Jyrki Taskinen

Prediction of physicochemical properties based on neural network modelling

The literature describing neural network modelling to predict physicochemical properties of organic compounds from the molecular structure is reviewed from the perspective of pharmaceutical research. The standard three-layer, feed-forward neural network is the technique most frequently used, although the use of other techniques is increasing. Various approaches to describe the molecular structure have been successfully used, including molecular fragments, topological indices, and descriptors calculated by semi-empirical quantum chemical methods. Some physicochemical properties, such as octanol-water partition coefficient, water solubility, boiling point and vapour pressure, have been modelled by several research groups over the years using different approaches and structurally diverse large training sets. The prediction accuracy of most models seems to be rather close to the performance of the experimental measurements, when the accuracy is assessed with a test set from the working database. Results with independent test sets have been less satisfactory. Implications of this problem are discussed.

Expression and Characterization of Recombinant Human UDP-glucuronosyltransferases (UGTs) UGT1A9 IS MORE RESISTANT TO DETERGENT INHIBITION THAN THE OTHER UGTs AND WAS PURIFIED AS AN ACTIVE DIMERIC ENZYME

Eight human liver UDP-glucuronosyltransferases (UGTs) were expressed in baculovirus-infected insect cells as fusion proteins carrying a short C-terminal extension that ends with 6 histidine residues (His tag). The activity of recombinant UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT2B4, UGT2B7, and UGT2B15 was almost fully inhibited by 0.2% Triton X-100. In the case of UGT1A9, however, glucuronidation of α-naphthol and scopoletin was resistant to such inhibition, whereas glucuronidation of entacapone and several other aglycones was sensitive. His-tagged UGT1A9 was purified by immobilized metal-chelating chromatography (IMAC). Purified UGT1A9 glucuronidated scopoletin at a high rate, whereas its glucuronidation activity toward entacapone was low and largely dependent on phospholipid addition. Recombinant UGT1A9 in which the His tag was replaced by hemagglutinin antigenic peptide (HA tag) was also prepared. Insect cells were co-infected with baculoviruses encoding both HA-tagged and His-tagged UGT1A9. Membranes from the co-infected cells, or a mixture of membranes from separately infected cells, were subjected to detergent extraction and IMAC, and the resulting fractions were analyzed for the presence of each type of UGT1A9 using tag-specific antibodies. In the case of separate infection, the HA-tagged UGT1A9 did not bind to the column. When co-infected with His-tagged UGT1A9, however, part of the HA-tagged enzyme was bound to the column and was eluted by imidazole concentration gradient together with the His-tagged UGT1A9, suggesting the formation of stable dimers that contain one His-tagged and one HA-tagged UGT1A9 monomers.

Characteristics of catechol O-methyltransferase (COMT) and properties of selective COMT inhibitors

The enzyme-catalyzed O-methylation of catecholamines was first described by Axelrod and coworkers in the late 1950’s [1–3]. They called the responsible enzyme catechol O-methyltransferase (COMT). During the subsequent 15 years the enzyme was partially purified, its distribution was established, several reaction mechanisms were proposed, and a number of inhibitors were described. The results of this study period were extensively reviewed by Guldberg and Marsden in 1975 [4].

X-ray Crystal Structure of Human Dopamine Sulfotransferase, SULT1A3 MOLECULAR MODELING AND QUANTITATIVE STRUCTURE-ACTIVITY RELATIONSHIP ANALYSIS DEMONSTRATE A MOLECULAR BASIS FOR SULFOTRANSFERASE SUBSTRATE SPECIFICITY

Humans are one of the few species that produce large amounts of catecholamine sulfates, and they have evolved a specific sulfotransferase, SULT1A3 (M-PST), to catalyze the formation of these conjugates. An orthologous protein has yet to be found in other species. To further our understanding of the molecular basis for the unique substrate selectivity of this enzyme, we have solved the crystal structure of human SULT1A3, complexed with 3′-phosphoadenosine 5′-phosphate (PAP), at 2.5 Å resolution and carried out quantitative structure-activity relationship (QSAR) analysis with a series of phenols and catechols. SULT1A3 adopts a similar fold to mouse estrogen sulfotransferase, with a central five-stranded β-sheet surrounded by α-helices. SULT1A3 is a dimer in solution but crystallized with a monomer in the asymmetric unit of the cell, although dimer interfaces were formed by interaction across crystallographic 2-fold axes. QSAR analysis revealed that the enzyme is highly selective for catechols, and catecholamines in particular, and that hydrogen bonding groups and lipophilicity (cLogD) strongly influenced K m . We also investigated further the role of Glu146 in SULT1A3 using site-directed mutagenesis and showed that it plays a key role not only in defining selectivity for dopamine but also in preventing many phenolic xenobiotics from binding to the enzyme.

