of CMTI-I indicating the backbone C, C", N, 0, as well as the disulfide C8 and S atoms: Fig.1 It was demonstrated that uncertainty in NMR structure determination can be eliminated by including stereospecific assignments and precise distance constraints in the definition of the structure. Crystal structure (coordinates of the crystal structure of the compounds were determined by X-ray crystallography). Results: The speed of the homolysis of the organometallic bond is 10" times higher in the apoenzyme bound coenzyme BIZ than in a homogenous solution. Structural changes occurring during the Co-C bond homolysis of the coenzyme BIZ leading from cobalt(II1) corrin to cobalt(I1) corrin were investigated. Fig. 1 shows the superposition of structures of the cobalt corrin part of the BIZ (dotted line) and of cob(II)alamin (solid line): Biological material: Apoenzyme, binding the coenzyme BIZ and the Data determined: . 1 shows that the crystal structure of BIZ and cob(I1)alamin are strikingly similar and offers no explanation for the mechanism of the protein-induced activation of homolysis. It was suggested that the Co-C bond may be labiiized by the apoenzyme itself and in addition to a substrate-induced separation of the homolysis fragments (which mights be supported by a strong binding of the separated fragments to the protein). 'H NMR (complete stereospecific assignments were carried out and proton-proton distance constrains were determined by the analyses of DQF-COSY, HOHAHA and NOESY spectra); NH, aH, OH [chemical shifts (ppm) of proton resonances of human ETI ; 3D structure (3D structure of ET was calculated using the distance geometry program DADAS based upon the NOESY proton-proton distance constrains determined by NMR spectroscopy); [root mean square distance (A) between 5 ET conformers calculated by distance geometry (DADAS)]. Results: The solution conformation of ET has been determined by the combined use of 2D 'H NMR spectroscopy and distance geometry calculations. Five structures of ET have been calculated from different initial conformations. The superposition of the backbone atoms the calculated structures is shown in Fig. 1 . The average RMS value in the core region for the main-cahin atoms was 0.46 A. Quant. Struct.-Act. Relat. 9, 234-293 (1990) The lack of specific interactions between the core and tail portions of ET and a characteristic helix-like conformation in the region from Lys' to Cys" was shown. Literature data indicated that neither the ETI -1 5 nor the ETl6-21 truncated derivatives of ET showed constricting or receptor binding activity suggesting that the ET receptor recognizes an active conformation consisting of both the tail and core portion. The present study, however, suggested that the receptor bound conformation of ET is probably different from that in solution because the lack of interaction between tail and core. The hydrophobic nature of the tail suggested the importance of a hydrophobic interaction with the receptor. Compounds: Triphenyl-methyl-phosphit cation (TPMP+). Biological material: Nicotinic acetylcholine receptor (AChR), a prototype of the type I of membrane receptor protein from the electric tissue of Torpedo and Electrophorus. Results: A computer model of the AChR ion channel has been proposed. Fig. 1 shows the side view of the ion channel model with five pore-forming MZ-helices and the channel blocking photoaffinity label (TPMP +) represented by the dotted sphere: Fig.1 It was supported by electronmicroscopy, electrophysiological-and biochemical experiments that the MZ-helices were formed by homologous amino acid sequences containing negatively charged amino acid side chains which act as the selectivity filter. The amino acid side chains may undergo conformational changes during the permeation of the cation. The predicted transmembrane folding of four transmembrane a-helices of type I receptors is shown in Fig. 2: Fig. 2 Energy profile calculations indicate that other transmembrane sequences of the receptor protein besides M2 may affect the ion channel. Source: CABIOS 1989, 5 (3), 219 -226. Results: An interactive computer program TEFOOJJ2 has been developed for drug design on IBM/PC and compatible computers. The program contains the following modules and performs the following calculations: a) Series design for selecting an optimal starting set of compounds using a modified version of Austel's method; b) Regression analysis calculating the Hansch's equation; c) Hansch searching