A two-state homology model of the hERG K+ channel: application to ligand binding

Homology models based on available K+ channel structures have been used to construct a multiple state representation of the hERG cardiac K+ channel. These states are used to capture the flexibility of the channel. We show that this flexibility is essential in order to correctly model the binding affinity of a set of diverse ligands. Using this multiple state approach, a binding affinity model was constructed for set of known hERG channel binders. The predicted pIC50s are in good agreement with experiment (RMSD: 0.56 kcal/mol). In addition, these calculations provide structures for the bound ligands that are consistent with published mutation studies. These computed ligand bound complex structures can be used to guide synthesis of analogs with reduced hERG liability.     

The open-access dataset for insilico cardiotoxicity prediction system

Drug cardiotoxicity is one of the main reasons of fatal drug related problem events and the subsequent withdrawals. Therefore, its early assessment is a crucial element of the drug development process. For the drug driven hERG inhibition assessment, which is assumed to be the main reason for toxicity, in vitro techniques are used. Gold standards are based on the Patch Clamp method with the use of various cell models but due to its low throughput, insilico models have become more appreciated. To develop a reliable empirical QSAR model, wide dataset containing a variety of cases has to be available. In this article, a freely available for download, set of data is described. It is based on literature peer-reviewed reports and contains hERG inhibition information expressed as IC50 value for 263 molecules described in 642 records. All studies were done with the use of three cell models (XO, CHO, HEK) and other elements describe the electrophysiological settings of the in vitro study. The above mentioned set was used for the successful development of the predictive models.     

Na+ permeation and block of hERG potassium channels

The inactivation gating of hERG channels is important for the channel function and drug-channel interaction. Whereas hERG channels are highly selective for K+, we have found that inactivated hERG channels allow Na+ to permeate in the absence of K+. This provides a new way to directly monitor and investigate hERG inactivation. By using whole cell patch clamp method with an internal solution containing 135 mM Na+ and an external solution containing 135 mM NMG+, we recorded a robust Na+ current through hERG channels expressed in HEK 293 cells. Kinetic analyses of the hERG Na+ and K+ currents indicate that the channel experiences at least two states during the inactivation process, an initial fast, less stable state followed by a slow, more stable state. The Na+ current reflects Na+ ions permeating through the fast inactivated state but not through the slow inactivated state or open state. Thus the hERG Na+ current displayed a slow inactivation as the channels travel from the less stable, fast inactivated state into the more stable, slow inactivated state. Removal of fast inactivation by the S631A mutation abolished the Na+ current. Moreover, acceleration of fast inactivation by mutations T623A, F627Y, and S641A did not affect the hERG Na+ current, but greatly diminished the hERG K+ current. We also found that external Na+ potently blocked the hERG outward Na+ current with an IC50 of 3.5 mM. Mutations in the channel pore and S6 regions, such as S624A, F627Y, and S641A, abolished the inhibitory effects of external Na+ on the hERG Na+ current. Na+ permeation and blockade of hERG channels provide novel ways to extend our understanding of the hERG gating mechanisms.     

Saxitoxin is a gating modifier of HERG K+ channels

Potassium (K+) channels mediate numerous electrical events in excitable cells, including cellular membrane potential repolarization. The hERG K+ channel plays an important role in myocardial repolarization, and inhibition of these K+ channels is associated with long QT syndromes that can cause fatal cardiac arrhythmias. In this study, we identify saxitoxin (STX) as a hERG channel modifier and investigate the mechanism using heterologous expression of the recombinant channel in HEK293 cells. In the presence of STX, channels opened slower during strong depolarizations, and they closed much faster upon repolarization, suggesting that toxin-bound channels can still open but are modified, and that STX does not simply block the ion conduction pore. STX decreased hERG K+ currents by stabilizing closed channel states visualized as shifts in the voltage dependence of channel opening to more depolarized membrane potentials. The concentration dependence for steady-state modification as well as the kinetics of onset and recovery indicate that multiple STX molecules bind to the channel. Rapid application of STX revealed an apparent "agonist-like" effect in which K+ currents were transiently increased. The mechanism of this effect was found to be an effect on the channel voltage-inactivation relationship. Because the kinetics of inactivation are rapid relative to activation for this channel, the increase in K+ current appeared quickly and could be subverted by a decrease in K+ currents due to the shift in the voltage-activation relationship at some membrane potentials. The results are consistent with a simple model in which STX binds to the hERG K+ channel at multiple sites and alters the energetics of channel gating by shifting both the voltage-inactivation and voltage-activation processes. The results suggest a novel extracellular mechanism for pharmacological manipulation of this channel through allosteric coupling to channel gating.     

