Artificial Neural Network Analysis of Pharmacokinetic and Toxicity Properties of Lead Molecules for Dengue Fever, Tuberculosis and Malaria.

Poor pharmacokinetic and toxicity profiles are major reasons for the low rate of advancing lead drug candidates into efficacy studies. The In-silico prediction of primary pharmacokinetic and toxicity properties in the drug discovery and development process can be used as guidance in the design of candidates. In-silico parameters can also be used to choose suitable compounds for in-vivo testing thereby reducing the number of animals used in experiments. At the Novartis Institute for Tropical Diseases, a data set has been curated from in-house measurements in the disease areas of Dengue, Tuberculosis and Malaria. Volume of distribution, half-life, total in-vivo clearance, in-vitro human plasma protein binding and in-vivo oral bioavailability have been measured for molecules in the lead optimization stage in each of these three disease areas. Data for the inhibition of the hERG channel using the radio ligand binding dofetilide assay was determined for a set of 300 molecules in these therapeutic areas. Based on this data, Artificial Neural Networks were used to construct In-silico models for each of the properties listed above that can be used to prioritize candidates for lead optimization and to assist in selecting promising molecules for in-vivo pharmacokinetic studies.

Structure activity relationships of 4-hydroxy-2-pyridones: A novel class of antituberculosis agents.

Pyridone 1 was identified from a high-throughput cell-based phenotypic screen against Mycobacterium tuberculosis (Mtb) including multi-drug resistant tuberculosis (MDR-TB) as a novel anti-TB agent and subsequently optimized series using cell-based Mtb assay. Preliminary structure activity relationship on the isobutyl group with higher cycloalkyl groups at 6-position of pyridone ring has enabled us to significant improvement of potency against Mtb. The lead compound 30j, a dimethylcyclohexyl group on the 6-position of the pyridone, displayed desirable in vitro potency against both drug sensitive and multi-drug resistant TB clinical isolates. In addition, 30j displayed favorable oral pharmacokinetic properties and demonstrated in vivo efficacy in mouse model. These results emphasize the importance of 4-hydroxy-2-pyridones as a new chemotype and further optimization of properties to treat MDR-TB.

The importance of molecular complexity in the design of screening libraries.

The one-dimensional model of Hann et al. (JChem Inf Comput Sci 41(3):856­864) has been extended to include reverse binding and wrap-around interaction modes between the protein and ligand to explore the complete combinatorial matrix of molecular recognition. The cumulative distribution function of the Maxwell­Boltzmann distribution has been used to calculate the probability of measuring the sensitivity of the interactions as the asymptotic limits of the distribution better describe the behavior of the interactions under experimental conditions. Based on our model, we hypothesized that molecules of lower complexity are preferred for target based screening campaigns, while augmenting such a library with moieties of moderate complexities maybe better suited for phenotypic screens. The validity of the hypothesis has been assessed via the analysis of the hit rate profiles for four ChemBL datasets for enzymatic and phenotypic screens.

Design, synthesis, and biological evaluation of indole-2-carboxamides: a promising class of antituberculosis agents.

Indole-2-carboxamides have been identified as a promising class of antituberculosis agents from phenotypic screening against mycobacteria. One of the hits, indole-2-carboxamide analog (1), had low micromolar potency against Mycobacterium tuberculosis (Mtb), high mouse liver microsomal clearance, and low aqueous solubility. Structure-activity relationship studies revealed that attaching alkyl groups to the cyclohexyl ring significantly improved Mtb activity but reduced solubility. Furthermore, chloro, fluoro, or cyano substitutions on the 4- and 6-positions of the indole ring as well as methyl substitution on the cyclohexyl ring significantly improved metabolic stability. 39 and 41, the lead candidates, displayed improved in vitro activity compared to most of the current standard TB drugs. The low aqueous solubility could not be mitigated because of the positive correlation of lipophilicity with Mtb potency. However, both compounds displayed favorable oral pharmacokinetic properties in rodents and demonstrated in vivo efficacy. Thus, indole-2-carboxamides represent a promising new class of antituberculosis agents.