method using the equation calculated by the regression analysis routine or the use of an input equation for the identification of the 10 most active compounds; d) Geometrical searching methods for finding the optimum substituents in the parameter space using the sphere, ellipse, quadratic or polyhedric cross algorithms with or without directionality factors; e) Space contraction for reducing the dimension of the parameter space by eliminating non-significant parameters; f) An example is given for the lead optimization of an aliphatic lead compound correctly predicting the n-pentane to be the optimum substituent. Results: A new expert system SPARC is being developed at EPA and at the University of Georgia to develop quantitative structure-activity relationships for broad compound classes: a) Classical QSAR approaches predict therapeutic response, environmental fate or toxicity from structure/property descriptors quantifying hydrophobicity, topological descriptors, electronic descriptors and steric effects; b) SPARC (SPARC Performs Automated Reasoning in Chemistry), an expert system written in Prolog, models chemistry at the level of physical organic chemistry in terms of mechanism of interaction that contribute to the phenomena of interest; c) SPARC uses algorithms based on fundamental chemical structure theory to estimate parameters such as acid dissociation constants (pK,s), hydrolysis rate constants, UV, visible and IR absorption spectra, and other properties; d) The information required to predict input data for broad classes of compounds is dispersed throughout the entire IR spectrum and can be extracted using Fourier transforms; e ) The accuracy of SPARC algorithm was demonstrated on the close match of calculated and experimental pK, values of 20 carboxylic acid derivatives near to the noise level of measurement. Abtstr. 347-351 Quant. Struct.-Act. Relat. 9, 234-293 (1990) Results: A new stand-alone molecular simulation program, NMRgraf integrating molecular modeling and NMR techniques has been introduced by BioDesign Inc. a) NMRgraf is a molecular modeling program utilizing force fields which incorporate empirical properties such as bond lengths and angles, dihedral, inversion and nonbonded interactions, electrostatic charges and van der Waals interactions; b) The molecular structural properties are combined with nuclear Overhouser effect (NOE) and J-coupling NMR data (experimental interproton distance constrains and bond angle data); c) The NMR proton-proton distance data are accurate only at relatively short distances (5 to 10 A) which restricts the use of NMR NOE approaches only for the analysis of molecules with known X-ray structure; d) The combination of NMR and molecular modeling approaches, however, makes it possible to model virtually any molecule even if its structure does not exist in the databases. Title: Electronic structure calculations on workstation computers. Results: The main features of the program system TURBOMOLE for large-scale calculation of SCF molecular electronic structure on workstation computers is described: a) The program system allows for SCF level treatments of energy, first-and second-order derivatives with respect to nuclear coordinates, and an evaluation of the ME? correlation energy approximation; b) The most important modules of TURBOMOLE are (i) DSCF performing closed and open shell RHF calculations; (ii) EGRAD used for analytical SCF gradient evaluations; (iii) KORA calculating direct two-electron integral transformation (iv) FORCE for the computation and processing of integral derivatives and the solution of CPHF equations; c) Comparison and evaluation of timings of representative applications of TURBOMOLE on various workstations showed that APOLLO DS 1O.OOO and IRIS 4D1210 were the fastest and comparable to the CONVEX C210 in scalar mode. Results: A new algorithm has been developed for the calculating and visualizing space filling models of molecules.The algorithm is about 25 times faster than a conventional one and has an interesting transparency effect when using a stereo viewer. a) The algorithm is briefly described and and the result is visualized on modeling a ribonucleotide unit; b) In the order of increasing atomnumbers, the (X,Y) sections of the hemispherical disks of the atoms are projected on the screen with decreasing value of the azimuthal angle (p) of the van der Wads radius and as the value of p decreases the projection is increasingly whitened to obtain shading