Predicting the potency of hERG K<Superscript>+</Superscript> channel inhibition by combining 3D-QSAR pharmacophore and 2D-QSAR models

Blockade of the hERG K⁺ channel has been identified as the most important mechanism of QT interval prolongation and thus inducing cardiac risk. In this work, an ensemble of 3D-QSAR pharmacophore models was constructed to provide insight into the determinants of the interactions between the hERG K⁺ channel and channel inhibitors. To predict hERG inhibitory activities, the predicted values from the ensemble of models were averaged, and the results thus obtained showed that the predictive ability of the combined 3D-QSAR pharmacophore model was greater that those of the individual models. Also, using the same training and test sets, a 2D-QSAR model based on a heuristic machine-learning method was developed in order to analyze the physicochemical characters of hERG inhibitors. The models indicated that the inhibitors have certain key inhibitory features in common, including hydrophobicity, aromaticity, and flexibility. A final model was developed by combining the combined 3D-QSAR pharmacophore with the 2D-QSAR model, and this final model outperformed any other individual model, showing the highest predictive ability and the lowest deviation. This model can not only predict hERG inhibitory potency accurately, thus allowing fast cardiac safety evaluation, but it provides an effective tool for avoiding hERG inhibitory liability and thus enhanced cardiac risk in the design and optimization of new chemical entities.     

Drug block of the hERG potassium channel: insight from modeling

Many commonly used, structurally diverse, drugs block the human ether-a-go-go-related gene (hERG) K(+) channel to cause acquired long QT syndrome, which can lead to sudden death via lethal cardiac arrhythmias. This undesirable side effect is a major hurdle in the development of safe drugs. To gain insight about the structure of hERG and the nature of drug block we have produced structural models of the channel pore domain, into each of which we have docked a set of 20 hERG blockers. In the absence of an experimentally determined three-dimensional structure of hERG, each of the models was validated against site-directed mutagenesis data. First, hERG models were produced of the open and closed channel states, based on homology with the prokaryotic K(+) channel crystal structures. The modeled complexes were in partial agreement with the mutagenesis data. To improve agreement with mutagenesis data, a KcsA-based model was refined by rotating the four copies of the S6 transmembrane helix half a residue position toward the C-terminus, so as to place all residues known to be involved in drug binding in positions lining the central cavity. This model produces complexes that are consistent with mutagenesis data for smaller, but not larger, ligands. Larger ligands could be accommodated following refinement of this model by enlarging the cavity using the inherent flexibility about the glycine hinge (Gly648) in S6, to produce results consistent with the experimental data for the majority of ligands tested.     

Predictive models in ecology: Comparison of performances and assessment of applicability