Preclinical evaluation of the antifolate QN254, 5-chloro- N'6'-(2,5-dimethoxy-benzyl)-quinazoline-2,4,6-triamine, as an antimalarial drug candidate.

Drug resistance against dihydrofolate reductase (DHFR) inhibitors-such as pyrimethamine (PM)-has now spread to almost all regions where malaria is endemic, rendering antifolate-based malaria treatments highly ineffective. We have previously shown that the di-amino quinazoline QN254 [5-chloro-N'6'-(2,5-dimethoxy-benzyl)-quinazoline-2,4,6-triamine] is active against the highly PM-resistant Plasmodium falciparum V1S strain, suggesting that QN254 could be used to treat malaria in regions with a high prevalence of antifolate resistance. Here, we further demonstrate that QN254 is highly active against Plasmodium falciparum clinical isolates, displaying various levels of antifolate drug resistance, and we provide biochemical and structural evidence that QN254 binds and inhibits the function of both the wild-type and the quadruple-mutant (V1S) forms of the DHFR enzyme. In addition, we have assessed QN254 oral bioavailability, efficacy, and safety in vivo. The compound displays favorable pharmacokinetic properties after oral administration in rodents. The drug was remarkably efficacious against Plasmodium berghei and could fully cure infected mice with three daily oral doses of 30 mg/kg. In the course of these efficacy studies, we have uncovered some dose limiting toxicity at higher doses that was confirmed in rats. Thus, despite its relative in vitro selectivity toward the Plasmodium DHFR enzyme, QN254 does not show the adequate therapeutic index to justify its further development as a single agent.

A small molecule fusion inhibitor of dengue virus.

The dengue virus envelope protein plays an essential role in viral entry by mediating fusion between the viral and host membranes. The crystal structure of the envelope protein shows a pocket (located at a "hinge" between Domains I and II) that can be occupied by ligand n-octyl-beta-D-glucoside (betaOG). Compounds blocking the betaOG pocket are thought to interfere with conformational changes in the envelope protein that are essential for fusion. Two fusion assays were developed to examine the anti-fusion activities of compounds. The first assay measures the cellular internalization of propidium iodide upon membrane fusion. The second assay measures the protease activity of trypsin upon fusion between dengue virions and trypsin-containing liposomes. We performed an in silico virtual screening for small molecules that can potentially bind to the betaOG pocket and tested these candidate molecules in the two fusion assays. We identified one compound that inhibits dengue fusion in both assays with an IC(50) of 6.8 microM and reduces viral titers with an EC(50) of 9.8 microM. Time-of-addition experiments showed that the compound was only active when present during viral infection but not when added 1h later, in agreement with a mechanism of action through fusion inhibition.

Fishing the target of antitubercular compounds: in silico target deconvolution model development and validation.

An in silico target prediction protocol for antitubercular (antiTB) compounds has been proposed in this work. This protocol is the extension of a recently published 'domain fishing model' (DFM), validating its predicted targets on a set of 42 common antitubercular drugs. For the 23 antiTB compounds of the set which are directly linked to targets (see text for definition), the DFM exhibited a very good target prediction accuracy of 95%. For 19 compounds indirectly linked to targets also, a reasonable pathway/embedded pathway prediction accuracy of 84% was achieved. Since mostly eukaryotic ligand binding data was used for the DFM generation, the high target prediction accuracy for prokaryotes (which is an extrapolation from the training data) was unexpected and provides an additional proof of concept of the DFM. To estimate the general applicability of the model, ligand-target coverage analysis was performed. Here, it was found that, although the DFM only modestly covers the entire TB proteome (32% of all proteins), it captures 70% of the proteome subset targeted by 42 common antiTB compounds, which is in agreement with the good predictive ability of the DFM for the targets of the compounds chosen here. In a prospective validation, the model successfully predicted the targets of new antiTB compounds, CBR-2092 and Amiclenomycin. Together, these findings suggest that in silico target prediction tools may be a useful supplement to existing, experimental target deconvolution strategies.

A small-molecule dengue virus entry inhibitor.