Aqueous solubility prediction of drugs based on molecular topology and neural network modeling.

A method for predicting the aqueous solubility of drug compounds was developed based on topological indices and artificial neural network (ANN) modeling. The aqueous solubility values for 211 drugs and related compounds representing acidic, neutral, and basic drugs of different structural classes were collected from the literature. The data set was divided into a training set (n = 160) and a randomly chosen test set (n = 51). Structural parameters used as inputs in a 23-5-1 artificial neural network included 14 atom-type electrotopological indices and nine other topological indices. For the test set, a predictive r2 = 0.86 and s = 0.53 (log units) were achieved.

Relationship between immobilised artificial membrane chromatographic retention and the brain penetration of structurally diverse drugs

Retention factors were determined for a set of 26 drugs, for which brain/blood concentration data are available, using immobilised artificial membrane (IAM) chromatography. The compound set represented acidic, basic and neutral drugs from various structural classes. The relationship between IAM retention and lipophilicity (n-octanol-water partition coefficient Koct), molecular size and acid/base character of the drugs and the relationship between brain distribution and IAM retention and the other parameters were analysed. IAM retention was increased with increases in lipophilicity and solute size, and decreased by the ionisation of acidic groups. Ionisation of basic groups had no significant effect. A three-parameter regression model with log Koct, molecular weight and an indicator parameter for the presence of carboxyl group explained 93% of the variation in log kIAM. The concentration ratio between brain and blood (log BB) was only weakly correlated with the IAM chromatographic retention or n-octanol-water partitioning. Three-parameter models taking ionisation and size into account, in addition to either log Koct or log kIAM, explained about 85% of the variation of log BB in the test set. Although IAM chromatography offers no advantage in these models, it seems to provide a better model than n-octanol-water partitioning for the membrane distribution of ionised compounds.

Quantitation of entacapone glucuronide in rat plasma by on-line coupled restricted access media column and liquid chromatography-tandem mass spectrometry.

A column-switching liquid chromatography-electrospray ionization-tandem mass spectrometric (LC-ESI-MS-MS) method was developed for the direct analysis of entacapone glucuronide in plasma. The plasma samples (5 microl) were injected onto a C18-alkyl-diol silica (ADS) column and the matrix compounds were washed to waste with a mixture of 20 mM ammonium acetate solution at pH 4.0-acetonitrile (97:3). The retained analyte fraction containing (E)- and (Z)-isomers of glucuronides of entacapone and tolcapone glucuronide (internal standard) was backflushed to the analytical C18 column, with a mixture of 20 mM ammonium acetate-acetonitrile (85:15) for the final separation at pH 7.0. The eluate was directed to the mass spectrometer after splitting (1:100). The mass spectrometer was operated in the negative ion mode and the deprotonated molecules [M-H]- were chosen as precursor ions for the analytes and internal standard. Collisionally induced dissociation of [M-H] in MS-MS resulted in loss of the neutral glucuronide moiety and in the appearance of intensive negatively charged aglycones [M-H-Glu]-, which were chosen as the product ions for single reaction monitoring. Quantitative studies showed a wide dynamic range (0.0025-100 microg/ml) with correlation coefficients better than 0.995. The method was repeatable within-day (relative standard deviation, RSD<7%) and between-day (RSD<14%) and the recovery (78-103%) was better than with the traditional, laborious pretreatment method. The use of tandem mass spectrometry permitted low limits of detection (1 ng/ml of entacapone glucuronide). The method was applied for the quantitation of (E)- and (Z)-isomers of entacapone glucuronide in plasma of rats used in absorption studies.