Ecological systems are governed by complex interactions which are mainly nonlinear. In order to capture the inherent complexity and nonlinearity of ecological, and in general biological systems, empirical models recently gained popularity. However, although these models, particularly connectionist approaches such as multilayered backpropagation networks, are commonly applied as predictive models in ecology to a wide variety of ecosystems and questions, there are no studies to date aiming to assess the performance, both in terms of data fitting and generalizability, and applicability of empirical models in ecology. Our aim is hence to provide an overview for nature of the wide range of the data sets and predictive variables, from both aquatic and terrestrial ecosystems with different scales of time-dependent dynamics, and the applicability and robustness of predictive modeling methods on such data sets by comparing different empirical modeling approaches. The models used in this study range from predicting the occurrence of submerged plants in shallow lakes to predicting nest occurrence of bird species from environmental variables and satellite images. The methods considered include k-nearest neighbor (k-NN), linear and quadratic discriminant analysis (LDA and QDA), generalized linear models (GLM) feedforward multilayer backpropagation networks and pseudo-supervised network ARTMAP.Our results show that the predictive performances of the models on training data could be misleading, and one should consider the predictive performance of a given model on an independent test set for assessing its predictive power. Moreover, our results suggest that for ecosystems involving time-dependent dynamics and periodicities whose frequency are possibly less than the time scale of the data considered, GLM and connectionist neural network models appear to be most suitable and robust, provided that a predictive variable reflecting these time-dependent dynamics included in the model either implicitly or explicitly. For spatial data, which does not include any time-dependence comparable to the time scale covered by the data, on the other hand, neighborhood based methods such as k-NN and ARTMAP proved to be more robust than other methods considered in this study. In addition, for predictive modeling purposes, first a suitable, computationally inexpensive method should be applied to the problem at hand a good predictive performance of which would render the computational cost and efforts associated with complex variants unnecessary.     

Probing the outer mouth structure of the HERG channel with peptide toxin footprinting and molecular modeling

Previous studies have shown that the unusually long S5-P linker lining human ether a-go-go related gene's (hERG's) outer vestibule is critical for its channel function: point mutations at high-impact positions here can interfere with the inactivation process and, in many cases, also reduce the pore's K+ selectivity. Because no data are available on the equivalent region in the available K channel crystal structures to allow for homology modeling, we used alternative approaches to model its three-dimensional structure. The first part of this article describes mutant cycle analysis used to identify residues on hERG's outer vestibule that interact with specific residues on the interaction surface of BeKm-1, a peptide toxin with known NMR structure and a high binding affinity to hERG. The second part describes molecular modeling of hERG's pore domain. The transmembrane region was modeled after the crystal structure of KvAP pore domain. The S5-P linker was docked to the transmembrane region based on data from previous NMR and mutagenesis experiments, as well as a set of modeling criteria. The models were further restrained by contact points between hERG's outer vestibule and the bound BeKm-1 toxin molecule deduced from the mutant cycle analysis. Based on these analyses, we propose a working model for the open conformation of the outer vestibule of the hERG channel, in which the S5-P linkers interact with the pore loops to influence ion flux through the pore.     

Prediction of hERG potassium channel affinity by traditional and hologram qSAR methods

Traditional and hologram QSAR (HQSAR) models were developed for the prediction of hERG potassium channel affinities. The models were validated on three different test sets including compounds with published patch-clamp IC(50) data and two subsets from the World Drug Index (compounds indicated to have ECG modifying adverse effect and drugs marked to be approved, respectively). Discriminant analysis performed on the full set of hERG data resulted in a traditional QSAR model that classified 83% of actives and 87% of inactives correctly. Analysis of our HQSAR model revealed it to be predictive in both IC(50) and discrimination studies.     

Investigation and identification of protein carbonylation sites based on position-specific amino acid composition and physicochemical features

Background

Protein carbonylation, an irreversible and non-enzymatic post-translational modification (PTM), is often used as a marker of oxidative stress. When reactive oxygen species (ROS) oxidized the amino acid side chains, carbonyl (CO) groups are produced especially on Lysine (K), Arginine (R), Threonine (T), and Proline (P). Nevertheless, due to the lack of information about the carbonylated substrate specificity, we were encouraged to develop a systematic method for a comprehensive investigation of protein carbonylation sites.