The incidence of dengue fever epidemics has increased dramatically over the last few decades. However, no vaccine or antiviral therapies are available. Therefore, the need for safe and effective antiviral drugs has become imperative. The entry of dengue virus into a host cell is mediated by its major envelope (E) protein. The crystal structure of the E protein reveals a hydrophobic pocket that is presumably important for low-pH-mediated membrane fusion. High-throughput docking with this hydrophobic pocket was performed, and hits were evaluated in cell-based assays. Compound 6 was identified as one of the inhibitors and had an average 50% effective concentration of 119 nM against dengue virus serotype 2 in a human cell line. Mechanism-of-action studies demonstrated that compound 6 acts at an early stage during dengue virus infection. It arrests dengue virus in vesicles that colocalize with endocytosed dextran and inhibits NS3 expression. The inhibitors described in this report can serve as molecular probes for the study of the entry of flavivirus into host cells.

Global Bayesian models for the prioritization of antitubercular agents.

To aid the creation of novel antituberculosis (antiTB) compounds, Bayesian models were derived and validated on a data set of 3779 compounds which have been measured for minimum inhibitory concentration (MIC) in the Mycobacterium tuberculosis H37Rv strain. The model development and validation involved exploring six different training sets and 15 fingerprint types which resulted in a total of 90 models, with active compounds defined as those with MIC < 5 microM. The best model was derived using Extended Class Fingerprints of maximum diameter 12 (ECFP_12) and a few global descriptors on a training set derived using Functional Class Fingerprints of maximum diameter 4 (FCFP_4). This model demonstrated very good discriminant ability in general, with excellent discriminant statistics for the training set (total accuracy: 0.968; positive recall: 0.967) and a good predictive ability for the test set (total accuracy: 0.869; positive recall: 0.789). The good predictive ability was maintained when the model was applied to a well-separated test set of 2880 compounds derived from a commercial database (total accuracy: 0.73; positive recall: 0.72). The model revealed several conserved substructures present in the active and inactive compounds which are believed to have incremental and detrimental effects on the MIC, respectively. Strategies for enhancing the repertoire of antiTB compounds with the model, including virtual screening of large databases and combinatorial library design, are proposed.

Lipiarmycin targets RNA polymerase and has good activity against multidrug-resistant strains of Mycobacterium tuberculosis.

OBJECTIVES: The aim of this study was to determine the in vitro activity of lipiarmycin against drug-resistant strains of Mycobacterium tuberculosis (MTB) and to establish the resistance mechanism of MTB against lipiarmycin using genetic approaches. METHODS: MIC values were measured against a panel of drug-resistant strains of MTB using the broth microdilution method. Spontaneous lipiarmycin-resistant mutants of MTB were tested for cross-resistance to standard anti-TB drugs, and their rpoB and rpoC genes were sequenced to identify mutations. RESULTS: Lipiarmycin exhibited excellent inhibitory activity against multidrug-resistant strains of MTB with MIC values of <0.1 mg/L. Sequence analysis of the rpoB and rpoC genes from spontaneous lipiarmycin-resistant mutants of MTB revealed that missense mutations in these genes caused resistance to lipiarmycin. Although both lipiarmycin and rifampicin are known to inhibit the bacterial RNA polymerase, the sites of mutation in the rpoB gene were found to be different in MTB strains resistant to these inhibitors. Whereas all six rifampicin-resistant MTB strains tested had mutation in the 81 bp hotspot region of the rpoB gene spanning codons 507-533, 16 of 18 lipiarmycin-resistant strains exhibited mutation between codons 977 and 1150. The remaining two lipiarmycin-resistant strains had mutation in the rpoC gene. CONCLUSIONS: Lipiarmycin has excellent bactericidal activity against MTB and lacks cross-resistance to standard anti-TB drugs. Furthermore, rifampicin-resistant strains remained fully susceptible to lipiarmycin, and none of the lipiarmycin-resistant MTB strains became resistant to rifampicin, highlighting the lack of cross-resistance.

Yellow fever virus NS3 protease: peptide-inhibition studies.