effect on the surfaces of the spheres; c) The transparency of the van der Waals' surfaces of atoms of a molecule makes possible to perceive almost the whole space filling structure and not only the surface, hiding the underlying atoms. Title: Supercomputers and biological sequence comparison algo-Authors: Core*, N.G. ; Edmiston, E.W. ; Saltz, J.H. ; Smith, R.M. rithms. Yale University School of Medicine New Haven CT 06520-2158, USA. Source: Computers Biomed. Res. 1989, 22(6) , 497 -515. Compounds: DNA and protein fragments. Chemical descriptors: Sequences of monomers. Results: A dynamic programming algorithm to determine best matches betweenpairs of sequences or pairs of subsequences has been used on the Intel iPSC/l hypercube and ,the Connection Machine (CM-I). Parallel processing of the comparison on CM-I results in run times which are 65 to 230 times as fast as the VAX 8650, with this factor increasing as the problem size increases. The CM-I and the Intel iPSC hypercube are comparable for smaller sequences, but the CM-I is several times quicker for larger sequences. A fast algorithm by Karlin and his coworkers designed to determine all exact repeats greater than a given length among a set of strings has been tried out on the Encore Multimax/32O. The dynamic programming algorithms are normally used to compare two sequences, but are very expensive for multiple sequences. The Karlin algorithm is well suited to comparing multiple sequences. Calculating a multiple comparison of 11 DNA sequences each 300-400 nucleotides long results in a speedup roughly equal to the number of the processors used. Source: CABIOS 1989, 5(4), 323. Results: A program has been developed for the prediction and display of the secondary structure of proteins using the primary amino acid sequence as database. a) The program calculates and graphically demonstrates four predictive profiles of the proteins allowing interpretation and comparison with the results of other programs; b) As a demonstration the sliding averages of n sequential amino acids were calculated and plotted for four properties of human interleukin 6: (i) plot of the probabilities of a-helix, P-structure and P-turns according to Chou and Fasman; (ii) @-turn index of Chou and Fasman; (iii) plot of the hydrophobicity index of Hopp and Woods; (iv) flexibility index of Karplus and Schulz; c) The regions of primary structure having properties which usually go together agreed reasonably well with each other, i.e. loops and turns with bend probability and hydrophilicity with flexibility. Title: 3DSEARCH. a system for three-dimensional substructure searching. Quant. Struct.-Act. Relat. 9, 234-293 (1990) Sci. 1989, 29(4) , 255 -260. Results: The search for threedimensional substructures is becoming widely used in 3D modeling and for the construction of pharmacophores for a variety of biological activities. A system (3DSEARCH) for the definition and search of three-dimensional substructures is described: a) Representation of atom types consists of five fields (i) element (He-U); (ii) number of non hydrogen neighbors (bonded atoms); (iii) number of 'K electrons; (iv) expected number of attached hydrogens; (v) formal charge; (vi) four type of dummy atoms are also used to define geometric points in space (e.g. centroid of a b) Definition of queries (i) definition of spatial relationship between atoms; (ii) matches in atom type (iii) preparation of keys (constituent descriptors); (iv) execution of key search; (v) geometric search including the handling of angleldihedral constrains and takes into account "excluded volume"; c) Time tests showed that a search of 3D structures with 3 to 5 atoms in large databases with more than 200,000 entries took only a few minutes (22 -492 s). Results: A database containing about 265,000 compounds in connection tables and 30,000 experimentally determined structures from the Cambridge Structural Database has been transformed into a database of low energy 3D molecular structures using the program CONCORD. The strategy for building the 3D database consisted of the following four steps: a) Generation of approximate 3D coordinates from connection tables (hydrogens were omitted); b) Assignment of atom types from connection table information characterized by five descriptors: (i) element type (He-U); (ii) number of attached non-hydrogen neighbors (0 -8); (iii) number of 'K electrons (0 -2); (iv) calculated number of attached hydrogens (0-4); (v) formal charge ( -1, 0, 