Results

After the removal of redundant data from multipe carbonylation-related articles, totally 226 carbonylated proteins in human are regarded as training dataset, which consisted of 307, 126, 128, and 129 carbonylation sites for K, R, T and P residues, respectively. To identify the useful features in predicting carbonylation sites, the linear amino acid sequence was adopted not only to build up the predictive model from training dataset, but also to compare the effectiveness of prediction with other types of features including amino acid composition (AAC), amino acid pair composition (AAPC), position-specific scoring matrix (PSSM), positional weighted matrix (PWM), solvent-accessible surface area (ASA), and physicochemical properties. The investigation of position-specific amino acid composition revealed that the positively charged amino acids (K and R) are remarkably enriched surrounding the carbonylated sites, which may play a functional role in discriminating between carbonylation and non-carbonylation sites. A variety of predictive models were built using various features and three different machine learning methods. Based on the evaluation by five-fold cross-validation, the models trained with PWM feature could provide better sensitivity in the positive training dataset, while the models trained with AAindex feature achieved higher specificity in the negative training dataset. Additionally, the model trained using hybrid features, including PWM, AAC and AAindex, obtained best MCC values of 0.432, 0.472, 0.443 and 0.467 on K, R, T and P residues, respectively.

Conclusion

When comparing to an existing prediction tool, the selected models trained with hybrid features provided a promising accuracy on an independent testing dataset. In short, this work not only characterized the carbonylated substrate preference, but also demonstrated that the proposed method could provide a feasible means for accelerating preliminary discovery of protein carbonylation.

     

Interaction between the cardiac rapidly (IKr) and slowly (IKs) activating delayed rectifier potassium channels revealed by low K+-induced hERG endocytic degradation

Cardiac repolarization is controlled by the rapidly (I(Kr)) and slowly (I(Ks)) activating delayed rectifier potassium channels. The human ether-a-go-go-related gene (hERG) encodes I(Kr), whereas KCNQ1 and KCNE1 together encode I(Ks). Decreases in I(Kr) or I(Ks) cause long QT syndrome (LQTS), a cardiac disorder with a high risk of sudden death. A reduction in extracellular K(+) concentration ([K(+)](o)) induces LQTS and selectively causes endocytic degradation of mature hERG channels from the plasma membrane. In the present study, we investigated whether I(Ks) compensates for the reduced I(Kr) under low K(+) conditions. Our data show that when hERG and KCNQ1 were expressed separately in human embryonic kidney (HEK) cells, exposure to 0 mM K(+) for 6 h completely eliminated the mature hERG channel expression but had no effect on KCNQ1. When hERG and KCNQ1 were co-expressed, KCNQ1 significantly delayed 0 mM K(+)-induced hERG reduction. Also, hERG degradation led to a significant reduction in KCNQ1 in 0 mM K(+) conditions. An interaction between hERG and KCNQ1 was identified in hERG+KCNQ1-expressing HEK cells. Furthermore, KCNQ1 preferentially co-immunoprecipitated with mature hERG channels that are localized in the plasma membrane. Biophysical and pharmacological analyses indicate that although hERG and KCNQ1 closely interact with each other, they form distinct hERG and KCNQ1 channels. These data extend our understanding of delayed rectifier potassium channel trafficking and regulation, as well as the pathology of LQTS.     

Fish distribution predictions from different points of view: comparing associative neural networks,geostatistics and regression models

Accurate prediction of species distributions based on sampling and environmental data is essential for further scientific analysis, such as stock assessment, detection of abundance fluctuation due to climate change or overexploitation, and to underpin management and legislation processes. The evolution of computer science and statistics has allowed the development of sophisticated and well-established modelling techniques as well as a variety of promising innovative approaches for modelling species distribution. The appropriate selection of modelling approach is crucial to the quality of predictions about species distribution. In this study, modelling techniques based on different approaches are compared and evaluated in relation to their predictive performance, utilizing fish density acoustic data. Generalized additive models and mixed models amongst the regression models, associative neural networks (ANNs) and artificial neural networks ensemble amongst the artificial neural networks and ordinary kriging amongst the geostatistical techniques are applied and evaluated. A verification dataset is used for estimating the predictive performance of these models. A combination of outputs from the different models is applied for prediction optimization to exploit the ability of each model to explain certain aspects of variation in species acoustic density. Neural networks and especially ANNs appear to provide more accurate results in fitting the training dataset while generalized additive models appear more flexible in predicting the verification dataset. The efficiency of each technique in relation to certain sampling and output strategies is also discussed.     