A recombinant form of yellow fever virus (YFV) NS3 protease, linked via a nonapeptide to the minimal NS2B co-factor sequence (CF40-gly-NS3pro190), was expressed in Escherichia coli and shown to be catalytically active. It efficiently cleaved the fluorogenic tetrapeptide substrate Bz-norleucine-lysine-arginine-arginine-AMC, which was previously optimized for dengue virus NS2B/3 protease. A series of small peptidic inhibitors based on this substrate sequence readily inhibited its enzymic activity. To understand the structure-activity relationship of the inhibitors, they were docked into a homology model of the YFV NS2B/NS3 protease structure. The results revealed that the P1 and P2 positions are most important for inhibitor binding, whilst the P3 and P4 positions have much less effect. These findings indicate that the characteristics of YFV protease are very similar to those reported for dengue and West Nile virus proteases, and suggest that pan-flavivirus NS3 protease drugs may be developed for flaviviral diseases.

Peptide inhibitors of West Nile NS3 protease: SAR study of tetrapeptide aldehyde inhibitors.

A series of inhibitors related to the benzoyl-norleucine-lysine-arginine-arginine (Bz-nKRR) tetrapeptide aldehyde was synthesized. When evaluated against the West Nile virus (WNV) NS3 protease, the measured IC(50) ranges from approximately 1 to 200 microM. Concurrently, a modeling study using the recently published crystal structure of the West Nile NS3/NS2B protease complex (pdb code 2FP7) was conducted. We found that the crystal structure is relevant in explaining the observed SAR for this series of tetrapeptides, with the S1 and S2 pockets being the key peptide recognition sites. In general, a residue capable of both pi-stacking and hydrogen bonding is favored in the S1 pocket, while a positively charged residue is preferred in the S2 pocket. This study not only confirms the importance of the NS2B domain in substrate-based inhibitor binding of WNV, it also suggests that the crystal structure would provide useful guidance in the drug discovery process of related Flavivirus proteases, given the high degree of homology.

Peptide inhibitors of dengue virus NS3 protease. Part 2: SAR study of tetrapeptide aldehyde inhibitors.

With the aim of discovering potent and selective dengue NS3 protease inhibitors, we systematically synthesized and evaluated a series of tetrapeptide aldehydes based on lead aldehyde 1 (Bz-Nle-Lys-Arg-Arg-H, K(i)=5.8 microM). In general, we observe that interactions of P(2) side chain are more important than P(1) followed by P(3) and P(4). Tripeptide and dipeptide aldehyde inhibitors also show low micromolar activity. Additionally, an effective non-basic, uncharged replacement of P(1) Arg is identified.

Finding new medicines for flaviviral targets.

With the incidence of dengue fever increasing all over the world, there is an urgent need for therapies. While drug discovery for any disease is a long and difficult process with uncertain success, dengue fever poses an additional complication in that most of the target patient population is young and lives in developing countries with very limited health care budgets. Recent progress in drug discovery for dengue and an analysis of approaches toward hepatitis C virus (HCV) therapeutics suggest that NS5 polymerase is the most promising target for dengue. Moreover such inhibitors may be useful for several other flaviviral diseases. NS3 proteases will be more challenging targets, especially if oral delivery is desired. Recent work has shown that potent inhibitors can be designed readily, but optimization of pharmacokinetic parameters will probably be a long an arduous task, especially since the primary binding pockets prefer to bind basic amino acids. NS3 helicase can also be considered a viable drug target for flaviviral diseases. It has however proved to be a challenging for HCV and selectivity issues versus human helicases must be overcome.

Dissociation of alkaliated alanine in the gas phase: the role of the metal cation.