1); c) Addition of three types of chemically meaningful dummy atoms for purposes of 3D substructure searching: (i) centroids of planar 5and 6-membered rings; (ii) dummy atoms representing the lone electron pairs; (iii) ring perpendiculars positioned orthogonal to and 0.5 A above and below each planar ring; d) Efficient storage of the resultant coordinate database indexing the compounds with identification number; e) The database can be used among others to deduce pharmacophores essential for biological activity and to search for compounds containing a given pharmacophore. Title Source: C&EN 1989, 67(43) , 18-20. Results: A Complex Carbohydrate Structure Database (CCSD) has been developed by the Complex Carbohydrate Research Center at the University of Georgia, having more than 2000 structures and related text files, with about 3000 more records to be added over the next two years. The following are the most important features of CCSD: a) In CCSD, database records include full primary structures for each complex carbohydrate, citations to papers in which sequences were published, and suplementary information such as spectroscopic analysis, biological activity, information about binding studies, etc; b) Structural display format visualize branching, points of attachment between glycosyl residues and substituents, anomeric configuration of glycosyl linkages, absolute configuration of glycosyl residues, ring size, identity of proteins or lipids to which carbohydrates are attached and other data; c) It is planned that CCSD will provide threedimensional coordinates, to visualize and rotate the structures in stereo and study their interaction with proteins or other biopolimers. Probing Bioactive Mechanisms p Commercial carbamate insecticides of type II, where R' = H, s-Bu Biological material: Acetylcholinesterase. Data determined: D [distance (A) between the serine oxygen of acetylcholinesterase and the methyl substituents of the carbamate and phosphate inhibitors Molecular modeling (models and minimum energy conformations of acetylcholine, aryl carbamate and phosphate ester inhibitors were created using the draw mode of a MACCS database and the PRXBLD modeling program) Results: Transition state modeling of the reaction of the serine hydroxyl ion of acetylcholinesterase with the methylcarbamoyl and dimethyl phosphoryl derivatives of 3,4-dimethyl-phenol showed that the active site binding for these two classes of acetylcholinesterase inhibitors should be different. The model shows that the distances between the serine oxygen and the ring substituents (meta-and para-me-thy1 groups) are different in both spacing and direction. Fig. 1 shows the transition state models of serine hydroxyl D values for the meta-and para-methyl substituents of N-methylcarbamate and dimethylphosphate were meta = 6.20, para = 8.10 and meta = 5.48, para = 7.03 A, respectively Title: A comparison of the CHARMM, AMBER and ECEPP potentials for peptides. 1. Conformational predictions for the tandemly repeated peptide (Asn-Ala-Asn-Pro)9 Biological material: Tandemly repeated peptide (Asn-Ala-Asn-Pro) which is a major immunogenic epitope in the circumsporozoite (CS) protein of Plasmodium falciparum Conformational analysis Dream or reality? A Authors: NBray-Szab6*, G.; Nagy, J.; BCrces, A. priori predictions for thrombin and ribonuclease mutants Molecular modeling (Geometric model of subtilisin and trypsin was built using Protein Data Bank coordinates and model of thrombin was built using the theoretical coordinate set of Graphic representations of the triad of the tetrahedral intermediate for the enzymes formed on the active side chain and residues (Ser-221, His-64 and Asp32 in subtilisin; Ser-195, His-64 and Asp-I02 in trypsin and thrombin; His-119, Lys-41 and His-I2 in ribonuclease A) were created using PCGEOM Electrostatic properties (electrostatic surfaces and fields of the molecules were calculated and displayed using AMBER and associated programs); [chemical shift (ppm) measured by NMR Results: A hypothetical model between rubredoxin and cytochrome c3 was built as a model for the study of electron transfer between different redox centers, as observed in other systems. Fig. I shows the main chains atoms of the proposed complex where the hemes of the cytochromes are shown along with the center of rubredoxin and stabilized by hydrogen bonds and charge-pair interactions (the nonheme iron of the rubredoxin is in close proximity to heme 1 of cytochrome c3):6 