Biomedical subjects

Hongmao Sun

Publications and source records attributed to Hongmao Sun.

8 recordsLinked to original sources

PubMed2006

Selective small-molecule inhibitor reveals critical mitotic functions of human CDK1.

CDK1 is a nonredundant cyclin-dependent kinase (CDK) with an essential role in mitosis, but its multiple functions still are poorly understood at a molecular level. Here we identify a selective small-molecule inhibitor of CDK1 that reversibly arrests human cells at the G(2)/M border of the cell cycle and allows for effective cell synchronization in early mitosis. Inhibition of CDK1 during cell division revealed that its activity is necessary and sufficient for maintaining the mitotic state of the cells, preventing replication origin licensing and premature cytokinesis. Although CDK1 inhibition for up to 24 h is well tolerated, longer exposure to the inhibitor induces apoptosis in tumor cells, suggesting that selective CDK1 inhibitors may have utility in cancer therapy.

Apoptosis↗

PubMed2006

An accurate and interpretable bayesian classification model for prediction of HERG liability.

Drug-induced QT interval prolongation has been identified as a critical side-effect of non-cardiovascular therapeutic agents and has resulted in the withdrawal of many drugs from the market. As almost all cases of drug-induced QT prolongation can be traced to the blockade of a voltage-dependent potassium ion channel encoded by the hERG (the human ether-à-go-go-related gene), early identification of potential hERG channel blockers will decrease the risk of cardiotoxicity-induced attritions in the later and more expensive development stage. Presented herein is a naive Bayes classifier to categorize hERG blockers into active and inactive classes, by using a universal, generic molecular descriptor system.1 The naive Bayes classifier was built from a training set containing 1979 corporate compounds, and exhibited an ROC accuracy of 0.87. The model was validated on an external test set of 66 drugs, of which 58 were correctly classified. The cumulative probabilities reflected the confidence of prediction and were proven useful for the identification of hERG blockers. Relative performance was compared for two classifiers constructed from either an atom-type-based molecular descriptor or the long range functional class fingerprint descriptor FCFP_6. The combination of an atom-typing descriptor and the naive Bayes classification technique enables the interpretation of the resulting model, which offers extra information for the design of compounds free of undesirable hERG activity.

Bayes Theorem↗

PubMed2006

Utilizing peptide structures as keys for unlocking challenging targets.

Three-dimensional structures of protein targets have proven to be extremely valuable for modern drug design and discovery. For cases where the structure of the protein is unattainable, such as G-protein coupled receptors (GPCRs), structural information on active ligands is still useful and helpful for deciphering the geometrical and chemical features of the active site. Peptides, constructed from easy-to-form amide backbones and featuring variable side-chains, have an inherent advantage in generating rapid quantitative structure-activity relationships (QSAR). Given the fact that peptides are natural ligands for many protein targets, structural investigation of a series of related peptides, typically carried out via nuclear magnetic resonance (NMR), can result in an accurate pharmacophore model. Such a model can be used for virtual screening, and to assist design of second-generation peptidomimetics with improved properties and design of non-peptidic leads. In this article, we will review examples in which a structural approach utilizing peptide ligands was employed to obtain a better understanding of the target active site. We will focus on cases where such information supplied guidance toward the discovery of small molecule ligands.

Drug Evaluation, Preclinical↗

PubMed2005

A naive bayes classifier for prediction of multidrug resistance reversal activity on the basis of atom typing.

Multidrug resistance (MDR), the ability of cancer cells to become simultaneously resistant to different drugs, remains an unsolved challenge in cancer chemotherapy. The use of MDR reversal (MDRR) agents is a promising approach to overcome this problem. For the design and development of such agents, it would be desirable to have a reliable model to estimate the MDRR activity of compounds. Presented here is a naive Bayes classifier to categorize MDRR agents into active and inactive classes, which uses a universal, generic molecular-descriptor system.(1) The naive Bayes classifier was built from a 424 compound training set, selected from 609 druglike compounds in the publicly available "Klopman set". The model correctly predicted MDRR activities for 82.2% of 185 compounds in a testing set. The cumulative probabilities were proven useful for prioritizing the compounds for testing. The impact of attribute dependences on the performance of the classifier was examined. As an unsupervised learner with no tuning parameters, a naive Bayes classifier is capable of providing an objective comparison of the effectiveness of different molecular descriptors. The relative performance of the classifiers constructed from either an atom-type-based molecular descriptor or the long-range functional-class fingerprint descriptors FCFP_6 or FCFP_2 was compared. Employing an atom typing descriptor with the naive Bayes classification, it enables the interpretability of the resulting model, which offers extra information for the rational design of MDRR agents.

Bayes Theorem↗

PubMed2004

A predictive pharmacophore model of human melanocortin-4 receptor as derived from the solution structures of cyclic peptides.