The dissociation of prototypical metal-cationized amino acid complexes, namely, alkaliated alanine ([Ala+M]+, M+ = Li+, Na+, K+), was studied by energy-resolved tandem mass spectrometry with an ion-trap mass analyzer and by density functional theory. Dissociation leads to formation of fragment ions arising from the loss of small neutrals, such as H2O, CO, NH3, (CO+NH3), and the formation of Na+/K+. The order of appearance threshold voltages for different dissociation pathways determined experimentally is consistent with the order of critical energies (energy barriers) obtained theoretically, and this provides the necessary confidence in both experimental and theoretical results. Although not explicitly involved in the reaction, the alkali metal cation plays novel and important roles in the dissociation of alkaliated alanine. The metal cation not only catalyzes the dissociation (via the formation of loosely bound ion-molecule complexes and by stabilizing the more polar intermediates and transition structures), but also affects the dissociation mechanisms, as the cation can alter the shape of the potential energy surfaces. This compression/expansion of the potential energy surface as a function of the alkali metal cation is discussed in detail, and how this affects the competitive loss of H2O versus CO/(CO+NH3) from [Ala+M]+ is illustrated. The present study provides new insights into the origin of the competition between various dissociation channels of alkaliated amino acid complexes.

Competition between pi and non-pi cation-binding sites in aromatic amino acids: a theoretical study of alkali metal cation (Li+, Na+, K+)-phenylalanine complexes.

To understand the cation-pi interaction in aromatic amino acids and peptides, the binding of M(+) (where M(+) = Li(+), Na(+), and K(+)) to phenylalanine (Phe) is studied at the best level of density functional theory reported so far. The different modes of M(+) binding show the same order of binding affinity (Li(+)>Na(+)>K(+)), in the approximate ratio of 2.2:1.5:1.0. The most stable binding mode is one in which the M(+) is stabilized by a tridentate interaction between the cation and the carbonyl oxygen (O[double bond]C), amino nitrogen (--NH(2)), and aromatic pi ring; the absolute Li(+), Na(+), and K(+) affinities are estimated theoretically to be 275, 201, and 141 kJ mol(-1), respectively. Factors affecting the relative stabilities of various M(+)-Phe binding modes and conformers have been identified, with ion-dipole interaction playing an important role. We found that the trend of pi and non-pi cation bonding distances (Na(+)-pi>Na(+)-N>Na(+)-O and K(+)-pi>K(+)-N>K(+)-O) in our theoretical Na(+)/K(+)-Phe structures are in agreement with the reported X-ray crystal structures of model synthetic receptors (sodium and potassium bound lariat ether complexes), even though the average alkali metal cation-pi distance found in the crystal structures is longer. This difference between the solid and the gas-phase structures can be reconciled by taking the higher coordination number of the cations in the lariat ether complexes into account.

Experimental validation of theoretical potassium and sodium cation affinities of amides by mass spectrometric kinetic method measurements.

In this study the theoretical Gaussian-2 K(+)/Na(+) binding affinities (enthalpies) at 0 K (in kJ mol(-1)) for six amides in the order: formamide (109.2/138.5) < N-methylformamide (117.7/148.6) < acetamide (118.7/149.5) < N,N-dimethylformamide (123.9/156.4) < N-methylacetamide (125.6/157.7) < N,N-dimethylacetamide (129.2/162.6), reported previously (Siu et al., J. Chem. Phys. 2001; 114: 7045-7051), were validated experimentally by mass spectrometric kinetic method measurements. By monitoring the collision-induced dissociation (CID) of K(+)/Na(+)-bound heterodimers of the amides, the relative affinities were shown to be accurate to within +/-2 kJ mol(-1). With these six theoretical K(+)/Na(+) binding affinities as reference values, the absolute K(+)/Na(+) affinities of imidazole, 1-methylimidazole, pyridazine and 1,2-dimethoxyethane were determined by the extended kinetic method, and found to be consistent (to within +/-9 kJ mol(-1)) with literature experimental values obtained by threshold-CID, equilibrium high-pressure mass spectrometry, and Fourier transform ion cyclotron resonance/ligand-exchange equilibrium methods. A self-consistent resolution is proposed for the inconsistencies in the relative order of K(+)/Na(+) affinities of amides reported in the literature. These two sets of validated K(+) and Na(+) affinity values are useful as reference values in kinetic method measurements of K(+)/Na(+) affinity of model biological ligands, such as the K(+) affinities of aliphatic amino acids.