A quantitative assessment of hERG liability as a function of lipophilicity

The impact of lipophilicity as a factor contributing to hERG potency is assessed for a large dataset of compounds of differing ionisation type. This dataset is derived from compounds tested in the IonWorks-based in vitro electrophysiology hERG assay at AstraZeneca. Using logistic regression, a quantification of the risk associated with increasing lipophilicity is presented. The anticipated differences between acidic, basic and neutral compounds are apparent in the data but lipophilicity is shown to be a stronger driver for hERG potency than might have been expected. Simple rules defining target lipophilicity values for minimizing hERG liability are derived.     

Strategies to develop radiomics and machine learning models for lung cancer stage and histology prediction using small data samples

Predictive models based on radiomics and machine-learning (ML) need large and annotated datasets for training, often difficult to collect. We designed an operative pipeline for model training to exploit data already available to the scientific community. The aim of this work was to explore the capability of radiomic features in predicting tumor histology and stage in patients with non-small cell lung cancer (NSCLC).We analyzed the radiotherapy planning thoracic CT scans of a proprietary sample of 47 subjects (L-RT) and integrated this dataset with a publicly available set of 130 patients from the MAASTRO NSCLC collection (Lung1). We implemented intra- and inter-sample cross-validation strategies (CV) for evaluating the ML predictive model performances with not so large datasets.We carried out two classification tasks: histology classification (3 classes) and overall stage classification (two classes: stage I and II). In the first task, the best performance was obtained by a Random Forest classifier, once the analysis has been restricted to stage I and II tumors of the Lung1 and L-RT merged dataset (AUC = 0.72 ± 0.11). For the overall stage classification, the best results were obtained when training on Lung1 and testing of L-RT dataset (AUC = 0.72 ± 0.04 for Random Forest and AUC = 0.84 ± 0.03 for linear-kernel Support Vector Machine).According to the classification task to be accomplished and to the heterogeneity of the available dataset(s), different CV strategies have to be explored and compared to make a robust assessment of the potential of a predictive model based on radiomics and ML.     

hERG K+ channel blockade by the antipsychotic drug thioridazine: An obligatory role for the S6 helix residue F656

The phenothiazine antipsychotic agent thioridazine has been linked with prolongation of the QT interval on the electrocardiogram, ventricular arrhythmias, and sudden death. Although thioridazine is known to inhibit cardiac hERG K(+) channels there is little mechanistic information on this action. We have investigated in detail hERG K(+) channel current (I(hERG)) blockade by thioridazine and identified a key molecular determinant of blockade. Whole-cell I(hERG) measurements were made at 37 degrees C from human embryonic kidney (HEK-293) cells expressing wild-type and mutant hERG channels. Thioridazine inhibited I(hERG) tails at -40mV following a 2s depolarization to +20mV with an IC(50) value of 80nM. Comparable levels of I(hERG) inhibition were seen with physiological command waveforms (ventricular and Purkinje fibre action potentials). Thioridazine block of I(hERG) was only weakly voltage-dependent, though the time dependence of I(hERG) inhibition indicated contingency of blockade upon channel gating. The S6 helix point mutation F656A almost completely abolished, and the Y652A mutation partially attenuated, I(hERG) inhibition by thioridazine. In summary, thioridazine is one of the most potent hERG K(+) channel blockers amongst antipsychotics, exhibiting characteristics of a preferential open/activated channel blocker and binding at a high affinity site in the hERG channel pore.