Using nuclear magnetic resonance (NMR) spectroscopy, we have determined the solution structures for a series of potent agonists for the human melanocortin-4 receptor (hMC4R), based on the cyclic peptide MT-II [Ac-Nle-cyclo-(Asp-Lys) (Asp-His-(D)Phe-Arg-Trp-Lys)-NH2]. Members of this series were designed to improve selectivity for MC4R versus the other melanocortin receptors, and to reduce the flexibility of the side chains. The most selective and rigid analog [penta-cyclo(D-K)-Asp-Apc-(D)Phe-Arg-(2S,3S)-beta-methylTrp-Lys-NH2] was found to be a full agonist of hMC4R with an EC50 of 11nM against hMC4R, and to exhibit 65-fold selectivity against hMC1R. This compound represents the most constrained hMC4R peptide agonist described to date. A beta-turn structure was conserved among all of the cyclic peptides studied. The rigidity of the analogs allowed an exceptionally well-defined pharmacophore model to be derived. This model was used to perform a virtual screen using a library of 1000 drug-like compounds, to which a small set of known potent ligands had been intentionally added. The utility of the model was validated by its ability to identify the known ligands from among this large library.

Humans↗

PubMedDate not supplied

A universal molecular descriptor system for prediction of logP, logS, logBB, and absorption.

Predictive models for octanol/water partition coefficient (logP), aqueous solubility (logS), blood-brain barrier (logBB), and human intestinal absorption (HIA) were built from a universal, generic molecular descriptor system, designed on the basis of atom type classification. The atom type classification tree was trained to optimize the logP predictions. With nine components, the final partial least-squares (PLS) model predicted logP of 10850 compounds in Starlist with a regression coefficient (r2) of 0.912, cross-validated r2 (q2) of 0.892, and root-mean-square error of estimation (RMSEE) of 0.50 log units. The PLS models for solubility (logS), blood-brain barrier (logBB), and a PLS-DA (discrimination analysis) model for HIA were established from the same atom type descriptors. The seven-component PLS model derived from a diverse set of 1478 organic compounds predicted a 21-compound test set designed by Yalkowsky with r2 = 0.88 and RMSEP (RMS error of prediction) = 0.64. A predictive r2 = 0.90 and RMSEE = 0.26 were achieved for logBB of a 57-compound "Abraham data set" with a three-component model. The first three components of a five-component PLS-DA model were sufficient to clearly separate the 169 drug molecules, collected by Abraham, into three classes, according to their percentage human intestinal absorption.

Absorption↗

PubMedDate not supplied

Prediction of chemical carcinogenicity from molecular structure.

Carcinogens represent a serious threat to human health. In vivo determination of carcinogenicity is time-consuming and expensive, thus in silico models to predict chemical carcinogenicity are highly desirable for virtual screening of compound libraries of both pharmaceutically and other commercially interesting molecules. In the present study, a PLS-DA (partial least squares discriminant analysis) model was developed to predict carcinogenicities in each of four rodent models: male mouse (MM), female mouse (FM), male rat (MR), and female rat (FR). The data set that was used contained over 520 compounds from both the NTP and the FDA databases. All the models were built from the same molecular descriptor system, which is based on atom typing [Sun, H. J. Chem. Inf. Comput. Sci. 2004, 44, 748-757], enabling the comparison of atomic contributions to carcinogenicity with respect to species and gender. Using four components, the models were able to achieve excellent fitting and prediction, with r(2) = 0.987 and q(2) = 0.944 for MM, r(2) = 0.985 and q(2) = 0.950 for FM, r(2) = 0.989 and q(2) = 0.962 for MR, and r(2) = 0.990 and q(2) = 0.965 for FR. The models were further validated by response permutation testing and external validation, and the results indicated that the models were both statistically significant and predictive. Variable influence on projection (VIP) analysis identified the key atom types and fragments that contributed to carcinogenicities and response differences across species and gender.

Animals↗

PubMedDate not supplied

Enhancing specificity and sensitivity of pharmacophore-based virtual screening by incorporating chemical and shape features--a case study of HIV protease inhibitors.

Virtual screening (VS), if applied appropriately, could significantly shorten the hit identification and hit-to-lead processes in drug discovery. Recently, the version of VS that is based upon similarity to a pharmacophore has received increased attention. This is due to two major factors: first, the public availability of the ZINC1 conformational database has provided a large selection pool with high-quality and purchasable small molecules; second, new technology has enabled a more accurate and flexible definition of pharmacophore models coupled with an efficient search speed. The major goal of this study was to achieve improved specificity and sensitivity of pharmacophore-based VS by optimizing the variables used to generate conformations of small molecules and those used to construct pharmacophore models from known inhibitors or from inhibitor-protein complex structures. By using human immunodeficiency virus protease and its inhibitors (PIs) as a case study, the impact of the key variables, including the selection of chemical features, involvement of excluded volumes (EV), the tolerance radius of excluded volumes, energy windows, and the maximum number of conformers in conformation generation, was explored. Protein flexibility was simulated by adjusting the sizes of EV. Our best pharmacophore model, combining both chemical features and excluded volumes, was able to correctly identify 60 out of 75 structurally diverse known PIs, while misclassifying only 5 out of 75 similar compounds that are not inhibitors. To evaluate the specificity of the model, 1193 oral drugs on the market were screened, and 25 original hits were identified, including 5 out of 6 known PI drugs.

Drug Design↗