Absolute potassium cation affinities (PCAs) in the gas phase.

The potassium cation affinities (PCAs) of 136 ligands (20 classes) in the gas phase were established by hybrid density functional theory calculations (B3-LYP with the 6-311+G(3df,2p) basis set). For these 136 ligands, 70 experimental values are available for comparison. Except for five specific PCA values-those of phenylalanine, cytosine, guanine, adenine (kinetic-method measurement), and Me(2)SO (by high-pressure mass spectrometric equilibrium measurement)-our theoretical estimates and the experimental affinities are in excellent agreement (mean absolute deviation (MAD) of 4.5 kJ mol(-1)). Comparisons with previously reported theoretical PCAs are also made. The effect of substituents on the modes of binding and the PCAs of unsubstituted parent ligands are discussed. Linear relations between Li+/Na+ and K+ affinities suggest that for the wide range of ligands studied here, the nature of binding between the cations and a given ligand is similar, and this allows the estimation of PCAs from known Li+ and/or Na+ affinities. Furthermore, empirical equations relating the PCAs of ligands with their dipole moments, polarizabilities (or molecular weights), and the number of binding sites were established. Such equations offer a simple method for estimating the PCAs of ligands not included in the present study.

A theoretical study of potassium cation binding to glycylglycine (GG) and alanylalanine (AA) dipeptides.

By combining Monte Carlo conformational search technique with high-level density functional calculations, the geometry and energetics of K(+) interaction with glycylglycine (GG) and alanylalanine (AA) were obtained for the first time. The most stable K(+)-GG and K(+)-AA complexes are in the charge-solvated (CS) form with K(+) bound to the carbonyl oxygens of the peptide backbone, and the estimated 0 K binding affinities (DeltaH(0)) are 152 and 157 kJ mol(-1), respectively. The K(+) ion is in close alignment with the molecular dipole moment vector of the bound ligand, that is, electrostatic ion-dipole interaction is the key stabilizing factor in these complexes. Furthermore, the strong ion-dipole interaction between K(+) and the amide carbonyl oxygen atom of the peptide bond is important in determining the relative stabilities of different CS binding modes. The most stable zwitterionic (ZW) complex involves protonation at the amide carbonyl oxygen atom and is approximately 48 kJ mol(-1) less stable than the most stable CS form. The usefulness of proton affinity (PA) as a criterion for estimating the relative stability of ZW versus CS binding modes is examined. The effect of chain length and the nature of metal cations on cation-dipeptide interactions are discussed. Based on results of this study, the interaction of K(+) with longer peptides consisting of aliphatic amino acids are rationalized.

Experimental validation of Gaussian-3 lithium cation affinities of amides: implications for the gas-phase lithium cation basicity scale.

Using a refined Gaussian-3 (G3) protocol, the highest level of ab initio calculations reported so far, we have established the Li+ cation binding enthalpy (affinity) at 0 K (in kJ mol-1) for formamide (195.7), N-methylformamide (209.2), N,N'-dimethylformamide (220.0), acetamide (211.7), N-methylacetamide (222.5), and N,N'-dimethylacetamide (230.1), with an estimated maximum uncertainty of +/-8 kJ mol-1. With these six theoretical lithium cation binding affinities as reference values, the absolute Li+ affinities of imidazole and dimethoxyethane were determined by the extended kinetic method, and by adopting the statistical data treatment protocol recently proposed by Armentrout. The Li+ affinities obtained for these two ligands are in good agreement (within 6 kJ mol-1) with recent values determined by the threshold collision-induced dissociation method, and consistent with the Li+ basicity values first reported by Taft and co-workers in 1990. Our study confirms that the previously suggested, and recently implemented, downward revision of Taft's original basicity scale by 10.9 kJ mol-1 is justified for ligands with revised basicities less than 151 kJ mol-1. However, for selected ligands with Li+ basicities greater than 151 kJ mol-1, including some of the six amides studied in this work, the reported discrepancy between theoretical and experimental estimates in the revised Li+ basicity scale of Burk et al. is likely to arise from experimental uncertainties.