spectroscopy].The model was consistent with the requirements of steric factors, complementary electrostatic interactions and NMR data of the complex. Comparison of the new model and the NMR data of the previously proposed flavodoxin-cytochrome c3 complex showed that both proteins interacted with the same heme-group of cytochrome c3. Title: Nuclear magnetic resonance solution and X-ray structures of squash trypsin inhibitor exhibit the same conformation of the proteinase binding loop.Authors: Holak*, T.A.; Bode, W.; Huber, R.; Otlewski, J.; Wilusz, T. Max-Planck-Institut fiir Biochemie D-8033 Martinsried bei Munchen, Federal Republic of Germany.Source: J. Mol. Biol. 1989, No.210, 649-654. Biological material: a) Trypsin inhibitor from the seeds of the squash Cucurbita maxima; b) P-Trypsin and trypsin inhibitor complex.Title: Retention prediction of analytes in reversed-phase high-performance liquid chromatography based on molecular structure. 5. Quant. Struct.-Act. Relat. 9, 234-293 (1990) Results: An expert system (CRIPES) has been developed for the prediction of RP-HPLC retention indices from molecular structure by combining a set of rules with retention coefficients stored in a database. The method underlying the system is based on the "alkyl aryl retention index scale" and aims to improve the reproducibility of prediction and compatibility between various instruments and column materials. The VP-Expert system shell from Microsoft was used for the development. The performance of CRIPES was demonstrated on several subtypes of substituted benzenes (phenacyl halides, substituted arylamines, arylamides and other types). In general the calculated and measured retention indices agreed well but relatively large deviations were observed between the IE and IC values for phenacyl bromides and chlorides, o-and p-bromo anilines, N-methylbenzamide and N,N-dimethylbenzamide and phthalate esters. The extension of the database with further interaction values was regarded as necessary for a consistently high accuracy at prediction. [out-of-plane bending energy (kcal/mol) given by the formula E, = kd', where d is the distance from the atom to the plane defined by its three attached atoms and k is a force constant]; Eb E, [torsional energy (kcal/mol .deg2) associated with four consecutive bonded atoms i,j,k,l given by the formula E, = ki,j.k,l(l fS/lSlCOS(lSlBi,j,k,~)), where B is the torsion angle between atoms i j , k and 1, s and k are constants]; E, [potential energy (kcallmol) (nonbonded energy term) associated with any pair of atoms which are neither directly bonded to a common atom or belong to substructures more than a specified cutoff distance away given by the formula E, = kij(l.0la" -2.0/a6), where a is the distance between the two atoms divided by the sum of their radii, and k is the geometric mean of the k constants associated with each atom]. Results: Model geometries produced by the Tripos 5.2 force field have been assessed by minimizing the crystall structures of three cyclic hexapeptides, crambin and 76 diverse complex organic compounds. Comparative force field studies of the Tripos 5.2, Amber and AmberlOPLS force fields were carried out by energy minimization of three cyclic hexapeptides starting from the crystal structures showed the Tripos 5.2 force field superior to the others with the exception of Amber, ECEP2 and LEVB force fields as published by other workers. A direct comparison between the performance of Tripos 5.2 and . Amber using isolated crambin showed that the bond and torsion angles of Tripos 5.2 averaged closer to the crystal structure than the angles calculated by Amber (rms = 0.025 A, 2.97 deg and 13.0 deg for bond lengths, angles, and torsions, respectively, and rms = 0.42 A for heavy atoms). Fig. 1 shows the superimposed structures of crambin before and after energy minimization:Fi 9.1 Tripos 5.2 was assessed for general purpose applications by minimizing 76 organic compounds starting from their crystal structures. The test showed that Tripos 5.2 had a systematic error in overestimating the bond lengths of atoms in small rings. Statistical analysis of the results showed that T r i p s 5.2 had an acceptable overall performance with both peptides and various organic molecules, however, its performance was not equal to the best specialized force fields. Title: New software weds molecular modeling, NMR. Author: Krieger. J. C&EN 1155 Sixteenth St., N.W., Washington DC 20036, USA. Source: C&EN 1990, 68(13) , 16.