Glucuronidation of Entacapone, Nitecapone, Tolcapone, and Some Other Nitrocatechols by Rat Liver Microsomes

AbstractPurpose. Nitrocatechol COMT inhibitors are a new class of bioactive compounds, for which glucuronidation is the most important metabolic pathway. The objective was to characterize the enzyme kinetics of nitrocatechol glucuronidation to improve the understanding and predicting of the pharmacokinetic behavior of this class of compounds. Methods. The glucuronidation kinetics of seven nitrocatechols and 4-nitrophenol, the reference substrate for phenol UDP-glucuronosyltrans-ferase activity, was measured in liver microsomes from creosote-treated rats and determined by non-linear fitting of the experimental data to the Michaelis-Menten equation. A new method that combined densitometric and radioactivity measurement of the glucuronides separated by HPTLC was developed for the quantification. Results. Apparent Km values for the nitrocatechols varied greatly depending on substitution pattern being comparable with 4-nitrophenol (0.11 mM) only in the case of 4-nitrocatechol (0.19 mM). Simple nitrocatechols showed two-fold Vmax values compared with 4-nitrophenol (68.6 nmol min−1 mg−1), while all disubstituted catechols exhibited much lower glucuronidation rate. Vmax/Km values were about 10 times higher for monosubstituted catechols compared to disubstituted ones. The kinetic parameters for COMT inhibitors were in the following order: Km nitecapone >> entacapone > tolcapone; Vmax nitecapone > entacapone > tolcapone; Vmax/Km tolcapone > nitecapone > entacapone. Conclusions. Nitrocatechols can in principle be good substrates of UGTs. However, substituents may have a remarkable effect on the enzyme kinetic parameters. The different behaviour of nitecapone compared to the other COMT inhibitors may be due to its hydrophilic 5-substituent. The longer elimination half-life of tolcapone in vivo compared to entacapone could not be explained by glucuronidation kinetics in vitro.

Direct analysis of nitrocatechol-type glucuronides in urine by capillary electrophoresis-electrospray ionisation mass spectrometry and tandem mass spectrometry.

Direct, quantitative capillary electrophoresis-electrospray ionisation mass spectrometric (CE-ESI-MS) and tandem mass spectrometric (CE-ESI-MS-MS) methods are described for the quantitation of 3-O-glucuronides of E- and Z-entacapone isomers (EEG and EZG) and tolcapone (TG) in urine. 3-O-Glucuronide of nitecapone was used as internal standard. Good separation of glucuronides was achieved with 20 mM ammonium acetate as separation solution at pH 6.84. Stacking was used to increase the sensitivity of the method by introducing samples in 5 mM ammonium acetate. CE-ESI-MS and CE-ESI-MS-MS methods are linear with correlation coefficients better than 0.9983 and 0.9982, and repeatable with relative standard deviations below 9 and 14%, respectively. The limit of detection (LOD) in CE-ESI-MS at signal-to-noise ratio 3 is 100 ng/ml for EEG and EZG and 250 ng/ml for TG. The CE-ESI-MS-MS method was the more sensitive; LOD was 7 ng/ml for all compounds, without any concentration of the sample.

CoMFA modeling of human catechol O-methyltransferase enzyme kinetics.

Three-dimensional QSAR models with different charge calculation methods (MOPAC-AM1-ESP, MOPAC-AM1-Coulson and Gasteiger-Hückel) were developed for predicting all three enzyme kinetic parameters Km, Vmax and Vmax/Km for catecholic substrates of human soluble catechol O-methyltransferase (S-COMT). The empirical parameters of 45 substrates were correlated to the steric and electronic molecular fields of the substrates utilizing Comparative Molecular Field Analysis (CoMFA). Alignment rules for CoMFA were developed based on the catalytic mechanism and crystal structure of S-COMT, and the analysis was optimized using an all-space search technique. Leave-one-out and leave-n-out cross-validation (with 5 and 10 cross-validation groups) was carried out, and all developed models proved to be statistically significant with q2 values up to 0.84. The models based on MOPAC charge calculations predicted the empirical values clearly better than the Gasteiger-Hückel method. The derived CoMFA coefficient contour maps of steric and electrostatic interactions correlated clearly with the S-COMT crystallographic structures.